55 Proc. 2nd International Conference on Problematic Soils, Malaysia, Dec. 2006, Singapore: CIPremier, pp. 5562 BONDED SOILS: PROBLEMATIC OR PREDICTABLE? D G Toll, Durham University, United Kingdom V Malandraki, formerly Durham University, United Kingdom Z Ali Rahman, Universiti Kebangsan Malaysia (formerly Durham University) D Gallipoli, Durham University, United Kingdom Abstract Bonded soils are often thought about as problematic soils as they do not fit into the frameworks of behaviour that have been developed for remoulded or unstructured soils. In reality, the role of bonding enhances the strength and stiffness of the soil. A comprehensive programme of triaxial testing has been carried out on artificially bonded sand in order to develop a framework of behaviour for bonded soils. The results of the tests are interpreted to investigate the conditions under which yielding of the bonds will occur. It is shown that bond breakdown can be predicted from a knowledge of the yield loci that define conditions under which yield will occur. It is found that a unique Critical State Line is not achieved, but that the Critical State is defined by a band of void ratios. Therefore, the ultimate state is difficult to predict for such a material. Nevertheless, normalisation of the data is possible, which is helpful in interpreting the behaviour. It is shown that, although complex, bonded soil behaviour can be explained, at least in part, by the frameworks of behaviour used for unstructured soils. Keywords: Bonded soil, triaxial tests, yield, Critical State 1. Introduction One of the most difficult areas of ground engineering involves what are generally considered to be the borderline materials between soils and rocks; those materials sometimes described as “hard soils soft rocks”. As Johnston & Novello (1994) have pointed out, we should not consider such materials as “borderline” since they represent the centre ground of a continuous spectrum ranging from uncemented soils at one end through to 'hard' rocks at the other. In reality, many natural soils are structured due to bonds between their particles (Leroueil and Vaughan, 1990) which means that they cannot be treated as purely particulate materials. Particulate soils reach this bonded state by diagenesis or secondary cementing. Weathered rocks move towards a weakly bonded state as weathering leaches out material and weakens the cementing bonds. Therefore it is essential we develop frameworks of behaviour that can describe the behaviour of weakly bonded geomaterials. They are often thought of as problematic soils as they do not easily fit into the frameworks of behaviour that have been developed for remoulded or unstructured soils. This paper considers some aspects of behaviour that will help us to understand bonded soils. It is based on a set of triaxial test data on artificially bonded sand. Similar studies looking at Critical State frameworks of behaviour for bonded soils (e.g. Coop and Atkinson, 1993; Cuccovillo and Coop, 1999; Cotecchia and Chandler, 2000) have focussed on soils that would be classified as metastable or stablecontractant according to the classification given by Vaughan (1988) i.e. they exist in a loose state that would be unsustainable except for the bonded structure (metastable) (and hence show large collapse deformations) or those that contract during shear. Such soils approach the Critical State from the “loose” side. In contrast, the data presented here is for a soil in the stabledilatant state. Hamidi and Haeri (2005) have also focused on this region. Such a soil does not show collapse or contraction due to bond breakdown, instead the behaviour is highly dilatant once the bonds start to break. Nevertheless, the bonded structure still plays a vital role in controlling the strength and stiffness of such a material.