Commercial solar & storage glossary

Tariffs & billing

Demand charge

A demand charge bills the highest average power draw (in kVA) recorded over any single half-hour of the billing month, separately from the energy consumed.

Commercial tariffs in Johannesburg split the bill: an energy component in c/kWh and a demand component in R/kVA. The demand meter integrates over 30-minute windows — the integration period — and the month's highest window sets the charge. A two-second motor inrush won't do it; three compressors running together through one half-hour will.

The network charges this way because cables and transformers are sized for your worst half-hour, not your average. It also means demand is billed in kVA, apparent power, so a poor power factor inflates the charge even at constant kW.

On a commercial account the demand and network-capacity lines are rarely a rounding error against the energy line, and they behave completely differently: consumption is a running total you can chip away at, demand is a single worst moment. That asymmetry is why a bill can fall much less than expected after a solar installation.

Solar clips demand only when the sun happens to be out during your peak window. If your peak lands at 07:00 on a winter morning, panels do nothing for it — that job belongs to batteries or load control.

See also: kVA, kW and power factor · Peak shaving · Notified Maximum Demand (NMD) · Time-of-Use (TOU) tariff

Tariffs & billing

kVA, kW and power factor

kW is real power doing work, kVA is apparent power the network must carry, and power factor is the ratio between them — demand charges are billed on kVA, so a low power factor means paying for power that does no work.

Motors, drives and fluorescent ballasts draw reactive current on top of the working current. The wires heat up for both, so the utility bills capacity in kVA. Power factor = kW ÷ kVA; unity is perfect, and industrial sites often sit between 0.80 and 0.95 uncorrected.

Power factor correction — capacitor banks, or the reactive capability built into many modern PV inverters — narrows the gap between kW and kVA. On a Johannesburg demand tariff it is frequently the fastest payback item on the whole bill, and it is routinely checked in the same audit that sizes a solar system.

Two billing mechanisms make it bite harder than the arithmetic alone suggests. Several tariffs levy a reactive energy charge on top of the demand charge, and some apply an explicit power-factor penalty once the measured figure falls below a stated floor. Check which applies to the specific supply before assuming correction is optional.

See also: Demand charge · Notified Maximum Demand (NMD)

Tariffs & billing

Notified Maximum Demand (NMD)

NMD is the maximum demand a customer notifies in writing and the supply authority accepts, per point of supply; it fixes the network capacity reserved for the connection and the basis of several fixed charges.

It is a contract, governed by the NMD rules approved by NERSA, not an informal ceiling. Exceed it and an excess network capacity charge applies to the portion above the notified figure; repeated exceedances can force a formal supply upgrade — new transformer, new cabling, long lead times. Sit far under it and you pay reserved-capacity charges on kVA you never use, which is common on sites that added efficiency measures but never revised the supply agreement.

Batteries and load control can hold measured demand under the NMD, which sometimes defers a connection upgrade entirely. On a constrained site that avoided upgrade is often worth more than the energy the battery ever shifts.

The same rules define the mirror-image figure on the generation side — Maximum Export Capacity — so a site with solar has a contracted number in each direction.

See also: Demand charge · Peak shaving · Maximum Export Capacity (MEC)

Tariffs & billing

Maximum Export Capacity (MEC)

MEC is the export-side twin of the NMD: the maximum capacity a customer notifies and the supply authority accepts for sending energy from an on-site generator into the network.

It is set by the same NERSA-approved NMD and MEC rules that govern the import side, and it is measured per point of supply. Where a connection is approved with an MEC of zero, the plant must hold everything behind the meter — which is what an export limiter enforces in practice.

MEC is a network constraint, not a commercial preference: the authority is protecting a feeder shared with other customers, and the limit reflects how much embedded generation that feeder can absorb. Two identical roofs on different feeders can be granted very different figures.

It matters at design stage rather than after. An array sized on roof area, then met with a low MEC at application, has to be redesigned around self-consumption or curtailed for its life — which is why the export position is confirmed before the layout is fixed, not after.

See also: Notified Maximum Demand (NMD) · Feed-in, net metering and net billing · Curtailment · SSEG (Small-Scale Embedded Generation)

Tariffs & billing

Time-of-Use (TOU) tariff

A Time-of-Use tariff prices electricity by time of day and season — on Megaflex-type structures a winter peak kWh can cost five or six times an off-peak kWh.

Eskom's Megaflex and the City Power business TOU tariffs that mirror it define peak, standard and off-peak periods, with weekday peaks in the early morning and evening, and a high-demand season (June to August) where peak rates step up hard. Exact windows and rates are in the current tariff booklet; the shape has been stable for years even as the numbers climb.

For solar, TOU means the value of a kWh depends on when it is displaced. Daytime generation mostly lands in standard periods — useful, but the brutal winter-morning peak sits before the sun is up. That mismatch, not total production, decides much of the business case, and it is why interval data has to precede sizing.

The calendar matters as much as the clock. Saturdays carry fewer peak hours than weekdays, and Sundays and public holidays are typically billed at the off-peak rate all day — so a seven-day operation and a five-day one face very different bills on identical consumption.

TOU is also what pays for storage arbitrage: charge off-peak, discharge into peak, and the spread less losses is the margin.

See also: Blended tariff · Demand charge · Load profile (interval data) · Round-trip efficiency

Tariffs & billing

Blended tariff

The blended tariff is the total electricity bill divided by total kWh consumed — the real average cost per unit once energy, demand, network and fixed charges are all counted.

It is the only honest basis for a savings model. Model against the winter peak rate and the payback looks spectacular; model against off-peak and solar looks pointless. Both are wrong. The blended rate is what the site actually pays per unit, and displaced consumption should be valued at (or below) it.

There is a subtlety worth carrying, though, because it cuts the other way. The blended rate is an average across every line on the account, including demand, network and fixed charges that solar does not reduce. Solar displaces energy, so on a demand-heavy account the blended rate can overstate what a displaced kWh is really worth. The rigorous question is what the next kWh avoided actually costs — which is why the blended rate is the ceiling for a savings model, not automatically the right number inside it.

This is the first thing to check in a competing proposal. If the savings line quietly uses the peak energy rate, the projected payback will not survive contact with the first year's bills.

See also: Time-of-Use (TOU) tariff · Demand charge

Tariffs & billing

Feed-in, net metering and net billing

Feed-in is exporting surplus solar to the grid; net metering offsets exports against imports kWh-for-kWh, while net billing — the norm where it exists in South Africa — credits exports at a much lower rate than the retail tariff.

Net billing is the direction of travel: it is the model NERSA's framework adopts, treating import and export as two separate transactions rather than one net meter reading, and most municipalities are implementing along those lines while the rules are finalised. It was chosen over net metering deliberately — kWh-for-kWh offsetting would hand the customer the full retail rate for exports, including the network costs the distributor still has to recover.

The gap matters. An imported kWh might cost R2.40 blended; the export credit, where a municipality offers one at all, is typically closer to wholesale. Some supply areas offer no credit, and some approve connections only with export limitation — the inverter is configured to hold export at zero, as provided for in NRS 097-2-1.

Confirm the export position and the approved Maximum Export Capacity with the supply authority before sizing, because it changes the answer. Under zero-export rules, capacity beyond daytime load produces energy with no home; under a decent credit, modest oversizing can still wash its face.

See also: Self-consumption ratio · SSEG (Small-Scale Embedded Generation) · Curtailment · Maximum Export Capacity (MEC)

System design

kWp vs kWh

kWp is a system's rated output under Standard Test Conditions; kWh is energy actually delivered over time — the rating sizes the plant, the energy pays for it.

The p is 'peak': panel output at STC — 1,000 W/m² irradiance, 25 °C cell temperature, AM1.5 spectrum. Real roofs see less irradiance most of the day and cells run far hotter than 25 °C, so a 500 kWp array almost never delivers 500 kW.

Datasheets usually carry a second rating alongside STC — NOCT or NMOT, measured at 800 W/m² and a cell temperature in the forties with light wind. It typically reads 10–15% below the STC figure, and it is the closer of the two to what a Highveld roof actually sees at midday. Comparing one supplier's STC number against another's NOCT number is a common way to be misled by a quote.

What it delivers over a year is the size times the site's specific yield. On well-oriented Johannesburg rooftops that figure commonly lands around 1,600–1,750 kWh per kWp — the Highveld's irradiance is generous — with orientation, tilt, shading and soiling deciding where in the range a given roof sits.

See also: Specific yield · Performance Ratio (PR) · P50 / P90 yield estimates

System design

Load profile (interval data)

A load profile is the half-hourly record of a site's demand from the utility's metering — it shows when energy is used, which monthly bills cannot.

Two factories with identical monthly bills can need completely different systems: one runs flat through daylight, the other spikes at 06:00 and again at 18:00. The first is a solar site; the second is a batteries-and-scheduling conversation.

The data already exists — City Power and Eskom large-power customers are billed off AMR meters logging every 30 minutes, and the profile can be requested. A feasibility study that starts from bills instead of interval data is guessing at the two things that drive the return: self-consumption and the demand profile.

See also: Self-consumption ratio · Time-of-Use (TOU) tariff · Demand charge

System design

Inverter clipping and DC:AC ratio

Clipping is what happens when the array's DC output exceeds the inverter's AC rating: the inverter caps at nameplate and the excess is never generated.

Commercial systems are deliberately built with more panel than inverter — DC:AC ratios around 1.2 to 1.3 are routine on commercial rooftops — because panels spend most of their lives below STC output. Accepting a sliver of clipping across a handful of clear midday hours buys better inverter utilisation through mornings, afternoons and winter.

The trade is quantifiable, which is what makes it a design decision rather than a preference. Oversizing of that order typically lifts annual production by roughly a tenth while clipping trims a low single-digit percentage off the midday peaks — so the ratio is chosen to keep clipping small and financially insignificant, not to eliminate it. A design with zero clipping has usually bought too much inverter.

Where the connection caps AC export, the logic inverts and gets stronger: if the Maximum Export Capacity fixes what may leave the site, adding DC is the only remaining way to raise total production, and clipping becomes the accepted price of filling the shoulders of the day.

Past a point the trade stops working: panels whose output is routinely thrown away are capital doing nothing. The right ratio falls out of the irradiance profile and the load profile together, not a rule of thumb.

See also: kWp vs kWh · Curtailment · String and MPPT

System design

String and MPPT

A string is a set of panels wired in series to an inverter input; the MPPT is the tracker that holds each string at its maximum power point as conditions change.

Series wiring stacks voltage, and string design is an engineering task constrained at both ends. Open-circuit voltage rises as cells get colder — the cold-corrected figure can run 10–25% above the datasheet's STC value — and it must stay under the inverter's maximum DC input on the coldest expected morning, or the inverter is damaged rather than merely tripped. At the other extreme, operating voltage falls as cells heat, and it must stay above the MPPT's minimum on the hottest afternoon or the tracker drops out and takes the string's output with it. Get either end wrong and the system either fails in June or quietly leaves energy on the table all summer.

Panels in a string carry the same current, so shade on two modules drags the whole string — bypass diodes limit the damage but don't remove it. On commercial roofs with plant rooms, parapets and multiple pitches, splitting orientations across separate MPPTs is what keeps one bad corner from taxing the rest.

String-level monitoring is the diagnostic payoff: a failed string is obvious against its neighbours. Inverter-level totals can hide it for months.

See also: Inverter clipping and DC:AC ratio · Performance Ratio (PR) · Soiling

System design

Grid-tied, hybrid and off-grid

Grid-tied systems run in parallel with the utility and must shut down when it fails; hybrids add batteries and can island; off-grid systems stand alone — so a plain grid-tied system does not ride through load shedding.

The shutdown is not a design shortcut. Anti-islanding protection under NRS 097-2-1 requires the inverter to disconnect on loss of grid — line crews cannot work on a network being back-fed by rooftop plants. No grid, no output, panels or not.

Riding through an outage takes a hybrid: batteries, an inverter that can form its own grid, and a changeover that separates the backed-up circuits. That resilience layer costs real money over a plain grid-tied system, which is why 'do we need to run through load shedding, and which loads exactly' is a scoping question, not an afterthought. Installations need a Certificate of Compliance under SANS 10142-1 either way.

See also: SSEG (Small-Scale Embedded Generation) · Peak shaving · Feed-in, net metering and net billing

Performance

Specific yield

Specific yield is annual energy per unit of installed capacity, in kWh per kWp — the number that lets a 100 kWp roof and a 700 kWp roof be compared on equal terms.

It rolls the site's whole story into one figure: irradiance, azimuth and tilt, shading, cell temperature, wiring and inverter losses, soiling. Total production can't tell you whether a site is healthy, because bigger systems always produce more; yield can.

South Africa sits at the generous end of the global range — a national average in the region of 1,680 kWh/kWp/year, with well-oriented Johannesburg rooftops around 1,700. Published figures for the city assume roughly a 22° tilt facing due north; a flat roof ballasted at 10° or an east–west layout trades some annual yield for a broader generation curve, which can be the better commercial answer even though the yield figure reads lower.

Comparing measured yield against the design estimate, month by month, is the plainest health check a system owner has.

See also: kWp vs kWh · Performance Ratio (PR) · Capacity factor · P50 / P90 yield estimates

Performance

Performance Ratio (PR)

Performance Ratio divides actual output by the output the measured sunlight should have produced, giving a weather-independent health figure — typically in the high 70s to mid 80s percent for a sound commercial system.

PR answers the question production alone can't: is this system underperforming, or was it just overcast? A pyranometer or satellite dataset gives plane-of-array irradiance; PR is metered kWh over what that irradiance times the array's kWp should theoretically deliver. Cloud cancels out of the ratio. It is a standardised measure, not a vendor invention — IEC 61724-1 defines how it is calculated and what the measurement chain must satisfy, which is what makes one plant's PR comparable to another's.

The residual reflects real losses — cell temperature (crystalline modules lose output at roughly 0.3–0.4% per °C above 25 °C, which is why PR dips in a Highveld summer), soiling, downtime, wiring. A stable PR through seasonal production swings is the signature of a healthy plant.

The signal to act on is drift. A PR sliding over months means soiling, a dead string, an inverter derating or new shading. Production can be rising while PR falls; winter-to-summer gains hide faults.

See also: Specific yield · Soiling · Degradation · O&M (Operations and Maintenance)

Performance

Self-consumption ratio

Self-consumption is the share of solar output used on site rather than exported, and it usually moves the financial return more than total generation does.

Every self-consumed kWh saves the blended tariff. An exported kWh earns the export credit — often small in Gauteng, sometimes zero. The ratio between those two values is why matching system size to the daytime load matters more than maximising the array.

Don't confuse it with self-sufficiency, which is the mirror question: self-consumption is the share of your generation that you use, self-sufficiency is the share of your consumption that solar covers. A modest array on a heavy industrial load scores near 100% self-consumption and single-digit self-sufficiency. Proposals sometimes quote whichever of the two flatters the design.

The commercial advantage here is structural. A weekday operation runs its load straight through the generation curve, and daytime commercial sites commonly self-consume most of what they make — a far better match than a typical household, which is empty when the sun is highest.

Raising self-consumption is often cheaper than adding hardware: shift compressors, chillers or batching into daylight, or add storage to soak up the midday surplus for the evening.

See also: Load profile (interval data) · Feed-in, net metering and net billing · Curtailment

Performance

Curtailment

Curtailment is output deliberately held below what the sun would allow — most often by a zero-export limiter when generation exceeds on-site load.

On a zero-export connection, or one held to a low Maximum Export Capacity, the inverter throttles the moment generation passes what may leave the site; a quiet December fortnight can show hours of it. That is the control system doing its job, not a fault.

The distinction matters when reading monitoring data: curtailed output looks like underperformance until you check the limiter flags. Persistent heavy curtailment is a different message — the array outgrew the load, and the surplus wants either a battery or a rethink of what runs during the day.

See also: Self-consumption ratio · Feed-in, net metering and net billing · Inverter clipping and DC:AC ratio

Performance

Soiling

Soiling is dust, bird droppings, pollen and industrial fallout on the glass, blocking light until rain or a cleaning crew removes it.

The Highveld's dry season does the damage: little rain from May to September, winter dust, and losses that build week on week precisely when TOU rates are at their highest. Sites near mines, unpaved roads or agricultural activity accumulate faster still, and bird droppings behave like hard shading on individual cells rather than a gentle haze.

Scale matters for budgeting. Studies put typical annual soiling losses in the low-to-mid single-digit percentages across most climates, rising sharply into the double digits on genuinely dusty or industrial sites — which is the difference between a nuisance and a line item.

Monitoring shows the classic sawtooth — output ratio drifting down, then snapping back after the first good rain. Soiling is the one loss that is fully recoverable for the price of a cleaning contract, so cleaning frequency is an economic decision: compare the energy recovered against the cost of the crew, per season, per site. Two to four cleans a year is the common cadence for a commercial array, weighted toward the dry months, with dusty sites justifying more. Budget honestly for rooftop access, too — cleaning a roof costs several times cleaning a ground-mount, and that gap is what decides the frequency.

See also: Performance Ratio (PR) · O&M (Operations and Maintenance) · Degradation

Performance

Degradation

Degradation is the slow, permanent loss of panel output with age — linear performance warranties typically guarantee somewhere between 80% and 92% of nameplate at year 25, depending on the module tier, which implies a guaranteed loss of roughly 0.3–0.5% a year after the first.

The warranted rate and the measured rate are different things. Long-run field studies put median crystalline degradation near half a percent a year, and the newer cell architectures do better than that — the spread between a commodity module and a premium one is real, and it compounds over 25 years into a meaningful energy difference. The warranty is the floor the manufacturer will defend, not the expectation.

Three mechanisms sit behind it. Light-induced degradation (LID) is a one-off drop of a percent or two in the first days of exposure, and it is specific to traditional boron-doped p-type cells — n-type architectures are essentially immune, which is part of what the price difference buys. LeTID is a slower light- and temperature-driven cousin. Potential-induced degradation (PID) is the one that can be severe: stack voltage and humidity conspire to leak current through the module, and it is controlled by module design and correct earthing rather than by accepting it.

Unlike soiling, none of it comes back with a wash. A 25-year cash flow built on year-one output overstates lifetime energy by several percent compounded; any honest yield model, LCOE or IRR carries the degradation curve in it.

See also: Performance Ratio (PR) · Soiling · LCOE (Levelised Cost of Energy)

Performance

Capacity factor

Capacity factor is average output over a period divided by nameplate running flat out — fixed rooftop PV in South Africa typically lands around the high teens to low twenties percent.

Nights and sun angle cap it: the resource simply isn't there 24 hours a day, which is a property of solar, not a defect of a plant. The figure earns its keep when comparing generation technologies or regions on one axis.

Compare like with like, though, because this is where the number is most often misused. Published South African solar capacity factors in the mid-to-high twenties describe utility plant — single-axis tracking, on high-irradiance Northern Cape sites. A fixed rooftop array in Gauteng lands closer to the high teens or low twenties, and that gap is tracking and location, not a worse system. Quoting the utility figure for a rooftop proposal overstates it by a third.

For judging a specific installation, PR and specific yield carry more information — capacity factor moves with the weather and says nothing about whether the plant extracted what the weather offered.

See also: Specific yield · Performance Ratio (PR)

Performance

P50 / P90 yield estimates

P50 and P90 are probability levels on a modelled yield: production should beat the P50 in half of all years and beat the P90 in nine years out of ten.

Read the phrasing precisely: P90 is not 'we are 90% confident the plant makes X'. It is 'in any given year there is a 90% chance the plant makes at least X'. The two sound alike and mean different things.

Simulation tools like PVsyst build the estimate from long-term satellite irradiance, then a combined uncertainty sets the distribution. That uncertainty is not just year-to-year weather variability — it also carries the error in the irradiance dataset itself and in the simulation model, combined as a root sum of squares. P90 then falls out arithmetically: roughly the P50 less 1.28 standard deviations. P50 is the central expectation; P90 is the conservative case lenders and PPA underwriters size debt against, because debt service has to survive a poor year.

The gap between a proposal's P50 and P90 is itself informative — a wide spread means either a volatile resource or thin data, and it is a direct read on how much the modeller trusts their own inputs. A proposal quoting one 'estimated production' figure with no probability attached is telling you less than it appears to.

See also: Specific yield · IRR (Internal Rate of Return) · LCOE (Levelised Cost of Energy)

Performance

O&M (Operations and Maintenance)

O&M is the ongoing work that keeps a plant at its designed output — monitoring, cleaning, thermographic and electrical inspection, fault response, and the component replacements a 25-year life implies.

No moving parts breeds complacency. Output erodes quietly — soiling, a tripped string, an inverter derating on a cooked plant room — and without monitoring the first notice is a bigger bill. The expensive failures are the silent ones: a single string down on an unmonitored roof burns more energy in a season than monitoring costs in years.

Annual thermographic scans catch hot connections and failing bypass diodes before they become burns; IV-curve tracing localises module faults; and the financial model should carry at least one inverter replacement across the asset life, because power electronics age faster than glass. The economics of the inspection are asymmetric in the owner's favour — a hot spot found on schedule is a module swap, while the same fault left for a year or two takes the string, the junction box and sometimes the roof with it.

A serious contract also reports monthly against a stated performance guarantee rather than simply confirming the plant is on. Reporting that cannot be checked against a number is not a contract term, it is a newsletter.

See also: Performance Ratio (PR) · Soiling · Degradation · LCOE (Levelised Cost of Energy)

Storage

Peak shaving

Peak shaving discharges a battery into short demand peaks so the metered half-hour maximum stays down, cutting the kVA on which the demand charge is billed.

It attacks the demand line, not the energy line, and the battery is sized accordingly: enough power (kW) to cover the height of the excursion, enough energy (kWh) to hold it through the half-hour integration windows in which it occurs. A 15-minute morning spike is a modest battery; a three-hour evening plateau is not peak shaving at all, it's load shifting.

Everything follows from the interval data — height, duration and frequency of the peaks. Where demand and NMD-related charges are a fat slice of the bill, this application can carry the battery's business case on its own, before any TOU arbitrage is counted.

See also: Demand charge · Notified Maximum Demand (NMD) · Depth of Discharge (DoD)

Storage

Depth of Discharge (DoD)

Depth of Discharge is how far into a battery's capacity a cycle goes; usable energy is nameplate capacity times the allowable DoD, and cycling deeper than specified shortens life.

Most commercial systems now run LiFePO₄ (LFP) cells, commonly rated around 90% DoD at 6,000 cycles or more — which is what makes daily cycling over a ten-year-plus horizon a reasonable plan rather than an optimistic one. The nameplate number on the datasheet is not the energy you plan around; the usable figure is.

It's a standard place for proposals to flatter themselves: quoting nameplate against a requirement stated in usable kWh describes a smaller battery than the buyer thinks they're getting. State the requirement in usable energy and let the supplier size the nameplate.

See also: Peak shaving · Round-trip efficiency · C-rate

Storage

Round-trip efficiency

Round-trip efficiency is energy out over energy in for a full charge–discharge cycle; for LFP systems including power conversion it commonly sits between 85% and 92%.

Losses live in the cells and in the power conversion system on both legs of the trip, so insist on a system-level figure — cell-only efficiency quietly ignores the inverter, and LFP cells alone will happily quote above 90%. The loss is a direct toll on any arbitrage strategy: it comes off the TOU spread before margin.

It is also not a fixed property of the battery. Pushing energy in or out harder raises internal losses as heat, so the same system measured at a hard peak-shaving duty returns less than it does on a gentle overnight charge. A quoted efficiency is only meaningful next to the C-rate it was measured at.

On a Megaflex-shaped winter tariff the spread is wide enough that the toll hardly matters; in summer, with peaks a fraction of winter's, the same battery earns far less per cycle. Storage economics in Gauteng are seasonal, and a model that annualises a winter spread flatters itself.

See also: Depth of Discharge (DoD) · Time-of-Use (TOU) tariff · Peak shaving

Storage

C-rate

C-rate is charge or discharge power relative to capacity: 1C empties the battery in an hour, 0.5C in two — it is the link between the kW a duty needs and the kWh the battery holds.

A 200 kWh battery at 1C delivers 200 kW for an hour; at 0.5C, 100 kW for two. The duty sets the rate: roughly 0.5C is the usual specification for two-hour energy shifting, and around 1C for one-hour duties. Short, hard peak-shaving excursions push higher still, and that becomes a selection criterion rather than a footnote.

Higher rates cost efficiency as well as money — more current through the same internal resistance means more heat and a lower round trip — so the rate is chosen, not maximised.

A specification that states kWh without the C-rate has defined half the machine. The duty cycle — how hard, how long, how often — is what the warranty and the thermal design are actually rated against.

See also: Peak shaving · Depth of Discharge (DoD) · Round-trip efficiency

Finance

PPA (Power Purchase Agreement)

Under a PPA a provider funds, owns and operates the solar plant on your site, and you buy its output at a contracted rate per kWh — typically below the utility tariff, escalating annually, over a 10-to-25-year term.

You commit no capital and carry no performance risk; the provider only earns on delivered energy. The trade is ownership economics: the provider's cost of capital and margin live inside the rate, so across the full term a cash purchase nearly always costs less in total. A PPA buys risk transfer and an untouched balance sheet, not the cheapest lifetime energy.

Read five clauses before comparing rates. The escalation: a PPA escalating at 6% against double-digit utility increases widens the saving every year, and the reverse arithmetic also exists — a low opening rate with an aggressive escalator is a common shape. Any take-or-pay minimum offtake. The end-of-term mechanics: transfer, renewal or removal. The buyout schedule, which typically opens a few years into the term and is where the real cost of exiting lives. And the clause almost nobody reads until it matters — what happens if the property is sold, since the agreement is normally ceded to the incoming owner or the plant removed, and either outcome is a live issue in the sale negotiation.

See also: Blended tariff · IRR (Internal Rate of Return) · Payback period

Finance

Payback period

Payback is the time for cumulative savings to repay the capital cost; simple payback divides capex by year-one savings and ignores everything that happens afterwards.

It communicates well and boards ask for it, so quote it — but know its blind spots. Tariff escalation makes later years worth more than year one, which simple payback ignores in your favour; degradation and O&M lean the other way. And it says nothing about the fifteen-plus years of returns after the crossover, which is where most of the asset's value sits.

Discounted payback is the stricter sibling: it asks when cumulative savings repay the capital once future rands are discounted back to today's money, and it carries tariff escalation and module degradation rather than assuming year one repeats. It always reads longer than simple payback, and the gap between the two is a fair measure of how much a proposal is leaning on the simpler method.

When comparing proposals, force the assumptions into the open: simple or discounted, what escalation, valued against blended or peak rates. Two honest engineers can quote paybacks a year apart off the same roof purely on assumptions.

See also: IRR (Internal Rate of Return) · LCOE (Levelised Cost of Energy) · PPA (Power Purchase Agreement)

Finance

IRR (Internal Rate of Return)

IRR is the annualised return implied by a project's full cash-flow profile — the discount rate at which its net present value is zero — and it is the figure that lets solar compete with other uses of the same capital.

Payback ranks projects by speed; IRR ranks them by return, over the whole life, with the timing of every cash flow counted — escalating tariffs, degradation, O&M, inverter replacement, residual value. That is the comparison a CFO actually needs against expansion capex or debt reduction.

In South Africa the after-tax picture matters: the Section 12B capital allowance accelerates depreciation on renewable generation assets and can lift after-tax IRR well above the pre-tax figure. The rules and rates shift — confirm current treatment with a tax adviser rather than a proposal PDF.

And interrogate the inputs before comparing two IRRs. A percentage built on 14% escalation, zero degradation and free maintenance is not the same animal as one built honestly; the assumptions are the number.

IRR has one structural blind spot: being a percentage, it says nothing about size. A small project can out-IRR a large one while creating a fraction of the value, which is why it is read next to NPV rather than instead of it.

See also: Payback period · NPV (Net Present Value) · LCOE (Levelised Cost of Energy) · PPA (Power Purchase Agreement) · P50 / P90 yield estimates

Finance

NPV (Net Present Value)

NPV is the value a project adds in today's money: every future cash flow discounted back at the investor's required rate of return, less the capital. Positive means the project clears the hurdle; negative means it does not.

It is the figure the other three metrics orbit. Payback says how fast, IRR says at what rate, LCOE says at what cost per kWh — NPV says how much richer the business ends up, which is what a capital allocation decision actually turns on.

The discount rate is the whole argument. Use the company's weighted cost of capital and NPV answers 'is this better than our other uses of money?'. Use a rate plucked to make the answer come out right and it answers nothing. Ask any proposal quoting an NPV what rate it discounted at; if the answer isn't immediate, the number isn't load-bearing.

For solar the arithmetic is unusually favourable to NPV as a lens, because the cash flows run for decades and escalate with the tariff. A project with a mediocre payback can carry a large NPV simply by continuing to save money long after the capital is repaid — which is exactly the value simple payback discards.

See also: IRR (Internal Rate of Return) · Payback period · LCOE (Levelised Cost of Energy)

Finance

LCOE (Levelised Cost of Energy)

LCOE is lifetime cost divided by lifetime energy — the all-in average cost per kWh a plant produces, comparable directly against a tariff or a PPA rate.

A meaningful LCOE carries everything: capital, financing, O&M, insurance, inverter replacement, and production net of degradation. Its virtue is the unit — R/kWh — which puts an owned asset, a PPA offer and the utility on one axis.

The detail that separates a real LCOE from a rough one is that both streams are discounted, not just the money. The formula divides the present value of lifetime costs by the present value of lifetime energy, discounting each year's kWh at the same rate as each year's rand. Discounting the costs but not the energy — the common shortcut — understates LCOE, because it weighs distant kWh as heavily as today's while shrinking the distant costs.

Its vice is how easily it is gamed. Capex divided by undegraded production is not an LCOE, and it will undercut every honest number in the room. Ask what's inside, and at what discount rate, before comparing.

See also: IRR (Internal Rate of Return) · NPV (Net Present Value) · Payback period · Degradation

Regulation

SSEG (Small-Scale Embedded Generation)

SSEG is generation connected on the customer's side of the utility meter; a grid-tied commercial system must be registered with its supply authority — City Power in most of Johannesburg, otherwise Eskom — before it may operate.

The application is engineering paperwork more than form-filling: a single-line diagram signed off by a professional, an inverter demonstrably compliant with the current edition of NRS 097-2-1 (the grid-interconnection standard covering protection settings and anti-islanding, revised in 2024) and appearing on the authority's approved list, a bi-directional meter in place of the existing one, and a Certificate of Compliance for the installation under SANS 10142-1, with SANS 60364 governing the installation design.

Budget calendar time, not just effort. Turnaround differs markedly between supply authorities and, in the larger metros, the metering step can stretch the process into months. It runs in parallel with procurement and construction, but only if it is started early — a plant standing complete and unregistered earns nothing.

Capacity thresholds, licensing exemptions and fee waivers have moved repeatedly in recent years and differ between supply authorities — several have run time-limited registration amnesties. Confirm the current rules and any open window for the specific connection rather than relying on last year's precedent.

Skipping registration is a false economy. An unregistered plant risks disconnection and complicates insurance, and it reliably surfaces in due diligence when the property is sold, refinanced or re-let.

See also: Grid-tied, hybrid and off-grid · Feed-in, net metering and net billing