الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 113 |  |  |
| | | from | 113 | | |
Abstract
All phases of construction projects include many risks. Those risks lead to serious cost escalation where Water Treatment Plant projects are not an exception. The Egyptian government spent around 4.44 billion Egyptian pounds on the water and sewer systems in 2014. The amount spent on Water Treatment Plants was around 35% of this sum (1.5 billion Egyptian). Water Treatment Plant Projects have significant importance due to their huge budget, tight schedules and obstacles. This leads to high level of risks in these projects. Therefore, the purpose of cost Contingency is to generate a reserve fund that is adequate to cover the ingrained risks in the project’s total budget and completion duration. Its presence removes the negative impact of unforeseen event.
The issue of setting appropriate Contingency is one that often poses difficulties for most Water Treatment Plant Projects as a vital constituent of a project budget is ”Cost Contingency”. Generally, a Contingency is represented as a fixed percentage of project cost. However, it is not appropriate to apply this deterministic method to Water Treatment Plants projects due to its huge variability. This variability makes every Water Treatment Plant a unique one of a type Project.
Developing an artificial neural network (ANN) model that can help contractors, during the tendering stage, to predict the Contingency percentage was the main purpose of the study. An ANN model was constructed, using MATLAB Neural Network Toolbox, to predict the contractor’s Contingency. This model has been trained, tested and validated with real data gathered from more than 80 constructed Water Treatment Plants. The sampling data was divided into three categories; 60% of the total input data was used for training the model, 20% for testing the model and 20% for validating the model. The performance graphs for various training algorithm, error plots between trial test and ANN anticipated values for Contingency percentage, comparison graphs for training, testing and validation (regression plots) have been discussed based on coefficient of correlation, R. The results demonstrate a decent understanding between the actual and ANN anticipated results which strongly support the development of such models for Contingency estimation of Water Treatment Plants projects. The model can be further simulated to assess contractors in Contingency estimation for novice data and provide them with more realistic results than the mere deterministic method.