Lately I’ve been particularly energized, likely due to it being the beginning of the year when new projects kick off and conversations revolve around taking fresh approaches. So it’s fitting that one of the projects I have been involved in deals with the analysis of smart meter data from the energy sector. The use of smart meters is a developing field in the United Kingdom; utilities are interested in gathering consumers’ energy usage data on a much more granular level than ever before. A so-called smart meter is installed in a household to allow consumers to view their household electricity consumption on a real-time basis, while also transmitting this information back to the energy supplier.
Ten thousand times the data
Today in the UK, meters are typically read on a monthly basis, since most electricity meters are situated inside and thus meter reading is reliant on someone being home to provide access. Instead, estimates are often relied on for billing and consumer budgeting. Smart meters can record data on a more granular level, providing, for example, half-hourly consumption information to the energy supplier via wireless data networks.
With approximately 30 million residential houses in the UK, getting half-hourly readings would equate to 1.6 trillion smart readings a year. This ten-thousand-times increase in readings means BIG data! With big data there’s great potential to mine for insights into customers’ behaviors.
Customer segmentation to find your best customers
For energy providers, the most costly consumers to serve are those who demand energy during a daily peak period, which typically occurs in the late afternoon or early evening. If utilities do not generate or contract enough energy to meet peak demand, they are forced to buy short-term contracts through the wholesale energy market. The wholesale cost of energy is highest during peak, so with current single-rate costs, energy companies are likely to be making little or no profit – maybe even a loss – during this peak period. To prevent this operating loss, energy suppliers want to identify those consumers who are lower than average users during this peak period, since they have the potential to be the most profitable customers. They may also want to identify heavy peak users to modify their usage patterns or avoid acquiring them as customers.
In a recent project, we applied segmentation analysis to half-hourly usage data, identifying the various types of usage patterns, for both “peaky” and “nonpeaky” customers. Since suppliers want to smooth out the demand made by peaky customers, they can offer them financial incentives to move their consumption to off-peak hours using time-of-use guidelines with lower rates during off-peak times. This could entice consumers, for example, to set their dishwashers and dryers to run during off-peak hours or set heaters in the winter to come on before arriving home from work.
Stress relief: Dual-purpose data
Another purpose of the smart meter is to reduce the strain on the system during heavy usage periods; as electricity cannot be economically stored, it has to be generated on demand – this means the more peaky the demand, the more unused (or wasted) generation capacity there is for other times of the day. Once the customer energy usage segments have been identified, other data such as socio-demographic information (e.g., household size, location, household income) regarding consumers can then be overlaid onto these segments to identify which consumer groups display similar consumption patterns.
An example segment (Figure 2), displays the standard usage curve (dotted line) and the consumption pattern identified in the segment (solid line). The standard usage curve is used throughout the industry as a representation of average usage across the population. It shows a slight peak in energy usage around 8-9 a.m. and a more extensive peak in the early evening. In the example segment, we identified that usage in the weekday evening period was 40 percent higher than the standard usage across the entire population. This clearly identifies a segment which is “peakier” than the normal peak, and maybe a customer group an energy supplier would want to target in order to try and spread energy usage throughout the day.
Consumers benefit from this segmentation through greater understanding of their own energy consumption, allowing them to better manage cost of their consumption. It also allows them to understand which energy rate may suit their consumption needs more cost-effectively. Suppliers benefit by meeting energy demand without having to build expensive power-generating capacity, improving efficiencies in load management and smoothing out demand by offering consumers more appropriate rates. Everyone benefits by better understanding and managing energy demand, so that we can ensure sustainability of energy supply for future generations. Analytics can also be used in forecasting future demand to better plan for strain on the network and determine when and where new generators need to be built.
There is a magnitude of potential within the field of smart meter analytics, and this is just the beginning of what can be achieved when analytical techniques are applied to mine a rich new stream of information.