A solution to SA's power crisis - TechCentral

A solution to SA’s power crisis

Dirk-de-Vos-180Load shedding and spiralling electricity prices are now part of South African daily life. What, then, can be done about it?

If one were to take a long view, one could confidently say that our electricity problems boil down to one thing: South Africa has never been able to price electricity correctly. Eskom’s own doubtful sustainability on multiple fronts — operational, technical, financial and indeed environmental — all come down to this one fact.

Between 1991 to 2008, when Eskom entered into a pricing compact, South Africa saw a progressive reduction in the price of electricity. A 2014 graph from a paper entitled “Repositioning electricity planning at the core: an evaluation of South Africa’s Integrated Resource Plan”, written by the Trade & Industrial Policy Strategies for the National Economic Development and Labour Council (Nedlac),  shows this historical trend. It also shows the steep increases since 2008 which, as we have seen, are continuing:


If one goes back to the 1970’s and 1980’s, there is a repeating pattern. When Eskom builds capacity, it builds far too much capacity, often with untested technology, and electricity prices shoot up.
The real problem we inherited from those years is that excess capacity needed to be mopped up. This was done in several ways: by attract energy-intensive industry like aluminium smelting, but also by pricing electricity less than it costs to produce over the long term. This sort of behaviour wreaks havoc with any sensible energy policy making.

This is not to say that there have not been attempts to develop policy. The 1998 white paper on energy, which sought to liberalise the energy sector, was formally adopted by national government. A cabinet resolution in 2001 prescribed that 30% of Eskom’s generation capacity be sold to the private sector, and in 2004 cabinet decided that 30% of new generation should be built by the private sector.

To support these policies, a moratorium on Eskom building its own capacity was imposed, which was lifted in 2004. None of these initiatives came to anything. Several reasons, including regulatory risk and protracted procurement processes, are blamed, but the real problem was low electricity prices. You simply can’t attract investment into a sector where the product is sold for less than it costs to produce. One can’t expect politicians to drive policy, which, if implemented, results in immediate increases in prices. The result is a slow drift to the precipice, which is where we are now.

It is not just national government policy. Local government, responsible for up to 40% of electricity supplied to end users, has used significant mark-ups on electricity to fund all sort of other projects. In larger urban areas, electricity makes up as much as 40% of total revenues and, although hard to establish, net surpluses from the sale of electricity amount to between 15% and 20% of these revenues. The sale of electricity does not only cross-subsidise other municipal expenses, it it also uses net surpluses generated by high-consumption users through an inclined block tariff to subsidise free and low-consumption users. It is not as though local reticulation has been maintained, with estimates of maintenance and refurbishment backlogs sitting at R27bn and growing by R2,5bn/year — and this excludes debt to Eskom of over R10bn.

It can’t continue. Most municipalities use Eskom’s MegaFlex tariff. While the trend in Eskom’s average pricing is sharply upward, a more significant change is in the differentials between off-peak and peak pricing, which is closer to R240/kWh, a differential of 14 times. In a constrained supply environment, this differential is set to grow as it should and is closer to the position in other jurisdictions. In a more liberalised electricity market, we would have got to these differentials a long time ago.

Recent figures released by national treasury and published by Statistics South Africa show the importance of the sale of electricity, particularly in the context of the more urban municipalities or municipalities that supply large industrial customers. We are also aware that Eskom’s tariff trajectory will be steeply upwards for the next few years.

Importantly, it is less about the upwards movement of the average tariff, but more in the increasing differentials between off-peak and peak pricing. Just a few years ago, the differential between peak and off peak in Eskom’s MegaFlex tariff (the tariff applicable to most municipalities) was around eight times; the differential now is more than 15 times. In future, the differential is likely to increase to 20 times. Electricity is expensive to store and, in general terms, has to be consumed as it is produced. The marginal costs of supplying electricity during peak times are much higher than at other times.

These changes to peak pricing are more significant for municipalities that are more exposed to the residential sector than Eskom. Although residential use of electricity in South Africa represents 18% of all electricity demand, at peak times residential demand represents over 35% of total demand.  At present, large municipalities are already supplying electricity to residents during winter peak times at a loss. The peak period, five hours of the day in winter, accounts for half of the total cost of all electricity supplied on the day in question. A stylised representation for a municipality might be as follows:


The problem becomes more acute when “losses” during peak times are not fully recovered from the sale of electricity during off-Peak and standard times. Higher flat rate residential tariffs mean that South Africans are having to save electricity where they can. But they save during standard and off-peak times which is where the surpluses are generated. Moreover, residents and businesses paying the highest inclined block tariffs are looking to move to solar water geysers and embedded generation (largely installing PV panels). Yet, during the early mornings and evenings (peaks), these customers are still drawing electricity from the grid, which makes them increasingly less profitable. Municipalities are not able to fully recover their position due to the flat tariff imposed on these customers.

The obvious answer is to move to time-of-use metering and charging for consumption accordingly so that municipal customers are aligned with the supply position faced by their local authorities. Any serious demand side management that does not destroy the electricity supply revenue model depends on time-of-use charging.

However, the installation of smart meters, where this has been carried out, has not produced the desired results. The tenders for smart meters issued by Johannesburg and the Tshwane municipality have garnered publicity for all the wrong reasons.  This problem is widely recognised and is likely the reason for an RFI issued by the department of co-operative governance & traditional affairs dealing with standardising the specifications of a national standard for smart meters. But we should be cautious about this, too. When dealing with a big change, there is a tendency to over-engineer things. We want smart meters that do all sorts of additional things like remote management of electrical appliances behind the meter. That is a mistake.

The other mistake is to issue large tenders for the roll-out of vendor-specific smart meters based on a tender process. The stakes become too high, and undue influence and bid rigging becomes inevitable. Even apparently low pricing on installation of end-to-end solutions results in vendor lock-in so that, in perpetuity, the municipality concerned becomes de facto dependent on the selected vendor for continued service. Moreover, there is no guarantee of future-proofing the smart meters.

Perhaps, then, it is time to move away from “smart” metering and focus on the main issue: time-of-use metering. This makes meters much simpler. A simple meter, costing about R800, might have just the following:


Any number of local manufacturers could build a time-of-use meter like the one above.

The mains issue, though, is this: electricity and the billing has to remain under the control of the municipality itself. The larger metros have recently committed huge resources to updating their billing systems. To remain in control, they must also be the custodian of the telemetry platform that connects and communicates with all smart/time-of-use meters. Importantly, this telemetry platform will use a cohesive but loosely coupled architecture that ensures any component can be replaced with no impact to the end-to-end solution. Using bulk purchasing, municipalities ought to be the contracting party with the cellphone networks that provide the connectivity to all the devices in the field. Both these elements should not be contracted out to any proprietary vendor or solutions provider.

A schematic of the time-of-use metering environment might therefore look like the following:


The strategically important issue here is that no one entity owns the complete supply chain. However the municipality controls the following:

  • The protocol to communicate to the time of use meters fitted/deployed.
  • The protocol between the GSM engine in the meter and the  management platform.
  • The protocol from the management platform to the existing ERP applications.

The municipality must also own and control the access network (for security and liability), the on-boarding process for meter manufacturers, and provide a certification process to approve meters as well as Sim and access networks.

There are several other inherent advantages to the suggested approach:

  • It keeps the electricity departments in control of electricity supply. In particular, it maintains full control of billing functions.
  • There is a need to build capacity internally and not commence a lengthy and overly complicated tender process.
  • Simpler process reduces change management to a minimum by reducing complexity.
  • The municipality is able to build local skills and Internet-of-things capacity within electricity departments, among local manufacturers and electronics assemblers.
  • It allows municipalities to expand telemetry (Internet of things) services to other applications — traffic lights, remote asset monitoring without requiring a complete, standalone solution for each of those.
  • The municipality’s bulk buying power can be leveraged to achieve the best discounts from providers of services (and cost drivers) within the value chain such as connectivity providers (the cellphone networks).

In time, the municipality will be able to use and develop meter data management and analytics engines to help with pricing, measuring demand response initiatives and so on.

The proposed approach also allows for piloting. Instead of getting a vendor to install, say, 100 000 meters, a refinement of requirements can be done through discrete pilot projects, which limits risk and builds internal municipality institutional buy-in. The piloting of smaller implementations across different types of municipalities gives valuable lessons for optimising large-scale deployments. In all cases, the municipality can decide how to proceed, but to do so from a position of knowledge and control.

The biggest benefit, though, is that the roll-out of time-of-use metering would be much, much cheaper than the smart metering implementations that we have seen to date. Indeed, a programme such as the one suggested here pays itself off, reducing the need for earmarked capital budgets for smart metering. Being able to implement time-of-use pricing may increase profitability on the sale of electricity, while consumers, by changing consumption habits, may be able to reduce their electricity bills.

The aligning of incentives creates another virtuous pattern. Presently, several energy-saving initiatives, such hot water geyers and energy-saving light bulbs are subsidised, often at great cost. The benefits of these subsidies are enjoyed entirely by the consumer who installs them (who derive the benefit of the savings). Time-of-use metering (and pricing) changes the dynamic. The capital costs of energy-saving devices or even just change of behaviour becomes the responsibility of consumers — as it is with any other product or service.

Charging electricity at the time that it is used is inevitable. Developing the capacity and means to do so need not be a series of expensive mistakes. There is a much better way.

  • Dirk de Vos is an energy expert. He is a founder member of IoTa, an Internet-of-things (IoT) accelerator that focuses on developing capacity for projects using IoT technology. He can be contacted on ddevos@qedsolutions.co.za


  1. The same could be said for Net Metering and home generation. If we simplified it and went back to the roots, we could have a system that works for all.

    Instead of regulating tariffs and introducing the scourge of the taxman, we could do the following.

    Think of “the grid” in terms of a battery. If you generate excess power, it needs to be stored. If you consider that a batter is only 75-80 percent efficient, then you would be happy for “the grid” to take a percentage to cover costs. This ties in neatly with peak charging and the desires of the provider to generate revenue.

    For example, contribute a kilowatt hour of electricity during peak period and receive 90% back as credit in kilowatt hours. Contribute that same kilowatt during off peak and receive only 70% credit. Any shortfall once the credit runs out can be purchased in a prepaid manner.

    No money changes hands, no regulation is required and the provider can sell the difference (taxable). Furthermore, this benefits the smallest and largest contributor equally.

  2. William Stucke on

    > For example, contribute a kilowatt hour of electricity during peak period and receive 90% back as credit in kilowatt hours. Contribute that same kilowatt during off peak and receive only 70% credit.

    Not quite, Simon. If we take Dirk’s 20:1 factor between the cost of peak and off peak electricity, it should be:

    “Contribute a kilowatt hour of electricity during peak period and receive 90% back as credit in kilowatt hours. Contribute that same kilowatt during off peak and receive nothing.” (or 4.5%, if you insist)

    Excess 3rd party power supplied during off peak periods has no value. It can’t be stored. All it means is that base load generation capacity has to be shut down, which costs more money than running it. The capacity for the peak demand still has to exist, but can only be amortised over a shorter portion of the day. Ergo, base load generation costs more, the more off peak power demand is reduced.

  3. William Stucke on

    I’m interested to know how you see consumer behaviour being changed, Dirk. A flashing light and buzzer during peak periods? An invitation to disable it, I’d say.

  4. Elegant solution and that’s just why the ANC gangster regime intent on capturing that which is still beyond the rickety gates of Nkandla has NO intention of taking this expert advice in any way shape or form. Besides, the overwhelming majority of the members-of-the-community have not the slightest intention of having their power metered by anything other than that which is gleaned from them by the ‘technician’ who installed their IC in the first place.
    (Thanks for the graphic on ESKOM’s tariffs Dirk, it was worth the entrance fee just on its own)
    IC?…Illegal Connection..Sadly it’s a common feature of life here in the NSA.
    Before I close I would like to make it known that the local ANC appointed Executive Mayor of Ekurhuleni has an equally ‘elegant’ way of dealing with the power crisis, it involves a mouse and a screen and it’s called DEBT WRITE OFF v1.1 He uses it regularly to forgive the R2bn or so Eskom debts run up by the members-of the-community here in the ANC heartland of Gauteng. Of course the fact that tens of thousands of honest dumb-meter PAYERS such as myself actually fund that write off simply doesn’t enter mayor Whatshisface’s mind. Maybe it does, but he couldn’t give a phlying phuk either way.

  5. I’m not sure where the 20:1 factor comes in between peak and off peak electricity. A cursory glance at worldwide off peak pricing indicates a 2 or 3:1 factor. At 20:1, we would all have batteries in our homes and with an 80% efficiency on those batteries, be using electricity for “free”.

    Certainly, if the municipalities introduced peak metering, it wouldn’t be at 20:1.

    My model for renewable generation is far from unique, although I’ll claim clean room in coming to my own conclusions.

    Either way, the idea being to flatten the demand and supply curve.

  6. The peak periods are planned for summer and winter, well in advance. If it cost you a quarter to run your pool pump at 1am instead of 8am, you’d rush out and change your timer immediately. Same applies to geysers, dishwashers, washing machines, tumble driers, etc.

    Even a simple example of 10000 pool pumps moving from peak to off peak would flatten the demand curve by 10 megawatts.

  7. Greg Mahlknecht on

    As Simon mentioned above, by pricing electricity cheaper after hours, consumers will actually use the timer and scheduling functionality on their devices. I had some Germans stay with us a while ago, and they were puzzled why we ran the dishwasher, dryer, washing machine, etc during the evening – and were flabbergasted that our electricity is the same price all the time. After speaking to people who have had the cheaper after hours electricity, it’s clear that consumer habits to change quite quickly. I’ve seen this first hand, and our behavior was strange to them, not the other way round.

    My father ran a metal works in Isipingo, and they’ve had the differential electricity pricing there for decades; the amount they saved on electricity allowed him to run a nighshift with basically free labour.

    When money’s on the line, people and industry adapt VERY quickly… we know this from the same ideas in cellphone contracts, ADSL accounts.. petrol price rises yield giant queues at petrol stations. These are not new ideas, we just need them implemented on the electrical grid and consumers will leap right in.

  8. William Stucke on

    > I’m not sure where the 20:1 factor comes in between peak and off peak electricity.

    From the article above. See:
    “a more significant change is in the differentials between off-peak and peak pricing, which is closer to R240/kWh, a differential of 14 times”, and

    ” Just a few years ago, the differential between peak and off peak in Eskom’s MegaFlex tariff (the tariff applicable to most municipalities) was around eight times; the differential now is more than 15 times. In future, the differential is likely to increase to 20 times. “

  9. William Stucke on

    > Either way, the idea being to flatten the demand and supply curve.
    I agree absolutely. However, many people are under the erroneous impression that installing solar PV on their home helps. It doesn’t, as explained elsewhere.

  10. William Stucke on

    Sure, I’m well familiar with the European model – I’ve used it. In both the case of your father and your German visitors. you’re talking about well educated people.
    As we have seen in the good ol’ RSA, uneducated people are also perfectly capable of picking the cheaper (cellular) tariffs. My question revolves around how those consumers will know when the peak periods are. This applies especially to prepaid electricity users, I’d imagine.
    A complicating factor is SA is not just peak periods, but the enforced blackouts laughingly known as “load shedding”. If those periods can also be indicated to consumers, and their behaviour modified sufficiently, it may even be possible to avoid black outs at all 😉
    If the difference is significant enough, AND people know about it, then I am quite sure that behaviour will change.

  11. William Stucke on

    I set my pool pump to only run in off-peak periods long ago. I spent >R1000 buying and installing timers for the 4 geysers on my property, so that they are off during peak times.
    What financial benefit do I get for this? Zero.
    Yes, peak demand is about double of-peak. That doesn’t mean that peak cost is only double. Not by a long chalk.

  12. Yes: For the most part, there would be differential pricing for known and pre-set times of the day. A simple LED display would also indicate which times are peak, standard, off-peak. As this information is communicated in real time to the billing engine, or subtracted from the pre-paid card, it would be easy to have this information relayed via an app as well.

    Time of Use metering is not specifically directed at loadshedding but it has application there in the context of load curtailment. If a city, instead of loadshedding the whole of a particular zone, can get everyone to drop 10% of their load at the time loadshedding is called (might merely mean switching off a kettle or oven that was on before), then the required load reduction for say, loadshedding levels 1 & 2 (the most common) is achieved and loadshedding of a whole zone can be avoided. Real time measurement allows a city to determine whether required load reduction is being achieved which is something they are unable to do at present.

  13. You are, indeed, a good citizen! My geysers are on timers that make it most efficient to receive the hot water at the time I want. If there was peak metering, the most efficient time would be during off peak hours.

  14. I must disagree with you on this idea of uneducated people being capable of picking cheaper cellular rates. You surely have been talking to more informed ones.
    The average man on the street just knows his MNO, and whether he/she is on contract or prepaid. And often tells you he/she is on MTN mahala, which gives 100% discount at 02h35, when the moon is blue.
    Not aware that the normal daytime rates when there is only 10 or 20% discount is actually very expensive.
    My safe bet is that only 20% of persons know what they actually pay for voice calls, data and SMSs. Mostly they are not even aware that there are already cheaper price plans from the same provider. The cell phone companies have also been masters in confusing customers about their real rates.

  15. A very nice article, starting with the historical background of the issues at Eskom and our power crisis.
    Surely these or similar time-of-use meters should be rolled out ASAP. Not sure what the smart meters can do above these ToU meters, and what the cost of these smart meters are.
    But what the first part makes clear is that the core of the problem lies with the management of Eskom, and even more with the politicians supervising it. So the only real solution must come from bold and strong decisions by the governing party, and not in elegant technical solutions. And as our politicians are ideologically totally retarded and seriously challenged in proper financial/economic management, we first of all need to convince a majority of voters to vote smarter. So the real solution lies in voter education and a better functioning of our democracy.
    Not exactly an easy quick-fix.
    Another issue that I think is much bigger than generally recognised, is that a considerable part of the households are simply not paying for any electricity, either through illegal connections or simply not paying their accounts and not being disconnected. The energy expert Chris Yelland once wrote that about 18% of electricity is not paid for IIRC, which indeed means that most households are not paying in SA.
    It is well known that of all power supplied to Soweto, only 20% is paid for. The head of the municipality in this small town, the Unit Manager once conceded to me that 85% of all households here are actually not paying anything for electricity, rates and taxes.
    So before we think about ToU meters or smart meters, lets first and above of all have at least 90% of households actually pay for their power, even when we continue to grant the indigent households 50 or in some towns even 100 Kwh free per month.

© 2009 – 2021 NewsCentral Media