A model for evaluating the effects of weather on construction projects

The thesis describes a study of the effects of weather on construction projects using a simulation model. The model uses project data supplied in the form of a network and historical weather data records to evaluate the effects of the weather elements—temperature, rainfall, windchill and windspeed—on the progress of construction work, using threshold limits specified for weather-sensitive activities. The overall effects of weather are measured by the losses in project production time. Thus by simulating a project over a long period of weather it is possible to obtain results which reasonably reflect the effects of weather on the project.The overall aim of the research was to develop a simulation model to study the causes of lost production time on construction projects. Historical weather data for 20 locations were obtained from the Meteorological Office, Bracknell, Berkshire. The computer program developed to perform these simulations has been named PROWESS, an acronym of PROject Weather Simulation System. The weather data is read from magnetic data tape and stored in the computer until required by the program. The project network is then analysed, and each activity is checked for sensitivity for that element. Where delays occur the project-duration is affected and the lost production time is recorded. Results for 16-year periods of simulation are presented in tabular form classifying the characteristics of the distribution of results for each project type, duration, start date at a location for a specific weather element. All losses in production time are analysed as percentages of the estimated basic values. The range of losses are interpreted to produce isograms for simple practical application. The results show that changes in project duration caused variations in lost production time on all project types. Project type was observed to have a marked influence on production losses. Start date had a very noticeable effect on short projects, starts in October and January produced mean percentage production losses exceeding 20% on concrete type projects at some locations while the corresponding values for April and July were much lower, and losses at the April start date were generally below 5%. The location had a marked effect on all projects. Attempts to forecast from a 16-year period of simulation were made by means of the Box-Jenkins model. The accuracy was only slightly better than adopting estimates based on the 95% confidence interval around the mean value of production losses obtained from sixteen years of simulation with the PROWESS method.