Optimizing energy and water use in data center facilities to achieve energy flexibility, reliability and life cycle cost-effectiveness.
Overview
Researchers at Lawrence Berkeley National Laboratory (LBNL) develop modeling tools that are used by data center designers, operators, and equipment providers to improve and de-risk data center energy, water, and control systems during design and operation. These tools were developed with input from industry practitioners.
Our Work
As modern data centers evolve to meet increasing and dynamic loads, designing and operating them to optimize energy usage while maintaining flexibility and reliability is a complex challenge. Computational tools can generate and evaluate strategies for design and operations, assisting data center designers, builders, and operators to make informed decisions to reduce costs and accelerate timelines. We collaborate with industry partners, universities, and other national labs to develop and validate these computational tools for data centers.
MOSTCOOL (Multi-Objective Simulation Tool for Cooling Optimization and Operational Longevity), developed under the ARPA-E COOLERCHIPS program, is a simulation software toolkit that can be used to optimize the design of data centers, including power and thermal management systems for lower cooling energy demand and lower cost, while maintaining high reliability and availability. The energy modeling capability (cooling systems and waste heat recovery) is enabled by the physics-based EnergyPlus engine.
We developed prototype air-cooled data center energy models using the EnergyPlus simulation engine and OpenStudio SDK. The models cover a small data center using CRAC units for cooling and a large data center using central CRAH air-handlers for cooling. The model assumptions reference ASHRAE standard 90.4 and input from industry practices as well as ASHRAE TC 9.9. The data center prototype models were implemented in the OpenStudio Standards gem as two new building types (small data center and large data center). OpenStudio Standards github repo: github.com/NatLabRockies/openstudio-standards.
LBNL collaborates with diverse Original Equipment Manufacturers, design firms, control providers and Modelica tool providers that have been using the Modelica Buildings Library to enable the full digitalization of data center energy and control system design, installation, and operation. The Modelica Buildings Library enables data-driven data center design, de-risks system integration, and enables real-time operations through the coupled dynamic modeling, simulation, and optimization of air- and liquid-cooled systems, energy storage, waste-heat reuse, geothermal systems, electrical systems and controls, all in one tool based on open standards in partnership with commercial and free Modelica tool providers. By enabling control modeling compliant with ANSI/ASHRAE Standard 231 and structuring design flows through Platform-Based Design, LBNL drives efficient and reliable design and operation for the next generation of digitalized energy infrastructure.
Related Publications & Resources
Digitalized control design and deployment based on ANSI/ASHRAE Standard 231, obc.lbl.gov
Wetter, Michael and Sulzer, Matthias and Nyenhuis, Eric and Hu, Jianjun and Fu, Hongxiang and Patureau, Remi and Dee, Virak and Tech, Andrew, Design, Verify, Deploy: The New Platform-Based Design Paradigm for Large-Scale Energy System Retrofits. dx.doi.org/10.2139/ssrn.6557422
