ePerf – 2nd Workshop on Performance and Modelling of Energy Systems
The workshop will take place at 8 am EST (2 pm CET) on Webex conferencing tool
Sustainability is a topic of great importance as it can help to address unprecedented environmental challenges that the world is facing today. This workshop aims to explore how improvements to or new uses of Information and Communication Technology (ICT) can improve the environmental sustainability and performance of smart energy systems in buildings, power grids, and transport systems.
This workshop will bring together leading researchers and practitioners from the Performance community and the Energy Systems and Informatics community to exchange technical ideas and practical experiences on issues related to the design and operation of smart energy systems. It will include invited talks and will serve as a forum for the Performance community to apply their techniques to this emerging and important area.
Workshop day: November 6, 2020
8.00 – 8.05 NYC / 14.00 – 14.05 Milan
Welcome by the organizer
8.05 – 8.40 NYC / 14.05 – 14.40 Milan
Highlight Talk 1: Solar-TK: David Irwin (University of Massachusetts Amherst, US)
Title: A Data-driven Toolkit for Solar PV Performance Modeling and Forecasting
Abstract: Solar energy capacity is continuing to increase. The key challenge with integrating solar into buildings and the electric grid is its high power generation variability, which is a function of many factors, including a site’s location, time, weather, and numerous physical attributes. There has been significant prior work on solar performance modeling and forecasting that infers a site’s current and future solar generation based on these factors. Accurate solar performance models and forecasts are also a prerequisite for conducting a wide range of building and grid energy-efficiency research. Unfortunately, much of the prior work is not accessible to researchers, either because it has not been released as open source, is time-consuming to re-implement, or requires access to proprietary data sources.
To address the problem, we present Solar-TK, a data-driven toolkit for solar performance modeling and forecasting that is simple, extensible, and publicly accessible. Solar-TK’s simple approach models and forecasts a site’s solar output given only its location and a small amount of historical generation data. SolarTK’s extensible design includes a small collection of independent modules that connect together to implement basic modeling and forecasting, while also enabling users to implement new energy analytics. We have released Solar-TK as open source to enable research that requires realistic solar models and forecasts, and to serve as a baseline for comparing new solar modeling and forecasting techniques. We compare Solar-TK’s simple approach with PVlib and show that it yields comparable accuracy. We then present three case studies showing how Solar-TK can advance energy-efficiency research.
8.40 – 9.15 NYC / 14.40 – 15.15 Milan
Highlight Talk 2: Dan Wang (Hong Kong Polytechnic University, Hong Kong)
Title: Chiller AIOps: Experiences Towards Data-driven Industry Operations
Abstract: With the success of AI technologies, recently, we have seen extensive advocates on applying AI technologies into the industry production, operation and maintenance processes, or the so-called Industry 4.0. Landing such a vision into a specific domain sector, however, is not straightforward. In this talk, we present our experiences in the domain sector of central HVAC systems. On the one hand, developing the machine learning models is non-trivial, and it heavily involves interdisciplinary knowledge. On the other hand, developing an analytics system that is portable across diverse buildings is also challenging. We further discuss our experiences with industry companies entering this domain sector. We found that their perspectives differ greatly; this introduces many challenges yet also provides great opportunities.
9.15 – 9.50 NYC / 15.15 – 15.50 Milan
Highlight Talk 3: S. Keshav, (University of Cambridge, UK)
Title: A Power-Efficient Smart Lighting System
Abstract: Lighting load accounts for a significant portion of overall energy consumption in office buildings. To reduce this load, we have designed and built a smart lighting control system that minimizes power consumption by quickly responding to changes in daylight and occupancy, while simultaneously providing personalized lighting comfort to each occupant. Our system measures illuminance and occupancy from sensors located at each workstation. Using an unobtrusive self-calibration process, it estimates the relationship between the dimming level of each bulb and the illuminance at each workstation. Subsequently, an adaptive control algorithm maintains the desired illuminance at work surfaces despite environmental fluctuations by periodically recalculating the power-efficient and comfort-preserving dimming level for each bulb. Based on a realistic deployment of our system, we find that our system quickly responds to changes in occupancy, daylight and user preferences. We also show, through extensive simulations using 7 months of collected daylight and occupancy data, that our system reduces energy consumption by about 40% compared to conventional LED lighting systems.
Joint work with Yerbol Aussat at the University of Waterloo
9.50 – 10.25 NYC / 15.50 – 16.25 Milan
Highlight Talk 4: Minghua Chen (City University of Hong Kong, Hong Kong)
Title: Demand-Aware Ride-Sharing as Network Utility Maximization
Abstract: Ride-sharing is a modern urban-mobility paradigm with tremendous potential in reducing congestion and pollution. Demand-aware design is a promising avenue for addressing a critical challenge in ride-sharing systems, namely joint optimization of request-vehicle assignment and routing for a fleet of vehicles, given the probability distributions of future demands. In this talk, we discuss a probabilistic demand-aware framework to tackle the challenge. The idea is to assign requests to vehicles in a probabilistic manner and post a network utility maximization formulation. It differentiates our work from existing ones and allows us to explore a richer design space to tackle the request-vehicle assignment puzzle with a performance guarantee but still keeping the final solution practically implementable. We focus on maximizing the expected number of passenger pickups. The optimization problem is non-convex, combinatorial, and NP-hard in nature. As a key contribution, we explore the problem structure and propose an approximation of the objective function to develop a dual-subgradient heuristic. We characterize a condition under which the heuristic generates a (1 − 1/e) approximation solution. The solution is simple and scalable, amendable for practical implementation. Simulation results based on real-world traces in Manhattan show that our demand-aware solution improves the passenger pickups by up to 46%, as compared to a conventional demand-oblivious scheme. The results also show that joint optimization at the fleet level leads to 19% more pickups than by separate optimizations at individual vehicles.
This is joint work with Qiulin Lin and Wenjie Xu from CUHK and Xiaojun Lin from Purdue University.
10.25 – 11.00 NYC / 16.25 – 17.00 Milan
Highlight Talk 5: Steven Low (California Institute of Technology, US)
Title: Caltech Adaptive Charging Network Research Portal
Abstract: We summarize a large-scale Adaptive Charging Network (ACN) for electric vehicles (EVs) at Caltech that was developed in collaboration with a Caltech startup PowerFlex and has been operational since 2016. We then describe ongoing work to build an open-source research facility based on ACN that provides (i) real fine-grained EV charging data, (ii) realistic EV charging simulators that are integrated with grid simulators such as OpenDSS, and (iii) a live testbed for testing new EV charging algorithms on real EVs. We present application examples of the ACN Research Portal, including learning user behavior, sizing charging facility, impact of large-scale EV charging on distribution systems, and charging profile classifications. Finally we design a scheme to price demand charge that is simple and efficient.
This is joint work with Zach Lee, Sunash Sarma, Tongxin Li, and John Pang.
Omid Ardakanian, University of Alberta, Edmonton, Canada