M. Clay Belcher
Bibliographic info:
Building Simulation, Vancouver, Canada, 1989, p. 53-59

Since methods incorporating the time value of money are the only ones that give an accurate picture of life costs of a system, they are the only methods appropriate for the analysis of building lighting systems. However, these procedures (such as the present worth method) and the algorithms for employing them are not without shortcomings. One problem with traditional methods is that they employ single point estimates as input values and result in a single value that is supposed to represent the present worth of the system. There is a strong temptation to adhere tenaciously to this number as the definitive answer without considering sensitivity of the output of the model to variations in input. One well?known technique for simultaneous analysis of the uncertainties of the various inputs to a present worth analysis involves Monte-Carlo simulation. First described in the finance and capital budgeting literature of the late 60s, this sophisticated approach is based on user input concerning the "odds" of a given cost taking on various values. The system being analyzed is repeatedly simulated. In each trial, samples are taken from the theoretical probability distributions for the cost inputs, and the present worth is calculated. The output from the model is then a distribution of how frequently present worth values were computed on each of many intervals. Such distributions contain significant information about variability and hence risk of the proposed investment. For maximum accessibility by lighting design professionals, Monte-Carlo software is needed that is specifically tailored to lighting system analysis and is designed for the PC environment. such characteristics are embodied in the MCSEALS* program developed by the author at the University of Kansas. This paper discusses the advantages of Monte?Carlo simulation for the economic analysis of lighting in buildings and describes the probability theory on which the MCSEALS software is based.