You are here:
Designing for Environment: A Multi-objective Optimization Framework Under UncertaintyEPA Grant Number: R828128
Title: Designing for Environment: A Multi-objective Optimization Framework Under Uncertainty
Investigators: Diwekar, Urmila M.
Institution: Carnegie Mellon University
EPA Project Officer: Karn, Barbara
Project Period: April 1, 2000 through March 31, 2003
Project Amount: $274,424
RFA: Technology for a Sustainable Environment (1999) RFA Text | Recipients Lists
Research Category: Sustainability , Pollution Prevention/Sustainable Development
Description:Designing for environment requires consideration of several indices of environmental impact, including ozone depletion and global warming potentials, human and aquatic toxicity, and photochemical, oxidation, and acid rain potentials. Current methodologies like the generalized waste reduction algorithm (WAR) provide a first step towards evaluating these impacts. However, to address the issues of accuracy and the relative weights of these impact indices, one must wrestle with the problem of uncertainties. Environmental impacts must also be weighed and balanced against other concerns, such as cost and long term sustainability. These multiple, often conflicting, goals pose a challenging and complex optimization problem, requiring multi-objective optimization under uncertainty. The proposal addresses the problem of quantifying and analyzing the various objectives involved in process design for environments. Towards this goal proposed is a novel multi-objective optimization framework under uncertainty. This framework is based on new and efficient algorithms for multi-objective optimization and for uncertainty analysis. The multi-objective optimization algorithm will provide a set of non-dominant designs where a further improvement for one objective will be at the expense of another. This approach will find a set of potentially optimal designs where trade-offs can be explicitly identified, unlike cost-benefit analysis, which deals with multiple objectives by identifying a single fundamental objective and then converting all the other objectives into this single currency. Our proposed approach is particularly valuable in a multi-objective situation where there are a large number of desirable and important objectives which are not easily translated into dollars (e.g., human health and life, environmental impacts, and plant flexibility). In this project we will develop a novel and efficient multi-objective optimization framework under uncertainty based on the algorithms described above to solve problems where objectives are difficulty to quantify, when there is a lack of data or inaccurate data.
Approach:One of the objectives of this study is to advance the state-of-the-art in the area of optimization under uncertainty so that real world problems of significance can be solved efficiently. This is achieved by (a) improving algorithms for the sampling over uncertain variables, and (b) developing a novel and efficient approach for multi-objective optimization. The efficiency of these new algorithms and the proposed framework will be exploited to tackle large scale real world problem of complex nature including: (1) process design and synthesis problems for continuous processes, and (2) solvent recycling problem in the batch industry. These case studies are identified by discussing with our industrial advisory group members.
Expected Results:The proposed framework will:
- Define and integrate environmental and economic objectives for process design in various simulation environments.
- Be flexible, so as to include new technologies as they mature.
- Quantify the technological and economic risks of new technologies.
- Identify when objectives are in synergy or in conflict and quantitate the trade-offs involved.
- Find designs which have minimal environmental impact and are cost-effective.