Search papers, labs, and topics across Lattice.
This paper introduces a scalable simulation framework for HVAC systems by coupling physics-informed neural ODEs (PINODEs) with differential-algebraic equation (DAE) solvers. The component-level PINODEs learn heat-exchanger dynamics by predicting conserved quantities and enforcing mass/energy balances, while system-level integration uses DAE solvers to enforce junction constraints. A lightweight corrector network further reduces system-level bias, achieving multi-fold speedups over high-fidelity simulation with low errors.
Physics-informed neural ODEs, when coupled with DAE solvers and a lightweight corrector, can simulate large-scale HVAC systems orders of magnitude faster than traditional methods while maintaining high accuracy.
We present a scalable, data-driven simulation framework for large-scale heating, ventilation, and air conditioning (HVAC) systems that couples physics-informed neural ordinary differential equations (PINODEs) with differential-algebraic equation (DAE) solvers. At the component level, we learn heat-exchanger dynamics using an implicit PINODE formulation that predicts conserved quantities (refrigerant mass $M_r$ and internal energy $E_\text{hx}$) as outputs, enabling physics-informed training via automatic differentiation of mass/energy balances. Stable long-horizon prediction is achieved through gradient-stabilized latent evolution with gated architectures and layer normalization. At the system level, we integrate learned components with DAE solvers (IDA and DASSL) that explicitly enforce junction constraints (pressure equilibrium and mass-flow consistency), and we use Bayesian optimization to tune solver parameters for accuracy--efficiency trade-offs. To reduce residual system-level bias, we introduce a lightweight corrector network trained on short trajectory segments. Across dual-compressor and scaled network studies, the proposed approach attains multi-fold speedups over high-fidelity simulation while keeping errors low (MAPE below a few percent) and scales to systems with up to 32 compressor--condenser pairs.
Stanford HAI