Estudio del establecimiento de áreas de influencia

The following paragraphs provides a brief descriptions of potential research options for improving the understanding of each of the knowledge areas listed in the previous sections.

Dynamic Time-of-Use (dToU) trials. The complexity and subjectivity of residential demand makes it unlikely that theory alone can inform on the effect of dToU signals. Understanding this must include an experimental approach. Attention to statistically significant sample sizes, appropriate social stratification and good experimental design will be crucial in ensuring the quality of results. This has been the primary contribution of the Low Carbon London (LCL) residential dToU trial.

Data on the types of loads shifted. The most direct approach would be to sub-meter ap-pliances at high resolution within the household. This would give unambiguous confirmation of the loads that were contributing to the measured response for a particular event. A less intrusive alternative might be identification and disaggregation of load signatures [83] from the power mea-surements of the primary meter, though this approach requires high resolution meamea-surements and more complex analysis techniques. However, both these approaches do not differentiate between actions that are the result of deliberate engagement with the DR programme, and coincidence. To gain more information here, surveys may be used to ascertain the appliances that the consumer finds easiest and most difficult to respond with.

Data on the effect of household occupancy. Statistically significant data on the effect of household occupancy can be obtained by ensuring the number of samples in future trials are such that, after grouping by the number of occupants in each household, each group still contains a significant number of households. The data necessary to determine the number of occupants in each household will necessarily have to be gathered via a survey.

Evidence and reasons for differing responses from vulnerable or low-income con-sumers. More data on low-income and vulnerable consumers may be obtained by targeting DR trial recruitment such that statistically significant numbers are selected from low income and vulnerable consumer classes. This could be augmented by consumer surveys to obtain data on the types of appliances that consumers are shifting. Data on the effects of household occupancy should be obtained in a similar manner, by choosing trial areas and targeting recruitment until statistically significant numbers of the chosen occupancy levels are obtained. This is now being studied in a related Low Carbon Network Fund (LCNF) project, Vulnerable Customers and Energy Efficiency, discussed in Section 3.4.

Evidence on persistence of DR. Data on the persistence of peak load reductions can only be obtained from longer trials or commercial offerings. Increasing trial duration or designing trials that may transition into commercially viable operations would provide valuable long term data.

Evidence on the effect of information provision and its vector. If the experiment is designed to test information interventions, all other variables should be held constant to as greater degree as reasonably possible. If the experiment is not designed to test such interventions, the information provided to consumers, and the vector by which it is provided, should remain a constant of the experiment.

Data on the response of consumers to different price levels. As can be see in Fig. 3.1, trials that were considered closely related to the UK context have tended to be conservative in their choice of peak to standard price ratio—there is currently little experimental response data beyond a price ratio of 3. Future trials should therefore aim to obtain data beyond this point.

More data on the effect of price differentials may be obtained by designing experiments with significant numbers of participants, using multiple tariff bands by using a sufficient range of price ratios to make it possible to deduce trends (should they exist) from results. As there are many time related variables which cannot be controlled in trials, ensuring the pricing schedule is designed to minimise noise while appropriately spanning seasons will be important.

Data on the effect and extent of the electrification of heat and transport. Until pene-tration levels for electric vehicles (EVs) increase, it will be difficult to obtain statistically significant experimental results on their DR potential, or indeed the network issues they may pose, from in-dividual trials. In order to make best use of the information available, current efforts are geared towards increasing penetration levels and the international sharing of research data. Green eMo-tion [84] is the largest such programme in Europe. Combining internaeMo-tional data with local data such as driver patterns and consumer preferences may be the best approach at this stage in EV roll-out. In the UK, plans to test small fleets of instrumented EVs in order to obtain information on driver patterns and charging preferences, as well as channeling investment into EV charging infrastructure, are underway in major cities and, indeed, were part of the LCL programme [85].

Data on the effect of heat pumps on the network and their potential to engage in DR is in a similar condition to that of EVs, though their situation is somewhat different. As heat pumps will be replacing conventional heating systems with little to no difference in service dynamics (unlike electric vehicles which have considerable reduced range and increased charging times relative the the incumbent), usage patterns can be derived from existing data. Furthermore, as these systems are effectively reversed air conditioning units, their aggregate load characteristics are relatively well understood. As load cycles may look similar to that of cooling loads in the USA, it may even be possible to infer their DR potential from existing data. Nevertheless, assumptions like this should

not be made without experimental verification. It should be noted that the scale of such trials should not need to be as large as those necessary to test consumer responsiveness to prices.