Predicting the Frequency and Cost of Hot and Cold Complaints in Buildings

Recalibration of the Complaint Prediction Model

“…To make analytical predictions, we assume that the three temperature distributions in Figure 1 are stationary (time-invariant statistics) and Gaussian. Federspiel (2000) showed that building temperatures are approximately Gaussian, and Fanger (1972) showed that probit (Gaussian) regression lines adequately predict dissatisfied thermal sensation votes. The standard level-crossing process is one in which a stationary Gaussian process crosses a fixed level.…”

“…It is for these two reasons that complaint levels are modeled as random processes. This methodology could be used for modeling other kinds of complaints, such as air quality complaints, but doing so would have less utility because they are much less frequent (Federspiel 2001). Figure 1 shows a graphical representation of the complaint model.…”

“…This paper describes an evaluation of the accuracy of the complaint model developed by Federspiel (2000) and the results of recalibrating it. The research involved collecting temperature time-series and complaint data from buildings, analyzing the data, assessing the accuracy of the original model, recalibrating the model, assessing the accuracy of the recalibrated model, and demonstrating the model's practical use.…”

“…The next sections contain the results and a discussion of those results and their applicability to design, operations, and thermal comfort standards. Federspiel (2000) proposed a complaint prediction model in which unsolicited thermal sensation complaints are modeled as stochastic temperature alarms. By stochastic, we mean that the complaint levels are random processes.…”

“…This makes it difficult to compare the outputs of these models with other important variables, such as energy use, and limits their application to operations and decision-making. Federspiel (2000) recently proposed a model that produces the Predicted Average Complaint Events (PACE) in an interval of time (days, weeks, months, or annually) based on three statistics of the space temperature in buildings. These statistics are mean space temperature, standard deviation of the space temperature, and standard deviation of the rate of change of the space temperature.…”