The essential first step in the design process is the development of a conceptual model of the ground and groundwater conditions. Conceptual models were introduced in Section 5.4 and are a non-mathematical representation of ground and groundwater conditions that are used to help understand and communicate geotechnical and hydrogeological conditions at a site.

Designers should use the conceptual model as the foundation for their subsequent choices and decisions about which of the methods in the designer’s ‘tool kit’ is appropriate for use for a particular project or site. If a ‘poor’ conceptual model - i.e. one that is not a good match for actual conditions - is used, then the outcome of the design process may be seriously flawed. The value of the conceptual model to dewatering design cannot be overstated.

The conceptual model is based on the available data on ground and groundwater conditions, the proposed construction project and the environmental sensitivities of the surrounding area. An advantage of developing the conceptual model at an early stage is that the designer will have to critically review the available data. This will help identify any significant gaps or uncertainties in the dataset.

Uncertainty is unavoidable when working below ground. Even if time and money were unlimited, it is not possible fully characterize ground and groundwater conditions. One common problem is that groundwater levels are often only monitored for a finite period, and the data available may not identify true maximum and minimum groundwater levels. The best conceptual models identify and highlight key areas of uncertainty that could affect groundwater control designs. This can help justify the need for further site investigation or allow designers to plan more robust or flexible dewatering systems to address the uncertainty.

The conceptual groundwater model depends on a very wide range of factors that are relevant to hydrogeological conceptual models, as described in Section 5.4, and listed here:

  • (i) Aquifer type(s) and properties
  • (ii) Aquifer depth and thickness
  • (iii) Presence of aquitards and aquicludes
  • (iv) Distance of influence and aquifer boundaries
  • (v) Initial groundwater level and pore water pressure profile
  • (vi) Presence of compressible strata
  • (vii) Geometry of the proposed works
  • (viii) Groundwater lowering technique
  • (ix) Period for which groundwater lowering is required
  • (x) Depth of proposed wells
  • (xi) Environmental constraints

If information is available to address most, if not all, of these factors, then a conceptual model can normally be developed. Conceptual models should not be unnecessarily complex (complexity in modelling is discussed in Section 5.3). On smaller projects in straightforward ground conditions, a conceptual model may simply be a list of the expected conditions. Figure 13.1 shows an example of a simple pro-forma to record key data. Conceptual models are outlined in each of the design examples given in Appendix 5.

Any consideration of groundwater flow in general, or of the equations presented later in this chapter, highlights that aquifer permeability is a critically important parameter. Chapter 12 has described the plethora of techniques available to estimate permeability, from the simple to the very complex. When assessing permeability values to be used in calculation, the designer should not visualize a single permeability value to be used in the conceptual model but rather, a range of realistic values to be used in sensitivity analyses. The range of permeability values may represent uncertainty due to natural variations in permeability, or limitations in the permeability test methods or results.

It is difficult to give simple, useful guidelines on the selection of realistic permeability ranges. There will always be some reliance on judgement and experience, but the following advice is relevant: [1]

Example of simple conceptual model for groundwater lowering system

Figure 13.1 Example of simple conceptual model for groundwater lowering system.

Always consider the important aspects of the design when selecting the permeability range to be used in calculations. Mistakes have been made when designers have focused too much on estimating the highest likely permeability to ensure a pumping capacity sufficient for the maximum possible flow rate. This is not always the most appropriate approach. It is true that in high-permeability aquifers, the total flow rate may be critical, and assessing permeability at the upper end of the possible range may be a robust approach. But alternatively, in soil and rock of low to moderate permeability, a critical case in design may be if the permeability is at the lower end of the possible range, when yields may be very low, necessitating unfeasibly large numbers of wells.

Once the conceptual model exists, an initial view must be taken of the dewatering method to be used and the likely geometry of the system.


It is important to be realistic with expectations of accuracy. Dewatering design should not be viewed as a precise analytical process resulting in single numerical values for key design outcomes such as pumped flow rates. The preceding parts of this chapter have repeatedly highlighted the likely uncertainty in design parameters and boundary conditions, and the resulting need to consider a range of design scenarios by means of sensitivity and parametric studies. In most circumstances, a dewatering design should not report a single value for calculated quantities such as pumped flow rate, time to achieve drawdown, estimated ground settlement or lowering of water levels at nearby groundwater-dependent features. The use of single values in design reports and calculation summaries may give the reader a false sense of precision; it is preferable in calculations and design reports that a range of values be quoted, for example ‘the predicted pumped flow rate will be between 10 and 15 1/s, based on the assumptions stated in the calculations’.

Figure 13.2 presents some interesting data from a study of around 20 pumped groundwater control systems in soils of low to moderate permeability, where dewatering was carried out by wellpoints, deep wells or ejector wells. For each case, the pumped flow rate (qc) was predicted by standard closed-form analytical methods based on a conceptual model and parameter values carefully derived from the available ground investigation information. These values were then compared with the pumped steady-state flow rate recorded in the field (qr). This study showed that even with high standards of conceptual modelling, design and parameter selection, the best that could be achieved was to predict the pumped flow rate within a factor of three times greater than or less than the flow rate observed in the field. In a small number of cases shown in Figure 13.2, the observed pumped flow rate varied from

Comparison of calculated and recorded flow rates from dewatering systems

Figure 13.2 Comparison of calculated and recorded flow rates from dewatering systems. (From Preene, M and Powrie, W, Geotechnique, 43, 191-206, 1993. With permission.) the observed value by a factor greater than three. In these cases, retrospective analysis indicated that the actual ground conditions differed significantly from those indicated in the site investigation information.

The data in Figure 13.2 should not be taken to indicate that methodical and rational dewatering design is futile. Rather, they show that the designer should avoid assigning undue precision to calculated values. The design process should identify the most likely value for calculated values but should also indicate the likely maximum and minimum values. These maximum and minimum values are important, as they need to be compared with the capabilities of the proposed dewatering system. For example, if the proposed dewatering system can cope with the predicted maximum and minimum values of flow rate, either in its nominal layout or with easily applicable modifications, then a robust dewatering design is likely to result. However, if the maximum and minimum predictions could not be handled by the dewatering system or would require expensive and time-consuming modifications, it may appropriate to re-visit the selection of the dewatering method and equipment and to develop a more flexible solution.


There are a number of pumped groundwater lowering methods available (see Section 9.5), and part of the design process is to select an appropriate technique that will satisfy the various constraints on the project in hand.

A useful starting point when selecting a technique is Figure 9.11. When the required drawdown and estimated soil permeability are known from the conceptual model, the appropriate method can be chosen. Where more than one method is feasible, the choice between them may be made on cost grounds, local availability of equipment, or expertise of those carrying out the works.

  • [1] Be aware that different methods of assessing permeability produce results of greateror lesser reliability. Table 12.3 provides some guidance. Consider the relative merits ofeach method when assessing permeability from the available results. • Always compare the permeability results with ‘typical’ values of permeability frompublished correlations with soil types (such as Table 4.1) or, even better, from experience at nearby sites. It is also good practice to compare permeability values usedin design with ‘non-quantitative’ indicators of permeability (see Section 12.5). Thisapproach is vital in excluding unrealistically high or low permeability results. Forexample, few experienced engineers would expect a slightly silty sandy gravel to havea permeability of 1 x 1CH m/s, yet falling head tests in such soils often produce resultsof that order. • If permeability estimates from various differing techniques produce broadly similarresults, in agreement with typical values for those soil types, then the design range ofpermeability could be assessed from the full range of data. If there are large discrepancies in the data from various methods, some of the data may have to be excluded fromthe assessment process. Again, Table 12.3 and Section 12.5 may be of help in assessingthe reliability of data.
< Prev   CONTENTS   Source   Next >