Developing a model for construction contractors pre-qualification in the Gaza Strip and West Bank

El Sawalhi, NIH

Abstract

Contractor pre-qualification is characterised as a multi-criteria problem with
uncertain inputs. The criteria used for pre-qualification includes qualitative and
quantitative information. Owing to the nature of pre-qualification, which
depends on subjective judgements of construction professionals, it becomes an
art rather than a science. Two approaches are found in the literature to model
the contractor's pre-qualification criteria; Linear and non-linear models.
The main aim of this research is to offer a rational method for contractor prequalification
that enables to pre-qualify the contractors who are able to achieve
the client's objectives.
The main question guiding the research is how to be sure that the selected
contractor is able to achieve the client's objectives. It is believed that there is
an indirect relationship between the contractor's attributes and the contractor's
ability to achieve the client's objectives. The time, cost and quality overruns of
a project have been used as indicators to measure the contractor's ability to
achieve client's objectives.
To achieve this aim, the methodologies used included literature review,
questionnaires, surveys, and hypothetical and real-life case studies.
This work suggested improvements to the previous contractor pre-qualification
models by using a hybrid model, combining the merits of Analytical Hierarchy
Process (AHP), Neural Networks (NN) and Genetic Algorithms (GA) in one
consolidated model called the Genetic Neural Network (GNN) model. AHP was
used to establish relative weights of the contractor's pre-qualification criteria;
NN was used as the main processing tool to find a relationship between the
contractor's attributes and his performance. The GA was used to select the
appropriate topology of the network.
The data collected from questionnaires 1 and 2 were utilized to establish
relative weights of contractors attributes. Hypothetical and real-life case studies
from executed projects in the Gaza Strip and West Bank were collected through
structured questionnaires. The actual evaluation of the contractor's attributes and the actual performance of the contractor in these projects in terms of
overrun of time, cost and quality were collected. The weighted attributes were
used as inputs to the GNN model. The corresponding time, cost, and quality
overruns for the same case were fed as outputs to the GNN model in a
supervised learning back propagation neural network. The adopted training and
testing processes to develop a trained model are presented.
The accuracy of the model was investigated using Average Absolute Error
*^
(AAE), Mean Square Error (MSE) and correlation co-efficient (R ). The
factors: AAE; MSE; and R2 showed a very good accuracy when comparing
model prediction with actual real-life cases.
The results revealed that there is a satisfactory relationship between the
contractor attributes and the corresponding performance in terms of contractor's
deviation from the client objectives. The GNN model is able to predict future
contractor performance in terms of time, cost, and quality overruns. Therefore,
the evolved model is able to predict the contractor performance.
Key words: Pre-qualification, Contractors, Neural Networks, Genetic
Algorithm, Model, Contractor Performance.

Citation

El Sawalhi, N. Developing a model for construction contractors pre-qualification in the Gaza Strip and West Bank. (Thesis). Salford : University of Salford

• Thesis Type: Thesis
• Deposit Date: Oct 3, 2012
• Award Date: Jan 1, 2007