We all know that not everyone uses energy the same way.
Some of us shut off all the lights when we leave in the morning until we get home at 6 p.m., others crank up the AC in the mid-afternoon. But, how does a utility go about uncovering these kinds of patterns for hundreds of thousands or millions of customers?
It starts with data of course. By combining detailed energy data along multiple dimensions — such as time, geography, and weather — you can then tease out key similarities and differences among types of energy users.
Working to extract insights from all those data is exciting. But to an untrained eye, it can also be overwhelming.
A quick glance at the below graph illustrates the point. The graph displays weather-normalized hourly electricity consumption from a random sample of 1,000 residential utility customers, for a typical weekday. It’s not hard for an onlooker to be intimidated by the blob-like result.
But with some clever machine learning logic and the right big data architecture (our data warehousing and processing frameworks run on tools like Hadoop and MapReduce), it’s not long before you can start finding signals in the noise. In recent months, we did so by analyzing usage data from 812,000 utility customers (for simplicity, a fraction of that dataset is displayed here) in 3 major US metropolitan regions. By applying advanced clustering techniques, such as vector quantization, we were able to identify a series of recurring patterns across the usage data.
Specifically, based on our statistical clustering, you can see around five distinct hourly electric load patterns start to emerge from what used to be a mere jumble. Each of these well-defined patterns can be described as a particular “load archetype.” There are about five weekday load archetypes discernible in the below graph.
A little color can help further illuminate different load archetypes extracted from the dataset. For example, below you can transparently distinguish between distinct categories of customers, such as those whose usage spikes in the morning ("the coffee makers" - black curve) versus those whose usage reaches a maximum around 5pm ("the late afternoon peakers" - dark blue curve).
An important discovery of our statistical analysis is that constructing load archetypes at scale — and classifying customers within them — is not only readily possible; it can also unlock new opportunities for utilities and their customers.
Utilities around the world rely on Opower’s customer engagement platform to use data insights, like load curve archetypes, to deliver the right message at the right time to the right customer. By coupling these data insights with personalized communications, utilities can improve the customer experience while at the same time boosting the impact and cost-effectiveness of their programs.
For example, if a utility can easily and quickly identify the customers in a region who most closely fall into a “late afternoon peaker” load archetype, then the utility can take a more targeted and direct approach to delivering peak reduction programs like behavioral demand response or smart thermostat management. In such a scenario, a utility saves time and money by focusing their efforts on customers who are best positioned to reduce peak load, and all customers are happier because they’re receiving offers that are most relevant to them (and not receiving offers that aren’t relevant).
Or imagine a load archetype that corresponds to electric vehicle owners who tend to charge their cars during the daytime. A utility could identify customers whose usage behavior falls within that archetype class, and deliver automated targeted outreach to them about special rate plans that incentivize car charging after midnight (when the grid has more excess capacity).
Load archetypes and the market segmentation possibilities that flow from it are just a couple ways that utilities can infuse data-driven personalization into the utility customer experience. For more on how advanced data insights are creating next-generation opportunities for utilities and their customers, check out this nifty disaggregation algorithm that is unlocking the power of the smart grid.