Modelling cost and time performance of public building projects in a terror impacted area of Nigeria

Examine the impact of construction cost-and-time-influencing factors on the production performance of public building projects in north eastern Nigeria and whether a predictive model could be devised in assessing this impact. The global poor performances of construction project cost and time, coupled with a dearth of studies into using machine learning systems as artificial neural networks for advanced cost and time impact predictions, has made researching into construction project performance imperative. Moreover, the Boko Haram insurgency in Nigeria's North East geopolitical zone intensified the disruptions of construction site programmes with consequent cost increases. Therefore, prediction and performance measurement tools for construction project cost and time can possibly be developed from the examination of data on initial contract sums, estimated construction duration, final cost, actual construction duration, and the influence of cost and time driving factors on public building projects in north eastern Nigeria. The research objectives include an assessment of the factors influencing the cost and time performance of public building projects in north eastern Nigeria and determination of the cost and time performance of selected public building projects in the study area. Others are the development of models for assessing the impact of cost and time influencing factors on the performance of cost and time in public building projects in the study area. Lastly validation of the developed cost and time performance impact assessment models of public building projects in the study area. A quantitative research approach that employs a questionnaire survey was adopted in sourcing primary and secondary data from purposively sampled construction industry professionals. The study used one-way between-groups Analysis of Variance with a post-hoc test, one-way repeated measures Analysis of Variance with Wilks’ lambda tests, multiple linear regressions (MLR), Factor analysis (FA), and Artificial Neural Network (ANN) to analyze the data collected. This data was about initial contract sums, estimated construction duration, final cost, actual construction duration, and the influence of the identified driving factors on public building projects, completed between 2012 and 2017. The study found that the mean percentage cost overrun of the projects studied decreases from uncomplicated projects to moderately complex projects; and increases from moderately complex to largely complex projects. Also, the mean percentage time overrun decreases with increases in project complexity. The significant cost influencing factors are the inexperience of the contract manager, payment delays to main contractors, unstable foreign exchange, variations to works, and corrupt practices. The time influencing factors include design errors, cash flow problems, payment delays to main contractors, contractors’ improper contract knowledge, and delay in building plans and approval. These factors found in the study area were used to develop MLR and ANN cost and time impact prediction models. The developed ANN impact prediction models were validated and compared using previous similar studies in terms of relative absolute deviations and mean absolute percentage errors. The MLR models, although better than the ANNs in terms of mean absolute percentage errors and relative mean absolute deviations, it yeilded poor explanations of the variances in the dependent variables (impacts or overruns) by the independent variables (multiple influence factors). The alternative ANN impact prediction models’ statistics are: (i) MAPE of the developed cost impact model prediction efficiency was found to be 93.54%. The Rel. MAD of the developed cost impact model was computed to be 1.46, in other words, plus or minus 1.46; (ii) the MAPE of the developed duration impact model iii prediction efficiency is 92.94%. The Rel. MAD of the developed duration impact model computed is 0.85, in other words, plus or minus 0.85. The ANN models compared favourably with previous similar studies in terms of relative absolute deviations and mean absolute percentage errors. The ANN’s capability of learning from examples represents an innovative approach to modelling. The study concludes that the developed ANN cost and time impact prediction models have the potential to aid the construction contractor in predicting the cost outcome of a project during the construction stage, by using the significant cost and time influencing factors. The study recommends that project managers, contractors, quantity surveyors, architects, builders and engineers should place priority on the significant factors identified in this study in their project planning, monitoring and control activities. Also recommended is the conversion of the developed models into Dashboards that construction professionals could use to promptly identify the factors influencing cost and time on construction projects, and to monitor performance.

Grantor

Department of Construction Economics and Management

Year dc:date.issued

2019

Author dc:creator

• Oboirien, Momoh Ohiomah

Advisors dc:contributor.advisor

• Windapo, Abimbola Olukemi
• Odeyinka, Henry Agboola

Handle dc:identifier.uri

http://hdl.handle.net/11427/30448

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/30448