Smart meters in the UK: an epic failure?

Some readers may conclude from my articles that I am less than impressed by current regulators and policymakers. I try to remember that I have been there and done that. I have my own history of regrets about advice and decisions that turned out badly. Often, outcomes are constrained by considerations which should have been set aside. None of us has a crystal ball and criticisms that rely heavily on hindsight are not convincing. Yet, some decisions are so foreseeably bad that no excuses should be accepted. This article deals with one such case.[1]

From a commercial perspective, smart meters offer two large benefits. First, and most immediate, they replace the need to employ an army of meter readers visiting customer premises at regular intervals. Self-reported meter readings are only a partial substitute, because, perhaps once a year, it is necessary to collect real meter readings to ensure that consumption is not being misreported, whether by accident or as deliberate fraud.

Second, they allow for the convenient adoption of time-of-day tariffs – such as peak/off-peak tariffs (most widely known in the UK as Economy 7 tariffs) and more sophisticated tariffs including ones linked to market prices that are called dynamic tariffs. Time-of-day tariffs are not new but the original requirement for separate meters (White meters) plus associated electrical circuits greatly limited their application in the UK.

From an academic perspective, smart meters are great. The capability to collect close to real-time data on demand and network utilization for energy, telecoms, and other networks offers the possibility of more efficient management of those networks. Network improvements can be better designed and, under some circumstances, customers may be able to adjust their usage of networks to reduce their bills or make better use of their infrastructure. However, the benefits of smart meters are not equal in all uses.

In the UK there has been a focus on installing smart meters for both electricity and gas even though the networks operate very differently. Gas networks provide a lot of storage and resilience via internal adjustments in pressure in both distribution pipes and transmission mains. There is no strong case for signalling changes in usage every 30 or 60 minutes for customers using small or medium volumes. In this respect, gas is closer to water than to electricity, so it would make more sense to combine volumetric smart meters for gas and water.

Electricity is different because electricity networks provide minimal storage and require continuous balancing. As intermittent sources of generation provide an increasing share of total generation, the case for signalling when marginal prices are low or high, reflecting surpluses or deficits in intermittent generation, becomes more important. Hence, in this article I want to move away from thinking from about energy use to thinking about meters as devices for data collection and communications. Our comparisons should focus on communications networks – telecoms – rather than gas and water networks.

As anyone who has designed or operated communications networks knows, it is rarely necessary to collect data at the level of individual consumers for the purpose of network management. It is sufficient to know what is going on at segment level, i.e. for groups of 5 to 50 connections. The value of smart meters or individual control units lies in handling customer interactions – e.g. billing, usage monitoring and troubleshooting.

This statement is based on practical experience of managing broadband networks, which have a lot in common with electricity networks. We are concerned with the traffic (load) at specific nodes in the network and passing along specific routes. The traffic from individual connections is constrained by the technical limits on those connections. If there are contractual limits on, for example, data usage, it is not necessary to rely on real-time monitoring to impose them. Checks every 5, 15 or 30 minutes are good enough.

Neither billing nor network management requires more than quite dumb smart meters which do little more than report a meter identifier plus usage over some interval to a central database. That is the way telephone networks and bills have been managed for decades. Indeed, that structure of communication is inevitable in any country like the UK which has chosen to use wireless networks for meter communication. The UK’s arrangement is simply a remote billing system that collects data on electricity consumption at specific intervals and is of minimal practical value for network management.

If we discount the irrelevant technical fluff about network management, the economic and technical merits of smart meters revolve around two main sets of questions.

The first question is whether they are a cost-effective way of replacing meter readers or alternative ways of obtaining meter readings. The almost certain answer is yes. One piece of evidence is that telephone systems went down the same route decades earlier and it is strange that electricity systems did not follow suit more quickly.

I am over-simplifying things. Even with party lines, telephone systems need identifiable endpoints. Thus, traffic identification is an inherent feature of telecoms, which is why we had many telephone exchanges, and that facilitates billing.  It matters which voice packets go where in communications networks, but it is not important which electrons go where in electricity networks.

In data processing terms, the fundamental feature of a smart meter is the capability to report upstream rather than just count downstream traffic. We should remember that every electricity network must have that capability down to some reasonably fine level as that is what grid and distribution meters do. The only question is whether and how we move the identifiers and reporting further downstream.

The second question is how to adopt time of use charging. Once we have upstream reporting of usage, it is (relatively) simple to introduce time of use charging. The difference between peak (work periods) and off-peak telephone tariffs was familiar to any telephone subscriber until very recently. Such tariffs are very simple. Everyone knew that taking advantage of lower rates meant not calling between 8 am and 6 pm on weekdays. Quite straightforward unless you had to call a business during regular working hours. However, it is rather more complicated for electricity networks. The main peaks in the UK are from 8 am to 10 am and from 4 pm to 8 pm. Boiling a kettle at 3.30 pm doesn’t work if you want to have tea after getting home at 4.30 pm. Unlike telephone systems whose peaks have usually been caused by business usage, the main peaks of most electricity systems reflect household usage.

Most experience with time of use tariffs is that they have a clear but small effect in shifting electricity from peak or high-price periods to off-peak or low-price periods. A few people running washing machines or driers after 11 pm makes little difference to peak demand. The biggest effects are found when some form of storage is available and reasonably large. In the past that was either hot water or heating systems. In future, electric vehicles and battery storage may be important, but the capital costs of acquiring storage are high and may be hard to justify in financial terms. There may be more scope for using buildings for heat storage but that would require a shift away from an obsessive focus on insulation and minimising heat loss.

Much can be learned from personal experience with smart meters in different locations and countries. Our family has two houses – one in Scotland and the other in the north of Italy. Both have smart meters for electricity and neither has a piped gas connection though our house in Italy has an LPG tank with wireless monitoring of usage. The smart meter in Italy was installed by the distribution company Enel more than 17 years ago and has worked without drama since then. It is an older model of smart meter that reports consumption for 3 time periods defined by the Italian regulator. The smart meter in Scotland was installed by a contractor employed by our energy supplier Octopus in July 2023. It has never worked properly, despite repeated complaints and follow-up visits. It does report half-hourly consumption to the central hub but from the consumer perspective it is even less informative than a non-smart meter.

The difference between the two cases highlights a critical feature of the UK program. In Italy smart meters were installed, first by Enel and then by other distribution companies, as a way of saving money. Enel’s cost-benefit suggested that they would save over €3 billion by eliminating meter readings, automating the preparation of bills, and collecting better information on usage. Installing smart meters was a commercial decision that took no account of claims about saving energy, etc. The distribution company owns the meters and managed the process of meter replacement to minimise costs.

In contrast, the UK program has been made technically and administratively more difficult by a regulatory decision made in the mid-2000s. The regulator Ofgem wanted to increase competition in the electricity sector by introducing competition for metering services. It transferred the ownership of meters from the distribution companies (DNOs – or DSOs in Euro-speak) to energy suppliers with the idea that suppliers would have more incentive to reduce costs by contracting with different providers of metering service. This sounded good on paper but has proved to be absurdly messy in practice.

More important it was a classic piece of short-sighted policymaking. By moving meters away from DNOs they had no reason to facilitate and invest in the transmission of data over their networks, which is what Enel and most smart meter programs do. Instead, smart meters in the UK rely on wireless transmission of data with all the attendant problems rather than a version of what in the domestic context is called powerline networking.

The problems caused by relying on wireless transmission have been exacerbated by the historic practice of locating electricity and gas meters inside houses, often in hard to access locations. Moving meters is difficult and expensive but wireless-read meters are more reliable and efficient when they are located on the outside of properties or, as in the case of our house in Italy, on the entrance to the property. That lesson is reinforced by our experience with a water meter in Washington DC where the water company replaced old meters inside buildings by wireless-read meters fitted to the supply pipe in the street or the outside of the property.

Another lesson from experience. For nearly 15 years we had an apartment in the Barbican Estate in the City of London, which consists of about 2,000 apartments in medium and high-rise blocks built between 1965 and 1976. Thames Water, our water company, was required to offer the option of metered water charges, which we requested. It turned out that it was impossible to install water meters for individual apartments because of the way the water pipe network in the building had been designed. All the electricity meters were installed in the basement of the building so that they would have to be replaced by smart meters at one time and wireless communication to in-apartment displays would be impossible.

In countries with a large legacy of old buildings it is details such as these which make a large difference to the success or failure of such projects. Maybe, 60% or even 80% of properties are straightforward but it is how the remaining 40% or 20% are dealt with that is crucial to the overall success of the program. The Barbican Estate is barely 50 years old, young by comparison with most of London, yet its internal infrastructure and design is poorly suited for modern services. In this respect, it is no worse than perhaps 90% of other multi-occupancy buildings constructed before 1980.

There is a final key questions, again illustrated by experience. Should electricity smart meters include two-way communication that would allow either the network operator or others to turn off the power? Often that capability is built-in, but it is the source of great suspicion in any system where it seems likely that power outages may become frequent in future. The suspicion is exaggerated, if only because imposing rolling outages is always possible by shutting down transformers or power connection nodes. It reflects a lack of trust in both governments and network operators that hinders the acceptance of smart meters and other methods of improved network control that require public cooperation.

In any case, switching off electricity is a very blunt and inefficient method of power management. When we were living in Washington DC in the 1990s we were on an electricity tariff that allowed the network operator to switch off our air conditioning system for a few hours at a time under specified conditions. This tariff included a substantial discount for regular consumption. The point was control of a key appliance rather than switching off the power supply to the house entirely. For such an arrangement we need either smart meters that are a lot smarter than the current ones or a standardised API or other means by which specific appliances – EV chargers, heat pumps, other heaters – can be controlled remotely.

Drawing upon experience from around the world, the smart meter program in the UK has been an expensive and foolish shambles. The fundamental reason for this mess was the decision to have it managed by energy suppliers rather than by distribution network operators. Almost everywhere implementation by DNOs has resulted in a transition that was less expensive and more rapid than in the UK.

The example of Italy was available when the decision about how to proceed was made in the UK. It was regulatory dogma and a perverse belief in British exceptionalism that led to the core mistake. That, in turn, determined the reliance on wireless communication that is the continuing weakness of the whole system. To achieve full coverage of smart meters, the UK will have to transition again to a system of powerline communication as in most other countries. That is some way ahead. For now the regulator and policymakers are intent on digging themselves ever deeper into a hole!

[1] I should declare an interest in this topic. Together with family and friends I have established and am running a company called Polare AI Ltd plus associated companies. Our goal is to develop both hardware and software that allows for the control of electrical appliances in response to variations in the market price of electricity.

One of the lessons from studies of time of use pricing is that households and businesses have limited interest or willingness to invest the effort and time required to adjust their electricity consumption to frequent price changes. Simple peak/off-peak pricing works but price structures with several time periods or hourly prices are difficult for even the best-informed consumers. This is a context where a combination of AI and proper software can achieve savings by remote control that are never attained by manual methods.

If you want to learn more about this, please see our website at https://polare.ai/. Remember that we are at a very early stage as the necessary technology is very new. If you are interested, you can sign up for our trial program.