Implementing smart metering and residential time-of-use tariffs in South Africa


By Kobus van den Berg

Many utilities and countries in the world are very aware of the limitations in their power generation and distribution capacity. South Africa is experiencing a severe shortage of generation capacity, mainly due to the underestimation of economic growth demand on energy sources as well as delayed investment in additional generation capacity. It is also the case that the cost of electricity has been relatively low in the past and a positive drive to decrease the “real” price of electricity and the “electricity for all” campaigns have finally caught up with us.

The problem encountered can be summarised as follows. The generation shortage currently experienced requires a major investment in generation capacity over the next 10 to 15 years. The immediate generation problem cannot be solved permanently before 2012. In the next few years alternative solutions will have to be employed to alleviate the situation. Urgent measures have to be taken to ensure that the economy and people of South Africa successfully survive the power crisis. What should be done to involve domestic consumers in the process?


Figure 1 – TOU energy component comparison

The main focus to address the immediate problem should be on the more efficient use of available electric energy and certainly the use of alternative energy sources like liquid petroleum gas and solar.

Demand side management methods will also have to be explored to reduce and shift load patterns. These methods include water heater control, efficient lighting and appliances, as well as time-of-use (TOU) tariffs linked to suitable new metering systems, in order to involve the customer in the process to employ more efficient methods in using electricity.

The design of a TOU tariff for domestic use will be a critical factor in its acceptance and understanding by customers. Customers will have to perceive the tariff as fair and practical. Many customers will perceive the implementation of the TOU tariff as an unnecessary complication of their environment and try to avoid it. If the tariff does not convey the pricing message, many customers, especially the more affluent ones (often with inefficient loads) will merely pay the increased electricity tariff and still enjoy their conveniences.

If Eskom’s TOU tariff for large users (Megaflex) is used as an indicator of the supply price signal, the ratio of energy cost levels compared to a typical domestic TOU tariff structure are as shown in Figure 1. The bulk tariff structures have been unbundled to a larger extent than the domestic tariffs, and thus the energy component of domestic tariffs will include some non-energy components. The data in Figure 1 indicates a strong correlation between the supply tariff and the domestic retail energy tariff. The price signal in the supply tariff has been preserved and changes in the supply cost can be passed on to, as well as motivated to, customers. The suggestion is thus that domestic TOU tariffs should include the same time slots as bulk supply tariffs. The metering systems can easily accommodate the tariff and customers will understand and respond to these signals if the system provides them with real time information.

Experience with a small smart meter pilot scheme in the CENTLEC distribution area (in Mangaung and Southern Free State) to test the functionality highlighted some of the essential features required in a smart meter (see box). A number of three phase units without direct feedback as well as a number of single phase units with an in-home LCD panel were installed at the homes of mainly employees of CENTLEC. The idea was to involve people who understand the consumption and cost of operation of appliances in their homes. The feedback panels display graphic information regarding immediate kW load, last one, seven and 28 days of consumption and cost. Numeric information regarding tariffs, cost and consumption is also available on the display. The meters were not used to bill the participants but everybody observed and reacted to the price and consumption indications.

To be effective in South Africa, smart meter systems will have to include at least the following functionality:

  • Extensive tariff capability
  • Suitable storage of historic half hour consumption data
  • In-house remote display for tariff, historic consumption, immediate load, cost of energy per hour, both numeric and graphic, audible alarm if super peak tariff activated, messaging facility as well as tariff indicator lights to indicate the current tariff activated
  • Multiple communications options
  • Remote programmability of meter, tariff, time
  • Tamper reporting
  • An extensive back office monitoring, management and control software and hardware system.

Figure 2 – Individual domestic consumption profiles (left house with gas hot water heater, right standard hot water heater)

The main observation in this pilot was the fact that without a direct feedback mechanism, smart metering does not involve the customer and one cannot expect any positive participation from them. A bar of red indicators during peak times appeared to be very intimidating and triggered reaction. Consumption data selected from the group illustrates that one cannot really define the consumption pattern in a domestic environment and generalise it to all households (Figure 2). The consumption in each house will be different due to number of people, working hours, school holidays, appliances, environmental temperature, and a large number of other behavioural factors. It is thus vital that the inhabitants are motivated to intelligently use and manage their own environment.


Table 1 – Typical household consumption (March 2008)

In another experiment the consumption of subgroups of electricity loads were measured (Table 1).

This data summarises the consumption of a typical household with three people with an average consumption level in the Bloemfontein area. The supply circuits were grouped in four circuits and measured with four meters. A smart meter system was fitted for monitoring purposes only. The “plugs” group included all the household appliances, like dishwasher, tumble drier, washing machine, microwave oven, kettle, pool pump, etc. The lighting consisted of compact fluorescent lamps (CFLs, including two security lights) and 40 W tube fluorescents where a higher level of illumination is required. The pool pump operation was kept to a minimum and hot water heaters switched on for only 4 hours per day via CENTLEC’s ripple control system. During the winter months gas space heating will be used and electric blankets utilised during sleeping hours in monitoring the consumption during this colder period.

The controllable load is thus limited to the pool pump and high consumption appliances like the dishwasher and tumble drier. The dishwasher is normally used once a day and is switched on at about 22h00. Load shifting has thus already been addressed. The inhabitants of this house are aware of electricity saving measures and thus attempt to use electricity effectively.

There may still be scope for savings but this will require major investments in a solar water heater and gas stove. The solar heater will have to be paid with the current R123 (US$15) per month used for water heating and the gas stove as well as the gas with the current R27 (US$3.50) for cooking. The solar water heater with an investment of R8,000 (US$1,000) will take at least seven years to break even at the current electricity tariff levels. The replacement of the electric stove with a gas stove is not a financially viable option.


Figure 3 – Effects of water heater control in CENTLEC
(blue 2007, red 2008)

In the CENTLEC case, water heaters are controlled via ripple control devices and the meters were not used to directly control any appliances in the homes of participants. The effect of drastic water heater control can be observed in the demand curve for the larger Bloemfontein area (Figure 3). To reduce the number of load shedding instances in the city, water heaters were switched off for 20 hours per day and heated up only from 03h30 to 05h30 and from 15h30 to 17h30. The switch-on and off effects can be seen. This process by itself reduced peak demand and overall consumption by a few percent. A project with Eskom will install another 15,000 units in the city. The advantage of water heater control is the immediate effect it has on the demand at a particular time. The drawback is the switchback peak it can cause if the heaters are all reconnected at the same time, and thus a phased process is required. This can be seen in Figure 3 where the switch-on process between 15h00 and 17h00 is controlled to result in a flat load curve to prevent the demand from exceeding a pre-determined value.

The customers are, however, not involved in the process. Smart metering can be used for the control of water heaters but if this is the prime objective it can be done more economically using remote control devices. It is thus necessary to apply smart metering in a much wider sense than switching off loads at will.

The main objections to smart meters at this stage are the lack of standardised protocols, functionality, and interoperability. The NRS 049 standard specification will address the problem to a certain extent but is still not a guarantee that metering systems will be interoperable to the extent that meters or systems can be mixed and matched without being locked into a particular supplier. A standard protocol based on the DLMS system will have to be developed by the utilities and specified as the only acceptable protocol to facilitate standardisation. The cost of metering systems is also very high.

It seems that the South African distribution industry is jumping on the smart meter bandwagon without proper evaluation and standardisation of requirements, with the prospect of learning the same incompatibility lessons experienced with prepayment metering systems in the early 1990s. That process resulted in a major re-investment in the replacement of meters after a suitable standard (STS) was adopted by all role players. The only difference during this round is that the investment will be significantly larger and replacement will not be a financially viable option in the short and medium term. With all the costly arrears in generation capacity it will be necessary to do a proper analysis of the installation cost, cost of ownership and lifespan of the systems. If the lifespan is shorter than 10 years it means that a re-investment will have to commence by 2015 (before the required generation capacity has been established effectively). Calculations to prove the financial viability of smart metering to save on the establishment of additional generation capacity will have to reflect this fact.

The customer segment currently served by prepayment meters in the CENTLEC distribution area on average consumes less than 200 kWh/house/month. This sector mainly uses electricity for lighting and home appliances like TVs and radios, but in most cases they do not have a water heater fitted. There is also still a tendency to use paraffin for cooking and heating. The main source of saving electricity is in the lighting sector. The savings required can be realised with the fitment of CFLs and that project is currently in progress. A calculation on the cost of prepayment meters, which are significantly cheaper than current smart meter systems, normally only breaks even after 10 to 12 years. To install smart meters in this segment is thus simply not financially viable at present.

The consumption of customers mainly served on credit metering systems in the CENTLEC distribution area averages around 900 kWh per month. A large percentage of these customers do have geysers and other appliances with higher consumption that should be managed more efficiently. This implies that in the domestic environment the credit meter customer should be concentrated on.

Larger homes and guest houses are normally supplied via three phase connections and consume significantly more, and should certainly be included in the project. It goes without saying that all business customers except the very smallest ones should be good candidates for TOU tariffs with smart meter systems.

If the total load in a domestic environment (900 kWh) has to be reduced by 10% this implies that 90 kWh per month or, at current prices, about R36 (US$4.50) must be saved. The investment in smart meters and other energy efficiency measures is thus worth about R36 to the customer. How much effort will anybody expend to save this amount? Even if the electricity price doubles only R72 (US$9.00) is at stake. It is true that the combined saving is worth much more on the supply side in terms of generation capacity but the customer will not directly see a major saving on his/her electricity bill. The customer will in many cases have to invest in water heater blankets, CFLs, solar water heaters and gas stoves and heaters to accomplish this saving – an investment that will take years to recover. How does one then convince the customer that all these expensive measures are saving him anything on the energy bill? Affluent customers might, after the initial savings hype, simply ignore the socalled saving measures.

In the case of CENTLEC a possible 35,000 domestic customers could be eligible for the installation of smart metering systems – effectively an investment of R105 million (US$1.4 million). After the completion of the current ripple relay DSM project most of these customers will be fitted with water heater control. It will thus be possible to control the main source of energy consumption in the residential environment constituting 40% of the average home’s consumption. The remaining 540 units consumed by the average home in this area are used for cooking, heating, lighting and other appliances. If a major percentage of these homes switch to gas for heating and some to gas for cooking and fit CFLs where possible, hardly any controllable load remains. The smart meter system will thus be used to effect a saving of say another 10% of the remaining 540 units, or 54 units per month. In effect this utility will have to spend R3,000 (US$375) plus all the additional communication and operational costs per household to control 54 units per month. With a more technically advanced metering system additional personnel will be required to operate and manage the system and more expensive technicians instead of electricians will be required to maintain the systems. The saving in units will also affect the annual revenue of the utility and negatively influence its financial viability.

The million dollar question thus remains: Why would a utility switch all customers with a consumption level of more 500 kWh per month with the general notion that smart metering is the ultimate solution? It would seem that smart metering is not necessarily the ultimate solution to the problem. We will have to go back to the drawing board and perform a detailed financial analysis in each utility to assess its viability in the South African scenario.

In closure it should be mentioned that many international distribution companies and governments use a medium to longer term implementation strategy and not a “big bang” approach. This approach combined with selective customer segment targeting will result in a much improved financial return on investment in smart meter systems.

 The author’s views in this article do not necessarily reflect the official view of CENTLEC.