Improving costing in infrastructure projects to accommodate uncertainties

Determining a reliable estimate for a construction project based on scant information during the early stage is quite challenging. It is all too usual to make incorrect estimates based on vague client needs and desires. Early cost estimate reliability is vital to the success of construction project delivery. It is widely acknowledged that one of the major factors affecting a country's economy growing is the presence of adequate social and economic infrastructure. Construction projects delivery management team therefore needs adequate and robust improvement in cost estimation at the early stage. There is need for holistic view of how the present-day project control and management professionals manage and deliver infrastructure projects to make it viable economically. Early cost estimation used in providing key decision in financing these infrastructure projects are known to be flawed due to inadequate information. This is followed by the worry that industry mandated risk management principles are ineffective in managing uncertainty, especially in complex project environments. Construction projects therefore have routinely overrun their estimates. The research identified that there is no unanimity on the reference point from which contingency estimate is produced at the early stage. Another identified problem is that there is insufficient uncertainty management during the early stage of the project. This thesis advocates the use of system thinking in identifying uncertainty factors during the early stage of project to improve cost estimate. A mixed method approach was used to fulfil the objectives of the study. Initially, semi-structure interviews were conducted to identify uncertainty factors that impact early project cost estimate and the importance of using system thinking in identifying them. Twenty respondents were selected from UK project control and management professionals involved in infrastructure project delivery. 300 questionnaires were distributed to professionals in the UK infrastructure project industry, including client, contractors, and subcontractors and 76 respondents were received. A snow-balling sample technique was used to gather the respective respondents. Their responses were analysed using statistical techniques, and some of the results served as input for the regression model produced in establishing relationship amongst system thinking, need for cognition scale scores and years of experience. Another quantitative study was done using secondary data (cost information) obtained from 31 infrastructure projects in the UK. These costs date was analysed using Generalized linear model and Bayesian hierarchical regression Model to produce 12 predictive models that estimate cost overrun and final cost of a given infrastructure project during the early stage.6 case-study firms were used for the validation The models produced take cognisant of project level random effects to account for uncertainties in parameter estimation which reduces the level of biases in the models. Parameter estimation is based on Markov chain monte carlo (MCMC) algorithms implemented within the stan framework. Models were assessed for convergence and goodness of fit using a constellation of model diagnostics and fit indices. The findings from all the analysis showed that the covariates are independent of the project level random effects and there is inadequate uncertainty management at the early stage. Additionally, the year of experience is independent of the system thinking and need for cognition scale scores. High system thinking scale scores will enable project control and management professionals practice holism efficiently during project cost estimation process at the early stage. The predictive cost estimating model would estimate the final cost and cost overrun of an infrastructure project at the early stage which will be useful in producing an effective Should cost model (SCM) for UK project delivery team. If utilized properly, could be used at the output definition and feasibility stage (GRIP framework) to inform the first business case (strategic outline case for project departments).