Manufacturing of physical models by 3D-scanning and -printing for fracture flow tests

Visual observation of flow and transport processes in fractures requires transparent replicas. Quite easily realized are parallel plate models, which pose only a quite rough approximation, though, and require certain geometric conditions. A better representation can be gained by impressions from real fractures, either by forming the free space, a common technique but rather effortful, or by epoxi imprints of the fracture surfaces. Accurate surface measurements and determination of contact pressures indicate, though, that several imprints of the same locations may show significant differences.

A rather new class of transparent physical models has been made possible with the introduction of reasonably accurate 3D-printers in combination with transparent resins.  Hydraulic tests with principle models of single fractures as well as DFNs have been established quite early. Realistic single fracture replicas still pose a problem, though.

Three steps are required to produce a fracture replica by this method:
1. 3D-scanning of the fracture surfaces
2. Preparation of a printable digital model of the fracture
3. 3D-printing of the digital model

This procedure has a lot of appeal as repeated printing of the same physical fracture model allows for parallel as well as repeated destructive tests.  Additionally, it rather elegantly avoids the problem of air enclosure and bubble evolution between resin and fracture surface which is a common problem with resin imprints. Moreover, it is possible to add features to the digital model that facilitate hydraulic tests such as connectors to inflow and outflow tubes. Since it was intended to cover the whole production process of this method, a 3D-scanner as well as a 3D-printer have been acquired accordingly.

However, new challenges appear also at all three stages of production. One obvious point is the accuracy. The coordinates of a fracture surface can of course only be sampled at a limited number of scanning points. On the same scale, also the dimensional accuracy of the 3D-Printer is restricted. Less evident is the problem of alignment of the two fracture surfaces. Snapping points of opposing fracture surfaces at a distance of less than one millimetre have been found in printed replicas of 7 x 10 cm size, suggesting a potentially serious impact on the aperture distribution by misalignment. Another point concerns the general ability of plastics to take up water, which affects the resin material to a considerable extent in that weight and size change with time. Details and solutions to these problems are addressed.

In closing, the repeatability of an actual tracer test in a printed fracture replica is investigated. The experimental setup consists of an upper and a lower part. Transport of a colored solution in the fully water-filled replica has been observed with an industrial camera and repeated three times. By post-processing the probability of the presence of the tracer at each pixel was evaluated. The resulting video reveals a reasonable degree of repeatability.