Numerical Implementation and Computer Simulation of Tracer Experiments in a Physical Aquifer Model

Abstract

The numerical model Sand Model was developed and used to replicate the laboratory sand tank model. The plots of the TBCs for the twelve (12) multilevel observation points within the model domain show a reasonable match of the simulated and observed tracer breakthrough curves. The calculated t value (Paired-t test statistic) for each well is less than the critical value of t ( tcritical = 2.16 for df=13), and the calculated P value for each well is greater than the chosen significance level (P=0.05), therefore the null (no difference) between observed and simulated data sets hypothesis is accepted. The analysis also reveals high to very high correlation between the observed and simulated values (Average R2 = 0.86). It took eight (8) and Twelve (12) hours respectively for the tracer to reach its peak concentrations of 3.81mS/cm and 1.55mS/cm respectively for the middle (observation point (BM) [x,y,z] [70,15,24] and down gradient observation point (EM) [x,y,z] [30,15,24] within the sand tank model. A sensitivity analysis showed that the time required for complete source depletion, was most dependent on the source definition and the hydraulic conductivity K of the porous medium. The 12000mg/l chloride tracer source was almost completely dispersed within 34 hours.

Keywords: Replication, Numerical simulation, Model validation, Solute
transport parameters, Tropical soil