Sensitivity analysis of upscaling prediction of the mass flux at DNAPL contaminated sites
Song Meiyu,Shi Xiaoqing,Kang Xueyuan,Wu Jichun
DOI: https://doi.org/10.19509/j.cnki.dzkq.tb20220262
2023-01-01
Abstract:Dense nonaqueous phase liquid（DNAPL） contamination is a growing problem. To assess the environmental risk of DNAPL-contaminated sites, the mass flux of the dissolved phase（mass flux） in the source zone of DNAPL contamination is often extrapolated using upscaling models. Due to the large number of parameters in the upscaling model and the high cost of investigation, the key parameters in the model need to be screened to guide the design of a reasonable observation data collection scheme for actual contaminated sites. In this paper, a global sensitivity analysis was first conducted on six parameters(mean groundwater velocity q, standardized concentration C 0 /C eq , the mass ratio of ganglia GF 0 , the fraction of the mass flux attributable to the ganglia dissolution f g , and fitting parameters β 1 and β 2 ) in the upscaling model to identify the key parameters, and then a local sensitivity analysis was used to quantify the impact of changes in the key parameters on mass flux prediction. The results showed that the parameters q, C 0 /C eq , GF 0 and f g had a large impact on the mass flux prediction. The sensitivities of q and C 0 /C eq were relatively high throughout the depletion process, while those of GF 0 and f g increased continuously with the depletion process, reaching peaks in the middle and late stages of depletion, respectively. For source zones with different structures, the increase in mass flux was essentially constant when q or C 0 /C eq increased. As the ganglia-to-pool（GTP） mass ratio increased in source zones, its effect on the mass flux prediction continued to increase or decrease when GF 0 or f g increased. Therefore, the investigation needs to focus on q and C 0 /C eq when predicting the mass flux, on GF 0 when reasonably designing the remediation plan of the source zone, and on f g when predicting the lifetime of the source zone. For all structural source zones, q and C 0 /C eq are the most important to investigate, and the investigation cost should be focused on GF 0 for source zones with large GTP and f g for source zones with small GTP.