A primer on the use of influence coefficients in building simulation

In order to create a simulation model of a building, it is usually necessary to make a number of assumptions and/or approximations about the building being simulated. Many physical quantities cannot be known precisely when the building is being simulated. For example, the real amount of infiltration is almost never known because it is impossible to predict accurately and difficult to measure. Other examples include quantity and type of internal mass, thermophysical properties of building materials, ground temperatures, and equipment efficiencies. In addition, occupancy often has a significant, but unpredictable influence on the building energy consumption. Such behavior as opening doors, leaving windows open, changing the thermostat settings, leaving shades open or closed and generating heat due to use of appliances all have an impact on building energy consumption. In order to gage the accuracy of a simulation, it is necessary to estimate the relevant significance of the assumptions made. By determining which assumptions have significant impact on the building energy consumption, it is possible to determine where effort should be made to refine the simulation. In addition, potential errors can be estimated. One method of determining the significance of the assumptions made when simulating a building involves the use of influence coefficients. An influence coefficient is the partial derivative of a simulation result with respect to a parameter. An example would be the partial derivative of total building energy consumption with respect to the effective solar transmissivity of a window shade. The simulation result could be any result of interest to the user, such as the total energy consumption, heating or cooling loads, annual energy costs, etc. The parameter could be any assumption that affects the simulation result. This paper describes the use of influence coefficients to estimate the significance of assumptions made in the building simulation process. A method for calculating and nondimensionalizing influence coefficients is presented. Examples are taken from a study comparing building energy performance of manufactured family housing units to conventionally-built family housing units at Fort Irwin, CA. The building simulation tool is the Building Loads Analysis and System Thermodynamics (BLAST) program. (1)