Written by Elizabeth Massey1 and Sarah Valovcin2
Utilities of all sizes are facing a world of systems with more complex technology and greater quantities of data, whether from Advanced Metering Infrastructure (AMI), Geographic Information Systems (GIS), or other platforms. Rather than simply installing new and complex systems and hoping for positive results, those utilities that can properly manage these systems and derive tangible value from the data and for their customers will have a distinct competitive advantage in the future.
The purpose of this project was twofold: 1) to develop a flexible advanced analytics platform that can accept data inputs from any source regardless of vendor, and 2) to run a series of real-world studies using advanced analytics to determine whether a smaller public power utility can benefit from this type of work.
There is a great deal of academic and industry-based literature on the research in AMI and digital data analytics, too vast to be included in this brief article. However, some resources of note include: early diffusion of smart meters in the US electric power industry is explained in (Strong, 2017); an excellent survey on the evolution of AMI technologies can be found at (Yu, 2013); and an overview of smart meter analytics including consumer behavior modeling, data aggregation and usage profiling and forecasting can be found at (Yi Wang, 2020). The US Department of Energy (DoE) has invested significantly in the digital grid infrastructure since 2009. Their report of the Smart Grid Investment Grant (SGIG) program can be found at (US Department of Energy, 2018) and at (Advanced Grid Research - Office of Electricity, US DoE, 2020).
Due to rapid retail load growth across their service territory, in 2015, a Pacific Northwest utility took the visionary position to look at innovations around energy efficiency. In addition, a resource capacity shortage began to surface in the Pacific Northwest of the United States around this time and conventional products for dealing with short-term power supply needs became scarce. As a result, the utility desired to study peak demand reduction through real-world demand savings programs and time of use rate design. A key component of the utility’s vision was to be an early adopter of a Data Analytics as a Service model. However, the utility was hesitant to go it alone on a significant R&D effort and successfully was awarded funding through the American Public Power Associations’ DEED grant program3.
To accomplish the first objective, "develop a flexible advanced analytics platform that can accept data inputs from any source regardless of vendor", the Energy Authority’s Connected Analytics data engineering team needed to understand the scope of requirements from its utility partner including data warehousing, verification, estimation and editing (VEE) functionality, data access, and data types to be collected and stored. The main sources of data included energy usage (kWh), customer information (name, phone number and account status)4 and weather data.
The TEA Connected Analytics Data Platform (CADP) is built using an all-cloud architecture on Microsoft Azure and using a Common Data Model (CDM) provides all the capabilities required by the TEA Connected Analytics team and its clients. Each client’s data is integrated, mapped, and loaded into the CDM. By leveraging a CDM, TEA Connected Analytics Solutions are more efficient and can be leveraged across clients, leading to reduced development and maintenance costs.
Ultimately, the TEA Connected Analytics Data Platform (CADP) total cost of ownership is lower than using third-party vendor solutions, while providing more flexibility and otherwise missing functionality. This aligns with TEA’s vision to be the leader in serving Public Power by providing a low-cost Public Power specific solution to the market.
For the second objective, run a series of real-world studies using advanced analytics to determine value to the utility, a series of use-cases were identified and prioritized. These included such use-cases as the ability of the utility to expand the customer usage profile horizons and overlay kWh time-series data with temperature information, transformer load management and analysis, meter tampering system connectivity mapping and energy efficiency marketing.
Real-world questions that were answered through this project included, “Can we identify customers with the highest potential for energy efficiency savings?” The Connected Analytics team at TEA used hierarchical clustering to group customers with similar household characteristics and to confirm the distribution profiles of daily consumption for customers in each household cluster. To find the truly "win-win" customers, the final customer segments were grouped based on hourly load profiles. Applying the same hierarchical clustering technique resulted in 6 final groupings. Of those 6 groups, 3 were chosen that best matched the utility’s natural morning peak. These 3 groups were then placed into a targeted energy efficiency program.
An early success story in this project resulted in finding a malfunctioning water heater in a customer’s home. The replacement project (as part of the EE program that resulted from this analysis) has been found to save customers an average of $100/year in electricity bills.
The analytics performed on this project has resulted in significant benefits to the utility in terms of showing potential savings opportunities and their access to customers, in other words, “getting a foot in the door” to increase response rates for program participation. This analysis leads to future work including confirming that a customer has indeed performed an energy efficiency measure, quantifying the actual savings resulting from energy efficiency measure and reducing peak demand by designing an optimized time of use rate structure.
- Director, Connected Analytics, The Energy Authority
- Senior Data Scientist, Connected Analytics, The Energy Authority
- More on the APPA DEED grant program can be found here: https://www.publicpower.org/deed-rd-funding
- Due to data privacy compliance, personally identifiable information was kept separately and used only during the approved research.
- Advanced Grid Research - Office of Electricity, US DoE. (2020). AMI in Review - Informing the Conversation. Washington D.C.: US Department of Energy.
- Strong, D. (2017). The Early Diffusion of Smart Meters in the US Electric Power Industry. Greensboro: University of North Carolina. Retrieved 12 23, 2021, from https://osf.io/axfqe/download
- US Department of Energy. (2018). Smart Grid System Report. Washington D.C.: US DoE. Retrieved 12 23, 2021, from https://www.energy.gov/sites/default/files/2019/02/f59/Smart%20Grid%20System%20Report%20November%202018_1.pdf
- Yi Wang, Q. C. (2020). Smart Meter Data Analytics, Electricity Consumer Behavior Modeling, Aggregation, and Forecasting. Singapore: Springer, Singapore. doi: https://doi.org/10.1007/978-981-15-2624-4
- Yu, D. A. (2013). Advanced analytics for harnessing the power of smart meter big data. International Workshop on Intelligent Energy Systems (IWIES). IEEE.
This article was edited by Shafi Khadem