Abstract:
Due to demands caused by climate change, the energy sector is undergoing a rapid transition. Energy systems for buildings and communities need to become decarbonized, grid-responsive, resilient, and adaptive to changes in usage, technology options, and markets. This leads to increased complexity in their design and operation. A successful energy transition requires that this new complexity can be tamed. After laying out the new challenges, we will present recent progress on new generation computational tools for building and district energy and control systems. We will also present new tool chains that allow for rapid system-level prototyping, model-based design flow and digitization, ranging from design to installation and operation. We will close with a discussion about what foundation should be built to meet design and operation challenges of decarbonized energy systems.
Bio:
Michael Wetter is a Senior Scientist at the Simulation Research Group at Lawrence Berkeley National Laboratory (LBNL). His research includes integrating building performance simulation tools into the research process, as well as their use for design and operation. He is leading the development of Spawn of EnergyPlus, a next-generation simulation engine for building and district energy and control systems, OpenBuildingControl, a project that digitizes the control delivery process, and the Modelica Buildings Library, the largest Modelica library for building energy and control systems. He has also been developing the Building Controls Virtual Test Bed software for co-simulation and model-based operation, co-simulation tools based on the Functional Mockup Interface standard and the GenOpt optimization program. He is the co-operating agent of IBPSA Project 1 and was co-operating agent of IEA EBC Annex 60, two multinational collaborations that develop new generation computational tools for buildings and community energy systems between 2013 and 2022. Prior to joining LBNL, he led the development of building system models at the United Technologies Research Center (UTRC). He did his dissertation at the University of California at Berkeley and at LBNL, where he created the GenOpt optimization program and the BuildOpt building simulation program and where he developed the first building energy optimization technique that provably converges to the optimal building design. He is a recipient of the bi-annual Outstanding Young Contributor Award of IBPSA and of the bi-annual Distinguished Achievements in Building Simulation Award of IBPSA-USA. He is the Chair of the College of Fellows of IBPSA, an IBPSA Fellow, and a member of the Board of Directors of the Modelica North America Users’ Group. He was Treasurer of IBPSA and President of IBPSA-USA.
Summary:
Urban challenges for energy transition
Cities are trying to transition to green technologies
But the transition can be material-intensive: human toll
Approach: deploy interoperable intelligent monitoring and management technologies
Focus: building efficiency
Different technologies applicable to different scales (room, building, neighborhood, city)
Very challenging to manage the complexity of the energy system transformation
Multi-energy carrier systems (electrical, gas, hydrogen, thermal)
Enlarged tech mix
Buildings are active nodes in the energy system (produce/consume energy, respond to guidance)
Constraints on system design
Turn on system intermittently only when it is needed
But many devices are most efficient when operating steadily at low rate
Tradeoff between size of energy storage and efficiency of storing energy
Physically larger battery can be loaded more efficiently (heat up large water tank to 40 deg C vs heat up small water tank to 90 deg C) but use up living space
Storing more power enables more time shifting of when battery is used. Less efficient storage process vs use during times when dirty power is available on grid.
Major challenge:
Current and future power supply/demand/control systems are very complex and multi-disciplinary
Need a new modeling/control workflow to design and couple models and then operate them
Math: Languages that allow reasoning
Standards: Compilers that generate specialization
Abstractions: Platforms that allow design reuse
Using Modelica modeling system
Have created a large Modelica library of power system components
Air-based HVAC,
Hydronic heating
Chillder plants
Solar thermal
Air flow
Room heat transfer
Room air flow
Electrical systems
Building envelope modeling is not as good a fit for Modelica
EnergyPlus used for building envelope
Modelica-based building control of EnergyPlus
Digitizes control delivery process
ASHRAE digital control algorithm standard
Built a translator from control standard to Modelica specification
Functional Mock-up Interface
Standard for exchanging simulators behind a single API
Approach makes it possible to create models of very large systems with heterogeneous components
Same house or block in different power grids
UC Merced use-case:
Conventional control of building chilled water plan charges storing tanks at night when power is not used
Optimal control algorithm charges during the day when solar power is available
BOPTEST: Building Operations Testing Framework
Compares different control approaches
Enables used of direct comparison or reinforcement learning
Compare different window technologies in your building
Thermal networks in Geneva
Expanded incrementally in modular fashion
Composed of different components/technologies
Need to be modeled modularly as well
Collaboration with Sidewalk labs on Toronto waterfront
Modeled many different designs for power/heating systems
Were able to optimize electricity use across many different designs
Optimization space is too complex for reinforcement learning but constraint optimization has been used to find good designs
Reducing complexity of design process via Platform-Based design
Hierarchy of spatial abstractions: city, block, building
Library of templates at all these abstractions