Development Of An Approximate Construction Time Prediction Models During The Project Planning Phase. A Case Study Of Selected Private Mixed Use And Public Office Building Projects In Addis Ababa.

Time, Cost and quality are key performance indicators of construction projects. However, failure to rely on standard duration prediction model lead towards delay of construction projects, which is currently an observed situation in the construction sector. Therefore, the Key objective of this study is to examine and validate the Bromillow’s Time-cost (BTC) model and the Love et al.’s Time-Floor (LTF) model to estimate contract durations for private mixed use and public office building projects in Addis Ababa. The study also propose an alternative duration prediction model by considering potentially influential project scope factors identified from literature by checking their appropriateness and comparing their prediction performances. The LTF model formulates the project duration in terms of gross floor area and floor numbers, while the BTC model formulates in terms of cost. Research data were collected from grade one consultants for private mixed-use buildings and from Addis Ababa City Administration Construction Bureau (AACACB) and Federal Government Buildings Construction Project Office (FGBCPO) for public office buildings. IBM SPSS statistics 26 and WEKA tool 3.8.5 were used to develop and validate the developed construction duration prediction models. Applying linear regression method, the study developed Bromillow’s Time-cost (BTC), Love et al.’s Time-Floor (LTF) and proposed ‘best-fit model’ indicating that the gross floor area is the sole predictor of duration. The model validation result also shows that BTC is superior model over linear regression models. Further, a multilayer perceptron neural network (MLP - NN) predictive model to the same data was applied to develop construction duration prediction models. As a result, (MLP - NN) model result shows significant improvement of the accuracy of the construction time prediction over linear regression models with mean absolute percentage error of 22.2% in private mixed use and 16.3 % in public office building projects. The practical implications of this study can help stakeholders participating in the construction industry to get multiple benefits such as client satisfaction, efficient and effective use of resources, minimizing occurrence of claims and healthy relationship among stakeholders in the construction business sector.

Get Full Work

Report copyright infringement or plagiarism