Algona Municipal Utilities in Iowa has completed a grant – under the American Public Power Association’s Demonstration of Energy & Efficiency Developments (DEED) research and development program – that enables small public power utilities to better utilize Advanced Metering Infrastructure (AMI) data to better understand customer behavior particularly as it pertains to rate classes.
The goal of the project, done in coordination with Iowa State University, was to improve on previous AMI analysis tools by using the unsupervised machine learning tool K-Means to make the tool easier to use while providing insights into customer load characteristics during critical peak periods.
“Many utilities, like Algona Municipal Utilities, lack the funds or staff expertise to analyze AMI data in a meaningful way,” John Bilsten, general manager of Algona Municipal Utilities, said in a statement. “Through the collaboration with Iowa State University, Algona Municipal Utilities and the APPA DEED Program, a simple tool has been created to make this data useful to small electric utilities.”
The current project built upon an earlier AMI analysis tool developed by Algona Municipal Utilities and Iowa State University with the aid of a separate DEED grant. That project enabled Algona Municipal Utilities to extract portions of its AMI data to better characterize and understand the variability of load behavior of different customer classes and the contribution of different rate classes on peak load and energy consumption.
The earlier grant project used, MATLAB (MATrix LABoratory), a language for technical computing that renders problems and solutions into familiar mathematical notation, that allows users to import AMI data, as well as customer class information and feeder identification.
The current project expanded the software tool by building in analyses that can better define customer groups in order to aid small utilities seeking to implement new rate definitions, or to consider reclassifying certain customers on the basis of their consumption patterns. The new tool also provides a check on the accuracy of AMI data being provided by billing and meter data management contractors.
By adding clustering techniques, the new tool can highlight “outlier” customers that do not match their current rate class, such as a rural residential load behaving like an industrial load or a large commercial customer not currently under a demand rate that has a load profile more typical of a demand customer. The tool will essentially confirm whether current rate classes are appropriate, given the load profile clusters of the utility’s customers.
“We now have the tools to make conclusions about customer behavior and the correct distribution of costs among our consumer owners. This will be particularly helpful as we analyze the impact of electric vehicles on our distribution system and rate structure,” Bilsten said.
One of the deliverables of the DEED grant was the development of a software tool using MATLAB, as well as a user manual with application examples. MATLAB Runtime software can run on computers that do not have MATLAB installed so that loyalty-free executables can be distributed to any computer quickly and securely. The source codes of the software tool will be hosted by Iowa State University and available to all DEED members.
DEED will host a webinar on the results of this project later this year with representatives from Algona Municipal Utilities and Iowa State University.