A Stage 4 outage is not one long cut. On most published schedules it works out to roughly eight hours of load shedding in a day, delivered as four separate blocks of about two hours each, and that changes what a battery bank is actually being asked to do. It is asked to discharge four times in a day with only partial recharge windows in between. So the number printed on the battery label, the nameplate kilowatt-hour figure, is almost never the number that keeps the lights on. Depth of discharge, inverter conversion losses, cold weather, cell ageing and the loads you leave running all subtract from that figure before a single plug sees anything.
The 04:10 block is the one that taught me this. I woke to the sound of the fridge compressor stopping, that particular silence a house makes when the compressor cuts and does not come back, and the little display next to the bed was blank. I lay there doing the arithmetic I always do: this is block one of four, and the inverter app said the bank was sitting at 78 percent. Not full. Seventy-eight, because the evening before had run late and the sun had only had a couple of clear hours to put anything back.
That is the thing nobody mentions when they say a system handles load shedding. Handling one block is easy. Handling the fourth block of the day is a completely different problem, and it is the one Stage 4 actually sets.
Stage 4 is a daily total, not a single event
Stage 1 is about two hours a day. Stage 2 is about four. Stage 3 about six. Stage 4 about eight, and the higher stages mostly keep adding blocks rather than stretching one long cut. There is usually a restoration buffer after each block, and the schedules rotate, so you do not get to pick which blocks land on you.
What that means in practice is that the question is not whether your battery can carry the house for two hours. Almost any properly built bank can. The question is whether it can carry four separate two-hour runs in a day and get back near full between each one. Those are not the same question, and a system that answers yes to the first can fail the second without anything being faulty.
I have watched it happen in my own house. First block, fine, everything stays on and I do not even think about it. Second block, fine, the bank gets back over 90 percent by lunchtime if the day is clear. Third block, mid-afternoon, still fine, but the recharge window is shorter than I would like because the afternoon geyser timer eats into it. Fourth block, around 18:00, and this is where the bank starts the evening at 70-something instead of full, and the evening is precisely when everything in the house wants to be on at once.
The number on the label and the number at the plug
Take a 48-volt battery rated at 100 amp-hours. That is 4.8 kilowatt-hours on paper. Very few of those kilowatt-hours are yours.
The first subtraction is depth of discharge. Lithium iron phosphate generally tolerates being cycled deeply and regularly without wrecking itself. Lead-acid types generally do not, and if you want a lead-acid bank to last, you treat the bottom half of its capacity as scenery rather than as energy. That single difference can move you from most of the nameplate being usable to roughly half of it, before anything else is counted.
The second subtraction is the inverter. Converting DC to AC is not free, and the inverter also carries a standing draw of its own. It is awake, running its electronics and its cooling, whether or not the house is pulling much. Over a day with four separate outages, that standby draw runs for eight hours, not for one, and it eats into the same pot of energy the kettle is competing for.
The third subtraction is temperature. A battery sitting in a garage on the Highveld in July does not hold what it holds in March. You do not get a warning about this. You just notice that the same block that used to end at 60 percent now ends at 45.
The fourth subtraction is age. Capacity fades over the years. A bank sized exactly to what one block needs in year one is not sized to what one block needs in year five, and nobody replaces a battery because a spreadsheet said so. They replace it when the lights go out at 10pm.
Run the illustration honestly and the picture moves. Say 4.8 kWh nameplate, cycled to 90 percent, through an inverter at roughly 90 percent efficiency, and about 3.9 kWh actually reaches your plugs. Knock off something for a cold garage and a bank that is not brand new, and you are closer to 3.4. Now the only remaining question is what you are running.
Divide by the load and everything changes
At 3.4 usable kilowatt-hours, the maths splits wildly depending on the house.
If the outage load is a fibre router, an ONT, two or three LED lights, a laptop charger and a television, you might be pulling 150 to 250 watts on average. Divide 3,400 watt-hours by 200 and you land somewhere near 17 hours. That is why a modest bank can carry a work-from-home setup through an entire Stage 4 day and barely notice it.
Now add the kitchen. A kettle is 1,800 to 2,400 watts for three minutes. A microwave is around 1,200. An induction plate can be over 2,000. A chest freezer cycles a compressor that draws maybe 100 to 150 watts while running, but starts with a surge several times that. A borehole pump is worse, and it runs longer. A geyser element is 3,000 to 4,000 watts and does not care about your plans.
Do the same bank against an average of 1,500 watts and you are looking at roughly two hours of coverage. Same battery, same day, same installer quote. One household gets through Stage 4 comfortably and the other one gets through one block and then goes dark in the second.
This is why I stopped asking installers what the runtime is. There is no runtime. There is a bank size divided by a load, and the load is a decision, not a specification.
What is actually running at seven in the evening
The reason the fourth block is the hard one is not only that the bank is lower. It is that the fourth block is usually the evening block, and the evening block is the heaviest one. Everyone is home. The lights are on. Someone is cooking. Someone else is charging something. The freezer door has been opened twice.
Early in a Stage 4 day I can be generous. By 18:00 I am not generous, I am deliberate. When a block starts, I walk the house and I do not guess at what things draw. The inverter display shows load in watts, so I switch something on, watch the number move, and switch it off again. That one habit has taught me more about my battery bank than any datasheet did. It is also how I found out that a device I had assumed was trivial was idling away a meaningful slice of an eight-hour outage day.
The other thing I do is sort loads into two groups in my head: things that can be interrupted and things that will not tolerate it. A freezer can go for hours if it stays shut. A router cannot be off if you are meant to be working. A pump filling a tank can wait for the grid. Once you have sorted that, you are not managing a battery anymore, you are managing a list.
Sizing against the day, not the block
If you are working through a quote, or trying to work out whether the system you have been offered will hold up, convert everything into a daily energy figure rather than a block figure.
Add up what you genuinely need to run during an outage day, in kilowatt-hours. Not the whole house meter reading, the essentials. Then get the usable capacity of the proposed bank, not the nameplate, and be specific with whoever is quoting about which depth of discharge they assumed and why that figure suits the chemistry they are proposing. Then look at the recharge side, because that is the half people skip. How many kilowatt-hours can the panels realistically put back into the battery in the windows you actually have, on a normal day rather than a perfect one.
If the daily energy you need sits close to the usable energy you have, the system will hold up in summer and embarrass you in winter, when the days are shorter, the battery is colder and the demand is higher. I would far rather see a bank that can do a full Stage 4 day and still finish the evening above 40 percent than one that can do four hours from full and nothing after that.
There is one more check worth doing. Ask how many discharge events the system was assessed against. If the runtime figure came from a single-block test, it tells you almost nothing about Stage 4, because Stage 4 is not one test. It is four tests with a recharge problem in the middle.
The awkward truth about the last block
By the time the fourth block lands, you are not running on the battery you bought. You are running on what is left of it after three cycles, a recharge window that was shorter than the spreadsheet assumed, a garage that got cold overnight, and a couple of years of capacity fade. That is the honest picture, and it is why I treat any single runtime number on a proposal with suspicion, whether it is mine or someone else's.
Stage 4 does not ask a battery bank for one big thing. It asks it for four medium things and a recovery, and the recovery is the part that usually fails. Subtract the depth of discharge, subtract the inverter, subtract the cold and the years, then divide by what you are genuinely willing to switch off. Whatever is left is your real number. It is smaller than the label, it will not look as good on a quote, and it is the only one that turns the lights back on at 19:00 on a Stage 4 evening.
