---
title: "Connected load, maximum demand and diversity factor"
description: "The three numbers get used as though they were one. Which to size on, typical demand factors by load type, and what getting it wrong costs on the bill."
date: "2026-03-03"
author: "Divakar B"
source: "https://energycalchq.com/blog/connected-load-maximum-demand-diversity"
license: "© 2026 Divakar B. Quote with attribution to https://energycalchq.com/blog/connected-load-maximum-demand-diversity"
---

Three terms turn up on every load schedule, and they get used interchangeably by
people who should know better. They are not the same number, they are not
interchangeable, and choosing the wrong one sizes a transformer, a DG set and a
DISCOM contract that you will pay for every month for twenty years.

Here is what each one means, what factors to apply, and where the money is.

## The three terms

**Connected load** is the sum of every nameplate on site. Every motor, every
light, every socket circuit, at full output, all at once. It is a real number
and it is easy to calculate, which is why people use it. It describes a
situation that never occurs.

**Maximum demand** is the highest average load over a defined interval —
normally 15 or 30 minutes, because that is what the utility's meter records.
This is what the supply must actually deliver, and it is the number to size on.

**Demand factor** connects them:

```
Demand factor = maximum demand / connected load
```

**Diversity factor** is the one that gets muddled. It is not the same thing:

```
Diversity factor = sum of individual maximum demands / coincident maximum demand
```

It is always **greater than or equal to 1**, because the parts of an
installation do not all peak at the same moment. The canteen peaks at lunch, the
HVAC peaks at three in the afternoon, the car park lighting peaks at night. Add
their individual peaks together and you get a number nobody ever draws.

If a figure you have been handed is less than 1, it is a demand factor whatever
the drawing calls it.

## Typical demand factors

Starting points for a commercial or light industrial building. They are
judgement, and the last section of this post explains how to replace them with
something better.

| Load | Demand factor | Why |
|---|---|---|
| Lighting, offices | 0.9 | Nearly all on together during the day |
| Lighting, warehouse | 0.6 | Zoned, and much of it stays off |
| Socket outlets, general | 0.2–0.3 | Sockets are provisioned, not used |
| HVAC / chillers | 0.8 | Runs continuously, modulates rather than stops |
| Lifts | 0.4–0.5 | Intermittent, and never all rising together |
| Kitchen equipment | 0.5–0.6 | Heavily cyclic |
| Process motors, continuous | 0.8–0.9 | Sized close to their duty |
| Process motors, batch | 0.5–0.7 | Depends entirely on the process |
| Welding sets | 0.3–0.5 | Very low duty cycle, very high peak |
| Fire pumps, standby | 0.0–0.2 | Excluded from normal demand — but see below |
| EV chargers | 0.6–1.0 | Rising, and load management changes it completely |

Socket outlets are where estimates diverge most. A floor with sixty twin sockets
has a connected load in the tens of kilowatts and a real demand of a few. Nobody
plugs a 13 A load into every socket.

Fire pumps and sprinkler sets are excluded from maximum demand because they only
run in an emergency — but the supply, the cable and the switchgear must still be
capable of carrying them. Exclude them from the demand calculation and include
them in the capability check.

## Worked example: an office building

| Load | Connected | Factor | Demand |
|---|---|---|---|
| Lighting | 45 kW | 0.90 | 40.5 kW |
| Small power / sockets | 60 kW | 0.30 | 18.0 kW |
| HVAC | 180 kW | 0.80 | 144.0 kW |
| Lifts, 3 × 15 kW | 45 kW | 0.50 | 22.5 kW |
| Canteen | 40 kW | 0.60 | 24.0 kW |
| Fire pumps (standby) | 30 kW | 0.20 | 6.0 kW |
| **Total** | **400 kW** | **0.64** | **255 kW** |

![Daily load profile showing 400 kW connected load against a 255 kW measured peak](/blog/load-profile-maximum-demand.svg "The curve is a real day. The upper dashed line is what a connected-load calculation would have bought.")

At 0.9 power factor that is 283 kVA. Add 15 % for future load and you are
looking for 326 kVA — a 400 kVA transformer, with the [transformer sizing
calculator](/tools/transformer-sizing) confirming the loading lands at a healthy
71 %.

Size the same building on connected load and you specify 400 kW / 0.9 = 444 kVA,
plus margin, and buy a 630 kVA transformer. That is one frame size larger, a
larger HT breaker, a larger incomer, and no-load losses running around the clock
for a capacity the building will never ask for.

## Where the money actually is

Oversizing costs more than the extra capital, and most of it is recurring.

**Contract demand.** Indian industrial and commercial tariffs bill a demand
charge on contracted kVA, every month, whether you draw it or not. Contract for
400 kVA and peak at 283 and you are paying for 117 kVA of nothing. Contract too
low and you pay a penalty on the excess — typically 1.5 to 2 times the normal
demand rate — so the correct number matters in both directions.

**Transformer no-load loss.** Present from the moment the HV is charged,
independent of load, and larger on a larger unit. Over a thirty-year life it is
a substantial figure.

**DG fuel.** A generator sized on connected load runs at a low load factor.
Below about 30 % load, a diesel wet-stacks, glazes its bores and costs more in
maintenance than the fuel it saved. The [DG sizing
calculator](/tools/dg-sizing) sizes against demand and against the largest motor
start, which is usually what governs anyway.

**Cable and switchgear.** Everything downstream is chosen from the same wrong
number.

## Stop guessing when you can measure

Every factor in this post is a substitute for data. If the installation exists
in any form, better data is available:

- **The electricity bill.** Indian commercial and industrial tariffs print
  recorded maximum demand every month. Twelve bills give you the annual peak and
  the seasonal shape for free.
- **The meter itself.** Most LT trivector and smart meters hold MD registers and
  a load survey, readable over [Modbus RTU](/blog/modbus-rtu-energy-meter-setup)
  or from the DISCOM's portal.
- **A clamp meter and a fortnight.** A logging clamp on the incomer, left for
  two weeks including a Monday and a weekend, beats any table.

For an extension, measure the existing demand and add the new load with its own
factor — do not re-estimate the whole building. The existing plant has already
told you what it draws.

## Three mistakes worth naming

**Applying a diversity factor twice.** Once at the sub-board level and again at
the main board is a common spreadsheet error, and it produces a supply that is
genuinely too small. Apply diversity once at each level of aggregation, and be
explicit about which level a factor belongs to.

**Forgetting that demand is measured in kVA, not kW.** The utility bills
apparent power. A 255 kW demand at 0.75 power factor is 340 kVA of contract
demand; at 0.95 it is 268 kVA. Correcting power factor reduces contract demand
directly — see [what the penalty
costs](/tools/power-conversion) — and on many tariffs that saving alone repays
the capacitor bank inside a year.

**Assuming diversity where there is none.** A process line where every motor
starts on the same signal has a demand factor near 1.0. A data centre runs flat
out at three in the morning. Diversity is an observation about how a building
behaves, not a discount you are entitled to.

Convert your demand figure into current for the supply side with the
[three-phase current calculator](/tools/three-phase-current), then size the
transformer, the incomer and the cable from that one number — measured if you
possibly can, estimated with the factors above if you cannot.