The emergence of horizontal directional drilling (HDD) technology has introduced an innovative approach for the installation of ground water wells. The horizontal configuration seems appealing for remediation scenarios where an affected area is quite large in lateral extent, and a horizontal well could potentially be more effective than a vertical well. The technology also may be beneficial in reaching areas under large buildings that would otherwise be difficult to access with vertical wells.

Horizontal wells have been installed for air sparging, soil venting, and ground water extraction since the late 1980s. Initial installations of these wells in both saturated and unsaturated sediments have produced encouraging results. Unfortunately, comprehensive performance studies of field installations are scarce, leaving the design aspects of these wells to either a trial-and-error approach or to computer models that have not been validated to this application.

A field investigation has been underway to determine the hydraulic behavior of a horizontal ground water extraction well in a shallow, unconfined, sandy aquifer.

Field Investigation

A horizontal ground water extraction well study site is located at a University of Guelph experimental research farm in Cambridge, Ontario. The horizontal well was installed in an unconfined aquifer, consisting of uniform medium sand. Boreholes indicate that the aquifer extends to a depth of 20 to 30 feet below ground surface, where it overlies a major till unit.

The ground surface overlying the well screen is a topographical low and has a thin layer of peat just below the surface. The water table has been observed to fluctuate between 18 inches and 3 feet below ground surface.

As depicted in Figure 1, the horizontal, screened section of the well was installed at a depth of 12 feet below ground surface. This screened section is 75 feet long, consisting of a PVC continuous wrapped wire screen. The PVC pipe on each end of the screen totals 130 feet, providing accessibility to the screened section from two points at ground surface.

The bore path was drilled into the ground at a slope of about 1:3.5, leveling out for a distance of 75 feet at the target depth of 12 feet below ground surface, and then sloping upwards to exit the ground.

The initial bore path was drilled with a 4-inch diameter steerable auger head. Steering was accomplished through control of the rotational position of the duckbill-shaped auger head. A walkover tracking system containing a receiver was used to identify the approximate location and depth of a sonde in the auger head, allowing the drill operator to guide it along the planned bore path.

Once the auger emerged from the exit end of the bore path, a series of two back reamers were attached to the drill rods. They were pulled through to expand the bore path to a diameter of 14 inches.

In order to protect the horizontal well screen from breaking or clogging, a carrier casing was used to drag the well assembly through the bore path. The horizontal well assembly was placed into the 8-inch polyethylene pipe carrier casing, which was attached to the second back reamer and pulled through to extend from the exit to the entrance of the bore path. Finally, the carrier casing was pulled out, leaving the horizontal well in place and completing the well installation. The well was flushed with water in order to dilute the drilling fluid and help to collapse the formation around the well.

Hydraulic Testing

To obtain a three-dimensional pressure distribution around the horizontal well during ground water extraction, a monitoring network consisting of 18 multilevel piezometer bundles was installed by a direct push method. A temporary casing was installed to the maximum desired vertical depth using a drive point. The piezometer bundle then was placed inside the casing, which was retracted, allowing the formation to collapse around the piezometer bundle. Following installation, water was pumped from each piezometer to create a natural sand pack around each screen.

The piezometer bundles were located along the length of the well screen and along two crossections perpendicular to the well screen. Each piezometer bundle was installed to monitor the same five vertical elevations spaced equally across the depth of the aquifer. Water levels were measured from these piezometers both prior to and during pumping with hand-held water level tapes.

A separate pumping test was performed for each piezometer bundle in the monitoring network. Each test was performed over a period of 40 minutes to determine the initial transient response of the aquifer. Following each test, the aquifer was allowed to recover for 40 minutes before the next text was initiated. One piezometer bundle was tested on three consecutive cycles of pump and recovery, and the total drawdown was not significantly different in any of the tests, indicating that this multiple pumping test methodology is an acceptable approach to observe the required spatial data.

Field Results

It is evident from the drawdown vs. time test data that the aquifer responds quickly at all depths. Even near the aquitard interface, the drawdown is significant. For this field scenario, a horizontal well could be very effective at capturing ground water from the entire depth of the aquifer.

The drawdown within the aquifer cross-section located at the center of the well screen demonstrated a quick response around the well, including even deep portions of the aquifer. In each test, ground water was extracted from the horizontal well by an above ground, centrifugal pump operating at a rate of 15 gal./min. The intake for the pump was located at the west end of the well screen.

Transient water level measurements were recorded, with one water level tape dedicated to each piezometer in a bundle. The measurements were duplicated for several bundles and found to be reproducible. A different drawdown effect occurred at the cross-section located at the east end of the well. The magnitude of drawdown is smaller at all locations along this cross-section, and the drawdown evolves in the upper portion of the aquifer.

So the drawdown along the length of the well screen varies, with the highest drawdown occurring at the two extreme ends of the well screen. One possible explanation for this observation is the variation in aquifer volume contributing water along the length of the well. At the center of the well, the contributing volume or area is the smallest, while at the ends of the well, water can be supplied to the well from a very large semi-circular volume or area. If the flow rate provided by the aquifer to the well is constant along the length of the well, then the drawdown at the center of the well would be the largest since it has the smallest contributing area.

Preliminary results from this horizontal ground water extraction well field test suggest that the drawdown effect due to pumping of the well is present across the depth of the aquifer, and the drawdown is not uniform across the length of the horizontal well screen.

Future Work

Further testing will be conducted to better understand the ground water hydraulics of flow to the horizontal well. A flowmeter and pressure transducer placed inside the well will be used to analyze the flow along the length of the well screen. Water table piezometers will be installed to monitor the actual water table response to pumping. A steady state test will be conducted to determine the late time response of the unconfined aquifer to pumping. The maximum pumping rate for the horizontal well will be determined. The possibility of performing slug tests on a horizontal well is presently under investigation.

In addition, a vertical well will be installed and tested to compare the aquifer response and to establish a basis of comparison to assess the performance of the horizontal well against.

Finally, data from the field experiments will be compared to results from a three-dimensional, unsaturated flow model.