Designing Project Systems in Presence of Variations
Construction projects are dynamic processes that usually face changing environment and requirements. Projects typically need cross boundary social entities to happen, in which teams are working to reach the project goals. These purposeful Inter-Organizational Entities (IOE) are working in a complex system consisting of inter-related set of activities, technologies, and requirements. The dynamic and complex nature of the project systems plus bounded rationality causes mismatch between what is being designed at the first place and what is actually executed. These variations can be very costly, reduce the productivity and detrimental to inter-organizational relationship as well as other issues such as moral hazard and hold up problems. This study started with the idea that although projects are being influenced by many factors, which in many cases are outside of the stakeholders' control, they are still predictable. On the other hand, the high rate of change orders show that there is a need for design improvement. Thus, the main objective of this study is to develop dynamic models for project design which are capable of addressing the temporariness, uniqueness, and variability attributes of projects. The models aim to improve the design practices to increase the project system resilience. At the same time they increase the project flexibility to make it more robust and agile. To show the predictability, this study used data from more than 2600 road construction projects tendered by Indiana Department of Transportation (INDOT) to build six different data mining models. By comparing these six machine learning Methods (Support Vector Machine, Logistic Regression, Multi-Layer Perceptron Neural Network, Decision Tree, K-Nearest Neighbor, and Naive Bayes Classifier) it was shown that the decision tree model has the highest accuracy and F-measure among the models in predicting the change order occurrence. This proves that although projects experience changes, these changes are predictable. In addition, it signifies that fact that projects with higher chances of variation require more attention regarding design review and control as well as agility and robustness consideration. In an attempt to address this issue, this research addresses project systems from two different, but complimentary, outlooks. First one, looks at projects as purposeful systems and model the projects using Functions-Based Systems Engineering (FBSE) method. Architecting the project enterprise to enhance the robustness of IOE during the project life cycle is one of the objectives of this study. The main idea is that any project is a combination of activities. The project enterprise is defining these activities and performing them. In another word the project system activities are the project enterprise functions. During the project life cycle the project enterprise is evolving itself to be able to perform the specific determined functions. These evolutions are captured and modeled using IDEF0 technique. State Based family of methods is another outlook that the project process can be modeled based on. System of Transient Modes (STMs) method is the second outlook which models the projects as states, transitions and modes. This perspective, makes the project design systematic, and it does help the project systems to be more robust and deliver their intended functionality in presence of variation in conditions. This second SE model was developed based on Wymore's system models (∀Z, Z&epsis; DSYSTEMS) and have discrete time scale to be able to designs modes and states of project. To validate the effectiveness of the developed conceptual framework, the model was tested by 48 undergraduate engineering students. The outcomes showe the ability of STMs framework in increasing the clarity of design, improving communication, and enhancing in seeing more details of the project. The outcomes of this research can assist the decision makers to make a more precise prediction. It also makes them able to implement systems engineering models in their practice to improve the design, communication, and documentation quality at different life cycle stages of projects. Also it helps the project enterprise design and its adaptation to changing environments. The field of project automation also can be benefited from this research since, both STMs and FBSE models can be implemented in automated project design and control. The models also can be used as part of the project designers' training programs as they are able to help the novice designers in problem scoping and problem solving.
Kandil, Purdue University.
Educational tests & measurements|Management|Engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our