Authors: Ben Anderson & Tom Rushby


The importance of flexibility for a zero-emissions electricity system continues to drive considerable research and socio-technical innovation across CREDS and its linked activities. Inherent in much, although not all, of this work is the notion that we need to find ways to enable flexibility at the consumer end of the line. But do we?

In energy research in general, this is often simplistically translated into a form of congestion-based charging. Here, price signalling via variable time-of-use tariffs will ‘naturally’ incentivise cost-optimising and economically rational domestic consumers to reduce and/or shift (flex) their demand in expensive (peak) demand periods. This model should, but doesn’t always, lead to important questions about equity and energy justice:

  • Will those who are less well-off be less able to flex and so penalised during periods of high prices? Or might they have additional flexibility capital which enables them to benefit?
  • If they do, will they have the skills and motivation to choose the ‘right’ tariff? The evidence on this is not good: even when provided with full information in a task that required no greater then primary school level maths, only 44% of a representative sample of GB bill payers selected the ‘best’ ToU tariff for a given consumption pattern (Table 14 of Nicolson’s PhD thesis). Crucially, ‘lower’ social grades perform even worse on this task (39%). This raises serious questions about the default ‘energy justice’ consequences of relying on ‘informed choice’ of ToU tariffs.
  • Will the consumers who are likely to benefit self-select into the ‘best’ tariff for them as many in the sector, including in regulation, appear to believe? On this evidence, probably not.

Distributional Effects of Time of Use Tariffs

A recent workshop on the “Distributional Effects of Time of Use Tariffs”, co-hosted by CREDS and the University of Reading’s DEePRED project, presented the results of a series of UK household time-use diary based models intended to explore these issues. Setting aside the methodological difficulty of reliably linking reported activities with actual electricity use, the time-use models suggested that “Complete electrification of cooking may disadvantage retired couples, couples with one child and single parents with one child if exposed to ToU” More generally the presence of children appeared to correlate with higher ‘peak time’ demand and also, therefore, with a more substantial ‘penalty’ if they were unable to reduce or shift. By extending the model to examine the ‘worst case’ scenario of no behavioural change under ToU pricing, the work also suggests that “ToU tariffs would financially benefit single parents in the low-income group” due to their increased levels of day-time occupancy. Although this, of course, rests on the assumption that such ‘in principle’ flexibility capital can actually be deployed. Intriguingly the models also suggest that there may need to be regional variation in the times at which higher prices are set due to regional variation in the timing of ‘peak time’ energy using activities. This is already one potential direction of travel highlighted in Ofgem’s ongoing network cost allocation review – although for slightly different reasons.

But does variable pricing deliver changes in behaviour?

Can variable pricing deliver flexibility? Again, the evidence is not strong: Recent reviews of responsiveness to price and non-price or combined incentives suggest that we can expect demand reduction of at most 5-15% with substantial variation between households and across studies. We should expect consumer-wide impacts to be at the lower end of this range as many of the reviewed studies used self-selecting and thus biased samples of consumers who were more likely to respond to incentives. We also know that consumers are particularly price insensitive in the evening peak period although there are indications that this varies by social group. Using price incentives may also lead to undesired effects, for example increasing demand outside of critical peak events, thus increasing peaks at other times.

We could continue to hope that variable pricing will deliver system flexibility and redouble our efforts to ‘engage’ the customer base. Surely they would make the ‘right’ decision(s) if we but gave them the ‘right information’, the ‘right graphics’ or a ‘better dashboard’. Or would they? Nicolson’s research suggests not, as does the reported flight of Italian consumers back from mandatory time of use tariffs to flat rate charging when given the choice.

Moreover it seems likely that the inherently inter-dependent sequences of energy-using practices actively hinders price-induced response. It may be technically feasible to shift appliance-use around – but for many people this would significantly impact comfort, convenience and the flow of everyday life. Activities (including those that use energy) are locked in sequences constrained by patterns of work, commuting, child-schooling, leisure, domestic tasks (cooking, laundry etc). When ‘stuff is done’ seems to have substantial inertia as studies of electricity demand during UK’s COVID-19 lockdown period showed. Even when ‘normal’ commuting/schooling/working time constraints were removed from a substantial proportion of the population, temporal patterns of residential energy use appeared to stay largely constant.

Perhaps we really do need to admit that consumers are generally not rationally acting cost optimisers. Expecting flexibility to emerge from the ability of residential customers to actively shift or adjust their real-time patterns of energy demand in response to anything other than punitive pricing appears unrealistic as our work on the Solent Achieving Value from Efficiency (SAVE) project suggests. Further, if there is some doubt as to whether informed choices are (correctly) made even with full information and support, effective commercial services, policy and regulation cannot end with this assumption. A different approach is required.

Whose flexibility is it anyway?

So where do we go from here? Perhaps we should re-ask ourselves: Who needs flexibility anyway? Is it the system, or the consumer? If the former, how can we nurture system flexibility without the apparently troublesome need to directly engage consumers in a ‘flexibility market’ at all? A recent review of two contrasting trials makes this clear. Drawing on dimensions of flexibility capital, it highlighted that substantially greater demand reduction might be secured where the consumer is not even aware that their demands on the system are being flexed. This is not news – decoupling demand from supply is equivalent to energy storage, but of course this doesn’t need to be direct storage of electrical energy. Flexibility solutions may be found in a range of socio-technical arrangements (products + practices) that provide automated demand response/direct control or otherwise de-couple energy use from energy demand. These include thermal (heat/cool/hot water) storage of various kinds, available in the thermal mass of many dwellings, and further enabled by building fabric upgrades and technologies such as heat batteries, phase-change materials and those hot water tanks we have been so keen to remove over the last 20 years. Our own research on LATENT (ResidentiaL HeAT As An Energy SysTem Service) is testing exactly these approaches.

Interestingly, these approaches feature strongly in the recently published UK Government smart systems and flexibility plan which concludes that “smart charging of electric vehicles and heat pumps combined with heat storage provide the largest potential for DSR [demand side response]”. The plan assumes that domestic smart appliance-based flexibility can deliver a mere 3% demand reduction in a given half hour, various forms of thermal flexibility offer up to 20% while smart EV charging can deliver up to 90%. What need here for consumers responding directly to variable price tariffs?

But of course variable price tariffs can provide a profitable space for others, even if consumers themselves are not directly in the loop. For example, New Zealand’s SolarCity installs PV & battery systems in homes at zero upfront cost to consumers using a commercial long term finance facility. The consumer contracts to buy electricity generated and/or stored by this ‘behind the meter’ Virtual Power Plant plus any ‘top up’ power required from SolarCity. New Zealanders are no strangers to direct, and usually invisible, hot water heating load ripple control by distribution network operators which routinely reduces evening peak demand by ~10%. Similarly, SolarCity report considerable success in decoupling grid demand from end-user consumption thus invisibly shifting load to both reduce peaks in demand and overall power bills. SolarCity’s scale is now sufficient that their “fleet of over three thousand five hundred batteries” are contracted to the New Zealand’s national grid operator’s Demand Response Programme and they are exploring ways to help manage location-specific low voltage network constraints. Anecdotally, SolarCity have seen substantial uptake by low-income households due to no up-front capital cost and lack of exposure to price ‘shocks’ of the kind that impacted Flick energy’s wholesale half-hourly price-following tariff. The latter demonstrated that rather than respond to sharp variable price tariff rises by reducing demand, like the Italian example cited above, the majority of customers simply switched (back) to a flat tariff supplier causing Flick to lose substantial market share in a very short space of time.

Conclusion

Perhaps the flexibility ‘solution’ is not therefore the mythical engaged, actively optimising price responsive consumer but optimisation based on automation where households are not expected to have to micro-manage (or resist) energy decisions. Perhaps the flexibility the system actually needs requires active disengagement so that the grid ‘sees flexibility’ but consumers can get on with their everyday lives? Yet if ‘active disengagement’ is the desired approach, the issue of potential injustices through unequal flexibility capital remains an important concern. What stake will consumers require in this emerging system in order to consent to differing levels of automation, and what level of control will they need – or want – to retain? Will a new ‘social licence to automate’ be required?


A version of this post also appeared on the CREDS blog.