Combining Sequential Gaussian Simulation with Linear Regression to Develop Rehabilitation Strategies Using a Hydrometallurgical Process to Simultaneously Remove Metals, PCP, and PCDD/F from a Contaminated Soil

ABSTRACT

In this study, a new approach to predicting the ability of a hydrometallurgical process to simultaneously remove metal(loid)s, pentachlorophenol (PCP), and polychlorodibenzodioxins and furans (PCDD/F) from contaminated soil is developed. The remediation process consisted of attrition and alkaline leaching steps applied for the coarse (> 0.250 mm) and fine (< 0.250 mm) fractions, respectively. First, a contaminant granulometric distribution-CGD model was established from granulo-chemical analyses performed on 5 selected sampling points collected from the contaminated site to estimate the levels of metallic and organic (PCP, PCDD/F) contamination in the coarse (> 0.250 mm) and fine (< 0.250 mm) fractions of the entire sample (24) and reduce the analytical costs. The accuracy of the CGD model for each contaminant in both fractions was then evaluated by cross-validation. The CGD model, sequential Gaussian simulation (SGS), and linear regression analyses were combined to predict the ability of the attrition and leaching processes applied to the coarse (> 0.250 mm) and fine (< 0.250 mm) soil fractions to simultaneously remove As, PCP, and PCDD/F from contaminated soil, respectively. The results showed that the attrition process could effectively remove the contaminants below the regulation standards to allow the industrial use of the rehabilitated site, as the coarse fraction represents an average proportion of 84 ± 2% of the total soil. However, the leaching process was ineffective in decontaminating the fine fraction (< 0.250 mm), which represented an average proportion of 14 ± 1% of the total soil. Based on these results, the most suitable strategy for this site can be established and a methodological reference for similar studies in risk assessment can be provided.

KEYWORDS:

• Contaminated site
• heavy metals
• PCDD/F
• sequential Gaussian simulations
• linear regression
• prediction

Additional information

Funding

This work was supported by the Natural Sciences and Engineering Research Council of Canada and IREQ under grant [RDC 463019-14] and the Canada Research Chairs Program.