Pumping Tests

25.1 General Principles

In principle, a pumping test involves pumping at a steady known flow from a well and observing the drawdown effect on groundwater levels at some distance away from the pumped well. In response to pumping, phreatic and piezometric levels around the pumping well will fall, creating a 'cone of depression'. The permeability of the ground is obtained by a study of the shape of the cone of depression, which is indicated by the water levels in the surrounding observation wells. The shape of the cone of depression depends on the pumping rate, the duration of pumping, the nature of the ground, the existence, or otherwise, of intermediate or other boundaries, the shape of the groundwater table, and the nature of recharge.

From the data obtained from the test, the coefficients of permeability, transmissivity and storage can be determined for a greater mass of ground than by the use of the borehole tests described in Chapter 21. The results can be used in the evaluation of dewatering requirements and groundwater resources, as well as in the design of positive groundwater cut-offs. It should be noted that a given coefficient of transmissivity can result from many different distributions of permeability with depth. If the test is intended for the evaluation of permeability in the design of dams and other similar projects where seepage is an important consideration, the use of down-the-hole velocity profiling at constant outflow can provide a permeability profile of the ground.

Pumping tests can be expensive, requiring adequately screened and developed pumping and observation wells, suitable pumping and support equipment, and personnel. Care should be taken therefore to design a suitable test programme. Before attempting to carry out a pumping test, reliable data should be obtained on the ground profile, if necessary by means of boreholes sunk especially for the purpose. The geological units encountered may then be grouped into hydrological units on the basis of permeability (Leach & Herbert, 1982).

The natural groundwater conditions should be determined by careful monitoring over a sufficient period before the pumping test. Ideally, the conditions should be stable during the test; if they are not, the fluctuations have to be recorded.

Fluctuations can be caused by rainwater infiltration, tides, groundwater extraction from wells, and nearby construction activities. This is particularly important in highly permeable ground subject to rapid recharge.

The interpretation of the data from a pumping test can be complicated and is much affected by the inferred ground conditions and by the influence of any boundaries. Where necessary, expert advice should be sought.

In Hong Kong, pumping tests have been used occasionally to determine hydrogeological parameters (as described above) but are more often carried out for the purpose of estimating the yield of water wells. They are also conducted occasionally to provide data for the design of major dewatering schemes associated with the construction of deep basements. The possible effects on adjacent ground and structures, e.g. settlements and inducement of negative skin-friction on piles, should be carefully considered before conducting a pumping test (see

Section 8.3.2(e)). Pumping test proposals for private developments must be submitted to the Buildings Ordinance Office for approval and consent prior to the commencement of works (see Appendix A.8).

25.2 Groundwater Conditions

There are two main types of groundwater conditions, confined and unconfined, and these should be recognized for analytical and design purposes.

(a) Confined. If the ground under investigation is fully saturated and the water is confined under pressure between two impermeable layers, then confined conditions are said to exist.

(b) Unconfined. If the original phreatic level is everywhere below the upper surface of the aquifer, then unconfined conditions are said to exist.

Intermediate between the above two groundwater conditions is a third called the semi-confined condition. In this case, fully saturated ground is overlain by material of significant but lower permeability, and significant leakage takes place across the boundary in response to pumping. Analysis of data from semi-confined conditions is possible, but the condition is less commonly encountered than the other two types.

The three types of groundwater conditions may be recognized by the test response (BSI,

2010). [Amd GG2/01/2017]

25.3 Test Site

Although the choice of test site may be dictated by practical considerations, such as access and availability of existing boreholes, the site should be representative of the area of interest. The hydrological conditions should not change appreciably over the site. It is essential that discharged water is not able to return to the ground under test.

25.4 Pumped Wells

Pumped wells should be of sufficient diameter to permit the insertion of a rising main and pump of a suitable type and capacity, together with a standpipe and velocity meter, if required. They should be provided with an adequate well screen, and filter pack where necessary, to prevent the withdrawal of fine particles from the surrounding soil. The minimum borehole diameter which will achieve this purpose is often 300 mm. It is desirable that they penetrate the full depth, of the water-bearing zone being tested. Where the ground is composed of two or more independent horizons, each should be tested separately. Where fully penetrating conditions do not exist, the data have to be corrected before analysis. In all cases, the screen intake area should be such as to ensure that the maximum velocity of water entering

the well is not greater than about 30 mm/s to ensure that hydraulic well looses are of an acceptable level.

If, during the test, changes in the shape of the cone of depression that are due to extraneous causes are a significant fraction of those due to pumping, then the resulting estimate of permeability may become unacceptable. Such influences can be corrected by monitoring (Walton, 1962), both before and during testing. Where possible, and within the limitations set by the permeability, the pumping rate should be chosen so that resulting changes in water levels are much greater than those due to extraneous causes, thus minimizing the effects of the latter on the results.

Suction pumps can be used where the groundwater does not have to be depressed by more than about 5 m below the intake chamber of the pump, and drawdown can be increased by setting the pump in a pit. For greater depths, submersible pumps are preferable. The more permeable the ground, the greater the pump capacity required to produce measurable drawdowns in the observation wells.

It is essential that the discharge is kept constant for the duration of the test and that all the water level observations are related to a time-scale referred to the onset of pumping.

It is particularly important to maintain a constant pumping rate when vertical flow velocities in the pumping well are being measured for the purpose of determining the relative permeabilities of specific horizons in the ground under test. The pumping rate may be controlled by a gate valve in the discharge line or by varying the speed of the pump, or both.

The rate of flow from the pump may be measured by a flow or orifice meter, or by a notch tank with automatic recording.

It is important that pumping wells should be adequately developed. Development of a well is the process by which particles surrounding the screen are rearranged, with coarsening grade and better uniformity towards the screen; it can be achieved in a number of ways (Johnson, 1982). Maximum development is indicated when the ratio of pumping rate to fall in water level in the pumping well reaches a maximum. Fine particles from the ground are removed during development, resulting in a stable, porous and permeable medium surrounding the well.

Successful well development results in reduced hydraulic head losses as the water enters the pumping well but, in any case, these losses (well losses) should be accounted for in the analyses of test results.

In Hong Kong, pumping tests are sometimes carried out in large diameter hand-dug caissons. This has several disadvantages, as the caisson may only partly penetrate the aquifer being tested, and the well storage is large. Also, high well losses are often incurred, and for this reason observation wells should always be used in conjunction with pumping tests in caissons.

25.5 Observation Wells

Observation wells should have an internal diameter large enough to permit insertion of a diameter or other water-level measuring device, but if the diameter is too large this may cause a time lag in drawdown. Standpipes with an internal diameter of 19 mm are often used.

Observation wells should penetrate the same ground as the pumping well and should permit entry of water from the full depth of ground being tested. If there is any risk that fine soil particles may clog the observation wells, they should be surrounded by a suitably graded filter material.

Although the permeability of the ground may be estimated from the pumping well drawdown data alone, more reliable values are obtained using data from one or more observation wells. The recommended minimum number of observation wells required to yield reasonably representative results is four, arranged in two rows at right angles to each other.

Their distances from the pumping well should approximate to a geometrical series. It may be necessary to add more wells if the initial ones yield anomalous data. If linear boundary conditions are associated with the site (e.g. river, canal or an impermeable subsurface bedrock scarp, fault or dyke), the two rows of observation wells are best arranged parallel and normal to the boundary.

The minimum distance between observation wells and the pumping well should be ten times the pumping well radius, and at least one of the observation wells in each row should be at a radial distance greater than twice the thickness of the ground being tested. However, unless the pumping rate is very high, and the duration of pumping long, particularly in low permeability ground under unconfined conditions, falls in water levels may be small at such distances. Preliminary calculations using assumed permeabilities estimated from borehole data will help to indicate the likely response in observation wells to pumping. Hence the appropriate distance of the observation wells from the pumping well and the timing of observations can be assessed.

In addition to the observation wells described above, it is desirable to have an additional standpipe inside the pumping well in order to obtain a reliable record of the drawdown of the well itself.

Depths to water levels should be measured to within ±5 mm. This usually means that measurement devices have to be checked at regular intervals against, for example, a graduated steel tape.

The water levels can be monitored with either an electrical dipmeter or an automatic well level recording system.

25.6 Test Procedures

Once the character of fluctuations and other extraneous influences has been established, the test programme designed and the wells developed, pumping of the ground at a constant rate should commence. Water levels in all wells are then measured with respect to time since commencement of the pumping. Typically the frequency of measurement might be at 1 min

intervals for the first 15 min and at regular logarithmic intervals thereafter. Sometimes, shorter intervals may be required initially. Therefore each well may have to be monitored by independent observers for the first 100 min, and then by one or more observers thereafter. In distant observation wells where head changes are small, automatic recorders can be used, although these generally require observation wells of 100 mm diameter or greater.

The measurements should be plotted during the course of pumping to evaluate the quality of the data, the nature of the response, and the required duration of pumping. Johnson (1982) and Kruseman & DeRidder (1980) have discussed the time requirements for both steady and non-steady state pumping tests carried out on confined, semi-confined and unconfined aquifers.

In all cases, water levels should continue to be monitored with respect to time from cessation of pumping until recovery of levels to the original values is complete. As in the drawdown phase, recovery data should be taken at 1 min intervals for 15 min following cessation of pumping and thereafter at regular intervals on a logarithmic scale.

25.7 Analysis of Results

There are two forms of analysis of pumping test data :

(a) Steady state. If pumping continues long enough, water levels cease to fall, and the hydraulic condition of the ground is said to be in a steady state with respect to time.

(b) Non-steady state. Before equilibrium is reached, water levels fall at a changing rate with respect to time and the hydraulic condition of the ground is said to be in a non-steady state.

The simpler form of analysis is the steady state type, but the necessary duration of pumping can be significantly longer than that necessary for non-steady state analysis. The analysis technique is also dependent on aquifer response, i.e. whether confined or unconfined conditions are present. A summary of some of the available analysis techniques is given in BSI (2012c), and these are further discussed by Johnson (1982) and Kruseman & DeRidder

(1980). [Amd GG2/01/2017]

A number of simplifying assumptions regarding ground conditions are required in whatever method of analysis is used, and it is therefore common that the actual drawdown data collected in the field may lead to ambiguities in the analysis. This may be caused by inhomogeneity and anisotropy in the aquifer, or the presence of unknown barriers to groundwater flow. In some cases, high flow velocities around the well may invalidate the use of Darcy's law, upon which most methods of analysis are based.