Underground Singapore 2023

Field Pumping Tests for Hydrogeological Studies

ABSTRACT: Permeability of the soil is one of the important parameters that affect the ground water
table during the tunnelling and deep excavation works. Understanding and determining the permeability
parameter is essential for controlling ground water drawdown during the deep excavation works.
Permeability is commonly measured during the site investigation by falling head/rising head tests, these
point permeability tests are not sufficient for detailed hydrogeological studies hence field pumping tests
which will give the mass permeability of the ground are essential. There is insufficient information in
existing literature on how to carry out pumping tests to determine mass permeability for typical deep
excavation and tunnelling projects.

This paper describes some pertinent details of field pumping tests at different sites in Singapore which
include practical considerations and challenges during planning and execution of the field pumping tests.
Determination of permeability from the field permeability test is also presented.

1 INTRODUCTION

Deep excavations and tunneling works carry significant hydrogeological risk as the groundwater
drawdown will affect the large areas around the works and result in the settlements in adjacent areas.
The current approach of using in-situ permeability tests together with laboratory falling head/constant
head tests to obtain the permeability is not fully representative due to non-homogenous nature of the
ground. In-situ permeability tests are carried out over a relatively short period of time, removing, or
adding a relatively small volume of water. Obtaining appropriate representative value of permeability
from these tests is difficult because the ground is generally anisotropic and heterogeneous, permeability
varies from one point to next point. One of the most reliable ways of estimating the effective
permeability of large ground mass is by field pumping tests. In field pumping tests, water is pumped out
from a well, after which pump flowrate and drawdown in surrounding areas are recorded from
observation wells and piezometers. From the test data a further analysis must be carried out to determine
the permeability values and influence zone of water draw down can be determined.

As the target drawdown in observation wells are recommended to be at least 10% of subsequent
drawdown during works in CIRIA C750, pumping wells should be sufficiently deep. The pumping well
is thus usually of diameter 300mm in order for a submersible pump to be able to fit. Submersible pumps
are required as surface pumps typically are unable to overcome more than 5m of water head.

The pumping well is typically a perforated PVC pipe, wrapped with filter material and surrounded by
gravel as filter material after installation in the borehole. The perforation of PVC pipe, geotextile wrap
and gravel have to be designed and checked to ensure a balance between preventing excessive well
losses and preventing fines from entering the well. Clean water should also be used as drilling fluid to
avoid affecting the permeability of the well.
 Underground Singapore 2023

Observation wells containing either piezometer should be installed at spacing that is able to capture the
groundwater drawdown profile. For ground with higher permeability, observation wells should be
placed closer to pumping well as the steady state cone of depression should have a steeper profile.

1.1.1 Pre-testing preparation

Soil investigation (SI) should be carried out at each pumping well and observation well location. Particle
size distribution test should be assigned as the results allow a preliminary estimate of permeability as
well. Particle size distribution information is also required in design of the well filter.

Where a confined aquifer is being targeted, borelogs have to be reviewed for design of well perforation
lengths such that the perforated zone matches the confined aquifer depths. Base readings of water levels
in all wells should also be taken for at least 1 week before test to determine that they have stabilised and
detect any fluctuations e.g. tidal fluctuations.

A schedule of pumping rates and duration should be calculated and proposed based on estimated
permeability, with the objective being to pump at highest possible rate to achieve target drawdown at
steady state seepage condition. The zone of influence of pumping test and expected drawdown should
also be calculated.

Disposal of water extracted should be arranged such that it does not return to the target aquifer. A
calibrated flow meter will be required to accurately record the flow rate.

1.1.2 Testing Stages and Termination Criteria

Figure 1. Stages of pumping test shown in plot of pumping well water level against time.

The pumping test typically consists of several stages as shown in Figure 1.

During the equipment test stage, the pump and full setup is being checked for functionality at the highest
and lowest flow rates. Thereafter, water level is allowed to recover to initial levels. By observing the
recharge rate during the equipment test, the suitability of step test initial flow rate can be re-evaluated.

The step test should consist of 5 steps with increasing flow rates with the objective being to estimate the
flow rate which leads to steady state flow at target drawdown depth. Water level should be allowed to
recover after the step test.

The constant rate discharge test usually lasts between 24 hours to 5 days depending on how quickly
steady state flow is achieved. When steady state flow is a test objective, the termination criteria of the
constant rate discharge test should be based on a determination that groundwater levels in observation
wells are stabilised and the cone of depression has stopped spreading. This is usually taken to be the
case when 3 consecutive readings in each observation well are of a certain difference e.g. 10mm, spaced
1 hour apart.

The final stage is the recovery test where readings should continue to be taken for analysis.

The rest of this paper describes the details and discuss the experience from such pumping tests carried
out in various LTA projects.
 Underground Singapore 2023

2 PUMPING TESTS CARRIED OUT

2.1 Location 1

Location 1 is located mostly in Bukit Timah formation (see Figure 2). A single pumping test was carried
out at to check the permeability of G(V) soil for groundwater flow behaviour to be analysed more
accurately. The pumping well consists of a 200mm diameter borehole containing a 150mm diameter
PVC perforated casing. Observation wells were arranged in approximately 5m-10m-20m-20m spacing
(see Figure 3). The pumping and observation wells were perforated along the length in G(V) soil. Both
water standpipes and piezometers were installed with readings recorded by datalogger. Grundfos
submersible pump (SP1A-14) with electric control panel was adopted. Based on borehole permeability
tests, the expected pumping rate was 2.5L/min for 58h to achieve a drawdown of 25m to -10mSHD. The
termination criteria for pumping was a minimum pumping duration of 72 hours or until steady state is
achieved, whichever earlier. Steady state flow was assumed to have been achieved when water levels in
observation wells maintain within 1cm over a duration of 1 hour.

During the step discharge test, pumping was carried out at increasing rates from 2.4L/min for 7.5h to
3.5L/min for 5.7h to 6L/min for 5h to 8.5L/min for 2.7 hours. Figure 4 shows the increasing rates of
pumping and corresponding decreasing water level in pumping well. As the drawdown exceeded the
target of -10mSHD, pumping rate was reduced to 8.0L/min and held constant for 30 hours, maintaining
the drawdown at 10mSHD. Upon termination of pumping, water level in the pumping well recovered
rapidly. The water level recovery was monitored for a period of 4.5 days.

The permeability value obtained from lab falling head permeability test was 1.47E-7m/s and 5.6E-8m/s
at RC/3091A and 7.4E-08 at RC/3091G. Meanwhile the borehole rising head permeability test was
1.74E-6m/s at RC/3091A and 6.3E-9, 4.0E-8, 7.4E-8m/s at RC/3091G. Permeability values obtained
from pumping test was approximately 4E-7m/s based on Theis approach (BS5930-1999) for non-steady
state analysis of unconfined aquifer. A PLAXIS FEM seepage analysis as shown in Figure 5 was used
to validate the calculated permeability. The model was able to simulate the step test and constant rate
test stages of the pumping test to provide similar drawdown values using permeability of 4E-7m/s. In
the PLAXIS model, model extent was varied from 200m to 1000m for sensitivity check. From Figure
6, it can also be noted from the groundwater head contours that the influence zone of the pumping well
is approximately 20m (radius).

By carrying out a pumping test focused on G(V) soil, a more representative permeability value has been
obtained. This allows more confidence in modelling groundwater behaviour in the subsequent design of
recharge systems.

Figure 2. Ground condition and location of submersible pump and piezometer tips.
 Underground Singapore 2023

Figure 3. Pumping well and observation well layout. Figure 4. Discharge rates and water drawdown.

Figure 5. Model inputs and boundary conditions. Figure 6. Model groundwater drawdown.

2.2 Location 2

Geology at Location 2 consists of Kallang Formation underlain by Old Alluvium. Kallang Formation
includes Upper Marine Clay, F1 Fluvial Sand, up to 12m at certain locations, followed by Lower Marine
Clay, F2 Fluvial Clay and F1 Fluvial Sand as shown in Figure 8. 2 pumping tests were carried out to
determine the permeability value of F1 and identify the groundwater flow behaviour in F1 for planning
of dewatering works and recharge well adequacy.

In the first pumping test, the pumping well was 200mm in diameter, containing a 100mm PVC
perforated pipe. Observation wells were arranged in spacing of 5m, 10m, 15m, 20m, 30m, 40m, 60m as
shown in Figure 9. Pumping and observation wells were both perforated in the F1 zone. Both standpipes
and piezometer were used, together with datalogger. Borehole investigation and permeability tests
indicated relatively low permeability for F1, ranging from 7E-6m/s to 1E-7m/s. Therefore, a pump with
lower pumping capacity was used (Grundfos 1A-14 submersible pump with electronic control panel).
The pumping rate was estimated to be around 1L/min, targeting a drawdown depth of approximately
19m-23m. Pumping termination criteria was initially set to 144h or when steady state flow is achieved,
whichever earlier.

Upon commencement of pumping, it became apparent that the pump was inadequate. The soil
permeability and groundwater recharge rate were higher than expected hence water level in pumping
well could not be drawn down. Overnight pumping at the maximum rate of 1L/min could only induce
1m of groundwater drawdown. Subsequently, a new higher capacity pump was procured and installed.
Thereafter, staged discharge test was carried out at 20L/min, 50L/min, 75L/min, 85L/min and constant
discharge test was carried out at 50L/min. A drawdown of 9m was maintained for 80h. The water levels
in pumping well and observation wells are shown in Figure 10.
 Underground Singapore 2023

The permeability value calculated using Dupuit/Thiem equation (1) for confined aquifer was
approximately 3E-5m/s, which is significantly higher than that of borehole falling head test.

2.3𝑄 𝑟2
Permeability k (m/s) = 2𝜋ℎ0 (𝑠1 −𝑠2 )
× 𝑙𝑜𝑔10 𝑟1 (1)

where Q is flow rate in m3/s.

Figure 7: Dupuit/Thiem formula for confined aquifer.

Figure 8. Ground Condition.

Figure 9. Pumping and observation well layout.

Figure 10. Discharge rates and water drawdown.

In the second pumping test, the pumping well was 300mm in diameter, containing a100mm PVC pipe.
A Grundfos submersible pump SP 5A-8 with electronic control panel with pumping capacity of
120L/min – 280L/min, was used. The layout of pumping well and observation wells are shown in Figure
11. The step discharge test rates adopted were 30L/min, 110L/min, 150 L/min while constant discharge
test was carried out at 45L/min for 60h. The water level in pumping well is shown in Figure 12.

Permeability was calculated using Dupuit/Thiem equation (1) for confined aquifer to be approximately
2.6E-5 m/s to 9.3E-5m/s.

The pumping tests carried out in Location 2 reinforced the understanding of the high permeability of F1
sands. In particular for Test 1, while the initial borehole permeability test reflected a lower permeability,
the subsequent pumping test demonstrated otherwise. Furthermore, the test also served to demonstrate
that while there was some F1 connectivity between the excavation area to nearby residential buildings,
the existing recharge wells in place were sufficient. Both tests were also able to provide ample data for
the design of an effective dewatering system for the excavation.

Figure 11. Second pumping test layout.

Figure 12. Discharge rates and water drawdown.

2.3 Location 3 and Location 4

Geology at Location 3 and 4 were similar. Both locations had Kallang Formation underlaid by Old
Alluvium. The geology are shown in Figures 16 and 17 respectively. 2 tests were carried out at each
location to determine the ground permeability and groundwater flow behavior for open cut excavation.
All pumping wells were 250mm in diameter containing 150mm diameter perforated steel casing.
Pumping well and observation well layout and spacing are shown in Figure 14 and Figure 15. Perforated
zone of pumping and observation wells were at the OA(B) and OA(C) layer. Grundfos submersible
pump (SQ 3-40) was used. Pumping rate was proposed as 30L/min while termination of pumping was
based on minimum duration of 72h or until steady state was achieved.

During the Location 3 pumping test A, pumping at a rate of 30L/min caused 18m of groundwater
drawdown in the pumping well within 10 min. Due to low permeability of the ground, when pumping
stopped, water level only recovered 0.24m after 3h. Subsequently, pumping at 3L/min caused another
10m drawdown within 30min. After the 10m drawdown was recovered, pumping at 1L/min caused 18m
drawdown in 3h. Water level recovery was monitored for 17 days.

During the Location 3 pumping test B, pumping at a rate of 1.5L/min, led to 7h of pumping with 29m
of groundwater drawdown. Subsequently, recovery was monitored for 2mth. The plot of water level in
pumping well against time is shown in Figure 18.

During Location 4 pumping test A, pumping was carried out at 1L/min, for 9h 34min, at which point in
time, the power generator broke down. Groundwater was drawdown to 25m. Recovery was monitored
for 1 month.

During Location 4 pumping test B, pumping was carried out at 1L/min, for 12h 40min, causing a
drawdown of 30m. Groundwater level recovery monitored over 1.5 month. The plot of water level in
pumping well against time is shown in Figure 19.

The permeability values obtained from lab falling head permeability tests were ranged widely, from
3.6E-6m/s to 2.2E-10 m/s. 1xE-8 m/s to 1xE-9 m/s. Permeability from pumping tests were calculated as
shown below.

At Location 3, due to relatively low recharge rates compared to pumping rate, the pumping test recovery
stage was analysed as a rising head test using Hvorslev formula (2) for borehole variable head test
(BS5930-1999). Permeability at Pumping Well A was calculated to be k = 1.76xE-9 m/s and
permeability at Pumping Well B was calculated to be k = 2.88E-10 m/s.
 Underground Singapore 2023

A 𝐻
Permeability k (m/s) = × 𝑙𝑜𝑔e 𝐻1 (2)
F(𝑡2 −𝑡1 ) 2

where
A is cross-sectional area of borehole casing or standpipe;
F is intake factor (BS 5930-1999, Fig 6 and 7);
H1 is the variable head measured at time t1 after test commences;
H2 is the variable head measured at time t2 after test commences.

At Location 4, permeability value calculated using modified Dupuit/Thiem equation (3) for unconfined
aquifer. Permeability at Pumping Well A was calculated to be k = 5.3xE-8 m/s and permeability at
Pumping Well B was calculated to be k = 2.5XE-8 m/s.

2.3𝑄 𝑟2
Permeability k (m/s) = 𝜋(ℎ22 −ℎ12 )
× 𝑙𝑜𝑔10 𝑟1 (3)

where Q is flow rate in m3/s.

Pumping tests at location 3 and 4 were able to demonstrate the low permeability of OA(C) and OA(B)
ground hence a cut-off wall targeting the necessary ground types could be designed. This was also
crucial in allowing an open cut excavation to be carried out without causing excessive groundwater
drawdown at the nearby buildings.

Figure 13. Dupuit/Thiem formula for unconfined aquifer.

Figure 14: Location 3 layout. Figure 15: Location 4 layout.

Figure 16: Location 3 soil profile. Figure 17: Location 4 soil profile.

Figure 18. Location 3 Pumping Test A discharge rates and water drawdown.

Figure 19. Location 4 Pumping Test B discharge rates and water drawdown.
 Underground Singapore 2023

3 DISCUSSION

3.1 Pump selection and pumping rates

As steady state flow is preferrable for simpler and more accurate analysis, a suitable pump must have a
range of flow rate that is high enough to induce a drawdown when recharge rates are high yet low enough
to be able to achieve steady state when recharge rates are low.

It was observed that in Location 2 pumping test 1, the pump selected was had insufficient capacity hence
the pump was unable to remove sufficient water to create drawdown in the well. In contrast, at Location
3 pumping well A and well B, the pumps selected and installed were of too high capacity hence when
operated at their lowest constant flow rate, the pump dried up the well, preventing a constant rate
discharge test from being executed.

3.2 Borehole stability for well construction

In Location 2’s first pumping test, during construction of pumping well, the open borehole with casing
for only the top 6m was left to stand overnight leading to a partial borehole collapse. As the pumping
well has to be sufficiently large to allow the submersible pump to be installed it is usually 200mm –
300mm in diameter. Furthermore, in order not to reduce the permeability of the borehole wall, only
clean water can be used as stabilising fluid. Hence care and planning should be taken to avoid leaving
an open borehole without casing to stand overnight.

3.3 Pre-pumping test monitoring

Pre-pumping test monitoring is essential to establish a baseline for interpreting subsequent data. Pre-
pumping test monitoring will allow existing trends in the groundwater levels to be identified and taken
into consideration. An example is the visible influence of tidal effect on the groundwater level of
Location 2 pumping test 1.

Figure 20. Location 2 pumping test 1, plot of piezometer readings against time.

3.4 Water disposal

BS EN ISO 22282-4 requires discharge water to be disposed such that they create a recharging effect.
Photo 1 shows Location 2 pumping test 1 where water extracted is being discharged into water tank on
site for removal. Photo 2 shows Location 2 target pumping test 2 where water is being discharged
directly into nearby canal which leads to the sea. This is appropriate as the pumping test is targeting the
F1 layer confined by Marine Clay.
 Underground Singapore 2023

Photo 1. Discharge into water tank. Photo 2. Discharge into canal leading to sea.

3.5 Surrounding construction activity

As surrounding construction activity can have an impact on the groundwater flow, it is preferrable for
the pumping test to be carried out before construction activities begin. During Location 2 pumping test
1, there were active recharge wells near to buildings (shown in Figure 17) hence the test could have been
affected.

Figure 21. Active recharge wells.

3.6 Measurement of flow rate

The flow rate together with the observation wells water levels are the main parameters to be measured.
In Location 2 pumping test A, the flowmeter had a minimum accuracy of 0.01m3/h, hence when pumping
at a lower rate than this, manual measurement of water quantity is required (as shown in Photo 3).

Photo 3. Location 2 pumping test 1

3.7 Pumping Test vs Borehole Permeability Test & Lab Sample Permeability Test

It must be noted that borehole permeability tests and lab sample permeability tests in the above locations
yielded wide ranging results that varied by a few magnitudes. This is consistent with the understanding
that borehole permeability tests only test a small volume of soil and is affected by soil disturbance from
drilling. Meanwhile lab sample permeability tests are also subject similar limitations of only testing a
small volume of soil and being affected by sample disturbance from the sampling process (CIRIA C750).
The heterogeneous nature of the ground combined with limited amount of ground being tested in
borehole and lab tests may hence lead to the uncertainty in overall ground permeability.

Comparing between the tests carried out at each location, it can be observed that at Location 1 and 2
where the ground was relatively more permeable, the pumping test yielded higher permeability values
than the average values derived from borehole permeability test and lab permeability test. Meanwhile at
Location 3 and 4 where the ground was relatively less permeable, the pumping tested yielded lower
permeability values than the average values derived from borehole permeability test and lab
permeability test. This may be taken to imply that the borehole and lab tests underestimate permeability
in highly permeable ground while overestimating permeability in less permeable ground.

4 CONCLUSION

Where an accurate value for the mass permeability of the ground is critical for geotechnical design or
impact assessment, a well-executed pumping test will be invaluable. The pumping tests carried out have
been shown to be able to provide more confidence in understanding the groundwater behavior in
particular the permeability. With such information, modelling of groundwater behavior to design cutoff
walls, dewatering and recharge systems can be carried out with greater confidence. Several pitfalls in
carrying out pumping tests e.g. pump selection, have also been highlighted. Hence it is hoped that the
information presented in this paper will be useful in helping engineers better plan, execute and analyse
pumping tests in local ground conditions.

REFERENCES

BS 5930:1999 Code of practice for site investigations