Evaluating the performance of an Inexact Newton method with a preconditioner for dynamic building system simulation

Published

December 26, 2021

Author(s)

Zhelun Chen, Jin Wen, Anthony Kearsley, Amanda Pertzborn

Abstract

Efficiently, robustly, and accurately solving systems of nonlinear differential algebraic equations (DAE) for dynamic building system simulation is becoming more important due to the increasing demand to simulate large-scale problems, including the integration of multiple buildings. Currently, many of the tools for dynamic building system simulation employ direct methods to solve simulation DAEs. These methods may fail to converge for stiff problems or when iterates are far from a solution. Moreover, as problem size grows, they are less likely to meet the increased memory requirements associated with large-scale problems. Newton-Krylov methods constitute an interesting option to satisfy the computational needs of large-scale dynamic building system simulation. Here, a preconditioned Newton-Krylov method is applied to a series of test problems, the numerical performance of which is presented and compared to the frequently employed Powell's Hybrid Method.

Citation

Journal of Building Performance Simulation

Volume

15

Issue

1

Pub Type

Journals

Download Paper

https://doi.org/10.1080/19401493.2021.2007285

Local Download

Keywords

Newton-Krylov methodHVAC simulationbuilding systemlarge-scale simulation

Citation

Chen, Z. , Wen, J. , Kearsley, A. and Pertzborn, A. (2021), Evaluating the performance of an Inexact Newton method with a preconditioner for dynamic building system simulation, Journal of Building Performance Simulation, [online], https://doi.org/10.1080/19401493.2021.2007285, https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=931065 (Accessed October 26, 2025)