Assessing the use of activated sludge process design guidelines in wastewater treatment plant projects: A methodology based on global sensitivity analysis

Design inputs (wastewater characteristics, operational settings, effluent requirements or safety factors, ...) need to be supplied when using activated sludge process design guidelines (ASPDG) to determine the design outputs (biological reactor volume, the dissolved oxygen demand or the different internal/external recycle flow-rates). The values of the design inputs might have strong effects on the future characteristics of the plant under study. For this reason, there is a need to determine how both design inputs and outputs are linked and how they affect wastewater treatment plant (WWTP) designs. In this paper we assess ASPDG with a methodology based on Monte Carlo (MC) simulations and Global Sensitivity Analysis (GSA). The novelty of this approach relies on working with design input and output ranges instead of single values, identifying the most influential design inputs on the different design outputs and improving the interpretation of the generated results with a set of visualization tools. The variation in these design inputs is attributed to epistemic uncertainty, natural variability as well as operator, owner and regulator decision ranges. Design outputs are calculated by sampling the previously defined input ranges and propagating this variation through the design guideline. Standard regression coefficients (SRC), cluster analysis (CA) and response surfaces (RS) are used to identify/interpret the design inputs that influence the variation on the design outputs the most. The illustrative case study uses the widely recognized Metcalf & Eddy guidelines and presents a didactic design example for an organic carbon (C) and nitrogen (N) removal pre-denitrifying activated sludge plant. Results show that the proposed GSA can satisfactorily decompose the variance of the design outputs (R-2 > 0.7): aerobic (V-AER) and anoxic (V-ANOX) volume, air demand (Q(AIR)) and internal recycle flow rate (Q(INTR)). Response surfaces are proposed to facilitate the visualization of how, when and why the design outputs may change when the most influential design inputs are modified. Finally, it is demonstrated that the proposed method is useful for process engineers providing a regional instead of a local picture of a design problem. (C) 2012 Elsevier Ltd. All rights reserved.