Graduation Year

2022

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Chemical Engineering

Major Professor

Aydin K. Sunol, Ph.D.

Committee Member

Scott W. Campbell, Ph.D.

Committee Member

John N. Kuhn, Ph.D.

Committee Member

Andres E. Tejada-Martinez, Ph.D.

Committee Member

Jeffrey A. Cunningham, Ph.D.

Committee Member

Gita T. Iranipour, Ph.D.

Keywords

Activated Sludge Process, Biological Nutrient Removal, Global Sensitivity Analysis, Model Calibration, Parameter Estimation

Abstract

The subareas of process systems engineering including process modeling, simulation, and optimization are becoming progressively substantial for understanding processes, making decisions, improving efficiency, and complying with stricter environmental and safety legislation. Additionally, the availability of user-friendly flowsheet simulators and computational resources with the current trend of data collection and analysis has allowed for a more accurate representation of complex processes such as wastewater treatment systems. Despite all the progress in the areas of process modeling and simulation, the first-principle mechanistic models for wastewater treatment processes are at best employed during the design stage, and therefore, not used optimally for day-to-day operational decision-making. However, for successful deployment of mechanistic models during the operational stage, the challenge that still remains is the development of robust calibrated models while accounting for process and operational uncertainty. Furthermore, wastewater treatment facilities can benefit from process optimization that allows for energy savings and improved effluent quality. Nonetheless, optimizing day-to-day operation can only be achieved through model-based optimization. Therefore, the overall objective of this research is to develop a generic framework for integrating dynamic process modeling with optimization. The overall objective can be summarized through two specific inter-related objectives. The first objective is to develop a framework for calibrating complex process models with numerous uncertain input factors while the second objective is to develop a framework for the deployment of the calibrated models for process optimization.

The developed approach in this work systematically addresses the challenges faced in the development and deployment of process models for advanced decision-making and process optimization. The integrated approach also aims at improving the understanding of the widely utilized mathematical models as well as quantifying the uncertainties related to the numerous model input factors. Although the application of the developed approach presented in this work is for wastewater treatment processes, the approach can also be utilized for complex chemical processes. Therefore, the developed approach serves as a framework for the integration of process engineering and optimization fields.

In order to achieve the overall objective of this work, a case study is utilized for one of the wastewater treatment facilities located in Hillsborough County, Florida. The integrated approach involves four main levels, process modeling, sensitivity analysis, parameter estimation, and process optimization. Each level of the approach has its own workflow and aims at addressing the challenges involved with that particular level. The output from each level serves as an input to the next level.

In the first level (process modeling) of the overall approach, defining the model structure and analyzing the collected data play a significant role in determining the overall performance of the model. Two different wastewater process modeling suites (BioWin and GPS-X) provided comparable results under controlled conditions. However, to accurately match facility performance, finer tuning of uncertain input factors is required. The second (sensitivity analysis) and third (parameter estimation) levels concluded that advanced techniques such as global sensitivity methods and optimization algorithms allow for effective model calibration and validation. The mathematical models used for the modeling of wastewater treatment processes exhibit highly non-linear behavior. However, based on the results from the second level (sensitivity analysis), some of the most influential input factors included maximum specific growth rates of heterotrophs, ammonia oxidizers and phosphorus accumulators, decay rates of heterotrophs and ammonia oxidizers, reduction factor for denitrification on nitrite, and rate constant storage for polyphosphate. The most influential factors allow for effective model calibration and validation as demonstrated in the third level (parameter estimation) of the overall approach.

The fourth and final level (process optimization) of the overall approach concluded that both steady-state and dynamic model-based optimization allows for significant operating cost reduction while improving the effluent quality. The dynamic model-based optimization allows for about 8.45% of daily cost reduction during the winter months and about 15.54% during the summer months. This accounts for about $40,000 savings in addition to $45,000 savings annually through a complete reduction of alum usage. Furthermore, the model-based analysis conducted for recommended optimum operating policies concluded that operating at lower dissolved oxygen concentrations increases the efficiency of simultaneous nitrification and denitrification resulting in improved effluent quality.

Download

Share

COinS