Masters Theses

Date of Award

12-1995

Degree Type

Thesis

Degree Name

Master of Science

Major

Chemical Engineering

Major Professor

Tse-Wei Wang

Committee Members

R. A. Reich, A. A. Rabbani, R. M. Counce, M. Ally

Abstract

This study presents the development of a dynamic simulator for a biotreatment wastewater bio-treatment plant and an operational and control analysis approach for evaluating potential control strategies for the plant-wide wastewater biotreatment plant patterned after a plant under construction by the DuPont Company in Victoria, Texas. The research first reviews existing modelling techniques and sensor technology. It then reviews concepts presented by the Activated Sludge Model No. 1 and utilizes these concepts in the creation of a dynamic simulator for a wastewater treatment plant using the Simulink™ package of Matlab™ as the programming platform. The dynamic simulator is designed to model a wastewater treatment plant based upon the Bardenpho system.

The two stage Bardenpho system features an anoxic tank, where nitrate is removed by conversion to nitrogen gas, and an aerobic tank, where carbon oxidation and nitrification occur. The output concentrations of interest in this plant are the effluent concentrations of carbonaceous waste, ammonia, and nitrate. The treatment objectives are to reduce the carbonaceous wastes and nitrate in water. The input variables available to control the output variables are the influent concentrations of ammonia and nitrate, the influent flow rate, the sludge age, and the recirculation ratio. (The recirculation ratio is defined as the ratio of the flow of water from the aerobic tank to the anoxic tank divided by the wastewater inlet flow into the plant.) If multiple but decoupled feedback control loops are to be considered, then an important consideration is to choose pairings of input (or manipulated) variables and output (or controlled) variables that will minimize interloop interactions and give acceptable performance characteristics.

This plant was evaluated for the performance and stability characteristics with 2 x 2 and 3 x 3 decoupled PID based feedback control loops in place. These properties include integral stabilizability and integral controllability and ability of the plant to maintain stabilizability and/or controllability when a sensor or actuator fails on-line. The study also includes a controller robustness analysis.

With the three controlled variables and five manipulated variables of interest, ten representative sets of input-output variables were studied. The ten sets of 3 x 3 systems were first arranged for optimum pairings, using the Relative Gain Array method. Of these ten optimum pairings, six were found to be not integral stabilizable, and two of the remaining four sets were not integral stabilizable when a particular critical sensor or actuator failed. The two remaining systems were integral controllable and maintained their integral controllability despite the failure of any sensor or actuator.

These two optimum pairing systems were very similar in configuration. The first set of optimum pairings uses the feed rate to control the carbonaceous waste in the effluent, the influent ammonia concentration to control the effluent nitrate concentration, and the recirculation ratio to control the effluent ammonia concentration. The second optimum set of pairings resembles the first, except the influent nitrate concentration replaces the recirculation ratio in controlling the effluent ammonia concentration.

In using a model to evaluate actual plant performance, there always will be some plant/model mismatch. Several selected parameters were chosen to illustrate the effect of their uncertainty on the robustness of the closed loop controller performance using the first set of optimum pairings of the input-output variables. While several parameters will differ from reality for any given model, or will change over time, evaluating these parameters singly or in pairs should help a control engineer to determine in which parameters changes are most critical to the controllability of the plant in terms of designing a controller.

Analysis approaches such as those presented in this research enhance greatly not only an understanding of the effect of various process input disturbances on the performance of the treatment plant, but also allows for the evaluation of various controller design scenarios as to their expected performance. Integration of upfront control analysis with process design would result in a much more operable plant that would be more resilient against external perturbations and internal process evolutions.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS