Evaluation of Groundwater Remediation Technologies Based on Fuzzy Multi-Criteria Decision Analysis Approaches

Hao Wang, Yanpeng Cai, Qian Tan, Yong ZengWater 2017 · DOI

TL;DR

This paper introduces a novel fuzzy multi-criteria decision analysis (FMCDA) framework to objectively evaluate and rank 18 different groundwater remediation technologies for oil contamination. By integrating fuzzy logic with four established MCDA methods and employing mean and Borda ranking for aggregation, the approach aims to provide a more robust and less subjective decision-making tool for selecting the most effective remediation strategies.

What Problem Does This Paper Address?

Oil pollution in groundwater poses a significant threat to human health and ecosystems. Choosing the most appropriate remediation technology from a wide array of options is a complex challenge for decision-makers. Traditional methods often suffer from subjectivity, contingency, and a one-sided perspective. This paper addresses the need for a systematic, comprehensive, and less subjective approach to evaluate and rank these technologies, considering multiple performance indicators and inherent uncertainties in the decision-making process.

Methods

The authors developed a fuzzy multi-criteria decision analysis (FMCDA) system. The core methodology involves:

  1. Indicator Selection: Identifying seven key performance indicators for evaluating remediation technologies, including aquifer depth, permeability, applicable scope and level of pollution, technology maturity, governance cost, and pollutant removal rate.
  2. Fuzzy Transformation: Converting descriptive linguistic variables (e.g., "high," "low," "medium") associated with these indicators into fuzzy sets and then into crisp numerical values using a left-right scoring approach. This step addresses the inherent uncertainty in qualitative assessments.
  3. Multi-Criteria Decision Analysis (MCDA): Applying four distinct MCDA methods to rank the technologies based on the transformed indicator data and assigned weights:
  4. Simple Weighted Addition (SWA): A straightforward method summing weighted normalized index values.
  5. Weighted Product (WP): Multiplies weighted index values, amplifying differences.
  6. Technique for Order Preference by Similarity to Ideal Solution (TOPSIS): Ranks alternatives based on their distance to ideal and worst solutions.
  7. Cooperative Game Theory (CGT): Aims to maximize the geometric distance to the worst scheme, with a caveat that it can exclude schemes with any minimum indicator value.
  8. MCDA Aggregation: Integrating the results from the four MCDA methods using two techniques:
  9. Mean Ranking: Averaging the ranks obtained from each MCDA method.
  10. Borda Ranking: A voting-based method that compares schemes pairwise to determine an overall ranking.

The integration of these methods is intended to mitigate the randomness and potential biases of individual MCDA approaches, leading to a more stable and accurate final ranking.

Datasets

The paper does not utilize pre-existing datasets for training or evaluation in the traditional machine learning sense. Instead, it constructs a dataset based on:

  • Expert Opinion: Data on the comparison of importance between indicators was gathered from 18 domestic experts.
  • Literature Review and Case Study: Information on 18 specific groundwater remediation technologies and their performance characteristics across the seven defined indicators was compiled from existing literature and applied to a case study of the Shengli oil field.
  • Fuzzy Consistent Judgment Matrix: This matrix, derived from expert input and indicator comparisons, provides the weights for each indicator.
  • Indicator Description Matrix: This matrix translates the qualitative descriptions of each technology's performance against each indicator (e.g., "small," "heavier," "applied widely") into mathematical language (numerical values).

Key Insights and Ideas

  • Bridging Qualitative and Quantitative: The paper effectively demonstrates how to translate subjective linguistic assessments of remediation technology performance into quantifiable data using fuzzy logic, a crucial step for objective comparison.
  • Robustness through Ensemble: Combining multiple MCDA methods (SWA, WP, TOPSIS, CGT) and then aggregating their results via mean and Borda ranking significantly enhances the reliability and reduces the subjectivity of the final technology selection.
  • Addressing MCDA Limitations: The authors explicitly acknowledge and aim to overcome the limitations of individual MCDA methods, such as the potential for strong randomness in SWA or the exclusion of potentially good schemes in CGT.
  • Practical Application: The framework is directly applied to a real-world scenario (Shengli oil field), showcasing its utility for practical decision-making in environmental management.

Key Results

The study evaluated 18 groundwater remediation technologies. While the specific ranking varied slightly between the individual MCDA methods, the integrated approach provided a consolidated priority order.

  • Top Performers: Across the integrated methods, technologies like Light Non-Aqueous Phase Liquid Recovery (A5), Surfactant Enhanced Remediation (A18), and Pump and Treat (A4) consistently appeared among the top-ranked options.
  • Integrated Ranking: The final consolidated ranking, considering both mean and Borda rankings, placed A5, A18, and A4 at the forefront. Technologies like Enhanced Bioremediation (A7) and In Situ Chemical Oxidation (A9) also showed strong performance.
  • CGT's Distinctive Behavior: The CGT method showed a notable difference in ranking, attributed to its characteristic of excluding schemes with any minimum indicator value, highlighting its conservative nature.

Limitations and Future Work

  • Subjectivity in Expert Input: While fuzzy logic helps, the initial indicator weights are still derived from expert opinions, which can introduce some level of subjectivity.
  • Data Availability: The accuracy of the evaluation relies on the quality and availability of data for each indicator for every technology.
  • Scope of Indicators: The chosen seven indicators might not capture all nuances of remediation technology performance.
  • Future Work: The authors suggest that this FMCDA framework can be applied to other decision-making problems involving multiple criteria and uncertainties. Further research could explore incorporating more sophisticated fuzzy logic techniques or expanding the set of evaluation indicators.

Summary updated

Abstract

Petroleum is an essential resource for the development of society and its production is huge. There is a great risk of leakage of oil during production, refining, and transportation. After entering the environment, the oil pollutants will be a great threat to the environment and may endanger human health. Therefore, it is very important to remediate oil pollution in the subsurface. However, it is necessary to choose the appropriate remediation technology. In this paper, 18 technologies are evaluated through constructing a parameter matrix with each technology and seven performance indicators, and a comprehensive analysis model is presented. In this model, four MCDA methods are used. They are SWA (Simple Weighted Addition Method), WP (Weighted Product Method), CGT (Cooperative Game Theory), and TOPSIS (Technique for Order Preference by Similarity to Ideal Solution). Mean ranking and Borda ranking methods are used to integrate the results of SWA, WP, CGT, and TOPSIS. Then two selection priorities of each method (mean ranking and Borda ranking) are obtained. The model is proposed to help decide the best choice of remediation technologies. It can effectively reduce contingency, subjectivity, one-sidedness of the traditional methods and provide scientific reference for effective decision-making.
BibTeX
@inproceedings{wang2017evaluation,
  author = {Hao Wang and Yanpeng Cai and Qian Tan and Yong Zeng},
  title = {{Evaluation of Groundwater Remediation Technologies Based on Fuzzy Multi-Criteria Decision Analysis Approaches}},
  booktitle = {Water},
  year = {2017},
  doi = {10.3390/w9060443},
}
Figure