Genset Peak Shaving in Malaysia: Does It Pay?

A 1,000 kW peak on a Medium Voltage account costs RM89,270 a month on the General option and RM97,060 on Time of Use, before a single kWh of energy is billed. Under RP4, TNB charges capacity and network against the single highest 30-minute demand interval of the month, so every kW you keep off that interval is worth RM89.27 to RM97.06. That arithmetic is why genset peak shaving keeps coming up in Malaysian plant meetings, usually with a standby set that already sits outside the switchroom doing nothing eleven months a year. The saving is real. The question is whether diesel costs less than the charge it avoids.

What does genset peak shaving actually save under RP4?

It saves the demand charge on whatever load the genset carries during the month's highest half-hour. Nothing more. The Medium Voltage capacity and network charges are RM29.43 plus RM59.84 per kW on General, and RM30.19 plus RM66.87 per kW on ToU. Shave 200 kW off the peak interval and you avoid RM17,854 or RM19,412 for that month. Current rates for all voltage classes are published on the TNB commercial and industrial pricing page and the myTNB tariff schedule.

Two conditions matter before you go further. If you are on Low Voltage, capacity and network are billed per kWh, not per kW, so there is no maximum demand charge to shave and the whole exercise collapses. And if you are on ToU, demand recorded outside 2:00pm to 10:00pm on weekdays is not charged at all, so a peak that lands at 6:00am or on a Sunday needs no genset, only a tariff review.

The other thing to understand is that this is a bet on a single interval. You do not know in advance which half-hour will be the month's highest, so the genset has to cover every interval that could plausibly become it. Miss one afternoon because the set failed to start, and the month's demand charge resets to the unshaved figure. A partial success is worth nothing.

What does a genset kWh cost in Malaysia?

Work it out from fuel first. The numbers below are illustrative assumptions, not quoted prices, and you should replace them with your own supplier invoices and logged consumption.

Unsubsidised diesel around RM3.30 to RM3.50 per litre, with a decently loaded set producing roughly 3.3 to 3.5 kWh per litre, puts fuel alone at about RM1.00 per kWh delivered. Against that, running the genset also avoids buying those kWh from TNB, and Medium Voltage energy rates under RP4 sit around 29 to 31 sen/kWh. So the net incremental cost of a shaved kWh is roughly 70 sen once you credit the avoided energy charge, before maintenance.

Maintenance is not a rounding error here. Service intervals on a standby set are counted in running hours, and peak shaving turns a machine that logs 30 hours a year into one that logs several hundred. Oil, filters, coolant, injectors and an overhaul reserve realistically add another 10 to 25 sen per kWh depending on the set, which lands the all-in cost somewhere around 80 to 95 sen per delivered kWh.

The break-even is a number of run hours, not a number of kW

Here is the useful part. Cost equals hours multiplied by kW shaved multiplied by RM per kWh. Saving equals kW shaved multiplied by RM per kW. The kW cancels out. Whether you shave 50 kW or 500 kW, the break-even is the same number of running hours per month.

At RM89.27/kW avoided and an all-in genset cost of about 85 sen/kWh, break-even is roughly 105 hours a month. On ToU at RM97.06/kW, roughly 115. Call it 100 to 140 hours depending on how kind your fuel price and your maintenance assumptions are.

Now compare that to your exposure. The full ToU peak window, 2:00pm to 10:00pm on weekdays, is about 176 hours a month. If your load profile means you have to run across that whole window to be safe, genset peak shaving loses money and you should stop there. If your peak is a genuinely predictable block, say a two-hour afternoon overlap between chiller pull-down and a batch process, that is around 44 hours a month and the set has a real margin. Short, predictable, infrequent peaks pay. Broad plateaus do not.

Can the genset run in parallel with the grid?

Only with Suruhanjaya Tenaga approval and the right protection and licensing, and that requirement shapes the whole design. Peak shaving is not the same duty as standby. A standby set uses open transition: the grid drops, the set starts, the ATS transfers, and the plant takes a short break in supply. That is fine for a power cut and unacceptable as a monthly routine on a production line or a data hall.

Shaving without dropping load means synchronising to the grid and soft-loading the set, which makes it an embedded generator rather than a backup. That brings interface protection, anti-islanding and reverse-power arrangements, an application to your TNB regional office, and approval and licensing under Suruhanjaya Tenaga. Check requirements for your installation rather than assuming your existing standby registration covers it, because it generally does not. Budget time for this, not just money.

Two more operational points. Diesel engines dislike sustained light loading: run a set below roughly a third of its rating for long stretches and you get wet stacking, glazed bores and unburnt fuel in the exhaust, so the shave block needs to be sized to load the machine properly, not trickled. And the set still has a life-safety job. Fuel that you burn on a Tuesday afternoon is fuel that is not in the tank when the grid actually fails.

Cheaper things to try before you buy fuel

Load staggering costs nothing. If your peak is created by three chillers, two compressors and a lift bank all ramping in the same half-hour, sequencing them is free and permanent, and it works whether or not you ever start a genset. Our guide to cutting TNB maximum demand charges walks through the sequencing and interlock options, and the maximum demand calculator will show you what a given shave is worth on your tariff.

Batteries change the cost structure rather than the size of it. A genset's cost is mostly opex, paid per kWh delivered, so it stays cheap only while total shaved energy stays small. A battery energy storage system for peak shaving is almost entirely capex, sized in kWh, and gets cheaper per event the more often you use it. That is the real dividing line: rare and short favours the genset, frequent and repeatable favours the battery, and structural all-afternoon demand favours fixing the process.

Your interval data decides this

Peak duration is the deciding input, and it is sitting in your meter. Pull twelve months of 30-minute interval data and answer three questions: how many intervals a month sit within 10% of your billed maximum demand, how long each of those excursions lasts, and whether they fall inside the ToU peak window. Ten minutes of overlap that you can predict is a genset case, or often a rescheduling case. Six hours of elevated demand every weekday is neither, and no amount of diesel will fix it economically.

CobiNeural reads that interval data continuously and tracks a live Max Demand KPI against your billed peak, so you can see excursions building and act before the half-hour closes rather than reading about them on next month's bill. It also logs what a shave actually delivered, which is the only way to know whether the fuel was worth it.

If you want a second opinion on the numbers, send us a recent TNB bill and a few months of interval data and we will map your peak profile against genset, battery and staggering economics before you commit to anything. Start at request a demo.