Reverse modelling of natural rock joints using 3D scanning and 3D printing

Abstract

In order to overcome the deficiency of natural joint specimens with the same surface morphology for experimental studies, we present a technical method for replicating natural joint specimens that incorporates two advanced techniques – three-dimensional (3D) scanning and 3D printing – using computer-aided design (CAD) as the ‘bridge’. This method uses an optical scanning apparatus and CAD techniques to reconstruct a virtual joint specimen with the natural rock’s joint morphology, and a 3D printer then manufactures a physical mould based on the virtual joint specimen for casting concrete or plastic specimens quickly and accurately. Quality verification clearly indicated that this method reduces the experimental errors originating from the discrepancies between replicating specimens containing natural joint’s morphology.

Introduction

The strength and stability of rock masses are significantly influenced by the presence of inherent joints and fractures. As has been thoroughly demonstrated, shearing failure is one of the common features of rock masses, and hence the shear strength that affects the deformation along natural joints or discontinuities is crucial to the stability of geotechnical engineering projects, such as underground tunnels, rock slopes and open-cut mines [1], [2], [3], [4], [5], [6], [7]. In the first ISRM International Congress on Rock Mechanics in 1966, Muller stated, ‘The deformation resistance of the material bridges takes effect at much smaller deformations than the joint friction: this joint friction makes up partly for lost strength’ [8], [9]. In the recent years, the frictional behaviour of rock joints has been of interest to many researchers and engineers [10], [11], [12], [13], [14], [15]. Shear and compression experiments are the most important ways to understand the mechanical behaviours of natural rock joints, such as shearing stress–displacement curves with peak strength, non-uniform shearing damage of natural surfaces and non-linear normal deformation during compression [16], [17], [18], [19], [20]. On the basis of these laboratory and field tests, several shearing strength criteria have been proposed to identify the strength of a natural rock joint, such as Patton’s model [21], Ladanyi’s empirical model [1], Barton’s empirical model [10], Amadei–Saeb’s analytical model [22], [23] and Grasselli’s three-dimensional (3D) model [24]. All of these distinguished works have indicated that the surface morphology of a rock joint, quantified as joint roughness coefficient (JRC), plays a key role in its shearing strength.

During the experimental investigation of natural rock joints, a good experimental procedure requires that the variables in the experiment can be controlled such that only one variable can be isolated and selectively changed. Because no two natural rock joint samples, even those from the same deposit, are truly identical, classical experiments cannot be conducted if the testing scheme requires multiple specimens [25], [26], [27], [28]. Therefore, the lack of sufficient joint specimens with the same natural surface morphology has always limited experimental studies. At present, joint samples are mainly produced using one of the following three general techniques [6], [14], [17], [29], [30], [31], [32], [33], [34], [35]: (i) tensile fractures, usually made in a manner similar to the Brazilian test, (ii) sawn flat joints with undulated or irregular surfaces or (iii) casts of natural or stylized joints with silicon moulds, mated silicon rubber and aluminium moulds.

Despite the widespread acceptance of the joint specimens produced using these techniques, there is an empirical concern that these methods lead to objective errors and increased time consumption, because most of these methods are unable to represent natural joints with complicated and irregular surface shapes or digitize their surface. This does not mean that the aforementioned traditional methods are inappropriate for rock joint behaviour studies, but rather they are more appropriate for local application with a lesser quantity than as the basis of a universal system [32], [36], [37], [38]. Moreover, these traditional methods are inconvenient for data collection and statistical analysis for joint shapes. In fact, the current dilemma in the ways of producing duplicate specimens has reduced the interest towards intensive studies on the mechanical properties of natural rock joints.

The recently developed 3D scanning (3DS) and 3D printing (3DP) techniques, combined with computer-aided design (CAD), acting as a bridge connecting 3DS and 3DP, provide a new way to manufacture experimental specimens with the same irregular joint morphology. The 3DS apparatus can help digitize and characterize the joint surfaces in three dimensions quickly, with high precision and no damage. On the basis of these digitization data for the natural joint, some 3D roughness parameters can be developed to characterize the joint surfaces and overcome the drawbacks of two-dimensional (2D) profiles [39], [40], [41], [42], [43], [44]. Furthermore, 3DP, that is, additive manufacturing or additive layer manufacturing, allows the automated generation of free-form solids directly from a computer file to a real object [45], [46], [47]. This 3DP provides a new way to make a physical mould with the morphology of a natural joint surface, because this technology can be used to print a joint surface with the same shape as the original joint. In addition, CAD can help reconstruct a virtual model body with any shape for experimental use from the scanned data of the natural joint surface. The clonal joint mould can be produced by 3DP using CAD technology. Therefore, on the basis of the printed joint mould, massive experimental specimens with the same natural joint morphology can be replicated for shearing or compression tests.

In this study, a new method for producing natural joint specimens that combines 3DS and 3DP techniques is presented. First, an optical scanning apparatus is used to digitize the 3D surface of a natural rock joint. With the resulting point cloud data, a virtual joint specimen with the natural joint’s morphology and given boundaries is reconstructed using CAD methods. The 3D printer then manufactures a physical mould quickly and accurately based on the virtual specimen. In this way, massive joint specimens can be replicated using concrete or plastic materials according to the printed mould. The error analysis between the original joint data, the printed joint surface and the actual surface of the physical specimen indicates that this reverse method can copy the natural morphology from the original rock joint to the artificial joint specimen with high efficiency and precision. Furthermore, direct shear tests show that the shearing displacement–force curves of the replicated joint specimens were similar to each other with little dispersion. These results clearly indicate that the presented method can reduce the experimental errors caused by the specimens themselves, and open a new door to further experimental study of natural joints.