Academic Paper·DeepMind·2022
A technical paper by DeepMind, Google, and Trane on controlling commercial-building chiller plants using reinforcement learning (RL). The authors describe BCOOLER, a system that generates safe recommendations for setpoints and equipment configurations based on BMS data, energy-use forecasts, and operating constraints. In real-world A/B experiments at two sites, the system reduced energy consumption by 9% and 13%, respectively, compared with Trane’s heuristic Sequence of Operations (SOO).
Reinforcement LearningAI-based HVAC controlChiller plantsData center cooling
Details →Government Report·Lawrence Berkeley National Laboratory·2016
This Lawrence Berkeley National Laboratory study, supported by the U.S. Department of Energy’s Federal Energy Management Program, estimates energy use by U.S. data centers from 2000 to 2020. The model covers servers, storage systems, networking equipment, and facility infrastructure, using data on the installed base of IT equipment and PUE by facility type. The report compares the current trend with a counterfactual scenario in which 2010 efficiency levels are maintained, and examines the potential of improved management, best practices, and workload migration to hyperscale data centers.
Data center energy consumptionPUEHyperscale data centersEnergy efficiency
Details →