Five kilowatt-hours of LiFePO4, spent honestly, runs a fridge, a router, a handful of LED lights, a television, a laptop and a couple of phone chargers through a four-hour load-shedding slot, and it may still have charge left when the grid returns. It will not carry that list plus a kettle plus a microwave at the same instant, because at that point the ceiling stops being energy and becomes current. The 5 kWh on the box is also not 5 kWh you can spend. Unusable capacity, inverter conversion losses and the inverter's own idle draw come off the top before the first appliance switches on.
20:04 on a Tuesday in July. The inverter in the passage gives its two-note beep and the house goes quiet in that particular way where the fridge compressor suddenly becomes the loudest thing in the room. The strip light over the counter stays on. The router's green LED doesn't even blink. I'm standing at the counter with the monitoring app open, and I'm not looking at the state of charge. I'm looking at the watts going out, because that one number tells me more about the next four hours than any percentage ever will.
That's the whole method, really. Not a headline figure. A list, a few labelled assumptions, and two or three divisions.
The label is nameplate, not spendable
A 5 kWh LiFePO4 pack is almost always a 51.2 V, 100 Ah unit, sixteen cells in series. Multiply the two and you get 5.12 kWh, and that is the number that lands on the datasheet. It gets quoted at a specific discharge rate, usually 0.2C or 0.5C, at 25 degrees Celsius, from a new cell in a laboratory. Move any of those conditions and the deliverable number moves with them. A pack that is cold in a Highveld garage in June, or being asked to work hard at a high C-rate, gives back less than the sticker.
Depth of discharge is the next cut. LiFePO4 tolerates going deep far better than lead-acid ever did, and plenty of manufacturers will let you run to 90 or even 100 percent depth of discharge. That doesn't make it a good planning number. Most people I know size on 80 percent, which leaves a fifth of the pack in reserve and keeps the cycle count honest, and some go as conservative as 70 percent on a pack they intend to keep for fifteen years. The number you pick is an assumption, and it should be written down where you can see it.
Then the inverter takes its share. Conversion efficiency on a decent hybrid inverter sits somewhere between 88 and 94 percent, and it is usually worst at low load, which is exactly where you'll be sitting at 03:00 when the only thing running is a router and the fridge. On top of that, the inverter draws power for itself just by being switched on. Twenty to fifty watts is a normal idle figure for a 5 kW hybrid, and over a four-hour slot that is 80 to 200 Wh gone before any appliance sees a joule of it.
Walking the house with the app open
I build the list at night, room by room, with the app open on my phone. What goes on the list is not what a plug says it can draw. It's what it actually draws, multiplied by how long it actually runs.
The two figures are rarely the same. A fridge-freezer might carry a 150 W rating on its plate, but the compressor cycles — it runs perhaps 35 to 40 percent of the time in a normal household with kids going in and out. So the fridge contributes an average of roughly 60 W, not 150 W. The router is the opposite: 15 W, honest, on for the entire slot. A television is somewhere in between, because nobody watches for the full four hours.
If the backup circuits are already installed, the inverter's own monitoring does most of this work for you. It logs what the essential loads actually pull, hour by hour, and that log is worth far more than any label on any appliance. If you haven't installed anything yet, a plug-in energy meter from the hardware shop, the kind that sits between the wall socket and the plug, will do the job one socket at a time. It takes an evening and it produces better numbers than a quote will.
The four-hour list, with every assumption on the table
Here is an illustrative list, not a prediction, and every number in it is a labelled assumption. It covers a modest household on a backup sub-board for four hours.
| Load | Draw | Effective hours in slot | Energy |
|---|---|---|---|
| Fridge-freezer, 250 l | 150 W running, roughly 40% duty | 1.6 h | 240 Wh |
| Chest freezer, 200 l | 120 W running, roughly 35% duty | 1.4 h | 168 Wh |
| Six LED downlights | 30 W total | 4 h | 120 Wh |
| Router and fibre ONT | 15 W | 4 h | 60 Wh |
| Television | 90 W | 3 h | 270 Wh |
| Laptop charger | 65 W | 2 h | 130 Wh |
| Two phone chargers | 12 W total | 3 h | 36 Wh |
| Microwave | 1 200 W | 0.1 h | 120 Wh |
| Total at the plug | 1 144 Wh |
Now push that back through the system, in the order the energy actually flows.
Loads: 1 144 Wh at the plug. Divide by 0.90 inverter efficiency, which is a middling assumption: 1 271 Wh drawn from the battery to serve those loads.
Inverter idle: 30 W times four hours equals 120 Wh drawn from the battery regardless of what is plugged in.
Battery energy consumed: 1 271 plus 120 equals 1 391 Wh.
Nameplate required at 80 percent depth of discharge: 1 391 divided by 0.80 equals 1 739 Wh, so about 1.75 kWh of nominal capacity.
That is the honest answer for this list, and it is smaller than most people expect. Change the assumptions and watch it move. At 90 percent depth of discharge it drops to about 1.55 kWh. If the inverter idles at 50 W instead of 30 W, it climbs to roughly 1.9 kWh. If the fridge duty cycle doubles because the kitchen door is open all evening, add another 240 Wh and roughly 300 Wh of nameplate. None of those variants is a promise. They are the edges of the estimate, and knowing where the edges are is the point.
What 5 kWh gives back
Run the same chain in reverse. A 5.12 kWh pack at 80 percent depth of discharge yields 4.10 kWh of usable energy. Take off 120 Wh of inverter idle over the slot and you have 3.98 kWh left to convert. Multiply by 0.90 and roughly 3.58 kWh reaches the appliances.
That is a little over three times the list above. In theory it is twelve hours of that household. In practice you will not get twelve hours of that household, and the reason has nothing to do with the battery chemistry. Load shedding doesn't arrive as one clean four-hour block. At Stage 4 to 6 you get two to four slots a day with two to four hours of charging between them, and if that gap falls before sunrise or during a rainy afternoon, the pack doesn't come back to full. Add cold cells, a pack that has aged a few hundred cycles, and the human habit of plugging in everything you own the moment the beep goes, and the practical window narrows.
So the working answer for a 5 kWh pack over a single four-hour slot is the base list, plus one energy-hungry appliance used with discipline. A 2 000 W kettle run five times is about 500 Wh. A slow cooker for the whole slot is more. Running the microwave and the television together for the full four hours alongside the fridge is a different calculation altogether, and a heavier one.
When the energy says yes and the battery says no
Energy is only half the constraint, and it is the half everyone calculates. The other half is how fast the pack can push current out, and it is where a 5 kWh battery quietly fails a 1.75 kWh job.
A 100 Ah pack at 51.2 V with a 100 A battery management system can deliver roughly 5.1 kW continuously, which is comfortable. Plenty of 5 kWh packs ship with a 50 A BMS, and that ceiling drops to about 2.5 kW. Now total up what switches on at the same instant: a 2 000 W kettle, a 1 200 W microwave, a fridge whose compressor surges for a second on start, and a 150 W television. That is over 4 kW of demand against a 2.5 kW ceiling, and the BMS will either shut the pack down or refuse the surge. The energy arithmetic never sees it coming, because over four hours the energy is fine.
When people in South Africa add a second battery in parallel, this is usually the real reason, even when they describe it as wanting more hours. Two packs in parallel double the current ceiling as well as the kilowatt-hours. If you are sizing for a kettle and a microwave on the same sub-board, work out the peak simultaneous load first and let that number, not the four-hour energy total, decide the pack count. Ask for continuous discharge current in amps rather than a C-rate, because a C-rate figure is meaningless unless you know which capacity it was quoted against.
Reading a quote against your own list
Once you have your own list and your own assumptions, a quote becomes something you can interrogate rather than something you either believe or don't.
Take the quoted kilowatt-hours and ask whether that is nameplate or usable. Ask at what discharge rate and at what temperature the figure was measured. Ask for the continuous and peak discharge current and the BMS rating, because those decide whether the kettle works. Ask what cycle life is claimed, at what depth of discharge, and down to what percentage of original capacity. Ask whether the pack can be paralleled later with the same model, because a pack that can't be expanded is a pack you replace.
Then compare the quoted figure to your calculation and expect the quote to be larger. It should be. Sizing on a single four-hour window with everything ideal is how people end up short in the second week of July. Treat the difference between your number and the installer's number as a conversation about assumptions, not as an error in either direction.
One last thing worth checking, because it sits outside the battery entirely: whether the quote includes the essential loads sub-board that carries the circuits you actually want backed up. A battery sized against a list that was never wired to the backup side is a battery sized against nothing. And whatever gets installed has to end up on a certificate of compliance, with the inverter and its changeover arrangement declared, because that paperwork is what stands between you and a problem at resale or at municipal registration.
The number that answers the question is not printed on the battery. It's on the back of an envelope at 20:04, written in your own handwriting, with the app open beside it. And it changes the moment you add one more appliance to the list.
