Abstracts – Browse Results

Search or browse again.

Click on the titles below to expand the information about each abstract.
Viewing 1 results ...

Badran, Y (2020) Analysis of construction and stakeholder risks for Public Private Partnership projects in developing countries: a comparative analysis using Artificial Neural Networks to determine the effect of poor stakeholder management and construction risks on the pr, Unpublished PhD Thesis, School of the Built Environment, University of Salford.

  • Type: Thesis
  • Keywords: artificial neural network; Australia; China; constructability; content analysis; coordination; developing countries; evaluation; government; economic growth; Hong Kong; India
  • ISBN/ISSN:
  • URL: http://usir.salford.ac.uk/id/eprint/58124/
  • Abstract:
    There has been a continuously increasing demand for public services and infrastructure all over the world especially in developing countries in order to respond to the rapidly growing population and the targeted economic growth in these countries. Accordingly resorting to the PPP scheme is a way for the governmental authorities to achieve the objectives of better services to the end user in energy, educational, water and wastewater, and transportation projects with the help and expertise of the private sector. While PPP was proven to be successful in several instances, there are also several failure stories where the PPP scheme was used. In order to avoid such problems and due to the complex nature of PPP projects and their extended life span, an adequate risk management technique should be performed for PPP projects to ensure their success. One of the crucial steps linked to risk management is stakeholder management. This research primarily aims to develop a mathematical model that analyzes the expected total effect of risks associated with poor stakeholder management during the construction phase on PPP projects’ schedule based on historical details of previous PPP projects in a comparative study with traditional construction projects using Artificial Neural Networks. In order to develop “the risks checklist” that will be inserted in the model, an extensive literature review of 30 sources was thoroughly studied in order to develop the list of the risks affecting PPP projects. To properly develop a comprehensive list of risks, the journal papers, research and publications that were studied covered the time span between 1998 until 2018. Furthermore, the literature review performed for the sake of developing the risk factors was covering different countries such as: the United Kingdom, Hong Kong, Scotland, China, Australia, India, Indonesia, Singapore, Iran, Malaysia, Thailand, Portugal and South Africa. These countries were chosen to encompass different levels of PPP experience. Accordingly, a comprehensive list of 118 risks was developed. In addition to the ranking and classification of risks into various risks categories, each one of the identified risks was mapped to its corresponding country. The purpose of this step is to determine the critical risks that the literature identified for each country in order to establish a cross-country comparison. From this mapping, it is found that most of the risks affecting PPP projects around the world are political, legal, stakeholder and construction risks. The inadequate PPP experience, lack of support from government, force majeure and permits delays are affecting PPP projects in all the countries included in this research. It is also noticed that risks affecting developed countries such as Hong Kong, China and UK are of similar nature to the risks affecting developing countries. The model was developed using Neural Designer ® Software. This software was used in particular as it is a powerful user-friendly interface able to make complex operations and build predictive models in an intuitive way with a graphical user interface. To build the model, the input variables were the 44 risk factors related to consttruction and stakehoders while the schedule Growth (or total project delay) was used as the target variable. The dataset contains 12 instances (or 12 projects) and was divided into three sets: a. Training comprising 66.7% of the projects (8 traditional projects) b. Selection (testing) comprising 16.7% of the projects (2 traditional projects) c. Validation comprising 16.7 of the projects (2 PPP projects) Once all the dataset information has been set, some analytics were performed in order to check the quality of the data. Model performance was detected using Mean Squared Error (MSE) and Normalized Squared Error (NSE) over the training, testing and validation datasets. Ten trials of the ANN model were performed using different training and testing strategies in order to be able choose the optimum model that has the best learning capabilities and delivering the least possible errors during the training and testing. Based on the different trials output, it is concluded that Model 4 delivers the smallest range of error (MSE and NSE) for training and testing. The architecture of this particular model is: 18 input nodes, three hidden neurons on two layers and one output. It was trained using a logistic function. It is noticed that having the hidden perceptron on two layers improved the model’s performance significantly and decreased errors for both training and testing. After performing training and testing of all models, and in all trials, it was noticed that the error decreased considerably by decreasing the number of input nodes. In order to validate the model’s performance, sensitivity analysis was performed to determine the cause and effect relationship between inputs and outputs of the ANN model. The most significant risk factor is “the lack of coordination” as it is the most important contributor to the model ?s ability to predict total project ?s delay. On the other hand, the least significant risk factor in this case is the “constructability” and the “protection of geological and historical objects”. Comparing the results of this sensitivity analysis to the risk mapping to different countries, it is noticed that the lack of coordination risk is not present in other countries such as Australia, Hong Kong and the UK. Based on the model ?s outcomes, correlations between all input and target variables ranked in descending order based on the best model out of the ten models were calculated. The maximum correlation (0.803336) is yield between the input variable “Delay in resolving contractual dispute” and the target variable “Schedule growth”. 37 risk factors out of the 44 have a high correlation factor (more than 0.1) with the total project ?s delay. Furthermore, a comparison was established between this new ranking and the ranking previously obtained from the literature review based on content analysis and on the ranking obtained from the sensitivity analysis. The following observations were drawn: • Based on the literature review, the material availability risk occupies the first position in terms of the most critical risks. This ranking is similar to a great extent to its ranking based on the correlation calculations according to which this risk occupies the third position. • The “Delay in resolving contractual dispute” occupies the highest rank in terms of correlation with the total project delay based on the ANN model’s outputs. This ranking is also similar to the results of the sensitivity analysis where it occupies the second position in terms of the risks having the highest contribution to the total project’s delay. On the other hand, the same risk is ranked 31st based on the results of the analysis of the literature review. Since the ANN model was based on real case projects, it makes more sense that this particular risk can be of detrimental effect to the project’s completion time. The same goes for the risk “Inadequate negotiation period prior to initiation”. This risk, based on the model’s deliverables, is ranking 11th and 17th in sensitivity analysis and correlation to the total project’s delay while, based on the literature review, is ranking 42nd out of 44. • The “Public opposition” risk is one of the most severe risks facing PPP projects based on the literature review as it occupies the second position based on the various sources taken into account. Nevertheless, based on the sensitivity analysis and on the correlation analysis, this risk occupies the 35th and 44th positions respectively. This difference in ranking can be caused by the relatively small sample size of PPP projects studied in this research. The dataset studied was not encompassing such risk as it was not faced in the projects that were analyzed. However, this does not mean that this risk is not significant especially for PPP projects. • For other risks such as “Constructability”, “staff crisis” and “subjective evaluation”, the literature review and the model deliverables produced very close results. • Based on the literature review, the material availability risk occupies the first position in terms of the most critical risks. This ranking is similar to a great extent to its ranking based on the correlation calculations according to which this risk occupies the third position. On the other hand, the ranking of this same risk is 31 based on the sensitivity analysis in terms of its effect and contribution to the total project delay. The “Delay in resolving contractual dispute” occupies the highest rank in terms of correlation with the total project delay based on the ANN model’s outputs. This ranking is also similar to the results of the sensitivity analysis where it occupies the second position in terms of the risks having the highest contribution to the total project’s delay. The “Public opposition” risk is one of the most severe risks facing PPP projects based on the literature review as it occupies the second position based on the various sources taken into account. Nevertheless, based on the sensitivity analysis and on the correlation analysis, this risk occupies the 35th and 44th positions respectively. A future destination for this study is to provide, in addition to the ANN model determining the contribution of the risks to the overall project delay, a tool assisting the public sector to choose and determine whether the PPP scheme in a particular project is the optimum scheme to use or not.