Lacking appropriate mathematical, mechanical, investigation and testing methods to realistically characterize the ground and predict the ground and system behaviours, for many centuries purely empirical methods for tunnel design have been used. Initially the used experience was bound to single persons. Later, formalized procedures have been developed, using a few, relatively easily collectable parameters, and by weighting them, arrive at a ground quality indicator. It is obvious that the complex properties of the ground cannot be adequately described by a single number or expression. Basis for the selection of parameters and weighting and rating is experience under specific conditions, making a use under other conditions questionable. The authors of some of the systems extended their classification systems to tunnel design tools by including additional parameters and input in the system. The assignment of e support measures is strongly biased, and does not consider different behavioural modes of the ground and its interaction with excavation and support. Thus tunnel designs based on such systems necessarily are inaccurate and sometimes even entirely wrong. From the legal point of view, the sole use of "classification systems" for tunnel design is inacceptable, as there is no proof of stability or serviceability. Due to the reasons mentioned above and a number of other shortcomings, like oversimplification, so called "classification systems" are not adequate to the problem, leading to inadequate designs, and thus to uneconomical construction.
The reasons for establishing a classification system can be manifold. The main purpose generally is to group items with similar features or properties into one category. This requires the definition of classification criteria. Those criteria need to be clear and meaningful. It is quite obvious that a small number of classification criteria leads to a quite heterogeneous distribution of individual members of one category.
As an example the classification of children, which should enter school may serve. The main criterion in most countries is that they are six years of age. If this is the only criterion applied, any six year old kind falls into the category of school beginners, regardless of their origin, mother tongue, weight and height, and what other differences might exist among that group.
To arrive at a more homogeneous group, additional criteria are required. Naturally the single criteria are not of the same importance, thus the different criteria need to be weighted. The weight of each criterion again depends on which specific requirements the target group should meet. It is quite clear that different persons would put different weights on the single criteria due to their personal preferences or experience. With adding more criteria to apparently arrive at a more homogeneous group the bias increases. There is no guarantee that the group eventually fits together, as one group member may have strong qualifications with respect to some criteria, while another group member has high qualification in other aspects.
The question arises why we need classification at all if the result probably is not satisfying? The answer is quite simple: to enable also inexperienced people to forma group, meeting certain requirements with some degree of probability of success. But while an expert can correct unwanted results from the classification process, a layperson uncritically has to believe that the result is appropriate.
