An Industry Perspective on Groundwater Flow Model Calibration

The purpose of this paper is to present an industry perspective on the calibration of groundwater flow models in view of the limitations imposed by budget and schedule constraints in engineering practice, considering also planning issues related to the execution of different phases of engineering projects. The majority of groundwater modeling applications in engineering practice fall in one of two categories, models that attempt to explain past behavior in order to design the remediation of a known existing problem, and models for sites of new facilities aimed at predicting changes in the groundwater regime due to the construction of new structures. A typical problem of the first category is the analysis of the migration of contaminant plumes originating from historic sources and the design of groundwater remediation solutions. The second category often includes large new industrial structures, such as new powerplants, where models are used to estimate hydrostatic loads used in the design, or to assess the risk from potential accidental releases of toxic or radioactive liquids that may enter the groundwater system during the life of the facility. Model calibration depends on the number and quality of available data. In engineering practice, a typical site investigation study includes the installation of several monitoring wells, aquifer tests and groundwater quality sampling. For studies in support of the construction of new facilities piezometric levels and water quality are typically monitored over a period of one to two years. Groundwater contamination studies may have much longer records. Aquifer testing is in most cases limited to slug tests. At sites where pumping tests are conducted, they are often limited to one or two tests. At construction sites, other available data often include data from borings drilled as part of geotechnical investigations. Sometimes data from geophysical investigations and laboratory analyses for the distribution coefficients of specific chemical compounds are also available. The starting point for any model is a credible conceptual model of the physical system. Conceptual groundwater models are constructed based on the understanding of the site geology and the interpretation of all available data. Because of the many uncertainties in this process, plausible alternative conceptual models are often considered.