One plant, two tariffs: what changes when you can see inside the panel

We simulated a mid-size auto-component plant minute by minute for two months and priced it under the Maharashtra and Karnataka industrial tariffs for 2026-27, once with only the monthly bill to go on and once with every feeder measured and reviewed each morning. Finding three ordinary problems the next day instead of from the bill made the two-month bill ₹1,32,540 lower in Maharashtra and ₹27,301 lower in Karnataka.

Simulated plant. The factory and its loads are synthetic, built from typical behaviour for each machine type. The tariff rules are real and cited to the page of each order. This shows a method, not a customer result or a forecast for any business. It makes no claim about meter accuracy, and all rupee figures are before the cost of monitoring and exclude fuel adjustment, electricity duty and taxes.

Full write-up: download the PDF, with every source and figure.

What is the plant?

A fictional auto-component plant in a Pune-area industrial estate on one HT connection with a contract demand of 540 kVA: four hydraulic presses, a servo press, a CNC bay, an air compressor, a chiller, a 110 kW tempering oven, and lighting and offices. It runs two shifts in October and one in November, when orders slow. Every feeder is modelled minute by minute for 61 days.

Three ordinary things go wrong:

14 October: to save power, the oven is switched off over lunch. At 12:00 it re-heats at full power just as the presses restart and the compressor refills.
15 October: the fuse feeding five of the six capacitor steps blows. Nobody notices.
16–21 November: a slow week; presses and the compressor are left running unloaded.

Stacked chart of one week of plant load by feeder, with a spike at noon on Wednesday 14 October when presses, compressor and oven restart together. — One week, feeder by feeder. The spike is the lunchtime restart on 14 October.

Which half-hour set the demand charge?

Both states bill industry on maximum demand: the highest average kVA in any 30-minute block of the month. On a normal day this plant peaks around 400–440 kVA. The lunchtime restart produced one half-hour at 492 kVA, about 50 kVA above any other block in October, and the whole month’s demand charge is billed on it.

Each state puts a floor under billing demand. Maharashtra bills the higher of recorded maximum demand or 75% of contract demand (405 kVA here). Karnataka bills the higher of maximum demand or 90% (486 kVA).

All half-hour demand blocks in October; one block on 14 October at 492 kVA stands above the rest, with the Maharashtra floor at 405 kVA and the Karnataka floor at 486 kVA marked. — All 1,488 half-hour blocks in October. One of them set the month's demand charge.

What did the failed capacitor cost?

Power factor drops from about 0.97 to 0.89 on 15 October and stays there. Maharashtra bills HT industry in kVAh, so the lower power factor raises the billed units every day. Karnataka bills in kWh and adds a surcharge only when the monthly average falls below 0.90. Here it stayed at 0.93 in October and 0.91 in November, so the Karnataka bill showed nothing.

Daily power factor falling from 0.97 to about 0.89 on 15 October and staying there through November. — Daily power factor. Same fault: a daily cost in Maharashtra, invisible on the bill in Karnataka.

Where does Nexura come in?

We ran the same two months again in two worlds.

Bill only. The lunchtime habit carries on, because the bill shows a kVA figure but not when or why. In Maharashtra the October bill prints a lower power factor and the capacitors are repaired on 12 November. In Karnataka no surcharge appears, so nothing prompts a repair.
With Nexura. Metrion measures every feeder inside the panel and Kairon prices the readings against the plant’s own tariff. Every morning our team reviews that analysis and emails the plant a short review of what it shows. The plant decides what to do; Kairon recommends and never switches anything.

Timeline: with only the bill, the restart habit and the capacitor fault run for weeks; with Nexura each problem is fixed within one or two days. — How long each problem runs in each world.

What happened	Seen with Nexura	Maharashtra	Karnataka
Lunchtime restart sets a new peak	Next morning; restarts staggered from 15 Oct	₹53,392	₹22,181
Capacitor fuse blows	16 Oct; fuse replaced 17 Oct	₹81,028	≈ ₹0
Machines left idling	17 Nov; stopped when idle from 18 Nov	₹5,681	₹4,847
Two months together	The individual values overlap	₹1,32,540	₹27,301

Monthly bills with and without Nexura for October and November in Maharashtra and Karnataka; the with-Nexura bill is lower in every month. — The same plant, with and without Nexura. Before the cost of monitoring.

Read this honestly. The 14 October half-hour still sets October’s peak in both worlds: monitoring cannot undo a half-hour that has already happened; it stops it happening again. And the gap between the states is real. In Karnataka the capacitor fault cost almost nothing over these two months, so the value there comes mostly from demand, and from whether the contract demand is set right.

What other changes were tested?

Bar chart of three changes priced under each tariff: staggering the restart, repairing capacitors, and cutting contract demand to 450 kVA, which raises the October bill in both states. — Each change re-priced under both tariffs. Red means the bill goes up.

Staggering the restart alone is worth ₹31,466 in October in Maharashtra and ₹2,184 in Karnataka, where the 90% floor absorbs most of the reduction. Cutting contract demand to 450 kVA after the quiet November looks tempting, but October’s 492 kVA would then exceed it and draw excess-demand penalties of ₹41,185 in Maharashtra and ₹14,785 in Karnataka. A contract-demand change is only safe once you know what sets your peak.

The bill, line by line

Monthly bill by component for the simulated plant under each tariff in October and November. — Base case (one restart, capacitor never repaired). Excludes fuel adjustment, electricity duty and taxes, so the totals are not a verdict on which state is cheaper.

How was it computed, and what does it not show?

Each feeder’s active and reactive power is generated minute by minute; a capacitor panel switches steps against the total; 30-minute demand blocks are averaged from the one-minute kVA.
Pricing is a pure function of the readings and each state’s tariff held as data. The two worlds use identical random draws, so every rupee of difference comes from when a problem was found and fixed.
It does not show meter accuracy, a real plant, a promised saving, or the full bill.

Sources

Maharashtra: MERC Order, Case No. 75 of 2025 (post-remand), 25 March 2026 — HT I(A) rates p.100, Time-of-Day Table 8 pp.76–77; MERC Case No. 217 of 2024 — billing demand p.787, excess demand p.785, kVAh billing para 7.16.7.
Karnataka: KERC Combined ESCOMs Tariff Order 2025 (27 March 2025), Annexure 9 — HT-2(a) rates p.550, billing demand and excess demand cl.1, power factor cl.2, p.545, Time-of-Day p.552; 30-minute maximum demand, Conditions of Supply cl.2.38.

Related: kVAh vs kWh billing explained.

Want to see this on your own panel?

Start with a free Bill X-ray: send your last three electricity bills and we return a written reading within two working days, with no obligation. If it shows something worth measuring, a 30-day paid pilot gives you a daily written review and a report reconciled to your next bill, and the fee counts towards your plan.

Get a free Bill X-ray