A Methodology for Global Sensitivity Analysis of Activated Sludge Models: Case Study with Activated Sludge Model No....

Area of Science:

• Environmental Engineering
• Computational Science
• Mathematical Modeling

Background:

• Activated sludge models are crucial for wastewater treatment but generate complex, time-varying outputs.
• Global Sensitivity Analysis (GSA) is challenging with time-dependent data due to fluctuating indices.
• Functional Principal Component Analysis (fPCA) offers a method to simplify complex, time-series data.

Purpose of the Study:

• To develop and demonstrate a computational approach integrating GSA with fPCA for activated sludge models.
• To address the issue of time-varying GSA indices by transforming time-dependent model responses.
• To evaluate computational factors influencing the performance of the proposed GSA-fPCA methodology.

Main Methods:

• Integrated Global Sensitivity Analysis (GSA) with Functional Principal Component Analysis (fPCA).
• Applied the GSA-fPCA methodology to the Activated Sludge Model No. 3 (ASM3).
• Aggregated time-dependent model response patterns into time-independent functional principal components (PCs).

Main Results:

• The first few functional PCs effectively captured up to 100% of the model's output patterns.
• GSA indices calculated from PC scores revealed inherent parameter sensitivity patterns within ASM3.
• The study evaluated the impact of parameter sampling range, simulation period, basis functions, and system state (batch/continuous).

Conclusions:

• The GSA-fPCA technique successfully removes the time-varying nature of sensitivity indices from dynamical models.
• This approach enhances the efficiency of capturing model response patterns using a reduced set of functional PCs.
• The GSA-fPCA method provides a robust numerical strategy for model analysis and calibration, adaptable to various model complexities and applications.