Firstly, rolling blackouts cut electricity use and that reduces municipal revenue from electricity. Secondly, as more and more of those who can afford to do so install solar power, municipalities lose income from those households and businesses.

Leave aside the fact that many municipalities don’t bother to pay Eskom and assume (a stretch) that municipalities are trying to be rational actors. What are they to do?

Eskom divides the day into three bands — off-peak, normal and peak — and charges municipalities differentially in each band partly to offset the high cost of peak generation and partly to discourage peak loading. Timing varies seasonally but peak time is anywhere from 7am–10am and 6pm–8pm to 6am–9pm and 5pm to 7pm, so generally peak time is about 5 hours per day.

Having perused a municipality’s Eskom electricity invoice, it is clear that peak charges are a big cost: 46% more than normal time and more than double off-peak charges. 

Why?

Eskom (when things are working correctly) has to deploy smaller-scale power plants to handle peak load and they are far more expensive per kWh than bigger plants.

Electricity charges are based on a cost x kilowatts x hours, or kilowatt-hours (kWh). To reduce large numbers, divide by 1,000 to get megawatt-hours (MWh).

Municipalities charge smaller consumers a standard rate; larger consumers are charged based on their maximum usage throughout the day. If you are a big fast food vendor with a flurry of take-home purchases during late afternoon peak, your highest-demand time will determine the rate you are charged all the time, an incentive to shift demand away from peak time or use other forms of energy (e.g., gas).

Since the municipality charges a fixed rate per kWh to all but the biggest users, they must absorb peak charges into the kWh rate they charge. Where solar power does not help them is that the solar peak does not coincide with consumption peaks: in the morning as households prepare breakfast and use hot water, and early evening when dinner is prepared.

As more households and businesses add solar power to cover rolling blackouts and to reduce electricity bills, this gets worse for a municipality.

Read more in Daily Maverick: Small Free State town in high court battle with Eskom over use of solar energy to reduce load shedding 

If a municipality puts in its own solar power with batteries, it can cut into its Eskom costs by releasing power from the batteries to the grid during peak hours (this is called peak shaving). Ideally, it should have enough capacity to cover all five hours of peak but any reduction in peak demand makes a difference.

As an example, from an invoice I have from a small municipality, in March 2022 they were charged for 7,700MWh, about 250MWh per day. Of this, 1,200MWh was peak consumption or less than 40MWh per day, or 8MWh per hour (peak is 5 hours per day). 

Solar panels today are available in capacities of 500W or more; 1,000 such panels would at peak produce 0.5MW. A realistic average over 24 hours is a quarter of peak power, taking into account nighttime and cloud. So 1,000 solar panels each of 500W capacity would average 3MWh per day. This is a decent fraction of peak power demand, given that this is the most expensive part of the municipal Eskom invoice.

How much space would all these panels take? With an approximate figure of 2.5 square metres per panel, 1,000 panels would require 2,500 square metres, not a huge chunk of land for a standalone installation.

Another trick could be used to shave peak even more. 

Eskom charges a very low rate for off-peak time, about 60% of the normal rate. If battery capacity was big enough to cover the entire peak time, batteries could be charged during off-peak hours as well as during sunny hours.

Read more in Daily Maverik: How to beat load shedding at home… and other ideas

How much battery? 

For my example, assuming the batteries can be discharged by 80% without damage (some lithium-ion batteries can discharge even more), you would need to cover the longest peak stretch, 3 hours. That means you need to be able to supply 8MW per hour for 3 hours, a total of 24MWh. So batteries would need to be able to store 30MWh.  This is a serious amount of battery — one supplier provides this capacity in 15 containers.

Fortunately, this sort of capability can be built up in stages with a modular design. Each time more is added, the cost to the municipality reduces.

An add-on to such a system is backup power. An uninterruptible power supply (UPS) is usually designed for short outages; with power off two hours or more at a stretch, a regular UPS will not last either in terms of run time or battery life. A generator is a poor option as it has ongoing running costs that are far higher than regular Eskom charges. For offices, a municipality could use a UPS on anything that cannot afford even a brief power glitch and run everything else off battery plus solar during rolling blackouts. This may reduce the ability to shave peak charges, but would keep services running.

What about heavy infrastructure requirements? The biggest is likely to be pumping water. In the municipality from which I harvested Eskom figures, the highest combined pump draw if all are operating at once is about 2MW. This could be supported out of the batteries at the cost of reducing peak shaving.

The really big question is not whether this is a good idea but whether municipalities have the will to implement such a solution. 

Read more in Daily Maverick: Informal settlements could drive South Africa’s rooftop solar revolution

From what I know of the disastrous state of municipal service delivery and infrastructure, implementing such a project could easily turn into flushing money down the toilet.

Since so many municipalities do not pay Eskom anyway, the incentive to reduce the size of their Eskom invoice is not what it should be.

What we really need is a government implementation agency that is built from the start with integrity checks to ensure that it does what it is supposed to do. Some may be sceptical that this can be done. But it is worth trying. DM

Philip Machanick is an emeritus associate professor of Computer Science at Rhodes University and a long-time climate and social activist.