Simulation of horizontal well performance using Visual MODFLOW

Environ Earth Sci (2013) 68:1119 1126 DOI 10.1007/s12665-012-1813-x ORIGINAL ARTICLE Simulation of horizontal well performance using Visual MODFLOW Wan Mohd Zamri W. Ismail Ismail Yusoff Bahaa-eldin E. A. Rahim Received: 15 August 2011 / Accepted: 19 June 2012 / Published online: 7 July 2012 Ó Springer-Verlag 2012 Abstract A proposed horizontal well or radial collector well installation in shallow aquifers to enhance water withdrawal rates in Pintu Geng well field in Kelantan, Malaysia was simulated using the Drainage Package of MODFLOW groundwater model. The modelling exercise aimed at identifying an optimum pumping rate that would safely achieve the desired drawdown of less than 2 m in an area of 300 m radius surrounding the Pintu Geng horizontal collector well. The model also would serve as a basis for the design of the horizontal well components. High degree of grid refinement for the well location is needed to simulate the real field installation. However, for the purpose of designing water withdrawal systems, it is important to obtain the correct production rate of these wells for a given drawdown. A transient groundwater flow model was calibrated and validated with few assumptions of the horizontal well hydraulic properties. The model demonstrates that under natural flow condition at -3 m depth, the six collectors (drains) tap a volume of 19,200 43,700 m 3 /day. A steady-state model was also developed to predict the capture zone delineation. Attention is also given to the impact of the well installation to the surrounding 300 m radius by inspecting the degree of the drawdown. Keywords Numerical model Groundwater Horizontal well Drain W. M. Z. W. Ismail Air Kelantan Sdn. Bhd., Tingkat 5, Bang. Perbadanan Menteri Besar Kelantan, Jln Kuala Krai, 15050 Kota Bharu, Kelantan, Malaysia I. Yusoff (&) B. E. A. Rahim Department of Geology, University of Malaya, 50603 Kuala Lumpur, Malaysia e-mail: ismaily70@um.edu.my Introduction Water quality improvements from riverbank filtration (RBF) have made it a desirable alternative to the traditional use of river water for water supply. Many water supply utilities are turning to RBF because of savings in treatment costs compared to the treatment of raw river water. Groundwater is typically a higher quality source than surface water, and usually requires less treatment than surface waters. Groundwater is naturally renewed through processes such as infiltration of precipitation and recharge from surface water sources into aquifers. However, in many areas, groundwater is not sufficient quantities to meet the water demand. Water from horizontal collector well (HCW) could be described as a hybrid source between surface water and groundwater. A typical RBF system consisting of one or a series of HCWs located adjacent to a river in an alluvial aquifer that is hydraulically connected to that river, that would deliver water in large quantities is the answer to the availability problem. The HCW collect water along lengths of screens at a low entrance velocity and takes advantage of the filtration capabilities of the naturally occurring alluvium. The collected water flows by gravity to the caisson where it is pumped to a water treatment plant, for example the Pintu Geng Water Treatment Plant, for minimal treatment as required by the law for potable consumption. A typical construction of HCW is illustrated in Fig. 1. This paper demonstrates the simulation of the horizontal collector well (HCW) or radial well performance using the Drainage Package of Visual MODFLOW for the groundwater modelling exercise, aimed at identifying an optimum pumping rate that is sustainable in a desired drawdown condition of less than 2 m in an area of 300 m radius surrounding the Pintu Geng horizontal collector well

1120 Environ Earth Sci (2013) 68:1119 1126 Fig. 1 Typical horizontal collector well construction (PGHCW). The calibrated and validated model would serve as a basis for the design of the horizontal well components. Review of hydrogeological data and conceptual model development The study area The area to be modelled for this project is in North Kelantan and the conceptual model of North Kelantan groundwater aquifer was developed as shown in Fig. 2. The regional conceptual model was developed for the area with few assumptions on the aquifer layers and hydraulic properties. The model boundary condition has been extended to the north and east of the well field in Pintu Geng to study the regional effect of the proposed horizontal well installation and was calibrated and validated against the measured water levels in field. The conceptual model A conceptual model is a pictorial representation of the groundwater flow system, frequently in the form of a block diagram or a cross section (Anderson and Woessner 1992). For the present study, there are few available conceptual models of the aquifer system that can be found in Saim (1997) and Mohamad Faizal (2002). The whole aquifer system may consist of few aquifer layers. As for the horizontal wells installation, the proposed model only models the first layer aquifer (the second layer from the ground surface) as illustrated in Fig. 3. The confining layer of silty clay above the first layer sand aquifer is also being modelled. The aquifer parameters and sources are given in Table 1. The top layer was defined as about 3 5 m silty clay layer, followed by a coarse sand layer that may reach 15 m depth. Numerical modelling Groundwater modelling is accomplished by using MOD- FLOW a modelling program developed by the USGS. This is the most widely used finite-difference groundwater model and considered a standard for groundwater modelling. The development of input files are compiled using Visual MODFLOW a commonly used pre-processor of data that is used to speed up and facilitate the development of the MODFLOW model. A three-dimensional representation of the site was created in Visual MODFLOW. This model domain was created as a 17,200 m by 17,200 m mesh in the X and Y (corresponding with east west and north south, respectively) directions, respectively, with a general uniform grid spacing of 345 m between grid nodes (Fig. 4). The horizontal wells were modelled using the Drainage package of MODFLOW model. The horizontal well (drain) is 0.2 m diameter and 50 m long. The horizontal wells flow to the central collector well having diameter of 3 m. The model setup for the horizontal and collector wells is shown in Fig. 4. The centre of the horizontal well (drain) was assigned as the drain elevation. The hydraulic conductivity within the drain was assigned thousand times higher compared to the aquifer value. The drain bed material was assumed to be the same as the aquifer material. Vertically, the grids are refined to:

Environ Earth Sci (2013) 68:1119 1126 1121 Fig. 2 Conceptual model of North Kelantan groundwater aquifer Fig. 3 Generalized hydrostratigraphy of North Kelantan aquifer

1122 Environ Earth Sci (2013) 68:1119 1126 Table 1 Model parameters and sources Parameter Source Model boundaries Saim (1997) and Mohamad Faizal (2002) Northern and Eastern boundary Western boundary Geology, hydrogeology map, JMG reports Southern boundary Model layer Saim (1997) and Mohamad Faizal (2002) Borehole information, JMG report Aquifer thickness and Borehole information, JMG report bottom elevation Saim (1997) and Mohamad Faizal (2002) Aquifer hydraulic Saim (1997) and Mohamad Faizal (2002) conductivity JMG reports, established value Recharge Estimation from Rainfall data (Jan 2000 Dec 2005) and (Jan 2009 Dec 2010) Water abstraction Air Kelantan Sdn. Bhd. JMG records Groundwater level JMG records 1. 0.2 m uniform spacing between nodes in the area around Pintu Geng well field (Fig. 4), 2. 3 m uniform spacing for the PGHCW and Pintu Geng well field including the following wells: PGN1 up to PGN8 (Fig. 5), 3. 23 m uniform spacing between nodes around the active pumping well fields. However, Pengkalan Chepa (inactive well field) was also included for prediction purpose, 4. 69 m uniform spacing between nodes of the pumping well fields in the area where groundwater development takes place. Horizontally, finer spacing is designed for the well field in Pintu Geng up to 8 in. (0.2 m) grid size in order to handle the horizontal well designed diameter. The horizontal refining process was only limited to immediate area around the Pintu Geng well field. The discretization of the modelled area consists of 50 columns and 50 rows. The areas surrounding the modelled Fig. 4 Proposed coverage of North Kelantan aquifer system with the model grid area. Model is bounded by river boundary condition to the east and west

Environ Earth Sci (2013) 68:1119 1126 1 Fig. 5 Layout of Pintu Geng well field and the six horizontal collector wells that drain water to PGHCW Table 2 Parameters to be used for the calibrated model Parameter Value Zone 1 (silt) Zone 2 (coarse sand) Zone3 (fine sand) Kx (m/s) 1E-7 0.006 0.0002 Ky (m/s) 1E-7 0.006 0.0002 Kz (m/s) 1E-8 0.0006 1E-5 Ss 1E-5 Sy 0.27 Recharge 12 % of rainfall Total porosity 0.20 Effective porosity 0.11 Kelantan River Pengkalan Datu Pengkalan Chepa Conductance (m 2 /day) 309 305 278 Kubang Kerian Pasir Hor Pintu Geng Pasir Tumboh Kg Sbg Kota Kg Puteh Pdp Groundwater pumping (m 3 /day) 10,800 14,680 8,000 4,200 7,800 10,000 27,200 8,800 Kg Kampung (village), Sbg Seribong, Pdp Penyadap area are made inactive. A digitized map of the study area was superimposed on the model as a base map. The model consists of two layers. These include layer 1 (*4 m of silty soil) and layer 2 (*10 m of coarse sand) which represent the shallow aquifer in the Kelantan area. The less permeable layer on the surface that partially confined the aquifers has been modelled as a continuous layer with a reasonable value of hydraulic conductivity (1E-7 m/s).

1124 Environ Earth Sci (2013) 68:1119 1126 Groundwater model and horizontal wells simulation The HCW was simulated using the MODFLOW drainage package. The South China Sea in the north-eastern boundary of the modelled area was assigned as constant head (Fig. 3). River boundaries were assigned for the northern, eastern (and south-eastern), and western edges of the modelled area which are bounded by three rivers, namely Pengkalan Chepa, Pengkalan Datu, and Kelantan. The boundary condition only allows water to enter the horizontal pipe screens by gravity flow when the water table is above the screens level. All the rivers have been modelled using the River Package of MODFLOW groundwater model. A value between 5 and 20 % of annual precipitation is recommended as an estimate of recharge when data is not available (Waterloo 2005). Based on this guidance, recharge was set equal to 12 % of average annual precipitation after subjection to revision throughout the model calibration and validation processes. The Visual MODFLOW model was run at transient state then calibrated to hydraulic heads recorded on a monthly basis from January 2000 to December 2005 at six groundwater monitoring wells for calibration. Table 2 presents parameters of the calibrated model. Most of the groundwater monitoring wells are located outside the well field and within the modelled area. Model calibration consists basically of modifying the recharge and conductance parameters to minimize the error between predicted and observed heads. Estimated recharge rate is 12 % of the rainfall. Fig. 6 Head versus time during the calibration period Fig. 7 Head versus time during the validation period

Environ Earth Sci (2013) 68:1119 1126 1125 Fig. 8 Capture zones of the present field The model validated to hydraulic heads recorded on a monthly basis from January 2009 to December 2010 at six groundwater monitoring wells located on-site. Estimated recharge rate is 12 % of the average rainfall. Results and discussion The model was calibrated and validated up to 95 % confidence interval. Model calibration is aimed at minimizing the normalized residual mean squared (nrms) error and maximizing the correlation coefficient (r) between predicted and observed groundwater heads. In contrast, the validated model exhibits nrms and r ranges between 20 and 46 % and 0.75 and 0.86, respectively. Figure 6 illustrates the simulated and observed head versus time. In contrast, the validated model exhibits nrms and r ranges between 20 and 46 % and 0.75 and 0.86, respectively. Figure 7 illustrates the simulated and observed head versus time. Delineation of capture zones The present pumping rate creates two dominant groundwater capture zones (Fig. 8). These include: 1. upstream capture zone developed due to the present pumping condition at present rate, 2. downstream capture zone developed due to natural flow gradient and groundwater discharge towards the South China Sea. However, some eight sub-capture zones of different diameters have been developed around the existing well fields within the upstream capture zone. These include: Table 3 Capture zones around each well field Well field Pasir Hor Penyadap Kg Puteh Pintu Geng Kota Kubang Kerian Seribong Pasir Tumboh Capture zone (km) North &2.5, North-west &1.5, South &0.7, South-east &4.0, East &0.5, West &2.0 North &0.5, South &0.6, West &1.7 East &1.5, South-west &2.2, West-south-west &3.0 North &4.5, South &1.5, East &3.0, West &1.5, North-east &4.5 North &1.0, South &0.5, East &1.5, West &2.0 North &0.7, South &2.0, East &1.5, West &0.8 North &0.6, North-east &2.2, South &0.3, East &1.4, West &0.3, South-east &2.5 North &0.5, North-east &1.2, South &0.4, East &1.0, West &0.4, South-east &2.0 North &0.8, North-east &1.8, South &0.9, East &1.7, West &0.4, South-east &1.0, South-east &2.2

1126 Environ Earth Sci (2013) 68:1119 1126 Table 4 Impacts of different scenarios of horizontal well placement at Pintu Gang s well field Item Number of radial well (drains) Drain elevation (m) Screen location Drains discharge range (m 3 /day) Water level in the PGHCW (m) Drawdown with 300 m radius from PGHCW Remarks Natural 2-0.5-0.4 to -0.6 5,700 to 13,300-0.5 Balanced water budget at zone 2 a Natural 6-0.5-0.4 to -0.6 6,200 to 14,100-0.47 Balanced water budget at zone 2 a Pumping at 9,000 m 3 /day 6-0.5-0.4 to -0.6 0.0 to 5,800-2.0 Up to 0.27 Balanced water budget at zone 2 a Natural 6-2.0-1.9 to -2.1 14,000 to 20,000-2.0 Balanced water budget at zone 2 a Natural 6-6.0-5.9 to -6.1 25,000 to 108,000-6.0 0.27 1.0 Balanced water budget at zone 2 Natural 6-3.3-3.2 to -3.4 19,200 to 43,700-3.0 Up to 0.5 Balanced water budget at zone 2 a Zone 2: capture zone of Pintu Geng well field Pasir Hor, Penyadap, Kg Puteh, Pintu Geng, Kota, Kubang Kerian, Kg Seribong and Pasir Tumboh (Table 3). Impacts of Pintu Geng horizontal collector well (PGHCW) Different scenarios were run to investigate the impact of the horizontal well placement on the groundwater head and drawdown inside the PGHCW itself and also within a radius of some 300 m from the Pintu Geng well field. The results are presented in Table 4. The model demonstrates that under natural flow condition at -3 m depth, the six collectors (drains) may tap a volume of groundwater ranging between 19,200 and 43,700 m 3 /day (12 % of stress periods are above 30,000 m 3 /day). The variation in the rate is due to the variation in the recharge input (12 % of the monthly rainfall). This rate is feasible considering that the model pumps for 24 h a day, while in reality pumping duration is less than that of the model. On the other hand, it seems that as much as deepening the location of the collectors (drains) is considered i.e. up to -6 m, the more water can be tapped (Table 4). This situation is basically due to the remarkable increase of hydraulic gradient which in turn increases the rate of groundwater flow into the drains regardless of the seasonal fluctuations of the water table. In both the cases the predicted drawdown of 0.5 1.0 m is within the safe limit (maximum limit is 2 m for this area around the PGHCW). Conclusions The calibrated and validated MODFLOW models were successfully established in the transient state for the North Kelantan shallow aquifer. The model demonstrated satisfactory simulation results and labelled with good statistic measures. The present pumping rate divided the modelled area into two dominant groundwater capture zones i.e. upstream and downstream. The upstream one is further subdivided into eight sub-capture zones that developed around each well field site due to the present pumping. The model demonstrates that under natural flow conditions at -3 m depth, the six radial wells (drains) may tap a volume of groundwater ranging between 19,200 and 43,700 m 3 /day (12 % of stress periods are above 30,000 m 3 /day). This rate is feasible considering that the model pumps for 24 h a day, while in reality pumping process is for a short duration than that for the model. The modelling exercise demonstrates that as much as deepening the location of the collectors (drains) is considered i.e. up to -6 m, the more water can be tapped. The predicted drawdown is in the range of 0.5 1.0 m for the area within 300 m radius from the PGHCW and this value is considered environmentally impact free. Acknowledgments The authors are grateful to the University of Malaya (Grant THEQS/2010A) and Air Kelantan Sdn. Bhd. (Kelantan State Water Company) for supporting this research project. References Anderson MP, Woessner WW (1992) Applied groundwater modelling: simulation of flow and advective transport. Academic Press, San Diego Mohamad Faizal TB (2002) Groundwater management for shallow aquifer in coastal area of Kota Bharu. Dissertation, University of Technology Malaysia (UTM) Saim S (1997) Groundwater protection in North Kelantan, Malaysia: an integrated mapping approach using modelling and GIS. Dissertation, University of Newcastle Tyne, UK Waterloo (2005) Visual MODFLOW v. 4.1 User s Manual. Waterloo Hydrogeologic, Inc. Waterloo