What 112 kWh of Battery Covers at a Food Processing Plant

When the grid drops at a food processing plant, the cold rooms start warming up before anyone reaches the switchboard. Raw material spoils on a timeline measured in hours, production lines stop mid-batch, and restarting them costs more than the electricity they were using.

For a plant in North Africa running on an unstable mains supply and a diesel generator, we prepared a PV + battery proposal to keep the core loads energised. The first page of that document calls it an off-grid system. It is not one, and the most useful thing we can do here is explain why — because the gap between what a system is called and what it actually does is where most of these projects go wrong. Off-grid sizing starts from a daily energy balance, and that balance is where a proposal either holds up or falls apart.

This is a sample project and a technical proposal at concept stage, not a delivered installation. The load figures are estimates, the generation figure is a design average, and every runtime below is a calculation rather than a measurement.

What the system is made of

Layer Equipment Quantity Key figures
Generation 720 W N-type modules 42 (30.24 kWp) 6 strings of 7; 1500 V system
Conversion 12 kW three-phase hybrid inverters 3 (36 kW) 2 MPPTs each, 80–400 V; 24 kW max PV input
Storage 16 kWh wall-mounted LFP batteries 7 (112 kWh) 51.2 V / 314 Ah; 8000+ cycles
Combining Common DC busbar 1 set Main fuse + 3 independent DC breakers
Distribution AC cabinet 1 Three-phase 400/230 V
System configuration diagram of a 30 kWp solar PV array, 112 kWh LFP battery bank and three parallel 12 kW hybrid inverters

The architecture is deliberately modular rather than a single large machine. Fourteen separate pieces of equipment, each replaceable, is more installation work than one 36 kW unit — and it is the reason a single failure does not take the plant down, which is covered further below.

The number that decides everything: 151 against 360

Here is the comparison that needs to be on the first page of any PV + storage proposal for an industrial site.

Average daily PV generation: about 151 kWh.

Estimated plant consumption: about 360 kWh per day — roughly 15 kW through the working day and 10 kW overnight.

That is 42%. The PV array, over a full year, covers less than half of what the plant uses.

Energy-flow diagram of PV, battery, hybrid inverters, mains and diesel generator supplying plant loads

A system sized like this cannot run a plant of that consumption with no grid and no generator, and no honest reading of the numbers says otherwise. What it can do is take over the plant’s core loads when the mains fails, shave the peaks the diesel generator was carrying, and put the plant through the longest part of each day without burning fuel.

The proposal states this in as many words: the system is positioned for supply continuity, peak support and lower diesel consumption. If the owner wants a higher share of self-supply, the same platform accepts more PV and more battery — the inverters’ PV input is running at 10.08 kW per unit against a 24 kW limit, so there is room to double the array without changing the conversion layer.

That spare capacity is worth knowing about before signing, not after. It is the cheapest expansion path on the site.

So what does the plant actually buy?

Three things, and it is worth being precise about each.

Continuity for the cold chain. The storage bank is sized to carry the night-time core load — cold-store holding plus lighting — from sunset to sunrise and beyond. That is the load that cannot be allowed to stop, and it is the one the sizing was built around.

A ceiling on diesel. The generator no longer runs to cover every shortfall. It runs when the battery reaches its low-SOC threshold, or through a stretch of consecutive overcast days. Fewer running hours, less fuel, less noise, less maintenance per kWh delivered.

A predictable daily cycle. PV supplies the load through the day, surplus charges the battery, the battery discharges overnight, and the grid or generator tops up when state of charge bottoms out. Four modes, closed loop, no operator.

What the plant does not buy is independence. Every claim about this system should be read against that.

Supply time is a division problem

The storage number that matters is not 112 kWh. It is 100.8 kWh — 112 kWh at the recommended 90% depth of discharge, with the remaining 10% held back for cell life.

Bar chart of battery supply time at 36 kW, 20 kW and 10 kW loads with 100.8 kWh usable energy

Divide the usable energy by the load, and the answer is immediate:

Load condition Calculation Continuous supply
Full load, 36 kW 100.8 ÷ 36 2.80 h
Typical plant average, 20 kW 100.8 ÷ 20 5.04 h
Night core load, 10 kW 100.8 ÷ 10 10.08 h

Two things about this table are easy to misread.

The first is that the 36 kW row is not a normal operating condition. It is the maximum continuous output of the three inverters combined, and it exists so that the plant can start and run everything at once. At that load the battery lasts under three hours, which is why transient high-power events are shared between PV, battery and mains rather than being carried by the battery alone.

The second is that these figures are battery-only. They assume no PV contribution at all, which is the conservative case. On a sunny day the array is supplying part of the load while the battery discharges, and the actual duration is longer.

The same arithmetic runs in the opposite direction when you ask about battery ageing. A battery at end of life delivers roughly 90% of nameplate, so the 10.08-hour figure becomes something closer to nine hours with the same load. If a specific duration has to be guaranteed for the life of the system, size against that number, not the day-one figure.

Why seven modules per string and not ten

The string design is where a proposal either matches its equipment or quietly does not.

Each string is 7 modules in series. At standard test conditions that gives an open-circuit voltage of 351.82 V and an operating voltage of 295.68 V, carrying 17.14 A and delivering 5.04 kW.

The inverter’s MPPT window is 80 to 400 V. The operating voltage sits comfortably inside it — not near the top, where a cold morning would push open-circuit voltage past the limit, and not near the bottom, where a hot afternoon would drop the array below the tracking range.

Two more checks come out of the same arithmetic. Each inverter takes two strings, so its PV input is 10.08 kW against a 24 kW limit — a 58% margin that leaves room for expansion. And the module’s maximum system voltage is 1500 V against a string that operates below 400 V, so there is no over-voltage risk at the array level.

The low-temperature open-circuit voltage needs to be recalculated for the final design from the module datasheet’s Voc temperature coefficient. It is the one number in this section that should not be taken from the concept document without rechecking, because the design margin at the coldest expected morning is what protects the inverter.

One busbar, three breakers, no single point of failure

The seven LFP batteries connect in parallel to a common DC copper busbar. Three DC cables leave that busbar, each with its own breaker, feeding the three inverters.

Wiring topology of six PV strings, common DC battery busbar and three-phase paralleled inverter outputs

That arrangement looks like plumbing. It is actually the redundancy design.

If one inverter fails or needs maintenance, its DC breaker opens. The other two inverters and the full battery bank keep running, supplying the plant at 24 kW — two thirds of capacity, no shutdown. The same logic holds on the PV side: six independent strings, no combining before the inverter, so a fault in one string does not drag the other five down with it. And a single battery can be isolated without losing the capacity of the other six.

On the AC side, the three inverters parallel through a cable that keeps voltage, phase and frequency synchronised, and the AC cabinet carries the breakers, surge arresters, residual-current protection and metering for the combined output. Parallel inverters that are not synchronised produce circulating current between themselves; the parallel comms cable is what prevents that, and it is worth confirming at commissioning rather than assuming.

The whole arrangement supports 100% unbalanced load, which matters in a plant where three-phase motors and single-phase lighting and control loads share the same cabinet. No separate transformer is needed to absorb the imbalance.

What the protection design has to survive

The site is hot, dusty and humid. That is the design condition, and it decides most of the equipment selection.

Modules are IP68 with a power temperature coefficient of −0.26%/°C, so they lose less output on a 45°C afternoon than a standard module would. Inverters are IP66 with forced-air cooling and a −20 to 60°C operating range, derating rather than tripping when the air is hot. Batteries are IP20 and belong indoors — the proposal puts them in a ventilated switch room, away from direct sunlight and away from direct air-conditioner airflow, with the room temperature recommended to stay under 35°C.

That last point is the one that gets skipped on site. An IP20 battery bank installed in a room that reaches 45°C in summer will lose capacity and life, and no amount of inverter quality compensates for it. The ventilation and the temperature monitoring are part of the system, not accessories to it.

Protection on the DC side is per-string: fuse, isolator and surge arrester on each of the six strings, a main fuse on the battery busbar, and BMS interlocks that stop charging and discharging across all three inverters when a cell goes out of range. On the AC side, breakers, surge arresters, residual-current protection and anti-islanding. Earthing is bonded across module frames, inverter enclosures, battery cabinets and the distribution cabinet, with an earth resistance target of 4 Ω or better.

How to test it before you accept it

A system this modular can be tested module by module, and the proposal sets out four tests that should happen before handover:

A stepped load test up to 36 kW. Load risen in stages, confirming that the three inverters parallel cleanly and that no single stage trips the AC protection.

A mains-cut transfer test. Open the incoming breaker with the plant running and confirm how long the battery takes to pick up the load. The inverter’s internal transfer is fast enough for most industrial loads; the question is whether the site’s own switchgear adds a gap on top of it.

A BMS interlock test. Force a cell out-of-range condition and confirm that all three inverters stop charging and discharging, rather than one of them continuing to push current into a battery it should not be charging.

A 72-hour continuous run. Three days covering day and night cycles, with the operating record retained. This is the test that catches the faults that only appear on the second night.

None of these are unusual. All of them are easier to run before the plant is producing than after.

What still has to be measured

The load figures in this proposal are estimates, and the design should be re-checked against a measured load curve before equipment is ordered. Specifically:

1. A 15-minute interval load profile, covering a full production week and a weekend, split by phase.

2. The actual night-time core load, since the 10 kW figure is what sets the storage capacity and the 10.08-hour claim.

3. Peak demand and start-up currents for the largest motors and compressors, including lift and cold-store compressor starts.

4. The mains failure pattern — how often, how long, and whether outages cluster in a season.

5. Site voltage and grid quality, including whether the mains is stable enough for grid-tied operation or mostly absent.

6. The diesel generator’s rating and control interface, if it is being retained.

If the measured night load comes in at 15 kW rather than 10, the storage duration falls to about 6.7 hours and the sizing conversation changes. That is not a reason to delay the proposal; it is the reason the proposal says measured values prevail.

FAQ

Q: Is this system really off-grid?

A: No. It has a mains interface and a diesel generator interface, and both are meant to be used. A genuinely off-grid design for a 360 kWh/day load would need an array and a battery bank several times this size. This system is sized for continuity of the core loads and for reducing diesel run hours, and it is honest about that limit.

Q: Why not just add more PV until the plant is self-sufficient?

A: You can, on the same platform — each inverter has 24 kW of PV input capacity and is using 10.08 kW, so the array could roughly double without new inverters. What changes is the cost per unit of energy and the amount of surplus you have to curtail or store in the middle of the day. Whether that is worth doing depends on the diesel price and the daytime load profile, and both need measuring first.

Q: Is 112 kWh enough for a cold store?

A: For the night-time core load at 10 kW, yes, with about ten hours of cover. For the whole plant at full 36 kW output, no — that is under three hours, and that is not what this battery is for. The number to design against is the load the plant cannot afford to lose, not its total connected load.

Q: What happens if one inverter fails on a Saturday night?

A: The plant keeps running on the other two, at 24 kW. Its DC breaker isolates the failed unit, the battery bank stays connected to the healthy pair, and the repair can be scheduled rather than treated as an emergency. That is the reason for three 12 kW units instead of one 36 kW unit.

Q: How long do the batteries last?

A: The cells are rated at 8000 cycles or more at 0.5C with 80% depth of discharge. Whether the installation reaches that depends mostly on temperature — a battery room held under 35°C will get far closer to the rating than one that is not. Cycle life claims are only meaningful alongside the conditions they were tested at.

About this proposal

This is a technical proposal preview prepared to show the design reasoning, not a delivered project. The load figures are estimates, the generation figure is a design average, and the equipment is described by reference specification. All supply durations are calculated values, and the on-site measured load curve prevails over every number in this article. The string voltage check should be recalculated from the module datasheet before the design is frozen.

If you are running a cold store or a processing line on an unstable grid, the most useful thing to send us is three months of interval load data and a description of which loads cannot stop. The battery size follows from those two answers far more directly than from the size of the plant.

This proposal was first shared on Facebook and LinkedIn.

YURI POWER Engineering Team · November 6, 2026