Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 94
PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON ENGINEERING COMPUTATIONAL TECHNOLOGY
Edited by:
Paper 96

Using Neural Networks to Assign the Maximum Capital Required Due Date for Construction Projects

E.M. Kassas1, H.H. Mohamed2 and H.H. Massoud1

1College of Engineering and Technology, AASTMT, Alexandria, Egypt
2Construction Engineering Department, Zagazig University, Egypt

Full Bibliographic Reference for this paper
E.M. Kassas, H.H. Mohamed, H.H. Massoud, "Using Neural Networks to Assign the Maximum Capital Required Due Date for Construction Projects", in , (Editors), "Proceedings of the Seventh International Conference on Engineering Computational Technology", Civil-Comp Press, Stirlingshire, UK, Paper 96, 2010. doi:10.4203/ccp.94.96
Keywords: maximum capital required, artificial neural networks, cash flow, prediction, maximum required capital due date.

Summary
A major problem always facing contractors when submitting a tender for a new construction project, is the estimation of the maximum capital required and its due date. The maximum capital required, represents the maximum amount of money which the contactor will need through the project construction period. The maximum required capital due date represents the date at which the maximum required capital will be required through the project construction period. Construction contractors should be able to predict both the maximum required capital and its due date at the project tender stage, where neither the detailed time schedule nor the project cash flow is known. At the early stages of the project no information about the contractor cash flow are known and also no time schedule for that project is available. Based upon this valuable data the contractor makes his own decision on the determination of the financing source for that new project and also decides how to deal with the peak requirement for money for the project construction. It also has a potential effect on the contractor's profit and as a consequence affects the contractor's decision to tender.

Studying previous cash flow models an identification for the most important factors affecting the maximum required capital and its corresponding due date is set. Data regarding these assigned factors are generic data about any new project under investigation and the market conditions through which the project will be executed. These factors were concluded in fifteen factor names: project type, project duration, estimated contract value, advanced payment, time lag, interest rate, mark up, time until first payment, retention, project location, weather condition, safety condition, possible increment in project duration, owner payment delay, inflation.

In this study an artificial neural network (ANN) is used to predict both the maximum required capital and its due date for a new construction project at the tendering stage. The ANN is used in solving problems where a number of input and output cases are available but there is no equation or formula to map between the inputs and outputs. Procedures have been developed that use historical data that has been gathered which relates the maximum required capital and its due date to the factors that influence their values which have already been mentioned. The validity of the proposed neural network model was evaluated by testing the model using data for another sample of past construction projects and comparing the model output with actual values.

Analysis has been undertaken on the cash flow for 42 pipe line projects, 138 industrial projects and 74 building projects constructed in Egypt during the time frame 1/1/2001 to 31/12/2007. Three different nets were developed for the considered construction projects. One for each project type. Five different structures were used for each of the proposed three nets. The tested alternative structures have 5, 10, 15, 20, 25 hidden neurons respectively. The RMS and the absolute difference percentage were calculated for the two output variables using the different five proposed structures for the proposed nets. After the nets training phase validation takes place.

The work presented shows that using the neural network technique is useful and reliable in predicting the maximum required capital and the corresponding due date for a new construction project at its tendering stage. The proposed model gives an average error less than 3.4%. In order to arrive at this accuracy a separate network has been used for each type of project. For the proposed model only one hidden layer was used to solve the problem so a very simple topology model structure was used.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £125 +P&P)