Building Simulation: An International Journal

Article Title

Optimization of variables in air conditioning control systems: Applications of simulations integrating CFD analysis and response factor method


response factor, CFD, optimization, temperature distribution, air conditioning control, thermal environment


The integration of three-dimensional spatial distributions into building simulations is of significant interest, and computational fluid dynamics (CFD) analysis is widely employed in building design processes. For example, based on the experience of architects and engineers, CFD analyses are often conducted under steady boundary conditions to determine the degree of attainment of indoor environments. However, CFD analyses have large calculation costs and cannot be often used for simulations with unsteady boundary conditions such as energy simulations in the building design processes. Thus, we developed a method that calculates sensitivities from heat sources to an arbitrary point in an indoor environment and integrates them into simulations with unsteady boundary conditions. In the proposed method, CFD analysis is employed under steady boundary conditions to calculate the response factors, and the resulting sensitivities are integrated into simulations under unsteady boundary conditions. In the present study, the proposed method was applied to optimize the variables of an air conditioning control system. With our method, temperature changes at a sensor over time are calculated from the time series of air supply temperature. In total, 800 calculations were conducted, and the optimal variables that allow the temperature at the sensor to reach the target value quickly were obtained. Except for the time required to calculate the response factors, the optimization in the present study took only a few seconds. If only CFD analysis was used for the optimization, the calculations would take a year. Thus, calculating the sensitivities via CFD analysis and utilizing the results in simulations is a useful approach for solving optimization problems. Moreover, the proposed method is applicable to simulations that require three-dimensional spatial distributions to enhance the accuracy of the calculation such as energy simulations.


Tsinghua University Press