Fuzzy approach to construction activity estimation

Past experience has shown that variations in production rate value for the same work item is attributed to a wide range of factors. The relationships between these factors and the production rates are often very complex. It is impossible to describe an exact mathematical causal relationship between the qualitative factors(QF) and production rates. Various subjective approaches have been attempted to quantify the uncertainties contained in these causal relationships. This thesis presents one such approach by adopting a fuzzy set theory in conjunction with a fuzzy rule based system that could improve the quantification of the qualitative factors in estimating construction activity durations and costs. A method to generate a Standard Activity Unit Rate(SAUR) is presented. A construction activity can be defined by combining the Design Breakdown Structure, Trade Breakdown Structure and Work Section Breakdown Structure. By establishing the data structure of an activity, it is possible to synthesis the SAUR from published estimating sources in a systematic way. After the SAUR is defined, it is then used as a standard value from which an appropriate Activity Unit Rate(AUR) can be determined. A proto-type fuzzy rule based system called 'Fuzzy Activity Unit Rate Analyser(FAURA)' was developed to formalise a systematic framework for the QF quantification process in determining the most likely activity duration/cost. The compatibility measurement method proposed by Nafarieh and Keller has been applied as an inference strategy for FAURA. A computer program was developed to implement FAURA using Turbo Prolog. FAURA was tested and analysed by using a hypothetical bricklayer's activity in conjunction with five major QF as the input variables. The results produced by FAURA iii show that it can be applied usefully to overcome many of the problems encountered in the QF quantification process. In addition, the analysis shows that a fuzzy rule base approach provides the means to model and study the variability of AUR. Although the domain problem of this research was in estimation of activity duration/cost, the principles and system presented in this study are not limited to this specific area, and can be applied to a wide range of other disciplines involving uncertainty quantification problems. Further, this research highlights how the existing subjective methods in activity duration/cost estimation can be enhanced by utilising fuzzy set theory and fuzzy logic.