07-10, 16:00–16:10 (Europe/Amsterdam), Else (1.3)
JuliaSim HVAC provides a comprehensive suite (based upon ModelingToolkit.jl) for the modeling and simulation of complex thermofluid systems via a library of pre-built component and refrigerant models that connect to advanced solvers that are customized to system behavior and are composable with the SciML ecosystem. We demonstrate typical workflows of the HVAC engineer: Parameter Estimation, Design Optimization, Control and Machine-Learning based acceleration.
Currently, industrial HVAC modelers use disparate tools for each step such as modeling and simulation, calibration, design optimization, control design, and machine learning. We provide an alternative approach leveraging the JuliaSim platform that enables all of these and more activities to be completed in one environment. The JuliaSim HVAC package provides the following features:
-
Industry-grade pre-built components:
• Configurable components for Tube-Fin Heat Exchangers, Compressors, Valves,
Fans, Conditioned spaces and Pipes
• Drag and drop in GUI for rapid prototyping -
Thermodynamic property models:
• Spline-based thermodynamic property models for several refrigerants
(such as R32, R1234YF, R290, R152a, R134a, R410A, R717)
• Dry Air and Moist Air -
Robust Solvers:
• Robust solvers that handle stiff nonlinear discontinuous dynamics
• Specialized DAE Initialization routines for large-scale models. -
Composability with Machine Learning Workflows:
• Integrate with the Model Optimizer tool for automated model calibration unleashing the power of
automatic differentiation of the simulator
• Leverage the Digital Echo tool for model acceleration and smoothing out of irrelevant dynamics using neural surrogates. -
Connect to Controls:
• Integrate with the Control packages for PID, Linear and Nonlinear Model Predictive Control (MPC).
We demonstrate some of these workflows highlighting the advantages of performing these in a Julia environment versus the current state-of-the-art.
Software Engineer - Simulation, Control and Optimization at JuliaHub