EN FR
GeoConferences.ca

Simulation of horizontal well depressurization - MODFLOW versus FEFLOW

Karl P. Lawrence, Willy Zawadzki, Ashley Mathai, Jeff Randall, Rob McLaren, Michel Mailloux

In the proceedings of: GeoMontréal 2013: 66th Canadian Geotechnical Conference; 11th joint with IAH-CNC

Session: General Hydrogeology I

ABSTRACT: Vertical wells have been historically used for depressurizing confined aquifers in mining and oil and gas applications. Advanced drilling technologies, however, permit well field optimization by efficient installation of horizontal wells lined by a pre-packed screen. This can enhance water extraction rates due to increased aquifer-well contact provided the aquifer has sufficient yield and the well design and placement is adequate. The potential advantage of horizontal wells is the focus of several ongoing oil and gas projects in Western Canada where detailed hydrogeological modelling is required in support of mine planning. Large scale mining and oil and gas operations usually rely on three-dimensional groundwater flow models to provide forecasting of hydrogeological conditions at various stages of mining. Representing horizontal wells in these models, however, is performed less regularly due to the high degree of grid/mesh refinement required. Groundwater extraction from horizontal wells also produces a lateral flow field in the aquifer (as opposed to a radial flow field for a vertical well) which is generally more difficult to interpret and necessitates the use of three-dimensional models to adequately represent the response. This paper assesses the implementation of horizontal wells in commonly used groundwater flow models - MODFLOW (McDonald et al., 1983), FEFLOW (Diersch, 2009), and HydroGeoSphere (Therrien et al., 2007). Several approaches for implementing these wells, including refined grid discretization, discrete features, and multi-node wells, in multiple groundwater flow packages are compared to benchmark analytical solutions from Saphir (Kappa, 2012). Recommendations for representing horizontal wells in complex flow models are provided. The limitations associated with each of the different approaches are also discussed. MODEL DESCRIPTION A block model is constructed in each of the three platforms (Visual MODFLOW, FEFLOW, HGS) to simulate the response of a 10 m thick confined aquifer. The aquifer is laterally extensive (30 km x 30 km) so boundary effects are not considered and the initial hydraulic head is set to 20 m above the top of the model. The hydraulic conductivity of the model is 7.5E-5 m/s laterally and one order of magnitude lower in the vertical direction. The specific storage is 5E-6 1/m. A three-dimensional representation of the model produced from FEFLOW is shown on Figure 1 (clipped along the centerline). A 500 m horizontal well with 0.1 m radius is simulated one meter from the base of the model in the center of the domain (laterally) using several implementation strategies (Figure 1). A constant rate pumping test is performed using this well for 30 days at 1000 m3/day. The hydraulic head is recorded at both ends of the horizontal well. The simulated drawdown from the groundwater flow models is compared to the analytical response produced using the pressure analysis package Saphir. HORIZONTAL WELL IMPLEMENTATION Two approaches were used to implement the horizontal wells. First, the numerical mesh was refined locally to the diameter of the well (laterally and vertically) and a highly conductive channel was physically implemented along the length of the well. Second, linear features (or discrete features or pipes) with high conductivities were used to simulate pipe flow along a series of grid nodes to represent the horizontal well. The actual implementation and terminology varies across the modelling platforms but the concepts are consistent within each. The grid refinement approach (referred to as channel in subsequent figures) was implemented in each of Visual MODFLOW, FEFLOW, and HGS. The discrete feature approach was implemented in FEFLOW and HGS. A tertiary simulation was conducted in Visual MODFLOW using the multiwell approach in which the rate was divided equally between 100 vertical wells (5 m spacing) positioned along the horizontal well (similar to the process in which the MNW2 package could be used to implement the horizontal well). Figure 1. FEFLOW model and mesh generated for horizontal well simulation (clipped to positive half-space)

Access this article:
Canadian Geotechnical Society members can access to this article, along with all other Canadian Geotechnical Conference proceedings, in the Member Area. Conference proceedings are also available in many libraries.

Cite this article:
Karl P. Lawrence; Willy Zawadzki; Ashley Mathai; Jeff Randall; Rob McLaren; Michel Mailloux (2013) Simulation of horizontal well depressurization - MODFLOW versus FEFLOW in GEO2013. Ottawa, Ontario: Canadian Geotechnical Society.

@article{GeoMon2013Paper691,author = Karl P. Lawrence; Willy Zawadzki; Ashley Mathai; Jeff Randall; Rob McLaren; Michel Mailloux ,title = Simulation of horizontal well depressurization - MODFLOW versus FEFLOW,year = 2013}