Battery storage for a Bangladeshi factory: sizing to the critical load, not the plant

Most factory battery projects are sized against the solar array, which is the wrong number. How to size against the load that cannot stop, pick a chemistry that survives the heat, site the battery room, and know when storage is not worth buying.

Ask a factory owner why they want batteries and the answer is usually that the solar should keep working when the grid goes. Ask which machines have to keep working and the room goes quiet. That gap is where most storage projects fail. The battery gets sized against the size of the solar array, and the array size is the one number in the building that has almost nothing to do with how much storage you need.

Start with the load that cannot stop

A 575 KWp rooftop, the size of the plant we commissioned at Ahad Jute Mills, does not imply a 575 kW battery. It implies nothing at all about the battery. The question is narrower: on a Tuesday afternoon when the utility supply drops, which circuits must stay energised, at what power, for how long, and what does it cost the factory if they do not?

In most Bangladeshi factories the honest answer is a short list. Process control PLCs and their input and output racks. The server room and the switches that carry production data. Emergency and stairwell lighting. The fire pump control and the alarm panel. Effluent treatment plant dosing and aeration, because letting an ETP go anaerobic costs far more to recover than it costs to keep running. Cold rooms holding product. In a poultry or feed operation, the ventilation fans, where the tolerance is measured in minutes.

Everything else, the looms, the extruders, the compressors, the moulding machines, is usually better served by the diesel generator the factory already owns. A battery that can run a production line is a different capital project from a battery that holds the plant safe until the generator picks up. Decide which one you are buying before anyone quotes.

• Meter before you size. Put a power logger on the sub-boards that feed the critical circuits and run it for a full production week, including a shift change and a Friday.
• Log the outages, not the average. The number that sizes the battery is the duration of the longest interruption you actually saw, not an annual availability figure.
• Separate standby loads from surge loads. A fire pump motor draws many times its running current at start. If it sits on the battery, the inverter is sized by that start and not by the running load.

Three numbers, not one

A battery system has a power rating and an energy rating, and they are bought separately even when they arrive as one line on a quotation.

• Power, in kW. Set by the largest instantaneous demand of the critical load, motor starting included. This sizes the battery inverter and the discharge rate the cells must sustain.
• Energy, in kWh. Critical load multiplied by the autonomy you want, divided by usable depth of discharge, then increased again for end-of-life capacity. A bank quoted at nameplate kWh will not deliver nameplate kWh on the last day of its warranty.
• Cycles and calendar life. A bank cycled daily for peak shaving ages differently from one that floats and discharges eight times a year. State the duty you intend in writing, because the warranty will be written around it.

Then set the autonomy honestly. Sixty seconds of ride-through until the generator synchronises is a completely different machine from four hours of evening operation. The first is a modest bank that pays for itself in avoided scrap and restart time. The second is a load-shifting project and has to be justified on arithmetic that only your own electricity bills can supply.

A sizing sequence that works

1. List the circuits that must stay live, with the board they come from and the breaker rating.
2. Log real power on those boards for a week. Record the peak, the mean during production, and the peak during a motor start.
3. Decide autonomy per circuit. Ventilation may need hours, PLCs may need thirty minutes, the ETP may need to run to the end of the shift.
4. Add the surge requirement and select the battery inverter rating from that.
5. Convert energy to cell capacity using usable depth of discharge and the guaranteed end-of-warranty retention, not the nameplate figure.
6. Re-check the answer against the generator. If the generator starts reliably inside a minute, everything past that first minute is a choice rather than a requirement.

Chemistry, and how each behaves in the heat here

Heat is the deciding variable in Bangladesh, and it is not the outdoor temperature that decides it. It is the temperature inside an unventilated plant room on the top floor of a factory in May, which is a good deal higher than anything the weather report shows.

Chemistry	How it behaves in a hot Bangladeshi plant room	Where it still makes sense
Flooded lead-acid	Needs topping up, ventilation and acid handling. Watering gets skipped in practice and the bank dies early.	Rarely defensible now for a new industrial installation.
VRLA and tubular gel	Sealed, familiar to local electricians, cheap per kWh at purchase. Battery makers commonly quote a rule of thumb that service life roughly halves for every ten degrees Celsius above the rated reference temperature. Shallow discharge only.	Small ride-through banks in an air conditioned room on standby duty, not cycling duty.
LFP (lithium iron phosphate)	The most thermally forgiving lithium chemistry in wide industrial use. Tolerates deeper discharge and daily cycling, and is less prone to thermal runaway than nickel-based lithium. Still needs the room held inside the maker's band or the management system derates charging.	The working default for factory storage now, particularly anything that cycles.
NMC	Higher energy density in a smaller footprint, less tolerant of high ambient temperature, and a more energetic failure mode. Demands real thermal management and a serious fire strategy.	Space-constrained sites where the room can be conditioned and the fire case can be argued properly.

Whatever the chemistry, ask for the derating curve rather than the datasheet headline. A bank that delivers rated capacity at its reference temperature and derates above it will spend most of its life derated in this climate, and the sizing has to allow for that. Ask what the battery management system does at high temperature as well. Many simply stop charging above a threshold, which means the plant looks healthy on the monitoring portal right up to the moment you need it.

Where the battery room goes

The battery room decision is usually made too late, after the modules are already on the roof and the only space left is a store cupboard beside the inverters.

• Not on the top floor if it can be avoided. Heat rises and the roof slab radiates. A ground floor room on an external wall runs cooler and is far easier for the fire service to reach.
• Floor loading. Battery banks are dense. Have the structural engineer confirm the slab before the racks are ordered, not after they arrive.
• Ventilation and cooling fed from the essential supply. Cooling that fails when the grid fails defeats the purpose of the battery it is protecting.
• Separation. Fire-rated separation from production areas and from the inverter room, a door opening outwards, and no other services routed through.
• Cable route. Long DC runs between battery and inverter cost copper and cost losses. Site the room near the inverters or carry the cable cost in the budget.
• Access for replacement. The bank will be replaced within the plant's life. If the only route in is a staircase with a turn in it, solve that on the drawing.

AC-coupled or DC-coupled

On an existing rooftop the choice usually makes itself. If the grid-tied inverters are installed and commissioned, an AC-coupled battery inverter is added on the AC side and the generating plant is left alone. It is simpler to permit, simpler to fault-find, and it does not disturb an approved net metering arrangement. The price is one extra conversion stage.

DC-coupled, where the battery shares the DC bus with the array through a hybrid inverter, is more efficient and is the better answer when plant and storage are designed together from the start. It is a poor answer when it means discarding working inverters. Retrofitting storage by changing the inverters on a plant already exporting under an approved agreement means going back to the utility, and that is a delay worth avoiding.

When storage is not worth buying

This is the part suppliers skip. There are sites where the correct recommendation is to spend the money somewhere else.

• When the generator already does the job. A well-maintained set that picks up inside a minute, on a process that tolerates that minute, leaves a battery very little to do.
• When the array is undersized against daytime load. If every unit the roof makes is consumed on site immediately, there is no surplus to store. Storage earns only where there is export you would rather keep, or an outage you cannot tolerate.
• When outages are long but rare. A bank sized for a four-hour interruption on a site that sees two a year is expensive insurance. Compare it honestly against the fuel cost of running the generator for those hours.
• When the room is wrong and cannot be made right. A lithium bank in a hot, unventilated, hard-to-reach room will not reach its warranty life, and the claim will fail on operating conditions.
• When nobody owns it. Storage needs someone reading the state of health report. If the electrical team is already stretched, put the service contract in the budget or leave the battery out.

The strongest case for storage in a Bangladeshi factory today is usually narrow and specific: a small, well-protected bank sized to hold the safety and control systems live through a transfer, plus whatever process load decides between a clean restart and a scrapped batch. That is a project with a defensible number attached to it. Sizing a battery to match the array is not.

Back to all Insights