Geological Uncertainties

Due to the complex character of geological materials there is always an element of uncertainty in geological investigation and testing. In this connection the following terminology is often used:

• Uncertainty or lack of absolute sureness. In geology it means that observations, measurements, calculations and evaluations are not reliable. As a consequence, the use of geological data often may involve some form of guesswork.
• Error is defined as the difference between computed or estimated result and the true value.
• A bias is the difference between the estimated value and the true value based on statistically random sampling. For example, joints sub-parallel to an outcrop have less chance of being sampled than joints perpendicular to an outcrop. This is a bias in sampling for orientation.

Geotechnical design problems often originate from a lack of knowledge, which contributes to uncertainties and difficulties in determining in advance the actual geology and the behaviour of a geotechnical structure. A lack of knowledge of the design conditions becomes evident in the execution phase, something that also was acknowledged by Terzaghi and Peck (1948) when they formulated the observational method. This procedure has been accepted and formalized in the new Eurocode 7 for geotechnical problems.

In principle the uncertainties in data are depending on the extent of investigations. However, in tunnelling it may be impossible to carry out any detailed investigate in advance. The most optimal solution may then be to carry out investigations during the excavation and apply the results to adjust the design if necessary.

Some comments on uncertainties can be seen here.
The paper Rock engineering and tunnelling – A Nordic approach includes description on geological uncertainty. 

Collapses of tunnels
Slides, cave-ins, flowing ground which sometimes occur in tunnels are some of the unexpected problems during tunnel excavations. Some comments on this are shown in the paper Collection of geo-data – limitations and uncertainties. 

See also abstract of the paper on ‘Slides in Norwegian water tunnels‘ and short descriptions on the tunnel collapses at Vinstra HEP  and Sundsbarm HEP.

Conclusion
Unexpected events are often caused by geological features because of lack of information from field investigations and collection:

1. Although extensive field investigation and good quality descriptions will enable the engineering geologist to predict the behaviour of a tunnel more accurately, it cannot eliminate the risk of encountering unexpected features.
2. A good quality characterisation of the rock mass will, however, in all cases except for wrong or incorrect interpretations, improve the quality of the geological input data for evaluation and analyses, and hence lead to better design.
3. The methods, effort and costs of collecting geo-data should always be balanced against the probable uncertainties and errors.

There are generally great difficulties to perform investigation from the terrain surface for a tunnel hundreds of metres below surface.  As long as the tunnel construction is found feasible, some of the investigations may be performed later from the tunnel face during tunnel excavation.

Some of the impact from uncertainty can be reduced by flexibility in tunnelling contracts by a risk-sharing system has obvious advantages. 

More on geological uncertainties is presented in the Rock Engineering book