Scientific Computation for Multiscale Modeling and Simulation Problem

Scientific Computation for Multi-scale Modeling and Simulation Problem Statement Key Deliverables § Simulation based design and optimization of whole buildings is feasible, but require domain specific reduced order models and computational algorithms to handle multi-scale and uncertain building system dynamics § Actionable visualization capability for rapid design decision making during design § New computational tools that bridge the gap between single room to whole building systems that can deal with time scale separated by >104, 100’s of rooms & 1, 000 s of parameters § Help identify and understand key dynamical behaviors for modeling and control design, and forensics/diagnostics during operation § Methods for optimal sensor placement and control design and performance verification Optimal Sensing Short term (<3 yrs; using existing computational resources) § Coupled structural and building thermal solver with visualization and analysis tools for design decision support (handle scales relevant to building massing → local indoor environment controls) § Parallel computational tools for sensitivity and adjoint based optimization and control at multiple building scales § Computational tools for simulation of open and closed -loop controlled building system, dealing with spatially extended feedback laws Long term (5 -10 yrs; using new computational resources) § Hierarchy of physics based models with multi-scale and multi-physics at the room, zone and building levels Benefits: § Whole building multi-scale simulation codes with • Improve speed of design exploration (e. g. ability to embedded capability for space sensitivity, uncertainty, explore 10 Xinvariant more lowstructure energy design alternatives) dynamical computations • Improve quality of design decision making (ability to assess robustness of designs and cost-performance sensitivity to design changes) • Re-usable models (improved flexibility for design) • More use of models and reduce deployment effort/cost