Load shedding does not just remove the grid for a few hours; it inserts an extra charge-discharge cycle into your battery on every day it runs, and that cycle is drawn against the very cycle count your warranty uses to decide how long the battery should have lasted. The stage number sets how many hours you sit without power, the hours set how deeply the battery drains, and the depth of discharge sets what fraction of a full cycle gets written into the inverter's log. Run that arithmetic across a bad winter and a battery can spend its rated cycle budget years before the calendar catches up.
I keep a small notebook on the shelf above the inverter, the one next to the torch that never gets put back properly. There is a page in it with three columns: date, cycles, state of health. I started it after a conversation with an installer who said something that stayed with me — that nobody phones him about a battery that has died, they phone him about a battery that "used to get us through the evening." That is the whole shape of this problem. It doesn't announce itself. It shrinks.
So on the first Sunday of each month I open the monitoring app, write the cycle number down, and compare it to the last line. It takes about forty seconds. Most months the number has moved by twenty-something. In the months when the schedule was ugly, it has moved by sixty, seventy, once by close to ninety. That is the entire cost of load shedding, expressed in the only currency the battery actually understands.
What a stage actually bills you
A stage is not a fixed two hours. Stage 1 is roughly two hours off in a day, and each stage up adds about two hours, so stage 4 is around eight hours spread across one or two slots depending on your block and your municipality. Some of us get a single fat four-and-a-half-hour slot; some get two shorter ones. The block schedule matters, because two separate outages mean two separate discharges, and two discharges mean two trips down the voltage curve with a recharge in between — which is harder on the cells than one long, slow draw.
Now translate hours into energy. I use round numbers here on purpose, because these are illustrative, not a spec sheet. Suppose your essentials — fridge, router, a few lights, the TV, a phone charger — pull around 300 W. A two-hour slot is 0.6 kWh. If your battery has 5 kWh usable, that is a little over 10 percent depth of discharge, or about a tenth of a cycle. A single four-and-a-half-hour slot at 500 W is 2.25 kWh, closer to 45 percent, so roughly half a cycle. Two of those in a day and you have spent almost a full cycle before the sun has done anything for you.
Here is the sentence I want you to keep: it is not the total that gets you, it is the rate. A solar-only household might cycle its battery around 250 times a year through ordinary evening self-consumption. If a rough load shedding year adds 100 to 150 equivalent cycles on top of that, you have not doubled your battery's life expectancy problem, you have increased the rate at which you spend it by something like forty to sixty percent. Same battery, same warranty document, much faster meter.
Shallow is not free, but it is cheaper
One thing the glossy brochures skate over: cycle ratings are usually quoted as equivalent full cycles to a stated depth of discharge, typically 80 or 90 percent. A shallow cycle counts as a fraction, not zero. So a stage 1 evening barely registers and a stage 6 evening registers hard. The gentlest version of load shedding is the one where the battery dips to maybe 60 percent, gets topped up by the sun by ten the next morning, and only comes down again the following evening. The harshest is stage 6 twice a day, on a cloudy week, where the pack goes to the floor, gets slammed back up on grid charge at whatever current the inverter allows, and sits warm in a garage that faces west.
Which brings in the other silent cost: heat and time spent sitting low. Both chemistries suffer for it. A lead-acid bank left partially charged between outages sulphates and loses capacity in a way that never shows on a cycle counter at all. Lithium iron phosphate tolerates the deep discharge far better and is rated for many more cycles, but it does not enjoy being parked at a low state of charge for hours in a hot room, and a garage in the Lowveld in February is a hot room. That is not a warranty argument you win. It is just wear that happens outside the arithmetic.
The number the warranty actually reads
When a claim goes in, the person assessing it does not look at your notebook. They pull the log — the inverter's history or the battery management system's record — and they read state of health, cycle count, maximum and minimum state of charge, temperature excursions and fault history. That log is the case file. Everything you remember about being careful is, unfortunately, not evidence.
This is why I treat documentation as part of the system, not paperwork. The invoice with the serial number. Proof of registration if the warranty required it within a set number of days. The commissioning record. The certificate of compliance for the installation, because that is what confirms the work was done to SANS 10142-1 in the first place. If the warranty conditions mention a maximum depth of discharge, an operating temperature band, or a periodic inspection, those are not decorative. Every one of them is a lever someone can pull to reduce or decline a claim, and none of them are things you can argue your way out of after the fact.
I am not going to tell you what your warranty says, because I have not read it and terms differ. What I will say is that the boring exercise of reading the conditions once, early, changes how you set up the system. If the depth-of-discharge floor is stated, set your inverter's shutdown state of charge above it with a margin for measurement error and cold. If there is a temperature band, the battery does not go on the west wall of the garage next to the geyser.
The quiet indicators, in the order I noticed them
The first thing that changes is runtime, and it changes so gradually you will not trust your memory. The evening that used to run until the 10pm slot ends now ends at 8:40. Then the recharge takes longer than it used to — you watch the app the next morning and the pack is still absorbing at 11am when last summer it was full by nine. Later you notice the voltage sag: the kettle goes on and the lights flicker in a way they didn't use to.
None of these is a diagnosis. They are trend lines. That is exactly why the notebook exists — I can look back eighteen months and see the runtime shrinking against a schedule that has not changed, and I know more than I would from a single bad evening.
Counting it forward
I do not think there is a clever trick that makes load shedding free for a battery. There is only how hard you lean on it. Smaller essential-load circuits mean shallower discharges. Moving the geyser off the backup side takes a huge, hot load out of the picture. Letting the sun do the recharge instead of the grid, where the schedule allows, means less current, less heat, less stress. Keeping the room ventilated is unglamorous and genuinely matters.
What I want you to take away is that this is accounting, not catastrophe. A battery does not fail on the evening the lights go out for good; it fails after a few thousand small withdrawals, most of them made during a load shedding slot you did not think twice about. The counter is in the log, it is moving faster than the brochure implied, and nobody is going to point that out to you. The least you can do is write the number down.
