Exported solar earns a credit. Self-consumed solar avoids a charge, and on a South African municipal account those two numbers are not the same. The unit you push out at midday is normally credited at a fraction of what the same unit costs you when it comes back through the meter in the evening. Establish that gap before anything else, because it decides how big the array should be, whether a battery earns its place on the roof, and whether exporting is a design goal at all or just whatever is left over.
I keep two columns on the back of an old invoice. The left column is headed "into the house", the right one "into the street". When someone asks me whether six more panels is a good idea, that envelope is what I reach for, because the left column converts at the retail rate and the right column converts at whatever the credit is, and the arithmetic is not symmetrical no matter how neatly the arrows are drawn on the quote.
Read the account as two different documents
I do this on a Sunday morning, when the account has been lying on the kitchen table long enough for me to stop being annoyed by it. Page one is what you pay. Page two, if you have one, is what you're paid. They are written in the same typeface and they are not the same kind of number.
Start on the charges. On a conventional municipal account there is the energy charge, usually a network or service component, sometimes a capacity or demand element, and VAT sitting on top of all of it. Every unit you avoid importing escapes the whole stack. That's the left column, and it is a stack.
Now find the export line. It's one line. It doesn't carry your network charge, it doesn't carry your service charge, and depending on how your municipality structures it, it may be an avoided-energy credit, a fixed feed-in tariff, or a net-billing arrangement closer to a wholesale figure. Different structures, same effect: one line sitting against a stack.
Then do the only calculation that matters at this stage. Put the credit over the retail unit and look at the fraction. If it comes out close to one, export is nearly as valuable as self-use and you can size generously. If it comes out as a fraction of a fraction, everything in the design tilts towards the house side of the meter.
The midday kilowatt-hour and the evening kilowatt-hour are not the same animal
Put yourself at the boundary wall at ten past twelve on a clear Highveld day. The inverter is at full tilt, nothing much is running inside, and the meter is spinning in the direction you paid a lot of money to make it spin in. That's the cheapest hour of the grid's day, and it's the hour you're selling into.
Now come back at twenty past six. Geyser element on, kettle on, stove on, someone in the shower. That's the most expensive hour of your household's day, and every one of those units is imported.
If you're on a time-of-use tariff, that spread is written into your rates. Even on a flat tariff, the export is happening in a window you don't control the price of, while the import is happening in a window where you're fully exposed.
This is the part that most sizing conversations skip. A geyser timer that heats water between ten and two converts an exported unit into an avoided evening import. That one change is often worth more than a row of extra panels, and it costs a fraction as much. A pool pump moved to the middle of the day does the same thing. So does the dishwasher, if your household can be persuaded. You are not generating more; you are relocating the value of what you already generate.
Batteries follow the same logic. They aren't there to store cheap midday units so you can sell them cheaply later. They're there to carry midday production into the expensive evening window. Whether that pays depends entirely on the spread on your tariff and on how many cycles a year the chemistry will actually deliver — which is a different article, but it's the same principle.
The credit lands on the account, not in your pocket
Here's the thing I had to learn the slow way. A feed-in credit is normally a credit against the municipal account. It isn't cash. Unless your municipality explicitly says otherwise, nobody is transferring money into your bank account, and the credit usually can't be banked indefinitely either.
So ask three questions of your own account before you design anything around export:
- Does the credit only offset the charges on that month's account, or does it carry forward?
- Is there a fixed monthly SSEG charge or a separate meter fee that appears whether you export one unit or a thousand?
- What happens in the month your array out-produces what the account can absorb?
That third one is where the arithmetic can go strange. If the credit resets and the fixed charge doesn't, a big array in a low-consumption month can cost you money to run. I've seen households with beautiful production numbers and an account that didn't move the way they expected.
Not every municipality lets you export at all
This is worth saying plainly, because a lot of quotes quietly assume it away. In some supply areas you're simply not permitted to push power back, and the installation has to be configured for zero export, with reverse-flow protection doing the enforcement. Elsewhere you can export, but only after an SSEG application is approved, the installation complies with NRS 097-2-1 where it applies, and the electrical work is signed off under SANS 10142-1 with a valid Certificate of Compliance.
Where export isn't on the table, the whole question collapses into self-consumption, and that's honestly not a bad place to be. You size against a load list instead of a credit line, and the system tends to come out smaller and cheaper and more predictable.
How I actually size the array
I work backwards, and I do it in this order.
First, the load list and its hours. What runs, when, and for how long. Not the total monthly consumption — the consumption that can realistically move into daylight.
Second, the inverter's AC limit. Panels beyond what the inverter can push through the terminals don't earn anything at midday; they just sit there producing into a bottleneck, or get deliberately clipped. DC oversizing has real benefits in winter and in cloudy weather, but it doesn't create extra export revenue out of nothing.
Third, the battery, sized against the evening peak it has to cover rather than the total evening consumption.
Fourth, and only fourth, the marginal panel. For that last group of modules, the question is: does this unit get consumed in the house, where it's worth the full retail rate, or exported, where it's worth the credit? If most of the marginal production lands on the export side, the marginal panel is being paid at the discounted rate, and you should be honest about whether that's worth the capital.
There's a divergence in the background too. Retail tariffs and feed-in credits don't move together — one tracks the cost of supplying you, the other tracks policy decisions that can change with a council resolution. I don't assume they'll stay in step, and I don't design as though they will. When they do drift apart, the drift always favours the self-consumption side of the meter.
The honest limitations
None of this means export is pointless. If your house is empty between eight and five, you have no pool, no timer-able geyser load, and no appetite for a battery, then export is the only way your array earns anything at all, and a discounted credit beats zero. That's a perfectly reasonable design — it's just a different design from the one the salesman with the big array on the brochure is describing.
It also depends on your meter. Prepaid and conventional metering behave differently, municipal supply and direct Eskom supply come with different paperwork, and the tariff structure you're on right now may not be the one you're on in three years.
Which is why the envelope on my desk has the left column in pen and the right column in pencil. The retail rate I avoid is fairly durable. The credit is a moving target, and I'd rather build a system that works when it moves against me than one that only works while it doesn't.
