Modeling of groudwater flow and contaminant transport in a karstic formation using finite diffference approach

Abstract

Lebanon is considered naturally water rich compared to other countries in the Middle East and North Africa region owing to heavy rainfall and snowfall in winter season and the prevalence of Karstic geology. Though, overexploitation and mismanagement of water resources along with poor sanitation practices have led to water shortage and groundwater contamination. Consequently, there is a need to assess the groundwater flow and contaminant transport in Karstic aquifers in order to have a better understanding of groundwater movement through karst conduits until reaching a certain discharge or abstraction point. The focus in this study is on fast flow regimes or flow through karst conduits and fractures as it governs most groundwater movement in Karstic aquifers. This assessment will help in generating remediation measures promoting safe water practices. In this study, a 2D experimental model, replicating karst conduits, was constructed and was used to assess the variation of hydraulic head throughout the prototype for different discharge nodes and pumping rates. Seven trials were conducted and hydraulic heads at different nodes were recorded. The prototype was also simulated on a finite difference software model, MODFLOW, with same dimensions and input parameters. The flow field generated in each trial was used to calculate hydraulic heads. In addition, the variation of ammonia contaminant concentration and contaminant travel time throughout the system were studied. A dyed contaminated water was injected into the system and water samples were collected from four collection nodes. Four trials were conducted, each with different discharge rate, injection rate, initial head, and initial contaminant concentration. For each trial, contaminant concentration at collection nodes were measured at two time periods using a spectrophotometer and medium range ammonia vials. Additionally, contaminant travel time to reach the collection nodes were recorded for three out of the four trials. The prototype was then simulated on MODFLOW, and the generated flow field in each trial was used along with MT3DMS and MODPATH packages to determine ammonia concentration variation and travel time, respectively. Hydraulic head and contaminant concentration variation, as well as contaminant travel time generated experimentally were compared to simulated ones. Linear trend line and 95% confidence curves were produced for each experiment and outliers were identified. Results showed that experimental values largely reflect numerical ones despite several encountered human and instrumental errors. Additionally, a regression model was generated to mathematically predict the contaminant concentration at any node throughout the prototype without the need to conduct experimental trials. A multiple linear regression equation, with six input variables, that best fits experimental data was generated. The precision of the regression model was validated by a strong coefficient of determination R2. Also, the model was proven to be useful in predicting the assessed value through the F-test. This latter showed higher F-value compared to F-critical and a very low F-distribution meaning that the pattern identified between the set of data is significant. The experimental and numerical models employed in this study along with the generated regression model can provide a powerful tool to predict the contaminant concentration at any location, knowing the contaminant source and discharge/abstraction point characteristics. This will help in identifying potential groundwater pollution and in designing safe water abstraction locations. A case study was conducted in order to assess chemical contaminant transport, in karstic formation, towards a contaminated well. Samples were collected from well in Chadra village and were tested in the laboratory using ASTM standard methods. Testing results showed out of range nitrate concentration value (20.54 mg/L > 10 mg/L). The well was modelled with its surrounding on MODFLOW software. A parametric study was conducted by changing porosity, time period, and well abstraction rate variables in order to determine their effects on contaminant path towards the well. These latter were generated using MODPATH package and showed an inversely proportional relationship with porosity, whereas time period and well abstraction rate showed a proportional relationship with contaminant travel distance. Next step was to validate laboratory testing results. Therefore, the nitrate concentration at the well level was calculated using MT3DMS package and was equal to 25 mg/L. Consequently, nitrate-transport modelling reflected the actual field results to a large extent. The evaluation of chemical contaminant transport towards a well is important to help developing groundwater management guidelines aiming towards preventing water contamination and reducing water usage. In this study, three wells in Chadra village were selected for testing and modeling. Physical, chemical, and bacteriological analysis were conducted on collected samples from each well to determine whether the well is contaminated or not. Then, the three wells were modeled with their surrounding using MODFLOW software and contaminant path lines were generated using MODPATH package. These path lines were used to generate protection measures related to safe well abstraction rate and safe setback distance from source to well. Accordingly, these measures were used to develop operational procedures related to pumping regime, sanitation system, and solid waste disposal. These procedures represent the water management guidelines that should be followed by community members and local authorities, along a transitional phase of 5 years, to ensure water safety.