§01 / 39

Electrical Theory Foundations

Read first if EE is new to you

Power Atlas assumes you know the fundamentals — what voltage is, why we use AC, what reactive power means physically. This page covers all of it. After reading, §02's formulas will make sense, not just be memorized.

The Three Quantities — Voltage, Current, Resistance

Quantity	Symbol	Unit	Plumbing analogy	What it physically is
Voltage	V	Volt (V)	Water pressure (PSI)	Electrical potential difference between two points. The "push" that wants to move charge. Measured between two points.
Current	I	Ampere (A)	Water flow rate (gal/min)	Charge flowing per second. Coulombs per second. Measured by clamp meter around a single conductor.
Resistance	R	Ohm (Ω)	Pipe friction	Opposition to current flow. A property of the conductor (size, length, material).
Power	P	Watt (W)	Water power (PSI × GPM)	Rate of energy transfer. Voltage × Current.
Energy	E	Watt-hour (Wh)	Total water moved	Power × time. What the utility bills you for (kWh).

Ohm's Law — The Foundational Equation

Voltage equals current times resistance. Memorize the triangle: cover what you want to find, the formula appears.

Three forms of one equation

V = I × R

I = V / R

R = V / I

i

Concrete example

A 12V car battery connected to a headlight bulb with R = 6 Ω draws I = 12/6 = 2 A. The bulb consumes P = V × I = 12 × 2 = 24 W. If the headlight runs 1 hour, energy used = 24 Wh = 0.024 kWh.

Kirchhoff's Laws — How Current and Voltage Behave in Circuits

Law	Statement	What it means in practice
KVL — Kirchhoff's Voltage Law	Sum of voltages around any closed loop = 0	Voltage drops across components in a loop add up to the source voltage. Why we calculate voltage drop along a feeder + branch + load = source voltage.
KCL — Kirchhoff's Current Law	Sum of currents entering a node = sum of currents leaving	Current splits at a junction in proportion to inverse impedance. Why parallel branches share current. Why neutral current is the unbalanced sum of phase currents.

DC vs AC — Why We Use Alternating Current

	DC (Direct Current)	AC (Alternating Current)
Direction of flow	One direction, constant	Reverses 60 times per second (US) or 50 (Europe)
Voltage transformation	Hard — requires DC-DC converters (inefficient at scale)	Easy — passive transformer steps up/down with ~ 99% efficiency
Long-distance transmission	Lossy at low voltage; requires HVDC for long distance (specialized + expensive)	Easy — step up to MV/HV, transmit, step down. I²R losses minimized.
Where used	Batteries, electronics, telecom, EV traction	Everything between the power plant and the wall outlet
Why it won (1880s)	—	The transformer made AC scalable. Tesla + Westinghouse beat Edison's DC for distribution.

The Sinusoid — What "AC" Actually Looks Like

RMS = the equivalent DC voltage that would produce the same heating in a resistor. Always use RMS for AC power calculations.

Frequency — Why 60 Hz?

60 Hz is the US/North America standard. 50 Hz is most of Europe + Asia. Higher frequency = smaller transformer cores (good) but more transmission line losses (bad). 60 Hz was Westinghouse's choice; 50 Hz was AEG's. Both work; the world settled on regional standards by ~1920.

Why Three-Phase Is the Standard

Three sinusoids, equal magnitude, 120° apart. Why this beats single-phase and 2-phase:

Property	Why 3-phase wins
Constant total instantaneous power	P₁(t) + P₂(t) + P₃(t) = constant. (For 1-phase, total power pulses at 120 Hz. For 2-phase, it pulses too.) Constant power = smooth motor torque, no vibration.
Minimum copper for transmission	3 wires for the same kW vs 2 wires for 1-phase = ~ 25% LESS copper per kW transmitted. (4 wires for 2-phase = WORSE.) This is why utilities use 3-phase everywhere.
Self-starting motors	3-phase produces a uniform rotating magnetic field. Motor starts on its own. (1-phase motors need start capacitors or shaded poles.)
Neutral can be omitted	Balanced 3φ has zero current in the neutral — can use 3 wires only. (Unbalanced loads need a neutral.)

The Magic of Transformers — Why AC Won

Faraday's Law (1831) — a changing magnetic field induces voltage in a nearby conductor. AC's constant change makes this practical:

AC current in primary winding creates a changing magnetic flux in the iron core.
The changing flux induces a voltage in the secondary winding.
The voltage ratio = the turns ratio: V₁/V₂ = N₁/N₂.
Power is conserved (minus tiny losses): V₁ × I₁ = V₂ × I₂. Step up voltage → step down current.

Why this matters: at the power plant, generate at 13.8 kV. Step up to 230 kV for transmission (low current = low I²R losses). Step down to 12.47 kV for distribution. Step down to 480V at the building. Step down to 120V at the outlet. Five voltage levels, five transformers, one path.

Iron core couples primary AC magnetic flux to secondary winding. Step up V → step down I.

Resistance + Inductance + Capacitance — The Three Passive Elements

Element	Symbol	What it does	How it acts in AC	Phase relationship
Resistor	R	Dissipates energy as heat. Pure friction.	Voltage and current rise/fall together. No energy storage.	V and I IN PHASE (PF = 1.0)
Inductor (L) — coil of wire, motor windings, transformer primary	L	Stores energy in magnetic field. Resists changes in current.	Voltage LEADS current by 90°. Current lags.	I lags V by 90° (PF = 0)
Capacitor (C) — two metal plates separated by insulator	C	Stores energy in electric field. Resists changes in voltage.	Voltage LAGS current by 90°. Current leads.	I leads V by 90° (PF = 0)

Why this matters: Reactive Power Explained Physically

An inductor (motor coil) doesn't dissipate energy — it stores energy in its magnetic field, then releases it. The current flowing back and forth creates this storage/release cycle. The current is real (you have to size wires for it), but no NET energy gets used.

This is what reactive power (Q, in kVAR) is — the apparent flow of power that just sloshes back and forth between source and load. It uses no fuel at the power plant but does use copper in the wires.

i

Why utilities care about reactive power

A factory drawing 100 kW at PF = 0.8 actually has 100 kVA × current passing through utility transformers (because S = P/PF = 125 kVA). The utility sized infrastructure for 125 kVA but only collected revenue for 100 kW. Power factor correction (capacitors) cancels the inductive sloshing, brings PF closer to 1.0, and frees up utility capacity.

Impedance — Generalized Resistance for AC

For AC, the opposition to current isn't just resistance. It includes inductive reactance (X_L) and capacitive reactance (X_C) too.

Impedance (Z) magnitude

|Z| = √(R² + X²)

where X = X_L − X_C (net reactance). Z replaces R in Ohm's Law for AC: V = I × Z.

Reactance	Formula	Notes
Inductive (X_L)	X_L = 2π × f × L	Higher frequency → more reactance. (Why high-frequency harmonics get blocked by inductors.)
Capacitive (X_C)	X_C = 1 / (2π × f × C)	Higher frequency → LESS reactance. (Why capacitors short out high-frequency noise.)

Phasors — Why Engineers Draw Triangles

Each AC voltage or current is a sinusoid with magnitude AND phase angle. Adding two sinusoids of different phases is messy with trigonometry. Phasors represent each sinusoid as an arrow (length = magnitude, direction = phase angle), then you add the arrows like vectors.

A phasor is just a 2D arrow in the complex plane. The power triangle (§01) is a phasor diagram.

Series + Parallel Circuits

Configuration	Resistance	Voltage	Current
Series (one path)	R_total = R₁ + R₂ + ...	V splits across each R	Same I through every R
Parallel (multiple paths)	1/R_total = 1/R₁ + 1/R₂ + ...	Same V across every R	I splits inversely to R
Two parallel resistors (special case)	R_total = (R₁ × R₂) / (R₁ + R₂)	—	—

Why this matters in power systems: branch circuits in a panel are in PARALLEL with each other (all share the bus voltage; current splits per load). Conductors in PARALLEL feeders share current proportionally to inverse impedance — equal-length matched conductors share equally; mismatched ones don't. NEC 310.10(H) requires identical termination + length for paralleled conductors precisely because of this.

Superposition Principle

For any LINEAR circuit with multiple sources, the response (voltage or current) at any point equals the SUM of responses caused by each source individually (with all other sources turned off — voltage sources shorted, current sources opened).

i

Where superposition applies in power systems

Multiple-source fault analysis: contribution from utility + on-site generation + motor back-feed. Sum each source's contribution to total fault current.
Power flow with co-generation: utility provides part of load, on-site PV provides rest. Solve each separately, sum.
Harmonic analysis: each harmonic frequency is solved separately, then summed (since circuit elements respond differently at each frequency).

Limitation: only works for LINEAR circuits. Doesn't apply to circuits with magnetic saturation (transformer cores at high flux) or semiconductor switches (VFD outputs).

Thevenin + Norton Equivalents

Any complex network of voltage sources, current sources, and resistors can be reduced to a single equivalent source with a single equivalent impedance, when viewed from any pair of terminals.

Equivalent	What it is	How to find
Thevenin equivalent (V_th, R_th)	Voltage source V_th in SERIES with resistance R_th	V_th = open-circuit voltage at the terminals. R_th = resistance looking into the terminals with all sources zeroed.
Norton equivalent (I_n, R_n)	Current source I_n in PARALLEL with resistance R_n	I_n = short-circuit current at the terminals. R_n = R_th (same as Thevenin).
Conversion	—	V_th = I_n × R_n · I_n = V_th / R_th

i

Why Thevenin matters: it's how the utility looks to you

When the utility tells you "available fault current at the service is 50 kA at 12.47 kV," they're giving you a Thevenin equivalent of the entire grid behind your meter. V_th = 12.47 kV (no-load voltage), R_th = 12.47 kV / 50 kA = 0.144 Ω. The 50 kA is the Norton current (short-circuit current). Same information, two notations.

Symmetrical Components — Intro

Real power systems aren't perfectly balanced. Single-phase faults, ground faults, broken conductors all create UNBALANCED 3-phase conditions. Analyzing them with regular phasors is messy. Symmetrical components decompose any unbalanced 3-phase set into THREE balanced sets:

Sequence	Symbol	What it is	When it exists
Positive sequence	V₁, I₁	Balanced 3φ rotating ABC	Always present in normal operation
Negative sequence	V₂, I₂	Balanced 3φ rotating ACB (reverse)	Created by phase-phase faults, single-phasing of motors, unbalanced loads
Zero sequence	V₀, I₀	3 in-phase quantities (no rotation)	Created by ground faults; flows in neutral; only exists in 4-wire systems

Each sequence has its own equivalent impedance for any piece of equipment. Faults are then analyzed by combining sequence networks. Full deep dive in §12 Short Circuit.

Power = Energy / Time — kW vs kWh

The most common confusion in EE literacy. Let's nail it:

	Power (kW)	Energy (kWh)
What it is	Rate of energy transfer (instantaneous)	Total energy moved over time
Unit	Watt = Joule/sec; kW = 1000 W	Watt-hour = 3600 J; kWh = 1000 Wh
Plumbing analogy	Flow rate (gal/min)	Total water moved (gallons)
Math	—	kWh = kW × hours
What you're billed for	Demand charge ($/kW)	Energy charge ($/kWh)
Typical magnitudes	House peak: 5-15 kW. Office: 50-500 kW. Atlas DC1: 5,000 kW.	House annual: 10,000 kWh. Office annual: 1M kWh. Atlas DC1 annual: 44M kWh.

A 100W lightbulb running for 10 hours uses 1 kWh of energy at a power rate of 0.1 kW. Same lightbulb running for 1 hour: still 0.1 kW power, but only 0.1 kWh energy.

Magnetic Field Basics — How Motors Work

Current creates magnetic field. A wire carrying current is surrounded by a circular magnetic field (right-hand rule). Coil the wire → magnetic field becomes a "bar magnet" with N + S poles.
Magnetic field exerts force on a current-carrying wire. If you put a current-carrying wire in another magnetic field, the two fields push the wire perpendicular to both (F = I × L × B).
Combine the two: a stator (stationary winding) creates a magnetic field. A rotor (moving winding or magnet) sits inside the stator and feels force → rotor spins. This is a motor.
Spin a rotor in reverse — get a generator. Mechanical input → magnetic field changes around rotor → induces voltage in stator → output power.
3-phase magic: three windings 120° apart in space, fed by 3-phase currents 120° apart in time, create a smoothly rotating magnetic field that drags the rotor along. This is why 3-phase induction motors are self-starting.

The Power System Hierarchy

Level	Voltage	What happens here	Where in Atlas DC1
Generation	13.8 kV typical (at the generator)	Hydro, gas, nuclear, wind, solar generators produce electricity	The utility's plants — not on Atlas DC1 site
Transmission	69 - 765 kV	Long-distance bulk power transfer. Step up to high voltage to minimize I²R losses over hundreds of miles.	—
Sub-transmission	34.5 - 69 kV	Intermediate distribution from transmission substations to local distribution substations	—
Primary distribution	4.16 - 34.5 kV (12.47 kV most common)	From distribution substation to neighborhoods/customers. Customer's MV service feeders.	Atlas DC1 utility service: 12.47 kV
Secondary distribution	120/240V (residential), 480Y/277V (commercial)	Service transformer step-down to utilization voltage	Atlas DC1 480Y/277V main bus
Utilization	120, 208, 240, 277, 415, 480V	The voltage your equipment runs on	Lighting (277V), receptacles (120V), motors (480V), IT (415Y/240V)

Generator → Wall Outlet — One Joule's Journey

Coal/gas/nuclear/water turns a turbine. Turbine spins a 60 Hz synchronous generator at 13.8 kV.
Step-up transformer brings it to 230 kV for transmission.
Power flows hundreds of miles via overhead transmission lines.
Substation step-down: 230 kV → 69 kV sub-transmission.
Distribution substation: 69 kV → 12.47 kV primary distribution.
Neighborhood pole-mount transformer: 12.47 kV → 120/240V single-phase to your house. (Or → 480Y/277V three-phase to a commercial building.)
Inside the building: panelboard → branch circuit → wall outlet.
Your laptop charger: 120V AC → AC-DC → DC-DC → 19V DC at the laptop.

Total efficiency from fuel to laptop: about 30%. (Most loss at the power plant — thermodynamic limits.) Transmission + distribution losses combined are typically only 5-10%.

Where to Go Next

Now that the theory is in place:

If you're new to design: read §01 (Conversions) → §02 (How Design Starts) → §03 (Load Analysis). The formulas will make sense, not just be memorized.
If you have an EE degree but it's been a while: skim §01, focus on §04 (Branch Circuits) onward.
If you're studying for the PE Power exam: the parent PE Power site has exam-focused practice with the same notation conventions.

Theory primer · The "why" behind the "what" · Skip if you have an EE background

§02 / 39

Conversions & Equations

Foundation · The math toolkit

Every section that follows references one of these formulas. Memorize them through reps. Once they're reflex, the rest of electrical design becomes choosing which one to apply.

The Power Triangle

The single most important visual in electrical engineering. Real power (P) does work. Reactive power (Q) supports magnetic fields in motors and transformers. Apparent power (S) is what the conductors actually carry. Conductors and transformers are sized for S (apparent), not P (real).

Conductors carry S. Loads consume P. Utilities bill on both.

Pythagorean relationship

S² = P² + Q²

Apparent power is the vector sum. Always.

Power factor

PF = cos θ = P / S

PF = 1.0 means all power is real (resistive load). PF = 0.8 means 80% of apparent power is real, 60% is reactive.

Solving for any side

P = S · PF

Q = S · sin θ

The Six Core Formulas — 1φ vs 3φ

Memorize these six. They cover ~90% of the conversion math you'll do for the rest of your career.

Quantity	Single-phase (1φ)	Three-phase (3φ)	Why the difference
Apparent S (kVA)	`V × I / 1000`	`√3 × V_LL × I / 1000`	3-phase has three conductors carrying current — the √3 captures the geometry of three sinusoids 120° apart. V_LL = line-to-line voltage.
Real P (kW)	`V × I × PF / 1000`	`√3 × V_LL × I × PF / 1000`
Reactive Q (kVAR)	`V × I × sin θ / 1000`	`√3 × V_LL × I × sin θ / 1000`
Solve for I from kVA	`I = kVA × 1000 / V`	`I = kVA × 1000 / (√3 × V_LL)`	Most common reverse — sizing conductors from a known load
Solve for I from kW	`I = kW × 1000 / (V × PF)`	`I = kW × 1000 / (√3 × V_LL × PF)`	If you only know P, you must include PF to get I
HP → kW	`kW = HP × 0.746` · `HP = kW × 1.341`		Output (mechanical) — does NOT account for motor efficiency or PF

√3

Why √3 in three-phase formulas?

Three sinusoidal voltages, equal magnitude, 120° apart. When you measure line-to-line voltage (between two phase conductors), the geometry of two phasors 120° apart yields √3 ≈ 1.732 × the line-to-neutral voltage. So V_LL = √3 × V_LN.

Examples that should be reflex: 277 × √3 = 480 · 120 × √3 = 208 · 2400 × √3 = 4160 · 7200 × √3 = 12,470.

The HP → kW → FLA Chain

The most-used calculation in motor work. Three steps. Each step needs one new piece of nameplate data.

For NEC purposes, FLA comes from NEC Table 430.250 not the nameplate (the table is conservative)

Conversions Worth Memorizing

HP → kW

×0.746

1 HP = 0.746 kW exactly. Reverse: × 1.341.

3φ Line-to-line

×√3

V_LL = √3 × V_LN. 277 → 480, 120 → 208, 2400 → 4160.

kVA at 480V 3φ

×1.203

kVA × 1.203 ≈ FLA. Useful sanity check on 480V loads.

kW → BTU/hr

×3,412

Server heat load → cooling requirement.

tons → kW (cooling)

×3.517

1 ton refrigeration ≈ 3.517 kW heat removed.

Δ% drop, 1φ

2VIL/CM

VD = (2 × L × I × R) / 1000. Use NEC Ch9 Table 9 ohms.

Worked Example 1 — Atlas DC1 Chiller Motor

Atlas DC1 has four 750-ton centrifugal chillers. Each is driven by a single induction motor. The mechanical engineer's MEL gives you the chiller, the manufacturer's cutsheet gives you the rest. You need FLA to start branch-circuit design.

Example 01 · Atlas DC1 spine Chiller CH-1 · 750-ton centrifugal · Trane CenTraVac equivalent

Given (from cutsheet)

Motor HP

450 HP nominal

Voltage

480V, 3φ, 60Hz

Efficiency η

96.0% (NEMA Premium)

Power factor

0.91 at full load

Step-by-step

HP → kW (mechanical output)

kW_out = 450 HP × 0.746 = 335.7 kW
kW (output) → kW (input) — divide by efficiency to get the electrical power the wire must deliver

kW_in = 335.7 / 0.96 = 349.7 kW
kW → kVA — divide by PF to get apparent power (what conductors carry)

kVA = 349.7 / 0.91 = 384.3 kVA
kVA → FLA — divide by √3 × V_LL

FLA = (384.3 × 1000) / (√3 × 480) = 384,300 / 831.4 = 462 A
Cross-check against NEC Table 430.250 — for 450 HP at 460V the table lists FLC = 480 A (close to our 462; NEC table is slightly conservative)

For NEC sizing use FLC = 480 A from NEC 430.250, not the calculated 462 A. NEC requires the table value (430.6).

!

NEC 430.6(A)(1) — the key trap

For motor sizing under NEC, the table values in 430.247–430.250 are used, NOT the nameplate FLA. The nameplate is for actual current measurement; the table is for design. They will differ by 5–15%, and the test/exam will check that you used the table value.

Worked Example 2 — Apartment Service Calculation

The math doesn't change for residential. Smaller voltages, single-phase split, but the same conversions. This example shows you the same tools applied at the other end of the spectrum.

Example 02 · Alternate scale A single 200A 120/240V apartment unit · NEC 220 Standard Method

Given

Service

200A, 120/240V, 1φ-3-wire

Connected load

28.4 kVA (after NEC 220 demand factors)

Find: actual service current

Single-phase formula: I = kVA × 1000 / V

I = (28.4 × 1000) / 240 = 118.3 A

Service breaker = 200 A is correctly sized — the demand load fits below the breaker, with margin for code-required 125% on continuous portions.
Why 240, not 120? Most large appliances (range, dryer, A/C, heat) use 240V. The service voltage for current calculation is the line-to-line value, even on split-phase.
Sanity check with HP-equivalent: 28.4 kVA at PF ≈ 1.0 = 28.4 kW = ~38 HP equivalent. A reasonable apartment runs ~5–8 kW peak; the 28.4 here represents NEC 220 demand load including diversity for a fully-equipped 2-bedroom unit.

i

Same formula, different scale

Atlas DC1's chiller calculation and this apartment calculation use the identical I = kVA / (k × V) formula. The only difference is k = 1 (single-phase) vs k = √3 (three-phase). Once the formula is muscle memory, scale is just substitution.

Per-Unit Math — The Foundation of Fault Analysis

Per-unit (pu) normalizes every quantity to a common base. Once normalized, you can add transformer impedances directly across voltage levels without conversion. Used in fault analysis (§12), load flow (§16), and protection coordination (§17).

Choose a base — typically

S_base = pick (e.g., 100 MVA or transformer kVA)

V_base = system voltage at the bus

Once chosen, every other base derives from these two.

Derived bases

I_base = S_base / (√3 × V_base)

Z_base = V_base² / S_base

Convert any quantity to pu

Z_pu = Z_actual / Z_base

Transformer %Z = 100 × Z_pu on the transformer's own base.

Convert pu impedance to a different base

Z_pu,new = Z_pu,old × (S_new / S_old) × (V_old / V_new)²

Critical when combining transformers of different sizes.

Example · Atlas DC1 spine Per-unit conversion of TX-A into a system-wide base

Pick a system base

S_base

100 MVA (arbitrary, chosen for clean numbers)

V_base at MV

12.47 kV

V_base at LV

480V

Step-by-step

TX-A nameplate impedance: 5.75% on TX-A's own base of 2,500 kVA.

Z_pu,TX on own base = 0.0575 pu
Convert to 100 MVA system base:

Z_pu,TX,new = 0.0575 × (100,000 / 2,500) × (12.47/12.47)² = 0.0575 × 40 × 1 = 2.30 pu
Compute base current at 480V bus:

I_base,LV = 100,000 / (√3 × 0.480) = 120,300 A
Fault current at 480V (assuming infinite source feeding TX-A):

I_fault,pu = 1.0 / 2.30 = 0.435 pu
I_fault,actual = 0.435 × 120,300 = 52,300 A

Same answer as the simple FLA / %Z approximation in §09 — because in this case, the utility was assumed infinite. With real utility impedance (per §12), the fault drops to ~50.3 kA.

i

Why per-unit matters

In multi-voltage systems (utility 12.47 kV → 480V → 415V → 240V), per-unit lets you add impedances without converting voltages back and forth. Every transformer's %Z just becomes a pu number on the system base, and you sum them. The MVA method (§12) is a shortcut version of per-unit math.

If You See THIS, Think THAT

If you see…	Think / use…
"480V" or "480Y/277V" on a cutsheet	3-phase. Use √3 in formulas. V_LL = 480, V_LN = 277.
"208Y/120V" or "120/208V"	3-phase wye. Most common commercial. V_LL = 208, V_LN = 120.
"120/240V"	Single-phase 3-wire (residential). NO √3. V_LL = 240.
HP given on motor nameplate	Mechanical output. Multiply by 0.746 to get kW; then divide by η × PF for kVA.
"Calculate motor branch circuit"	Use FLC from NEC Table 430.250, NOT the nameplate FLA. (Per NEC 430.6.)
kVA given but you need amps	3φ: `I = kVA × 1000 / (√3 × V_LL)` · 1φ: `I = kVA × 1000 / V`
kW given (no PF mentioned)	Stop. You need PF before you can find kVA or amps. If load is purely resistive (heaters, incandescent), PF = 1.0 and kW = kVA.
"Per-unit" in a problem	Quantities are normalized to a base. Always find S_base and V_base first; then I_base = S_base / (√3 × V_base).
Server load in kW for cooling sizing	Multiply kW × 3,412 for BTU/hr, or kW / 3.517 for cooling tons. (Atlas DC1: 2.5 MW IT load = 711 tons of cooling.)
"Tons" on chiller cutsheet	Refrigeration tons. 1 ton = 3.517 kW heat removed. Motor input is different — see HP→FLA chain.

Drill — Quick Self-Check

Five problems. Hide answers; work them mentally; reveal to check. The goal is reflex, not deliberation.

Drill 01 · 1φ → amps

A 5 kVA, 240V single-phase load. What is the current?

Drill 02 · 3φ → amps

A 75 kVA load at 480V 3φ. What is the current?

Drill 03 · HP → kW

A 100 HP motor. What is mechanical output in kW?

Drill 04 · kW → kVA

A 100 kW load at PF = 0.85. What is kVA?

Drill 05 · Atlas DC1 quick math

Atlas DC1 IT load = 2.5 MW. PF for the IT load = 0.95 (modern PSU). What kVA does the UPS see?

Drill 06 · cooling math

Atlas DC1 IT load = 2.5 MW. How many tons of cooling does it need?

§03 / 39

How Design Starts

MEL · SLD · Voltages · Cutsheets

The electrical engineer doesn't choose what loads exist. Other disciplines do. Your work begins when you receive a list of equipment that needs power — and ends when every single one is wired, protected, and labeled.

The Workflow — Where Each Document Lives

Every project follows the same five documents. They get bigger and more detailed as design progresses, but the order doesn't change. Memorize this flow — every section in Power Atlas slots into one of these stages.

An MEL row that's missed in Stage 1 becomes a missing breaker in Stage 5 — and a 3 a.m. RFI from the field

Meet Atlas DC1 — Your Reference Facility

Every section in this handbook returns to Atlas DC1. It's a representative 2.5 MW colocation data center built in 2N redundant topology — large enough to span every electrical concept, small enough to hold in your head all at once. Here is the full one-line. Spend a minute reading it. We'll dissect every piece across the next 31 sections.

IT load

2.5 MW

Critical load. PF ≈ 0.95 → ~2,632 kVA delivered by UPS at full load.

Topology

2N

Two independent paths (A and B). Either path alone carries the full IT load.

Tier rating

III Uptime

Concurrently maintainable — any single path can be removed for service without dropping load.

Service

12.47 kV

Utility primary. Two pad-mount transformers step down to 480V building distribution.

Genset capacity

5 MW

2 × 2500 kW. Each genset alone carries one full A-or-B path. Backup for utility loss.

Cooling

3000 tons

4 × 750-ton centrifugal chillers. One in standby (N+1 mech inside the 2N elec).

The MEL — What You Receive

The MEL is the input document. Other disciplines fill it out. Mechanical lists pumps and chillers. Process lists production equipment. Plumbing lists water heaters. IT lists racks and PDUs. Each row is a load you must power.

Anatomy of an MEL Row

Real MELs vary by employer, but the essential columns are universal. Here is a row from Atlas DC1's MEL — Chiller CH-1 — with each column annotated.

MEL Column	Atlas CH-1 value	What you do with it	Source
Tag / Equipment ID	CH-1	Becomes the panel-schedule row label, the SLD callout, the cable schedule reference	Mechanical assigns
Description	750-ton centrifugal chiller	Confirms load type (motor) and informs duty cycle for power-quality + protection sections	Mech / process
Quantity	4 (CH-1, CH-2, CH-3, CH-4)	Multiply for total capacity; consider redundancy / standby in load study	Mech
Voltage	480V, 3φ, 60Hz	Determines which panel/MCC it connects to; selects correct calculation formula	Mech (per cutsheet)
HP / kW	450 HP nominal	Starting point for FLA calculation; starting point for load study before efficiency & PF	Mech (per spec)
FLA / FLC	480 A (NEC 430.250)	Direct input to wire size, breaker size, panel/MCC bus size	You fill in (or cutsheet)
MCA	600 A (1.25 × FLC)	Minimum wire ampacity; sets conductor size	You calculate
MOCP	1200 A (250% × FLC, NEC 430.52)	Maximum breaker size; set actual ≤ this	You calculate (or cutsheet)
Starting type	VFD (variable freq drive)	Affects starting current, harmonic mitigation, branch circuit conductor choice	Mech / electrical
Location	Mech room MR-1	Determines feeder length → voltage drop calc; raceway routing	Architect / mech
Service factor	1.15	Affects motor overload setting; allows brief overloads	Cutsheet
Code letter	F	Locked-rotor kVA per HP — only relevant for DOL starting (less important for VFD)	Cutsheet
Notes	Standby duty (N+1)	One chiller is N+1 standby — load study uses 3 running, not 4	Mech

ANSI Standard Voltages

Voltages aren't arbitrary. ANSI C84.1 defines the discrete standard voltages used in North America — utilization (at the load), system (utility-supplied nominal), and the tolerances around each. You'll see these exact numbers on every cutsheet.

Class	System V (LL)	Phase / Wire	Where used	Typical Atlas DC1 use
Low (≤ 600 V)	120 / 240 V	1φ-3W	Residential, light commercial, control circuits	Office spaces
	208Y / 120 V	3φ-4W	Small commercial, retail, downstream of step-down	Office, lighting at 277V (when 480Y) or 120V branches
	480Y / 277 V	3φ-4W	Industrial, commercial mech room, MV-fed buildings	Atlas main 480V bus — chillers, UPS input, large pumps
Medium (1 – 35 kV)	2400 / 4160 V	3φ-3W or 4W	Large industrial motors, in-plant distribution	—
	13.8 kV	3φ-3W	Common industrial primary, large MV motors	Used on bigger DCs (>10MW); not Atlas DC1
	12.47 kV / 7.2 kV	3φ-4W	Utility distribution primary	Atlas DC1 utility service
High (35 – 230 kV)	69 kV	3φ	Sub-transmission, large industrial direct service	Hyperscale (≥50MW) DCs
High (35 – 230 kV)	115 / 138 / 230 kV	3φ	Transmission, utility substation primary	Hyperscale campuses with on-site substations

i

Utilization vs system voltage

Cutsheets specify utilization voltage (e.g., 460V, 230V, 200V — what the equipment actually expects to see at its terminals after voltage drop). The system voltage is the nominal source value (e.g., 480V, 240V, 208V). The standard pairs them: 480 system → 460 utilization. Don't confuse the two when reading nameplates.

SLD Symbol Legend

A single-line diagram is a schematic shorthand. Three-phase systems get drawn with one line, even though three conductors exist. Equipment is symbolic. Here are the symbols you'll see on every commercial / industrial drawing.

SLDs use single lines for 3-phase circuits. Counts (3φ, 3W vs 4W) are noted in text or with tick marks across the line.

Worked Example 1 — Atlas DC1 MEL Walk-Through

Below is a fragment of Atlas DC1's MEL exactly as you would receive it. Three rows: a chiller, a pump, and an IT-room PDU. Walk through what each row tells you.

Example 01 · Atlas DC1 spine Three rows from the MEL — what to extract

Tag	Description	V	HP/kW	FLA	Starter	Location	Notes
CH-1	Centrifugal chiller	480V 3φ	450 HP	480 A	VFD	MR-1	N+1, water-cooled
CWP-1	Cond water pump	480V 3φ	75 HP	96 A	VFD	MR-1	1-per-chiller
PDU-A1	IT power dist unit	480→415Y/240V	500 kVA	602 A in	n/a	IT Hall A	Side A · UPS-fed

What each row tells you

Row 1 — CH-1: A 450 HP motor on a 480V 3φ system, started by a VFD. Located in mechanical room MR-1. There is a fourth identical chiller (N+1 standby).

Action: branch circuit on the 480V SWGR — A bus. FLA = 480 A from NEC table. Wire ≥ 600 A (MCA = 1.25 × FLA). Breaker ≤ 1200 A (MOCP = 250%, NEC 430.52). Connect ahead of VFD with output reactor — see §14 Motors.
Row 2 — CWP-1: Smaller motor, 75 HP. Same voltage, same MCC bus. Co-located with chiller (MR-1).

Action: branch circuit on same MCC. FLA 96 A → MCA 120 A → wire #1 AWG (per NEC 310.16, 75°C). Breaker ≤ 240 A (per 430.52). Pump cycles with chiller via control logic — confirm with Mech.
Row 3 — PDU-A1: 500 kVA Power Distribution Unit. 480V input, 415Y/240V output. Located in IT Hall A.

Action: feed from UPS-A1 output, NOT directly from 480V SWGR (it's a critical IT load). Input current 602 A — requires 800 A breaker on UPS-A1's distribution panel. PDU is itself a step-down transformer plus distribution; see §09 Transformers.

!

The "Notes" column is where the danger lives

Half the field RFIs come from electrical engineers ignoring MEL notes. "Standby" means N+1 — adjust load study. "Water-cooled" means there's a cooling tower load you might miss. "UPS-fed" means a different feeder source than the rest of the room. Read every notes column twice.

Worked Example 2 — Tracing Power Flow on the Atlas DC1 SLD

Once you have the SLD, you should be able to start at any load and trace the full path back to the utility — naming every device, voltage, and protection along the way. This is the single most useful skill in commissioning, troubleshooting, and arc-flash work.

Example 02 · Atlas DC1 spine Trace one server's power: from rack PDU back to the utility

The path (server → utility)

Server power supply — 415V or 240V input
Rack PDU — 415Y/240V outlet on the rack
RPP / branch circuit — a 30A or 60A breaker in the row power panel
PDU-A1 — 500 kVA distribution unit, 480V→415Y/240V step-down + sub-distribution
UPS-A1 output bus — Critical UPS Bus A, 480V (before the PDU step-down)
UPS-A1 — 1250 kVA double-conversion static UPS, with VRLA battery string
480V SWGR — A — main switchgear bus, 4000A rated, fed from TX-A or GEN-A via ATS-A
ATS-A — automatic transfer switch, normal position = TX-A, transfers to GEN-A on utility loss
TX-A — 2500 kVA pad-mount transformer, 12.47kV→480Y/277V, %Z = 5.75
MV SWGR primary feeder — 12.47 kV, 3φ, 50kA fault current available
Utility source — local distribution feeder, 12.47kV grounded-wye

What this trace gives you

i

Voltage transitions

12.47 kV → 480V → 415V → 240V at the load. Four voltage levels, three transformations. Each step = a separately derived system requiring its own grounding (see §13).

⚠

Arc flash points

Eight equipment locations on this trace need an arc flash label (NEC 110.16). Incident energy varies wildly — highest at PDU-A1 distribution panel (close to UPS source, fast trip), lowest at MV switchgear (longer trip times).

!

Why 2N matters in this trace

Side B is identical and independent. Each server has dual PSUs (A + B). If anything on the A path fails, the server pulls from B with no interruption. This is why the cross-tie breaker between SWGR-A and SWGR-B is normally open — closing it would couple the two sides and defeat 2N.

If You See THIS, Think THAT

If you see…	Think / use…
"MEL" or "Equipment Schedule" handed to you	You're at Stage 1. Extract V, HP/kW, FLA, MOCP, location for every row before drawing a single line.
HP given but no FLA on the MEL	Use NEC Table 430.250 for FLC. Don't calculate from HP × 0.746 / V / PF for NEC sizing — that's nameplate, not table.
"480V" on a cutsheet	Means 480Y/277V 3φ-4W in commercial. Confirm wire count: 4-wire if 277V loads exist, 3-wire if motor-only.
"460V" on a motor nameplate	Utilization voltage — equipment expects 460V at terminals. System is 480V, with ~4% drop accepted.
Single-line shows a circle	Rotating machine — motor (M), generator (G), or sync condenser (SC). Letter inside identifies.
Single-line shows two intersecting circles	Two-winding transformer. Configuration (Δ-Y, Y-Y, etc.) labeled separately or shown with explicit windings.
Dashed line on the SLD	Either emergency/standby (often copper-colored), or a normally-open device. Read the label.
"2N" topology mentioned	Two completely independent paths. Cross-ties normally open. Each path sized for full load.
"N+1" topology mentioned	One redundant unit. Less expensive than 2N. Doesn't tolerate the failure of more than one unit.
"Tier III" in DC documentation	Concurrently maintainable — any single component can be taken offline for service without dropping load. 2N or 2(N+1) typical.
"PDU" in a data center	Power Distribution Unit — typically a 480→415Y/240V step-down transformer with downstream sub-panels. NOT a "plug strip." It's an entire piece of switchgear.
"RPP" in a data center	Remote Power Panel — a panelboard at the row level, fed from a PDU. Provides the actual rack-level branch circuits.
"Behind the UPS" or "critical bus"	Load is non-interruptible. Must be fed from UPS output, not utility-direct. Coordinated independent of mech loads.

§04 / 39

Load Analysis

Connected · Demand · Continuous · Diversity

The MEL gives you a list of loads. Load analysis turns that list into the numbers that size every transformer, every feeder, every breaker. Two loads matter most: the one if everything ran at once (connected), and the one that actually happens (demand).

Load Type Definitions — One Place

NEC and engineering practice define many overlapping load terms. Here they are, all in one table.

Term	Definition	Source
Continuous load	Maximum current expected to continue for 3 hours or more	NEC Article 100 (definitions)
Non-continuous load	Loads not classified as continuous (cyclic, intermittent, brief peaks)	Implicit from NEC 100
Connected load	Sum of all nameplate ratings, treating every load as if running at 100%	Engineering practice
Demand load	Maximum kW (or kVA) the system actually carries at peak — connected × demand factor	NEC 220, IEEE 141
Demand factor (D_f)	Max demand / total connected load. Always ≤ 1.0. NEC 220 publishes specific values per occupancy.	NEC 220
Diversity factor (D_v)	Σ individual peaks / system peak. Always ≥ 1.0. Applied across multiple feeders.	IEEE 141
Coincidence factor (C_f)	1 / D_v. Inverse of diversity. Common in residential utility load research.	IEEE 141
Load factor (L_f)	Average demand / peak demand over a period. Indicates how "flat" usage is.	Utility tariffs
Coincident load	Loads that DO peak together (heater + lighting both peak in winter evening)	Engineering judgment
Noncoincident load	Loads that CANNOT peak together (heating vs cooling, NEC 220.60)	NEC 220.60
Inrush current	Brief peak (typically 6-12× FLA) when energizing motors or transformers. Lasts less than 1 second.	Motor/transformer characteristics
Locked-rotor current (LRA)	Current a motor draws if rotor cannot turn. Typically 6-8× FLA. Sustained until protection trips.	NEC 430.7
Starting current	Current during motor acceleration. Decreases as motor reaches rated speed.	Motor characteristics
Cyclic load	Load that turns on/off in a regular pattern (elevators, welders, AC compressors)	Engineering judgment
Intermittent load	Brief operations followed by rest periods. NEMA defines duty cycles by ratio.	NEMA MG 1
Standby load	Load that's normally OFF but ready to operate (backup pumps, redundant equipment)	Engineering practice
Critical load	Load that must remain energized at all times (IT, life safety, process)	Engineering judgment
Sheddable load	Load that can be dropped without significant impact (lighting, comfort HVAC)	Demand response (§27)
Linear load	Load that draws current proportional to voltage (resistive heaters, incandescent)	Power quality (§15)
Nonlinear load	Load that draws current in non-sinusoidal pulses (rectifiers, VFDs, LEDs, servers). Generates harmonics.	Power quality (§15)

Connected vs Demand Load

Connected load is what you'd see if every load nameplate ran at 100% simultaneously. Demand load is what the system actually pulls at peak — after accounting for the fact that not everything runs, not everything runs at full output, and not everything peaks at the same moment.

Connected = sum of all nameplates. Demand = what the system actually carries at peak.

The Two Numbers Side by Side

	Connected Load	Demand Load
Definition	Sum of every load's nameplate, as if all ran simultaneously at 100%	Maximum kW (or kVA) the system actually sees at peak
Always larger by…	1.0× (reference)	0.4× to 1.0× (depends on diversity, demand factor)
Used for	Equipment room space estimate · transformer thermal limit ceiling · fault current · MCC bus design	Feeder ampacity · service entrance · transformer kVA · utility metering · generator sizing
NEC reference	—	Article 220 — demand factors per occupancy & load type
Atlas DC1	2.5 MW IT + 2.5 MW mech + 0.3 MW BOP ≈ 5.3 MW	2.5 MW IT (designed at 100%) + ~1.8 MW mech (at peak) + 0.2 MW BOP ≈ 4.5 MW

i

Atlas DC1 is unusual — IT is 100% demand

For most facilities, demand << connected because not everything peaks together. But IT load in a fully-loaded data center IS demand load — every server runs continuously. This is why DC1 is sized to deliver 2.5 MW continuously (not 2.5 MW occasionally). The mech load, however, has demand diversity: chillers ramp with cooling need, not all motors at full speed all the time.

Demand Factor vs Diversity Factor

Both reduce a number; they reduce different numbers and they live in different parts of the calculation. Distinguishing them is the first thing the PE exam tests in load analysis.

Factor	Formula	Typical range	Where applied
Demand factor	D_f = max demand / connected load	0.4 – 1.0	One single load category (e.g., the lighting demand factor for a warehouse). NEC 220 publishes specific values.
Diversity factor	D_v = Σ individual peaks / system peak	1.0 – 3.0+	Across multiple feeders/buildings — accounts for the fact that different consumers peak at different times.
Coincidence factor	C_f = 1 / D_v	0.3 – 1.0	Inverse of diversity — sometimes published this way (especially in residential service/utility work).
Load factor	L_f = average demand / peak demand (over a period)	0.3 – 1.0	Energy/revenue planning — not used directly for sizing, but tells you how "flat" or "peaky" your usage is. Atlas DC1: ≈ 0.95 (very flat).

NEC 220 Demand Factors — by Load Category

NEC Article 220 publishes the demand factors you must use for code-compliant sizing of feeders and services. Below are the most-used categories. Use these for the standard method (220.40).

Load type	Threshold / Tier	Demand factor	NEC reference
General lighting (dwelling)	First 3,000 VA	100 %	Table 220.45
	3,001 – 120,000 VA	35 %
	Above 120,000 VA	25 %
General lighting (warehouse)	First 12,500 VA / remainder	100 % / 50 %	Table 220.45
General lighting (hospitals)	First 50,000 / remainder	40 % / 20 %	Table 220.45
Receptacles (non-dwelling)	First 10 kVA	100 %	220.44
Receptacles (non-dwelling)	Remainder	50 %	220.44
Cooking equipment (commercial)	1–6 units / 6+ units	100 % / 65 % / down to 50%	Table 220.56
Range/oven (dwelling)	1 unit ≤ 12 kW	8 kW (per Table 220.55 — Column C)	Table 220.55
Dryers (dwelling)	1–4 / 5+	100 % / dropping per table	Table 220.54
Motor feeder (mixed motor)	Largest motor	125 %	430.24
Motor feeder (mixed motor)	All other motors	100 %	430.24
HVAC (largest of)	Heating OR cooling — pick the larger	100 % (largest noncoincident)	220.60

!

220.60 — Noncoincident loads

When two loads will never run at the same time (heating & cooling are the classic example), only count the larger one. Same applies to electric heat vs gas heat backup. Don't double-count loads physically incapable of operating simultaneously.

Continuous vs Non-Continuous — The 3-Hour Rule

NEC 100 defines a continuous load as one whose maximum current is expected to continue for 3 hours or more. This trips a different multiplier than demand factor — the 125% rule for sizing wire and breakers.

_max current CONTINUOUS — runs at I_max ≥ 3 hr 3-hr threshold NON-CONTINUOUS — peaks < 3 hr 3 hours determines the 125% multiplier

If max current persists ≥ 3 hours → continuous → 125% sizing on conductor + OCPD

For continuous loads — NEC 210.19(A) & 210.20(A)

I_wire ≥ 1.25 × I_load

I_OCPD ≥ 1.25 × I_load

125% multiplier applied to both the conductor ampacity and the overcurrent device. NOT the same as demand factor — these are separate calculations that stack.

Mixed continuous + non-continuous

I_OCPD ≥ 1.25 × I_cont + 1.0 × I_non

Add the two pieces with their respective multipliers.

i

Why 125%? — The thermal answer

Conductors and standard thermal-magnetic breakers are rated to carry 100% of nameplate continuously only at 80% of their maximum capability — for thermal margin. If you want to actually run 100% continuous load, you must derate the breaker/wire to 80%, which is the same as up-sizing by 1/0.80 = 125%. It's the same math, expressed two ways.

What Counts as Continuous?

Continuous loads (apply 125%)

Office & commercial lighting (open all day)
Retail floor lighting
Outdoor / street lighting
Server / IT loads (always on)
Refrigeration compressors (continuous duty)
Process equipment running production shifts
Battery chargers (sustained float)
Heaters running through cold weather
EV charging (often hours of continuous draw)

Non-continuous (no 125% multiplier)

Receptacles (cycle on/off through the day)
Welders (intermittent, duty-cycle sized)
Elevators (cyclic motor loads)
Cooking ranges (peaks < 3 hr)
Most HVAC (cycling on thermostat)
Snow melt, water heaters (typically cyclic)
Garage door openers, lift gates
Equipment with manufacturer-specified duty cycle

Motor Loads — Why They Have Their Own Rules

Motors get their own NEC article (430) because they violate two assumptions: (1) their starting current is 6–8× FLA for a few seconds, which would trip a normal-sized breaker; (2) they are typically continuous duty in industrial settings. The 125% multiplier appears, but for a different reason.

Motor calc	Single motor	Multiple motor feeder	NEC reference
Conductor ampacity (MCA)	1.25 × FLC of motor	1.25 × largest motor FLC + 1.00 × all other motor FLCs + other loads	430.22 / 430.24
Branch-circuit OCPD (MOCP)	Per Table 430.52 (e.g., inverse-time CB ≤ 250% × FLC)	Largest motor's MOCP + sum of other motor FLCs + other loads	430.52 / 430.62
Overload protection	Separate device (in starter/MCC) at 115–125% of FLA, NOT in branch breaker	Each motor has its own overload	430.32
FLA source	NEC Table 430.247–250 (FLC), NOT nameplate	Same — table values	430.6(A)

!

Motor branch circuits — three protections in one

A motor branch has three devices: (1) short-circuit / ground-fault protection (the breaker — set high, up to 250% FLC), (2) overload protection (in starter, set at 115–125% FLA), (3) disconnecting means. The breaker is NOT the overload — those are separate devices. Confusing them is the most common motor-circuit error.

Worked Example 1 — Atlas DC1 Load Study (Side A)

Build the load study for one side of Atlas DC1. This sizes the 480V SWGR-A bus, the TX-A transformer, GEN-A, and the feeder from the utility to the building.

Example 01 · Atlas DC1 spine Build the load study for Side A · feeds half the IT + half the mech

Side A loads (per the MEL)

Load	Connected (kW)	Cont.?	Demand factor	Demand kW	Multiplier	Sized kW
UPS-A1+A2 input (1.25 MW IT, 96% UPS η, 0.95 PF)	1,302 kW	Yes	1.0	1,302 kW	1.25 (continuous)	1,628 kW
CH-1 + CH-2 chillers (450 HP × 2)	674	Yes	1.0 (peak)	674	1.25 largest = 1.0625 ave	716
CWP-1, CWP-2 (75 HP × 2)	112	Yes	1.0	112	1.0 (other motors)	112
CRAH fans (50 HP × 4)	149	Yes	0.95 (modulating)	142	1.0	142
Lighting (mech + IT halls)	22	Yes	1.0	22	1.25	28
Receptacles & misc	15	No	0.5 (NEC 220.44 above 10kVA)	9	1.0	9
TOTAL — Side A	2,318 kW	—	—	2,292 kW	—	2,708 kW (sized) → 2,851 kVA at PF 0.95

Step-by-step

Convert IT load to electrical (input) kW. 1.25 MW IT (mechanical/computational output equivalent) needs to account for UPS efficiency (~96%) and PSU efficiency (~94%) — net ~10% loss between utility and useful load.

UPS input kW = 1250 / 0.96 / 0.94 ≈ 1,386 kW. But for sizing the upstream we use the UPS rating itself: 2 × 1250 kVA × 0.95 PF = 2,375 kVA × 0.95 = ~2,256 kW. Pick the larger — UPS rating governs.
Apply 125% to continuous loads. Per NEC 210.20 / 215.3, the OCPD must be sized to 125% of continuous load.

UPS feeders: 1,316 × 1.25 = 1,645 kW for OCPD sizing. Lighting: 22 × 1.25 = 28 kW.
Apply NEC 430.24 to motor feeder portion. Largest motor (CH-1 at 337 kW) gets 125%, all others at 100%.

Motor feeder: (337 × 1.25) + 337 + 56 + 56 + 142 = 421 + 337 + 56 + 56 + 142 = 1,012 kW for motor portion sizing.
Sum to size the 480V SWGR — A bus + TX-A.

Total demand at 480V = ~2,652 kW. Divide by PF (0.95 average) = 2,791 kVA. TX-A spec'd at 2,500 kVA — close to the line. Real designs would either oversize TX-A to 3,000 kVA or accept slight overload at full load (only happens during commissioning + 100% IT loading + max mech).
Cross-check: 480V FLA at the SWGR.

FLA = (2,791 × 1000) / (√3 × 480) = 3,357 A — fits within the 4,000 A bus. ✓
Generator sizing. GEN-A must carry 100% of Side A demand on utility loss. 2,791 kVA × ~1.10 starting margin (motor inrush) → ~3,070 kVA.

GEN-A is spec'd at 2,500 kW × ~0.85 PF = 2,941 kVA — borderline. Real design would step up to 3,000 kW (3,750 kVA) to provide motor-starting margin. Atlas DC1 currently uses load-shedding logic to drop non-critical loads on genset operation.

!

kW vs kVA — keep them straight

All "Sized (kW)" values above are real power AFTER continuous + motor multipliers. Convert to kVA for transformer/genset sizing by dividing by PF. Conductors and transformers are sized for kVA (apparent), not kW (real). See Atlas DC1 Canonical Specs for the reconciled load study with TX-A sizing called out.

⚠

Real-world sanity check

A real Atlas DC1 load study spans 50+ rows with sub-totals per panel, per UPS module, per ATS source, and per phase. The condensed version here shows the methodology. The two questions every load study answers: "will the equipment fit thermally?" and "will the genset start the load?"

Worked Example 2 — 50-Unit Apartment Building (NEC 220 Standard)

Apartment building service sizing is the textbook NEC 220 problem. Same principles, different demand factors, simpler geometry. Watch how the stacked diversity reduces the connected load to a fraction of itself.

Example 02 · Alternate scale 50 dwelling units · 1,200 sq ft each · electric range & dryer in each · 208Y/120V 3φ service

Per-unit connected load (NEC 220.42 + 220.55)

Item	Value	Notes
General lighting + receptacles	1,200 ft² × 3 VA/ft² = 3,600 VA	NEC 220.41
2 small-appliance circuits	2 × 1,500 = 3,000 VA	NEC 220.52(A)
Laundry circuit	1,500 VA	NEC 220.52(B)
Range (12 kW nameplate)	8 kVA (Table 220.55, Col C)	NEC 220.55
Dryer (5 kW nameplate)	5 kVA	NEC 220.54
Per-unit connected:	21,100 VA	—

Step-by-step (multi-family demand)

Total connected lighting + small appliance + laundry across 50 units

(3,600 + 3,000 + 1,500) × 50 = 405,000 VA
Apply NEC 220.45 demand factors: 100% of first 3,000 + 35% of next 117,000 + 25% of remainder.

3,000 + (117,000 × 0.35) + (285,000 × 0.25) = 3,000 + 40,950 + 71,250 = 115,200 VA
Range demand for 50 ranges (Table 220.55, Col C, 50+ units):

Per Table 220.55 — for 50 units of 8 kW each: total demand = 90 kW + (0.75 × 50) = 90 + 37.5 = 127.5 kW = 127,500 VA
Dryer demand for 50 dryers (Table 220.54):

First 4 at 100% = 4 × 5 = 20 kW. Next 8 at 85% = 8 × 5 × 0.85 = 34 kW. Next 8 at 75% = 30 kW. Continue per table → roughly 110,000 VA total dryer demand.
Sum total demand load:

115,200 + 127,500 + 110,000 = 352,700 VA = 352.7 kVA
Convert to amps at 208V 3φ:

I = (352,700) / (√3 × 208) = 352,700 / 360 = 980 A

Service entrance sized at 1200 A, 208Y/120V 3φ with appropriate margin.
Compare to connected: 50 units × 21,100 VA = 1,055,000 VA = 1,055 kVA connected. Demand is 33%. Two-thirds of the apparent load disappears via NEC 220 demand factors — the diversity of human behavior across 50 households.

i

Why the demand factor stack works

Across 50 units, not every range fires at 6 PM. Not every dryer runs Saturday morning. Not every TV + microwave + lights peak together. NEC's tables encode decades of utility metering data into fixed multipliers. Trust them — but only for the load types they cover. An assisted living facility, hospital, or industrial plant uses different factors.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Continuous load

An office lighting circuit draws 12 A continuously. What's the minimum breaker size?

Drill 2 · Demand factor — receps

A commercial building has 30 kVA of receptacles. What's the demand load?

Drill 3 · NEC 430.24 motor feeder

Motors on one feeder: 50 HP (FLC 65A), 25 HP (FLC 34A), 10 HP (FLC 14A). What's the minimum feeder ampacity?

Drill 4 · Noncoincident loads

A panel has 50 kW of heating + 30 kW of cooling. NEC 220.60 demand?

Drill 5 · Atlas DC1

Atlas DC1 Side A demand was 2,708 kW after sizing multipliers. At PF 0.95 and 480V 3φ, what's the FLA?

If You See THIS, Think THAT

If you see…	Think / use…
"Connected load" in a problem	Sum of all nameplates. NO demand factor. Use for fault analysis, equipment-room ceiling, transformer thermal limit.
"Demand load"	What the system actually carries at peak. Use NEC 220 factors. Use for feeder, service, and transformer sizing.
Load operates ≥ 3 hours at max	Continuous → apply 125% to wire AND breaker (NEC 210.19, 210.20).
"Sum of connected load" + "demand factor"	That demand factor is per NEC 220 Tables. Multiply category-by-category, not in bulk.
Multiple motors on a feeder	NEC 430.24: 125% of largest motor FLC + 100% of all other motor FLCs + other loads. Largest motor only gets the bonus.
Heating AND cooling on the same panel	NEC 220.60: only count the larger one (noncoincident). They can't run together.
"Diversity factor" mentioned	Greater than 1 — applied to peaks across multiple feeders. Don't confuse with demand factor.
Receptacles in commercial	NEC 220.44: 100% of first 10 kVA, 50% of remainder.
50+ dryers in apartment building	Table 220.54 — demand factor drops below 50%; very significant savings.
"Service factor" of 1.15 on motor	Allows brief overload up to 115% — affects overload protection setting (430.32), NOT branch circuit sizing.
"Largest motor" called out in feeder problem	NEC 430.24 applies. Tag it; the 125% bonus belongs to it.
"Load factor" mentioned	Energy-efficiency / utility metric. NOT a sizing factor. Don't apply to feeder calc.

§05 / 39

Branch Circuit Design

MCA · MOCP · NEC 430 · wire & breaker sizing

Every branch circuit answers two questions: how big is the wire, and how big is the breaker. MCA tells you the first. MOCP tells you the maximum for the second. The cutsheet usually tells you both — when it doesn't, you calculate them from FLA.

MCA vs MOCP — The Two Numbers That Govern Everything

Every motor cutsheet, every package HVAC unit, every commercial appliance lists these two numbers. They look similar — they're not. One sizes the wire. One caps the breaker. Mixing them up causes nuisance trips or undersized conductors, both bad outcomes.

A motor branch has THREE protections: short-circuit (breaker), overload (in starter/VFD), and disconnect.

Side-by-Side Definition

MCA · Minimum Circuit Ampacity

What it sizes: the conductor (wire).

For motors: MCA = 1.25 × FLC (NEC 430.22)

For continuous load: MCA = 1.25 × I_load (NEC 210.19)

Why 1.25×: conductor must carry continuous current without exceeding 75°C / 90°C insulation rating

Rule: wire ampacity ≥ MCA

MOCP · Maximum Overcurrent Protection

What it caps: the breaker / fuse rating.

For motors: per NEC Table 430.52 — typically up to 250% × FLC for inverse-time CB

Why so large: motor inrush is 6–8× FLC for ~1 sec; breaker must let it through

Where overload protection lives: separate device, in the starter/MCC/VFD

Rule: breaker ≤ MOCP, rounded up to next standard size

NEC Table 430.52 — Motor Branch-Circuit Protection

This table publishes the maximum percentage of motor FLC for the branch-circuit OCPD by device type. Memorize the four rows. They are tested.

Protective device	Single-phase & 3φ AC squirrel-cage / Δ-connected synchronous	Wound-rotor	DC (constant V)	Notes
Non-time-delay fuse	300%	150%	150%	Fast acting — only used for non-motor work usually
Dual-element (time-delay) fuse	175%	150%	150%	Most common motor fuse — handles inrush gracefully
Instantaneous-trip CB	800%	800%	250%	"Magnetic-only" CB. Used in MCCs with separate overload.
Inverse-time CB	250%	150%	150%	Standard thermal-magnetic CB. Most common in panelboards.

!

NEC 430.52(C)(1) Exception 1 — round up

If the calculated max % doesn't land on a standard breaker size (NEC 240.6: 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 125, 150, 175, 200, 225, 250, 300, 350, 400, 450, 500, 600, 700, 800, 1000, 1200, 1600, 2000, 2500, 3000, 4000, 5000, 6000), you may round UP to the next standard size. So a calculated 480 × 250% = 1200 A → use 1200 A breaker (already a standard size).

The Branch Circuit Design Sequence

Five steps. Always in this order. Most cutsheets give you the answer to steps 2 & 3 — but knowing how to derive them lets you size circuits when the cutsheet is missing or wrong.

For non-motor continuous loads, replace step 1 with "load current" and skip 430.52 — use the simpler 125% rule.

Standard Sizes — Breakers and Wire

The two reference tables you'll consult on every branch circuit.

Standard breaker sizes (NEC 240.6)

Range	Standard amps
15 – 60 A	15, 20, 25, 30, 35, 40, 45, 50, 60
70 – 200 A	70, 80, 90, 100, 110, 125, 150, 175, 200
225 – 600 A	225, 250, 300, 350, 400, 450, 500, 600
700 – 2500 A	700, 800, 1000, 1200, 1600, 2000, 2500
3000 – 6000 A	3000, 4000, 5000, 6000

NEC 310.16 — copper THWN-2 (75°C col)

Wire	Ampacity	Common use
#14 AWG	15 A	Lighting branches (rare)
#12 AWG	20 A	Standard receptacle
#10 AWG	30 A	Dryer, A/C, water heater
#8 AWG	40 A	Range, mid-size A/C
#6 AWG	55 A	Sub-panel feeders
#4 AWG	70 A	Small motor feeders
#2 AWG	95 A	—
1/0 AWG	125 A	—
3/0 AWG	175 A	—
250 kcmil	215 A	—
500 kcmil	320 A	—
750 kcmil	400 A	—

For 90°C insulation column or aluminum, see full NEC 310.16. Always verify temperature rating of equipment terminals (typically 75°C for ≥ 100A).

Conductor Derating — When 75°C Isn't 75°C Anymore

NEC 310.16 ampacity assumes ideal conditions: 30°C ambient, ≤ 3 current-carrying conductors in raceway. Real life isn't ideal. Two correction factors stack.

NEC 310.15(B)(1) — ambient temperature

Ambient °C	Factor (75°C)	Factor (90°C)
21–25	1.05	1.04
26–30	1.00	1.00
31–35	0.94	0.96
36–40	0.88	0.91
41–45	0.82	0.87
46–50	0.75	0.82
51–55	0.67	0.76

NEC 310.15(C)(1) — adjustment for #conductors

Current-carrying cond.	Adjust factor
1–3	1.00
4–6	0.80
7–9	0.70
10–20	0.50
21–30	0.45
31–40	0.40
> 40	0.35

Neutrals usually don't count as current-carrying — except in 3φ-4W systems carrying nonlinear/harmonic loads (then they do).

Final allowable ampacity (the derating stack)

I_allow = I_{NEC 310.16} × C_temp × C_fill

Take the wire's tabulated ampacity, multiply by both correction factors. Result must still be ≥ MCA. Otherwise you upsize the wire and recheck.

Worked Example 1 — Atlas DC1 Chiller Branch Circuit

The CH-1 chiller from §02. Now we size its actual branch circuit end-to-end.

Example 01 · Atlas DC1 spine CH-1 · 450 HP @ 480V 3φ · VFD-driven · 75 ft from 480V SWGR-A

From the cutsheet / NEC

Motor HP

450 HP

FLC

480 A (NEC 430.250 for 450 HP @ 460V)

Voltage

480V 3φ

Starting

VFD (no inrush — VFD ramps current)

Run length

75 ft conductors in EMT, ambient 30°C

Conductors

3φ + EGC (4 wires, but only 3 current-carrying)

Step-by-step

Step 1 — Get FLC. Per NEC 430.6(A)(1), use Table 430.250 not nameplate. For 450 HP at 460V, FLC = 480 A.

FLC = 480 A
Step 2 — MCA per NEC 430.22. Single motor → 125% of FLC.

MCA = 1.25 × 480 = 600 A
Step 3 — MOCP per NEC 430.52. Inverse-time CB on 3φ AC squirrel-cage = 250%.

MOCP = 2.50 × 480 = 1,200 A (already a standard size — no rounding needed)
Step 4 — Pick breaker. Standard sizes: 800, 1000, 1200. Pick the largest ≤ MOCP that gives reasonable coordination.

Use 1,200 A inverse-time CB. Note: VFD doesn't need 250% — VFD soft-starts the motor. A smaller 800 A breaker would also pass NEC and provide tighter protection. Designer's call. Industry practice with VFDs: size at 175–200% × FLC for tighter protection.
Step 5 — Pick wire. Wire ampacity ≥ MCA = 600 A.

From NEC 310.16 (75°C copper THWN-2): one set of 750 kcmil = 400 A. Need parallel runs. Two sets of 350 kcmil per phase = 2 × 310 = 620 A ≥ 600 ✓
OR: two sets of 250 kcmil = 2 × 255 = 510 A — fails ✗
Choose: 2 sets of 350 kcmil per phase, parallel in 2 conduits
Apply derating. Each conduit holds 3 current-carrying conductors (no neutral). Ambient = 30°C → no temp derating. → Each set × 1.00 × 1.00 = 310 A. Two sets parallel = 620 A. Still ≥ 600. ✓

Final design summary

Item	Spec
Branch breaker	1200 A inverse-time CB (or 800 A for tighter VFD coordination)
Phase conductors	2 sets of 350 kcmil THWN-2 copper, in 2 separate 4" EMT
Equipment ground	1/0 AWG copper per NEC 250.122 (sized to OCPD)
Disconnect	1200 A fused or non-fused, within sight of motor (NEC 430.102)
Overload	In VFD (set at 115% of FLA, with motor temp sensor input)

i

Why parallel runs?

For currents above ~400 A, single conductors get unwieldy (#3/0 = 200A, 750 kcmil = 400A is the practical max). Paralleling per NEC 310.10(H) — equal length, same conductor type, terminated identically — is standard for >400 A. The sets must be in separate raceways or grouped per phase to avoid magnetic imbalance.

Worked Example 2 — Office Lighting Branch (Non-Motor, Continuous)

Not every branch is a motor. Most aren't. The non-motor continuous-load case uses a simpler logic: the 125% rule applies once.

Example 02 · Alternate context Office lighting branch · 277V 1φ from 480Y/277V panel · 22 LED fixtures × 60W

Given

Voltage

277V 1φ

Load

22 fixtures × 60W = 1,320 W

Continuous?

Yes (office lighting, runs 8+ hours)

Distance

100 ft to first fixture

Step-by-step

Convert to amps.

I = 1,320 W / 277V = 4.77 A
Apply 125% (continuous). NEC 210.19(A): conductor must carry ≥ 125% of continuous load.

MCA = 4.77 × 1.25 = 5.97 A
OCPD per NEC 210.20(A).

OCPD ≥ 1.25 × 4.77 = 5.97 A → next standard size = 15 A breaker. (Could also use 20 A.)
Pick wire. #14 AWG = 15 A — meets MCA. But for 100 ft at 4.77 A, voltage drop:

VD = (2 × 100 × 4.77 × 3.07Ω/kft) / 1000 = 2.93V = 1.06% — well within NEC 3% recommended. #14 AWG is fine.
Final design.

15 A 1-pole CB · #14 AWG copper THWN-2 · #14 EGC · 1/2" EMT

i

Notice what's different from the motor case

One multiplier, not two. The 125% appears in step 2 AND step 3 — but it's the same 125% applied to both, NOT 1.25 for the wire and 2.50 for the breaker. Non-motor continuous loads don't have starting inrush, so OCPD doesn't need the 250% headroom from NEC 430.52. This single-multiplier logic governs lighting, heaters, EV chargers, server farms — anything continuous but not a motor.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · MCA from FLC

A 30 HP, 480V 3φ motor (FLC = 40 A from NEC 430.250). What's the MCA?

Drill 2 · MOCP — inverse-time CB

Same 30 HP motor. What's the maximum branch breaker (inverse-time CB)?

Drill 3 · Continuous non-motor

An office lighting branch at 16 A continuous. Wire + breaker?

Drill 4 · Derating — many conductors

9 current-carrying #10 AWG conductors in one EMT, ambient 30°C. Derated ampacity (75°C)?

Drill 5 · Round up — round down?

Calculated MOCP = 287 A. Standard CB sizes: 250, 300, 350, 400. Pick:

i

Data center branches — NEC 645 may apply

For Information Technology Equipment (ITE) rooms qualifying under NEC Article 645, branch circuits to IT equipment have relaxed wiring method rules — under-floor cabling is permitted without conduit, and certain cable types (CL2, CMP, CMR) may be used. Trade-off: the entire room must comply with 645.4 (single emergency disconnect for ALL ITE + HVAC, smoke/heat detection, restricted access). Most large data halls forgo NEC 645 to use standard wiring methods; smaller IT closets often invoke 645 for cabling flexibility. See §32 Codes Reference for full Article 645 detail.

If You See THIS, Think THAT

If you see…	Think / use…
"1-pole" / "2-pole" / "3-pole" breaker — which to pick?	Mnemonic: "One pole, one hot. Two poles, no neutral. Three poles, three phases." 1P = line-to-neutral (120V/277V) — lighting, receps. 2P = line-to-line (240V/480V) — range, dryer, A/C, 1φ heater. 3P = three-phase load — motor, MCC, sub-feeder.
"MCA" on a cutsheet	Wire ampacity floor. Conductor must be rated ≥ MCA after derating.
"MOCP" on a cutsheet	Breaker ceiling. Round UP to next standard size if MOCP is between standard sizes.
Both MCA and MOCP listed	Use them. They override your calculation. Manufacturer tested the equipment.
Only HP given (motor)	Look up FLC in NEC Table 430.250 (3φ) or 430.248 (1φ). Calculate MCA = 1.25 × FLC, MOCP = up to 250% × FLC for inv-time CB.
Only kW given (non-motor)	Calculate I = kW × 1000 / (V × PF for 1φ, or √3 × V × PF for 3φ). If continuous: wire AND breaker = 1.25 × I.
VFD-driven motor	VFD soft-starts → no need for full 250% MOCP. Industry practice: 175–200% × FLC. Add input/output reactors per harmonic concerns (§15).
Multiple motors on a single branch	NOT typical — usually one branch per motor. If multiple are required, NEC 430.53 has special rules (group motor protection).
Conductor in 50°C ambient	Apply temp correction: 0.75 × tabulated 75°C ampacity. Recheck MCA.
9 conductors in one conduit	Apply 0.70 fill factor (NEC 310.15(C)(1)). Recheck MCA.
Long branch run (>100 ft at high I)	Voltage drop check. NEC informational note: ≤ 3% on branch, ≤ 5% total. May need to upsize beyond MCA.
Range / cooktop / dryer (residential)	NEC 220.55 / 220.54 demand factors apply. Use breaker per nameplate, wire per Table 220 demand.
"Group fuse" / "group motor"	NEC 430.53. Specialized — multiple motors share one fuse. Limited applicability.

§06 / 39

Panel Schedules

Anatomy · phase balancing · bus sizing · main breaker

A panel schedule is one document that captures every branch circuit on a panel: which breaker, which wire, which load, which phase. It is the deliverable that contractors use to actually install the work.

Panelboard vs Switchboard vs Switchgear vs MCC

Before reading a panel schedule, know which kind of "panel" you're looking at. The word panel covers four very different pieces of equipment.

Equipment	Typical use	Voltage	Bus rating	Construction	Atlas DC1
Panelboard	Branch-circuit distribution from a feeder	≤ 600V	≤ 1200 A	Wall- or floor-mounted; molded-case breakers; NEMA PB1 / UL 67	Office lighting, RPPs in IT halls
Switchboard	Service entrance, feeder distribution	≤ 600V	800 – 5000 A typical	Free-standing; molded-case or insulated-case CBs; NEMA PB2 / UL 891	—
Switchgear	Main distribution at service or sub-station	LV ≤ 600V or MV 1–38 kV	800 – 6000 A LV; up to 4000 A MV	Free-standing; drawout breakers; protective relays; UL 1558 (LV) or IEEE C37 (MV)	480V SWGR-A & B (4000A) · 12.47kV MV SWGR
MCC (Motor Control Center)	Motor starters & VFDs grouped together	≤ 600V (LV) or 5kV (MV)	800 – 5000 A bus	Free-standing; modular "buckets" — combination starter, VFD, soft-starter; NEMA ICS18 / UL 845	Mech room MCC for chillers/pumps (often bolted to SWGR)
PDU (data-center context)	480→415Y/240V step-down + sub-distribution to racks	480V in / 415Y/240V out	225 – 1000 kVA typical	Cabinet with isolation transformer + integral panelboard	PDU-A1 (500 kVA), PDU-B1, etc.
RPP (Remote Power Panel)	Row-level distribution from a PDU to racks	415Y/240V or 208Y/120V	225 – 400 A	Slim panelboard at row end; sub-metered branches	One per IT row, fed from PDU

Anatomy of a Panel Schedule

The panel schedule is a tabular document. Below is a real-format Eaton/Square-D-style schedule for a 42-circuit panelboard. Each row is one breaker; the table layout encodes the phase rotation.

PANEL: RPP-A1-1 · 415Y/240V 3φ-4W · 400A MCB · 42 circuits · NEMA 1 · Cu bus · IT Hall A · Fed from PDU-A1
Ckt#	Description	Wire	Trip	P	A (W)	B (W)	C (W)	P	Trip	Wire
1	Rack A1-01 · servers	#10	30A	1	5760	—	—	1	30A	#10
2	Rack A1-02 · servers	#10	30A	1	—	5760	—	1	30A	#10
3	Rack A1-03 · servers	#10	30A	1	—	—	5760	1	30A	#10
4	Rack A1-04 · servers	#10	30A	1	5760	—	—	1	30A	#10
5	Rack A1-05 · servers	#10	30A	1	—	5760	—	1	30A	#10
6	Rack A1-06 · servers	#10	30A	1	—	—	5760	1	30A	#10
7	Rack A1-07 · GPU node	#6	60A	1	11520	—	—	1	60A	#6
8	Rack A1-08 · GPU node	#6	60A	1	—	11520	—	1	60A	#6
9	Rack A1-09 · GPU node	#6	60A	1	—	—	11520	1	60A	#6
10	PDU controls	#12	20A	1	800	—	—	1	20A	#12
11	PDU monitoring	#12	20A	1	—	800	—	1	20A	#12
12	Hot-aisle lighting	#12	20A	1	—	—	800	1	20A	#12
… (circuits 13–42 follow same pattern)		—	—	—	—	—	—	—	—	—
PHASE TOTALS (W) →					23,840	23,840	23,840	Σ = 71,520 W = 71.5 kW = 75.3 kVA @ 0.95 PF
PHASE CURRENT (A) →					99.3	99.3	99.3	Bus loaded to 25% of 400A — well within 80% target

What every column tells you

Column	What it captures	Why it matters
Ckt #	Position in panel (odd numbers left, even right)	Phase rotation: 1=A, 2=A, 3=B, 4=B, 5=C, 6=C — repeats. Enforces balance by geometry.
Description	What the breaker feeds	Field labeling, troubleshooting, future modifications
Wire	Conductor size + type	Branch circuit conductor — sized per MCA
Trip	Breaker amp rating	OCPD — sized per MOCP, rounded to standard
P (poles)	1, 2, or 3 pole	1P = 277V or 120V, 2P = 240V or 480V, 3P = 3φ load
A / B / C (W or VA)	Watts (or VA) attributed to that phase	Used to calculate phase total + balance check
Phase total	Sum of all loads on each phase	Balance check: phases should be within ~5–10% of each other
Phase current	VA/V calculation per phase	Ensures no phase exceeds bus rating

Phase Balancing — Why Circuit Numbers Are Geometry

The odd/even circuit numbering on a panel isn't decorative. The bus bars physically alternate A-A-B-B-C-C top-to-bottom. As long as you fill circuits sequentially with similar loads, the panel auto-balances itself.

Three-phase panels rotate A-A-B-B-C-C-A-A-B-B-C-C ... by circuit pair, top to bottom

Balance rules

Spread similar loads across phases. If you have 30 identical 30A breakers, fill them sequentially. The geometry handles the balance.
2-pole breakers connect to consecutive bus stabs (positions 1+3 = A+B, 3+5 = B+C). They span two phases.
3-pole breakers connect to all three (positions 1+3+5). Ideal for 3φ loads — inherently balanced.
Target imbalance: ≤ 10% between heaviest and lightest phase. Below 5% is excellent.

Imbalance percentage

imb % = (I_max − I_avg) / I_avg × 100

NEMA limit for motor operation = 1% (motors derate quickly). Panel balance target = 10%.

Bus & Main Breaker Sizing

The panel's bus must carry the worst-case phase current. The main breaker (MCB) protects the bus. Or, with a Main Lug Only (MLO) panel, the upstream OCPD protects.

Decision	Rule	NEC reference
Bus rating ≥	1.0 × (heaviest phase current after demand factor)	NEC 408.30
Bus rating ≥	1.25 × continuous load on the heaviest phase	NEC 215.3 (applies to feeder OCPD)
Main breaker (MCB) ≤	Bus rating	NEC 408.36
Main breaker (MCB) ≥	Same logic as feeder OCPD: 1.25 × cont + 1.0 × non-cont	NEC 215.3
MLO panel	Upstream feeder breaker provides the bus protection. Bus must equal or exceed feeder breaker rating.	NEC 408.36(A) Exception
Number of branch breakers ≤	42 per panelboard (lighting & appliance)	NEC 408.54 (deleted in 2008+ but many AHJs still enforce; otherwise UL 67 governs)

i

Standard panelboard bus sizes

100 · 125 · 150 · 200 · 225 · 250 · 400 · 600 · 800 · 1000 · 1200 A. For data center RPPs (415Y/240V), 400A and 600A are standard. For office panelboards (208Y/120V), 100–225A is typical. PDU sub-panels: 400–800A.

Worked Example 1 — Atlas DC1 RPP-A1-1 Panel Schedule

Continue the panel schedule shown earlier. We need to verify bus and main breaker are correctly sized.

Example 01 · Atlas DC1 spine RPP-A1-1 · 415Y/240V 3φ-4W · serves 8 racks + supporting loads

Panel summary (from schedule)

Voltage

415Y/240V 3φ-4W (downstream of PDU-A1)

Connected loads

71.5 kW total (continuous — server load)

Phase A total

23.84 kW = 23,840 / 240 = 99.3 A

Phase B total

99.3 A

Phase C total

99.3 A (perfectly balanced — sequential rack loading)

Bus + main breaker sizing

Apply 125% to continuous (server) load.

99.3 × 1.25 = 124 A per phase required
Bus selection. Standard sizes ≥ 124 A: 125, 150, 200, 225, 400.

Choose 400 A bus to allow future capacity (you don't want to swap the panel when racks expand). Loaded to 31% currently — leaves ~270 A spare.
Main breaker. Sized to protect bus AND ≥ 1.25 × continuous load. Standard breakers ≤ 400: 400, 350, 300, 250, 225, 200, 175, 150, 125.

400 A MCB matches the bus. Could also use 200 A MCB and call it derated for current load, but a 400A MCB allows the panel to grow.
Feeder from PDU-A1 to RPP-A1-1. Sized to 125% × 99.3 = 124 A minimum.

Feeder breaker at PDU-A1 = 400 A (matches downstream MCB). Wire = 500 kcmil Cu THWN-2 (320 A × 1.25 derating headroom = 400 A capable). Or 1/0 Al MC cable.

!

Why oversize the panel for a data center?

Server racks grow over time. A "24 kW rack" today becomes a "44 kW rack" in 18 months when GPU nodes get upgraded. Sizing the bus at 4× current load is industry standard — replacing an RPP requires shutting down the row, and DC operators avoid that at all costs. Oversize at install; never resize after.

Worked Example 2 — Apartment Unit Panel

A single 200A residential panel for a typical 2-bedroom unit. The panel schedule for residential is simpler — fewer phases (just split-phase A-B), but more circuit types.

Example 02 · Alternate scale 200 A · 120/240V 1φ-3W · 30-circuit · MCB

PANEL: UNIT-101 · 120/240V 1φ-3W · 200A MCB · 30 circuits · Square D Homeline
Ckt	Desc	Wire	Trip	P	A (W)	B (W)	P	Trip
1, 3	Range — 50A 240V	#6	50	2	8000	—	—	—
5, 7	Dryer — 30A 240V	#10	30	2	—	5000	—	—
9, 11	Water heater — 30A 240V	#10	30	2	4500	—	—	—
13, 15	A/C condenser — 40A 240V	#8	40	2	—	3500	—	—
17	Kitchen receps #1 — 20A	#12	20	1	1500	—	1	20
19	Kitchen receps #2 — 20A	#12	20	1	—	1500	1	20
21	Bath receps GFCI	#12	20	1	800	—	1	20
23	Bedroom receps AFCI	#14	15	1	—	600	1	20
25	Living receps AFCI	#14	15	1	600	—	1	20
27	Lighting	#14	15	1	—	800	1	15
29	Laundry recep	#12	20	1	1500	—	1	20
PHASE TOTALS (W) →					16,900	11,400	Σ = 28,300 W
CONNECTED PHASE I (A) →					141	95	After NEC 220 demand: ~ 118 A actual peak
IMBALANCE →					39%		Re-balance: move dryer + WH to opposite phases

Issues caught by the schedule

Heavy imbalance. Phase A = 16.9 kW, Phase B = 11.4 kW. 39% imbalance — well above 10% target.

Fix: swap range and dryer phase positions. Range is single-phase 240V (uses both A and B equally — already balanced). Move water heater (4500W) from A to B; result: A = 12.4 kW, B = 15.9 kW. Imbalance now ~ 14%. Better. (Perfect balance impossible with discrete loads of different sizes.)
200A MCB sizing check. Heaviest phase = 141A connected. With NEC 220 demand factors applied, peak ~ 118A. Within 200A bus. ✓

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Bus sizing

Panel demand on the heaviest phase = 156 A continuous. Minimum bus rating?

Drill 2 · Phase rotation

On a 3-phase panelboard, what phase does circuit 7 connect to?

Drill 3 · Imbalance check

Phase A = 80 A, B = 65 A, C = 75 A. Avg = 73.3. % imbalance?

Drill 4 · Panel vs switchgear

A 3,000 A free-standing distribution panel with drawout breakers — is it a panelboard, switchboard, or switchgear?

Drill 5 · Atlas RPP

Atlas DC1 RPP-A1-1 had 99.3 A on each phase (continuous). Required main breaker minimum?

If You See THIS, Think THAT

If you see…	Think / use…
How a breaker's poles map to the panel bus	"One pole, one hot. Two poles, no neutral. Three poles, three phases." 1P touches one phase (A, B, or C). 2P spans two adjacent positions (A+B, B+C, or C+A). 3P spans all three. 3P loads are inherently balanced; 1P loads must be distributed across phases for balance (see Phase Rotation section above).
"Panel schedule" requested	Tabular doc with one row per breaker. Captures wire, breaker, phase, load. Used by contractor to install.
"42-circuit panelboard"	21 left + 21 right. Lighting & appliance branch panels traditionally limited to 42; modern UL 67 allows more.
"MCB" vs "MLO"	Main Circuit Breaker (panel has its own main); vs Main Lug Only (no main, fed protected from upstream).
2-pole breaker on 3φ panel	Spans 2 phases (e.g., A-B, B-C, A-C). Used for 240V or 480V single-phase loads (split-phase or 3φ panel).
3-pole breaker	Spans all 3 phases. Used for 3φ motor / pump / panel-feed loads.
"PDU" in DC	Power Distribution Unit — 480→415Y/240V step-down + integrated panelboard. Not a power strip.
"RPP" in DC	Remote Power Panel — branch panel at the row/aisle, fed from PDU.
"MCC" — Motor Control Center	Free-standing modular cabinet with starter/VFD/disc combo "buckets" per motor.
Phases imbalanced > 10%	Re-arrange branch positions to redistribute load. The panel layout itself is the lever.
Bus loaded > 80%	Either upsize to next standard bus, OR re-shed loads to another panel. Don't run panels at 100%.
"Series-rated" breakers	Downstream CB has lower interrupting rating than upstream — only valid if combination is UL-listed for series rating. Many AHJs prohibit; verify.

§07 / 39

Feeder Design

NEC 430.24 · voltage drop · raceway · derating

A feeder carries power between two pieces of distribution equipment — service to switchgear, switchgear to panel, panel to PDU. Sized like a branch circuit but with one big difference: the largest motor on the feeder gets a 25% bonus.

Branch vs Feeder vs Service — NEC Definitions

Three terms, three different sets of rules. NEC Article 100 defines them precisely.

Conductor type	Definition (NEC 100)	NEC article	OCPD multiplier	Atlas DC1 examples
Branch circuit	Conductors between the final OCPD and the outlet/equipment	210, 430 (motors)	125% × cont (NEC 210.20). Motors: per Table 430.52	RPP breaker → rack PDU; chiller branch CB → motor
Feeder	Conductors between service equipment / source and the final branch-circuit OCPD	215, 430.24 (motor feeders)	125% × cont (NEC 215.3). Motors: 125% largest + 100% rest (430.24)	SWGR-A → UPS-A1 input; PDU-A1 → RPP-A1-1; SWGR → MCC
Service conductors	Conductors from utility supply to service equipment	230	Calculated per Article 220 demand	Utility 12.47kV → MV switchgear primary
Tap conductor	Smaller conductor tapped from a larger feeder, with restrictive rules on length and termination	240.21(B)	Special — 10ft / 25ft / 100ft tap rules	Disconnect taps in switchgear sections

NEC 430.24 — The Motor Feeder Rule

This is the most-tested motor sizing rule in the PE exam, and one of the most-used in real practice. Whenever a feeder serves multiple motors (an MCC, a mech-room sub-panel, a chilled-water plant), the largest motor gets the 125% bonus and all others contribute their FLC at face value.

NEC 430.24 — multi-motor feeder ampacity

I_feeder ≥ 1.25 × FLC_largest + Σ FLC_{other motors} + other loads

"Other loads" = lighting + receptacles + non-motor equipment, each per their respective demand rules.

Only the largest motor's FLC gets multiplied by 1.25. Every other motor contributes its FLC at face value. All currents must be at the same voltage — if motors are at different voltages downstream of a transformer, convert all currents to the feeder's voltage first.

!

Why only the largest motor?

The 125% accommodates the worst case: the largest motor starting last, drawing inrush current, while every other motor is already running at full FLC. Smaller motors starting don't change the picture meaningfully because their inrush is small. NEC 430.24 captures this in a clean formula instead of dynamic load-flow analysis.

Voltage Drop on Feeders

NEC 215.2(A)(1) Informational Note recommends ≤ 3% on feeder, ≤ 5% combined feeder + branch. This is a recommendation, not a requirement, but most AHJs and engineering specs treat it as mandatory. Long feeder runs frequently dictate wire size more than ampacity does.

Single-phase voltage drop (NEC Ch 9 Table 8 — DC method)

VD = (2 × I × R × L) / 1000

L in feet. R = ohms/1000 ft from NEC Ch 9 Table 8 (DC) or Table 9 (AC). Factor of 2 = round trip.

Three-phase voltage drop

VD = (√3 × I × R × L) / 1000

No factor of 2 — neutral current is 0 in balanced 3φ. R from NEC Ch 9 Table 9 if at typical 75°C.

Convert to %VD

%VD = VD / V_nom × 100

Use system voltage (480 for 3φ, 240 or 120 for 1φ).

i

Quick rule of thumb (copper)

For any conductor: VD ≈ I × L / (Cu factor). The Cu factor for typical 75°C copper is about 22 × CM × 1000. So VD% on 480V 3φ ≈ (1.732 × I × L × 12.9) / (CM × 480), where CM = circular mils of conductor.

NEC Ch 9 Table 9 — AC resistance (75°C, copper, in steel conduit)

Conductor	R (Ω/1000 ft)	X (Ω/1000 ft)	Effective Z @ 0.85 PF
#12 AWG	2.0	0.054	1.74
#10 AWG	1.2	0.050	1.05
#8 AWG	0.78	0.052	0.69
#6 AWG	0.49	0.051	0.44
#4 AWG	0.31	0.048	0.29
#2 AWG	0.20	0.045	0.19
1/0 AWG	0.12	0.044	0.13
3/0 AWG	0.079	0.042	0.094
250 kcmil	0.054	0.041	0.073
500 kcmil	0.029	0.039	0.054
750 kcmil	0.021	0.038	0.048

Worked Example 1 — Atlas DC1 Feeder from UPS-A1 to PDU-A1

The feeder between UPS-A1 (480V output) and PDU-A1 (480V input) — 250 ft of cable. Sized for the full UPS output rating, not just current load.

Example 01 · Atlas DC1 spine UPS-A1 (1250 kVA, 480V 3φ) → PDU-A1 (480V input) · 250 ft route through cable tray

Given

UPS rating

1250 kVA at 480V 3φ

UPS full-load current

I = 1,250,000 / (√3 × 480) = 1,504 A

Run length

250 ft (cable tray + conduit)

PDU input

500 kVA = 602 A actual; UPS sized at 1250 kVA = 1,504 A possible

Continuous?

Yes (IT load, 24/7)

Step-by-step

Decide the feeder ampacity basis. Must serve worst-case downstream load. Since UPS could feed multiple PDUs in real designs, size feeder to UPS full-output current (1,504 A), not just one PDU.

Feeder I_basis = 1,504 A (UPS rating). Apply 125% continuous → 1,880 A.
Conductor selection. 1,880 A is well above any single conductor. Need parallel runs.

Try 4 parallel sets of 600 kcmil (each rated 350A at 75°C) = 4 × 350 = 1,400 A — fails.
Try 4 sets of 750 kcmil = 4 × 400 = 1,600 A — fails.
Try 5 sets of 600 kcmil = 5 × 350 = 1,750 A — fails.
Try 5 sets of 750 kcmil = 5 × 400 = 2,000 A. ✓ (Or 6 sets of 500 kcmil = 1,920 A.)
Voltage drop check. R for 750 kcmil = 0.021 Ω/1000 ft (Table 9).

VD = (√3 × 1,504 × 0.021 × 250) / 1000 / 5 (parallel paths divide R) = (1.732 × 1504 × 0.021 × 250) / 5000 = 13,684 / 5000 = 2.74 V.
%VD = 2.74 / 480 × 100 = 0.57% — excellent. Well below 3%.
Feeder OCPD at UPS-A1 output.

125% × 1,504 = 1,880 A. Round up to standard: 2,000 A molded-case CB. Or use a 1,600 A breaker if downstream coordination allows.
Final spec.

2,000 A CB at UPS · 5 sets of 750 kcmil THWN-2 Cu in 5 separate 4" EMT/cable tray runs · 350 kcmil EGC per NEC 250.122 (Table — for 2000A breaker, EGC = 4/0 Cu)

i

Cable tray cuts conductor cost dramatically

For 5 parallel sets of 750 kcmil at 250 ft, the conductor cost alone is > $50,000. Cable tray (NEC 392) lets you use Type TC or MC cable instead of conduit-wired THWN, with substantial savings on installation labor. We explore this in §08.

Worked Example 2 — Apartment Building Service Feeder

The 50-unit apartment building from §03 had a calculated demand of 980 A at 208V 3φ. Now we size the actual service feeder from the utility transformer.

Example 02 · Alternate scale 50 units · 980 A demand at 208Y/120V 3φ-4W · 80 ft underground service from pad-mount

Given

Demand load

980 A at 208Y/120V 3φ-4W (per §03 calc)

Run length

80 ft underground (PVC duct under the building)

Voltage drop allowed

≤ 3% feeder

Step-by-step

Service entrance breaker (or fuses) + bus. Round 980 A up to standard size: 1,200 A.
Conductor sizing. Need 1,200 A in parallel.

3 parallel sets of 500 kcmil (each 320A) = 960 A — fails for 1,200A breaker.
4 sets of 500 kcmil = 1,280 A. ✓
Or 3 sets of 750 kcmil = 1,200 A — exactly meets.
Neutral conductor. 3φ-4W with 120V residential loads = significant unbalanced neutral current. Often sized 100% of phase conductors for residential, even though NEC allows reduction.
Voltage drop check (3 sets of 750 kcmil, R = 0.021 Ω/1000 ft):

VD = (√3 × 980 × 0.021 × 80) / 1000 / 3 = 2,852 / 3000 = 0.95 V. %VD = 0.95 / 208 = 0.46%. ✓
Final spec.

1,200 A main service breaker (or 1,200 A fuse) · 3 sets of 750 kcmil Cu THWN-2 in 4" PVC underground · neutral 750 kcmil (full size) · 4/0 Cu EGC. (Or aluminum at lower cost — see §07.)

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Branch vs feeder

A 200 A breaker feeds a sub-panel which has its own branches. The 200 A circuit is a:

Drill 2 · Voltage drop — 3φ

350 ft of 4/0 Cu (R = 0.062 Ω/kft) carrying 180 A at 480V 3φ. %VD?

Drill 3 · Tap rule — 10 ft

Can you tap a 400 A feeder with #6 AWG (75 A) for 8 feet without sizing the wire to 400 A?

Drill 4 · Nonlinear neutral

A 4-wire feeder serves 100% nonlinear server load. Neutral sizing?

Drill 5 · Atlas UPS feeder

Atlas DC1 UPS-A1 = 1500 A. Continuous. Min feeder ampacity?

If You See THIS, Think THAT

If you see…	Think / use…
"Feeder" between switchgear and panel	NEC Article 215. Sized: 125% × continuous + 100% × non-continuous, with NEC 220 demand factors applied.
"Multiple motors on a feeder"	NEC 430.24: 125% × largest motor FLC + 100% × all other motor FLCs + other loads.
"Mixed motor + non-motor on feeder"	Above formula PLUS lighting + receps with their NEC 220 demand factors. Sum.
Long feeder run (> 100 ft, large I)	Voltage drop check. ≤ 3% target. May need to upsize beyond ampacity-only sizing.
Feeder current > 400 A	Almost always parallel runs. Watch terminations and cable management.
"Tap conductor" (NEC 240.21)	Special rules: 10ft tap (no termination protection), 25ft tap (with restrictions), 100ft tap (industrial only). All have specific size minima.
"Service entrance"	NEC Article 230. Different rules from feeder — service has no upstream OCPD inside the building. Sized for full demand load.
"Heavy unbalanced 3φ-4W"	Neutral can carry phase current or more (with harmonics). Don't undersize neutral.
"Harmonic loads on the feeder" (servers, VFDs, LEDs)	Neutral becomes a current-carrying conductor for derating purposes. NEC 310.15(C). Often size neutral 200% in pure nonlinear feeders.
Underground feeder in PVC	NEC 310.60 different ampacity table for direct burial / duct bank. Soil thermal resistivity matters.

§08 / 39

Conductor Types Decoded

Insulation letter codes · temperature ratings · copper vs aluminum · cable types

The two-to-five letters stamped on a wire's jacket tell you everything: temperature rating, wet vs dry, oil resistance, where it's allowed. Decode the letters once and you'll never specify the wrong wire again.

The Insulation Letter Code

Every conductor type code is built from a small alphabet of letters. Each letter encodes one property. Stack them in order and you've described the wire.

Letter	Meaning	Example in code
T	Thermoplastic insulation (typically PVC)	THHN, THWN
H	Heat-resistant — 75°C rating	THWN
HH	Higher heat resistance — 90°C rating	THHN
W	Wet-location rated	THWN, XHHW
N	Nylon outer jacket (oil + abrasion resistance)	THN, THWN
X	Cross-linked polyethylene (XLPE) insulation	XHHW, XHHW-2
-2	90°C wet AND dry (the "-2" denotes wet-location 90°C, vs. -W which is wet-only 75°C)	THWN-2, XHHW-2
R	Rubber insulation (older types)	RHH, RHW
U	Underground service entrance rated	USE-2, UF
SE	Service entrance cable	SEU (round), SER (round)
MV-	Medium voltage cable (followed by temp rating: 90 or 105)	MV-105 (105°C)

i

How to read THWN-2

T = thermoplastic (PVC) insulation
H+W = heat-resistant 75°C, wet-location rated
N = nylon outer jacket
-2 = upgraded to 90°C in both wet and dry locations

→ THWN-2 = PVC + nylon, 90°C wet/dry, the workhorse of modern commercial & industrial wiring.

The Conductor Types You'll Actually See

There are dozens of NEC-recognized types. In real practice you specify maybe ten of them, ever. These are the ones.

Type	Insulation	Temp dry / wet	Where used	Where NOT to use
THWN-2	PVC + nylon	90°C / 90°C	Most common. Conduit-wired branches and feeders, indoors and out, wet or dry. Default for new construction.	Not for direct burial; not for cable tray (use TC); not for free-air without conduit
THHN	PVC + nylon	90°C / —	Dry locations only. Often replaced by THWN-2 (better rating, similar cost).	Wet, damp, exterior, underground
XHHW-2	XLPE (cross-linked)	90°C / 90°C	Premium feeder/branch wire. Better insulation toughness than PVC. Often used for service entrance and large feeders.	Slightly more expensive than THWN-2; usually no functional difference for indoor work
USE-2	XLPE	90°C / 90°C wet	Direct-burial service entrance. Underground feeders.	Some types not labeled for indoor wiring methods — check label for dual rating
NM-B (Romex)	PVC	90°C / —	Residential interior wiring. Dwellings, multi-family ≤ 3 stories.	Commercial buildings (most jurisdictions); wet locations
MC (metal-clad)	Conductors in aluminum/steel armor	90°C	Commercial & industrial in cable tray, exposed, or raceway-free runs. Replaces conduit-wired systems for labor savings.	Direct burial without specific MC-HL types
AC (BX)	Conductors in flexible armor (no separate ground)	90°C	Older commercial wiring. Largely replaced by MC.	Wet locations; new construction generally prefers MC
TC-ER	Tray cable, exposed-run rated	90°C / 90°C	Cable tray with exposed runs (NEC 392). Industrial and DC distribution.	Direct burial; check NEC 336 for limitations
SO / SOOW (cord)	Rubber	90°C / 90°C	Portable equipment, drop cords, temporary connections, generators	Permanent wiring (NEC 400.8 prohibits cord as substitute for fixed wiring)
MV-105	EPR or XLPE, shielded	105°C / —	Medium voltage cable (5kV, 15kV, 35kV). Substations, MV feeders, utility-side primary.	LV applications (overkill); requires special terminations
SE / SER	Multiple conductors in one cable	90°C / 90°C	Service entrance, residential. Sub-feeders inside dwellings.	Commercial service in most jurisdictions
UF	PVC, direct burial rated	60°C / 60°C	Direct burial, residential outdoor branch circuits	Aerial; conduit (use THWN-2 instead); commercial direct burial (use USE-2)

Temperature Ratings — and the Termination Trap

A conductor with 90°C insulation can carry more current than the same wire size at 75°C. But you can't always use the 90°C ampacity column — because the device the wire terminates on has its own temperature limit.

Termination type	Max temp rating	Use which NEC 310.16 column	Common scenarios
Equipment ≤ 100A circuits	60°C	60°C column	Most residential breakers; small device terminals
Equipment > 100A circuits	75°C	75°C column	Commercial breakers, panelboard mains, motor terminations
Equipment marked 90°C	90°C	90°C column (rare)	Some specialty equipment; check the label, don't assume
NEC 110.14(C) exception	—	You may use 90°C ampacity for the derating calculation, but final allowable can't exceed the termination column	This is how 90°C wire derates more gracefully in conduit fill / high ambient cases

!

Why 90°C wire doesn't help most of the time

If you use THWN-2 (90°C rated) on a 200A breaker (75°C terminations), you must size by the 75°C column. The 90°C rating only helps when you're applying derating factors — the higher 90°C ampacity acts as a buffer before the derating drops you below the 75°C column value. For straight ampacity sizing without derating, the 75°C column governs.

Copper vs Aluminum — When Each Wins

Aluminum is roughly half the cost of copper for the same ampacity but requires larger conductor sizes (lower conductivity), specific terminations, and antioxidant compound. For large feeders, aluminum saves significant money. For branch circuits, copper is universal.

Property	Copper	Aluminum
Conductivity	1.0× (reference)	0.61× — needs larger size for same ampacity
Cost (commodity)	Higher	~50% of copper for equivalent ampacity
Weight	Heavy	~1/3 of copper — easier installation on long runs
Termination	Direct connection acceptable	Requires AL-rated lug, anti-oxidation compound (Penetrox/Noalox), torque per spec
Cold-flow (creep)	Stable	Connections loosen over time if not properly torqued — reason for residential aluminum failures in 1970s
NEC small wire restriction	OK at #14, #12	NEC 310.106(B) — minimum #12 for AL conductors generally; for branch circuits, #6 is the practical floor due to terminations
Common application	Branch circuits, all sizes; sensitive equipment	Service entrances, feeders ≥ #6, MV cable, utility distribution
Atlas DC1 examples	All branches, panelboards, UPS internal	Service entrance from utility (12.47 kV); some 480V feeder runs > 200ft

Aluminum Sizing Comparison (Same Ampacity)

Ampacity (75°C)	Copper size	Aluminum size (one step bigger)	Cost savings (rough)
100 A	#3 AWG	#1 AWG	~30%
200 A	3/0 AWG	4/0 AWG	~35%
400 A	500 kcmil	700 kcmil	~40%
600 A	750 kcmil (or 2×4/0)	2×500 kcmil parallel	~45%
1000 A	2×500 kcmil parallel	2×800 kcmil parallel	~45%

Worked Example 1 — Atlas DC1 Conductor Selection Across the System

One reference facility, six different conductor types — each appropriate for its place in the system.

Example 01 · Atlas DC1 spine Specifying conductor type for each system zone

Atlas DC1 location	Application	Conductor specification	Why this type
Utility 12.47kV → MV switchgear	MV primary feeder	15kV MV-105 EPR shielded, 3/c with concentric neutral, AL conductor	MV requires shielding. Aluminum economical at this size. EPR insulation for thermal toughness.
TX-A → 480V SWGR-A (mech room)	Transformer secondary feeder, 4000A	Multiple parallel sets of 750 kcmil Cu THWN-2 in cable tray	High ampacity, indoor, dry. THWN-2 is the workhorse. Could substitute XHHW-2.
SWGR-A → MCC-MR1 (chillers)	Motor MCC feeder	3 sets of 350 kcmil Cu XHHW-2 in 4" EMT	Tougher insulation handles repeated mechanical stress; better resistance to oil/coolant in mech rooms.
SWGR-A → UPS-A1	UPS feeder, 1500A	5 sets of 750 kcmil Cu THWN-2, separate raceways	Critical load — copper for terminating quality; separate raceways prevent magnetic imbalance.
UPS-A1 → PDU-A1 (IT hall)	UPS output to PDU, 1500A	5 sets of 750 kcmil Cu THWN-2 in cable tray (or TC-ER cable)	Cable tray reduces install labor 30-50% vs conduit. TC-ER cable rated for tray.
PDU-A1 → RPP-A1-1 (row level)	Sub-feeder, 400A	1 set of 500 kcmil Cu THWN-2 in 3" EMT	Single set fine at 400A. EMT for indoor finished space.
RPP → rack PDU strip	Branch circuit, 30A	#10 AWG Cu THWN-2 in 1/2" EMT, or MC cable in tray	Standard branch wiring. MC cable for faster row turn-up.
Site exterior → outdoor lighting	Underground branch, 20A	#12 Cu USE-2 direct buried, or THWN-2 in PVC conduit	USE-2 direct-burial rated and saves the conduit. THWN-2 in PVC is the conduit alternative.
Generator paralleling cabinet	Control wiring	#14 Cu MTW or TFFN, color-coded for control circuits	MTW (Machine Tool Wire) or TFFN for tight bends inside control cabinets.

i

Why one facility uses six different conductor types

Insulation choice follows environment + ampacity + termination + install method. A facility-wide standard ("we use THWN-2 everywhere") is unrealistic — MV cables, direct-burial outdoors, control circuits, and tray installations each have specific needs. Specifying the right type per location is part of professional design.

Worked Example 2 — Residential Service + Branches

Example 02 · Alternate scale Single-family home · 200A service · interior wiring + outdoor branches

Location	Specification	Why
Utility transformer → meter base (overhead)	Triplex (USE-2) AL service drop, sized to NEC 310.12	USE-2 weather + UV resistant; AL economical for utility-scale distribution. NEC 310.12 = "residential 83% rule".
Meter → main panel (interior)	4/0 AL SER cable	SE/SER cable is the residential service entrance standard. AL for cost savings.
Main panel → sub-panel (laundry / garage)	4-conductor #6 Cu (3 hot + 1 ground) NM-B (or 4-cond MC for garage)	NM-B = "Romex" for residential interior. MC required where physical protection needed.
Branch circuits (outlets, lighting)	#14, #12 Cu NM-B	Standard residential branch wire. Cu-only at small sizes per NEC.
Range, dryer (240V)	#6 (range) or #10 (dryer) Cu NM-B with separate ground	Modern residential 240V branches require 4-wire (2 hot + neutral + ground).
Outdoor receptacles, garage door	#12 Cu UF-B direct burial OR THWN-2 in PVC conduit	UF saves conduit cost. PVC + THWN-2 is more rigorous and easier to repair.
Pool / hot tub circuits	#10 or #8 Cu THWN-2 in PVC, GFCI protected	Specialized rules per NEC 680. PVC conduit (no metal in pool area).

!

NEC 310.12 — the residential 83% rule

For 100A through 400A 1φ-3W residential services only, NEC 310.12 lets you use one wire size smaller than NEC 310.16 would require. Example: 200A service with 4/0 AL (NEC 310.12) instead of 250 kcmil AL (NEC 310.16). Saves significant cost. Applies ONLY to residential service entrance.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Decode the letters

Decode XHHW-2:

Drill 2 · Termination temp

Equipment marked '75°C' rated. Wire is THWN-2 (90°C). Which ampacity column?

Drill 3 · Cu vs AL

Need 200 A ampacity. Cu = 3/0 AWG. AL ≈ ?

Drill 4 · Direct burial

Underground conductor without conduit, residential branch — type?

Drill 5 · Residential 83% rule

200A 1φ residential service. NEC 310.16 calls for 250 kcmil AL. Per NEC 310.12, what's allowed?

If You See THIS, Think THAT

If you see…	Think / use…
"THWN-2" called out	PVC + nylon, 90°C wet/dry. Default for new commercial/industrial conduit-wired work.
"THHN" only	Dry locations only — 90°C dry but no wet rating. Largely obsolete in favor of THWN-2.
"XHHW-2"	XLPE insulation (tougher than PVC), 90°C wet/dry. Premium choice for large feeders, MV transitions.
"USE-2" in residential	Direct-burial service entrance. NEC 338 — also rated as RHH/RHW-2 for indoor use when so labeled.
"MV-105"	Medium voltage cable, 105°C. 5/15/35 kV applications. Requires shielding + special terminations.
"NM-B"	Romex. Residential interior only — most jurisdictions ban from commercial buildings.
"MC" (metal-clad)	Conductors in metal armor. Tray-rated, cable-managed install. Replaces conduit-wired systems for labor savings.
"TC-ER" or "TC"	Tray cable, exposed-run rated. NEC 392 cable tray installations.
Termination ≤ 100A	NEC 110.14(C): 60°C ampacity column. Even if wire is 90°C-rated.
Termination > 100A	NEC 110.14(C): 75°C column. THWN-2 / XHHW-2 90°C rating used only for derating margin.
Aluminum conductor specified	Use AL-rated lugs, antioxidant compound (Penetrox/Noalox), torque per spec. NEC 110.14 enforces.
Direct burial without conduit	USE-2 (for service or feeder); UF-B (for residential branches). NOT THWN-2.
Residential service ≤ 400A 1φ	NEC 310.12 — smaller residential service conductor allowed (the "83% rule").
"TC-MC" or "MC-HL"	Specialized variants: TC-MC for tray + low temp; MC-HL for hazardous (Class I Div 1) locations.

§09 / 39

Cable Tray & Busway

NEC 392 (tray) · NEC 368 (busway) · routing alternatives to conduit

When you have many large conductors going the same direction, conduit becomes ridiculous — labor cost dwarfs material cost. Cable tray (open support) and busway (factory-built bus bars) are the alternatives. Each has its own NEC article and economic sweet spot.

Cable Tray vs Busway vs Conduit — Decision Matrix

Three ways to route many conductors from one place to another. Each wins in different conditions.

Feature	Conduit (EMT, RMC, PVC)	Cable Tray (NEC 392)	Busway (NEC 368)
Construction	Pipe with conductors pulled through	Open support carrying cable	Factory bus bars in metal enclosure
Best ampacity range	Up to ~600 A (single set)	200 A to 5,000+ A	225 A to 6,300 A
Install labor (rel.)	1.0× (baseline)	0.4×–0.6× (much faster)	0.3×–0.5× (modular)
Material cost (rel.)	1.0×	1.1×–1.3×	1.5×–2.0×
Future modifications	Difficult — pull additional conductors	Drop new cables in easily	Plug-in: tap anywhere
Atlas DC1	RPP feeders, branches	UPS-to-PDU feeders, MV cables	Possible 480V SWGR riser

Cable Tray Types — NEC 392

Tray type	Description	Pros	Cons	Common use
Ladder	Side rails with rung crossmembers	Best ventilation, easy cable drops, lightest	Cable can sag between rungs	Industrial, MV, large bundles
Solid bottom	Continuous metal bottom	Maximum cable support, EMI shielding	Heat trap (worst ventilation)	Sensitive electronic cables, RF environments
Ventilated bottom	Perforated bottom	Compromise: support + airflow	Heavier than ladder	Default for commercial work
Wire mesh	Welded wire grid	Cheapest, lightest, fastest install, full ventilation	Lower load capacity	Data centers (telecom + power), modern commercial
Channel	U-shape, single cable	Easy single cable runs	Limited capacity	Single MV feeder, single fiber bundle

NEC 392 Tray Fill Rules

Cable type	NEC section	Fill limit
Multiconductor (MC, TC) ≥ 4/0 AWG	392.22(A)(1)	Sum of cable diameters ≤ tray width — single layer
Multiconductor < 4/0 AWG	392.22(A)(2)	Sum of cross-sectional areas per Table 392.22(A) — multi-layer OK
Single conductor ≥ 1/0 AWG (tray-rated)	392.22(B)	Specific tables per AWG range — only marked types (e.g., XHHW-2)

i

When tray ampacity needs derating

NEC 392.80 — for TC and MC cables with continuous covers, follow Table 310.16 (raceway) ampacities. For uncovered tray with few cables, free-air ampacities apply (much higher).

Busway — NEC Article 368

Factory-built bus bars in a metal enclosure, sold in 10-ft sections that bolt together. Two main types:

Busway type	Description	Where used
Feeder busway	No tap openings; point-to-point distribution	SWGR-to-SWGR, vertical risers in tall buildings, large feeder runs
Plug-in busway	Tap openings every 24" for plug-in switches	Industrial overhead distribution, manufacturing floors with movable equipment
Sandwich (low-impedance)	Bus bars stacked tightly with insulation between → very low impedance	Data centers, sensitive electronic facilities

Standard Busway Ampacity Sizes

225 · 400 · 600 · 800 · 1000 · 1200 · 1600 · 2000 · 2500 · 3000 · 4000 · 5000 · 6000 A

Worked Example 1 — Atlas DC1 UPS-to-PDU Feeder

Example 01 · Atlas DC1 spineUPS-A1 → PDU-A1 (1500A, 250 ft) — three options compared

Method	Spec	Material	Labor	Total
Conduit (5 sets THWN-2)	5× 4" EMT, 5 sets 750 kcmil Cu, fittings	~$60K	1.0× ~$80K	~$140K
Cable Tray (TC-ER cable)	250 ft 18" wire mesh tray + 5 runs of 3/c 750 kcmil Cu TC-ER	~$70K	0.5× ~$40K	~$110K
Feeder Busway (1600A)	250 ft of 1600A AL feeder busway, 4 fittings	~$120K	0.4× ~$32K	~$152K

Result: Cable tray with TC-ER cable wins. Conduit rejected (too much labor for 5 parallel sets). Busway rejected (premium not justified for fixed point-to-point).

i

Why DC operators love cable tray

Beyond cost: visibility. Cable tray exposes every conductor for inspection. When something fails or needs replacement, you see it. Conduit hides everything.

Worked Example 2 — Manufacturing Floor Plug-In Busway

Example 02 · Alternate scaleIndustrial plant overhead distribution — 800A plug-in busway feeding movable CNC machines

Why busway here: Plant rearranges machines every few months. Conduit-fed branches require new pulls each time. Plug-in busway taps every 24" — plug a switch in anywhere.
Sizing: 12 CNC × 30 HP × 3 = ~108 kW continuous demand. With 1.25× + diversity → ~140 kW = 169 A. Round up: 800 A plug-in busway (significant future capacity).
Routing: 200 ft along overhead truss. Standard 10-ft sections + corner fittings. Each machine's plug-in switch is a fused disconnect with branch-circuit OCPD.
Cost: The flexibility (any machine moves, any branch added/removed without electrician callout) often pays the busway premium in the first year.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · When tray wins

5 parallel 750 kcmil feeders, 250 ft route. Conduit or tray?

Drill 2 · Tray fill — TC cable

8" wide ladder tray, 4" usable for cables. Cable diameter 1.5". Max cables side-by-side?

Drill 3 · Plug-in busway

Industrial floor with movable equipment. Best routing method?

Drill 4 · Standard busway sizes

Need ~ 1,400 A continuous. Standard busway sizes?

Drill 5 · TC vs MC

Cable in tray, exposed run. Type?

If You See THIS, Think THAT

If you see…	Think / use…
Many parallel feeders going same direction	Cable tray. Conduit gets unwieldy past 3-4 parallel sets.
"Plug-in busway"	NEC 368. Industrial floor distribution where tap points needed.
"Feeder busway"	Point-to-point factory bus bars. Vertical risers in skyscrapers, SWGR-to-SWGR.
"Wire mesh tray"	Cheapest, fastest tray to install. Common in DCs.
"Solid-bottom tray"	Heat trap — derate cable ampacity per NEC 392.80. Use only when EMI/EMC requires.
"TC-ER" cable	Tray Cable, Exposed-Run rated. Designed for cable tray (NEC 336).
"MC" cable in tray	NEC 330 + 392 — MC is tray-rated when armored.
Single conductors in tray	NEC 392.10 — only ≥ 1/0 AWG, only specific marked types (XHHW-2).
Frequent equipment moves expected	Plug-in busway. Otherwise tray or conduit.

§10 / 39

Transformers

kVA · %Z · Δ-Y configurations · inrush · NEC 450

Every voltage transition in your system has a transformer behind it. Sizing comes from the load study; %Z determines fault current downstream; the winding configuration determines grounding rules. Get all three from the cutsheet.

The Three Numbers Every Transformer Cutsheet Has

Sizing is from the load study (kVA). %Z determines downstream fault current. The winding configuration determines grounding rules. Get all three from the cutsheet — every other characteristic follows.

Parameter	What it does	Atlas DC1 TX-A
kVA rating	Maximum continuous output. Sized at ~110-125% of demand load to allow thermal cycling.	2,500 kVA
Voltage ratings	Primary / secondary nominal voltages. Determines turns ratio and tap settings.	12,470 / 480Y/277V
%Z (impedance)	Per-unit impedance. Lower %Z → higher fault current downstream. Standard values 4.5–7%.	5.75%
Winding configuration	Δ-Y, Y-Y, Δ-Δ, Y-Δ. Determines neutral availability and grounding strategy.	Δ-Y (delta primary, wye secondary, neutral grounded)
Cooling class	How heat is removed. ONAN (oil-natural air-natural), ONAF (oil-natural air-forced), KNAN (less-flammable fluid), Dry-type.	KNAN (less-flammable fluid for indoor use)
Insulation class	Temperature rise rating. 65°C standard for new equipment.	65°C rise
Tap settings	±2.5% no-load taps for fine voltage adjustment. 4 taps each side of nominal typical.	±5% in 2.5% steps

Standard kVA Sizes

Class	Standard sizes (kVA)
Single-phase	1, 1.5, 3, 5, 7.5, 10, 15, 25, 37.5, 50, 75, 100, 167, 250, 333, 500
Three-phase	15, 30, 45, 75, 112.5, 150, 225, 300, 500, 750, 1000, 1500, 2000, 2500, 3000, 5000, 7500, 10,000

Winding Configurations — Δ vs Y

Configuration	Where used	Pros	Cons
Δ-Y (delta-wye)	Most common. Distribution transformers, all utility step-downs to commercial/industrial	Wye secondary provides neutral for 1φ loads. Delta primary blocks zero-sequence currents from secondary fault → quieter primary.	30° phase shift between primary and secondary (leading or lagging by 30°, depending on labeling)
Y-Y (wye-wye)	Some utility distribution, autotransformers, industrial step-downs where primary and secondary both need neutrals	No phase shift. Both sides have neutrals available.	Requires careful 3rd-harmonic management; ground faults transfer between primary and secondary
Δ-Δ (delta-delta)	Industrial 480V-480V step transformers, isolation transformers	No phase shift. Open-delta operation possible (one bank can fail and system continues).	No neutral. Cannot serve 1φ phase-to-neutral loads.
Y-Δ (wye-delta)	Step-up transformers (generator to grid), some industrial applications	Generator side has neutral. Delta secondary blocks 3rd harmonics from grid.	30° phase shift (opposite direction from Δ-Y).

Why %Z Matters — Fault Current Downstream

%Z (per-unit impedance) determines how much current the transformer can deliver into a downstream fault. Lower %Z = higher fault current. This sets the AIC requirement for downstream switchgear.

Fault current at transformer secondary (infinite primary bus assumption)

I_fault = I_FLA / (%Z / 100)

Quick approximation assuming infinite primary bus (utility impedance = 0). Real fault current is somewhat lower because the utility has real impedance — see §12 for the rigorous MVA method that includes utility contribution. The error is small (~5%) for stiff utilities, larger (20%+) for weaker connections.

Inrush Current — Why Upstream Protection Sees the Pain

When energizing a transformer, the inrush can be 8–12× rated current for the first half-cycle, decaying to normal in 6-10 cycles. Upstream OCPD must allow this without tripping.

Aspect	Detail
Magnitude	8-12× FLA peak first half-cycle; decays in 6-10 cycles to normal
Cause	DC offset in flux when energized — depends on point-on-wave of switching
Mitigation	Upstream OCPD picked to coordinate above inrush curve. NEC 450.3 specifies primary protection ≤ 250% of rated primary current for transformers ≥ 1000V.
Sympathy inrush	Energizing a new transformer can trigger inrush in already-energized adjacent transformers — must consider in protection coordination

Worked Example 1 — Atlas DC1 TX-A Sizing & Fault Current

Example 01 · Atlas DC1 spine2,500 kVA, 12.47kV-480Y/277V, %Z = 5.75 — what does this mean for the system?

Why 2,500 kVA? Side A demand was ~2,791 kVA (from §03). 2,500 kVA appears slightly undersized — but real installation oversizes the genset side and accepts brief overload at full IT loading. Many real DCs would spec 3,000 kVA.
Secondary FLA:
FLA = 2,500,000 / (√3 × 480) = 3,007 A
Fault current at 480V bus (using %Z):
I_fault = 3,007 / 0.0575 = 52,300 A symmetric
→ 480V SWGR-A bus must be rated for ≥ 52 kA AIC. Standard ratings: 65 kA. ✓
Inrush: 10× × 3007 = ~30 kA peak first half cycle. Primary breaker (12.47 kV side) must let this through.
Primary protection (NEC 450.3 for transformers ≥ 1000V): 250% × primary FLA = 2.5 × (2,500,000 / (√3 × 12,470)) = 2.5 × 116 = 290 A. Primary CB sized at 300 A or fuse at 250 A. ✓

Worked Example 2 — Office Building Step-Down (480→208/120V)

Example 02 · Alternate scale75 kVA, 480-208Y/120V, %Z = 5%, dry-type indoor — sized for office lighting + receptacles

Office demand load: 50 kW lighting + 10 kVA receptacles + 5 kVA misc = ~65 kVA total demand. → Use 75 kVA standard size.
Primary FLA: 75,000 / (√3 × 480) = 90 A
Secondary FLA: 75,000 / (√3 × 208) = 208 A
Primary protection (NEC 450.3(B), < 1000V): 125% of primary FLA → 113 A → use 125 A breaker.
Secondary protection (NEC 450.3 not required if primary is sized at <125%): Many designs add secondary protection anyway — 225 A panelboard MCB matches the 208 A secondary FLA.
Fault current at 208V bus: 208 / 0.05 = 4,160 A. 208V panelboards routinely rated for 10 kA AIC — ample margin.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · FLA from kVA

1,500 kVA, 480V 3φ secondary. FLA?

Drill 2 · %Z fault current

1,500 kVA, %Z = 5.5%. Approximate fault current at secondary (infinite primary)?

Drill 3 · Δ-Y vs Δ-Δ

A transformer secondary serves both 3φ motors AND 1φ-N lighting. Configuration?

Drill 4 · NEC 450.3 primary

1,500 kVA at 4160V (primary). NEC 450.3 max primary OCPD?

Drill 5 · Atlas TX-A

Atlas TX-A: 2,500 kVA, 480V secondary, %Z = 5.75. Fault at secondary (infinite primary)?

Worked Example 3 — Setting Transformer Taps for Voltage Optimization

Real utility primary voltage often runs slightly off nominal. Distribution transformers have ±2.5% no-load taps to compensate. Setting the tap correctly delivers nominal voltage at the secondary loads.

Example 03 · Tap optimizationAtlas DC1 TX-A — utility primary measured at 12,830V instead of nominal 12,470V

The problem

Utility primary measured = 12,830V (2.9% high above 12,470V nominal). Common when transformer is close to substation.
TX-A secondary with all taps in nominal position would deliver: 12,830 / 12,470 × 480 = 494V at light load.
Per ANSI C84.1, utilization range is 456V to 504V (95-105% of 480V). 494V is within range but tight, especially if drops accumulate downstream.

Tap selection

TX-A has 5 no-load primary taps in 2.5% steps: +5%, +2.5%, Nominal, −2.5%, −5%.

Choose the +2.5% tap. This raises the primary turns by 2.5%, requiring 2.5% more primary voltage to deliver the same secondary voltage.
New secondary at no-load:
V_sec = (12,830 / 12,470) × (1 / 1.025) × 480 = 1.029 × 0.9756 × 480 = 482V
Centered in the 456-504V range. Plenty of margin for downstream voltage drops.

!

"No-load" taps must be set with transformer DE-ENERGIZED

No-load (NLT) taps require the transformer to be completely de-energized — they cannot be switched under load. Load Tap Changers (LTC) are different equipment that can switch under load (used in utility substation transformers, not pad-mounts). For Atlas DC1 pad-mount TX-A, tap changes happen during scheduled maintenance with full LOTO (§29).

i

Why measure first, set tap second

Don't set taps from the design assumption. Wait for actual primary voltage measurements over several days/weeks (some utilities have seasonal variation). Then set the tap that centers your secondary in the ANSI band.

If You See THIS, Think THAT

If you see…	Think / use…
"%Z" or "5.75%" on cutsheet	Per-unit impedance. Drives fault current. Lower %Z = more fault current downstream.
"Δ-Y" winding	Standard utility/commercial config. 30° phase shift. Wye secondary has neutral.
"Y-Y" winding	Both sides have neutrals. Watch for 3rd harmonic issues. Less common.
"Δ-Δ" winding	No neutral. Industrial 480-480V isolation. Cannot serve 1φ-N loads from secondary.
"K-factor 4" or "K-13" transformer	Designed for harmonic loads (servers, VFDs, LEDs). Larger neutral, special core. Used in DCs.
"Pad-mount" transformer	Outdoor utility-grade. 12.47kV/480V typically. Used at service entrance for commercial/industrial.
"Dry-type" transformer	Indoor, no oil. NEMA 1 enclosure. Lower %Z = louder. Cooling: AA (ambient air), AFA (forced air).
"ONAN/KNAN" transformer	Liquid-cooled. ONAN = mineral oil, KNAN = less-flammable fluid (FM-200, Envirotemp). KNAN is required for indoor liquid-cooled.
"Inrush current"	8-12× rated for ~half cycle. Upstream OCPD must coordinate above inrush curve. NEC 450.3 sizing.
NEC 450.3	Transformer overcurrent protection. ≥ 1000V: ≤ 250% primary; < 1000V: ≤ 125% primary (with exceptions).
Tap settings shown	±2.5% no-load taps. Adjust if utility primary voltage is consistently high or low.

§11 / 39

Service Entrance & Utility Coordination

NEC 220 · NEC 230 · primary vs secondary metered · utility coordination

Service entrance is the boundary where utility responsibility ends and yours begins. NEC Article 230 governs everything between the utility connection point and the first overcurrent device inside the building. Utility coordination is the longest schedule pole on most projects.

Where the Building Meets the Grid

Service entrance is the boundary where utility responsibility ends and yours begins. NEC Article 230 governs everything between the utility connection point and the first overcurrent device inside the building.

Service component	What it is	Whose responsibility
Service drop / lateral	Conductors from utility supply to service point (overhead = drop, underground = lateral)	Utility (typically up to weatherhead/pad)
Service point	Demarcation between utility and customer ownership	Defined by tariff agreement
Service entrance conductors	From service point to service equipment	Customer (electrical engineer designs)
Metering equipment	CTs/PTs and meter — secondary or primary metered	Utility owns; customer provides space + cabinet
Service disconnect	Main breaker(s) that disconnect the entire building	Customer; up to 6 disconnects allowed (NEC 230.71)
Service overcurrent device	Protects service entrance conductors	Customer; sized per NEC 230.90

NEC 220 — Standard Method vs Optional Method

Method	NEC reference	Where used	Result
Standard Method	NEC 220 Part III (sections 220.40–220.61)	Universal — works for any occupancy. Required for nontypical loads.	Detailed line-by-line load tabulation with NEC table demand factors
Optional Method (dwelling)	NEC 220.82	Single-family dwellings only. Simpler.	Apply 100% to first 10 kVA of total connected, 40% to remainder. Plus additional rules for HVAC.
Optional Method (existing dwelling)	NEC 220.83	Existing dwelling adding load (e.g., HVAC retrofit)	Lets you check if existing service is adequate without recalculating from scratch
Optional Method (multi-family)	NEC 220.84	3+ unit dwellings only	Per-unit demand factor table for entire building
Optional Method (school)	NEC 220.86	Schools — stadium lighting, athletic loads	Special demand factors

Primary vs Secondary Metering

Metering type	Description	When used	Pros	Cons
Secondary	Meter is on the LV (customer) side of the service transformer. Utility owns transformer.	< 500 kVA typically. Small commercial.	Simpler installation. Utility owns/maintains transformer.	Customer pays for transformer losses (heat = wasted energy on customer side).
Primary	Meter is on the HV side of the service transformer. Customer owns transformer.	Larger services (≥ 500 kVA typical). Atlas DC1 case.	Lower kWh rate (utility passes through transformer loss savings). Customer can choose transformer specs.	Customer responsible for transformer maintenance, replacement, fault.

Coordinating with the Utility — What to Bring to the Meeting

You need from the utility	You bring to the utility
Available fault current at service point (kA at primary, kA at secondary)	Single-line diagram showing service equipment, transformer, main switchgear
Voltage at service point (utility's nominal) and tolerance band	Estimated demand load (kW + kVA + PF)
Service voltage class options (12.47kV, 4160V, 480V)	Any large motor starting kW (for voltage flicker check)
Metering location requirements	Construction schedule (need by date)
X/R ratio and impedance to bus	Site plan with proposed building location
Backfeed / generator paralleling rules (UL 1741, IEEE 1547)	Backfeed plans (PV, ESS, generator paralleling)
Tariff / billing rate options + special rate qualifications (TOU, demand)	Anticipated load growth over 5-year horizon
Demand limit for service voltage class (some utilities require step-up to MV for ≥ 1MW)	Required reliability tier (e.g., dual feeders for hospital/data center)

!

Schedule reality — utility coordination is the long pole

Utility transformer lead times can run 12-18 months in 2025-2026. For data centers and large industrial, the utility coordination meeting must happen before design is finalized. Utility engineers typically prefer 6-12 months notice for any service ≥ 500 kVA, more for MV.

Worked Example 1 — Atlas DC1 Service Coordination

Example 01 · Atlas DC1 spine12.47 kV primary metered service · 5 MW total demand · two 2500 kVA pad-mount transformers (2N)

Service architecture

Why MV primary metered: Total demand > 1 MW makes 480V impractical (would need 6 separate 1500 kVA transformers and a parallel switchgear lineup). MV at 12.47 kV is utility's standard distribution voltage in this region.
Why two transformers: 2N redundancy. Either TX-A or TX-B alone serves the full IT load. Each transformer fed from a different utility distribution feeder for true redundancy.
Utility data received: Available fault current at service point = 50 kA RMS symm at 12.47 kV. X/R ratio = 8.5. Voltage 12.47 kV ±5%.
Metering: Customer provides utility metering cabinet. CT/PT compartment in MV switchgear. Utility installs meter; customer never touches it (sealed).
Required tariff: Large General Service - Time of Use (LGS-TOU). Demand charges + energy charges. Penalties for power factor below 0.95 (Atlas DC1 spec'd for 0.95 PF correction).
Generator paralleling: Atlas DC1 does NOT parallel generators with utility (open-transition ATS). Avoids IEEE 1547 / UL 1741 compliance complexity.

Worked Example 2 — 50-Unit Apartment Service (NEC 220 Standard)

Example 02 · Alternate scale50 dwelling units · NEC 220 standard method · secondary metered 480-208Y/120V utility transformer

Demand calculation completed in §03: 352.7 kVA = 980 A at 208V 3φ.

Service voltage: 208Y/120V 3φ-4W. Secondary metered (utility owns 480-208 transformer; building gets 208V at the meter).
Service equipment: 1,200 A main service disconnect (or fused). Service entrance bus rated 1,200 A. Each apartment unit fed from a 200 A panelboard via meter stack on the exterior.
NEC 230.71: Up to 6 service disconnects allowed without separate switch. Common configuration: 1 main service disc + 50 unit disconnects via meter stack. Most jurisdictions require single main service disconnect.
Coordination: Utility installs the pad-mount transformer (typically 750 kVA for this load). Customer installs the meter stack and unit panels.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Article scope

Which NEC article covers service-entrance conductors?

Drill 2 · NEC 230.71

How many service disconnects allowed per NEC 230.71?

Drill 3 · GFP threshold

When is NEC 230.95 GFP required?

Drill 4 · Primary vs secondary metered

5 MW commercial facility — primary or secondary metered?

Drill 5 · NEC 220 method

Single-family dwelling load calc — which method?

If You See THIS, Think THAT

If you see…	Think / use…
"Service entrance"	NEC Article 230. Conductors from utility to first OCPD. Different rules from feeder.
NEC 230.71 — "up to 6 service disconnects"	Allowed but most modern jurisdictions require a single main disconnect.
NEC 230.95	Ground fault protection required at service for 480V/277V services with main ≥ 1000A.
"Primary metered" service	≥ 500 kVA typical. Customer owns transformer. Lower kWh rate.
"Secondary metered" service	Smaller services. Utility owns transformer.
NEC 220 Standard Method	Universal load calculation method. Always works.
NEC 220 Optional Method	Dwelling-specific shortcut. NEC 220.82 (single), 220.84 (multi-family).
"NEC 310.12"	Residential service entrance conductor "83% rule" — smaller AL conductor allowed.
"Available fault current at service"	Need from utility to size service equipment AIC + arc flash inputs.
"Utility transformer lead time"	12-18 months in 2025-2026. Coordinate utility EARLY.
"Backfeed" or "PV/ESS interconnection"	Triggers IEEE 1547 / UL 1741 utility approval. Adds 3-9 months to schedule.

§12 / 39

Overcurrent Protection & Coordination

Fuses vs CBs · TCC curves · selective coordination · NEC 700.27

Every conductor in your system has an OCPD upstream. Picking the right one isn't just about ampacity — it's about what trips first when something faults. Selective coordination keeps a single fault from taking down half the building.

Fuses vs Circuit Breakers

Two technologies, both meeting NEC requirements, very different operating characteristics. Coordination strategies depend on which type you choose.

Property	Fuses	Circuit Breakers
Operation	Sacrificial — element melts on overcurrent	Reusable — mechanical contacts open
Speed (low fault)	Slower (thermal element)	Faster (thermal-magnetic)
Speed (high fault)	Faster (current-limiting fuses can clear in < 1/4 cycle)	Slower (must wait for half cycle minimum)
Coordination	Easier — fuse curves naturally cascade	Harder — requires careful selection or zone-selective interlocking
Replacement	Stock 3 fuses, replace blown ones	Reset, no inventory
Single-phase tripping	Single fuse blows on single-phase fault → motor singles-out	3-pole CB trips all 3 phases together
Cost (per device)	Lower for fuse + holder	Higher for breaker
Typical use	Industrial, MV, high-fault situations, motor branches	Commercial buildings, panelboards, lighting branches

Time-Current Curves (TCC) — How to Read Them

A TCC plots how long a device takes to trip vs the current flowing through it. Both axes are logarithmic — covers 6+ decades on a single chart. Reading a TCC is the foundation of every coordination study.

Two CBs. Branch (green) is to the LEFT of feeder (copper) at every current — coordinated.

Selective vs Cascading Coordination

Coordination type	Description	Pros	Cons	Where used
Selective	Downstream device opens FIRST for any fault current. Upstream remains closed.	Minimum disruption. Only the faulted branch loses power.	More expensive equipment. May require larger upstream breakers.	Hospitals (NEC 700.27), data centers, life-safety systems
Cascading	Upstream device may also trip on high faults. Downstream sometimes never opens.	Less expensive. Upstream protects downstream rated lower than fault current.	Larger sections lose power on fault. Some equipment may not get isolated.	Most commercial buildings (cost-driven)
Series-rated	UL-listed combination where downstream CB has lower interrupting rating than fault current.	Allows lower-rated downstream CBs in high-fault systems.	NEC 240.86: must use UL-listed combination. Many AHJs question this.	Sometimes residential service entrance (200A 22 kA breaker behind 100kA fault).
Zone-Selective Interlocking (ZSI)	Modern electronic CBs communicate. Downstream CB tells upstream "I see the fault, don't trip."	Selective coordination AT FULL FAULT levels. Best of both worlds.	Requires modern electronic CBs and signal wiring.	New construction in critical facilities; data center MV switchgear.

!

NEC 700.27 — Selective coordination required for life safety

For NEC Article 700 emergency systems (life safety), selective coordination is mandatory. A fault in one part of the emergency system cannot trip an upstream device serving other emergency loads. This is a code requirement, not optional. Hospitals, schools, places of assembly all touch this.

Worked Example 1 — Atlas DC1 Coordination Cascade

Example 01 · Atlas DC1 spineCoordinating UPS-A1 output → PDU-A1 input → RPP-A1-1 main → branch breaker

The chain (top to bottom)

Position	Device	Trip A	Why this rating
1 (UPS output)	2,000 A static-trip CB	2,000 A	Sized for full UPS-A1 output (1,500 A × 125% = 1,875 → round up to 2,000)
2 (PDU primary)	800 A LSIG (electronic) CB	800 A	PDU-A1 input current 602 A × 125% = 753 → round up to 800. Electronic trip allows instantaneous setting tuned for selectivity.
3 (RPP main)	400 A MCB	400 A	RPP bus rated 400 A (from §05 calc). 124 A demand × 125% = 155 → 400 A bus allows future growth.
4 (branch)	30 A 1-pole	30 A	Server rack: 24 A continuous × 125% = 30 A.

Coordination check

Test fault at branch (rack PDU): Available fault ~12 kA at 240V branch.
30A CB clears in < 0.01 s (instantaneous region). 400A MCB instantaneous pickup at 5× = 2,000 A → no trip on 240V fault current. ✓ Branch isolates only.
Test fault at RPP main: Available fault ~18 kA at 415V.
400A MCB clears in 0.1-1 s depending on settings. 800A LSIG must wait at least 0.3 s before tripping. Coordinated if settings tuned. ✓
Test fault at PDU input: Available fault ~25 kA at 480V.
800A LSIG clears in 0.2-0.5 s. 2000A static trip set for short delay 0.5 s. Coordinated. ✓

Result: Full selective coordination achieved. A fault anywhere isolates only the affected branch.

Worked Example 2 — Hospital Life Safety Coordination (NEC 700.27)

Example 02 · Alternate scaleHospital essential electrical system — life safety branch must be selectively coordinated per NEC 700.27

System: 600 A generator → 400 A ATS → 225 A panelboard MCB → 20 A branch breakers (egress lighting, exit signs, fire alarm).
Coordination requirement: NEC 700.27 — for any fault current available, the OCPD closest to the fault must clear before any upstream OCPD operates. This is at ALL fault levels, not just bolted faults.
Method: Use fuses in the chain (their curves naturally cascade), or use ZSI-equipped electronic CBs.
Documentation: Submit a coordination study showing TCC plots with NO overlap at any fault level. AHJ reviews.

i

Why fuses won this market

Hospitals adopted fuses heavily after NEC 700.27 because their curves naturally coordinate without engineering analysis. A 100 A fuse won't blow before a 60 A fuse upstream — physics. Modern ZSI breakers achieve the same result electronically but require system design and commissioning.

TCC Plot — Real Coordinated Cascade (Atlas DC1)

This is the TCC plot for Atlas DC1's UPS → PDU → RPP → branch coordination. Each curve shows trip time vs current. Curves to the LEFT trip first.

Each curve to the LEFT of upstream curves at all fault levels. Selectivity = no overlaps. ✓

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · TCC reading

Two breakers on a TCC: A is to the LEFT of B at 5 kA. Which trips first at 5 kA fault?

Drill 2 · Selective coordination

A fault at branch level. Which breaker should open?

Drill 3 · NEC 700.27

When is selective coordination MANDATORY?

Drill 4 · Maintenance switch

Adjustable instantaneous trip set lower during energized work — what's it for?

Drill 5 · Atlas DC1 chain

Atlas DC1 fault at RPP. With 30A → 400A → 800A → 2000A chain, which opens?

If You See THIS, Think THAT

If you see…	Think / use…
"Coordination study"	TCC plot showing every protective device. Verify no upstream curve overlaps a downstream curve at any current.
"Selective coordination required"	NEC 700.27 — life safety. Mandatory. Only branch closest to fault opens.
"Cascade" or "non-selective"	Multiple devices may trip on a fault. Lower cost, more disruption.
"Series-rated combination"	NEC 240.86 — UL-listed combination only. Verify with manufacturer documentation.
"ZSI" or "Zone-Selective Interlocking"	Electronic CBs that communicate. Modern approach to selective coordination.
"LSI" or "LSIG" trip unit	Long-time, Short-time, Instantaneous (+ Ground for G). Adjustable trip settings on electronic CBs.
Inverse-time CB curve	Standard thermal-magnetic. Slower at low current, fast at high current.
"Current-limiting fuse"	Special fuse that opens in less than ¼ cycle. Limits let-through energy. Used where fault currents very high.
"Maintenance switch" on a breaker	Reduces instantaneous setting during maintenance. Lowers arc flash incident energy. (See §18.)

§13 / 39

Short Circuit Analysis

MVA method · X/R ratio · AIC ratings · symmetric vs asymmetric

Available fault current at each bus determines what equipment ratings you need. Get it wrong and the equipment can fail catastrophically. The MVA method gives you a quick answer; per-unit gives you the rigorous answer.

Why Fault Current Matters

Three things depend on the available fault current at every bus in your system:

Use	Detail	Section reference
Equipment AIC rating	Switchgear, breakers, panelboards must withstand the fault current without exploding. AIC = Amperes Interrupting Capacity.	§05, §09
Conductor withstand	Conductors can be damaged by fault current. Larger sizes withstand more. Per IEEE 242 / NEC 110.10.	§07
Arc flash incident energy	Fault current is one of the two key inputs to IEEE 1584 calculation (other is trip time).	§18
Coordination study	TCC plots overlay against fault current to verify selectivity at all fault levels.	§11

Symmetrical vs Asymmetrical Fault Current

Type	Description	When it matters
Symmetrical RMS	Steady-state AC fault current — what the meter reads ~6 cycles after the fault	Equipment AIC ratings (interrupting), continuous bus rating
Asymmetrical (DC offset)	First half-cycle includes DC offset from the inductive system. Peak can be 2.7× RMS symmetric.	Equipment momentary withstand (closing into a fault), arc flash calculation
X/R ratio	System reactance / resistance. Higher X/R = more DC offset = higher asymmetric current. Typical: 6 (LV) to 30 (MV).	Multiplier on symmetric fault to get asymmetric

The MVA Method — Fast Hand Calc

For radial systems, the MVA method gives a quick, accurate fault current at any bus. Convert every impedance to its MVA contribution, combine in series/parallel, divide into voltage to get fault current.

Source MVA from utility

MVA_util = √3 × V × I_fault / 1,000,000

If utility says 50 kA at 12.47 kV: MVA = 1.732 × 12,470 × 50,000 / 1,000,000 = 1,080 MVA

Transformer MVA on its base

MVA_xfmr = kVA × 100 / %Z

Atlas TX-A: 2500 × 100 / 5.75 = 43.5 MVA

Combine in series (utility + transformer)

1 / MVA_total = 1 / MVA_util + 1 / MVA_xfmr

Like resistors in parallel — the smaller one dominates.

Fault current at the bus

I_fault = MVA_total × 1,000,000 / (√3 × V)

Equipment AIC Ratings — Standard Levels

Equipment class	Standard AIC ratings (kA)
Residential breakers	10, 22
Commercial molded-case (MCCB)	14, 18, 22, 25, 35, 65, 100
Insulated-case (ICCB) and low-voltage power CB	35, 65, 85, 100, 200
Medium voltage (5kV/15kV)	25, 40, 50, 63 (kA RMS sym)
Current-limiting fuses (LV)	200, 300 (interrupt rating)

Worked Example 1 — Atlas DC1 Fault Current at 480V SWGR-A (MVA Method)

Example 01 · Atlas DC1 spineFault current at 480V bus — utility through TX-A

Step 1 — Utility MVA at 12.47kV (given by utility):
I_util = 50 kA at 12.47 kV
MVA_util = √3 × 12,470 × 50,000 / 10⁶ = 1,080 MVA
Step 2 — TX-A MVA on its base:
MVA_TX-A = 2,500 × 100 / 5.75 = 43.5 MVA
Step 3 — Combine in series:
1/MVA_total = 1/1080 + 1/43.5 = 0.000926 + 0.02299 = 0.02391
MVA_total = 1 / 0.02391 = 41.8 MVA
Step 4 — Fault current at 480V bus:
I_fault = 41.8 × 10⁶ / (√3 × 480) = 41,800,000 / 831 = 50,300 A symmetric RMS
Step 5 — Equipment selection:
480V SWGR-A bus and breakers must be rated for ≥ 50 kA. Standard rating: 65 kA AIC. ✓
Step 6 — Asymmetric peak (for arc flash):
X/R at 480V bus from TX-A ≈ 8. Multiplier = ~1.4 for first half cycle. Peak asymmetric = 50,300 × 1.4 = ~70,400 A peak

!

Don't ignore the utility contribution

A common error: using only the transformer impedance and assuming the utility is "infinite." For Atlas DC1, the transformer alone would give MVA = 43.5 → fault = 52.4 kA. Adding the real utility brings it down to 41.8 MVA → 50.3 kA. The error is small here (~5%) but can be 20%+ on weaker utility connections.

Worked Example 2 — Office Building 200A Service Fault

Example 02 · Alternate scaleOffice building · 200A service · 480-208Y/120V utility transformer 75 kVA at %Z = 5%

Utility MVA (assumed infinite at 480V primary):
Treat as infinite — typical assumption for small-service calc.
Transformer MVA:
MVA_tx = 75 × 100 / 5 = 1,500 kVA = 1.5 MVA
Fault at 208V bus:
I_fault = 1,500,000 / (√3 × 208) = 4,160 A
Equipment:
Standard 200A panel = 10 kA AIC. Significantly above the 4.2 kA available — ample margin. ✓

Why small services rarely have AIC issues: small transformers limit fault current. Most residential and small commercial work doesn't even need a fault study — code-minimum equipment ratings suffice.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · MVA from utility

Utility says 25 kA at 12.47 kV. Source MVA?

Drill 2 · Transformer MVA

1000 kVA, %Z = 5%. MVA on its base?

Drill 3 · Series combination

Utility 540 MVA + Transformer 20 MVA. Combined MVA?

Drill 4 · AIC required

Fault current at 480V bus = 35 kA symmetric. Equipment AIC?

Drill 5 · Asymmetric peak

X/R = 10. Symmetric = 30 kA. Approximate asymmetric peak?

Worked Example 3 — Per-Unit Fault Analysis (Multi-Source)

The MVA method is fast but breaks down when you have multiple sources or want to track voltages across transformations. Per-unit handles both. Here's the rigorous version of the Atlas DC1 fault current calc.

Example 03 · Atlas DC1 spinePer-unit fault analysis at 480V SWGR-A — utility + TX-A + motor contribution

Pick a base

S_base

100 MVA (system-wide reference)

V_base,MV

12.47 kV

V_base,LV

480V

Convert each source impedance to system pu

Utility: 50 kA at 12.47 kV → MVA_util = √3 × 12.47 × 50 = 1,080 MVA

Z_pu,util = 100 / 1,080 = 0.0926 pu on 100 MVA base
TX-A: 5.75% on 2,500 kVA own base. Convert to 100 MVA system base.

Z_pu,TX = 0.0575 × (100,000 / 2,500) = 2.30 pu
Motor contribution (running induction motors back-feed fault for first ~ 4 cycles): Atlas DC1 Side A has chillers on VFDs. VFD-driven motors do NOT back-feed (rectifier blocks reverse current). Motor contribution = 0 here. (For DOL-fed motors, would add ~ 6× motor FLA at 4-6 cycles.)

Combine in series + solve

Series total impedance:

Z_pu,total = 0.0926 + 2.30 = 2.39 pu
Per-unit fault current:

I_pu = 1.0 / 2.39 = 0.418 pu
Base current at 480V bus:

I_base,LV = 100 × 10⁶ / (√3 × 480) = 120,300 A
Actual fault current:

I_fault = 0.418 × 120,300 = 50,300 A

Same answer as the MVA method — because per-unit and MVA are the same math, expressed differently. The per-unit method is more flexible when you have generators, motor contributions, multiple parallel paths, or want to track voltages.

i

When per-unit beats MVA method

(1) Multiple sources (utility + on-site generators). (2) Looped systems where current flows multiple paths. (3) When you need post-fault voltages at every bus. (4) Software tools (SKM, ETAP) work in per-unit internally. (5) Motor contributions — easier to add as parallel impedances.

Motor Contribution to Fault — When It Matters

Running induction motors don't drop fault current to zero instantly when faulted. Their inertia keeps the rotor spinning briefly, generating back-EMF that contributes fault current for the first ~ 4 cycles.

Motor type	Contribution magnitude	Duration
Induction motor (DOL-started)	4-6× motor FLA	4-6 cycles
Induction motor (VFD-driven)	~ 0 (rectifier blocks back-feed)	—
Synchronous motor (excited)	Behaves like a generator: 4-8× FLA	Continuous (until field collapses, ~ 30 cycles)
Synchronous condenser	Highest contribution	Continuous

!

Motor contribution can be the difference between adequate + inadequate AIC

A 480V system with utility-only fault = 35 kA can rise to 50 kA when adjacent 1,000 HP synchronous motors back-feed the fault. Equipment AIC must withstand the combined value. Always include motor contribution in industrial fault studies.

Symmetrical Components Decomposition

Real fault currents are rarely balanced. Phase-phase faults, ground faults, and open conductors all create unbalanced 3φ conditions. Symmetrical components let us analyze unbalanced cases using three independent BALANCED systems.

Component	Description	Equipment behavior
Positive (1)	Normal balanced 3φ ABC rotation. Always present in healthy operation.	Equipment positive-sequence impedance Z₁ = nameplate %Z (for transformers/generators)
Negative (2)	Balanced 3φ ACB (reverse) rotation. Caused by unbalance.	Z₂: rotating machines have Z₂ ≠ Z₁ (negative-sequence stator current creates a counter-rotating field — induces 2× rotor losses, hence motor 46 protection)
Zero (0)	3 phasors in-phase (no rotation). Flows only when neutral path exists.	Z₀: depends on grounding scheme. Δ-connected windings BLOCK zero-sequence (no path back).

Sequence Networks for Common Faults

Each fault type connects the three sequence networks differently. The interconnection determines the fault current.

Fault type	Sequence network connection	Fault current formula	Magnitude vs 3φ bolted
3φ symmetrical (3LG or 3L)	Positive sequence only — negative + zero networks not involved	I = V / Z₁	1.00× (reference)
Single line-to-ground (SLG)	Z₁ + Z₂ + Z₀ in series	I = 3V / (Z₁ + Z₂ + Z₀)	Often higher than 3φ on solidly grounded systems (especially close to transformer)
Line-to-line (L-L)	Z₁ + Z₂ in series	I = √3 × V / (Z₁ + Z₂)	~ 0.87× of 3φ
Double line-to-ground (LL-G)	Z₁ in series with (Z₂ ∥ Z₀)	Complex — usually computed by software	Variable; often higher than L-L

!

SLG can EXCEED 3φ fault — and it's common

On solidly grounded systems with a Δ-Y transformer (the typical Atlas DC1 setup), the SLG fault on the wye side can be 1.0-1.3× the 3φ fault current. Why? Because the zero-sequence current from a near-fault has very low impedance (just the transformer + neutral path) compared to the source impedance. Always check both 3φ AND SLG when sizing equipment AIC.

Sequence Impedances of Common Equipment

Equipment	Z₁ = Z₂ ?	Z₀ behavior
Transmission line	Yes	Z₀ ~ 3× Z₁ (return path through ground)
Cable	Yes	Z₀ varies with shielding/grounding
Transformer Δ-Y	Yes	Wye side: Z₀ = Z₁. Delta side: blocks zero-sequence (Z₀ = ∞)
Transformer Y-Y (both grounded)	Yes	Z₀ = Z₁ typically
Synchronous generator	NO — Z₁ < Z₂	Z₀ typically < Z₁ (small but nonzero)
Induction motor	NO — Z₂ ≈ locked-rotor Z (~ 1/6 of Z₁)	No path (no neutral connection in delta or ungrounded wye)

This is why induction motors contribute heavily to negative-sequence currents when faults occur — and why phase loss / single-phasing damages them quickly (the 2× counter-rotating field induces extreme rotor losses).

If You See THIS, Think THAT

If you see…	Think / use…
"Available fault current"	Symmetric RMS at the bus. Drives equipment AIC + arc flash inputs.
"AIC" or "Interrupting Rating"	kA the equipment can safely interrupt. Must equal or exceed available fault current.
"X/R ratio"	System reactance / resistance. Higher X/R = more asymmetric current. Affects equipment momentary withstand.
"Asymmetric" or "peak current"	First half-cycle. Includes DC offset. Peak ≈ 2.7× RMS sym for X/R = 30; ≈ 1.4× for X/R = 8.
%Z given for transformer	I_fault ≈ FLA / %Z (infinite primary). Real fault is somewhat less.
"MVA method" / "per-unit method"	Two equivalent ways to combine impedances and compute fault current. MVA = faster hand calc; per-unit = rigorous + multi-source.
"Motor contribution to fault"	Large motors momentarily contribute fault current (4-8× motor FLA). Add to utility contribution for MV system fault calc.
"Fault current at end of long cable"	Cable impedance reduces fault. For long branches, may be much less than panel-bus value. Use voltage-drop ohms in calc.
Generator fault contribution	Much lower than utility (subtransient ~ 6-8× generator FLA). On-genset fault current may not trip downstream OCPD designed for utility fault — coordination headache.

§14 / 39

Grounding & Bonding

NEC 250 · system vs equipment · solidly grounded · HRG · GFP

Grounding gives fault current a return path. Bonding equalizes potential between metal parts. NEC Article 250 governs both. The grounding scheme you pick — solidly grounded vs HRG vs ungrounded — has consequences for protection, arc flash, and operations.

System Grounding vs Equipment Grounding

Two completely different things, both called "grounding." Confusing them is the most common NEC 250 error.

	System Grounding	Equipment Grounding
What it grounds	The neutral of the source (transformer secondary, generator)	Metal enclosures, conduits, equipment cases
Purpose	Establishes a reference voltage; provides a low-impedance path for fault current to trip OCPD on ground fault	Bonds all metal parts together so they're at the same potential — prevents shock hazard from energized metal
NEC reference	NEC 250 Part II	NEC 250 Part VI
Conductor name	Grounding electrode conductor (GEC) — connects neutral to earth electrodes	Equipment grounding conductor (EGC) — runs with circuit conductors
Sized by	NEC Table 250.66 (size of largest service entrance conductor)	NEC Table 250.122 (size of OCPD protecting the circuit)
Where joined	Joined to EGC at the service equipment via main bonding jumper. Only ONE point.	—

Three System Grounding Schemes

Scheme	Description	Pros	Cons	Where used
Solidly Grounded	Neutral bonded directly to ground (zero impedance)	Simple. Standard equipment. Ground faults trip overcurrent immediately.	Ground fault current = high (50%+ of 3φ fault). Causes equipment damage.	Standard for nearly all commercial/industrial. Atlas DC1.
Ungrounded (delta)	No neutral connection to ground	First ground fault doesn't shut down system — service continues. Important for continuous-process plants.	Hard to detect first fault. Second fault = phase-phase fault (catastrophic). Transient overvoltage risks.	Older industrial plants; declining use.
High-Resistance Grounded (HRG)	Neutral grounded through a resistor that limits ground-fault current to ~1-10 A	Faulted system continues operating. Easy to detect ground fault (alarm + indication). Solves ungrounded problems.	Requires HRG cabinet + monitoring. Doesn't trip OCPD — must be located + cleared manually.	Industrial continuous-process (chem plants, refineries, steel mills, paper mills). Mining.
Low-Resistance Grounded	Neutral through a resistor that limits ground fault to ~100-1000 A	Limits ground-fault damage. Still trips OCPD.	Specialized; less common.	Some industrial MV applications.

Grounding Electrode System (NEC 250.50)

Every service requires a grounding electrode system. NEC 250.50 lists the acceptable types — if any are present, ALL must be bonded together. You don't choose just one.

Electrode type	Description	NEC reference
Metal underground water pipe	10+ ft in earth, must be supplemented with another electrode	250.52(A)(1)
Metal building frame	Effectively grounded — large structures	250.52(A)(2)
Concrete-encased electrode (CEE / "Ufer")	20+ ft of #4 AWG bare copper or ½" rebar in concrete footing. Modern best practice — REQUIRED in new construction.	250.52(A)(3)
Ground ring	20+ ft of #2 AWG bare copper buried 30" deep around perimeter	250.52(A)(4)
Ground rod / pipe	5/8" × 8 ft minimum, driven 8 ft deep. Single rod requires resistance ≤ 25Ω OR add second rod.	250.52(A)(5), (A)(7)
Ground plate	2 sq ft minimum, buried 30" deep	250.52(A)(7)

Equipment Ground Conductor (EGC) Sizing — NEC 250.122

EGC size is based on the OCPD protecting the circuit, not the conductor size.

OCPD rating (A)	Cu EGC	Al EGC
15	#14	#12
20	#12	#10
60	#10	#8
100	#8	#6
200	#6	#4
400	#3	#1
600	#1	2/0
800	1/0	3/0
1200	3/0	250 kcmil
2000	250 kcmil	400 kcmil

Ground Fault Protection (GFP) — NEC 230.95

For 480Y/277V services with main breaker ≥ 1000A, NEC requires Ground Fault Protection of equipment (GFPE). This is separate from GFCI for personnel and trips on ground faults below the breaker's normal trip threshold.

Aspect	NEC 230.95 GFPE	GFCI (NEC 210.8)
Purpose	Protect equipment from arcing ground faults that wouldn't trip OCPD	Protect personnel from electrocution
Trip current	1200 A maximum setting	4-6 mA (5 mA typical)
Where required	480Y/277V services ≥ 1000A main	Wet locations, kitchens, bathrooms, outdoors, etc.
Who tests	Performance test required at installation per NEC 230.95(C)	Test button monthly

Visual — Three Grounding Schemes Side by Side

Three schemes, three behaviors under ground fault. The choice depends on whether continuous operation or fast trip matters more.

Worked Example 1 — Atlas DC1 Grounding System

Example 01 · Atlas DC1 spineSolidly grounded 480Y/277V system with concrete-encased electrode + ground ring

System grounding: Each transformer secondary (TX-A, TX-B) is solidly grounded via a Main Bonding Jumper at its associated 480V switchgear. Two separately derived systems → two MBJs.
Grounding electrode system: Per NEC 250.50, all available electrodes bonded.
• Concrete-encased electrode (CEE) — 100 ft of #4 bare Cu in foundation pour
• Ground ring — 250 ft of 4/0 bare Cu around building perimeter, 30" deep
• Building steel — bonded at multiple points
All bonded together with #2/0 Cu GEC.
Ground fault protection (NEC 230.95): Each 480V SWGR main is 4000A → GFPE required. Typical setting: 1200 A pickup, 0.3 s delay. Tested per NEC 230.95(C) at commissioning.
EGC sizing: For 1200A feeder breaker → 3/0 Cu EGC per Table 250.122. For 30A branch → #10 Cu EGC.
Each PDU as separately derived system: PDU contains a 480-415Y/240V isolation transformer. The 415V side is a separately derived system → its own MBJ + GEC bonded to building grounding electrode system.
IT equipment: Server racks bonded to a "Signal Reference Grid" (SRG) — separate from but bonded to the building EGC. Per ANSI/TIA-942 (data center standard).

i

Why DCs use separately derived systems for IT loads

PDU isolation transformers create a fresh neutral and grounding point at each PDU. This isolates IT-room ground from the rest of the building — keeping ground noise out of sensitive electronics. NEC 250.30 governs the grounding requirements for these separately derived systems.

Worked Example 2 — Industrial HRG System

Example 02 · Alternate contextChemical plant 480V industrial system — High-Resistance Grounded for continuous operation

Why HRG: A chemical reactor that must not trip on a single ground fault. Sudden shutdown = product loss + safety hazard. HRG converts a ground fault from a trip event to an alarm event.
Resistor sizing: Limit ground fault current to 5 A. R = V_LN / I = 277 / 5 = 55 Ω. Continuous-rated resistor in HRG cabinet.
Detection: Resistor + voltage sensor across resistor. When ground fault occurs, voltage appears across resistor → alarm.
Operation: First ground fault = alarm. Operator dispatches maintenance to find the fault. Production continues. Second ground fault on different phase = phase-to-phase fault → trips main. Avoid this — clear the first fault promptly.
Trade-off: Lose ground-fault tripping and need active monitoring + skilled maintenance to chase faults. Gain: continuous production through faults.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · System vs equipment

Which grounding bonds the SOURCE NEUTRAL to ground?

Drill 2 · EGC sizing

200 A breaker. NEC 250.122 EGC (Cu)?

Drill 3 · MBJ count

How many Main Bonding Jumpers per service?

Drill 4 · HRG vs solidly grounded

Industrial process plant cannot tolerate trips on ground fault. Best scheme?

Drill 5 · Separately derived

Each transformer secondary in Atlas DC1 — separately derived system?

Insulation Testing — Megger

Insulation degrades over time from heat, moisture, contamination. Insulation testing applies a high DC voltage (500-5000 V) and measures leakage current → insulation resistance in megohms (MΩ). Required at commissioning and periodic maintenance.

Test	Voltage applied	What it tells you
Insulation Resistance (IR)	500-5000 V DC (matched to equipment voltage rating)	Single-point measurement. Pass/fail vs minimum acceptable.
Polarization Index (PI)	Same DC voltage	Ratio of 10-min reading / 1-min reading. PI ≥ 2 = good. < 1 = wet, contaminated.
Dielectric Absorption Ratio (DAR)	Same DC voltage	60-sec reading / 30-sec reading. ≥ 1.4 = good for thermoset insulation.
Step Voltage	Stepped (500, 1000, 2500, 5000 V)	If IR drops at higher voltage, insulation has weak spots
Breakdown Test (Hipot)	2× operating voltage + 1000 V (DC), or AC equivalent	Destructive — used for verification of new equipment only

Acceptance Criteria (rough)

For motor + transformer windings, IEEE 43 (2013) gives minimum IR (corrected to 40°C):

1 MΩ + 1 MΩ per kV of operating voltage for motors built before 1970
100 MΩ minimum for modern thermoset insulation systems
5 MΩ minimum for thermoplastic insulation systems

Atlas DC1 480V motor IR: minimum acceptable ≈ 100 MΩ. Typical reading on healthy motor: 1,000-10,000 MΩ.

Ground Resistance Testing

NEC 250.53 requires single ground rods to achieve ≤ 25 Ω resistance to earth — or add a second rod (no further test required). For substations and critical facilities, much lower resistance is sought (≤ 5 Ω, often ≤ 1 Ω).

Test method	How it works	Best for
Fall-of-Potential (3-point)	Inject current via auxiliary electrode at distance D. Measure voltage at intermediate electrode at varying positions. Resistance plateau at 62% of D = true ground resistance.	Single ground rods + small grounding systems. The classical method.
Clamp-on (induced-current)	Inductive clamp around grounded conductor. Measures resistance via induced current loop. No disconnection required.	Quick spot checks. Limited accuracy.
Slope method	Multiple fall-of-potential measurements at fractions of D. Resolves geometry of large grounding systems.	Substations and large facilities (when 62% rule fails).
4-point (soil resistivity)	Four equally-spaced electrodes (Wenner method). Calculates soil resistivity ρ in Ω·m.	Pre-construction site characterization. Drives ground design.

!

Soil resistivity is the dominant variable

Sandy/gravelly soil: ~ 1,000 Ω·m. Wet clay: ~ 30 Ω·m. Rocky: 10,000+ Ω·m. The same ground rod in different soils can give 10× different resistance. Always do 4-point soil testing on critical projects before designing the grounding system.

If You See THIS, Think THAT

If you see…	Think / use…
"System grounding"	Bonding the source neutral to ground. NEC 250 Part II.
"Equipment grounding"	Bonding metal enclosures together via EGC. NEC 250 Part VI.
"GEC" (Grounding Electrode Conductor)	From transformer/service neutral to grounding electrode system. Sized per NEC 250.66.
"EGC" (Equipment Grounding Conductor)	Runs with circuit conductors. Sized per NEC 250.122 (based on OCPD).
"Main Bonding Jumper" (MBJ)	The single bond between neutral and ground at the service equipment. Only ONE per service.
"Separately derived system"	Every transformer secondary (and generator). Has its own MBJ + GEC. NEC 250.30.
"Solidly grounded"	Standard. Neutral bonded directly to ground at the source.
"HRG" or "high-resistance grounded"	Industrial scheme that limits ground fault to ~5 A and uses alarm instead of trip.
"Ungrounded delta"	Older system. No neutral. First fault doesn't trip but creates monitoring requirement.
"GFP" or "GFPE" (NEC 230.95)	Required on 480Y services with ≥ 1000A main. Trips on arcing ground faults below normal OCPD threshold.
"GFCI" (NEC 210.8)	Personnel protection (5 mA). Required in wet/damp locations.
"CEE" or "Ufer" ground	Concrete-encased electrode. Modern best practice. NEC 250.52(A)(3).
"Ground rod ≤ 25Ω" or "two rods"	NEC 250.53(A)(2) — single ground rod must achieve ≤ 25Ω OR you add a second rod.

§15 / 39

Motors & Motor Control

Motor types · nameplate · DOL/Soft/VFD selection · MCCs

Motors are the largest single load category in industrial buildings. NEC 430 covers their branch circuits. The starter you pick — DOL, soft start, VFD — determines starting current, harmonics, and lifetime cost.

Motor Types — A Field Guide

Motor type	Description	Pros	Cons	Common use
Squirrel-cage induction	3φ AC, no slip rings, fixed-cage rotor	Cheapest, simplest, reliable, no maintenance	Hard to start large sizes (high inrush). Limited speed control without VFD.	Default — 95% of all industrial motors
Wound-rotor induction	3φ AC with slip rings + external resistance	Soft starting via external resistance. Variable torque/speed control.	Higher cost, slip rings + brushes need maintenance	Crushers, hoists, large-inertia loads (legacy)
Synchronous	3φ AC with separately excited DC field	Constant speed regardless of load. Can correct PF by adjusting excitation.	More complex (DC excitation system). More expensive.	Large pumps, compressors, paper mills, steel mills
DC motor	Brushed or brushless DC	Excellent speed control. High starting torque.	Brush wear (brushed). Cost (brushless).	Traction (cranes, EV propulsion), older industrial
Permanent-magnet AC (PMSM)	Brushless AC with permanent-magnet rotor — needs VFD	Highest efficiency. Compact. Excellent speed control.	Expensive. Requires VFD always.	HVAC fans, EV motors, modern HE pumps
Stepper	Discrete-step rotation, no feedback needed	Precise positioning without encoder	Limited torque. Inefficient at high speed.	3D printers, CNC tooling, small actuators

Nameplate Decoded

A motor nameplate has 12-15 fields. Each tells you something specific. Here's what to look for.

Nameplate field	What it means	What you do with it
HP (or kW)	Mechanical output rating	Starting point for FLA calc + load study
Volts	Utilization voltage (e.g., 460V on 480V system)	Confirms compatibility with system voltage
Amps (FLA)	Full load amperes at rated output	For overload setting; NEC sizing uses NEC Table FLC, NOT this
RPM	Full-load speed	Determines pole count: 1800 = 4-pole, 3600 = 2-pole, 1200 = 6-pole (60 Hz)
Hz	Operating frequency (60 in US)	Confirms frequency match. VFD can run at any frequency.
SF (Service Factor)	Multiplier on continuous overload capability	Common: 1.0 (no overload allowed) or 1.15 (15% momentary OK). Affects overload setting.
Code Letter	Locked-rotor kVA per HP. A=lowest, V=highest.	Determines starting current. Code F = 5.6 kVA/HP. Important for DOL starting.
Design Letter	NEMA torque/speed characteristics: A, B, C, D	Design B = standard (90% of motors). C = high starting torque. D = very high inertia loads.
Insulation Class	Max temperature rise: A=60°C, B=80°C, F=105°C, H=125°C	F or H standard for industrial.
Frame Size	NEMA frame number — physical dimensions	Determines mounting hole pattern. 56 (small), 143T-449T (medium-large).
Enclosure	ODP, TEFC, TENV, XP	ODP = open drip-proof. TEFC = totally enclosed fan-cooled (most common). XP = explosion-proof.
Efficiency	η at full load. NEMA Premium ≥ 95% for medium motors.	Required by ASHRAE 90.1. Affects energy cost over motor life.

Starting Methods — DOL vs Soft Starter vs VFD

Method	Starting current	Cost	When to use	When NOT to use
DOL (Direct-On-Line)	6-8× FLA for ~1 sec	Lowest — just a contactor + overload	Small motors (≤ 50 HP usually). When inrush is acceptable to upstream system.	Large motors where inrush stresses utility / causes voltage flicker.
Star-Delta (Y-Δ)	~33% of DOL inrush (2-3× FLA)	Medium — extra contactor	Mid-size motors with light starting load. Fans, low-inertia pumps.	High starting torque required. Variable-speed needed.
Autotransformer	Adjustable: 50-80% of DOL	Higher — autotransformer in starter	Mid-large motors needing controlled inrush.	Variable speed. Frequent starts.
Soft Starter (SCR)	Adjustable: 200-400% FLA. Ramped voltage.	Medium-high	Mid-large motors needing controlled torque ramp. Pumps preventing water hammer.	True variable speed needed (VFD instead).
VFD (Variable Frequency Drive)	Just FLA — no inrush at all	Highest — but pays back via energy savings	Modern default for any large motor or any application that benefits from variable speed.	Constant-speed simple loads where VFD cost not justified.

MCC (Motor Control Center) Anatomy

An MCC is a standalone, modular cabinet that houses all the motor starters for a process area. Each motor gets a "bucket" — a removable drawer with starter, overload, control circuits, and disconnect.

Bucket type	Contents	Use
Combination starter (NEMA size 1-5)	Disconnect (fused or unfused), contactor, overload relay, control transformer	Standard FVNR (Full-Voltage Non-Reversing) control of small/mid motors
Reversing starter	Two contactors mechanically + electrically interlocked	Conveyors, hoists, anything that needs both directions
Soft starter	SCR-based reduced-voltage starter	Mid motors needing controlled ramp
VFD	AC drive with input filter, output reactor option	Large or variable-speed motors. Atlas DC1 chillers.
Feeder bucket (no motor control)	Disconnect + breaker only	Subfeed to remote panel/MCC
Lighting xfmr / control bucket	Step-down transformer + control circuit distribution	120V control supply for MCC controls

VFD Considerations — Beyond Speed Control

Issue	Cause	Mitigation
Harmonics on input	VFD rectifier draws non-sinusoidal current → 30-40% THDi	Input line reactor (cheap, 5%), 12-pulse rectifier (better), active front end (best, expensive). See §15.
Reflected wave on motor cable	Long cable + high dV/dt = voltage doubling at motor terminals → insulation damage	Output reactor (slows dV/dt), dV/dt filter, sinewave filter for very long runs
Bearing currents	Common-mode voltage from VFD induces shaft current → bearing fluting	Insulated NDE bearing, shaft grounding ring, bearing-isolating output filter
Heat dissipation in motor	Motor running below 60Hz has reduced self-cooling fan effect	Inverter-duty motor with separately powered cooling fan, or oversized motor frame
Overspeeding the load	VFD can run motor > 60 Hz → mechanical limits exceeded	Set max output frequency in VFD parameters; verify mechanical rating

Worked Example 1 — Atlas DC1 Chiller VFD Selection

Example 01 · Atlas DC1 spineCH-1 chiller motor: 450 HP, 480V 3φ — VFD selection vs alternatives

Why VFD won here

DOL would draw 450 × 6 = 2,700 A inrush. Atlas TX-A is 2500 kVA = 3007 FLA. Inrush would cause significant voltage dip on the 480V bus during startup, with cascade impacts on adjacent UPS and IT loads. Unacceptable in a 2N data center.
Star-delta would still pull ~900 A inrush + couldn't modulate. Chiller load varies with cooling demand — fixed speed wastes energy.
Soft starter limits inrush but no speed control. Doesn't help with energy savings.
VFD wins on every count: Zero inrush. Modulates with cooling load. ~30% energy savings annually. Soft start preserves chiller bearings.

VFD spec (per cutsheet)

Parameter	Value
VFD type	6-pulse PWM, integral input reactor (5% impedance)
Rating	500 HP, 480V (oversized 1 frame for thermal margin)
AIC rating	65 kA (matches Atlas DC1 fault current)
Output reactor	3% reactance, on output (cable length ~30 ft, on the safe side)
Communication	BACnet/IP for BMS integration

Branch circuit reconciliation

From §04: chiller branch CB sized at 1,200 A (250% × 480 FLC per NEC 430.52). With VFD, this is overkill — VFD soft-starts. Industry practice with VFDs: size CB at 175-200% × FLC for tighter protection. Could use 1,000 A here; 1,200 A still fine.

Worked Example 2 — Industrial Conveyor with Soft Starter

Example 02 · Alternate context75 HP conveyor motor — needs controlled ramp to avoid product damage

Why soft starter (not VFD): Conveyor runs at fixed speed during operation. No need for variable speed. Soft starter cheaper than VFD by ~40%.
Why not DOL: Sudden start jerks product on belt. Mechanical stress on gearbox + sprockets.
Soft starter spec: 100 HP rated SCR-based unit. Adjustable ramp 5-30 sec. Bypass contactor closes after ramp completes (eliminates SCR conduction loss in run mode).
Branch circuit: 75 HP × 1.25 = MCA = 117 A → 4/0 Cu THWN-2. Branch CB sized 250% × 96 FLC = 240 A.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · DOL inrush

20 HP motor at 480V 3φ. FLA = 27 A. DOL inrush?

Drill 2 · HP → kW

100 HP → kW?

Drill 3 · Code letter F

Code F = 5.0-5.6 LR-kVA/HP. 50 HP at code F: locked rotor kVA?

Drill 4 · VFD vs soft starter

Variable speed needed AND want energy savings?

Drill 5 · Atlas chiller

Atlas DC1 CH-1 = 450 HP. Why VFD chosen over DOL?

Worked Example 3 — Motor Starting Voltage Dip + Flicker

Large DOL-started motors cause voltage dips on the bus during start. Excessive dip causes lights to flicker, contactors to drop out, and IT equipment to reset. IEEE 1453 / IEEE 141 govern acceptable flicker.

Example 03 · Industrial flicker200 HP motor at Code F, DOL-started on 480V bus fed by 750 kVA transformer (%Z = 5%)

Inputs

Motor

200 HP, 480V, NEC FLC = 240 A, Code letter F

Locked-rotor kVA per HP

5.6 (NEMA Code F)

Transformer

750 kVA, %Z = 5%, 480V secondary

Other load on bus

~ 200 kVA constant

Step-by-step

Motor locked-rotor kVA:

LRkVA = 200 × 5.6 = 1,120 kVA momentary inrush
Locked-rotor current at 480V:

I_LR = 1,120,000 / (√3 × 480) = 1,348 A (vs FLA of 240 — 5.6× as expected)
Transformer base impedance:

Z_base = V² / S = 480² / 750,000 = 0.307 Ω
Z_actual = 0.05 × 0.307 = 0.0154 Ω
Voltage drop during inrush:

V_drop = √3 × I_LR × Z = √3 × 1,348 × 0.0154 = 35.9V = 7.5% of 480V

7.5% dip on the 480V bus during the ~ 1-second start.
Acceptable? IEEE 141 (Red Book) recommends < 10% momentary dip. IEC 61000-3-3 says < 4% for sensitive loads. 7.5% is acceptable for general industrial but excessive for sensitive electronics or commercial environment.
Mitigation if needed:

• Switch to soft starter → drops to ~ 2-3% dip
• Use VFD → drops to ~ 0% dip
• Larger transformer (1500 kVA at %Z 5%) → dip drops to 3.7%
• Star-delta starting → ~ 33% of DOL inrush → dip ~ 2.5%

i

The data center connection

Atlas DC1 chose VFDs for chillers specifically to AVOID this calculation. A 450 HP DOL-started chiller on the 2,500 kVA TX would cause ~ 15% bus dip — guaranteed to drop UPS into bypass and trip downstream contactors. VFDs eliminate inrush entirely.

Synchronous Machine Theory

Synchronous machines (motors AND generators) lock to the grid frequency. Speed = 120 × f / poles regardless of load. Different physics from induction machines.

Aspect	Synchronous machine	Induction machine (for contrast)
Rotor	DC-excited field winding (separate excitation system)	Squirrel cage (no excitation needed)
Speed	Locked to line frequency (synchronous speed)	Slip below sync speed (typically 1-5% slip)
Starting	Cannot self-start (needs auxiliary or pony motor or VFD start)	Self-starting
Power factor	Adjustable via excitation — leading, unity, or lagging	Always lagging (~ 0.85 typical)
Cost	Higher (excitation system, controls)	Lower
Where used	Large pumps + compressors (≥ 1000 HP), generators, sync condensers (PFC)	Universal — 95% of motors

The V-Curve — Sync Motor PF vs Field Current

A sync motor's power factor depends on its DC field current. At one specific field current, PF = 1.0 (unity). Less excitation → motor draws lagging reactive (looks like an inductor). More excitation → motor delivers leading reactive (acts like a capacitor — a "synchronous condenser").

Plotting armature current (Y) vs field current (X) gives a V-shaped curve, one V per load level. The bottom of the V is unity PF.

Capability Curve

The capability curve plots the operating envelope of a synchronous machine in P-Q space (real vs reactive power). It's bounded by:

Boundary	What limits it
Stator (armature) current limit	Thermal limit on stator winding — defines a circle of constant kVA
Field current (rotor) thermal limit	Thermal limit on rotor winding — defines a curve in the lagging region
Stator end-iron heating	Limits leading PF operation (under-excited end of curve)
Steady-state stability limit	Theoretical maximum — practical limit is below this
Prime mover (turbine) limit	Mechanical limit on real power output (horizontal line)

Wound-Rotor + Deep-Bar Induction Motors

Type	Design	Pros	Cons	Where used
Standard squirrel cage	Cast aluminum or copper bars in rotor slots	Cheap, simple, reliable	High inrush, fixed speed	95% of all motor applications
Wound rotor	3φ winding on rotor + slip rings + external resistance	Soft starting (insert R, reduce inrush). Speed control by varying R.	Slip rings + brushes need maintenance. Higher cost.	Crushers, hoists, large-inertia loads (pre-VFD era)
Deep-bar squirrel cage	Deep, narrow rotor bars	High starting torque + low starting current (skin effect at slip frequency)	Slightly lower running efficiency	NEMA Design B (most common)
Double-cage squirrel cage	Two cages in same rotor — outer high-R for start, inner low-R for run	Best starting + best running performance	Expensive to manufacture	NEMA Design C (high starting torque)

i

Why "deep-bar" matters

At standstill (high slip, slip frequency = line frequency 60 Hz), skin effect pushes rotor current to the OUTER edge of the deep bar — high effective resistance → high starting torque, low starting current. At run speed (low slip, slip frequency < 5 Hz), no skin effect → current uses full bar cross-section → low resistance → high efficiency. The motor self-tunes between starting and running modes.

NEMA Design Letters (Squirrel Cage)

Design	Starting torque	Starting current	Slip	Use
A	Normal (~150% rated)	HIGH	Low (≤ 5%)	Rarely specified — high inrush
B (most common)	Normal	Normal (~600-650% rated)	Low (≤ 5%)	Default for general purpose — fans, pumps, compressors
C	HIGH (~200% rated)	Normal	Low (≤ 5%)	Loads with high-inertia start: conveyors, crushers, large compressors
D	VERY HIGH (~275% rated)	Normal	HIGH (5-13%)	Punch presses, oil-well pumps, anything with cyclic peak loads + flywheel

If You See THIS, Think THAT

If you see…	Think / use…
"Code letter F" or similar	Locked-rotor kVA per HP. F = 5.6. Determines starting current — important for DOL.
"Design letter B"	NEMA standard motor (90% of motors). Normal starting torque, normal slip.
"Service factor 1.15"	Allows 15% continuous overload. Affects overload setting (NEC 430.32).
"NEMA Premium efficiency"	Required for new equipment per ASHRAE 90.1 + DOE rules. ≥ 95% for medium motors.
"VFD-driven motor"	Sized smaller branch CB (175-200% vs 250%). Worry about harmonics, reflected wave, bearing currents.
"Inverter-duty motor"	Designed for VFD use. Better insulation, often separately cooled.
"TEFC enclosure"	Totally enclosed fan-cooled. Most common industrial. Good for dirty/dusty.
"XP enclosure"	Explosion-proof. Required for Class I Div 1 hazardous locations (§21).
Large motor in 2N facility	Use VFD or soft starter to avoid inrush impact on critical loads.
"6-pulse" vs "12-pulse" VFD	6-pulse cheap, 30%+ THDi. 12-pulse cleaner, more expensive. Active front end = cleanest.
Reflected wave / bearing currents	VFD on motor with long cables. Add output reactor or dV/dt filter.

§16 / 39

Power Quality

Harmonics · IEEE 519 · PFC · capacitor switching · K-factor

Modern loads (servers, VFDs, LED drivers) are nonlinear — they pull current in pulses, generating harmonics. Harmonics cause neutral overheating, transformer derating, capacitor failures, and revenue meter errors.

What Is Power Quality?

An ideal power system delivers a perfect 60 Hz sinusoidal voltage at exactly the rated magnitude. Real systems deviate. Power quality covers all the deviations: harmonics, voltage sag/swell, flicker, transients, imbalance, frequency drift.

Disturbance	Cause	Effect	Mitigation
Harmonics	Nonlinear loads (rectifiers, VFDs, LEDs, servers)	Neutral overheating, transformer derating, capacitor failure	K-factor xfmr, harmonic filter, isolation, 12-pulse drives
Voltage sag (dip)	Large motor start, fault clearing on adjacent feeder	Sensitive electronics drop out, contactor chatter	UPS, dynamic voltage restorer, ride-through circuits
Voltage swell	Capacitor switching, load drop	Insulation stress, electronics damage	Surge protection (§24), tighter voltage regulation
Transients (impulses)	Lightning, switching, capacitor energization	Equipment damage, electronics failure	SPDs (Type 1/2/3), good grounding
Flicker	Repetitive load fluctuations (welders, arc furnaces, motors)	Visible light flicker, occupant discomfort	Static var compensator (SVC), STATCOM, larger transformer
Imbalance	Uneven phase loading (1φ loads on 3φ system)	Motor derating (NEMA 1% rule), neutral overload	Phase balancing in panel design (§05)
Frequency deviation	Generator islanded operation, grid disturbance	Motor speed/torque variation, sensitive equipment dropout	UPS isolation, generator governor tuning

Harmonics — The Modern Power Quality Issue

Most modern loads (servers, VFDs, LED drivers, EV chargers) are nonlinear — they pull current in pulses, not smooth sinusoids. The pulses decompose into a fundamental (60 Hz) plus harmonic frequencies (5th = 300 Hz, 7th = 420 Hz, 11th, 13th, etc.).

Harmonic source	Dominant harmonics	Typical THDi
6-pulse rectifier (typical VFD input)	5th, 7th, 11th, 13th	30-40%
12-pulse rectifier (better VFD)	11th, 13th, 23rd, 25th	10-15%
18-pulse rectifier (best passive)	17th, 19th, 35th, 37th	5-8%
Active Front End (AFE) drive	switching frequency artifacts only	< 5%
Single-phase server PSU (modern PFC)	3rd, 5th, 7th	5-15%
Single-phase server PSU (older, no PFC)	3rd dominant — large neutral current	30-80%
LED driver (cheap)	3rd, 5th, 7th	10-30%
EV charger Level 2	5th, 7th	5-10%
EV charger DCFC (Level 3)	5th, 7th, 11th, 13th — depends on rectifier	5-15% (with filter), 25-30% (without)

THD vs TDD — Same Distortion, Different Reference

Total Harmonic Distortion (THD) — % of fundamental

THD = √(I₂² + I₃² + I₄² + ...) / I₁ × 100%

I₁ = fundamental (60 Hz). I_n = nth harmonic. Used by power quality monitors and IEEE standards for current.

Total Demand Distortion (TDD) — % of maximum demand current

TDD = √(I₂² + I₃² + I₄² + ...) / I_L × 100%

I_L = maximum demand current at the point of common coupling (PCC). IEEE 519 limits use TDD, not THD.

IEEE 519-2022 Limits at the PCC

IEEE 519 sets harmonic limits at the Point of Common Coupling (PCC) — the boundary between user and utility. Limits depend on the short-circuit ratio (SCR = I_SC / I_L): stronger source = more harmonic-tolerant.

SCR (I_SC/I_L)	Individual TDD limit	Total TDD limit
< 20	4%	5%
20 - 50	7%	8%
50 - 100	10%	12%
100 - 1000	12%	15%
> 1000	15%	20%

Power Factor Correction

Inductive loads (motors, transformers) cause current to lag voltage → lower PF → utility bills demand penalty. Capacitor banks supply reactive power locally to bring PF closer to 1.0.

kVAR needed to correct from PF₁ to PF₂

kVAR = kW × (tan(arccos PF₁) − tan(arccos PF₂))

Example: 1000 kW load at PF=0.80 (lag), correct to 0.95: kVAR = 1000 × (0.75 − 0.329) = 421 kVAR capacitor bank

Capacitor Switching Transients

Issue	Cause	Mitigation
Energization transient	Closing into uncharged cap → 2× nominal voltage spike	Pre-insertion resistor in capacitor switch, synchronous switching
Voltage magnification	Cap energization at primary causes higher voltage at customer secondary if customer has caps too — resonance	Coordinate utility + customer cap switching, avoid same kVAR ratings
Restrike on opening	Cap voltage tries to reverse during open → arc restrike → repeated transients	Vacuum or SF6 caps switches with restrike-resistant designs
Resonance with system harmonics	Cap + system inductance form parallel resonant circuit at a harmonic frequency → magnification	Detuning reactors (5% L in series with cap), shifts resonance below dominant harmonic

!

The PFC + harmonics catastrophe

A facility with significant 5th harmonic (300 Hz) and a power factor correction cap bank can hit parallel resonance — caps + system L resonate at the 5th. Result: 5th harmonic current magnification 5-20×, capacitor failure, transformer overheating, equipment damage. Always study harmonics before installing PFC caps in modern (server-heavy or VFD-heavy) facilities.

Worked Example 1 — Atlas DC1 Server Harmonics

Example 01 · Atlas DC1 spine2.5 MW IT load — characterizing harmonic environment + mitigation strategy

Source: 2.5 MW of modern servers with PFC PSUs. THDi at the rack ~ 8-12%. THDi at PDU level (after diversity averaging across thousands of PSUs) ~ 6-8%.
At the UPS output: Servers' harmonics flow back through the UPS → reflected on UPS DC bus → minor flow back upstream of UPS. UPS isolation reduces upstream THD significantly.
At the 480V SWGR (PCC): Combination of UPS-fed loads + chiller VFDs. Chiller VFDs are 6-pulse with 5% input reactor → individual TDD ~15%. Total TDD at SWGR ~ 6-9% (after diversity).
IEEE 519 check: Atlas DC1 SCR at PCC ≈ 50,300 / 3,007 = ~17 (fault current / demand current). At SCR = 17, TDD limit = 5%. Atlas DC1 ~ 7% — over limit.
Mitigation: Upgrade chiller VFDs to 12-pulse (drops their TDD to ~10%) → facility TDD drops to ~ 4%. Within 5% limit. ✓
K-factor transformer at PDU: PDU isolation transformers spec'd as K-13 — designed to handle harmonic neutral currents from 1φ servers without overheating.

i

Why DCs avoid PFC capacitors

Server PSUs already have integral PFC — server load presents PF ≈ 0.95 to utility naturally. No external cap bank needed. Adding caps risks resonance with the strong 5th + 7th harmonic content — a recipe for destroyed capacitors and fires. Atlas DC1 has NO PFC cap banks.

Worked Example 2 — Industrial Motor Plant PFC

Example 02 · Alternate contextManufacturing plant — 600 kW motor load at PF = 0.78 — utility penalizes below 0.90

Penalty avoidance: Utility charges $5/kVAR over the 0.90 PF threshold per month. Worth correcting.
Calc: kVAR_corr = 600 × (tan(arccos 0.78) − tan(arccos 0.95)) = 600 × (0.802 − 0.329) = 284 kVAR
Bank size: Round up to standard 300 kVAR. Multi-step (50 + 100 + 150 kVAR) for variable load.
Harmonic check: Plant has 3 VFDs (75 HP, 100 HP, 150 HP). Harmonics present. Resonance risk if cap bank not detuned.
Solution: Detuned PFC bank with 7% reactor. Shifts parallel resonance to ~3.8th harmonic — well below dominant 5th and 7th. Safe.
Result: PF rises from 0.78 to 0.95. Penalty eliminated. Payback ~ 14 months.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · THD vs TDD

What's the difference?

Drill 2 · Triplens

Which harmonics add in the neutral?

Drill 3 · PFC kVAR

300 kW load at PF 0.80 lag → correct to 0.95. kVAR needed?

Drill 4 · K-factor xfmr

Server farm has 30% THDi. Transformer rating?

Drill 5 · Cap + harmonics

Adding PFC caps to a plant with VFDs — risk?

Reactive Compensation Beyond Capacitors

For dynamic, fast, or large-scale reactive support, simple capacitor banks aren't enough. Three competing technologies — each with its own trade-offs.

Technology	How it works	Response time	Reactive range	Cost	Where used
Fixed cap bank	Switched in/out by contactors	Cycles to seconds	Discrete steps	Lowest	Steady industrial loads, light-duty PFC
Switched cap bank (auto)	Multiple stages switched by PF controller	Seconds	Stepwise	Low-medium	Variable industrial loads
SVC (Static Var Compensator)	Thyristor-controlled reactor + switched cap banks	1-2 cycles (~ 33 ms)	Continuous over wide range	Medium	Arc furnaces, light flicker mitigation, voltage control on transmission
STATCOM (Static Synchronous Compensator)	VSC (voltage source converter) + DC link cap; behaves like adjustable AC source	1/4 cycle (~ 4 ms)	Continuous over full range, including DURING faults	High	Severe disturbance support, wind farms, HVDC, modern utility
Synchronous condenser	Synchronous motor with no shaft load; absorbs/delivers reactive via field excitation	10-30 cycles (slow)	Continuous	Highest (mechanical machine + foundation)	Inertia + reactive support at large substations; ride-through enhancement

SVC — How Thyristor Control Works

An SVC pairs a thyristor-controlled reactor (TCR — variable inductive reactance via firing angle) with thyristor-switched capacitor banks (TSC — discrete capacitive blocks). By varying TCR firing angle and switching TSC blocks, the net reactive power output can be smoothly varied from full inductive to full capacitive.

STATCOM — Why Modern Grids Prefer It

A STATCOM is essentially a large IGBT-based inverter connected to the grid via a step-up transformer. It synthesizes a sinusoidal output voltage with controllable magnitude and phase. By adjusting the magnitude relative to the grid voltage, it absorbs (Vstatcom < Vgrid) or delivers (Vstatcom > Vgrid) reactive power.

i

STATCOM's killer feature: works during voltage sags

Capacitor banks deliver reactive power proportional to V² — when grid voltage sags during a fault, cap output crashes. SVCs have similar issue. STATCOM output is independent of grid voltage — it can deliver full reactive support DURING the fault when it's needed most. This is why wind farms (which must ride through grid faults per IEEE 1547) use STATCOM-based reactive support.

Synchronous Condenser — The Comeback

A sync condenser is a synchronous motor running with no shaft load. By varying its DC field excitation, it can absorb or deliver reactive power to the grid. Slow response (mechanical inertia), but offers something nothing else does: real spinning inertia. As renewable inverter-based generation displaces synchronous generators, grid inertia drops — sync condensers are being installed at major substations to restore inertia + provide ride-through.

If You See THIS, Think THAT

If you see…	Think / use…
"THD" specification	Total Harmonic Distortion vs fundamental. Used for voltage limits and current at the load.
"TDD" or IEEE 519	Total Demand Distortion vs maximum demand. IEEE 519 limits at PCC.
"6-pulse VFD"	Default cheap drive. ~30-40% THDi. Add 5% input reactor → ~25%.
"12-pulse" or "active front end"	Cleaner drive. 12-pulse ~ 10-15% THDi. AFE ~ < 5%.
"K-factor transformer" (K-4, K-13)	Designed for harmonic loads. Larger neutral, 60 Hz–rated for harmonic heating. Used in DCs.
"Power factor correction" / cap bank	Add capacitive kVAR to offset inductive load. Watch for resonance with harmonics.
"Detuned" PFC bank	Has a reactor in series with caps to shift resonance away from harmonics. Required in modern plants with VFDs.
"Active filter" (active harmonic filter)	Real-time injects opposite-phase harmonics. Most flexible mitigation. Expensive.
"Voltage flicker"	Repetitive load swings. SVC, STATCOM, or larger source impedance.
"Sag" / "dip"	Brief voltage drop. UPS provides ride-through.
"Triplens" or "third harmonic in neutral"	3rd, 9th, 15th harmonics add in neutral instead of cancel. Can be 173% of phase current. Always size neutral 200% in pure 1φ-3W server farms.

§17 / 39

Load Flow Analysis

Bus voltages · branch currents · radial vs looped · software tools

Load flow tells you the voltage at every bus, the current in every feeder, and where power factor sags. Done by hand for small systems; done with software (SKM, ETAP, EasyPower) for everything else.

What Load Flow Analysis Computes

Load flow (a.k.a. power flow) is the steady-state solution of the power system: voltage at every bus, current in every feeder, real and reactive power flow at every branch. Without it, you're guessing at voltage drop and PF on complex systems.

Output	What you do with it	Section reference
Bus voltages (magnitude + angle)	Verify each load receives within tolerance (±5% per ANSI C84.1)	§01, §06
Branch currents	Confirm wires not overloaded; check transformer loading	§06, §09
Real + reactive power at each node	Identify where reactive power is generated/consumed; PFC placement	§15
Transformer tap recommendations	Adjust no-load taps to optimize voltage profile	§09
Generator dispatch (if multiple sources)	Determine which gen carries which load	§19
Loss analysis	Find inefficient feeders; size correction	—

Radial vs Looped vs Networked

Topology	Description	Hand calc?	Typical use
Radial	Single source feeding tree of loads. No closed loops.	Yes — work from source to ends	99% of commercial / industrial buildings, residential
Looped	Two sources or feeders meet, with a normally-open tie	Possible but tedious — two cases (each tie position)	Critical commercial (hospitals, data centers); urban distribution
Networked	Multiple sources, multiple paths, any load can be supplied through several routes	Software only — Newton-Raphson or similar iterative solver	Utility transmission, downtown urban (network protector grids)

Hand Calc — 3-Bus Radial

For radial systems, work from source to ends. At each bus, sum the downstream loads, apply the upstream impedance, calculate voltage drop, repeat.

₂? L2 100 kVA PF=0.9 F2: 4/0, 100ft R = 0.012 Ω Bus 3 — V₃? L3 75 kVA PF=0.95

Three-bus radial example — solve for V at Bus 2 and Bus 3

Solution

Currents at each bus.
I₂ (load 2) = 100 × 1000 / (√3 × 480) = 120.3 A at PF 0.9 lag
I₃ (load 3) = 75 × 1000 / (√3 × 480) = 90.2 A at PF 0.95 lag
Current in F1 = sum of downstream loads.
I_F1 = I₂ + I₃ = 120.3 + 90.2 = 210.5 A (assuming similar PFs)
Voltage drop on F1.
VD_F1 = √3 × I × R = √3 × 210.5 × 0.011 = 4.0 V → V₂ = 480 − 4 = 476 V (0.83% drop) ✓
Current in F2 = I₃ only.
VD_F2 = √3 × 90.2 × 0.012 = 1.9 V → V₃ = 476 − 1.9 = 474 V (1.25% total drop) ✓
Total voltage drop budget: ≤ 5%. Atlas system has plenty of margin.

When You Need Software

Hand calc works for small radial. For real systems, use load flow software:

Software	Use case	Note
SKM PowerTools	Industry standard for industrial / commercial	Steep learning curve. Comprehensive.
ETAP	Industrial focus. Strong for arc flash + protection coordination integration	Most popular for power plant + petrochemical work
EasyPower	User-friendly. Good for new engineers.	Excellent integration of load flow + arc flash + coordination
PowerWorld	Transmission system focus	Used by utilities + ISOs
PSS/E or PSCAD	Utility / transmission planning + transient analysis	Specialized

Worked Example 1 — Atlas DC1 Voltage Profile

Example 01 · Atlas DC1 spineVoltage at every bus from utility through to rack PDU

Bus	Voltage	Drop from upstream	%VD running total
Utility 12.47 kV (PCC)	12.47 kV	—	0%
TX-A primary	12.47 kV	negligible (short MV cable)	0%
TX-A secondary (480V SWGR-A)	478 V	5.75% × loading × cos(impedance angle) = ~0.4%	~0.4%
UPS-A1 input	477 V	0.6% (250 ft feeder)	~1.0%
UPS-A1 output (regulated)	480 V	UPS regulates to setpoint — eliminates upstream variation	0% (re-referenced)
PDU-A1 input (480V)	477 V	0.6% (250 ft from UPS)	~0.6%
PDU-A1 output (415V at xfmr secondary)	413 V	3.5% × loading at PDU xfmr ~ 0.5%	~1.1% from UPS
RPP-A1-1 (415V)	411 V	0.6% (50 ft from PDU)	~1.7%
Rack PDU strip (240V phase-neutral)	237 V	0.4% (10 ft branch)	~2.1%

i

UPS as a voltage reset

The UPS is more than backup power — it actively regulates output voltage regardless of input. Upstream voltage variations (utility sag, transformer drops) don't propagate downstream. Total VD from UPS to rack: ~2.1%, well within IT equipment tolerance (±10% typical).

Worked Example 2 — Apartment Building VD Across Service

Example 02 · Alternate scale50-unit apartment · 980 A demand · 80 ft service feeder · check VD at panel

Service: 1200 A breaker, 3 sets of 750 kcmil Cu, 80 ft.
VD on service feeder (from §06): 0.95 V at 980 A. = 0.46% VD
VD on per-unit feeder (200 A panel, 50 ft, #2/0 Cu):
VD = 2 × 100 (1φ) × 0.13 × 50 / 1000 = 1.3 V on 240V = 0.54%
VD on branch circuit (worst — 30A range, 50 ft):
VD = 2 × 30 × 0.78 (#8) × 50 / 1000 = 2.34 V on 240V = 0.98%
Total worst-case VD (utility transformer secondary to range outlet): 0.46 + 0.54 + 0.98 = ~2.0%. Well within 5% NEC recommendation.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · VD on feeder

200 ft of #2 Cu (R = 0.20 Ω/kft), 100 A at 480V 3φ. %VD?

Drill 2 · Radial vs looped

Hand-calc possible for which topology?

Drill 3 · Voltage profile

Atlas DC1 utility 12.47kV → server. How many transformations?

Drill 4 · PCC

What is the Point of Common Coupling?

Drill 5 · UPS as voltage reset

Voltage at UPS output regardless of input variation?

Voltage Regulation — The Theory

Voltage drop and voltage regulation sound similar but mean different things in formal practice.

	Voltage Drop (%VD)	Voltage Regulation (%VR)
Definition	(V_source − V_load) / V_nominal × 100	(V_no-load − V_full-load) / V_full-load × 100
Reference	Nominal voltage	Load-side full-load voltage
Used for	Conductor sizing	Transformer + generator performance
Typical limit	≤ 3% feeder, ≤ 5% combined (NEC 215.2 IN)	≤ 3-5% for most transformers (depends on application)

Two-Bus Power Flow Equations

For a transmission line connecting two buses (sending S, receiving R) with line impedance Z = R + jX and angle δ between bus voltages:

Real power transferred (lossless line approximation)

P = (|V_S| × |V_R| / X) × sin δ

δ = phase angle between sending and receiving voltages. Maximum theoretical transfer at δ = 90° (steady-state stability limit).

Reactive power received

Q_R = (|V_R|/X) × (|V_S| × cos δ − |V_R|)

Reactive power flow depends on voltage MAGNITUDE difference. Real power flow depends on voltage ANGLE difference. This is the fundamental decoupling that makes power flow analysis tractable.

i

The PV-PQ decoupling — why power flow is solvable

Real power transfer is governed by voltage ANGLE differences (δ). Reactive power transfer is governed by voltage MAGNITUDE differences (|V|). For typical lines (X >> R), they're nearly independent. This decoupling is why we can solve P and Q separately in iterative load flow algorithms (Newton-Raphson, fast-decoupled).

Surge Impedance Loading (SIL) — for transmission

When real power transfer = SIL = V_line² / Z_c (where Z_c is line surge impedance), the line has zero net reactive power along its length. Above SIL → line absorbs reactive (looks inductive). Below SIL → line delivers reactive (looks capacitive). This concept governs reactive compensation strategy on transmission systems.

Steady-State Stability Limit

From the two-bus equation, P_max = |V_S||V_R|/X at δ = 90°. Beyond 90°, the system becomes unstable (small perturbation leads to larger perturbation). Practical operating limits keep δ < 35-40° for adequate stability margin.

If You See THIS, Think THAT

If you see…	Think / use…
"Load flow analysis"	Steady-state V, I, P, Q at every node. Software for complex; hand calc for radial.
"Voltage profile"	V vs distance plot. Tells you where to add tap adjustments or upsize feeders.
"Voltage regulation"	(V_NL − V_FL) / V_FL × 100. NEC informational note ≤ 5% total.
"Reactive power flow" / VARs	Inductive loads sink VARs; capacitors source them. Flows from generator/cap → load.
"Looped" or "networked" system	Software required. Hand calc impractical.
"Radial system"	Hand calc OK. Work source-to-load.
"PCC" (Point of Common Coupling)	Boundary between user + utility. IEEE 519 limits apply here.
"SKM/ETAP/EasyPower"	Power system software. SKM = legacy industrial; ETAP = industrial focus; EasyPower = user-friendly.
UPS in the system	Voltage reset point. Upstream variation doesn't propagate downstream.

§18 / 39

Protection & Relaying

ANSI device numbers · 51/50/87/27/49 · differential · numerical relays

Breakers are dumb — they trip when current exceeds a setting. Relays are smart — they decide WHEN and WHY. Every protection device has an ANSI device number. Coordination studies plot the curves and verify selectivity.

ANSI/IEEE Device Numbers

Every protective device has a number. The IEEE C37.2 standard assigns 1-99 (some up to 999) to specific functions. Memorize the dozen most-used; the rest are looked up.

Device #	Function	Where used
21	Distance relay	Transmission line protection
25	Synchronism check	Generator paralleling, ATS closed-transition
27	Undervoltage	Motor protection, generator dropout
32	Reverse power (directional power)	Generator protection (motoring), prevent backfeed
37	Undercurrent	Motor loss-of-load protection (e.g., loss of cooling)
40	Loss of field (excitation)	Synchronous motor / generator protection
46	Negative sequence (current unbalance)	Motor protection — phase loss, unbalanced load
47	Phase sequence / phase reversal	Verify correct phase rotation on incoming feed
49	Thermal overload (machine)	Motor + transformer protection
50	Instantaneous overcurrent	Universal — fastest trip on high faults
51	Time-overcurrent (inverse-time)	Universal — coordinated overcurrent
50G / 51G	Ground fault (50 = inst, 51 = time)	Detects ground faults; required by NEC 230.95 for large 480V services
50N / 51N	Residual neutral overcurrent	Ground fault on Y-grounded systems
59	Overvoltage	Generator protection, capacitor protection
67	Directional overcurrent	Looped systems where fault current can flow either direction
79	Auto-reclose	Distribution feeder breakers; reclose after temporary fault
81	Frequency (under or over)	Generator protection, load shedding, intentional islanding
87	Differential	Transformer (87T), bus (87B), motor (87M), generator (87G) — fastest, most selective protection
87L	Line differential	Transmission line — pilot protection

Inverse-Time vs Definite-Time vs Instantaneous Tripping

Tripping characteristic	Description	Where used
Instantaneous (50)	No intentional delay — trip in < 1 cycle when current exceeds setpoint	High-fault region — clears bolted faults fastest
Definite-time (51 with definite-time setting)	Fixed delay regardless of current magnitude (after pickup)	Backup protection — coordinates above downstream device's clearing time
Inverse-time (51)	Higher current → faster trip. IEC and IEEE curves: standard inverse, very inverse, extremely inverse	Universal time-overcurrent. Coordinates naturally with downstream OCPDs at all current levels
Pickup current	The current threshold that "starts" the timing element	Set above maximum normal load current with margin
Time dial	Multiplier on the curve — shifts curve up/down	Coordinated with downstream; lower TD = faster operation

Differential Protection (87)

Differential measures current entering a zone vs current leaving. If they don't match, current is going somewhere it shouldn't — internal fault. Trips immediately. The fastest protection available, with no coordination delay needed because it only operates on faults INSIDE its protection zone.

Differential type	Protected zone	Operation
87T Transformer differential	Inside the transformer windings	CTs on primary + secondary. Compensates for turns ratio + winding configuration. Trips on any internal fault.
87B Bus differential	Inside the switchgear bus	CTs on every bus connection. Sum should be zero. Trips on any bus fault — clears in < 1 cycle, prevents catastrophic arc flash.
87M Motor differential	Inside the motor windings	CTs on phase + neutral connections. Detects winding-to-winding fault.
87G Generator differential	Inside generator windings	Same principle — most generators have 87G as primary protection.
87L Line differential (pilot)	Transmission line	Communication channel between line ends compares currents. Telecomm-dependent.

Modern Numerical Relays

Old-school electromechanical relays (cup-and-disk) are being replaced everywhere by numerical relays — microprocessor-based devices that combine many ANSI functions in one box, with communication, event logging, and remote access.

Manufacturer	Common product line	Notes
Schweitzer Engineering Labs (SEL)	SEL-351, 387, 411, 421, 487	Industry leader. Strong cybersecurity. Engineering-friendly programming.
GE / Multilin	F60, F35, MIF II, T60	Strong utility presence. UR family.
ABB	Relion 615, 620, 630, 670 series	European-strong; 60870-5-103/104 native.
Siemens	SIPROTEC 4 / 5	European-strong; integrated DIGSI software.
Eaton	EDR 5000, MP-3000, MP-4000	Industrial focus.

Coordination Study — The Deliverable

A coordination study plots every TCC for every OCPD on a single log-log chart, with the available fault current marked. The result: visual confirmation that for any fault, only the closest device opens.

Component	What's shown
Source impedance line	Available fault current at each bus
OCPD curves	Each device's TCC at its protected location
Cable damage curve	Conductor I²t damage threshold (NEC 110.10)
Transformer damage curve	Per IEEE C57.109 / ANSI
Motor inrush region	For motor branches, plot inrush curve to ensure CB doesn't trip on starting
Selectivity bands	Time gap between upstream and downstream curves (≥ 0.3 sec typical for fuses, ≥ 0.4 sec for CBs)

Worked Example 1 — Atlas DC1 MV Switchgear Protection Scheme

Example 01 · Atlas DC1 spine12.47 kV switchgear protecting TX-A and TX-B + utility incoming

Protection functions on each device

Position	Protection (ANSI #s)	Why
Utility incoming CB (12.47 kV)	50, 51, 50G, 51G, 27, 59, 81	Standard incoming protection: overcurrent, ground, voltage, frequency
TX-A primary CB (12.47 kV)	87T (with TX-A secondary CT input), 50, 51, 50G, 51G, 26 (sudden gas pressure)	Differential primary protection of TX-A — trips on any internal fault. Backup overcurrent.
TX-A secondary CB (480V)	50, 51, 50G, 51G, GFP per NEC 230.95	Backup feeder protection for 480V SWGR
Bus differential 87B	One zone per side (A bus and B bus)	Clears bus fault in < 1 cycle — minimizes arc flash

Why each device matters

87T transformer differential: A winding-to-winding fault inside TX-A would draw fault current from utility but the fault location is inside the transformer enclosure. Without 87T, only the slower 51 element trips → significant transformer damage. With 87T, trip in 1-2 cycles.
87B bus differential: A fault on the 480V bus (e.g., insulation failure from a falling tool) would otherwise wait for transformer 51 to time out (~ 100 ms+). At 50 kA fault current, that's massive incident energy. 87B clears in 4 cycles → 90% reduction in incident energy.
50G/51G ground fault: Detects ground faults on solidly-grounded system before they escalate to phase-phase faults. Sensitivity: 100-1200 A typical setting.

i

Why DCs invest heavily in protection

A typical commercial building's MV switchgear has 51 and 50G — that's it. Atlas DC1 adds 87T, 87B, 27, 59, 81 because every cycle of clearing time matters: less arc flash, less downtime, less chance of cascading failure into the IT load. The cost is justified by the value at stake.

Worked Example 2 — Industrial Motor Protection (49, 51, 46, 27)

Example 02 · Alternate context500 HP induction motor on 480V — protected by integrated motor relay

ANSI #	Function	Setting	Why
49	Thermal overload	Per NEC 430.32 — 115% of FLA for SF=1.0; 125% for SF=1.15	Protect motor windings from thermal damage
50	Instantaneous OC	~ 130% of locked-rotor	Bolted fault on motor leads
51	Time-OC	120-130% FLA pickup, time dial coordinates with upstream	Backup to thermal overload
46	Negative sequence	Pickup at ~ 5% I₂/I₁	Detects phase loss and unbalance — both very damaging to induction motors
27	Undervoltage	~ 80% of nominal	Drop motor on sustained undervoltage to prevent stall and overheating
37	Undercurrent	Custom per application	Optional — detect loss of load (broken pump shaft, etc.)

One numerical relay (e.g., SEL-710) provides all of these functions plus event recording and Modbus communication. Old electromechanical equivalent would be 4-6 separate panels.

Coordination Plot Example — Atlas DC1 MV Protection

The TCC plot for the MV switchgear protection chain: utility 51 → TX-A primary 51 + 87T → 480V SWGR-A 51 + 87B.

Differential (87B, 87T) clears INSIDE its zone faster than any 51 element — sub-cycle vs hundreds of milliseconds. Critical for arc flash.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Top ANSI numbers

What are 50, 51, 87?

Drill 2 · 87T speed

Why is 87T (transformer differential) faster than 51?

Drill 3 · Pickup vs time dial

Two settings on a 51 element?

Drill 4 · Negative sequence (46)

What does 46 detect?

Drill 5 · Atlas MV

What protection does Atlas DC1's MV switchgear use?

Instrument Transformers — CT + PT

Protective relays and meters can't measure thousands of amps or thousands of volts directly. CTs and PTs step these down to safe, standardized values (5 A and 120 V respectively).

Current Transformers (CTs)

Parameter	Description
Ratio	Primary:secondary, e.g., 1200:5 means 1200 A primary → 5 A secondary at full load
Burden	Load on the secondary (relays + wiring + meters). Specified in VA. Higher burden = more saturation risk.
Accuracy class	For metering: 0.3, 0.6, 1.2 (% error at rated burden). For relaying: C100, C200, C400, C800 (relaying class — voltage at saturation).
Polarity	Marked terminals (H1 + X1). Critical for differential protection — wrong polarity = immediate trip on energization.
Saturation	At very high primary current (faults), CT iron core saturates → secondary output stops following primary. Causes incorrect relay operation. Sized to avoid saturation at maximum fault current.
Open secondary danger	Never open a CT secondary while energized! With no burden, voltage rises to thousands of volts → arcing + insulation failure + lethal. Always short-circuit before disconnecting.

Potential Transformers (PTs / VTs)

Parameter	Description
Ratio	Primary:secondary, e.g., 14400:120 means 14.4 kV primary → 120 V secondary
Burden	Same concept as CT but secondary is voltage-limited not current-limited
Accuracy class	0.3, 0.6, 1.2 metering. Various relaying classes.
Connection types	Wye-wye (most common), open-delta, V-V (used when delta primary system has no available neutral)
Capacitive Voltage Transformers (CVTs)	Used at very high voltages (≥ 138 kV) — capacitive divider + tuning circuit. Cheaper than full magnetic PT.

Ladder Logic — Relay Programming Basics

Ladder logic is a graphical programming language designed to mimic the wiring diagrams of relay-based control panels. PLCs use it; modern protective relays often use a similar logic syntax.

Symbol	Meaning
`--\| \|--`	Normally open contact (input). True when input is energized.
`--\|/\|--`	Normally closed contact (input). True when input is NOT energized.
`--( )--`	Output coil. Energized when the rung's logic is true.
`--(L)--`	Latching output coil. Stays energized after one true cycle.
`--(U)--`	Unlatching coil. Resets a latched output.
Rungs in series	AND logic — all conditions must be true
Rungs in parallel	OR logic — any condition true energizes output

Boolean Algebra — The Math Behind Ladder

Operation	Symbol	Truth	Ladder equivalent
AND	· or &	1·1 = 1; else 0	Series contacts
OR	+ or \|	0+0 = 0; else 1	Parallel contacts
NOT	Bar over var	NOT(0) = 1; NOT(1) = 0	Normally-closed contact
NAND	NOT(AND)	NOT(1·1) = 0; else 1	Series of NC contacts
NOR	NOT(OR)	NOT(0+0) = 1; else 0	Parallel of NC contacts
XOR	⊕	1 if exactly one input is 1	(A·NOT B) + (NOT A·B)

If You See THIS, Think THAT

If you see…	Think / use…
"51"	Time-overcurrent. Universal coordinated protection.
"50"	Instantaneous OC. Fastest trip on high fault.
"87T"	Transformer differential. Internal-fault protection. Sub-cycle clearing.
"87B"	Bus differential. Critical for arc flash reduction.
"50G", "51G"	Ground fault elements. NEC 230.95 requires for 480V services ≥ 1000A.
"49" thermal overload	Motor protection. NEC 430.32 sets the limits.
"46" negative sequence	Detects phase loss / unbalance on motors. Very valuable — prevents motor damage from single-phasing.
"27" undervoltage	Motor protection (drop on UV) or generator protection.
"81" frequency	Generator protection or load shedding logic.
"25" synchronism check	ATS closed-transition. Generator paralleling.
SEL-351 / SEL-787 / SEL-787-3	Schweitzer relays — feeder, transformer, multi-phase. Industry standard for new installations.
"Coordination study"	Plot of all TCCs. Verify selectivity at all fault levels.
"Pickup" + "time dial"	The two settings on every 51 element.

§19 / 39

Arc Flash

IEEE 1584 · NFPA 70E · NEC 110.16 labels · PPE categories · mitigation

An arc flash is an explosion of plasma at fault — incident energies of 8 cal/cm² can cause 3rd-degree burns, 40 cal/cm² is lethal. IEEE 1584-2018 calculates the energy at every bus; NFPA 70E governs PPE; NEC 110.16 requires the labels.

What Is Arc Flash?

An arc flash is a plasma explosion at electrical fault — temperatures exceed 19,000°C, pressure waves up to 720 mph, intense UV/IR radiation. NEC 110.16 requires labels at every panel; NFPA 70E governs how workers approach energized equipment; IEEE 1584 calculates the energy.

Hazard	Source	Effect
Thermal (incident energy)	Plasma radiation + heated air	3rd-degree burns at > 1.2 cal/cm²
Pressure wave	Air superheated to plasma — explodes outward	Concussion, blown out enclosure, blunt-force injury
Molten metal projectiles	Vaporized + recondensed copper, aluminum	Penetrating burns, eye damage
UV / IR radiation	Plasma emission	Eye damage (arc-eye), accelerated burns
Toxic gases	Vaporized insulation	Inhalation injury
Acoustic shock	Sub-millisecond pressure pulse	Eardrum rupture, hearing damage

IEEE 1584-2018 — The Calculation Standard

IEEE 1584 publishes an empirical model for arc flash incident energy. Inputs go in, energy at working distance comes out.

Input	Description	Source
Bolted fault current (kA)	3-phase fault current at the location	Short-circuit study (§12)
Trip time (cycles or sec)	How long until upstream OCPD clears the fault	Coordination study (§11) — read from TCC at the bolted fault current
Voltage (system)	208, 480, 4160, 12,470 V	Per system design
Electrode configuration	VCB, VOA, VCBB, HCB, HOA — vertical/horizontal, in box / open air, with/without barrier	Per equipment construction
Gap between conductors	Standard: 25mm at 600V, 32mm at 5kV, 102mm at 15kV	NESC defaults
Box dimensions	Width × height × depth	Per equipment cutsheet (typical: 508×508×508 mm for 480V switchgear)
Working distance	Distance from arc to worker's chest	18" (455mm) for LV, 36" (915mm) for MV typical

i

Trip time matters MORE than fault current

Incident energy is approximately linear in trip time but only weakly dependent on fault current. Doubling trip time doubles incident energy. Halving trip time halves it. This is why every mitigation strategy targets faster clearing.

PPE Categories

NFPA 70E defines PPE categories based on incident energy. The category determines what the worker must wear when working on or near energized equipment.

Category	Incident energy (cal/cm²)	Required PPE
0 (eliminated)	< 1.2	Long-sleeve work clothing, safety glasses, hard hat. (Threshold below 2nd-degree burn.)
1	1.2 – 4	4 cal arc-rated (AR) shirt + pants OR coverall + face shield + balaclava
2	4 – 8	8 cal AR clothing + AR face shield with balaclava OR full hood
3	8 – 25	25 cal AR suit + full hood + AR gloves
4	25 – 40	40 cal AR suit (heavy) + full hood + heavy AR gloves. Maximum allowed.
> 40 ("dangerous")	> 40	NO PPE PROVIDES PROTECTION. Equipment must be de-energized before work.

Boundaries — Shock vs Arc Flash

Boundary	Definition	Distance basis
Limited approach (shock)	Crossing requires being qualified worker or escorted	Per NFPA 70E Table 130.4(E)(a) — voltage-based
Restricted approach (shock)	Crossing requires shock PPE + work permit + protective equipment	Per NFPA 70E Table 130.4(E)(a)
Arc flash boundary (AFB)	Distance at which incident energy drops to 1.2 cal/cm² (2nd-degree burn threshold)	Calculated per IEEE 1584 — depends on fault and trip time

NEC 110.16 — Label Requirements

Every piece of electrical equipment likely to need examination, adjustment, servicing, or maintenance while energized must be labeled. Two label tiers — minimum NEC 110.16 and detailed per NFPA 70E.

Label content	NEC 110.16(A) generic	NEC 110.16(B) detailed (since 2017 NEC for service ≥ 1200A)
Warning of arc flash hazard	✓	✓
Nominal voltage	—	✓
Available fault current	—	✓
Clearing time of upstream OCPD	—	✓
Date of label	—	✓
Incident energy + PPE category (NFPA 70E)	—	Per NFPA 70E 130.5(H), site-specific labels
Arc flash boundary	—	Per NFPA 70E 130.5(H)

Mitigation Strategies — How to Reduce Incident Energy

Strategy	How it works	Reduction
Maintenance switch	Reduces instantaneous trip setting during energized work. After work, restored to normal.	50-90% reduction
Zone-Selective Interlocking (ZSI)	Upstream CB asks downstream "do you see this?" — if no, upstream trips immediately	Selective coordination at full speed; large reduction at upstream buses
Current-limiting fuses	Open in < 1/4 cycle on bolted fault. Limits let-through energy.	Up to 90% on the protected zone
Arc-resistant switchgear	Equipment vents arc upward through ducts. Workers in front are protected.	Eliminates worker-side hazard, but doesn't reduce energy
Remote racking / remote operation	Worker is outside the arc flash boundary during racking	Removes worker, not energy. Best practice combined with other mitigations.
Higher-impedance transformer	Reduces fault current at secondary	Linear with %Z increase — but increases voltage drop
Optical arc flash detection	Photo sensors detect arc flash light, command upstream CB to trip in < 1/2 cycle	Drastic reduction (90%+) — newer technology
De-energize for work	The only true elimination	100% reduction

Visual — Boundaries Around an Arc Flash Source

Three boundaries are independent. Arc flash boundary (AFB) varies with incident energy; shock boundaries are voltage-dependent (NFPA 70E Table 130.4(E)).

Worked Example 1 — Atlas DC1 480V SWGR Arc Flash

Example 01 · Atlas DC1 spine480V SWGR-A (4000A bus) — arc flash analysis with and without mitigation

Inputs

Voltage

480V (LL)

Bolted fault

50.3 kA (per §12)

Working distance

18" (455 mm)

Box dimensions

508 × 508 × 508 mm

Configuration

VCB (vertical conductors in box)

Scenario A — without mitigation

Trip time: Upstream is the TX-A primary breaker (12.47 kV side). Looking at TCC at 50.3 kA reflected to primary (~5 kA at 12.47 kV side): inverse-time trip ~ 0.2 sec (12 cycles).
IEEE 1584 result: Incident energy ≈ 18 cal/cm². AFB ≈ 6 ft.
PPE Category: 3 (between 8 and 25 cal). Heavy AR suit + hood required.

Scenario B — with maintenance switch enabled

Maintenance switch lowers instantaneous setting: Trip in 4 cycles (0.067 sec) at 50 kA fault.
IEEE 1584 result: Incident energy ≈ 6 cal/cm². AFB ≈ 4 ft.
PPE Category: 2 (between 4 and 8 cal). Standard AR shirt+pants+face shield. Much lighter PPE.

Scenario C — with 87B bus differential trip

Bus differential clears in 4 cycles always. Doesn't depend on coordination time delay.
IEEE 1584 result: Same as Scenario B — ~6 cal/cm² with no maintenance switch action required.
Atlas DC1 chose this approach: Permanent 87B reduces normal-operation incident energy. (All cal/cm² values shown are illustrative — real numbers come from running IEEE 1584-2018 with site-specific inputs. See Atlas DC1 Arc Flash Profile.)

!

Atlas DC1 label content

Per NFPA 70E 130.5(H): "WARNING — Arc Flash and Shock Hazard. 480V. Available Fault Current 50.3 kA. Trip time 0.067 sec. Incident Energy 6.0 cal/cm² @ 18". Arc Flash Boundary 4 ft. Limited Approach 3.5 ft. Restricted Approach 1 ft 2 in. PPE Category 2. Calculated 2026-01-15."

Worked Example 2 — PDU Panel Arc Flash

Example 02 · Atlas DC1 spinePDU-A1 480V distribution panel — high incident energy case

Surprising result: PDU panel often has HIGHER incident energy than upstream switchgear. Why? Lower fault current (less impedance margin) but also slower upstream trip time.
Inputs: 480V, 25 kA bolted fault (after PDU isolation transformer), upstream OCPD at UPS-A1 output is 2000A LSIG. At 25 kA, it trips in ~ 1 sec (long-time region).
IEEE 1584 result: Incident energy ≈ 12 cal/cm² at 18". Category 3 PPE.
Mitigation: Lower upstream LSIG instantaneous to ~ 4× pickup (8000A) — trips in 0.05 sec. New incident energy ≈ 2.5 cal/cm² → Category 1.
Trade-off: Lower instantaneous = better arc flash but might trip on motor inrush. Coordination check required.

i

Don't assume lower voltage = lower hazard

PDU panels (480V) routinely calculate HIGHER incident energy than upstream MV switchgear, because MV CBs trip much faster and MV gear is often arc-resistant. Always run the calculation; never assume.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · PPE Cat

Incident energy = 6 cal/cm². PPE category?

Drill 2 · Working distance

What working distance for LV vs MV?

Drill 3 · AFB definition

What is the Arc Flash Boundary?

Drill 4 · Trip time impact

Trip time DOUBLES from 100 ms to 200 ms. Incident energy?

Drill 5 · > 40 cal/cm²

Equipment shows 50 cal/cm². Action?

IEEE 1584-2018 — Walking Through the Formula

IEEE 1584 provides the empirical equations for arc flash incident energy. The 2018 version is significantly different from the 2002 version (which was the standard for 16 years). Here's how the actual calculation works.

Step 1 — Calculate arcing current

The arcing current (I_arc) is less than bolted fault current because the arc itself adds impedance. For Voltage 600V or below:

Arcing current (480V VCB configuration)

log₁₀(I_arc) ≈ k + 0.662 log₁₀(I_bf) + 0.0966V + 0.000526G

Where k is a constant per electrode config, I_bf = bolted fault kA, V = system voltage kV, G = gap mm. Typical I_arc ≈ 50-90% of I_bf.

Step 2 — Calculate normalized incident energy

The normalized energy (E_n) is at standardized conditions (610 mm working distance, 0.2 sec arc duration).

Normalized incident energy (general form)

log₁₀(E_n) ≈ k₁ + k₂ × log₁₀(I_arc) + k₃ × G

k₁, k₂, k₃ are empirical constants from IEEE 1584-2018 Table 1, depending on electrode configuration (VCB, VOA, VCBB, HCB, HOA).

Step 3 — Adjust for actual working distance + arc duration

Actual incident energy at worker's chest

E = E_n × (t / 0.2) × (610 / D)^x

t = arc duration (sec) — the upstream OCPD trip time at I_arc. D = working distance (mm). x = distance exponent (~ 1.6 for VCB at 480V, varies by config).

Step 4 — Convert to cal/cm²

If E is in joules/cm², divide by 4.184 to get cal/cm². This is what gets compared to PPE category.

Electrode Configurations — How They Affect E

Code	Configuration	Where used	Relative E
VCB	Vertical Conductors in metal Box	Standard switchgear, panelboards	Reference (1.0)
VOA	Vertical Conductors in Open Air	Outdoor disconnects, exposed buses	Lower than VCB (0.7-0.85×)
VCBB	Vertical Conductors in metal Box w/ Barrier	Sectioned switchgear with insulating barrier	Higher than VCB (1.2×) — barrier directs arc forward
HCB	Horizontal Conductors in metal Box	Some bus configurations, MCC buckets	Higher than VCB (1.2-1.4×)
HOA	Horizontal Conductors in Open Air	Rare — outdoor horizontal bus	Lower (0.7×)

!

2018 vs 2002 — why the change matters

2002 IEEE 1584 had only one electrode configuration. 2018 added five, because real-world tests showed enclosure geometry and electrode orientation have major effects on energy direction. Pre-2018 calculations may underestimate energy by 30-50% for HCB/VCBB configurations. Re-run any pre-2018 arc flash study before relying on the labels.

Worked Example 3 — IEEE 1584 Sensitivity Study

Example 03 · Atlas DC1 spine480V SWGR-A — how each input affects incident energy

Baseline case (per spec)

Bolted fault

50 kA

Arc current (calc)

~ 35 kA (about 70% of bolted)

Trip time (with 87B)

0.067 sec

Working distance

18" / 455 mm

Configuration

VCB (vertical in box)

Box dimensions

508×508×508 mm

Result

~ 6 cal/cm² (PPE Cat 2)

What if we change ONE variable?

Variable changed	New value	New incident energy	Change
Trip time	0.2 sec (no 87B)	~ 18 cal/cm²	3× higher
Trip time	0.5 sec (only main 51 backup)	~ 45 cal/cm² 🚨	7.5× — Cat 4+ — DANGEROUS
Working distance	36" (1.6× farther)	~ 2.5 cal/cm²	~ ⅖ — Cat 1
Configuration	VCBB (with barrier)	~ 7.2 cal/cm²	+20%
Configuration	HCB (horizontal)	~ 8.4 cal/cm²	+40%
Bolted fault	30 kA (smaller TX)	~ 4 cal/cm²	−33% — Cat 1
Bolted fault	65 kA (larger TX)	~ 7.5 cal/cm²	+25% — still Cat 2

Key insight: Trip time has a roughly LINEAR effect on incident energy. Bolted fault current has a much weaker effect. Halve the trip time → halve the incident energy. This is why every mitigation strategy targets faster clearing.

If You See THIS, Think THAT

If you see…	Think / use…
"IEEE 1584-2018"	Current arc flash calculation standard. Replaced 2002 version. Different formulas + electrode configs.
"NFPA 70E"	Workplace electrical safety. Drives PPE selection + work practices. Updated every 3 years.
"NEC 110.16"	Required arc flash labels on equipment. Generic + (since 2017) detailed for ≥ 1200A services.
"Incident energy" or "cal/cm²"	Energy at working distance. 1.2 = 2nd-degree burn threshold. 8 = serious.
"PPE Category 2 / 3"	Required protective clothing. Cat 2 = 8 cal AR. Cat 3 = 25 cal AR suit.
"Arc Flash Boundary" (AFB)	Distance at which incident energy drops to 1.2 cal/cm². Workers must wear PPE inside this boundary.
"Working distance" (typically 18" or 36")	Distance from arc to worker's chest. Affects calculation.
"Maintenance switch"	Lowers instantaneous trip setting during energized work. Reduces incident energy 50-90%.
"Arc-resistant switchgear"	Vents arc upward. Eliminates worker-side hazard for closed-door operation.
"Optical arc flash detection"	Photo sensor + ZSI signal. Sub-cycle clearing. Modern mitigation.
Incident energy > 40 cal/cm²	"Dangerous" — no PPE provides protection. Equipment must be de-energized.

§20 / 39

Emergency & Standby Systems

NEC 700/701/702 · ATS types · UPS topologies · generator paralleling

When utility power fails, emergency systems take over. NEC 700/701/702 distinguish required emergency (life safety), legally required standby, and optional standby. Generator paralleling adds complexity above single-genset systems.

NEC 700 / 701 / 702 — Three Tiers of Standby

NEC Article	Type	Loads served	Transfer time	Wiring requirements
700	Emergency System (life safety)	Egress lighting, exit signs, fire alarm, fire pumps, smoke control	≤ 10 sec from utility loss	Separate from all other systems. Selectively coordinated. Listed equipment only.
701	Legally Required Standby	Sewage handling, communication, ventilation for first responders, certain HVAC	≤ 60 sec	Separate from optional but can share emergency. Selectively coordinated.
702	Optional Standby	Anything you want continuous power on — data centers, manufacturing, comfort	No code requirement	Standard wiring methods. No selective coordination requirement.
Critical Operations Power Systems (COPS)	NEC 708 — only certain critical infrastructure (financial, security, emergency communications)	Specialty	Highest level — bunker construction	Separate from all other; resistance to physical attack

Generator Sizing — More Than Just Demand kW

Generators must handle (1) the connected demand load, (2) the largest motor's starting kVA, (3) step-loading transients during sequential ATS transfers, and (4) harmonic non-linear loads. The biggest of these governs sizing.

Sizing factor	Calculation	Atlas DC1 example
Demand load (kW)	Sum of all loads at peak	Side A demand ≈ 2,652 kW
Demand kVA	kW / system PF	2,652 / 0.95 ≈ 2,791 kVA
Motor starting kVA	Largest motor LRkVA / system damping	VFD-driven chillers — no inrush. If DOL: 5.6 kVA/HP × 450 = 2,520 kVA momentary.
Step loading	Largest single-step load increase during ATS sequence	Atlas DC1 transfers IT load (UPS pre-loaded → step transfer ~ 1.25 MW)
Nonlinear load impact	Generator alternator must handle harmonic currents — derate ~ 10% if > 30% nonlinear load fraction	Atlas DC1 ~ 50% nonlinear (UPS, VFDs) → derate 15%
Final size	Largest of above + future capacity headroom	2,791 / 0.85 derate ÷ 0.95 PF ≈ 3,460 kVA → spec'd 2,500 kW (3,125 kVA at 0.8 PF). Marginal — real Atlas would step up.

ATS — Automatic Transfer Switch

Type	Operation	Pros	Cons	Where used
Open Transition	Break-before-make. Short outage on transfer (50-200 ms typical).	Simple. Cheaper. Cannot backfeed utility.	Brief power loss on transfer.	Standard for most installations including Atlas DC1. IT loads ride through via UPS.
Closed Transition	Make-before-break. Generator paralleled with utility for 100 ms.	No power interruption.	Requires generator + utility synchronization (25). Utility approval (IEEE 1547 / UL 1741).	Critical applications without UPS: HVAC, hospitals.
Delayed Transition	Open transfer with intentional 1-3 sec delay in middle	Allows certain motor loads (centrifuges, etc.) to coast down before transfer — prevents out-of-phase reconnection.	Unusable for sensitive loads.	Specialty industrial.
Bypass-Isolation	ATS can be removed for service while load is fed via bypass switch	Maintainability. Required for Tier III/IV data centers.	More expensive.	Atlas DC1 (Tier III equivalent).

UPS Topologies

Topology	Operation	Pros	Cons	Where used
Offline / Standby	Load fed from utility; battery + inverter take over on outage	Cheapest. Highest efficiency (~ 99%).	Brief 4-10 ms transfer. No conditioning of utility power.	Small UPS (≤ 3 kVA) — desktop applications
Line-Interactive	Always-on autotransformer regulates voltage; battery + inverter for outage	Voltage regulation. Better than offline. Efficient (~ 97%).	Brief transfer on outage (~ 4 ms).	Mid-size UPS (3-50 kVA) — small server rooms
Online Double-Conversion	Always running through rectifier → battery → inverter. Load NEVER sees utility directly.	Zero transfer time. Perfect output regardless of utility quality. PFC + harmonic filtering inherent.	Lower efficiency (~ 94-96%). Higher cost.	Industry standard for large UPS — Atlas DC1.
ECO mode (eco-conversion)	Bypass mode unless utility poor; switches to double-conversion when needed	~ 99% efficiency in normal mode (savings on losses)	Brief transfer when switching modes	Some modern data center UPS — e.g., Mission Critical Eco mode
Rotary UPS	Diesel + flywheel + AC alternator — no batteries	No battery maintenance. Long life.	Limited ride-through (10-20 sec). Lots of moving parts.	Some hyperscale DCs (Active Power, Hitec)

Generator Paralleling & Synchronization

For systems with multiple generators (large data centers, hospitals, industrial), the generators must paralleled to share load. Synchronization is the critical step before paralleling.

Synchronization parameter	Tolerance for paralleling	Why
Frequency	Within 0.1 Hz (or 0.2%)	Frequency mismatch causes power oscillation
Voltage magnitude	Within 5%	Voltage mismatch causes reactive power circulation
Phase angle	Within 10° (some apps require < 5°)	Phase mismatch causes large transient current and torque jolt on alternator
Phase rotation	Must match exactly	Wrong rotation = catastrophic short circuit

Paralleling Switchgear

A paralleling switchgear lineup includes synchronization relays (25), governor controls, voltage regulator interfaces, load sharing controls, and an HMI. Industry vendors: Caterpillar, Cummins, Generac, Russelectric, Aspen, Pioneer.

Worked Example 1 — Atlas DC1 Generator System

Example 01 · Atlas DC1 spine2 × 2500 kW gens, 2N (one per side), no paralleling — vs alternative architectures

Architecture: 2N independent (chosen)

Each side has its own genset. GEN-A serves Side A only via ATS-A. GEN-B serves Side B only via ATS-B. No cross-tie of generator outputs.
Why no paralleling: Avoids IEEE 1547 / UL 1741 utility approval if closed-transition. Avoids paralleling switchgear cost. Avoids common-mode failures from synchronization controls.
Trade-off: If GEN-A fails during utility outage AND IT load is high, Side A loses power → IT loses 50% capacity (still operational on Side B). Acceptable per 2N design philosophy.
Genset sizing margin: Each gen sized at 2,500 kW = 3,125 kVA at 0.8 PF. Side A demand was 2,791 kVA. Tight. Real Atlas would size 3,000 kW.

Alternative architecture: paralleled generators (rejected)

Architecture: Both gens parallel onto a common bus. Each side fed from common gen bus through its ATS.
Pros: Either gen carries either side. Total capacity 5 MW shared.
Cons: Common-mode failure (paralleling SWGR fails) → no gen power at all. Cost: ~ $500K extra for paralleling SWGR + controls.
Decision: 2N independent wins on simplicity + reliability for this size.

Worked Example 2 — Hospital Essential Electrical System

Example 02 · Alternate contextHospital — NEC 517 essential electrical system + life safety transfer ≤ 10 sec

Three branches per NEC 517:
• Life safety (egress lighting, exit signs, fire alarm) — NEC 700, ≤ 10 sec transfer
• Critical care (operating rooms, ICU, ED, dialysis) — NEC 700, ≤ 10 sec
• Equipment (HVAC for critical areas, elevators, etc.) — NEC 701 or 700
Architecture: Single 1500 kW diesel genset → 3 ATSs → 3 essential branches. Each ATS sized for its branch's full demand.
Why selective coordination matters here: A short on one operating room's panel cannot trip the genset main. Per NEC 700.27.
Test schedule: NFPA 110 — monthly test under load (≥ 30%). Annual test at 100% load for 4 hours. Failures must be documented and corrected.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · NEC 700 transfer time

Maximum transfer time for emergency system?

Drill 2 · ATS open vs closed

Brief outage on transfer?

Drill 3 · UPS topology

Zero transfer time UPS?

Drill 4 · Gen sizing

Genset must handle (1) demand, (2) inrush, (3) step load. Largest of these governs:

Drill 5 · Atlas DC1 paralleling?

Does Atlas DC1 parallel its 2 generators?

If You See THIS, Think THAT

If you see…	Think / use…
"NEC 700 system"	Emergency / life safety. ≤ 10 sec transfer. Selectively coordinated. Listed equipment.
"NEC 701"	Legally required standby. ≤ 60 sec transfer.
"NEC 702"	Optional standby. Most commercial / industrial / DC backup falls here.
"Open-transition ATS"	Brief outage on transfer. Standard for most installations. UPS rides through.
"Closed-transition ATS"	Generator paralleled with utility briefly. No outage. Requires utility approval.
"ATS bypass-isolation"	ATS removable for service. Required for Tier III/IV.
"Online double-conversion UPS"	Industry standard for large UPS. Zero transfer time.
"Eco mode UPS"	Higher efficiency. Brief transfer when switching modes.
"Rotary UPS"	Diesel + flywheel. No battery. 10-20 sec ride-through.
"Generator paralleling"	Synchronization (25) + load sharing controls. Significant cost.
"Paralleling switchgear"	Custom lineup with sync controls. Atlas DC1 doesn't have this.
NFPA 110	Standard for emergency + standby power systems. Test requirements.
NFPA 111	Stored energy systems (UPS, batteries).
"Step loading" of generator	Maximum kW the gen can pick up in one step. Limits ATS transfer sequencing.

§21 / 39

DC Systems & Battery Sizing

NEC 480 · battery chemistries · IEEE 485 sizing · float charging · room ventilation

Substations, data center UPS, telecom, and industrial controls all need DC backup. NEC 480 governs storage batteries. Sizing requires Ah calculation across the worst-case duty cycle plus aging and temperature factors.

Where DC Systems Live

Application	Voltage	Why DC
UPS battery strings (data centers)	240, 480, 540V (depending on inverter)	Battery storage requires DC; inverter converts back to AC
Substation battery (control + protection)	125V (most common); 250V; 48V	Powers protective relays + breaker trip coils. Must operate during AC outage.
Telecom (DC plant)	-48V (negative grounded)	Legacy from telephone era. Equipment standardized worldwide.
Solar PV systems	~ 600-1500V DC string	PV cells produce DC; inverter to AC for utility tie
Modern data center DC distribution (emerging)	-380V or +380V	Eliminates DC-AC-DC conversion losses for IT loads
Industrial control circuits	24V DC	PLC inputs/outputs, sensors, contactors. Safer than 120V AC for control wiring.
Backup lighting (egress)	12V or 24V DC battery integral to fixture	Battery-backed exit signs / egress lights

Battery Chemistries

Chemistry	V/cell nominal	Typical use	Pros	Cons
VRLA / AGM (Valve-Regulated Lead-Acid)	2.0V	UPS, telecom, generator starting	Sealed (no maintenance). Spillproof. Affordable.	~ 5-7 yr life. Hydrogen evolution under abnormal conditions.
Flooded lead-acid	2.0V	Substation batteries, large industrial	~ 20 yr life. Field-rebuildable. Tolerates abuse.	Requires water addition. Hydrogen evolution always. Spill containment.
Li-ion (LFP, NMC)	3.2V (LFP); 3.7V (NMC)	Modern UPS, EV, ESS, residential storage	10-15 yr life. Higher energy density. Faster recharge. Less weight.	Expensive. Thermal runaway risk (NMC moreso). Dedicated BMS required.
NiCd (Nickel-Cadmium)	1.2V	Substation, harsh environment, aviation	20+ yr life. Cold weather tolerance. Deep discharge OK.	Expensive. Cadmium toxicity. Memory effect.
Flow batteries (Vanadium, Zinc)	varies	Grid-scale ESS, long-duration storage	Decoupled power and energy. Long cycle life.	Bulky. New technology — limited deployment.

Battery Sizing — The Ah Calculation

For a UPS or substation battery, sizing requires defining the duty cycle (load profile vs time), then translating to Ah needed. IEEE 485 (lead-acid) and IEEE 1184/1188 govern.

Simple sizing — constant load

Ah = (Load A × Hours) × (Aging factor) × (Temp factor) × (Design margin)

Aging: 1.25 (battery loses capacity to 80% over life). Temp: 1.0 at 77°F, increases for cold. Design margin: typically 1.10.

Atlas DC1 UPS Battery — Worked

UPS-A1 = 1250 kVA, 480V output. Battery string at ~ 540V DC (270 cells × 2V). Required ride-through: 5 minutes (long enough for genset to start and ATS to transfer).

Step	Calculation	Result
1. UPS DC current at full load	I = 1,250,000 W / (540V × 0.96 inverter η) = 2,411 A DC	2,411 A
2. Energy for 5 min	2,411 × (5/60) = 200.9 Ah	200.9 Ah at full discharge
3. Aging factor (1.25)	200.9 × 1.25 = 251.1 Ah	251.1 Ah
4. Temp factor (1.0 at 77°F)	251.1 × 1.0 = 251.1 Ah	251.1 Ah
5. Design margin (1.10)	251.1 × 1.10 = 276.2 Ah	276.2 Ah
6. Round to next standard cell size	VRLA available: 100, 150, 200, 300 Ah cells	300 Ah cell × 270 cells per string

Float Charging

Battery is kept at full charge by a continuous low-voltage float charge from the rectifier. Voltage is set above battery resting voltage but below gassing voltage.

Battery type	Float V/cell	Equalize V/cell (occasional)
VRLA	2.25-2.30 V	2.35-2.40 V (some types — most don't need)
Flooded lead-acid	2.20-2.25 V	2.45-2.55 V (monthly)
NiCd	1.40-1.45 V	1.55 V
Li-ion (LFP)	3.40 V (or charge to 3.40 then float at lower)	None — not needed

Battery Rooms — Hydrogen Hazard

Lead-acid and NiCd batteries evolve hydrogen during charging — especially during equalization. Concentration must stay below 2% (50% of 4% LEL).

Code	Requirement
NEC 480.10(A)	Battery rooms must have ventilation to prevent hydrogen accumulation
NFPA 1	Ventilation rate: 1 cfm/sq ft floor minimum (with active monitoring)
IEEE 1635 / ASHRAE 21	Calculate hydrogen evolution rate; size ventilation
NEC 500.5(B)(1) (impl.)	Battery rooms typically Class I Div 2 Group B (hydrogen). See §21.
Spill containment	Required for flooded lead-acid (electrolyte). VRLA exempt.

Worked Example 1 — Atlas DC1 UPS Battery (Continued)

Example 01 · Atlas DC1 spineFull battery string design + room ventilation + protection

Battery configuration

Item	Spec
Cells per string	270 × VRLA 2V cells = 540V nominal
Cell capacity	300 Ah at 8-hour discharge rate (~ 280 Ah at 5-min discharge rate after rate derating)
Strings per UPS	2 (parallel) for N+1 redundancy at the string level
Total per UPS	540 cells × 300 Ah
Batteries for full Atlas DC1	4 UPS × 2 strings × 270 cells = 2,160 cells

Battery room sizing

Hydrogen evolution rate (per IEEE 1635): ~ 0.0006 cfh/cell at float; ~ 0.05 cfh/cell at equalize. For 2,160 cells:
At equalize: 0.05 × 2,160 = 108 cfh = 1.8 cfm hydrogen production
Ventilation to keep H₂ < 1% (safety factor under 2% LEL/2):
Ventilation rate = 100 × hydrogen production = 180 cfm minimum
Real: 500 cfm (active fan with H₂ sensor monitoring)
Hazardous location classification: Class I Div 2 Group B (hydrogen). Equipment in battery room must be Div 2 rated. See §21.

DC protection

Protection	Detail
String breaker (DC)	Each string protected by a DC-rated CB. Sized for full discharge current. AIC = available DC fault current.
Ground fault detection	DC ungrounded systems use ground fault monitoring. Per NEC 480.10(D), each string monitored for ground.
Battery monitor	Modern systems monitor cell voltage + impedance to predict failure. Albers, Eagle Eye.

Worked Example 2 — Substation Battery (125V DC)

Example 02 · Alternate contextIndustrial substation — 125V DC battery for protective relay tripping power

Application: Powers all protective relays + breaker trip coils on a 13.8 kV substation. Must operate during AC outage.
Duty cycle (per IEEE 485): Normal load 5 A continuous (relay supplies). Trip load: 30 A for 0.1 sec when CB trips. End: 5 A for 8 hr ride-through (battery sized to last until utility restored).
Sizing: Continuous 8 hr × 5 A = 40 Ah base. Add aging 1.25 + temperature + margin. Size: ~ 60 Ah at 8-hour rate. Use 60 Ah NiCd or 80 Ah flooded lead-acid.
Why NiCd preferred for substations: 20+ yr life vs lead-acid 7. Tolerates outdoor temperature swings. Lower long-term cost.
String configuration: 92 NiCd cells × 1.2V = 110V (charging brings to 125V).

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · VRLA float V

Float voltage per VRLA cell?

Drill 2 · Telecom DC

What voltage is telecom DC plant?

Drill 3 · Battery sizing (IEEE 485)

Constant 100 A load for 5 min. Add aging 1.25, temp 1.0, margin 1.10. Ah?

Drill 4 · Battery room hazard

Lead-acid battery room. Hazardous location class?

Drill 5 · Li-ion advantage

Why DCs migrate to Li-ion UPS?

If You See THIS, Think THAT

If you see…	Think / use…
NEC 480	Stationary storage batteries. Governs battery rooms, ventilation, ground fault.
"VRLA" / "AGM"	Sealed lead-acid. Most common UPS battery. 5-7 yr life.
"Flooded lead-acid"	20 yr life. Substation choice. Requires maintenance + spill containment.
"Li-ion" or "LFP"	Modern data center UPS. Lighter, longer life, higher cost. Requires BMS.
"NiCd"	Long life (20+ yr). Substations + harsh environment.
IEEE 485	Lead-acid sizing. Standard since 1983.
IEEE 1184 / 1188	UPS battery sizing + maintenance.
IEEE 1635 / ASHRAE 21	Battery room ventilation calc.
"Float voltage"	Continuous low charge to keep battery at full state of charge. ~ 2.25-2.30 V/cell for VRLA.
"Equalize charge"	Periodic higher voltage to balance cells. NOT needed for VRLA or Li-ion typically.
"-48V" telecom	Negative-grounded DC. Telecom worldwide standard. Powers radios, switches.
"Battery room" + Class I Div 2	Hydrogen evolution → hazardous location. Equipment must be Div 2 rated for Group B.

§22 / 39

Hazardous Locations

Class I/II/III · Division 1/2 · Zone system · equipment marking · protection methods

Where flammable gases, dusts, or fibers exist, equipment must be specially rated. The Class/Division system (US legacy) and the Zone system (international) both define the hazard. Equipment ratings get cryptic — but the system underneath is logical.

Class / Division System (US Legacy)

NEC defines hazardous locations by what makes them hazardous (Class) and how often the hazard is present (Division).

Class	Hazard	Examples
Class I	Flammable gases, vapors, liquids	Refineries, gas stations, paint booths, aircraft hangars, battery rooms (hydrogen)
Class II	Combustible dusts	Grain elevators, sawmills, sugar refineries, coal handling, pharmaceutical mfg
Class III	Ignitable fibers / flyings (no longer combustible-suspended)	Cotton mills, woodworking, textile finishing

Division	Definition	When the hazard is present
Division 1	Hazard present under normal operating conditions — continuously, intermittently, or periodically	The vapor space inside a fuel tank; the cyclone of a dust collector while running
Division 2	Hazard present only under abnormal conditions — accidental release, equipment failure	The 5-foot zone around a closed valve; an enclosed area near an open Class 1 Div 1 location

Group System (Class I & II)

Different gases and dusts ignite at different energies. NEC subdivides into Groups based on this.

Class	Group	Material	Notes
I	A	Acetylene	Most ignitable Class I material
I	B	Hydrogen, ethylene oxide	Battery rooms = Group B
I	C	Ethylene, ether
I	D	Methane, propane, gasoline, alcohol	Most common gas group
II	E	Conductive metallic dusts (aluminum, magnesium)
II	F	Carbon-based dusts (coal, charcoal, coke)
II	G	Other combustible dusts (grain, plastic, sugar, wood)	Most common dust group

Zone System (IEC / International — also acceptable in NEC)

Increasingly used in US under NEC 505 / 506 as alternative to Class/Division. Similar concept; different naming.

Zone (gas)	Description	Class/Div equivalent
Zone 0	Hazard present continuously or for long periods	(part of Class I Div 1)
Zone 1	Hazard present periodically under normal operation	(part of Class I Div 1)
Zone 2	Hazard present only under abnormal conditions	Class I Div 2
Zone 20, 21, 22	Same hierarchy for dust (NEC 506)	Class II Div 1 / 2

Equipment Marking — Decoded

Equipment for hazardous locations is marked with Class, Division (or Zone), Group, and temperature code.

Example marking

Class I, Div 2, Groups B, C, D, T3

Suitable for flammable gas (Class I), abnormal-only hazard (Div 2), groups B (hydrogen), C, D — and operates at temperature T3 (≤ 200°C surface).

Temperature Codes (T-codes)

T-code	Max surface temp
T1	450°C
T2	300°C
T3	200°C
T4	135°C
T5	100°C
T6	85°C

Equipment T-code must be lower than the auto-ignition temperature of the gas/dust present.

Protection Methods

Method	Code	How it works	Where used
Explosion-Proof (Flameproof)	XP, Ex d	Heavy enclosure contains internal explosion; flame quenched at flange path before reaching outside	Class I Div 1 motors, switches, junction boxes
Purged / Pressurized	X, Ex p	Continuous purge with inert/clean gas keeps flammable atmosphere out of enclosure	Large enclosures: control panels, motors, analyzers
Intrinsically Safe	IS, Ex i	Energy in circuit is too low to cause ignition under any fault condition	Field instruments, sensors, low-power control loops
Encapsulated	Ex m	Components potted in resin — physically isolated from atmosphere	Solenoids, small electronics
Oil-immersed	Ex o	Components submerged in oil isolated from atmosphere	Switchgear (legacy)
Sand-filled	Ex q	Quartz sand fills enclosure — components isolated	Older equipment
Non-incendive (Div 2 only)	Ex n	Cannot ignite under normal operation. Cheaper than IS.	Most Class I Div 2 equipment — common LV equipment

Area Classification — How Zones Are Determined

The Owner (with help from chemical engineers) creates an area classification drawing showing zones around each potential leak source. NFPA 497 (gases/vapors) and NFPA 499 (dusts) provide the methodology.

Source	Typical zone classification
Inside fuel tank vapor space	Class I Div 1 (Zone 0)
5 ft cylinder around tank vent	Class I Div 2 (Zone 2)
10 ft horizontally + 18 in vertically around dispensing nozzle (gas station)	Class I Div 1
Beyond 10 ft, within 25 ft	Class I Div 2
Aircraft hangar floor up to 18" (lighter-than-air spillage)	Class I Div 1
Aircraft hangar above 18"	Class I Div 2
Inside grain elevator (silo, processing)	Class II Div 1
Outside grain elevator (10 ft buffer)	Class II Div 2

Visual — Area Classification Around a Vapor Source

Each piece of equipment must be rated for its zone. Boundaries determined by vapor density, ventilation, and source rate per NFPA 497.

Worked Example 1 — Atlas DC1 Battery Room

Example 01 · Atlas DC1 spineLead-acid battery rooms — Class I Div 2 Group B (hydrogen)

Hazard analysis

Hazard: Lead-acid batteries evolve hydrogen during normal float charging (very small amount) and during equalization (significant). Hydrogen LEL = 4%.
Why Div 2 (not Div 1): With proper ventilation maintaining concentration < 1%, the hazard exists only under abnormal conditions (loss of ventilation + simultaneous overcharging).
Class I Group B: Hydrogen requires Group B equipment (most demanding flammable gas group).

Equipment requirements

Item	Requirement
Lighting fixtures	Class I Div 2 Group B rated. Most LED industrial fixtures qualify.
Conduit + boxes	Standard EMT/RMC OK in Div 2 (vs. specialty XP boxes required in Div 1)
Receptacles + switches	Non-incendive Div 2 rated, OR standard rated outside the classified area with 18" buffer
Battery monitoring equipment	Mounted outside classified area when possible; Div 2 rated when inside
Ventilation fan	Non-sparking design. Run continuously with H₂ sensor backup.
Hydrogen gas detector	Continuous monitoring; alarm at 1% (25% of LEL); alarm + auxiliary ventilation at 2%

i

Why Li-ion can avoid Class I Div 2

Li-ion batteries don't evolve hydrogen during charging. Many DCs migrate to Li-ion specifically to eliminate the Class I Div 2 classification (and its equipment cost overhead). The trade-off: thermal runaway risk requires different fire protection (NFPA 855).

Worked Example 2 — Refinery Pump Station

Example 02 · Alternate contextPetroleum refinery — Class I Div 1 Group D pumping station with 50 HP centrifugal pump

Equipment	Spec	Why
Motor	Class I Div 1 Group D, T3, XP enclosure	Hydrocarbon vapors during normal operation. XP contains internal arcing.
Local disconnect	Class I Div 1 Group D XP	Located within sight of pump per NEC 430.102.
Conduit + fittings	RMC with explosion-proof fittings (sealed at boundary between hazardous and non-haz areas per NEC 501.15)	Prevents transmission of explosion through conduit system.
Lighting	Class I Div 1 Group D LED fixture	Fixtures with high IP rating + XP rating.
Field instrument (flow meter)	Intrinsically Safe	IS allows lower-cost field instruments. Energy too low to ignite even under fault.
Control wiring (4-20 mA)	IS rated; isolated barrier in non-haz control room	Ensures safe energy levels reach field.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Class meaning

Class II hazard?

Drill 2 · Division logic

Continuously present hazard during normal operation?

Drill 3 · Group B

Which gas is Group B?

Drill 4 · T-code

Equipment T3 — max surface temperature?

Drill 5 · XP vs IS

Cheaper protection for low-power field instruments?

If You See THIS, Think THAT

If you see…	Think / use…
"Class I" location	Flammable gas/vapor. Refinery, hangar, battery room, gas station.
"Class II" location	Combustible dust. Grain, sawmill, sugar, pharmaceutical.
"Division 1"	Hazard present in normal operation. Strictest equipment requirements (XP, IS, etc.).
"Division 2"	Hazard present only abnormally. Many standard rated equipment options.
"Zone 0/1/2" or "Zone 20/21/22"	IEC system. NEC 505/506 alternative. Increasingly used.
"Group B"	Hydrogen, ethylene oxide. Battery rooms.
"Group D"	Methane, propane, gasoline. Most common gas group.
"Group G"	Combustible dust (grain, sugar, plastic). Most common dust group.
"T3 / T4 / T6"	Max equipment surface temp. Lower number = hotter (allowed). Must be below auto-ignition temp.
"XP" or "Ex d" or "explosion-proof"	Heavy enclosure for Class I Div 1.
"IS" or "intrinsically safe" or "Ex i"	Energy too low to ignite. Field instruments. Cheaper than XP.
"Purged" or "X" or "Ex p"	Pressurized with clean gas. Larger enclosures.
"Conduit seal" / NEC 501.15	Mandatory at boundary of Class I area. Prevents flame propagation.
NFPA 497 / 499	Area classification recommended practice for gases/dusts.
NFPA 30 (flammable liquids)	Storage + handling code; impacts area classification around tanks.

§23 / 39

Substations & Switchyards

Equipment hierarchy · bus configurations · indoor/outdoor/GIS · IEEE 80 grounding

When facilities exceed ~5 MW, on-site substations become economical. Outdoor switchyards step down transmission to distribution voltages. Indoor unit substations integrate transformer + MV switchgear + LV switchgear in one lineup.

When to Build a Substation

For most commercial buildings, the utility provides a pad-mount transformer at the property line and the building has a 480V or 208V service. For larger facilities, the customer takes MV directly and steps it down on-site — a substation.

Facility size	Typical service	Substation type
≤ 1 MW	Pad-mount transformer at LV (480Y/277V)	None — utility pad-mount sufficient
1-5 MW	Pad-mount or vault primary; secondary unit substation	Indoor secondary unit substation (Atlas DC1 falls here)
5-20 MW	Customer-owned MV switchgear + multiple LV transformers	Indoor substation lineup
20-50 MW	Outdoor substation, customer-owned MV bus	Outdoor switchyard, padmounted or station-class transformers
≥ 50 MW (hyperscale DC, large industrial)	Direct from sub-transmission (69-138 kV)	Full outdoor switchyard with breakers, disconnects, lightning protection

Substation Equipment Hierarchy

Equipment	Function	Typical voltage range
Disconnect switch	Visual break for safe isolation. Operated only off-load (no fault interruption capability).	All voltages
Power circuit breaker	Interrupts load + fault current. Air-magnetic, vacuum, SF6, or oil insulated.	All voltages
Recloser	Distribution feeder breaker that automatically attempts re-closure 1-3 times after fault	5-38 kV
CT (Current Transformer)	Step-down current for metering + protection	Per voltage class
PT / VT (Potential / Voltage Transformer)	Step-down voltage for metering + protection	Per voltage class
Surge arrester (lightning arrester)	Diverts lightning + switching surges to ground	All voltages — sized to system MCOV
Power transformer	Step up/down. Pad-mount, station-class, autotransformer.	Per service
Metering / control building	Houses meters, relays, communications, batteries, station service	—
Grounding mat	Mesh of buried conductors limiting touch + step potential per IEEE 80	Per facility size

Bus Configurations

Configuration	Description	Reliability	Cost	Where used
Single bus	One main bus, breakers connect lines to bus	Lowest — bus fault drops everything	Lowest	Small distribution sub
Sectionalized bus	Single bus split by tie breaker into 2-3 sections	Medium — fault on one section doesn't drop other	Medium	Medium distribution sub
Main and Transfer bus	Main bus + auxiliary transfer bus, lines can move via bypass switches	Allows breaker maintenance	Medium-high	Older transmission
Ring bus	Ring of breakers; each line / transformer is between two breakers	High — any breaker can be removed without dropping load	High	Substations 69-230 kV; modern medium-large
Breaker-and-a-half	3 breakers per 2 circuits — middle breaker shared	Highest — any breaker or bus can be removed without losing load	Highest	Critical transmission, large generation
Double bus, double breaker	Two complete buses, each circuit has 2 breakers (one to each bus)	Highest — but expensive	Highest	Critical applications, less common in US

Indoor vs Outdoor vs GIS

Type	Description	Pros	Cons
Indoor metal-clad MV switchgear	Drawout breakers + bus in a metal enclosure, indoor location	Weather-protected. Compact. Easier maintenance.	Building cost. Limited voltage (≤ 38 kV).
Outdoor metal-enclosed MV switchgear (padmount)	Same as indoor but in a weatherproof enclosure	No building. Quick install.	Larger footprint. Weather exposure.
Outdoor station-class (open-air switchyard)	Air-insulated equipment on steel frames, outdoor	Cheap per MVA at high voltage. Standard for utility substations.	Large land area. Lightning exposed. Visual impact.
GIS (Gas-Insulated Switchgear)	SF6 gas-insulated metal enclosure	~ 10% footprint of air-insulated. Reliable. Indoor or outdoor.	Expensive. SF6 gas concerns (greenhouse). Specialized maintenance.

Grounding Mat — IEEE 80

A buried mesh of bare copper covering the substation footprint, bonded to all equipment. Limits touch potential (hand-to-feet) and step potential (foot-to-foot) during a fault to safe levels (≤ 250-1000 V depending on body weight + soil resistivity).

Aspect	Detail
Mesh	Typical 10×10 ft to 20×20 ft squares of bare copper or copper-clad steel
Conductor size	Per IEEE 80 fault current calc — 4/0 AWG to 500 kcmil typical
Burial depth	18-30" deep
Crushed rock surface	4-6" of high-resistivity crushed stone — increases foot resistance, reduces touch potential
Calculation	IEEE 80 — touch potential V_touch ≤ k × (1.16 + 0.7 × ρ_s) / √t · IEEE 80 design

Worked Example 1 — Atlas DC1 Service Topology (Indoor Secondary Unit Substation)

Example 01 · Atlas DC1 spine12.47 kV utility → 2× pad-mount transformers → 480V indoor switchgear (no on-site substation)

Why no on-site substation: Atlas DC1 is 5 MW total. Utility provides 12.47 kV. Two 2,500 kVA pad-mount transformers (utility-furnished or customer-owned) step down to 480V. The 480V switchgear IS the customer's main distribution.
Service architecture: Two utility distribution feeders (radial from different substations). Each feeds one TX → one Side. Total 2N redundancy from utility through to UPS.
What a true on-site substation would add: Customer-owned MV switchgear + multiple smaller LV transformers + paralleling capability. Justified ≥ 10 MW typically.

Worked Example 2 — Hyperscale DC Substation (Alternative Scale)

Example 02 · Alternate scale100 MW hyperscale data center campus — 138 kV utility service via on-site outdoor switchyard

Component	Spec
Utility service	138 kV from 2 separate utility substations (true 2N at the transmission level)
Switchyard configuration	Ring bus — 6 breakers, 3 line positions + 3 transformer positions
Substation transformers	3 × 75 MVA, 138-13.8 kV, %Z 8% — Δ-Y grounded
13.8 kV distribution	Customer-owned MV switchgear lineup — 6 outgoing feeders to data hall PDUs
Each data hall	Data hall has its own 13.8 kV → 480V step-down transformers (multiple per hall)
Standby generation	30 × 3 MW gensets, paralleled via paralleling switchgear, sync to 13.8 kV bus
Grounding mat	Per IEEE 80 — fault current 30 kA at 13.8 kV requires ~ 250 ft × 250 ft mesh of 4/0 bare Cu

A facility this size requires civil + electrical + utility coordination over 2-3 years before energization. The substation alone is a $20M-50M scope.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · When to substation

≥ 5 MW facility — typical?

Drill 2 · Bus configurations

Highest reliability bus configuration?

Drill 3 · GIS vs air-insulated

Compact + indoor MV switchgear?

Drill 4 · IEEE 80

Substation grounding mat standard?

Drill 5 · Atlas DC1 substation?

Does Atlas DC1 have an on-site substation?

If You See THIS, Think THAT

If you see…	Think / use…
"Substation" or "switchyard"	Customer-owned voltage transformation. ≥ 5 MW typical.
"Unit substation"	Integrated transformer + LV switchgear in one product. Typical for < 5 MW.
"Pad-mount transformer"	Outdoor weatherproof enclosure. Most common utility-supplied transformer.
"Station-class transformer"	Large outdoor transformer, generally ≥ 5 MVA. Open construction with cooling radiators.
"Ring bus"	6-breaker ring. High reliability for medium-large substation.
"Breaker-and-a-half"	Highest reliability. 3 breakers per 2 circuits. Critical transmission.
"GIS" (Gas-Insulated Switchgear)	SF6 insulated. Compact. Premium price.
"Recloser"	Distribution feeder breaker with auto-reclose. 1-3 attempts after temporary fault.
IEEE 80	Substation grounding (touch + step potential).
"Lightning arrester" (surge arrester)	Required at substation entry. Diverts lightning. See §23.
"CT" + "PT" or "VT"	Current + Voltage transformers for metering and protection.
"Drawout breaker"	Removable from cubicle for maintenance without dropping load (with bypass).
"Auto-transformer"	Single-winding transformer. Used 138-69 kV connections, common in transmission.

§24 / 39

Lightning Protection

NFPA 780 · rolling sphere · air terminals · down conductors · bonding

NFPA 780 governs structure protection from lightning. The rolling-sphere method determines where air terminals must go. Equipment grounding alone isn't enough — large structures need a proper LPS.

NFPA 780 — Standard for Lightning Protection

NFPA 780 governs lightning protection systems (LPS) for structures. It is NOT in the NEC — separate code, but referenced. Provides the methodology for sizing air terminals, down conductors, and ground systems.

NFPA 780 component	Purpose
Air terminals (Franklin rods)	Provide preferred attachment point for lightning strikes
Down conductors	Carry lightning current from air terminal to ground
Ground termination	Disperses lightning current into earth
Bonding	Equipotential bonding of all metallic objects to prevent flashover
Surge protection (SPDs)	Protects electrical/electronic equipment from induced surges (§24)

The Rolling Sphere Method

Imagine rolling a 150-ft (Class I structure) or 100-ft (Class II) sphere over the building. Wherever the sphere touches the building is exposed to a lightning strike. Air terminals must be placed so the sphere can't touch any vulnerable area.

Class	Sphere radius	Air terminal spacing	Application
Class I	150 ft	20 ft on protected periphery; 50 ft within	Buildings ≤ 75 ft tall
Class II	100 ft	20 ft	Buildings > 75 ft tall, structures with explosive contents, hazardous occupancies

Down Conductors

Spec	Class I	Class II
Conductor size (Cu)	32 AWG (~ 12 in² total cross section)	2/0 AWG (~ 67 in²)
Spacing on periphery (max)	100 ft (60 ft in seismic zones)	60 ft
Number minimum	2 per protected structure	2 per protected structure
Routing	Direct vertical path; avoid sharp bends (radius ≥ 8")	Same

Bonding (Equipotential)

All metal objects within 6 ft of the LPS down conductor must be bonded — pipes, gutters, cable trays, antenna masts, fences. Otherwise lightning current can side-flash from the down conductor through the metal object → equipment damage or fire.

Risk Assessment — IEC 62305 / NFPA 780 Annex L

Some structures are more important to protect than others. Risk assessment quantifies acceptable risk. Considers structure type, contents, location, lightning flash density.

Lightning Protection Level (LPL)	Sphere radius	Description
I (highest)	20 m (66 ft)	Critical infrastructure: nuclear, hospitals, explosive storage
II	30 m (98 ft)	Hazardous chemical / biological / industrial
III	45 m (148 ft)	Standard commercial / industrial
IV (lowest)	60 m (197 ft)	Residential, low-value assets

Worked Example 1 — Atlas DC1 LPS

Example 01 · Atlas DC1 spine2.5 MW data center, 200 ft × 300 ft × 25 ft tall — full NFPA 780 LPS

Classification: 25 ft tall, contains critical IT load. Class I structure (≤ 75 ft) but high-value contents argue for Class II treatment per Annex L.
Air terminals: 20 ft spacing on periphery + 20 ft on roof grid. For 200 × 300 ft roof: ~ 90 air terminals. Stainless steel rods, 12" tall above mounting structure.
Down conductors: 60 ft spacing on perimeter. 200 ft long side / 60 ft = 4 down conductors per long side; 300 ft / 60 = 5 per long side. Plus corners: ~ 18 down conductors total. 2/0 bare Cu, surface mounted.
Ground termination: Each down conductor terminates to a 10-ft ground rod, all bonded to a perimeter ground ring (4/0 bare Cu, 30" deep, 250 ft × 350 ft circuit).
Bonding: All metallic equipment within 6 ft of down conductors bonded — HVAC roof units, satellite dishes, ladders, fire escape, gas pipes.
Integration with electrical grounding: NFPA 780 ground ring bonded to building grounding electrode system (NEC 250.50). Single equipotential ground.
Surge protection: Type 1 SPDs at MV switchgear; Type 2 at 480V switchgear; Type 3 at PDUs. (See §24.)

i

Why DCs invest heavily in lightning protection

A direct strike on an unprotected building can drive 30,000-200,000 A into the structure. Without an LPS, that current finds whatever path exists — including through the IT equipment, network, and people. NFPA 780 LPS adds maybe 0.1% to building cost; insurance pays back many multiples in claim avoidance.

Worked Example 2 — Telecom Tower

Example 02 · Alternate context300-ft self-supporting telecom tower with rooftop equipment shed

Hazard: Tower height = highest object for miles → ~ 100% strike attractor. Strikes 5-15 times per year typical.
Lightning conductor: Tower steel itself acts as down conductor. Top of tower has air terminal mast extending 5+ ft above antennas. Steel must be electrically bonded throughout.
Tower base ground: Multiple ground rods radiating from tower base, all bonded together. Counterpoise (radial ground wires) extending 50-100 ft. Soil resistivity tested per IEEE 81.
Equipment protection: Coax cables from tower-top antennas pass through coax surge arresters (gas-discharge tube type) at the bulkhead entry to the equipment shed.
AC service grounding: Service neutral bonded to tower ground at single point only (NEC 250.30 separately derived rules apply). Avoids creating alternate paths for lightning current.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · NFPA 780 standard

Standard for lightning protection systems?

Drill 2 · Rolling sphere

Sphere radius for Class I structure?

Drill 3 · Air terminal spacing

Maximum spacing on protected periphery?

Drill 4 · Down conductor

Minimum number per structure?

Drill 5 · Bonding to LPS

Metallic equipment within how many ft must be bonded?

If You See THIS, Think THAT

If you see…	Think / use…
"NFPA 780"	Lightning protection system standard. Not in NEC.
"Air terminal" / "Franklin rod"	Vertical metal rod on roof — preferred attachment point for strikes.
"Rolling sphere method"	NFPA 780 design technique. 150 ft sphere for Class I; 100 ft for Class II.
"Down conductor"	Cu cable from air terminal to ground. 2/0 typical.
"Bonding to LPS"	Connecting nearby metal to down conductor. Prevents side-flash.
"Side flash"	Lightning jumps from down conductor through nearby metal — root cause of LPS failures.
"Counterpoise"	Radial ground wires extending from tower base. Disperses lightning current.
"LPL" (Lightning Protection Level)	IEC 62305 risk-based design. LPL I = highest protection.
"Coax surge arrester"	Gas-discharge or solid-state device protecting coax from lightning entering via antennas.
"Lightning flash density" or Nₐ	Strikes per km² per year. Used in risk assessment. Highest in southern US.
"ESE" (Early Streamer Emission)	Active air terminal — controversial. NFPA 780 doesn't recognize. UL doesn't list. Some jurisdictions accept.

§25 / 39

Surge Protection (SPDs)

NEC 285 · Type 1/2/3 SPDs · NEC 230.67 mandate · cascading layers

Surge Protective Devices clamp transient overvoltages from lightning, switching events, and motor stops. NEC 285 governs application. NEC 230.67 (2020+) now requires Type 1 or 2 at every service.

What SPDs Do

Surge Protective Devices clamp transient overvoltages from lightning, switching events, and motor stops. Without SPDs, transients reach equipment as 2-10× nominal voltage spikes for microseconds — destructive to electronics.

Source	Magnitude	Duration
Direct lightning strike (rare on building)	30,000 - 200,000 A	Microseconds, single shot
Indirect lightning (induced)	500 - 10,000 V transient	Microseconds
Utility switching	2-3× nominal V	Cycles to seconds
Capacitor switching	1.5-2× nominal V	~ 1 cycle
Inductive load switching (motor stop)	10s of kV depending on size	Microseconds
Welding equipment	kV transients	Repetitive

SPD Types — NEC 285

Type	Location	UL standard	Where used
Type 1	Line side of service disconnect (between utility transformer and service main)	UL 1449 Type 1	Hardwired to service entrance — handles direct lightning. Required by NEC 230.67 for some services.
Type 2	Load side of service disconnect — at service equipment or panelboard main	UL 1449 Type 2	Most common. Whole-building protection. Often combined with main breaker.
Type 3	Point-of-use — > 30 ft from service	UL 1449 Type 3	Plug strips, UPS input, sensitive equipment
Type 4	Component (no enclosure)	UL 1449 Type 4	OEM use inside equipment — not field-installed

NEC 230.67 — SPD Required at Every Service (2020+)

The 2020 NEC introduced 230.67, which requires Surge Protective Devices on most new services. The 2023 NEC expanded the requirement.

Service	NEC 230.67 (2020)	NEC 230.67 (2023)
Dwelling unit ≤ 1000V	Type 1 or Type 2 SPD required at service equipment	Same — required
Other occupancies	Not addressed	SPD required if essential systems present (e.g., fire pumps, life safety)
Industrial / commercial > 1000V	Not addressed by 230.67 (good practice still applies)	Same

SPD Ratings to Look For

Rating	Meaning	Typical values
kA per phase	Maximum surge current the SPD can handle (8/20 µs waveform)	40, 80, 120, 160, 200, 300, 400 kA per phase
VPR (Voltage Protection Rating)	Voltage that passes through SPD during a surge — what the equipment sees	Lower is better. 600V VPR for 480V service is excellent.
Nominal Voltage (V_n)	System voltage SPD is designed for	120, 208Y/120, 240, 480Y/277, 600Y/347, 12.47kV
MCOV (Maximum Continuous Operating Voltage)	Sustained voltage SPD can withstand without operation	~ 115% of nominal
SCCR (Short Circuit Current Rating)	Available fault current SPD can be installed in	10, 22, 65, 100, 200 kA
Nominal Discharge Current (I_n)	Current the SPD can repeatedly discharge without damage	5, 10, 20 kA per phase

Cascading SPDs

Best practice: layered protection. Type 1 at service handles the biggest surges; Type 2 at distribution panels reduces remaining; Type 3 at sensitive equipment provides final filtering.

Layer	Location	Typical kA	VPR
1 — Service entrance	Type 1 at MV switchgear or service equipment	120-300 kA	1500-2000V (480Y/277V)
2 — Distribution	Type 2 at major distribution panels	40-120 kA	1000-1500V
3 — Branch / point-of-use	Type 3 at sensitive equipment	10-40 kA	600-900V

SPD Technology — MOV vs SAD

Technology	Mechanism	Pros	Cons
MOV (Metal Oxide Varistor)	Zinc oxide ceramic — non-linear V/I	Cheap. Handles high energy. Self-resetting.	Wears out with surges. Eventual end-of-life. Visual indicator required (NEC).
SAD (Silicon Avalanche Diode)	Solid-state semiconductor	Faster clamping. Tighter VPR.	Lower energy capability. More expensive.
GDT (Gas Discharge Tube)	Spark gap in gas-filled tube	Very high current handling.	Slow response. High let-through during firing.
Hybrid	Combines MOV + SAD + filter	Best of all worlds.	Most expensive.

Worked Example 1 — Atlas DC1 SPD Cascade

Example 01 · Atlas DC1 spineLayered SPD protection from MV through to rack PDU

Location	SPD type	Spec	Why
12.47 kV MV switchgear	MV surge arrester (not "SPD" per UL — different standard)	15 kV class, 12 kV MCOV, 10 kA discharge	Handles direct lightning entry from utility primary
480V SWGR-A main	Type 1 SPD	200 kA per phase, VPR 1500V, hardwired ahead of main breaker	Whole-building protection. NEC 230.67 even though commercial.
480V distribution panels	Type 2 SPD	120 kA per phase, VPR 1200V	Reduces surges reaching downstream equipment
UPS-A1 input	Type 2 SPD	80 kA, VPR 1000V	Protects UPS rectifier (most sensitive component)
UPS-A1 output (415V)	Type 2 SPD	40 kA, VPR 800V	Protects critical IT downstream
PDU-A1 distribution panel	Type 2 SPD	40 kA, VPR 800V	Last stage before IT racks
Rack PDU strips	Type 3 SPD (integral)	10 kA, VPR 600V	Final point-of-use filtering
Generator paralleling cabinet	Type 2 SPD on each gen output	40 kA, VPR 1500V	Protects gen alternator from switching transients

i

SPD diagnostics matter

MOVs degrade with each surge. NEC 285.4 requires SPDs to have a status indicator (red/green LED). Atlas DC1 SPDs are monitored via Modbus/SNMP — when status changes, the BMS alerts maintenance for replacement before the next surge can reach equipment.

Worked Example 2 — Residential Service SPD (NEC 230.67 Compliance)

Example 02 · Alternate scaleSingle-family home · 200 A 1φ service · 2020+ NEC compliance

NEC 230.67 (2020): Type 1 or Type 2 SPD required at the service equipment of every dwelling unit. Mandatory.
Type 2 selection: 200 A panel, 120/240V 1φ. SPD ratings: 40-80 kA per phase, VPR ≤ 1500V on 240V (or ≤ 600V on 120V circuits).
Mounting: Hardwired to a 2-pole 30A breaker in the panel. Some panels have factory-installed integral SPD.
Status indicator: Green LED when active, red LED when end-of-life. Per NEC 285.4.
Cost: Whole-house SPD: $50-200 retail, $300-500 installed. Saves $$$ in claims after the first nearby strike.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · SPD types

Type 1, 2, 3 difference?

Drill 2 · NEC 230.67

Required at every dwelling service since which NEC?

Drill 3 · MOV vs SAD

Which has tighter VPR?

Drill 4 · VPR — meaning

Voltage Protection Rating: what does it mean?

Drill 5 · Cascade design

Layered SPDs — service kA vs point-of-use kA?

If You See THIS, Think THAT

If you see…	Think / use…
"SPD" or "Surge Protective Device"	NEC 285. Clamps transient overvoltages.
"NEC 230.67"	Mandatory SPD at every dwelling service since 2020 NEC. Some commercial in 2023.
"NEC 285"	Surge protective device application rules.
"Type 1 SPD"	Hardwired ahead of service disconnect. Handles biggest surges.
"Type 2 SPD"	Load side of service. Most common. Whole-building protection.
"Type 3 SPD"	Point-of-use. Plug strips, UPS input.
"VPR"	Voltage Protection Rating. Lower is better. What gets through SPD.
"MOV" (Metal Oxide Varistor)	Most common SPD technology. Wears out with each surge.
"SAD" (Silicon Avalanche Diode)	Faster, tighter clamping. Lower energy.
"Hybrid SPD"	MOV + SAD + filter. Premium.
"MCOV"	Max Continuous Operating Voltage. ~ 115% of nominal.
"SPD status indicator"	Red/green LED. Required by NEC 285.4. Visual end-of-life signal.
UL 1449	SPD product standard. Verify Type listing matches application.

§26 / 39

PV & Energy Storage

NEC 690 · NEC 705 · NEC 706 · 120% rule · rapid shutdown · ESS · NFPA 855

Solar PV is now standard on commercial buildings. NEC 690 (PV) and 705 (interconnection) govern. Energy Storage Systems (ESS) under NEC 706 are the fast-growing companion.

NEC Articles for Renewable Energy

Article	Covers
NEC 690	Solar PV systems — modules, inverters, DC wiring, rapid shutdown
NEC 691	Large-scale PV systems (≥ 5 MW utility-scale)
NEC 692	Fuel cell systems
NEC 705	Interconnected power production sources (PV + utility, ESS + utility, etc.)
NEC 706	Energy Storage Systems (ESS) — batteries, flywheels
NEC 712	DC microgrids
NEC 750	Energy management systems
NFPA 855	Standard for installation of stationary energy storage systems (fire safety)

PV System Architectures

Architecture	Description	Pros	Cons
String inverter	Multiple PV modules in series → single central inverter	Cheapest. Simple.	One module shading drops the whole string. No module-level data.
String + DC optimizer (per module)	String inverter + per-module DC-DC optimizer	Module-level mppt + monitoring. Better partial-shade tolerance.	Optimizer cost.
Microinverter (per module)	One inverter per module → AC immediately	Module-level redundancy. AC distribution simpler. Module-level monitoring.	Most expensive. Many small inverters to maintain.
Central inverter (utility-scale)	Large 1-2 MW inverters serving large arrays	Best economics at scale.	Single point of failure for large array.

The 120% Rule — NEC 705.12(B)(2)

When PV (or any source) is back-fed into a busbar that's also fed by utility, the rule says: (utility breaker rating) + (PV breaker rating) ≤ 120% of busbar rating. This prevents busbar overload during simultaneous full feed from both sources.

Example application

200 A bus + 200 A main + ___ A PV ≤ 120% × 200 = 240 A

Therefore PV ≤ 40 A. A 40 A breaker (back-fed PV) on a 200 A panel with a 200 A main is the maximum.

Rapid Shutdown — NEC 690.12

Required since 2014 NEC. Within 30 sec of activation (turning off a switch at building exterior or fire-alarm system signal), all DC voltage on the array conductors must drop to safe levels.

Voltage limit	Where measured	Time
≤ 30 V (2017+ NEC)	Within array boundary	30 sec from activation
≤ 80 V (between conductors and to ground outside array)	Outside array — connections to inverter	30 sec from activation

Achieved via module-level rapid shutdown devices (a small switch at each module that opens on signal loss) or string-level devices.

Energy Storage Systems (ESS) — NEC 706

NEC 706 (introduced 2017) covers stationary battery systems. Combined with NFPA 855 for fire safety. Li-ion is dominant chemistry now.

Application	Why ESS
Peak shaving (commercial demand)	Discharge battery during peak rate hours → reduce demand charges
Solar self-consumption	Store daytime PV → use at night
Backup power	Replace diesel genset for some applications
Frequency regulation (utility-scale)	Sub-second response for grid stability
Behind-the-meter (commercial & residential)	Reduce demand + provide backup combined

Worked Example 1 — Atlas DC1 PV + ESS

Example 01 · Atlas DC1 spine200 kW rooftop PV + 500 kWh ESS for peak shaving — interconnection design

PV system: 480 modules × 420 W = 200 kW DC. 4× 50 kW string inverters with module-level rapid shutdown devices.
Inverter output: 4× 50 kW = 200 kW AC at 480Y/277V 3φ.
Interconnection point: 480V SWGR-A bus (4000 A). 200 kW = 240 A back-fed.
120% rule check (NEC 705.12): 4000 A bus + 4000 A main + 240 A PV = 8240 A ≤ 1.2 × 4000 = 4800 A?? NO — fails.
Resolution options: (1) Reduce main breaker — not feasible, sized for full IT load. (2) Add a "supply-side connection" — interconnect PV ahead of the main breaker per NEC 705.11. Choose (2): PV breaker becomes a separate service-side disconnect.
ESS interconnection: 500 kWh Li-ion battery system in dedicated room, NFPA 855 fire-rated walls. 200 kW power conversion system (PCS) ties to UPS bus (or 480V bus depending on architecture).
Fire protection: NFPA 855 — Li-ion ESS requires sprinklers, gas detection, ventilation, max 50 kWh per Li-ion ESS unit (without separation), 3 ft separation between units.

i

Why DCs are aggressive on PV + ESS

A 2.5 MW data center pays $5K-50K/month in demand charges depending on rate. Peak shaving via ESS can cut this 30-50%. PV provides additional offset and ESG marketing value. Modern hyperscale DCs include 50-200 MW of co-located PV + 100+ MWh ESS.

Worked Example 2 — Residential Solar (NEC 705.12 Standard Application)

Example 02 · Alternate scaleSingle-family home · 200A panel · 8 kW PV system

PV array: 25 modules × 320 W = 8 kW DC. 7.6 kW AC inverter (single-string, 240V 1φ).
AC current: 7,600 / 240 = 31.7 A → use 40 A back-fed breaker.
120% rule check: 200 A bus + 200 A main + 40 A PV = 240 A ≤ 1.2 × 200 = 240 A. Just fits. ✓
Where to land the PV breaker: Opposite end of busbar from main breaker (NEC 705.12(B)(2)(3)b — minimum spacing for 120% rule).
Rapid shutdown: Per NEC 690.12, module-level rapid shutdown devices. Activated by AC service disconnect or fire alarm.
Labeling (NEC 705.10): Service entrance equipment requires a label noting "PHOTOVOLTAIC SYSTEM PRESENT" with locations.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · 120% rule

200 A bus + 200 A main + ___ A PV ≤ ?

Drill 2 · Rapid shutdown V

Inside array boundary, must drop to:

Drill 3 · NEC 690 article

PV system installation rules?

Drill 4 · ESS fire code

Standard for ESS installation?

Drill 5 · Supply-side connection

PV exceeds 120% rule allowance. Alternative?

If You See THIS, Think THAT

If you see…	Think / use…
NEC 690	Solar PV. Modules, DC wiring, inverters, rapid shutdown.
NEC 705	Interconnection of any power source (PV, gen, ESS) with utility.
NEC 706	Energy Storage Systems. Batteries, flywheels.
NFPA 855	Stationary ESS fire code. Spacing, ventilation, suppression.
"120% rule" / NEC 705.12(B)(2)	Bus + main + PV ≤ 120% of busbar. Limits back-fed PV.
"Supply-side connection" / NEC 705.11	Connect PV ahead of main breaker. Bypasses 120% rule. Becomes a service-side disconnect.
"Rapid shutdown" / NEC 690.12	30-sec drop to safe voltage at array boundary. Required since 2014.
"String inverter"	Multiple modules in series. Cheap. One module shaded = whole string affected.
"Microinverter"	One per module. Module-level redundancy. Premium.
"DC optimizer"	Module-level DC-DC + monitoring. Hybrid approach (string + module benefits).
"PV breaker back-fed"	Conducts power INTO panel from PV inverter. Sized for PV inverter rated AC current × 125%.
"IEEE 1547" / "UL 1741"	Inverter standards for utility interconnection. Anti-islanding requirements.
"Anti-islanding"	Inverter must shut off within 2 sec when utility loss detected. Prevents backfeeding into "dead" utility — keeps line workers safe.

§27 / 39

EV Charging

NEC 625 · L1/L2/DCFC · continuous load · EVEMS

EV charging is now standard on every commercial project. NEC 625 governs. Demand factors for EVSE differ from general receptacles. Energy Management Systems (EVEMS) reduce service-size requirements.

NEC Article 625 — EV Charging Equipment

EV charging is now standard on every commercial project. NEC 625 governs it. Demand factors differ from general receptacles. EVEMS (Energy Management Systems) reduce service-size requirements through load shedding.

Charging Level	Voltage	Current	Power	Use
Level 1	120V 1φ	12-16 A	1.4-1.9 kW	Residential, slow trickle. ~ 4-6 mi/hr added range.
Level 2	208V or 240V 1φ; some 480V 3φ commercial	16-80 A	3.3-19.2 kW	Standard residential + most commercial. ~ 10-60 mi/hr.
DC Fast Charging (DCFC)	480V 3φ in; DC out	varies	50-350+ kW	Highway corridor, fleet refueling. ~ 100-400 mi/hr.
"Megawatt Charging" (MCS)	1000V+ DC	3000+ A	1-3.75 MW	Heavy duty trucks, buses (emerging)

EV Charging Is Always Continuous

Per NEC 625.41, EV charging is classified as a continuous load regardless of duration. So 125% multiplier applies to wire and breaker. This is true even for a 30-min DCFC session.

Branch circuit sizing

OCPD ≥ 1.25 × (EVSE rated input current)

Example: 40A EVSE → 50 A breaker minimum, with #6 Cu THWN-2.

EVEMS — Energy Management Systems (NEC 625.42)

NEC 625.42 allows demand factor reduction via Energy Management Systems (EVEMS). Without EVEMS, sum of all EVSE branches is treated as 100% continuous. With EVEMS, the system can dynamically limit total simultaneous charging power → service size much smaller.

Approach	Demand calculation	Service size impact
No EVEMS — full simultaneous	100% × N stations × max kW each × 1.25 continuous	Largest. 50 stations × 7.2 kW = 360 kW + 1.25 = 450 kW
EVEMS — dynamic load sharing	Configured maximum kW (sum < service capacity)	Smallest. EVEMS limits total to e.g. 100 kW shared across all stations
EVEMS — load-shedding hierarchical	EVEMS sheds EV load when other building loads peak	Allows EV charging on tight existing services

DCFC — Special Considerations

Aspect	Detail
Power level	50, 100, 150, 175, 250, 350 kW per stall typical. 480V 3φ input.
Service requirement	A 4-stall 350 kW DCFC site = 1.4 MW peak. Often requires utility upgrade.
Demand factor	Per NEC 625.42(B), allowable diversity for > 1 station based on charging session statistics. Real-world: rarely all stalls full at full power.
Harmonic content	DCFC is a large rectifier — significant harmonics. Often passive or active filter required at site to meet IEEE 519.
Fault current	Service often upgrades fault current at site. Equipment AIC must accommodate.
Coordination with utility	For sites > 250 kW, often requires custom rate + demand charge structure. Utility approval lead time.

Worked Example 1 — Atlas DC1 EV Charging Station

Example 01 · Atlas DC1 spineAtlas DC1 office building parking — 4 Level 2 + 1 DCFC for fleet

Equipment list

Item	Qty	Spec	Branch
Level 2 EVSE — staff parking	4	208V 1φ, 40 A continuous	50 A breaker, #6 Cu in 1" EMT
DCFC — fleet vehicles	1	480V 3φ, 75 kW (90 A input)	125 A breaker, 1/0 Cu in 1.5" EMT

Service impact analysis

Without EVEMS — full simultaneous load:
4 × (208 × 40 × 1.25) + 75 × 1.25 / 0.95 = 41.6 + 99 = 140 kW peak
With EVEMS: Shed Level 2 stations during DCFC session. Max total = 75 kW DCFC OR 4 × 8.3 kW = 33 kW Level 2 = max 75 kW peak (one or other).
Atlas DC1 chose EVEMS: Existing service has limited EV capacity allocation. EVEMS reduces peak from 140 to 75 kW = 50% smaller transformer needed for EV.
Branch from 480V SWGR-A: Dedicated EV panel fed from 480V SWGR. 200 A branch CB, 4/0 Cu sub-feeder to EV panel.

Worked Example 2 — Highway Corridor DCFC Site

Example 02 · Alternate contextStandalone highway DCFC site — 4 stalls × 350 kW each

Site demand: 4 × 350 kW = 1.4 MW peak. With 1.25 continuous = 1.75 MW.
Service: Customer-owned 12.47 kV utility service with on-site step-down. 2 MVA pad-mount transformer at 480V.
Switchgear: 480V switchgear with 4 × 600 A feeder breakers (one per DCFC stall).
EVEMS: Often NOT used for highway DCFC because customers expect full power on demand. Service sized for full simultaneous use.
Harmonics: 4 × 350 kW DCFC = significant 5th and 7th harmonic injection. Active filter required to meet IEEE 519 at PCC.
Utility coordination: 12+ months lead time. Site-specific demand rate negotiation.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Continuous classification

EV charging is what kind of load per NEC 625.41?

Drill 2 · Level 2 voltage

Standard Level 2 EV charging voltage?

Drill 3 · DCFC voltage

DC Fast Charging input?

Drill 4 · EVEMS purpose

What does EVEMS reduce?

Drill 5 · EVSE vs EV

Where is the actual battery charger?

If You See THIS, Think THAT

If you see…	Think / use…
NEC 625	EV charging equipment installation rules.
"Level 1" charging	120V residential. 1.4-1.9 kW. Slow.
"Level 2" charging	208V or 240V. 3.3-19.2 kW. Standard residential + most commercial.
"DCFC" or "Level 3"	480V 3φ in, DC out. 50-350+ kW. Highway corridor.
"EVSE"	Electric Vehicle Supply Equipment. The "charger." (Actual battery charger is in the car.)
"EVEMS"	Energy Management System per NEC 625.42. Limits total simultaneous EV charging power. Allows smaller services.
EV charging classified as	ALWAYS continuous load. 125% rule applies.
"4-stall 350 kW DCFC site"	1.4 MW peak. Customer-owned MV service required typically.
NEMA 14-50 outlet	Common Level 2 receptacle (50 A 240V).
"Anti-islanding for V2G"	Vehicle-to-Grid bidirectional charging. Inverter standards apply.
"OCPP"	Open Charge Point Protocol. Industry standard for EVSE communication.

§28 / 39

Demand Response & Load Shedding

Load priority tiers · ATS-based shedding · PMS · utility programs

When the system can't carry full load (utility curtailment, generator transfer, scheduled maintenance), shed loads in priority order. ATS-based shedding works for simple cases; PMS handles complex prioritization.

Why Shed Load?

Trigger	What's happening	Action
Utility outage → genset on	Genset capacity may be less than full building load	Shed non-essential to keep gen within rating
Genset failure during outage	Remaining gen capacity insufficient	Cascading shed to match remaining capacity
Utility demand response event	Utility paying customers to reduce demand during peak	Voluntary shed of pre-defined loads for incentive payment
Peak shaving (cost optimization)	Demand charges based on monthly peak kW	Automatic shed during forecast peaks; ESS discharge fills gap
Scheduled maintenance	Switchgear or transformer offline	Pre-shed loads on the affected feeder
Equipment overload	Local feeder approaching capacity	Shed lowest-priority load on that feeder

Load Priority Tiers

Tier	Description	Examples
1 — Life Safety	NEC 700 — never shed	Egress lighting, fire alarm, fire pumps, smoke control
2 — Critical Process	Mission-critical loads	IT (data center), surgical (hospital), refrigeration (food storage)
3 — Important	Significant disruption if dropped	HVAC for occupied spaces, security systems, communication
4 — Optional	Comfort, convenience	Office HVAC, parking lot lighting, EV charging, decorative lighting
5 — Sheddable	First to shed; comfortable to lose	Forecast HVAC pre-cooling, EV charging during peak, water heaters

Implementation — How Shedding Actually Works

Method	Description	Where used
ATS-based shed	Auxiliary contacts on ATS open shedding contactors when on genset position	Simple emergency systems, hospitals, small DCs
PMS (Power Management System)	Centralized controller monitors all loads + sources, dynamically prioritizes	Large DCs, complex industrial, hyperscale
Smart panels / EVEMS	Panel-level controllers shed branch circuits based on programmed priority	Modern commercial, residential demand response
Frequency-based shed	Underfrequency relays (81U) drop loads when genset starts to slow under load	Last-resort shed when other systems fail
Utility load control switches	Utility-installed device that cycles AC compressor or water heater on demand	Residential utility programs (often opt-in for rate discount)

Utility Demand Response Programs

Program	How it works	Customer benefit
Time-of-Use (TOU)	Higher rates during peak hours	Shift consumption to lower-rate periods
Critical Peak Pricing	Even higher rates on critical days (5-15 days/yr)	Major reduction in usage during called events
Direct Load Control	Utility cycles HVAC or water heater during emergencies	Reduced rate; some loss of comfort
Demand Response (DR)	Utility pays for committed reduction during event (1-100 events/yr)	Significant payment for reliable reduction
Real-time pricing	Wholesale market price passed through hourly	Sophisticated customers shed when price spikes
Capacity programs	Customer commits to be available for grid needs (4 hr advance notice)	Annual capacity payment + per-event payment

Worked Example 1 — Atlas DC1 Load Shed Sequence

Example 01 · Atlas DC1 spineUtility loss → genset start → load shed sequence (Side A)

Time sequence

T (sec)	Event	Loads on
T = 0	Utility power lost on Side A	UPS-A1, UPS-A2 ride through on battery. Mech loads off (no switchgear power).
T = 1	ATS-A senses utility loss, signals GEN-A to start	Same — UPS still on battery. Cooling beginning to lose pressure.
T = 10	GEN-A starts and reaches rated voltage + frequency	Same.
T = 12	ATS-A closes to genset position	Side A bus re-energized from gen. UPS rectifiers come back on; battery resting.
T = 13	Load priority controller checks: can GEN-A carry full Side A load? Yes (2,500 kW gen vs 2,300 kW demand). No shed needed normally.	Full load.
T = 13 (alternative)	If GEN-A capacity insufficient: shed CRAH fans (Tier 4) → 240 kW reduction	UPS + chillers + critical loads only.
T = 30	Chiller plant restart sequence begins (CH-1 → CWP-1 → CRAH return)	Cooling restored. IT load uninterrupted throughout.

Why this works: UPS battery ride-through (5 min) + genset (10 sec start) = 30 sec total cooling outage. IT thermal mass tolerates this without shutting down.

Worked Example 2 — Commercial Building Peak Shaving via ESS

Example 02 · Alternate context200,000 sq ft office — peak demand 800 kW · ESS + load shed for cost reduction

Demand profile: Peak 800 kW occurs 2-5 PM weekdays in summer (HVAC plus afternoon office). Off-peak: ~ 250 kW.
Demand charge: $20/kW/mo. Annual demand cost = 800 × $20 × 12 = $192,000.
Strategy: 500 kWh / 250 kW Li-ion ESS + automated load shed.
Discharge profile: ESS discharges 250 kW for 2 hr during peak (3-5 PM) → reduces measured peak from 800 to 550 kW.
Load shed (back-up): If ESS depleted, shed parking lot lighting (50 kW) + half of office HVAC (200 kW) for last 30 min of peak hour.
Savings: Peak reduction 250 kW × $20 × 12 = $60K/yr saved. Plus ESS arbitrages day/night rates: ~$15K/yr. Total: ~$75K/yr. Payback ~ 5-7 yr.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Tier 1 loads

What load tier is NEVER shed?

Drill 2 · Why peak shave

Why is reducing peak demand valuable?

Drill 3 · ATS vs PMS

Centralized controller for complex shedding?

Drill 4 · Underfrequency

What ANSI device sheds load when generator slows under load?

Drill 5 · Atlas first to shed

What load class sheds first in Atlas DC1?

Demand Charges — How They Work + Math

Most commercial + industrial utility tariffs include both an ENERGY charge ($/kWh) and a DEMAND charge ($/kW). Demand is what drives peak shaving + load shedding economics.

How Demand Is Measured

Metric	Definition
Demand interval	15-min, 30-min, or hourly window over which average kW is computed
Monthly peak	Highest demand interval value during the billing month
Ratchet clause	Some tariffs lock the billed peak to the highest of the past 11 months — one summer peak charges all year
Demand charge	Monthly peak (kW) × $/kW rate
Time-of-Use (TOU) demand	Different $/kW rates for on-peak vs off-peak hours

Worked Example — Peak Shaving Math

Example · Office buildingDemand reduction via ESS + load shed — annual savings calc

Baseline tariff

Demand rate

$22 / kW / month (on-peak window 12pm-8pm weekdays)

Energy rate

$0.09 / kWh on-peak; $0.05 / kWh off-peak

Ratchet

Billed peak = max(current month, 0.75 × past 11-month peak)

Building peak

800 kW (summer, 3 PM)

Without intervention

Annual demand cost

800 kW × $22 × 12 = $211,200

Annual energy cost

~ $180,000 (varies by usage profile)

Total annual

~ $391,200

With 250 kW × 2-hr ESS

ESS discharge during peak

250 kW for 2 hr → reduces measured peak by 250 kW

New measured peak

800 − 250 = 550 kW

New annual demand cost

550 × $22 × 12 = $145,200

Annual demand savings

$211,200 − $145,200 = $66,000 / year

Plus arbitrage savings (energy)

ESS charges off-peak

250 kW × 2 hr × 365 days × ($0.09 − $0.05) − round-trip losses (15%) = ~ $6,200/yr

ROI

Total annual savings

$66,000 + $6,200 = $72,200 / year

ESS install cost (500 kWh / 250 kW Li-ion)

~ $400,000 (2026 prices)

Simple payback

~ 5.5 years

i

Why ESS often pencils out for commercial

Demand charges are typically 30-50% of a commercial bill. Even small peak reductions deliver large $/yr savings. Combined with utility incentives (often 30-50% capex rebate) + ITC/MACRS depreciation, real payback can drop to 3-4 years.

If You See THIS, Think THAT

If you see…	Think / use…
"Demand response" / DR program	Utility pays for load reduction during peak. 1-100 events/yr typical.
"Load shedding"	Dropping loads in priority order. Triggered by overload, gen capacity, or utility request.
"Peak shaving"	Reducing peak demand charge via ESS, shed, or generation.
"PMS" (Power Management System)	Centralized controller. Required for complex systems > 2 MW typically.
"Time-of-Use" rate	Different prices throughout the day. Drives arbitrage opportunities.
"Tier 1 load"	Life safety. NEVER shed.
"Underfrequency shedding" (81U)	Last-resort load shed when generator can't keep frequency.
"Direct Load Control"	Utility-installed device for residential AC/water heater. Opt-in.
"Critical Peak Pricing" (CPP)	Utility rate event that may happen 5-15 days/yr. Major price increase.
EVSE on commercial service	Often the biggest sheddable load. EVEMS implements automatic shed.

§37 / 32

BESS Deep Dive

LFP vs NMC · PCS · BMS · NEC 706 · NFPA 855 · use cases

Battery Energy Storage Systems serve a different role than UPS batteries — energy arbitrage, peak shaving, grid services over hours instead of minutes. LFP chemistry won this market. NEC 706 + NFPA 855 + UL 9540 govern installation.

BESS — Beyond UPS Batteries

Battery Energy Storage Systems (BESS) at utility scale serve a different purpose than UPS batteries. UPS = ride-through during outages (minutes). BESS = energy arbitrage, peak shaving, frequency regulation, renewable smoothing (hours to days).

	UPS battery (§20)	BESS
Purpose	Bridge utility loss until generator starts	Energy arbitrage, peak shaving, grid services
Discharge duration	5-15 minutes	2-4 hours typical (some 8+ hours)
Cycle frequency	Rare (hopefully never beyond float)	Daily (hundreds of cycles/year)
Chemistry	VRLA (Atlas DC1), Li-ion (modern UPS)	Almost universally Li-ion (LFP preferred for cycle life)
Sizing	Ah for full IT load × duration	kWh for energy stored + kW for power output
Round-trip efficiency	~ 90% (mostly idle)	~ 85-90% (cycled regularly)
Code	NEC 480 + NFPA 855	NEC 706 + NFPA 855 (more rigorous than 480)

The Stack — From Cell to BESS

Level	What it is	Atlas DC1 example (500 kWh BESS for peak shaving)
Cell	Single LFP cell. ~ 3.2V, 50-300 Ah depending on form factor.	~ 50 Ah pouch cell × ~ 300 cells
Module	Pre-assembled group of cells (e.g., 16-24 cells in series), with monitoring + balancing	16-cell module = 51.2V × 50 Ah = 2.5 kWh
Rack	Vertical assembly of modules + BMS (Battery Management System)	10 modules per rack × 2.5 kWh = 25 kWh per rack
Container	20 ft or 40 ft ISO container with multiple racks + thermal management + fire suppression	20 racks per container = 500 kWh container (roughly the Atlas DC1 size)
PCS (Power Conversion System)	DC-AC inverter that connects BESS to AC bus. Bidirectional.	250 kW PCS for Atlas DC1 BESS
EMS (Energy Management System)	Software optimizing when to charge/discharge based on rates, signals, forecasts	Peak shaving algorithm that learns building load profile
SCADA	Operator interface; integrates with BMS + EMS + utility communications	—

Lithium Chemistry Comparison — Why LFP Won for BESS

Chemistry	Energy density	Cycle life	Thermal stability	Cost (2026)	Use
NMC (Nickel Manganese Cobalt)	HIGH (200-265 Wh/kg)	1,000-2,000 cycles	LOWER (thermal runaway risk)	~ $130/kWh	EVs (higher density needed for range)
LFP (Lithium Iron Phosphate)	LOWER (90-160 Wh/kg)	3,000-6,000+ cycles	HIGHER (much safer)	~ $80/kWh	BESS standard. Better safety + cycle life trumps density for stationary storage.
LTO (Lithium Titanate)	LOWEST (50-80 Wh/kg)	10,000+ cycles	VERY HIGH	~ $250/kWh	Niche (high-cycle apps; grid frequency regulation)
NCA (Nickel Cobalt Aluminum)	HIGH (200-260 Wh/kg)	1,000-2,000	MODERATE	~ $150/kWh	EV (Tesla)
Solid-state	VERY HIGH (300+ Wh/kg projected)	5,000+ projected	VERY HIGH	Not commercial yet	Future EV + premium BESS

BESS Use Cases

Use case	How BESS earns its keep	Discharge cycle
Peak shaving (commercial)	Discharge during peak demand hours → reduce demand charge ($)	2-4 hours daily during peak window
Energy arbitrage	Charge off-peak ($), discharge on-peak ($) → captures rate spread	Daily
Demand response participation	Utility pays for committed reduction during called events	1-100 events/year, 1-4 hours each
Solar self-consumption	Store daytime PV → use at night (when PV not generating)	Daily
Backup power	Replace or supplement diesel genset	Rare (during outages)
UPS augmentation	Extend ride-through beyond traditional UPS battery	Rare
Frequency regulation (utility-scale)	Grid operator pays for sub-second response to frequency excursions	Constant micro-cycles (millions/year)
Voltage support	Inject/absorb reactive power to stabilize voltage	Continuous (low energy throughput)
Renewable smoothing	Smooth wind/PV output to meet contractual ramp limits	Continuous (small but constant cycling)
Microgrid islanding	Maintain power to a microgrid when disconnected from utility	Variable (depends on renewable + load)

NEC 706 + NFPA 855 — Code Requirements

Code/Standard	Requirements
NEC 706.20 — Disconnects	Each ESS unit must have a readily accessible disconnect for emergency service
NEC 706.21 — Overcurrent protection	Both DC and AC sides protected; sized for rated current
NEC 706.31 — Grounding	Per NEC 250; some chemistries require special grounding considerations
NFPA 855 §8 — Spacing	3-ft separation between Li-ion ESS units (some local jurisdictions require more)
NFPA 855 §9 — Containment	Ventilation, drainage for thermal runaway gases
NFPA 855 §12 — Detection + suppression	Heat + smoke detection; Class C-rated suppression (NOT water for Li-ion)
UL 9540 + UL 9540A	System-level listing (9540) + thermal runaway test (9540A) — required by AHJ for permitting
NEC 706.40 — Safety controls	BMS cutoff for over-charge, over-discharge, over-current, over-temperature

Worked Example — Atlas DC1 BESS for Peak Shaving

Example · Atlas DC1 spine500 kWh / 250 kW LFP BESS for peak shaving + ride-through augmentation

Why this BESS

Demand shaving: Atlas DC1 utility tariff has $22/kW demand charge. Reducing peak by 250 kW saves $66K/year.
Augmenting UPS: Could discharge to UPS bus during extended outages (beyond 5-min battery ride-through), bridging gap to fuel resupply for gensets
Future PV pairing: When Atlas DC1 adds 200 kW rooftop PV, BESS stores excess for night use

System spec

Energy

500 kWh (2 hr at 250 kW discharge)

Power

250 kW continuous, 350 kW peak

Chemistry

LFP (lithium iron phosphate) — safer + longer cycle life than NMC

Form factor

Outdoor 20-ft ISO container (Tesla Megapack-style)

PCS

250 kW bi-directional inverter (480Y/277V grid tie)

Round-trip efficiency

~ 88%

Cycle life

6,000 cycles to 80% capacity (~ 16 years at daily cycle)

Tie point

480V SWGR-A side (or shared via cross-tie)

Capex (2026)

$300-400K (system installed)

Operational logic

Charge: 11 PM - 4 AM (off-peak rate)
Standby: 4 AM - 12 PM (idle, monitoring)
Discharge (peak shaving): 12 PM - 8 PM weekdays summer
Emergency: If utility loss + UPS battery depletes before genset restored, BESS bridges to UPS bus (manual switchover, not seamless)

Annual savings

Peak demand reduction

250 kW × $22 × 12 = $66,000/yr

Energy arbitrage

~ $6,000/yr (off-peak/on-peak spread × duty cycle × efficiency)

Total

~ $72,000/yr

Simple payback

$300-400K / $72K = 4.2-5.6 years

i

Why BESS economics keep getting better

Li-ion cell prices dropped from $1,200/kWh in 2010 to under $100/kWh in 2024. By 2030, BESS at scale is projected at $50/kWh. Combined with utility incentives + ITC tax credit (30% federal in US through 2032), real payback can drop to 2-3 years for commercial.

If You See THIS, Think THAT

If you see…	Think / use…
"BESS"	Battery Energy Storage System. Utility-scale or commercial. NEC 706 + NFPA 855.
"LFP" (Lithium Iron Phosphate)	BESS standard chemistry. Safer + longer cycle than NMC.
"NMC" (Nickel Manganese Cobalt)	EV chemistry. Higher density but more thermal runaway risk.
"PCS" (Power Conversion System)	DC-AC bidirectional inverter. Connects BESS to AC bus.
"BMS" (Battery Management System)	Per-cell monitoring + balancing + safety cutoffs
"EMS" (Energy Management System)	Software optimizing charge/discharge schedule
"UL 9540" / "UL 9540A"	System listing + thermal runaway test. Required by AHJ.
"NFPA 855"	Stationary ESS fire safety standard. Spacing, ventilation, suppression.
"Megapack" / "PowerPack"	Tesla container BESS products (Megapack = 1.9 MWh; PowerPack discontinued)
"Round-trip efficiency"	Energy out / energy in. ~ 85-90% for Li-ion.
"State of Charge (SoC)"	Current battery charge as % of capacity
"Cycle life"	Number of full charge/discharge cycles before capacity drops to 80% of original
"Frequency regulation"	Sub-second BESS response to grid frequency. Highest-value utility service.

§35 / 32

Liquid Cooling Electrical

DLC · RDHx · immersion · CDU · busway · UPS implications

Air cooling stops at ~ 20 kW/rack. AI/HPC workloads need 30-100+ kW/rack. Liquid cooling becomes mandatory — and it changes the electrical design significantly. Branch circuits become busway. UPS topology shifts. Server power density increases.

Why Liquid Cooling Now

Air cooling worked great when servers drew 5-10 kW per rack. Modern AI/HPC workloads (GPU clusters) push densities to 30-100+ kW per rack. Air cannot remove heat at this density without impractical airflow rates. Liquid cooling becomes mandatory.

Cooling type	Max rack density	Typical PUE	Industry use
Traditional CRAC + raised floor	~ 10 kW/rack	1.6-2.0	Legacy DCs
Hot aisle / cold aisle containment	~ 20 kW/rack	1.4-1.6	Standard modern DCs (Atlas DC1)
In-row cooling	~ 30 kW/rack	1.3-1.5	Mid-density colocation
Rear-door heat exchanger (RDHx)	40-50 kW/rack	1.2-1.3	Higher-density traditional + early AI
Direct liquid cooling (DLC) — cold plates	50-100+ kW/rack	1.1-1.2	NVIDIA H100/H200 clusters, custom AI accelerators
Immersion cooling — single-phase	100-200+ kW/rack	1.05-1.10	Hyperscale AI training (Microsoft, Meta)
Immersion cooling — two-phase	200-400+ kW/rack	1.02-1.05	Cutting-edge research (3M Novec)

Liquid Cooling Architectures

Rear-Door Heat Exchanger (RDHx)

A liquid-cooled coil mounted on the back of the rack. Hot exhaust air passes through the coil before returning to the room — heat transferred to chilled water. Server fans still push air; servers remain air-cooled.

Aspect	RDHx detail
Cooling capacity	30-50 kW per rack typical
Server modifications	None — works with stock air-cooled servers
Plumbing	Each rack needs supply + return chilled water connections
Failure mode	If coil fails, hot air dumps into room — adjacent racks may overheat
Water leakage protection	Drip pans + leak detection sensors at rack level
Electrical impact	None directly — server power same as air-cooled
Best for	Density bumps without liquid in IT room (water still in coil only)

Direct Liquid Cooling (DLC) — Cold Plates

Coolant circulated through metal cold plates mounted directly on hot components (CPU, GPU, memory). Coolant absorbs heat at the chip and carries it to a CDU (Coolant Distribution Unit) that exchanges with facility chilled water.

Aspect	DLC detail
Cooling capacity	50-100+ kW per rack
Server modifications	Required — server vendor builds with DLC option (NVIDIA HGX H100, Intel Xeon Max, AMD EPYC liquid)
Coolant types	Treated water (most common), water-glycol, dielectric fluids (3M Novec 7000)
CDU (Coolant Distribution Unit)	Heat exchanger between server-side coolant loop and facility chilled water; pumps + filtration
Manifolds	Plumbing inside each rack distributes coolant to server cold plates
Quick disconnects	Drip-free quick disconnects allow server pull/swap without draining the loop
Electrical impact	Reduces server fan power → ~ 5-10% IT power reduction → also reduces total facility power
Adoption (2026)	Standard for new AI deployments; retrofit of air-cooled facilities is complex

Immersion Cooling

Servers fully submerged in dielectric fluid. Fluid removes heat directly from all components. No fans, no dust, no humidity issues. Single-phase keeps fluid liquid throughout; two-phase boils at chip temperature for higher heat transfer.

Aspect	Single-phase immersion	Two-phase immersion
Coolant	Mineral oil, synthetic dielectric (Engineered Fluids ElectroSafe)	3M Novec 7000-series (boils at 34-61°C)
Heat transfer	Convection	Phase change (boiling) — higher coefficient
Density	100-200 kW/rack	200-400+ kW/rack
PUE	~ 1.05-1.10	~ 1.02-1.05
Server modifications	Remove fans, replace thermal paste with immersion-rated, optionally remove HDDs (use SSDs only)	Same + heat-spreader plates on chips for boiling surface
Adoption	Growing (research + early hyperscale)	Limited (cost + complexity)
Concern	Fluid procurement, disposal, environmental (PFAS regulations on Novec)	Same + 3M discontinuing some Novec products

Electrical Implications of Liquid Cooling

Implication	Detail
Higher rack power → busway not branch circuits	At 50-100 kW/rack, conventional branch circuits become impractical. Use bus duct (NEC 368) running down each row with plug-in tap-offs at each rack.
CDU electrical load	Each CDU has its own pump (10-30 kW typical) — adds to mech load, fed from PDU
Reduced server fan power	~ 5-10% reduction in IT power (server fans gone or minimal). Improves PUE.
Different IT redundancy model	DLC servers cannot tolerate even brief power loss — coolant pump must continue. Requires UPS for both server AND CDU.
Leak detection	Required at every CDU + manifold + rack. Tied to BMS for alarms; some systems auto-shutoff valve on leak.
Plumbing-electrical separation	Water near electrical = bad. Code-compliant separation (NEC 110.26 working space, drip pans, sub-floor drainage)
Hot water reuse	DLC return water at 35-50°C is hot enough for building heat reuse — improves ERE metric
Facility chilled water temp can be HIGHER	Air-cooled DC needs 7°C chilled water. DLC works with 30-40°C — enables free cooling year-round in moderate climates

Worked Example — Atlas DC1 Future AI Hall

Example · Atlas DC1 spineRetrofitting one of Atlas DC1's IT halls for AI workloads with DLC

Current state (one row of Atlas DC1 IT Hall A)

Existing density

12 kW/rack × 104 racks = 1.25 MW

Cooling

Cold aisle containment, CRAH air-cooled

Branch circuits

42-circuit RPP per row, 30A branches

AI conversion target

Target density

60 kW/rack with DLC

Target rack count

21 racks (instead of 104) — 60 × 21 = 1.26 MW (same total)

Cooling

DLC with CDUs for each rack pair

Power distribution

2,000 A busway down row instead of 42-circuit panel

Electrical changes required

Replace RPP with busway. Existing 400 A panelboard → install 2,000 A busway (Square D Powerlink or Eaton Pow-R-Line) along ceiling of row
Plug-in switches per rack. Each rack gets 100 A plug-in fused disconnect (60 kW × 1.25 = 75 kVA / 415V × √3 = 104 A)
Re-route CDU power. Each CDU draws ~ 20-30 kW; fed from same busway or separate PDU
UPS sizing. Total row load = 21 × 60 kW + 11 CDU × 25 kW = 1,535 kW. Existing UPS-A1 sized for 1,250 kVA — undersized for AI conversion. Need to upsize UPS or split row across both UPS sides.
Plumbing. Run chilled water mains with 30°C supply (warmer than existing 7°C — can reduce chiller energy)
Leak detection. Add water sensors under raised floor per rack. Auto-shutoff valves at row level.
Cost estimate for one row conversion: $500K-1M (busway, plumbing, CDUs, leak detection, network upgrade, sub-flooring)

i

Why retrofitting is expensive vs greenfield AI DC

A purpose-built AI data center is designed from day one for liquid cooling. Atlas DC1 was built for air. Retrofitting requires running plumbing through existing finished space, replacing distribution equipment, and disrupting operations. Modern hyperscale AI campuses are built liquid-cooled from the start.

If You See THIS, Think THAT

If you see…	Think / use…
"DLC" (Direct Liquid Cooling)	Cold plates on chips. 50-100 kW/rack. Most common modern AI cooling.
"RDHx" (Rear-door heat exchanger)	Coil on back of rack. 30-50 kW/rack. Air still flows through servers.
"Immersion cooling"	Servers submerged in dielectric fluid. 100-400 kW/rack.
"CDU" (Coolant Distribution Unit)	Heat exchanger + pump between server-side loop and facility chilled water
"Quick disconnect"	Drip-free coupling allowing server pull without draining loop
"Two-phase immersion"	Coolant boils at chip surface (Novec). Highest density. Newest.
"Bus duct" / "busway" in IT halls	For 30+ kW/rack. Branch circuits don't scale that high.
"Chilled water 30°C return"	DLC enables this. Massive PUE improvement vs traditional 7°C.
"PFAS regulations on Novec"	Two-phase immersion fluids facing regulatory pressure (3M phasing out)
"Heat reuse" in DC context	DLC return water hot enough to heat adjacent buildings

§36 / 32

AI/HPC Data Center Design

AI workloads · 30-100+ kW racks · DLC · InfiniBand · SuperPODs

AI/HPC workloads have flipped data center design. Atlas DC1 is air-cooled and traditional. AI clusters need liquid cooling, busway distribution, sub-millisecond network fabric, and 5-100× the power density. This section maps the gap.

What Makes AI/HPC Different

Atlas DC1 was designed for traditional cloud + enterprise workloads — distributed servers, mixed densities, web/database work. AI/HPC is fundamentally different. Training a single large language model can use 25,000+ GPUs in tightly-coupled clusters with sub-millisecond network coordination. The design constraints flip:

Dimension	Traditional DC (Atlas DC1)	AI/HPC
Rack density	~ 12 kW/rack	30-100+ kW/rack (training); up to 200+ for inference
Cooling	Air (CRAH + containment)	Liquid (DLC, immersion) — mandatory
Power per row	1-1.5 MW	5-20+ MW
Network	10-100 Gbps Ethernet, ms latency OK	InfiniBand or NVLink, sub-ms latency required
Workload pattern	Bursty (web requests come/go)	Constant (training runs for weeks at full load)
Failure tolerance	Application-level (web servers fail individually)	Cluster-level (one server failing can halt 1000-server training run)
Power continuity	UPS ride-through 5 min OK	Same — but checkpointing failures can cost days of training time
PUE target	1.3-1.5	1.05-1.2
Capital cost / MW	$15-22M/MW	$25-50M/MW (cooling + network premium)
Build timeline	18-24 months	24-36 months (custom mech, complex commissioning)

The AI Compute Stack — What's Inside an "AI Cluster"

Layer	Component	Power per unit
Accelerator	NVIDIA H100 (700W), H200 (1000W), B100/B200 (~ 1200W), AMD MI300 (750W), custom (Google TPU, AWS Trainium, Meta MTIA, Microsoft Maia)	700-1200W per chip
Server (DGX-style)	8 GPUs + 2 CPUs + memory + NICs	10-12 kW per server (NVIDIA HGX H100 = 10.2 kW)
Rack	4-8 servers per rack (with cooling)	40-80+ kW/rack
Pod / Cluster	16-128 racks tightly coupled by InfiniBand	1-10+ MW per pod
SuperPOD / SuperCluster	Multiple pods coordinated for very large training (NVIDIA SuperPOD = 32-127 DGX systems)	10-50+ MW per SuperPOD
Hyperscale AI campus	Multiple SuperPODs (xAI Memphis = 100,000+ H100 = 100+ MW)	100 MW - 1+ GW

The Network Fabric — Why It Matters for Power

AI training requires GPUs in different servers to share gradients millions of times per second. Standard Ethernet has too much latency. Two competing technologies:

Technology	Use	Power impact
NVLink (NVIDIA)	GPU-to-GPU within and between servers — 900 GB/s per link, sub-microsecond latency	Switch racks (NVLink switches) consume 5-20 kW each
InfiniBand (Mellanox/NVIDIA)	Server-to-server within pod — 400 Gbps per port, microsecond latency	IB switches consume 1-3 kW each
Ethernet (RoCE)	Alternative for scale-out; emerging Ultra Ethernet	Lower than InfiniBand
Optical interconnect	Cross-pod cabling at 800 Gbps+ optical	Optical transceivers add 10-30W per port

For a 10 MW AI cluster, the network fabric alone can consume 5-10% of total power — not negligible.

Power Distribution Architecture for AI/HPC

Element	Atlas DC1 (traditional)	AI/HPC equivalent
Service voltage	12.47 kV utility	Same OR higher (138 kV for hyperscale campuses)
Service transformers	2 × 2,500 kVA	Multiple 5-30 MVA transformers (per pod)
Distribution voltage	480Y/277V to 415Y/240V	480V or 415V → some hyperscale exploring 800V DC for direct-to-server feed
Per-row distribution	RPP panelboard (400 A)	Bus duct (2,000-4,000 A)
Per-rack delivery	30-60A branch circuits	100-225A plug-in disconnect from busway
UPS ride-through	5 minutes	Same OR shorter (some designs use rotary UPS for inertia + ride-through)
Redundancy	2N (dual-fed servers)	2N OR distributed redundant (4N3) at hyperscale; some accept N+1 at module level
Cooling power	~ 30% of IT	~ 5-15% of IT (DLC much more efficient)

Worked Example — A 10 MW AI Pod

Example · NVIDIA HGX H100 SuperPODSizing electrical for a typical AI training cluster

The cluster

Compute

128 × HGX H100 servers (8-GPU each) = 1,024 H100 GPUs

Server power

10.2 kW each × 128 = 1,306 kW (just GPUs + CPUs)

Network

36 InfiniBand switches at 1.5 kW = 54 kW

Storage

Parallel filesystem (Weka, Lustre): 100 kW

Total IT load

~ 1,460 kW = 1.46 MW

Sized facility load

IT load × 1.25 (continuous):
1,460 × 1.25 = 1,825 kW
Mech load (DLC + facility):
10% of IT (vs 30% for air-cooled) = 146 kW + facility overhead 50 kW = ~ 200 kW
Total facility demand:
1,825 + 200 = ~ 2,025 kW
PUE achieved:
2,025 / 1,460 = ~ 1.39 (could be lower with optimized DLC)

Electrical infrastructure

Service transformer

2,500 kVA pad-mount (or 2 × 1,500 if 2N)

UPS

2 × 1,250 kVA online double-conversion (2N for IT)

Generators

2 × 2,500 kW Tier 4 diesel

Power distribution

480Y/277V busway (4,000 A) down each row

Per-rack feed

100 A plug-in disconnect (10 kW × 1.25 = ~ 30 A safety margin built-in)

Cooling

Direct liquid cooling (CDUs serving multiple racks); chilled water 30°C supply

i

Why this single pod is bigger than half of Atlas DC1

Atlas DC1 = 2.5 MW total. This single AI pod = 1.46 MW IT (2.0 MW total facility). One pod consumes more power than HALF of Atlas DC1's design capacity. Modern hyperscale AI campuses might have 50-100 of these pods coordinating on a single training run.

The Frontier — Coming Architectures

Trend (2026)	Implication
800V DC distribution	Eliminates AC-DC-AC conversion at every PSU. Pioneered by Open Compute Project (OCP). Adopted by hyperscale.
Battery backup IN the rack	Replace centralized UPS with batteries at each rack — eliminates UPS losses, simplifies redundancy
Microgrid + on-site generation	Pair AI campus with on-site PV + ESS + gas turbines. 100+ MW microgrids becoming common.
Submersion / two-phase immersion	Pushing rack densities to 200-400 kW/rack
Heat reuse to district heating	Datacenter waste heat (50-80°C with DLC) feeds neighboring buildings or even municipal heat grids (Helsinki, Stockholm)
Modular AI pods	Factory-built pods shipped to site; deploy in 6 months instead of 24
Co-location with renewables	Build AI campus next to wind/solar farms; long-term PPAs lock in low-cost clean power

If You See THIS, Think THAT

If you see…	Think / use…
"AI/HPC data center"	30-100 kW/rack · DLC mandatory · sub-ms network · single training run uses 1000s of GPUs
"NVIDIA HGX H100" / "DGX"	NVIDIA's reference 8-GPU server. ~ 10 kW. Standard AI building block.
"SuperPOD"	NVIDIA terminology for 32-127 DGX systems coordinated by InfiniBand
"InfiniBand"	Required for tight GPU coordination. Higher cost than Ethernet but required for training.
"NVLink switch"	NVIDIA's GPU-to-GPU interconnect within and between servers
"800V DC"	Open Compute Project standard. Direct DC to server. Hyperscale-only currently.
"Liquid cooling" in 2026 context	Almost certainly DLC (cold plates), increasingly immersion. See §35.
"PUE 1.1" or lower	DLC or immersion. Air-cooled cannot achieve this.
"Hyperscaler"	AWS, Google, Microsoft, Meta, Apple, Alibaba, Tencent. Operate own DCs.
"Cloud GPU on-demand"	End-user accesses these AI clusters via cloud APIs. The DC is hyperscaler's; the GPUs are rented by hour.

§29 / 39

Reading Drawings & Specs

E-sheet structure · CSI Division 26 · schedules · symbols · RFIs

An electrical drawing set has 30+ sheets, organized by a precise convention. Knowing which sheet number to look at when you have a question is half the skill. Specifications (Division 26) tell you HOW to install — drawings tell you WHAT.

The Electrical Drawing Set

An electrical project's drawing set is organized by sheet number. The number tells you what type of information to expect.

Sheet number	Content	What you find here
E001	Cover sheet, drawing index	Project info, sheet list, applicable codes, abbreviations
E002	Symbols + general notes	Legend of all electrical symbols used; project-wide notes
E003-E099	Site / utility	Service entrance, site lighting, utility coordination
E101-E199	Floor plans — power	Receptacles, equipment locations, panel locations
E201-E299	Floor plans — lighting	Light fixtures, controls, emergency egress lighting
E301-E399	Floor plans — systems	Fire alarm, security, telecom, audio/visual
E401-E499	Single-line diagrams	Power distribution SLD, riser diagram
E501-E599	Schedules	Panel schedules, transformer schedule, MCC schedule, fixture schedule
E601-E699	Details	Mounting details, grounding details, service entrance detail
E701-E799	Special systems	Telecom rooms, AV rooms, equipment rooms
E801-E899	Demolition	Existing-to-remove (renovation projects only)

i

When you have a question, look here

"Where does this circuit feed from?" → Panel schedule (E501). "What's the fault current at this bus?" → SLD (E401). "How is this panel mounted?" → Detail (E601). Drawing-set fluency = knowing where each question lives.

The Specification Document

Drawings tell you WHAT to install. Specifications tell you HOW. The CSI MasterFormat 50-Division system is the industry standard for organizing specifications.

CSI Division	Subject	Electrical relevance
Division 1 — General Requirements	Project administration, submittals, etc.	Read first — applies to all trades
Division 26 — Electrical	All electrical work	Your home base
Division 27 — Communications	Voice + data + AV cabling	Often coordinated with electrical
Division 28 — Electronic Safety + Security	Fire alarm, security, access control	Integrated with electrical service
Division 23 — HVAC	Mechanical equipment	You provide power for their equipment per MEL
Division 33 — Utilities	Site utilities	Coordination with utility company

Division 26 Sub-Sections (Most Common)

Section	Content
26 05 00	Common Work Results for Electrical (general requirements)
26 05 19	Low-Voltage Electrical Power Conductors and Cables
26 05 26	Grounding and Bonding
26 05 33	Raceways and Boxes
26 09 23	Lighting Control Devices
26 09 43	Network Lighting Controls
26 12 00	Medium-Voltage Transformers
26 13 00	Medium-Voltage Switchgear
26 18 00	Medium-Voltage Distribution
26 22 00	Low-Voltage Transformers
26 24 13	Switchboards
26 24 16	Panelboards
26 24 19	Motor-Control Centers
26 27 26	Wiring Devices (receptacles, switches)
26 28 13	Fuses
26 28 16	Enclosed Switches and Circuit Breakers
26 29 13	Enclosed Controllers (motor starters)
26 32 13	Engine Generators
26 33 53	Static Uninterruptible Power Supply
26 36 00	Transfer Switches
26 41 13	Lightning Protection for Structures
26 43 13	Surge Protective Devices
26 51 00	Interior Lighting
26 56 00	Exterior Lighting

Three-Part Specification Format

Every Division 26 spec section follows the CSI 3-part format:

Part	Content	What you do with it
Part 1 — General	References, submittal requirements, quality assurance, warranty	Read first — applies to entire section
Part 2 — Products	Approved manufacturers, technical specifications, options	Tells you exactly what equipment to buy + what's substitutable
Part 3 — Execution	Installation, testing, commissioning, training	Field installation rules

Drawing-Spec Discrepancies

When drawings and specs disagree (and they often do), which governs?

!

Generally: specifications govern (per Division 1)

Most contracts state specifications take precedence over drawings, and large-scale drawings take precedence over small-scale. ALWAYS submit an RFI when discrepancy is found — never assume which is correct.

Schedules — Where Equipment Lives

Schedule type	Content	Sheet location
Panel Schedule	Every breaker, wire, load, phase. Per panel.	E501-E599 typically
Transformer Schedule	kVA, voltage, %Z, configuration, location, OCPD	E501
MCC Schedule	Each bucket: starter type, motor served, FLA, CB size	E501
Switchgear Schedule	Each compartment: breaker rating, function, connection	E501
Lighting Fixture Schedule	Each fixture type: model, watts, lumens, mounting, voltage	E501-E599
Cable Schedule	Each cable run: from, to, type, size, length	E501-E599 (industrial only)
Conduit Schedule	Each conduit run: type, size, fittings	Industrial only
Equipment Schedule	Each piece of major equipment: tag, V, HP/kW, FLA, location	The MEL — usually mech-provided, electrical-augmented

Worked Example 1 — Atlas DC1 Drawing Set

Example 01 · Atlas DC1 spineAtlas DC1's 87-sheet drawing set + 250-page Division 26 specification

Drawing set (excerpts)

Sheet	Content
E001-E002	Cover, index, codes, symbols
E003-E010	Site plan, utility coordination, ground ring
E101-E110	Floor plans — IT halls power layout (PDU + RPP locations)
E111-E115	Mech room power (chillers, pumps, MCCs)
E201-E210	Lighting plans (IT halls, mech, office, exterior)
E301-E305	Fire alarm + emergency systems
E401	Main SLD (the canonical Atlas DC1 one-line)
E402-E405	Detailed SLDs for each side, UPS, generator paralleling
E501-E520	Panel schedules (every panel + RPP)
E521-E523	Transformer + MCC + UPS schedules
E601-E620	Mounting details, grounding details, service entrance
E701-E705	UPS room layouts, battery room ventilation, cable tray routes

Specification (excerpts from 250-page Division 26)

Section	Excerpt
26 13 00 (MV switchgear)	3-piece arc-resistant gear, vacuum CBs, withstand 50 kA. Manufacturers: Eaton, Siemens, ABB. Coordination study by mfr.
26 24 13 (Switchboards)	Square D/Eaton/GE acceptable. Bus 4000A Cu. 65 kA AIC. ANSI/NEMA PB2 compliant.
26 32 13 (Generators)	2 × 2500 kW Tier 4 final, sound-attenuated enclosure, sub-base fuel tank 24 hr. 0.85 PF. Manufacturers: Caterpillar, MTU, Cummins.
26 33 53 (UPS)	Static double-conversion UPS, 1250 kVA, VRLA battery, 5 min ride-through. SCCR 65 kA. Eaton, Schneider, Vertiv accepted.
26 36 00 (ATS)	Bypass-isolation construction. Open transition. 4000 A. NEMA 1.

Result: Contractor uses these to bid + procure equipment. Engineer reviews submittals against the spec.

Worked Example 2 — Reading a Spec for First Time

Example 02 · Skill walkthroughWhere to start when handed a 200-page Division 26 spec

Read 26 05 00 first. "Common Work Results" — applies to entire division. Submittal requirements, codes, warranties.
Then read 26 05 19 (cables) + 26 05 33 (raceways) + 26 05 26 (grounding). The basic infrastructure spec.
Match each piece of equipment on drawings to its spec section. Switchboard on E401? Read 26 24 13. Panel on E501? Read 26 24 16. Etc.
Look for "approved manufacturers" lists. Tells you who's bidding. If only one mfr listed = single-source spec.
Look for "Owner-furnished, contractor-installed" (OFCI). Common for IT switchgear in DCs — owner buys, contractor installs.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Sheet number

Where do you find the SLD?

Drill 2 · Division 26

What's Division 26 in CSI MasterFormat?

Drill 3 · 3-part spec

What's in Part 2 of a CSI spec?

Drill 4 · Discrepancy

Drawings + specs disagree. Generally which governs?

Drill 5 · OFCI

OFCI means?

If You See THIS, Think THAT

If you see…	Think / use…
"Division 26"	Electrical specifications. Companion to drawings.
"E001" or "E101" sheet	Cover/index or first floor power plan respectively.
"E401"	Single-line diagram. The system map.
"E501"	Panel schedules + transformer + MCC schedules.
"E601"	Details — mounting, grounding, service entrance.
"OFCI"	Owner-Furnished, Contractor-Installed. Common for sensitive equipment.
"OFOI"	Owner-Furnished, Owner-Installed. Rare in commercial.
"RFI" (Request for Information)	Contractor question — engineer must respond formally.
"Submittal"	Contractor's documentation showing equipment selected. Engineer reviews + stamps.
"As-built" or "As-recorded"	Final drawings showing what was actually installed (vs. designed).
"IFC" (Issued for Construction) stamp	Drawing version released for construction. Subsequent revisions tracked.
"Revision cloud"	Marks a change region on a drawing. Triangle marker shows revision number.
"NTS" or "Not To Scale"	Drawing not to scale — measure dimensions, don't measure off the drawing.
"Coordinate with [other discipline]"	Issue resolved between trades during construction. Common note on drawings.

§30 / 39

LOTO

NFPA 70E Article 120 · OSHA 1910.147 · 7-step sequence · group LOTO

Lockout/Tagout is the procedure for safely isolating equipment for service. NFPA 70E and OSHA 29 CFR 1910.147 govern. Authorized vs affected workers have different responsibilities. Energy isolation devices vary by source type.

Lockout/Tagout — The Standard

Lockout/Tagout is the procedure for safely isolating equipment for service. NFPA 70E and OSHA 29 CFR 1910.147 govern. The goal: make absolutely sure no energy can reach the equipment while a worker is in contact with it.

Standard	Scope
OSHA 29 CFR 1910.147	"Control of Hazardous Energy" — the federal LOTO standard. Applies to general industry.
OSHA 29 CFR 1926.417	Construction-specific lockout requirements.
NFPA 70E Article 120	"Establishing an Electrically Safe Work Condition" — the electrical LOTO procedure.
ANSI Z244.1	National consensus standard for control of hazardous energy.

The 7-Step LOTO Sequence (NFPA 70E 120.5)

Step	Action	Detail
1	Identify all sources	Use SLDs, schedules, walk-down. Multiple sources = backfeed possible.
2	Notify everyone affected	Operators, adjacent workers, customer.
3	Open disconnects + breakers	Both line + load side of equipment.
4	Apply locks + tags	One worker = one lock. Personal padlock + tag with name + date.
5	Discharge stored energy	Capacitors, springs, batteries, hydraulic accumulators, compressed air, thermal.
6	Verify de-energization	Test before touch — voltmeter on a known live source first, then on equipment, then on known live source again.
7	Apply temporary protective grounds (MV/HV)	For voltages > 600V — induced voltage can re-energize the line.

Authorized vs Affected vs Other Workers

Role	Definition	Training requirement
Authorized	Workers who lock out + work on energy-isolated equipment	Full LOTO training. Authorized to apply + remove their own locks.
Affected	Workers whose job uses the equipment being LOTO'd	Awareness training. Cannot apply locks. Must be informed of LOTO.
Other	Workers in the area	General awareness. Recognize a locked-out condition.

Energy Source Types

Energy	How to isolate	How to verify
Electrical	Disconnect / breaker open + locked	Voltmeter test (live-dead-live)
Hydraulic	Block valves closed + locked, pressure relieved	Pressure gauge at zero
Pneumatic	Air valve closed + locked, line vented	Pressure gauge at zero
Mechanical	Block in place, springs released	Visual inspection
Thermal	Allow cool down, isolate hot/cold sources	Temperature measurement
Chemical	Block valves on chemical lines	Sniff testing for vapors, line break point
Stored (capacitors, springs)	Discharge to ground via resistor; relax springs	Voltmeter on capacitor; visual on springs

Equipment for LOTO

Item	Purpose
Padlock	Physically holds disconnect open. Each worker has unique key — only that worker can remove their lock.
Lockout tag	Attached to padlock. Shows worker name, date, reason.
Hasp / multi-lock device	Allows multiple workers to apply locks to the same disconnect (group LOTO).
Breaker lockout	Specific device for circuit breakers — slides between breaker handle and ON position.
Plug lockout	Encases the plug of a portable cord, prevents reconnection.
Valve lockout	Devices to lock pneumatic/hydraulic valves in closed position.
Voltage detector	Used in step 6 verification. Cat III rated for the voltage being tested.
Temporary protective ground (TPG)	For MV/HV — connects line to ground after isolation. Required for > 600V.

Group LOTO + Complex Procedures

Situation	Procedure
Single worker on simple equipment	Standard 7-step. One lock per worker.
Multiple workers, one piece of equipment	Group LOTO. Each worker applies their own lock to a hasp. Equipment cannot be re-energized until ALL workers remove their locks.
Multiple workers, multiple disconnects	Group LOTO with key-lock box. Master lock on disconnects holds keys; each worker locks the box.
Long-duration project (multiple shifts)	Shift transfer of LOTO. Strict sign-off + verification at each shift change.
Worker cannot remove their own lock	NEC + OSHA exception process — supervisor verifies worker is gone, then removes after multiple confirmations.

Worked Example 1 — Atlas DC1 LOTO with 2N Topology

Example 01 · Atlas DC1 spineServicing UPS-A1 — LOTO procedure that doesn't drop IT load

Why this is interesting

UPS-A1 serves IT Row A. In a non-2N facility, LOTO of UPS-A1 = drop IT load. In Atlas DC1's 2N topology, IT Row A is also fed from UPS-B1 via redundant paths in each rack PDU.

Verify 2N path operational: Before starting, verify Side B (UPS-B1) is fully operational and able to carry full Side A IT load. Coordinate with operations team.
Identify all sources of UPS-A1: Input from 480V SWGR-A. Output to PDU-A1. Battery string. Bypass static switch. UPS controls power.
Notify: Site operations, IT operations, customer, fire alarm panel monitor.
Open + lock disconnects: Input CB at SWGR-A (2000 A). Output CB at PDU-A1. Battery string DC CB. Static bypass disconnect.
Discharge stored energy: Wait minimum 5 minutes for DC bus capacitors to discharge below 50V (per UPS manufacturer). Voltmeter verify.
Verify de-energized: Live-dead-live test on UPS terminals.
Battery string isolation: Battery DC voltage typically 540V. Even after AC removed, battery is still energized. Apply lock to DC CB.
Work begins. If multiple workers, group LOTO with each applying their own lock to a multi-lock hasp.

Throughout this procedure

IT load remains fully powered via UPS-B1 redundant rack feeds.
Genset coordination: Verify GEN-A startup not commanded during LOTO (it would energize Side A through ATS-A but UPS-A1 is locked out anyway).
Bypass switch: If maintenance bypass switch present, operators may transition to bypass before LOTO — but for full LOTO, bypass also locked.

Worked Example 2 — Standard Industrial LOTO (Motor)

Example 02 · Alternate contextServicing 100 HP industrial pump motor — standard 7-step LOTO

Identify sources: Power from MCC bucket. Control wiring from PLC. Mechanical: pump impeller. Hydraulic: process flow upstream.
Notify: Operators, adjacent shift workers, control room.
Open disconnects: MCC bucket disconnect (combination starter has pull-out handle). Open + lock.
Apply control lockout: Lock the auto/manual selector switch in OFF position so PLC cannot send start signal.
Isolate process: Close + lock isolation valves upstream + downstream. Bleed line pressure.
Verify: Voltmeter (live-dead-live) on motor terminals. Pressure gauge on process line at zero.
Work begins.
After work: Close access panels. Remove locks in reverse order. Open valves. Restore power. Test run.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · LOTO standard

Federal LOTO standard?

Drill 2 · Authorized worker

Who can apply locks?

Drill 3 · Live-dead-live

Three steps of voltage verification?

Drill 4 · Stored energy

Capacitor bank in UPS — safe to touch immediately after disconnect?

Drill 5 · MV/HV grounding

Required additional step for ≥ 600V LOTO?

If You See THIS, Think THAT

If you see…	Think / use…
"LOTO" / "Lockout/Tagout"	Procedure for safe energy isolation. NFPA 70E + OSHA 1910.147.
"NFPA 70E Article 120"	Electrical-specific LOTO. The electrically safe work condition procedure.
"Authorized worker"	Trained on full LOTO. Can apply + remove own lock.
"Affected worker"	Awareness only. Cannot apply locks but must be informed.
"Live-dead-live test"	Verify voltmeter on live source first, then absent on equipment, then live again. Confirms voltmeter still works.
"Group LOTO"	Multiple workers each apply own lock to a multi-lock hasp.
"Stored energy"	Capacitors (UPS, VFDs), springs, batteries, hydraulic accumulators. Must discharge before work.
"Temporary protective ground" (TPG)	For MV/HV — bonds line to ground after isolation. Mandatory above 600V.
"Voltmeter Cat III rated"	Test instrument rated for the voltage being tested. Cat III for distribution; Cat IV for service entrance.
"Single point of control" / dual disconnect	Two independent isolation methods for higher-risk work. Used for life-critical systems.
"Energized work permit"	NFPA 70E 130.2 — required if working on energized equipment is justified (rare). Documented hazard analysis + PPE.

§31 / 39

Working with the AHJ

Plan review · inspection stages · NEC 90.4 equivalency · best practices

The Authority Having Jurisdiction (typically the local building/electrical inspector) approves your design and inspects construction. Failed plan reviews and inspections delay schedules. A relationship with the AHJ is the most underrated skill in electrical engineering.

Who Is the AHJ?

The Authority Having Jurisdiction is the person or office responsible for enforcing electrical code in your project's jurisdiction. Usually a city/county building department or state-level electrical board. Sometimes the fire marshal, insurance carrier, or owner's representative.

Type of AHJ	What they enforce
City building department electrical inspector	NEC + local amendments (most common AHJ for commercial)
State electrical board (some states)	NEC + state-level amendments (e.g., Massachusetts MA250)
Fire marshal	NFPA 70E (workplace), NFPA 13 (sprinklers), NFPA 855 (ESS), NFPA 780 (lightning)
Insurance underwriter (FM Global, etc.)	Insurance-driven standards (sometimes stricter than NEC)
Owner's representative	Project-specific requirements (often more stringent)
Federal AHJ (for federal projects)	NEC + agency-specific (DOD UFC, GSA standards)
Utility	Service entrance + interconnection only (not entire building)

Plan Review — What the AHJ Looks At

Before a permit is issued, the AHJ reviews the construction documents for code compliance. Drawings + specifications + calculations must address every code-required element.

Review item	What AHJ checks
NEC compliance overview	Latest NEC version (or jurisdiction's adopted version) properly applied
Service sizing (NEC 220)	Load calculation submitted; service entrance properly sized
Available fault current (NEC 110.24)	Documented at service equipment; equipment AIC adequate
Overcurrent protection (NEC 240)	Properly sized; selective coordination if NEC 700.27 applies
Grounding (NEC 250)	Service grounding electrode system; equipment grounding sized; GFP if required
Working space (NEC 110.26)	Clearances around equipment; egress routes
Hazardous locations (NEC 500-516)	Area classification drawing; equipment ratings
Emergency systems (NEC 700)	Selective coordination study; transfer time compliance; fuel supply
Special occupancies	NEC 517 (healthcare), 518 (assembly), 547 (agricultural), 680 (pool/spa)
Local amendments	Jurisdiction-specific add-ons (CA Title 24, NYC, Chicago, etc.)

Common Plan Review Rejections

Rejection reason	How to avoid
Missing fault current (NEC 110.24)	Show available fault current at service + at major buses on SLD
Missing surge protection (NEC 230.67)	Add Type 1/2 SPD at every dwelling service (2020+ NEC)
Working space (NEC 110.26) violations	Show clearances on plans. Don't put equipment in tight closets.
Inadequate grounding	Detail grounding electrode system + EGC sizing
Missing arc flash labels (NEC 110.16(B))	Required for service equipment ≥ 1200 A. Specify in spec.
Demand calc errors	Apply NEC 220 demand factors correctly per occupancy type
EVSE without 125% rule	EV charging is continuous load. Apply 125% to wire + breaker.
Selective coordination not shown	NEC 700.27 — life safety systems require coordination study
PV interconnection violating 120% rule	Use supply-side connection (NEC 705.11) if needed

Inspection Stages

Inspection	When	What's checked
Underground / rough-in	After conduit + boxes installed, before backfill or drywall	Conduit routing, box mounting, support, depth (if underground)
Service entrance	After service installed, before energization	Grounding, conductor sizing, breaker selection, working space, labels
Rough-in (general)	After all conductors pulled, before drywall closes walls	Conductor sizing, splice locations, box fill, support
Service connection / utility coordination	Before utility energizes service	Service per NEC 230, grounding, working space
Final / occupancy	End of construction	Devices installed, labels in place, panel schedules complete, GFCI/AFCI tests pass
Re-inspection	After failed inspection corrected	Specific items previously failed
Specialty (medium voltage, hazardous)	Per project — usually outside the regular cycle	Specific to specialty (MV terminations, area classification)
Performance test (NEC 230.95(C))	Before energizing service with GFP	Field test of GFP system

NEC 90.4 — Equivalencies + Variances

The AHJ has the authority to approve methods that aren't strictly per NEC, when equivalent safety is demonstrated. Per NEC 90.4: "The authority having jurisdiction shall be permitted to grant equivalent provisions."

i

When to ask for a variance

Industrial / process / data center installations sometimes need to deviate from NEC for technical reasons. Document the equivalent safety. Submit formal variance request to AHJ before construction. Common cases: HRG systems (instead of GFP), specialized cable types, working space alternatives.

Working with the AHJ — Best Practices

Practice	Why it matters
Pre-application meeting	Discuss complex aspects before submitting drawings. Surface concerns early.
Cite NEC sections in your design	Show the AHJ you applied the code, by reference. Reduces back-and-forth.
Use the inspector's preferred forms	Some AHJs require specific submittal forms.
Be present at inspections	Address questions on the spot. Avoid second visits.
Don't argue — ask for the code section	If you disagree with an inspector, ask politely for the specific NEC reference. Often the difference is interpretation.
Build relationships	Your reputation with the local AHJ matters. A good track record earns trust.

Worked Example 1 — Atlas DC1 AHJ Coordination

Example 01 · Atlas DC1 spinePermit + inspection process for a 2.5 MW data center

Stages

Pre-application meeting (6 months before submittal): Meet with city electrical inspector + fire marshal. Present concept SLD + proposed approach for: medium voltage service, 2N redundancy, generator location, fuel storage (per IFC), Li-ion ESS rooms (NFPA 855), arc flash analysis methodology.
Drawing submittal: 87-sheet drawing set + 250-page Division 26 spec + load calc + arc flash study + coordination study + grounding study. Plan review fees ~ $25,000.
First plan review comments (4 weeks later):
• Need details on UPS battery room ventilation (NFPA 1)
• Working space at MV switchgear marginal — verify NEC 110.34
• Selective coordination study for UPS-fed loads (verify NEC 700.27 not required since this is NEC 702 optional standby)
• Lightning protection drawings (NFPA 780) — separate submittal package
Resubmit with responses (2 weeks): Address each comment with detail or documentation. Most resolved.
Permit issued. Construction begins.
Inspections: Underground (foundation grounding ring), rough-in (conduit + cable installations), service entrance (MV), final.
Specialty inspection: Fire marshal walks through ESS rooms (NFPA 855), battery rooms (Class I Div 2 verification), Type 1 SPD at MV switchgear.
Pre-energization: NEC 230.95(C) GFP performance test. Witness by AHJ.
Final / occupancy: Verify all panel schedules complete, arc flash labels installed, working space clear, emergency lighting tested.

Worked Example 2 — Residential Permit (Smaller Scale)

Example 02 · Alternate scaleSingle-family home addition — adding 200A subpanel

Permit application: Form + sketch showing existing service + new subpanel location. ~ $200-500 fee.
Plan review: Plans checker verifies basic requirements: load calc shows existing service handles new load, conductor sizing, GFCI/AFCI requirements.
Permit issued. Often same-day for simple residential.
Rough-in inspection: Inspector verifies conduit, boxes, conductors, before drywall.
Final inspection: Devices installed, GFCI/AFCI test, panel cover labeling.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · Plan review

Top reason for plan review rejection?

Drill 2 · NEC 90.4

AHJ authority to approve alternatives?

Drill 3 · Inspection stages

When does service-entrance inspection happen?

Drill 4 · Pre-application meeting

When is it most valuable?

Drill 5 · Atlas DC1 special inspection

Who inspects Atlas DC1's Li-ion ESS room?

If You See THIS, Think THAT

If you see…	Think / use…
"AHJ"	Authority Having Jurisdiction. Usually city/county electrical inspector.
"NEC 90.4 equivalency"	AHJ authority to approve alternative methods. Document equivalent safety.
"Plan review"	AHJ reviews drawings + specs before permit. Common rejection reasons fixable upfront.
"Rough-in inspection"	After conduit + cable installed, before drywall. Verifies physical installation.
"Final inspection"	End of construction. Verifies devices, labels, tests pass.
"NEC 110.16 label"	Arc flash warning required at every panel.
"NEC 110.24 fault current"	Available fault current must be marked on service equipment.
"NEC 230.95(C) GFP test"	Performance test required before energizing 480V service ≥ 1000A.
"FM Global rules"	Insurance carrier standards. Often stricter than NEC. Sometimes the AHJ.
"Local amendment"	City/state-specific rules added to NEC. Always check.
"Pre-application meeting"	Best practice for complex projects. Surfaces concerns before formal submittal.
"NEC 700.27" + life safety	Selective coordination required for life safety. AHJ enforces.

§33 / 32

Revit for Power Atlas

BIM modeling · families · clash detection · schedule auto-generation

Revit is how modern electrical design gets documented. Not just drawing — modeling. The model auto-generates schedules, catches clashes before construction, and serves as the single source of truth for plans, sections, and specifications.

What Revit Is + Why Electrical Engineers Use It

Revit is a Building Information Modeling (BIM) software by Autodesk. Unlike CAD (which draws lines), Revit models 3D parametric objects that know what they are. An electrical panel in Revit isn't a rectangle — it's a "Panelboard" object with electrical connectors, voltage, ratings, and bidirectional links to its panel schedule.

Aspect	CAD (AutoCAD/MEP)	Revit
Drawing primitive	Lines, arcs, blocks	3D parametric families (Wall, Panel, Light Fixture, etc.)
Schedule generation	Manual data entry	Auto-generated from model objects
Coordination with other disciplines	Overlay drawings; manual conflict checking	Federated models; automated clash detection
Single source of truth	Every drawing is independent	Plans, sections, schedules all live-update from one model
Learning curve	Moderate	Steep (8-12 weeks for proficiency)
Industry adoption (2026)	Legacy; declining for new buildings	Standard for commercial + industrial new construction

The Five Things Electrical Revit Modelers Do

#	Activity	Atlas DC1 example
1	Place equipment families — switchgear, transformers, panels, ATSs, UPS, gensets, PDUs	Drop TX-A pad-mount family in mech yard, set kVA + voltage parameters
2	Run cable tray + conduit — define routes through the building	Tray from 480V SWGR-A → UPS-A1 (250 ft route)
3	Wire branch circuits — connect equipment loads to source panels	Connect each rack PDU to RPP-A1-1 with branch wire
4	Generate schedules — panel schedules, transformer schedules, fixture schedules, equipment schedules	RPP-A1-1 panel schedule with all 42 circuits + phase totals
5	Coordinate with other trades — clash detection vs mechanical, structural, plumbing	Verify cable tray doesn't conflict with chilled water piping in mech room

Atlas DC1 Revit Project — How It's Organized

Model file	Owner	Contains
Atlas-DC1-Architectural.rvt	Architect	Walls, doors, ceilings, room boundaries
Atlas-DC1-Structural.rvt	Structural engineer	Beams, columns, foundations, floor decks
Atlas-DC1-Mechanical.rvt	HVAC engineer	Chillers, CRAH, ducts, chilled water pipes
Atlas-DC1-Plumbing.rvt	Plumbing engineer	Domestic water, sanitary, fire suppression piping
Atlas-DC1-Electrical.rvt	You (electrical engineer)	Switchgear, transformers, panels, ATSs, UPS, gensets, PDUs, conduit/tray, branch wiring, lighting fixtures, fire alarm, telecom
Atlas-DC1-Federated.rvt	BIM coordinator	Links all discipline models together for coordination + clash detection

Electrical Family Library — What You Need

Family category	Examples	Source
Distribution Equipment	Switchgear, switchboards, MCCs, panelboards, ATSs, UPS, generators	Manufacturer (Eaton, Schneider, ABB) or in-house library
Transformers	Pad-mount, dry-type, secondary unit substation	Manufacturer libraries
Wiring Devices	Receptacles, switches, GFCI/AFCI outlets, occupancy sensors	Default Revit + manufacturer (Hubbell, Leviton)
Lighting Fixtures	2×4 LED troffers, downlights, exit signs, emergency packs	Manufacturer (Lithonia, Philips, Acuity)
Cable Tray / Conduit	Ladder, ventilated, wire mesh tray; EMT, RMC, PVC conduit	Default Revit + manufacturer (B-Line, T&B, Cooper)
Fire Alarm	Smoke detectors, pull stations, NACs (notification), FACPs	Manufacturer (Siemens, Notifier, Edwards)
Communications	Data outlets, patch panels, racks	Default + custom

Worked Example 1 — Modeling Atlas DC1's RPP-A1-1 Panel

Example 01 · Atlas DC1 spineFrom schedule design (§05) to Revit panel object

Workflow

Place panelboard family. In Revit, Insert → Load Family → Electrical → Power → Panelboard. Drop into IT Hall A row position.
Set type properties. Distribution System: 415Y/240V 3φ-4W. Bus rating: 400 A. AIC: 14 kA. Mounting: Surface.
Set instance properties. Panel name: RPP-A1-1. Mark: RPP-A1-1. Source: PDU-A1 panel (link to upstream).
Connect upstream. Wire from PDU-A1 sub-panel breaker to RPP-A1-1 main lug. Revit assigns power load automatically based on connected branch circuits.
Place branch loads. For each rack, drop a "Rack PDU" family (custom — likely an electrical equipment object). Set load: 5,760 W per circuit.
Wire branches. Use Revit's Power Wire tool. Connect rack to RPP-A1-1. Revit auto-assigns to first available circuit, balancing across phases.
Generate panel schedule view. Right-click panel in Project Browser → Edit Panel Schedule. The schedule auto-populates with all connected branches, currents, phase balance.
Verify against design. Total panel load matches §05 calc (99.3 A per phase). Phase balance within ±5%. Bus loaded to 25%.

Worked Example 2 — Cable Tray Routing

Example 02 · Atlas DC1 spineCable tray from 480V SWGR-A to UPS-A1 — Revit routing

Workflow

Decide tray spec from §08. 18-inch wire mesh tray, ladder type, 4-inch tall.
Pick endpoints. Cable tray riser from 480V SWGR-A (in electrical room) to UPS-A1 (in adjacent UPS room).
Place tray. Systems → Electrical → Cable Tray. Select type. Click start point + end point. Revit auto-routes orthogonally with bends.
Set elevation. Properties → Reference Level + Offset. Set tray bottom at 12'-0" AFF (above floor) — clear of mechanical piping.
Coordinate with mechanical. Open the federated model. Clash detection (Navisworks or Revit native) → check tray vs chilled water pipes, ducts, structural beams.
Resolve clashes. Adjust tray elevation, route around obstacles, or coordinate with mech to lower a duct.
Add cables to tray. Power cables from 480V SWGR-A breaker → UPS-A1 input lug, 5 sets of 750 kcmil Cu THWN-2 (per §06). Right-click tray → Add Cables. Revit verifies fill ratio against NEC 392.
Generate cable schedule. View → Schedules → Cable Schedule. Lists every cable: from, to, type, length, conduit/tray.

Coordination + Clash Detection

The biggest single value Revit delivers vs CAD: catching conflicts BEFORE construction.

Common clash	Cost if caught in field	Cost if caught in Revit
Cable tray through HVAC duct	$15-50K (rework + delay)	$0 (move in model)
Conduit through structural beam	$5-25K + structural rework	$0
Lighting fixture in HVAC plenum	$2-10K	$0
Panel within 110.26 working space of door swing	$5-20K + AHJ fail	$0
Branch circuit where wall has no stud	$1-5K + drywall patch	$0

Tools used: Revit native clash detection, Navisworks Manage (Autodesk), Revizto, BIM 360 Coordinate. Most projects use one of these in addition to the design Revit model itself.

Common Workflow Tools

Tool	Use	Vendor
Revit	The design model itself	Autodesk
Navisworks Manage	Federated model viewer + clash detection. The "go-to" for combining all-discipline models.	Autodesk
BIM 360 / Autodesk Construction Cloud	Cloud collaboration; review + markup	Autodesk
Revizto	Issue tracking + coordination meetings; alternative to BIM 360	Revizto
Bluebeam Revu	PDF markup + redlining of construction documents	Bluebeam
Dynamo	Visual programming inside Revit — automate repetitive tasks (place 1000 receptacles, batch-update parameters)	Autodesk (built-in)
Civil 3D	Site engineering — utility routing outside building	Autodesk
SKM PowerTools / ETAP / EasyPower	Power system analysis (load flow, fault, coordination, arc flash) — does NOT integrate directly with Revit; manual data transfer	SKM, ETAP, ESA

If You See THIS, Think THAT

If you see…	Think / use…
"BIM execution plan (BEP)"	Project's coordination plan for who-models-what + sharing rules. Read this before starting.
"Federated model"	All discipline models linked together. Used for coordination + clash detection.
"Worksharing"	Multiple users on same Revit model simultaneously via central file
"LOD" (Level of Development)	How detailed is this model? LOD 100 = generic placeholder; LOD 500 = as-built. Spec'd in BEP.
"Family"	Revit's term for parametric object library item
"Type" vs "Instance" parameters	Type = applies to all of that family (e.g., "10A breaker" applies to all 10A instances). Instance = unique to placed object (location, mark).
"Schedule view"	Revit's auto-generated tabular view of model objects (panel schedule, equipment schedule, etc.)
"Distribution system"	Revit's voltage/configuration object (e.g., "480Y/277V 3φ-4W"). Connect panels to a distribution system.
"Power Wire" tool	Revit's tool for connecting load to source panel — assigns to next available circuit.
"Detail Level: Coarse / Medium / Fine"	How much detail Revit shows in views. Affects performance.
"Phasing"	Revit's way to model existing-vs-new construction (renovation projects)
"Rendered" vs "shaded"	Display options. For working, "shaded" is faster.

§34 / 32

Closeout & Commissioning

Cx Levels 1-5 · NETA ATS/MTS · IST · as-builts · O&M

Commissioning is the structured testing that verifies installed systems work as designed. Closeout transfers the facility from contractor to owner with documentation, training, and warranty in place. The bridge from \"design\" to \"operating.\"

Project Phases — Where Closeout + Commissioning Live

Phase	Duration (Atlas DC1)	Electrical engineer's role
Schematic Design (SD)	2-3 months	Loads, SLD, room layout, equipment selection
Design Development (DD)	3-4 months	Detailed sizing, coordination, drawings 50% complete
Construction Documents (CD)	3-4 months	IFC drawings + specs (Division 26)
Bid + Award	2-3 months	RFI responses; verify contractor qualifications
Construction Administration (CA)	12-18 months	RFIs, submittals, change orders, site visits, punch lists
Commissioning (Cx)	3-6 months (overlaps with end of construction)	Witness factory acceptance tests + on-site testing + final acceptance
Closeout	1-3 months	As-builts, O&M manuals, warranties, training, final invoice
Operations / Warranty	1-year warranty period	Warranty walk-through at 11 months

Commissioning — The Full Picture

Commissioning (Cx) is the structured process of verifying that installed systems work as designed. For Atlas DC1, electrical commissioning runs from FAT (Factory Acceptance Test) of major equipment through 5-Level testing to final integrated systems test (IST).

Cx Level	Name	What's tested	When
Level 1	FAT (Factory Acceptance Test)	Each major equipment piece tested at manufacturer's factory before shipment	Before delivery
Level 2	Site Receipt + Storage	Equipment received undamaged; stored properly	At delivery
Level 3	Component-Level Static Test	Insulation tests, mechanical operation, calibration of individual devices (no power)	Pre-energization
Level 4	Subsystem Functional Test	Each subsystem energized + tested independently (UPS, gen, ATS individually)	System energization
Level 5	Integrated Systems Test (IST)	Full facility tested under simulated failure scenarios — utility loss, gen failure, UPS failure, fault scenarios	Pre-occupancy / pre-handover

NETA Acceptance Testing

InterNational Electrical Testing Association (NETA) publishes the standard test procedures for electrical equipment. Used at Cx Level 3 (component static testing) by independent electrical testing firms.

NETA standard	Scope
NETA ATS (Acceptance Testing Specifications)	Tests on NEW equipment before energization
NETA MTS (Maintenance Testing Specifications)	Tests on EXISTING equipment for periodic maintenance
NETA STD (Standard for Electrical Power Equipment Maintenance)	Maintenance frequency + procedures
NETA ETT (Electrical Testing Technician)	Certification standards for the testing personnel

What's Tested at Each Cx Level — Atlas DC1 Examples

Level 3 — Component Tests (NETA ATS)

Equipment	NETA tests required
Cables (medium voltage)	Insulation resistance + DC withstand (hipot at 80% factory test) + partial discharge if cable > 1000 ft
Cables (low voltage)	Insulation resistance only
Transformers (TX-A, TX-B)	Insulation resistance + winding ratio (TTR) + DC winding resistance + power factor (Doble) test + oil testing (dielectric breakdown, DGA)
Switchgear (12.47 kV MV SWGR)	Operation test (open/close), insulation resistance, contact resistance, primary current injection of CTs, secondary trip testing of relays
Circuit breakers	Insulation resistance, contact resistance, time-current calibration (primary current injection)
UPS units	Vendor commissioning (battery acceptance, transfer testing, harmonic verification)
Generators	Insulation resistance, governor calibration, voltage regulator setup, 4-hr load bank test, parallel testing if applicable
ATSs	Manual + automatic transfer cycles, time delay verification, load testing
Grounding system	Ground resistance test (3-point fall-of-potential), continuity verification of grounding electrode system
Protective relays	Setting verification + secondary current injection at each pickup level
Surge arrestors	Insulation resistance, leakage current

Level 5 — Integrated System Test (Atlas DC1)

Full facility tested under simulated failure scenarios. Owner/operator + Cx authority + design engineer all witness.

Test scenario	What's verified	Pass criteria
Utility loss → genset start (Side A)	ATS-A senses loss, signals GEN-A start, gen reaches voltage + frequency, ATS transfers	Total time < 12 sec; UPS rides through; IT load uninterrupted
Utility loss → both sides simultaneously	Both ATSs transfer to gens within target time	Both gens sync with their respective UPS within 15 sec
Genset failure on Side A while running	UPS-A1 detects gen output loss, transitions to battery	5-min battery ride-through verified; load shed sequence (CRAH first) initiates
UPS-A1 fault	Static bypass switch activates; load picked up by utility direct	Transfer < 4 ms (no IT load impact)
480V SWGR-A bus fault	87B bus differential trips main	Trip time < 4 cycles (67 ms) measured at oscillograph
Loss of cooling (chiller failure)	BMS detects, redundant chiller starts	Server inlet temp stays within ASHRAE TC 9.9 envelope
Load bank test at 100% IT design	Full IT load (2.5 MW) drawn for 4 hours	All systems stable; PUE measured + recorded
EPO (Emergency Power Off) test	EPO button drops ALL ITE + HVAC per NEC 645.10	All loads de-energized within 1 sec

Closeout Deliverables

Deliverable	Description	Who provides
As-built / record drawings	Final drawings showing what was actually installed (vs designed)	Contractor → Engineer reviews + stamps
O&M manuals	Operating + maintenance manuals for every piece of equipment	Contractor compiles from manufacturer data
Warranties	Manufacturer warranties (typically 1-2 years on equipment) + contractor workmanship warranty (typically 1 year)	Contractor
Training	Owner staff trained on operating + maintaining systems	Contractor + manufacturer reps
Spare parts inventory	Critical spares stocked on-site (relays, fuses, gaskets, capacitors)	Contractor purchases per spec
Final inspection sign-off	AHJ signs off on final electrical inspection (ready for occupancy)	Contractor coordinates with AHJ
Cx report	Comprehensive report of all Level 3-5 tests, results, deficiencies + resolutions	Cx authority
Final invoice + lien releases	Contractor's final billing + waiver of all subcontractor liens	Contractor
Punch list completion certification	All construction defects corrected and signed off	Contractor + Owner walk-through

Worked Example — Atlas DC1 Energization Day

Example · Atlas DC1 spineFirst-time energization sequence — months of prep collapse into one day

Sequence (one week before "first power")

T-7 days: NETA test firm completes all Level 3 component tests. Submits report.
T-5 days: Engineer reviews NETA report; punch list issued to contractor for any deficiencies.
T-3 days: Punch list cleared. Contractor + Cx authority + utility coordinator confirm energization day schedule.
T-1 day: Pre-energization walkthrough. Verify barriers, signage, PPE on hand. EPO buttons functional. AHJ pre-walkthrough.

Energization day timeline

Time	Action	Witness
06:00	Crew on site; safety briefing; LOTO permits issued	—
07:00	Utility coordinator energizes 12.47 kV primary feeder; verifies voltage at MV switchgear	Utility + Engineer + AHJ
07:30	Energize TX-A primary; verify secondary voltage at 480V SWGR-A (no load)	Engineer + Cx + AHJ
08:00	Repeat for TX-B and SWGR-B	Same
08:30	NEC 230.95(C) GFP performance test on each main breaker	AHJ witnesses
09:00	Energize ATS-A and ATS-B in normal (utility) position	Engineer + Cx
10:00	Energize UPS-A1 in bypass mode; verify input voltage. Then on rectifier; battery comes up to float.	Vendor + Engineer
11:00	Repeat UPS-A2, UPS-B1, UPS-B2 sequentially	Same
12:00	Lunch + first-energization debrief	—
13:00	Energize PDUs + RPPs to no-load voltage. Verify each panel.	Engineer + Cx
14:00	Energize chiller plant + start CH-1; verify cooling tower operational; chilled water reaches setpoint	Mech engineer + Cx
15:00	Atlas DC1 energized + idle. No IT load yet.	—
15:30	Begin Level 5 IST: simulated utility loss → ATS transfers → gen runs → IT load (load bank) ride-through verified	Owner + Cx + Engineer + Vendor
17:00	Run all IST scenarios from Cx test plan	Same
20:00	IST complete. Full day's results documented.	—
21:00	Daily debrief; punch any issues; plan tomorrow's IT load bank ramp testing	—

"First customer rack" gets installed weeks later after Cx is fully signed off. There's no rush — this day is about verifying the facility CAN power critical IT load. Actual IT load comes when the customer is ready.

If You See THIS, Think THAT

If you see…	Think / use…
"Cx" or "commissioning"	Structured testing process — Levels 1-5
"FAT" (Factory Acceptance Test)	Cx Level 1 — at manufacturer's factory before shipment
"NETA ATS"	Independent electrical testing at Cx Level 3
"IST" (Integrated Systems Test)	Cx Level 5 — full facility tested under failure scenarios
"Punch list"	List of construction deficiencies to be corrected before final acceptance
"As-built drawings" / "Record drawings"	Final drawings showing what was actually installed (vs design IFC)
"Substantial completion"	Date when facility is fit for intended use; warranty period starts
"Final completion"	All punch list resolved; final invoice releasable
"O&M manual"	Operating + maintenance manual for installed equipment
"Lien release"	Subcontractor waives right to file lien against owner property; required for final payment
"Warranty walkthrough"	Inspection at 11 months — final chance to get warranty work done
"EPO test"	Emergency Power Off — drops all ITE + HVAC per NEC 645.10. Must be tested at Cx Level 5.
"Hot cutover"	Energizing while old system is still live — used in retrofit projects, riskier than greenfield

§38 / 32

Specifications & Construction Admin

Division 26 · 3-part spec · RFIs · submittals · change orders · CA

Once design is done, the engineer transitions to advisor. Division 26 specifications govern HOW work is installed. Construction Administration runs from RFI #1 to substantial completion. Get the process right and the project ships clean.

Specifications — Division 26

Drawings show WHAT to install. Specifications show HOW. CSI MasterFormat Division 26 is the standard organization for electrical specs.

The Three-Part Format

Every Division 26 spec section follows this structure:

Part	Title	What goes here
Part 1	General	References, definitions, submittals, quality assurance, warranty, delivery + storage
Part 2	Products	Approved manufacturers, materials, equipment specifications, technical performance requirements
Part 3	Execution	Installation, testing, commissioning, training, demonstration, cleaning, protection

Atlas DC1 Division 26 Sections (typical)

A 250-page Division 26 spec for Atlas DC1 might include:

Section	Subject	Approx pages
26 05 00	Common Work Results for Electrical (general; applies to all sections)	15-20
26 05 19	Low-Voltage Conductors + Cables	10-15
26 05 26	Grounding + Bonding	8-12
26 05 33	Raceways + Boxes	10-15
26 05 39	Lighting Control Devices	5-8
26 05 53	Identification (labels, signs)	3-5
26 09 23	Lighting Control Devices	8-12
26 12 13	Medium-Voltage Switchgear	20-30
26 22 13	Medium-Voltage Transformers (pad-mount)	10-15
26 24 13	Switchboards (LV)	10-15
26 24 16	Panelboards	5-8
26 24 19	Motor Control Centers	10-15
26 27 26	Wiring Devices (receptacles, switches)	5-8
26 32 13	Engine Generators	15-20
26 33 53	Static UPS	15-20
26 36 00	Transfer Switches	8-12
26 41 13	Lightning Protection (NFPA 780)	5-8
26 43 13	Surge Protective Devices	5-8
26 51 00	Interior Lighting	15-20
26 56 00	Exterior Lighting	10-15
27 (Comm)	Communications + telecom (Div 27, distinct from electrical)	30-50
28 (Safety/Security)	Fire alarm, security, access control (Div 28)	30-50

Construction Administration (CA) — The Engineer's Role During Construction

Once construction starts, the design engineer transitions from designer to advisor. CA activities:

Activity	Description	Typical frequency
RFI responses	Contractor asks formal questions; engineer responds within 2-7 days. Documented.	20-100/month
Submittal review	Contractor submits product cut sheets + shop drawings; engineer reviews + stamps (Approved / Approved as Noted / Rejected / Revise + Resubmit)	50-500 over project
Site visits	Periodic walkthroughs to verify installation quality + answer field questions	1-4/month, more during energization
Change orders	Owner-requested or contractor-discovered changes requiring scope adjustment	10-50/project
Pay application review	Engineer reviews contractor monthly progress payment requests; approves % completion	Monthly
Punch list management	Engineer walks site near substantial completion; documents deficiencies	Substantial completion + warranty walk
Witness testing	Engineer attends Cx Levels 4-5 to verify performance	Project end
As-built review	Verify contractor's record drawings reflect actual installation	Project end

RFI Management — A Skill in Itself

A poorly-managed RFI process tanks projects. Best practices:

Practice	Why
Respond within 5 business days	Slower = construction delay. Owner pays for delays.
Clear yes/no on the question, plus reasoning	Avoids back-and-forth
Reference NEC article + spec section + drawing	Defensible; helps contractor understand
Distinguish design intent vs additional cost	If response triggers cost, document as change order trigger
CC all relevant parties (architect, structural, owner)	Prevents downstream coordination issues
Use a tracking system (Bluebeam Studio, Procore, Submittal Exchange)	50+ RFIs/month requires tracking; can't manage in email

Submittal Review — What Engineers Look For

Item	Engineer checks
Switchgear submittal	Bus rating, AIC, breaker types + ratings, dimensions vs allotted space, single-line accuracy, GFP per spec, ANSI/IEEE compliance
UPS submittal	kVA rating, battery + runtime, SCCR, harmonics performance, communications interface, warranty
Generator submittal	kW rating, fuel type + tank, sound attenuation, emissions tier (Tier 4 final), enclosure rating, controls + paralleling capability
Lighting fixture submittal	Wattage, lumens, color temp, CRI, dimming compatibility, warranty, certifications (DLC, UL)
Conductors	Type (THWN-2, XHHW-2, etc.), insulation rating, manufacturer listing, voltage rating
Cable tray	NEMA load class, finish, size, fittings, support spans
Coordination study	Provided by contractor's switchgear vendor — engineer verifies plot meets selectivity requirements
Arc flash study	IEEE 1584-2018 method, working distances, electrode configurations, label data

Change Orders

Construction inevitably encounters surprises. Three types:

Type	Initiated by	Engineer's role
Owner change	Owner adds scope (e.g., "extend service to future addition")	Design + spec the change; review contractor's cost
Field-discovered change	Contractor discovers something unforeseen (e.g., conduit needs to route around buried foundation)	Confirm change is necessary; verify cost
Design clarification (no cost)	Engineer issues additional drawings/details to clarify intent without scope change	Issue ASI (Architect's Supplemental Instruction) or engineer's directive — typically no contractor compensation

Substantial Completion + Final Acceptance

Milestone	What it means	What follows
Substantial Completion	Facility is fit for its intended use (per architect/engineer certification). Owner can occupy.	Warranty period starts. Punch list issued. Final retainage held until punch complete.
Final Completion	All punch list items resolved. All closeout deliverables submitted.	Retainage released. Final pay app approved.
Warranty Walkthrough	At ~ 11 months, engineer + owner walk facility	Contractor corrects any warranty issues before warranty expires (12 months)

If You See THIS, Think THAT

If you see…	Think / use…
"Division 26"	Electrical specifications. CSI MasterFormat.
"Three-part spec format"	Part 1 General · Part 2 Products · Part 3 Execution
"RFI" (Request for Information)	Contractor's formal question. 5-day response target.
"Submittal"	Contractor's product proposal. Engineer stamps approval.
"Approved as Noted"	Submittal approved with engineer's notes; contractor must comply with notes
"Revise + Resubmit"	Submittal not acceptable; contractor must address comments + resubmit
"ASI" (Architect's Supplemental Instruction)	Clarification with no scope change — typically no cost impact
"PR" (Proposal Request)	Owner asks contractor to price a potential change
"Change order"	Approved scope change with contractor cost impact
"Substantial Completion"	Project fit for use. Warranty starts. Punch issued.
"Punch list"	List of construction defects to be corrected
"Retainage"	Portion of contractor payment withheld pending final completion (typically 5-10%)
"Specs vs Drawings discrepancy"	Specs typically govern (per Division 1). Always RFI when found.
"Procore" / "Submittal Exchange" / "Bluebeam Studio"	RFI + submittal tracking platforms

§32 / 39

Energy Codes

ASHRAE 90.1 · LPD by space · controls · ASHRAE 90.4 for DCs

ASHRAE 90.1 and IECC govern the energy efficiency of buildings. Lighting power density, HVAC efficiency, controls, metering — all are mandatory minimums. Many jurisdictions add stretch codes (Title 24 in CA, NYStretch in NY).

Energy Codes — The Two That Matter

Standard	Scope	Adoption
ASHRAE 90.1	Energy Standard for Buildings Except Low-Rise Residential. The commercial/industrial energy code.	Adopted by most US states (sometimes via IECC reference). Updated every 3 years (2019, 2022, 2025).
IECC (International Energy Conservation Code)	Includes residential + commercial chapters. References ASHRAE 90.1 for commercial as alternate.	Adopted by many states.
California Title 24	California-specific energy code. Stricter than ASHRAE 90.1.	California only. Often the most aggressive code.
ASHRAE 90.4	Specific energy standard for data centers (since 2016)	Adopted with caveats by some jurisdictions for data center energy compliance.

ASHRAE 90.1 Coverage Areas

Topic	Key requirements
Lighting Power Density (LPD)	Maximum W/sq ft by space type. Office: ~ 0.7 W/sf. Warehouse: ~ 0.3 W/sf. Patient room: ~ 0.5 W/sf.
Lighting controls	Mandatory: occupancy sensors, daylight harvesting (perimeter zones), automatic shutoff, multilevel switching.
Receptacle controls	50% of receptacles in offices must auto-shutoff (NEC 406.4(D) + 90.1 coordinate)
HVAC efficiency	Minimum efficiency for chillers, boilers, fans, pumps, heat pumps
Building envelope	Minimum insulation, fenestration U-factor + SHGC
Service water heating	Minimum efficiency for water heaters, pipe insulation
Energy metering + monitoring	Submeters required for buildings > 50,000 sf or 25,000 sf in some adoptions
Renewable energy provisions	Some adoptions require PV-ready or PV install
Transformer efficiency	NEMA TP-1 / DOE 2016 minimum efficiency for distribution transformers
Motor efficiency	NEMA Premium efficiency required for new motors

Lighting Power Density (LPD) by Space Type

Space type	ASHRAE 90.1-2022 LPD (W/sf)
Office (open)	0.59
Office (private)	0.81
Conference room	0.87
Classroom	0.71
Lobby	0.87
Restroom	0.61
Storage	0.42
Warehouse (medium-bulky)	0.31
Mechanical room	0.43
Patient room (hospital)	0.55
Operating room (hospital)	2.20
Server room (data center)	0.39
Retail	0.84-1.62 (sales area type-specific)
Parking garage (open)	0.13 (interior); 0.04 (exterior)

Mandatory Lighting Controls

Control type	Where required (ASHRAE 90.1)
Occupancy sensors	Most spaces. Auto-off when unoccupied (15-30 min delay typical).
Manual on / partial on	Many private spaces — must turn on manually or to ≤ 50% automatically.
Daylight responsive controls (DRC)	Within 15 ft of windows — daylight zones must reduce lighting based on natural light.
Automatic shutoff	All buildings — automatic time-based shutoff (after hours).
Multilevel control	Most spaces — minimum 3 levels (off, ~ 50%, full).
Egress lighting controls	Always-on emergency lighting per life-safety code (independent of energy code).
Exterior lighting controls	Photocell + curfew controls. Multi-level for parking lot.

Energy Metering Requirements

What's metered	When required
Whole building	Always — utility meter
Lighting subsystem	Buildings > 25,000 sf (some adoptions)
HVAC subsystem	Buildings > 25,000 sf
Receptacle subsystem	Buildings > 25,000 sf
Process loads (kitchen, lab, manufacturing)	Buildings with significant process loads
Tenant submetering	Often required for multi-tenant — allows tenant accountability
Renewable energy	Always — track production separately

ASHRAE 90.4 — Data Center Energy

Data centers consume so much energy that they got their own ASHRAE standard. ASHRAE 90.4 addresses the energy efficiency of the data center components themselves, not just the building shell.

ASHRAE 90.4 metric	Description
MLC (Mechanical Load Component)	Cooling system efficiency relative to IT load. Lower is better.
ELC (Electrical Loss Component)	Electrical distribution losses (UPS, transformers, conductors) relative to IT load.
PUE (Power Usage Effectiveness)	Total facility power / IT power. Industry metric (not officially in 90.4).
Climate-zone-specific limits	MLC + ELC limits vary by climate zone (warmer climates = higher MLC allowed).

Worked Example 1 — Atlas DC1 Energy Compliance

Example 01 · Atlas DC1 spine2.5 MW data center — ASHRAE 90.1 + 90.4 compliance

Building energy compliance

Lighting (ASHRAE 90.1): Server rooms 0.39 W/sf, mech 0.43 W/sf, office 0.59 W/sf. All compliant with LED lighting.
Lighting controls: Occupancy sensors in all unoccupied DC areas (server rooms, mech rooms, electrical rooms). Daylight controls in office perimeter. Manual override only in occupied spaces.
Receptacle controls: 50% of office receptacles auto-off. (Not applied in IT halls — every server is critical.)
Transformer efficiency: All transformers spec'd to NEMA TP-1 / DOE 2016 efficiency.
Motor efficiency: All motors NEMA Premium.
Submetering: Each PDU + each chiller + each major mech equipment.

Data center energy (ASHRAE 90.4)

Metric	Atlas DC1	Target
PUE	1.4 (typical for 2N-redundant)	< 1.5 for modern, < 1.3 for hyperscale
MLC	0.20 (chiller plant efficient)	Per ASHRAE 90.4 climate zone
ELC	0.16 (UPS double-conversion + 2N)	Per ASHRAE 90.4

Trade-off: 2N redundancy increases PUE (more conversion losses) but reduces downtime risk. ASHRAE 90.4 acknowledges this trade-off.

Worked Example 2 — Office Building LPD Compliance

Example 02 · Alternate context50,000 sq ft office — verify lighting power compliant with ASHRAE 90.1

Building total LPD allowance (whole-building method):
Office buildings: 0.61 W/sf × 50,000 sf = 30.5 kW total lighting power
Designed lighting: 30,000 sf open office × 0.59 = 17.7 kW. 5,000 sf private office × 0.81 = 4.05 kW. 5,000 sf circulation × 0.66 = 3.3 kW. Etc.
Total designed: ~ 28 kW. Within 30.5 kW allowance. ✓
If over limit: Reduce LPD. Switch from fluorescent to LED (40-60% reduction). Or use space-by-space method which can be more flexible.

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · ASHRAE 90.1

Commercial energy code standard?

Drill 2 · Office LPD

Open office LPD per ASHRAE 90.1-2022?

Drill 3 · Receptacle controls

% of office receptacles must auto-off?

Drill 4 · PUE

Power Usage Effectiveness for modern DC?

Drill 5 · Daylight harvesting

Required within how many ft of windows?

Illumination Calculations — Lumen Method + Point Method

Lighting design has two layers: meeting the energy code limit (LPD W/sf, ASHRAE 90.1) AND meeting the illuminance requirement (footcandles, IES). Both must be satisfied simultaneously.

Footcandle Requirements by Space (IES Lighting Handbook)

Space type	Recommended (fc)	Notes
Office (general)	30-50 fc on work plane	Higher for reading-intensive tasks
Office (computer-only)	20-30 fc	Reduce glare on screens
Conference room	30-50 fc	Dimmable for video presentations
Corridor / lobby	10-20 fc	Lower than work areas
Restroom	10-20 fc	—
Storage / warehouse	10-30 fc	Higher in active picking aisles
Operating room (hospital)	1,000-2,000 fc on patient	Special task lighting
Server room (data center)	20-30 fc	Per ASHRAE TC 9.9 — minimal lighting; off when unoccupied
Retail (sales floor)	50-100 fc	Higher for merchandise display
Parking garage (interior)	5-10 fc minimum	Per IES RP-20
Parking lot (open)	2-5 fc minimum	Per IES RP-20
Stair / egress	10 fc minimum (NFPA 101)	Continuously lit

Lumen Method — General Lighting Design

Used to determine the number of fixtures needed for general illumination of a uniformly-lit area.

Number of fixtures required

N = (E × A) / (L × CU × LLF)

N = number of fixtures · E = required illuminance (fc) · A = area (sq ft) · L = lumens per fixture · CU = coefficient of utilization (geometry-dependent, ~ 0.5-0.8) · LLF = light loss factor (~ 0.7-0.85, accounts for dirt + lamp depreciation)

Worked Example — Office Lighting Design

Example · 5,000 sq ft open officeLumen method — fixture count + LPD compliance

Inputs

Area

5,000 sq ft

Required illuminance

40 fc on work plane (general office)

Fixture

2×4 LED troffer, 4,500 lm output, 30 W

CU

0.65 (10×10 grid spacing, 80% reflectance walls)

LLF

0.80 (clean office, modern LED)

Calc

Fixture count:
N = (40 × 5,000) / (4,500 × 0.65 × 0.80) = 200,000 / 2,340 = 85 fixtures
Connected lighting load:
85 × 30 W = 2,550 W = 0.51 W/sf
ASHRAE 90.1-2022 LPD limit (open office):
0.59 W/sf maximum.
0.51 ≤ 0.59 → COMPLIANT ✓

i

Why LED won the energy code race

10 years ago this same office would have needed ~ 45-watt T8 fluorescent fixtures, totaling ~ 3,825 W = 0.77 W/sf — exceeding the 0.59 W/sf limit. LED converted the lighting industry overnight by enabling code compliance without sacrificing footcandles.

Point Method — for Specific Locations

Used when you need to verify illuminance at a specific point (operating table, security camera location, parking garage corner). Computed using inverse-square law for direct light + lumen method for reflected.

Illuminance from a single source (point method, direct only)

E = (I × cos θ) / d²

E = illuminance (fc) at point · I = candela toward the point · θ = angle from normal · d = distance (ft). Sum across all fixtures contributing to the point.

If You See THIS, Think THAT

If you see…	Think / use…
"ASHRAE 90.1"	Commercial/industrial energy code. Universally relevant.
"IECC"	Building energy code. Often references ASHRAE 90.1.
"Title 24" (CA)	California-specific. Strictest. Distinct from NEC.
"ASHRAE 90.4"	Data center-specific energy standard.
"LPD" (Lighting Power Density)	Watts per sq ft limit by space type. Mandatory cap.
"PUE" (Power Usage Effectiveness)	Data center metric. Total / IT power. 1.0 = perfect, 2.0 = inefficient.
"Daylight harvesting"	Reduce artificial lighting based on available daylight. Required within 15 ft of windows.
"Occupancy sensor"	Auto-off when unoccupied. Required in most spaces by ASHRAE 90.1.
"NEMA Premium efficiency"	Highest efficiency tier for motors. Required by DOE for new motors.
"NEMA TP-1" / "DOE 2016"	Distribution transformer efficiency standards. Required.
"Submetering"	Per-tenant or per-subsystem electrical metering. Required for buildings > 25,000 sf often.
"Performance path" vs "Prescriptive path"	Two ways to comply with ASHRAE 90.1: meet specific limits (prescriptive) or demonstrate equivalent overall energy use (performance, more flexible).

§32 / 39

Codes & Standards Reference

NEC navigation · IEEE Color Books · NFPA standards · UL · ANSI · OSHA

The library of standards that govern electrical engineering. NEC is the foundation; the IEEE Color Books explain how to apply it; NFPA 70E covers worker safety; ANSI standards cover everything else.

Navigating the NEC

The NEC (NFPA 70) is organized into 9 chapters covering different aspects of installation. Knowing the chapter structure lets you find anything in seconds.

Chapter	Topic	Articles
1	General	90 (introduction), 100 (definitions), 110 (general installation)
2	Wiring + Protection	200-285 (grounding, overcurrent, services, feeders, branches, etc.)
3	Wiring Methods + Materials	300-399 (raceway, cable, conductor, box types)
4	Equipment for General Use	400-490 (cords, fixtures, switches, receptacles, transformers, motors, generators, capacitors, etc.)
5	Special Occupancies	500-590 (hazardous locations, healthcare, places of assembly, residential, agricultural, mobile homes, RV parks, etc.)
6	Special Equipment	600-695 (signs, X-ray, induction heating, electric vehicles, swimming pools, fire pumps, etc.)
7	Special Conditions	700-770 (emergency, optional standby, COPS, energy storage, fire alarm, comms)
8	Communications Systems	800-840 (radio, TV, comm, fiber)
9	Tables	Conductor properties, conduit fill, etc. — referenced from other chapters

Most-Referenced NEC Articles

Article	Subject	When you go here
110	Requirements for installation	Working space (110.26), labels (110.16), fault current (110.24), termination temp (110.14)
210	Branch circuits	Sizing (210.19), OCPD (210.20), GFCI (210.8), AFCI (210.12)
215	Feeders	Sizing (215.2), OCPD (215.3), VD (215.2 IN)
220	Branch + feeder + service load calc	Demand factors, optional methods, service sizing
225	Outside branches + feeders	Outdoor wiring rules
230	Service entrance	Service conductors, disconnects (230.71), GFP (230.95), SPD (230.67)
240	Overcurrent protection	OCPD types, sizes (240.6), tap rules (240.21), series-rated (240.86)
250	Grounding + bonding	System grounding, equipment grounding, GEC + EGC sizing
285	Surge protective devices	SPD types + application
310	Conductors	Ampacity tables (310.16), derating, insulation types
314	Outlet, device, junction boxes	Box fill, mounting
344	Rigid metal conduit	RMC requirements
348-358	Other raceway types	EMT, FMC, LFMC, ENT, etc.
368	Busways	Busway installation
392	Cable trays	Tray fill, ampacity, allowed cables
406	Receptacles	Receptacle types + GFCI + tamper-resistant requirements
408	Switchboards + panelboards	Bus sizing (408.30), 42-circuit limit (legacy)
430	Motors	Branch circuit + feeder + overload + disconnect for motors
450	Transformers	OCPD, location, ventilation
480	Storage batteries	Battery installation, ventilation, grounding
500-516	Hazardous locations	Class/Division system + equipment + wiring
517	Healthcare	Hospital electrical systems
625	EV charging	EVSE installation + EVEMS
690	Solar PV	PV system installation, rapid shutdown
700-708	Emergency + standby + COPS	Emergency systems, generators, ATS

IEEE Color Books — The Industry Bible Series

Color	IEEE #	Subject
Red Book	IEEE 141	Recommended Practice for Electric Power Distribution for Industrial Plants
Green Book	IEEE 142	Recommended Practice for Grounding of Industrial and Commercial Power Systems
Buff Book	IEEE 242	Recommended Practice for Protection and Coordination of Industrial and Commercial Power Systems
Brown Book	IEEE 399	Recommended Practice for Industrial and Commercial Power Systems Analysis
Gray Book	IEEE 241	Recommended Practice for Electric Power Systems in Commercial Buildings
White Book	IEEE 602	Recommended Practice for Electric Systems in Health Care Facilities
Yellow Book	IEEE 902	Guide for Maintenance, Operation, and Safety of Industrial and Commercial Power Systems
Bronze Book	IEEE 739	Recommended Practice for Energy Management in Industrial and Commercial Facilities
Emerald Book	IEEE 1100	Recommended Practice for Powering and Grounding Electronic Equipment

NFPA Standards Beyond the NEC

Standard	Subject
NFPA 70 (NEC)	Electrical installation
NFPA 70E	Electrical safety in the workplace (PPE, LOTO, arc flash work practices)
NFPA 70B	Recommended practice for electrical equipment maintenance
NFPA 110	Standard for emergency + standby power systems (test requirements)
NFPA 111	Standard for stored electrical energy emergency + standby power systems
NFPA 13	Standard for installation of sprinkler systems
NFPA 25	Inspection, testing, maintenance of fire protection systems
NFPA 72	National Fire Alarm + Signaling Code
NFPA 101	Life Safety Code (occupant safety, egress)
NFPA 780	Installation of lightning protection systems
NFPA 855	Standard for installation of stationary energy storage systems
NFPA 30	Flammable + combustible liquids
NFPA 497	Recommended practice for classification of flammable liquids/gases/vapors and area classification
NFPA 499	Recommended practice for area classification of combustible dusts
NFPA 1	Fire Code
NFPA 75	Standard for fire protection of information technology equipment (data centers)

ANSI/UL Standards

Standard	Subject
UL 67	Panelboards
UL 489	Molded-case circuit breakers
UL 508	Industrial control equipment
UL 508A	Industrial control panels (SCCR ratings)
UL 845	Motor control centers
UL 891	Switchboards
UL 924	Emergency lighting + power equipment
UL 1008	Transfer switch equipment
UL 1449	Surge protective devices
UL 1558	Metal-enclosed low-voltage power CB switchgear
UL 1741	Inverters, converters, controllers + interconnection equipment
ANSI C84.1	Electric power systems and equipment — voltage ratings
ANSI/IEEE C37	Switchgear standards series
ANSI/IEEE C57	Transformer standards series

OSHA References

OSHA reference	Subject
29 CFR 1910 Subpart S	Electrical (general industry)
29 CFR 1910.147	Control of hazardous energy (LOTO)
29 CFR 1910.331-335	Electrical safety-related work practices
29 CFR 1926 Subpart K	Electrical (construction)
29 CFR 1910.269	Electric power generation, transmission, and distribution (utility)

NEC Adoption Cycle + Local Amendments

NEC is updated every 3 years (2020, 2023, 2026, ...). Each state/jurisdiction adopts their own version on their own schedule — could be 2020, 2017, or earlier. Always confirm which NEC version is enforced in your jurisdiction.

Jurisdiction	Common variations
California	Title 24 modifications + state-specific amendments
New York City	NYC Electrical Code (NEC + extensive local amendments)
Chicago	Chicago Electrical Code (formerly required RMC everywhere; relaxed)
Massachusetts	MA250 — state amendments
Houston	Frequent NEC adoption + few amendments
Federal projects	UFC for DOD; GSA standards for federal buildings

NFPA 70E vs NEC — The Distinction

	NEC (NFPA 70)	NFPA 70E
Scope	Electrical installation	Workplace electrical safety
Audience	Engineers + electricians installing equipment	Workers operating + maintaining equipment
Force of law	Adopted by states/AHJs as code	Adopted by OSHA as workplace safety rule
Examples of content	Wire sizing, breaker sizing, grounding	PPE, LOTO procedures, arc flash boundaries
Update cycle	3 years	3 years

Drill — Quick Self-Check

Work each problem mentally; reveal to check. Goal: reflex, not deliberation.

Drill 1 · NEC organization

How many chapters in NEC?

Drill 2 · Most-cited NEC

NEC article for most-violated rule (working space)?

Drill 3 · Color books

IEEE Red Book covers?

Drill 4 · NEC vs NFPA 70E

Which covers worker electrical safety?

Drill 5 · Local amendments

Which take precedence over NEC?

NEC Article 645 — Information Technology Equipment

NEC 645 is a special article governing electrical installations in Information Technology Equipment (ITE) rooms — primarily computer rooms and data centers. It MAY be invoked instead of standard NEC requirements (645 + Chapter 7 emergency systems) when specific conditions are met.

NEC 645 Provision	What it allows / requires
645.4 — Conditions for compliance	Room must have: (1) approved automatic disconnect for ITE + HVAC, (2) heat detection, (3) only ITE personnel access, (4) separation from other occupancies by fire-rated walls, (5) listed ITE per UL 60950 / UL 62368
645.5 — Supply circuits + interconnecting cables	Permits power supplies + interconnect cables NOT in raceway (relaxed from Chapter 3 requirements). Allows under-floor wiring of certain types.
645.10 — Disconnecting means	Single emergency disconnect must drop ALL ITE + HVAC. Located outside room (or at room exit). Required to mount at every exit door.
645.11 — UPSs allowed without grouping	Multiple UPS units in the room are permitted without the grouping requirements of NEC 700
645.27 — Fire / smoke spread	Cables in plenum spaces must be plenum-rated (CMP) per Chapter 8

i

When 645 vs not 645?

If a room qualifies as an ITE room AND the design wants the relaxed wiring rules (under-floor cabling, etc.), 645 applies. Otherwise, standard NEC chapters apply (raceway requirements, etc.). Most modern hyperscale data halls do NOT invoke 645 because they want building-grade fire suppression + standard wiring rules; smaller IT closets often DO invoke 645 for the cabling flexibility.

Industry Standards Matrix — Who Governs What

Beyond NEC, multiple standards apply to electrical design. Knowing which body owns which subject saves time chasing references.

Standard	Body	Scope	Where applied
NEC (NFPA 70)	NFPA	Electrical installation	All buildings, all occupancies (US)
NFPA 70E	NFPA	Workplace electrical safety (PPE, LOTO, arc flash work)	Worker-facing operations
NFPA 110	NFPA	Emergency + standby power testing	Generator + UPS test programs
NFPA 75	NFPA	Fire protection of IT equipment	Data centers, server rooms
NFPA 76	NFPA	Fire protection of telecom facilities	Telecom central offices
NFPA 780	NFPA	Lightning protection systems	Tall buildings, critical infrastructure
NFPA 855	NFPA	Stationary energy storage installation	Battery rooms, ESS facilities
IEEE Color Books (141, 142, 242, 399, 1100)	IEEE	Industrial + commercial power systems	Engineering reference for design
IEEE 519	IEEE	Harmonic limits at PCC	Industrial, data centers, utility-customer interface
IEEE 1547	IEEE	Distributed energy interconnection	PV, ESS, generation interconnect
IEEE 1584	IEEE	Arc flash incident energy calculation	All arc flash studies
IEEE 80	IEEE	Substation grounding (touch + step)	Substation design
IEEE 485	IEEE	Lead-acid battery sizing	UPS + substation batteries
IEEE 43	IEEE	Insulation resistance testing	Motors + transformers
ANSI C84.1	ANSI	Standard electrical voltages + tolerances	All voltage classes
ANSI/IEEE C37	ANSI/IEEE	Switchgear + protection devices	Switchgear specs + protective relay device numbers
ANSI/IEEE C57	ANSI/IEEE	Transformer standards	Transformer specs + testing
ASHRAE 90.1	ASHRAE	Energy efficiency in commercial buildings	Lighting LPD, HVAC, motors, transformers
ASHRAE 90.4	ASHRAE	Data center energy efficiency	DC-specific energy metrics + design
ASHRAE TC 9.9	ASHRAE	Mission-critical environments (DC thermal)	DC inlet/outlet temperature ranges
TIA-942-B	TIA	Data center facility design + ratings	DC infrastructure rating + pathway design
BICSI 002	BICSI	Data center design + implementation best practices	DC engineering practice
Uptime Institute Tier Standard	Uptime Institute	DC reliability tier classification	DC marketing + design certification
UL standards (489, 508A, 845, 891, 924, 1008, 1449, 1558, 1741)	UL	Equipment listing + testing	All electrical product certification
OSHA 29 CFR 1910 / 1926	OSHA	Workplace safety (general industry / construction)	Federally mandated worker safety
IEC standards (60269, 60364, 61439)	IEC	International electrical standards	Outside US; some US adoptions

!

Hierarchy when standards conflict

When standards conflict, the hierarchy is generally: (1) Federal law (OSHA, NRC), (2) State + local code (adopted NEC + amendments), (3) AHJ-required standards (NFPA, etc.), (4) Project specs (Division 26), (5) Industry recommended practices (IEEE, BICSI). Higher in the list overrides lower. Always confirm specific hierarchy with project AHJ.

If You See THIS, Think THAT

If you see…	Think / use…
"NEC" / "NFPA 70"	Electrical installation. The starting reference for design.
"NFPA 70E"	Workplace safety. PPE, LOTO, arc flash procedures.
"NEC chapter 9"	Tables — conductor properties, conduit fill. Referenced from other chapters.
"NEC 110.26"	Working space requirements. Most-violated rule.
"IEEE Red Book" / IEEE 141	Industrial power distribution. Most-cited engineering reference.
"IEEE Green Book" / IEEE 142	Grounding. Definitive reference.
"IEEE Buff Book" / IEEE 242	Protection + coordination.
"NFPA 110"	Emergency power test requirements.
"UL 1741"	Inverter standard. Critical for PV + ESS interconnection.
"UL 489"	Molded-case circuit breakers.
"ANSI C84.1"	Voltage standards (120, 208, 240, 480, etc.).
"OSHA 1910.147"	Lockout/Tagout. Federal LOTO requirement.
"Local amendment"	Jurisdiction-specific NEC modifications. Always check.

★

Power Atlas — Complete

You've reached §32, the final section. The handbook now covers every concept from MEL through final code reference, all built around Atlas DC1 as the spine. Use the navigation panel on the left to revisit any section. Future updates will add a printable PDF version, more worked examples, and a coordinated set of practice problems.

Power Atlas

Table of Contents

The Three Quantities — Voltage, Current, Resistance

Ohm's Law — The Foundational Equation

Kirchhoff's Laws — How Current and Voltage Behave in Circuits

DC vs AC — Why We Use Alternating Current

The Sinusoid — What "AC" Actually Looks Like

Frequency — Why 60 Hz?

Why Three-Phase Is the Standard

The Magic of Transformers — Why AC Won

Resistance + Inductance + Capacitance — The Three Passive Elements

Why this matters: Reactive Power Explained Physically

Impedance — Generalized Resistance for AC

Phasors — Why Engineers Draw Triangles

Series + Parallel Circuits

Superposition Principle

Thevenin + Norton Equivalents

Symmetrical Components — Intro

Power = Energy / Time — kW vs kWh

Magnetic Field Basics — How Motors Work

The Power System Hierarchy

Generator → Wall Outlet — One Joule's Journey

Where to Go Next

The Power Triangle

The Six Core Formulas — 1φ vs 3φ

The HP → kW → FLA Chain

Conversions Worth Memorizing

Worked Example 1 — Atlas DC1 Chiller Motor

Given (from cutsheet)

Step-by-step

Worked Example 2 — Apartment Service Calculation

Given

Find: actual service current

Per-Unit Math — The Foundation of Fault Analysis

Pick a system base

Step-by-step

If You See THIS, Think THAT

Drill — Quick Self-Check

The Workflow — Where Each Document Lives

Meet Atlas DC1 — Your Reference Facility

The MEL — What You Receive

Anatomy of an MEL Row

ANSI Standard Voltages

SLD Symbol Legend

Worked Example 1 — Atlas DC1 MEL Walk-Through

What each row tells you

Worked Example 2 — Tracing Power Flow on the Atlas DC1 SLD

The path (server → utility)

What this trace gives you

If You See THIS, Think THAT

Load Type Definitions — One Place

Connected vs Demand Load

The Two Numbers Side by Side

Demand Factor vs Diversity Factor

NEC 220 Demand Factors — by Load Category

Continuous vs Non-Continuous — The 3-Hour Rule

What Counts as Continuous?

Motor Loads — Why They Have Their Own Rules

Worked Example 1 — Atlas DC1 Load Study (Side A)

Side A loads (per the MEL)

Step-by-step

Worked Example 2 — 50-Unit Apartment Building (NEC 220 Standard)

Per-unit connected load (NEC 220.42 + 220.55)

Step-by-step (multi-family demand)

Drill — Quick Self-Check

If You See THIS, Think THAT

MCA vs MOCP — The Two Numbers That Govern Everything

Side-by-Side Definition

NEC Table 430.52 — Motor Branch-Circuit Protection

The Branch Circuit Design Sequence

Standard Sizes — Breakers and Wire

Standard breaker sizes (NEC 240.6)

NEC 310.16 — copper THWN-2 (75°C col)

Conductor Derating — When 75°C Isn't 75°C Anymore

NEC 310.15(B)(1) — ambient temperature

NEC 310.15(C)(1) — adjustment for #conductors

Worked Example 1 — Atlas DC1 Chiller Branch Circuit

From the cutsheet / NEC

Step-by-step

Final design summary