Two Bloemfontein rooftops can have the same cost per watt and very different payback paths. One sends most midday surplus to the grid. The other runs the geyser and pool pump while the sun is high. The second pays itself off sooner, not because the panels are better, but because more kilowatt-hours avoid the high purchase tariff. The export credit is low, so every kilowatt-hour that leaves is worth less than one used inside. That split often moves payback more than adding another module.
The Filters a Payback Calculation Must Pass
Before trusting any quote, four filters should be applied. Payback numbers hide assumptions.
- Separate used energy from exported energy.
- Value avoided purchase at the top block still being bought.
- Treat self-consumption as a range.
- Let the result move when the export credit moves.
If a filter fails, the calculation is telling a story about generation, not money.
Two ways of reading a quote matter here. The total-generation way values every kilowatt-hour near the purchase tariff. The self-use way splits the flows, prices them differently, and tests the timing.
Does It Split Used Energy from Exported Energy?
The total-generation way fails. It treats all generation as if the house consumed it. On a clear Bloemfontein weekday, the system exports strongly while nobody is home, and the evening grid draw stays nearly the same. The inverter display shows generation; the bill does not move by the same amount.
The self-use way passes because it has two buckets. Once export is its own bucket, a larger array no longer automatically means faster payback. An array sized to yearly consumption may simply push more midday energy into a low-value export stream.
Total-generation is eliminated.
Does the Tariff Reflect the Block the House Would Have Bought?
Most total-generation calculations fail here too. They use an average tariff. Municipal block tariffs are not flat. A self-consumed kilowatt-hour saves the highest block still being purchased, not the average. Export is priced separately.
The self-use way passes only when it tracks the marginal block. A midday pool pump alone does little if the geyser is still heating in the evening and the household remains in the same top block. Move both loads into midday, and the marginal avoided energy becomes visible. That timing change affects payback without changing the system size.
Any method using a single average tariff is out.
Is the Self-Consumption Percentage Treated as a Range?
Even the self-use way can fail here. A fixed 35% or 60% hides the real uncertainty. The same system can move between the two depending on when appliances run. Two arrays with similar cost per watt can invert their payback order once self-use differs by more than twenty percentage points.
Batteries can lift self-consumption, but they add capital cost and round-trip losses. They do not automatically improve payback. The model should show what happens when self-consumption shifts by 10 or 20 percentage points before it ranks anything.
A fixed self-consumption assumption fails.
Does Payback Move When the Export Credit Moves?
A split calculation can still be static. It must be stressed against different credit levels. Export credits are usually below the avoided purchase tariff. When that gap is wide, payback is highly sensitive to self-consumption. When the credit sits closer to the purchase tariff, exported energy matters more, but that is not the usual South African starting point.
The self-use way passes if it recalculates at two or three credit ratios. This is not about knowing today's municipal credit rate. It is about seeing whether the result depends on the export rate. If it does, panel size becomes secondary.
After this, only the self-use way remains.
The Reading That Survives
The self-use reading survives. In its strongest form, it values avoided purchases at the marginal block, separates export, uses a self-consumption band, and checks the credit gap. It avoids the trap of a single payback year based on total generation.
When an Export Split Cannot Be Measured
Some inverter monitoring shows total production but not how much left the property. The self-use method can then fail the first filter. The honest adjustment is to relax the demand for an exact split, not the split itself. Model 30%, 40%, and 50% self-consumption and present payback as a range. The cost is a band instead of a single year. That band is more useful than pretending the system never exports.
What Survives the Gauntlet
The self-use method survives. The split test removes more options than the export-credit test because exporting energy must be priced separately before any comparison works. In many homes, moving the pool pump and geyser into the solar window changes payback more than adding panels would.
