Back Analysis

32.1 General

Natural or man-made conditions on a site sometimes create phenomena which may be used to assess parameters that are otherwise difficult to assess or which may be used as a check on laboratory measurements. Examples of such phenomena are slope failures and settlement of structures. It may then be possible, starting from the observed phenomena, to perform a back analysis, for example, in the case of a slope failure, to arrive at shear strength parameters which fit the observed facts. Back analysis of settlements is also possible, but care is required in assessing actual loadings and the times for which they have acted.

All applications of back analysis should be accompanied by rigorous geological and geotechnical investigations, which should include a thorough review of the history of the problem and examination of relevant climatic and groundwater records. Back analysis should only be used if it is applicable to the problem in hand and the ground conditions encountered.

All parameters that can have a significant effect on the analysis should be carefully considered.

Since it is very rare for a unique analytical solution to be obtained, sensitivity studies are normally carried out to assess the effect of parameters that cannot be obtained by direct means.

The pitfalls of back analysis are discussed further by Leróueil & Tavenas (1981).

32.2 Failures and Design of Slope Preventive or Remedial Works [Amd GG2/01/2017]

In Hong Kong, numerous landslides are caused by intense rainfall every year, and back analysis is sometimes carried out to derive shear strength parameters as part of the design procedure for slope remedial or preventive works. However, a note of caution is necessary regarding the interpretation and use of the results. Although the failure itself can be studied in great detail after the event, it is extremely rare to have accurate information on the specific ground conditions at the time of failure, particularly with regard to pore water pressures (Hencher et al, 1984). For this reason, back analysis may be just as useful in permitting a rational, qualitative assessment of the failure mechanism as in deriving information specifically on shear strength parameters for use in design (Hencher & Martin, 1984).

Whilst the back analysis approach serves as a useful yardstick for assessing the mass shear strength, this approach should not be used in isolation. The application of the back analysis approach for the design of slope preventive or remedial works should be accompanied by rigorous geotechnical investigation, including ground investigation, laboratory testing and engineering geological mapping in order to provide a more reasonable basis for making judgement on suitable mass shear strength parameters. [Amd GG2/01/2017]

Extreme care must be taken to avoid the adoption of unduly high shear strength parameters derived from back analysis which may have incorporated the effect of suction.

Where the ground mass is relatively homogeneous with no corestones or boulders, binding effect of tree roots, three-dimensional effects of the slope geometry, etc., back-analysed shear strength parameters which are higher than those determined from laboratory tests on saturated

specimens should not be used. [Amd GG2/01/2017]

The notion that the continued stability of an existing slope may be proven by past rainstorms should be treated with caution. Factors such as deterioration of the slope condition, progressive slope deformation and possible changes in environmental conditions (e.g. enhanced infiltration through a dilapidated hard surface cover, leakage from water-carrying services, etc.) should be carefully considered before one can confidently count on past performance in the use of the back analysis approach. In particular, the use of the back analysis approach to derive mass shear strength parameters for designing upgrading works should be done with extreme caution where the proposed works involve the replacement of a hard surface cover with a

vegetated cover. [Amd GG2/01/2017]

When back analysis is used, it should be for deriving the possible range of strength parameters. If more than one predominant soil type (e.g. fill and CDG) is involved, the back analysis approach, if used, should be applied to assess the strength parameters for the particular soil type with greater uncertainty (e.g. high spatial variability, difficulty in retrieving representative samples for laboratory testing, etc.), whilst the strength parameters for the other soil type(s) should be determined by other means such as conventional laboratory tests.

[Amd GG2/01/2017]

The back-analysed shear strength parameters for a certain slope section should not be applied to other sections of the slope in the case of a sizeable slope with highly varied characteristics (e.g. in terms of slope geometry, geology, catchment characteristics, etc.) along its length. Different sets of strength parameters may need to be obtained by back-analysing

suitably selected sections. [Amd GG2/01/2017]

Back analysis of a landslide or a slope with signs of distress can be useful where there is sufficient and reliable information on the ground conditions and the geometry of the rupture surface. This can assist in the diagnosis of failure mechanisms and examination of the likely combinations of soil shear strength and groundwater conditions at the time of failure. For the design of remedial works, however, the back-analysed groundwater conditions prevailing at the time of the landslide should not necessarily be taken as the design groundwater conditions.

This is because the groundwater response for rainstorms with differing characteristics in terms of intensity, duration and return period may be different, and because the hydrogeological setting of the failed slope may be altered by the landslide and/or the proposed remedial works.

[Amd GG2/01/2017]

The above guidance on the use of the back analysis approach is also generally applicable to the design of preventive or remedial works for retaining walls. [Amd GG2/01/2017]

32.3 Other Cases

Although back analysis is carried out typically when slope failures or significant ground movements have occurred, it can be useful in other cases where conventional predictive methods may not lead to realistic design solutions. An example of this application in Hong Kong is in the design of preventive works for existing steep slopes formed in soils derived from insitu rock weathering. In such cases, conventional slope stability analysis will often yield factors of safety less than unity even when a slope has stood safely without signs of distress for many years. Shear strength parameters obtained from back analysis of hypothetical failure

surfaces through the existing slope often allow a more realistic form of stability improvement to be made than would be possible from conventional analysis. However, it is very important to check that the assumptions made in the back analysis are valid; for example, the failure surfaces selected should be realistic and the proposed works should not result in any substantial change to the form of or loadings on the slope. Similarly, in the prediction of settlements in variable ground, or in ground from which it is impossible to retrieve representative samples, back analysis of settlement data from an adjacent site in similar materials may be the only satisfactory means of producing sensible predictions.