The implementation of EU Directive 91/271/EEC concerning urban wastewater treatment promoted the construction of new facilities and the introduction of nutrient removal technologies in areas designated as sensitive. The need to build at a rapid pace imposed economically sound approaches for the design of the new infrastructures and the retrofit of the existing ones. These studies relied exclusively on the use of heuristic knowledge and numerical correlations generated from simplified activated sludge models. Hence, some of the resulting wastewater treatment plants (WWTPs) were characterized by a lack of robustness and flexibility, bad controller performance, frequent microbiology-related solids separation problems in the secondary settler, high operating and maintenance costs and/or partial nutrient removal, which made their performance far from optimal. Most of these problems arose because of inadequate design, making the scientific community aware of the crucial importance of the conceptual design stage. Thus, these traditional design approaches should turn into more complex assessment methods in order to conduct integrated assessments taking into account a multiplicity of objectives an hence ensuring a correct plant performance. Despite the importance of this fact only a few methods in the literature addressed the systematic evaluation of conceptual WWTP design alternatives using multiple objectives. Yet, the decisions made during this stage are of paramount importance in determining the future plant structure and operation. The main objective pursued in this thesis targets the development of a systematic conceptual design method for WWTP using multiple objectives, which supports decision making when selecting the most desirable option amongst several generated alternatives. This research work contributes with a modular and evolutionary approach combining techniques from different disciplines such as: a hierarchical decision approach, multicriteria decision analysis, preliminary multiobjective optimization using sensitivity functions, knowledge extraction and data mining techniques, multivariate statistical techniques and uncertainty analysis using Monte Carlo simulations. This is accomplished by dividing the design method into 4 different blocks: (1) hierarchical generation and multicriteria evaluation of the design alternatives, (2) analysis of critical decisions, (3) multivariate analysis and, finally, (4) uncertainty analysis. The first block of the proposed method, supports the conceptual design of WWTP combining a hierarchical decision approach with multicriteria analysis. The hierarchical decision approach breaks down the conceptual design into a number of issues that are easier to analyze and to evaluate while the multicriteria analysis allows the inclusion of different objectives at the same time. Hence, the number of alternatives to evaluate is reduced while the future WWTP design and operation is greatly influenced by environmental, technical, economical and legal aspects. Also, the inclusion of a sensitivity analysis facilitates the study of the variation of the generated alternatives with respect to the relative importance of the objectives. The second block, analysis of critical decisions, is tackled with sensitivity analysis, preliminary multiobjective optimization and knowledge extraction to assist the designer during the selection of the best alternative amongst the most promising alternatives i.e. options with a similar overall degree of satisfaction of the design objectives but with completely different implications for the future plant design and operation. The analysis provides a wider picture of the possible design space and allows the identification of desirable (or undesirable) WWTP design directions in advance.
The third block of the proposed method, involves the application of multivariate statistical techniques to mine the complex multicriteria matrixes obtained during the evaluation of WWTP alternatives. Specifically, the techniques used in this research work are i) cluster analysis, ii) principal component/factor analysis, and iii) discriminant analysis. As a result, there is a significant improvement in the accessibility of the information needed for effective evaluation of WWTP alternatives, yielding more knowledge than the current evaluation methods to finally enhance the comprehension of the whole evaluation process. In the fourth and last block, uncertainty analysis of the different alternatives is further applied. The objective of this tool is to support the decision making when uncertainty on the model parameters used to carry out the analysis of the WWTP alternatives is either included or not. The uncertainty in the model parameters is introduced, i.e input uncertainty, characterising it by probability distributions. Next, Monte Carlo simulations are run to see how those input uncertainties are propagated through the model and affect the different outcomes. Thus, it is possible to study the variation of the overall degree of satisfaction of the design objectives, the contributions of the different objectives in the overall variance to finally analyze the influence of the relative importance of the design objectives during the selection of the alternatives. Thus, in comparison with the traditional approaches the conceptual design method developed in this thesis addresses design/redesign problems with respect to multiple objectives and multiple performance measures. Also, it includes a more reliable decision procedure that shows in a systematic, objective and transparent fashion the rationale way a certain alternative is selected and not the others. The decision procedure provides to the designer/decision maker with the alternative that best fulfils the defined objectives, showing its main advantages and weaknesses, the different correlations between the alternatives and evaluation criteria and dealing with the uncertainty prevailing in some of the model parameters used during the analysis. A number of case studies, selection of biological nitrogen removal process (case study #1), optimization of the setpoints in two control loops (case study #2), redesign to achieve simultaneous organic carbon, nitrogen and phosphorus removal (case study #3) and evaluation of control strategies at plant wide level (case studies #4 and #5), are used to demonstrate the capabilities of the conceptual design method.