Use this URL to cite or link to this record in EThOS:
Title: Operational decision support in the presence of uncertainties
Author: Arsene, Corneliu Titus Cezar
ISNI:       0000 0001 3429 8173
Awarding Body: Nottingham Trent University
Current Institution: Nottingham Trent University
Date of Award: 2004
Availability of Full Text:
Access from EThOS:
Access from Institution:
The purpose of this research project is to investigate the implications of the loop equations formulation of the state estimation procedure for the implementation of decision support (DS) systems in operational control of water networks. The proposed DS system comprises a number of standalone applications which could be grouped in the mathematical modeling and simulation of the water systems and how to deal with the information uncertainty in the decision-making process. Both tasks are essential in order to supply safely, treated water to consumers. The mathematical modeling and simulation foiTns the basis for detailed optimization of the water network operations while the unceifainty based reasoning is used to reduce the complexity of the network system and to increase the credibility of its model. This research attempts to integrate the two aspects of operational decision support into a single computational framework of loop equations. The prototype DS system will be validated using case studies taken from the water industry. The optimal control of water systems is a challenging problem because the models are non-linear and large-scale and measurements are noisy and frequently incomplete. The problem of steady state analysis of water distribution systems is studied in the context of a co-tree flows simulator algorithm that is derived from the basic loop coiTective flows algorithm. It is shown that the co-tree formulation has some inherent advantages over the original formulation due to the use of the spanning trees. This allows a rapid determination of the necessary input data for the simulator (the loop and topological incidence matrixes and the initial flows) as well as the fast calculus of the nodal heads at the end of the simulation. A novel Least Square (LS) state estimator that is suitable for on-line monitoring of the water distribution systems is presented. The state variables are both the loop corrective flows and the variation of nodal demands. It is shown that the input data necessary to build the network equations can be derived from the spanning tree obtained for the co-tree flows simulator and so there is a natural connection between the novel state estimator and the simulator algorithm. In spite of the increased size of the state vector, a satisfactory convergence is obtained through an enhancement in the Jacobian matrix. Furthermore a fine-tuning of the inverse of the tree incidence matrix is carried out in order to avoid the lack of numerical stability characteristic to the nodal heads state estimators. A very efficient and effective loop flows LS state estimator is developed that is tested successfully on realistic water networks. Based on the novel state estimation technique, Confidence Limit Analysis (CLA) algorithms that are quantifying the measurement uncertainty impact on the state estimates, are developed. They include a formulation of an Experimental Sensitivity Matrix method (ESM), a sensitivity matrix method within the loop equations framework and an Error Maximization technique (EM). The performances of these algorithms are assessed in terms of their computational complexity and the accuracy of the results that they produce. It is shown that the computational efficiency and the accuracy of results of the EM method renders it suitable for online decision support applications. Finally, it is shown that the novel state estimation technique and the confidence limits analysis algorithm are connected to a previous developed pattern classification module. The overall system is used for fault detection and identification for a realistic 34-node water network. Both, the "loop-equations based" state estimates and the variations of the nodal demands, together with their confidence limits are used as input data to the classification module in order to decide on the operational status of the 34-node water network. The extensive performance studies for 24 hour of water network operations with particular emphasis on detection and correct location of leakages are earned out.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available