A Pseudo-Transient Method for Modeling Flow Networks

A flow network is a modeling technique often used to model fluid flow in buildings and HVAC systems. This technique allows for fluid flows and pressures to be simulated without resorting to a full CFD analysis. A flow network is simulated by applying the steady-state conservation of mass principle at discrete points called nodes to create a system of nonlinear equations which can be solved by a variety of methods. Most solvers for building simulation flow networks are based on some derivative of Newton’s method. These solvers can show fast convergence when good initial guesses are provided but can be slow or divergent otherwise, especially in large networks. This paper introduces a new fictitious transient term to the nodal conservation of mass equations which allows for the equations to be decoupled and solved using explicit methods. This effectively removes the need for a multivariable nonlinear solver when simulating flow networks. The effects of a single model input parameter on stability and accuracy was investigated. Test cases were used to compare the new modeling technique to a traditional steady-state model using a Levenberg Marquardt solver. The results show that the new method creates almost no loss in accuracy while greatly improving the simulation time, especially for shorter time steps and larger models.