Dissertation > Excellent graduate degree dissertation topics show

The Development of Stochastic Networks Analysis Technique and Its Application in Project Risk Analysis

Author: LvYou
Tutor: LiChunHao
School: Jilin University
Course: Project management
Keywords: stochastic networks GERT integration of multi-variable simulation utility index project risk analysis
CLC: F284
Type: Master's thesis
Year: 2009
Downloads: 209
Quote: 0
Read: Download Dissertation

Abstract

Projects are both the application of traditional engineering and an important vector of modern science and technology. There are a large number of uncertain factors and risks within the construction of projects, and the intrinsic relationship between these risk factors is complex. The cross-impact between various risks factors and external factors makes the risk multilevel and the loss cost by the risk is more and more. As project managers, we will be most concerned about whether or not for us to find an effective project risk management methods or tools to better predict and to avoid uncertainties arising from such risks.There are many types and models of network planning technique used in risk management, and CPM and PERT are frequently used. CPM and PERT are all belong to the scope of positive network in essence, which are inadequate in the system with decision-making risk factors.PERT makes a number of assumptions for the calculating of distribution function and expectations,the variance of random variables in the network. These assumptions include: all project activities are independent of each other, completion time is estimated; critical path can be identified and there is sufficient number of project activities to facilitate the application of central limit theorem; ignore the impact of non-critical path activities on the project period; duration of activities can be described byβdistribution, and useμ=(a+4m+b)/6 andσ2=(b-a)2/36 to estimate the expectation and the variance respectively, in which a, b, m, respectively, are the most pessimistic、the most likely and the most optimistic estimates on the subject period obeyed by the distribution activities.The assumptions of PERT limit the scope of its application, for example, the active time of many project can not be known exactly at the beginning, very few projects can be expected to be completed before that time, especially in large-scale projects. This comes from increasing risk factors partly; in project management practice it is also widely acknowledged a point of view - the early activities have great impact on the duration and project’s quality of a later stage and the loss of most of the time is often cost by rework rather than a specific phase of the issues, which are not reflected in PERT and CPM; the critical path obtained from expectations can not fully represent the total project of network; when the conditions and environment changes, this method can not appropriately reflect the new state, the results used the method of analytic estimates have great difference with actual values.Thus, while the PERT makes certain considerations about probability in the activity time. however, PERT and CPM network models still have great limitations, mainly :First of all, each activity in the network must be achieved, so there is no chance to choose from branchs with probabilities. After the realization of each activity, there is no room of activities-choosing or decision-making.Secondly, it does not allow the existence of any loop in PERT and CPM network, which virtually rules out a very rich part of the feedback.The emergence of Graphical Evaluation and Review Technique (GERT) is a turning point of the development of network technology; it is an extension of Generalized Activity Networks (GAN).In GERT, not only the various parameters (such as time, cost, etc.) of the activities have randomicity, and GERT allows the realization of activities to have randomicity, which greatly enriched the content of network technology and expand its scope of application.GERT stochastic network can contain nodes with different logic characteristics, the extraction point of nodes allows a number of probability branches and it allows loop and self-loop circuit in the network. The period of each activity can select any kind of probability distribution and so on. The analysis algorithm of GERT network and its software system GERT-E were already forming. This marks an important stage in the development of the GERT network. And compared with PERT, CPM, the modeling function and application of GERT have a significant expansion so that it makes the network technology moves from the general project area into the research and development area, from a one-time projects into the bulk and mass production process, and are widely used in stochastic service system.Through the comprehensive analysis of risk management techniques at home and abroad, the writer found that it used the stochastic network technology based on the graphical evaluation technique (GERT) in the economic risk analysis of the project, it combined risk analysis and stochastic network technology together, it has been proved to be a valuable analysis method that using a random network analysis and Monte-Carlo simulation methods for risk estimation and evaluation, it have a certain potential for development and research space. There is a wide range of applications in engineering area and scientific research for the standard GERT network, that is, a stochastic network with single goal.The research method of random network usually based on the basic GERT network algorithm, which is time expectation and variance control of a random network of the single target; the exploration about the impact of the risk of project under multi-objective combined effect is only in the area of the second elements of linear correlation. Even through the establishment of multi-additive variable stochastic network model, it just gets the result of the construction period of the project,the average and variance of the cost at the same time, it is easier than the standard single-element stochastic GERT network solution but there is no essential difference between the two methods. Their elements which have been discussed are still in the range of linear function. In this paper, the writer makes further study based on the standard GERT Network and makes the standard GERT Network change from single-objective controlled model into a stochastic network simulation model of a number of nonlinear related objects controlling at the same time. Through writing the stochastic network program and using multi-attribute utility function, the writer build the risk control system closer to the actual project production model and it offers the decision-makers with a more humane, and practical methods of risk prediction.In this paper, the simulation of the stochastic GERT network is achieved through a computer program. The core of the simulation process is the Monte-Carlo simulation method. For the random network nodes which have many investment projects, the progress, costs, quality is not fixed; there exists a certain probability of occurrence. In this paper, the writer proved the stochastic simulation process through a computer language and solved the randomness problem of the variables through a large number of simulation tests.For the control of integrated risk of the stochastic network, this paper uses the risk analysis model of integration of multi-variable,takes advantage of multi-attribute integrative utility index and the weight which is determined by swing weighting to meet the different risk factors for the preferences of decision makers. Through repeated simulation, the writer find the combination of the project, costs, satisfaction nearest to expected utility of decision-makers. It gives the decision-makers the expected benefit and test the effectiveness of subjective reality at the same time.Finally, this paper takes the risk analysis model of integration of multi-variable for the actual projects and it confirmed that the method has good operability and implementation and it gives a new idea for the integrated control of project risk.In the view that the risk estimation and evaluation of the project is a very complicated problem of system evaluation, in this paper the writer uses the methods of logic reasoning, comparative analysis and system analysis, meanwhile the writer uses management, operations research, mathematical statistics, computer network technology and so on to solve the foundation and realization of the multi-disciplinary stochastic network model so that this paper could provide reference for decisionmakers and makes the decision more reasonable and scientific.

Related Dissertations

CLC: > Economic > Economic planning and management > Infrastructure the economy > Organization and management of infrastructure
© 2012 www.DissertationTopic.Net Mobile