Solar industry is at a critical point in its lifetime. The industry is experiencing exponential growth (residential solar market doubled in 2013). At the same time, as the large-scale deployment of residential solar systems was triggered by financial incentives starting in 2007, a large number of systems are entering their preventive and corrective maintenance periods. Typical residential solar contracts require solar companies to operate and maintain the systems for up to 20 years. With hundreds of thousands of systems distributed across several states and climates, the role of data in the operation of solar networks is becoming increasingly important.
This presentation explains data-driven operation of one of the largest residential solar fleets in the United States. The focus is on developing the data infrastructure and analytical methods to quickly and proactively identify operational issues and their root cause. The data-driven insights are fed back into operations, customer relations management, sales, marketing, product, and design teams. We demonstrate how advanced statistical methods can be deployed to predict production issues and hypothesize their root causes. In particular, we explain a novel method to estimate the long-term, gradual decrease in solar systems productivity, known as performance degradation. The process involves a full cycle of data collection, integration, hypotheses setup, field experiments, calibration, and visualization.
The audience of this talk includes data scientists, operations analysts, investors, and executives of large, distributed renewable energy networks.
Amir is the first data scientist at Sunrun, where he leads performance analytics on one of the nation's largest distributed renewable energy networks. He develops statistical methods and machine learning algorithms to continuously monitor and optimize the operation of the network to maximize production and happiness of solar customers. Prior to joining Sunrun, Amir was a doctoral student in Engineering at Stanford University, where he worked on high-profile projects such as the ARPA-E project and PG&E's smart meter initiative. Amir's research involved statistical analysis of smart meter data and household profiles to rank buildings energy efficiency and identify the drivers of energy efficiency among households. Amir passionately believes that data can help better manage our energy supply and demand. He works on data-driven solutions to reduce the costs of operating distributed energy systems, improve service quality, increase customer engagement and happiness, and increase the sustainability of energy networks.
Amir's expertise and experiences span across statistics, machine learning, management of distributed energy systems, and behavioral analysis of energy consumers. He has a Masters in Construction Management from Stanford and a B.Sc. in Civil Engineering from Sharif University of Technology. He has experience working on large public projects as well as policy making in energy and transportation sectors, both in the United States and in developing countries. When not working, Amir enjoys mountaineering and landscape photography.