Smart Meter Q&A with Nur Sunar
Thursday February 18, 2021
Dr. Nur Sunar is an Assistant Professor of Operations at UNC Kenan-Flagler
Q1. Can you explain what you’re studying in this paper? How does data play a role in your analysis and in this industry?
In this paper, we design an innovative clustering and pricing policy for an electric utility company (utility) that sells retail electricity to end users (customers).
Smart meters and other innovative technologies such as the Internet of Things (IoT) and smart home appliances provide highly granular data on customer consumption patterns to utilities. In addition to the large-scale smart meter data, utilities can have access to weather data (e.g., temperature, humidity, pressure) and various household-level information, including demographics (e.g., time spent at home), and home characteristics (e.g., heating, ventilation, air conditioning, foundation).
Considering these, in this paper, we design a data-driven clustering and personalized dynamic pricing policy for a utility to help it maximize its profits by exploiting rich information in such large data sets. To examine the practical impact of our policy, we conduct case studies based on real-life smart meter data from Texas. Our data-driven study reveals that relative to the company policy, our policy increases the company profits by 146 percent over a three-month period.
With the rapid deployment of smart meters, the energy sector needs to embrace advanced data analytics more than ever. Sophisticated data analytics techniques can be utilized in many areas of the energy sector, such as load profiling, demand forecasting, renewable energy forecasting and outage prediction.
Data analysis is at the core of our work. In this paper, we first uncover a cluster structure based on weather and household-level data using spectral clustering, which is a machine learning technique. Each cluster can be considered as a separate customer segment. After clustering, we gradually learn the customers’ price sensitivity and willingness to pay within each segment. By calibrating the consumption model using real-life smart meter data from Texas, we compare our policy with the company’s historical pricing decisions. Overall, our paper emphasizes the value of advanced data analytics for utilities.
Q2. How did you identify this particular relationship between underlying cluster structure and the corresponding consumption models? What led you to look at this specific opportunity for profit maximization?
Intuitively, electricity consumption is closely related to weather, household-related characteristics, and building-related characteristics. For example, consumers who have a larger house or spend more time at home tend to consume more electricity. We formulate all relevant information that might affect energy consumption as features, and perform a clustering analysis on features. Depending on these features, different households might respond differently to a change in the retail electricity price. So, each cluster represents a collection of features that defines a different electricity consumption behavior with respect to price. With this in mind, we identify the relationship between the cluster structure and electricity consumption, and discover that different clusters exhibit different price sensitivities and different willingness to pay.
Our approach enables real-time feature-based dynamic pricing strategies. These observations and advantages motivated us to take this specific opportunity for profit maximization.
Q3. With an increase of company profits by 146 percent over a three-month period, it seems your policy would be widely adopted. But what hurdles do electric utility companies face in actually adopting and using smart meters on a large scale?
Our results are very promising; I am excited. We envision that because of its benefits, dynamic pricing of retail electricity, in a similar spirit of our policy, would be widely adopted in the near future.
I am optimistic about the adoption and usage of smart meters on a large scale. In 2019, there were more than 80 million advanced (smart) metering infrastructure installations by the U.S. residential customers. The E.U. targeted to have 72 percent of its electricity end-users install smart meters by the end of 2020. In fact, several European countries, such as Spain and Sweden, are expected to reach a 100 percent smart meter adoption rate in the near future.
The broad installation of smart meters enables dynamic pricing of retail electricity. The energy sector is subject to various regulations; one might think that the regulations could create challenges in the implementation of dynamic pricing policies. However, in many parts of the world, regulatory bodies seem to support this policy. For example, policymakers in the E.U. seem to strongly support dynamic pricing of retail electricity. (The E.U.’s Internal Market for Electricity Directive favors dynamic retail electricity pricing, and requires that in every member state, there is at least one energy supplier that offers a real-time pricing program to customers.)
There is growing interest in the dynamic pricing of retail electricity in the U.S., as well. Several utilities have launched pilot programs, and some of them have implemented a form of dynamic pricing. For example, Commonwealth Edison, which is the largest utility in Illinois, has implemented real-time pricing (hourly pricing), under which residential electricity rates change every hour. In addition, U.S. regulators seem to increasingly support smart meter investments and the design of advanced pricing policies. These are great developments that give more hope about the potentially transformative impact of smart meters.
There are things to consider or hurdles to overcome during the implementation of any innovative business strategy. And this is valid for smart meters, as well. Companies should be well prepared to take advantage of their massive smart-meter data. Automation and data analytics tools should be carefully integrated to business processes before offering a data-driven pricing policy to customers. Utilities should also consider launching customer assistance programs that provide information about smart meters and specifics of their dynamic pricing policies. These programs can increase customer awareness and enhance process transparency. Moreover, with smart meters, protecting data privacy becomes critical for utilities. Thus, utilities might benefit from investing in cybersecurity before launching services or products related to the smart meter technology.
Q4. Thinking about emerging technologies and the growing power of data analytics, do you think these innovations are changing business or just business processes? Looking at utilities in particular, are these forces driving incremental or more radical change?
There are many exciting emerging technologies in the energy sector: smart meters, residential solar panels and solar-plus-storage technology to name a few. Some of these technologies have already changed business processes. As these emerging technologies are adopted more, I expect to see more radical changes in the energy sector. I believe we will see the radical changes not only in business strategies but also in policy making. For example, the rapid deployment of smart meters is likely to give rise to new energy policies that promote the efficient use of energy.
In the long run, I think business model innovation will be the key for the success of utilities. In our paper, we are designing one innovative business program for utilities.