Technical Report NTB 90-07

Grimsel Test SiteGeophysical methods for the detection of discontinuities ahead of a tunnel face(out of print)

In 1988 NAGRA started a research and development program at Grimsel Test Site (GTS) to investigate the potential of geophysical remote sensing methods (radar and seismics) to predict discontinuities ahead of the tunnel face.

This task was initiated as an attempt to provide a more complete knowledge of the rockmass for a safe construction of a tunnel. If we can tell in advance where the main fracture zones are to be encountered during excavations, the safety risks as well costs of tunneling will be greatly reduced.

A number of practical goals were set to guide the studies and experiments during the project:

  • The measurements should be performed during breaks of the normal construction operation. No additional stand-by times should result for construction work.
  • The investigation depth should be in the order of at least 100 m to allow enough time for data processing and interpretation before the tunnel arrives at the located discontinuities.
  • The preparation for the measurements should not interrupt or interfere with the construction work.
  • The method should be applicable for conventional drill and blast techniques as well as for full face drilling operations

The first investigations were done using a subset from an existing data base, obtained from both radar and seismic experiments, originally recorded for a tomography project at GTS. Although the data recording, station spacing etc. were optimized for the tomographic application and far from ideal for a reflection survey, it was possible to reach some meaningful results. Reflection events were identified which could be attributed to known fracture zones and lamprophyre dykes.

Radar measurements give good results in rocks with high electrical resistance (e.g. granite, rock salt) but have a very limited investigation depth in materials with lower resistance (e.g. marls, clays). NAGRA's near future interests being focused on measurements in sediments and considering the results of the first tests, we decided to concentrate our efforts on seismics rather than radar.

After discussions with several experts, it was agreed that in this case the most adequate approach for the seismic measurements is a modified "Vertical Seismic Profiling (VSP)" technique which is well known in borehole geophysics. A more suggestive name could have been, for example, "Horizontal Tunnel Profiling" but as we liberally refer to routines developed initially for boreholes, we prefer to conserve the generic name of VSP. For a more objective evaluation of the results, conventional seismic measurements along the tunnel were also recommended.

Theoretical calculations (FD-modeling) were done to determine the optimal test layout. This study showed that, ideally, the source and the receivers should be placed at a certain distance (several meters) away from the tunnel wall. In practice this is hardly possible as it would require many boreholes to be drilled radially from the tunnel. The best compromise between theoretical and operational aspects is to place the source into a side wall drillhole at a minimum distance of about 4 m from the tunnel wall and to attach the receivers directly to the tunnel wall. In this case we expect to reduce the tunnel wave and still achieve a realistic operational time. However, as the measurements at GTS were primarily a scientific experiment, we decided not to measure only the most promising layout according to the theory, but to proceed with different test configurations to prove the validity of the theoretical results optimize the operational routine.

The following VSP configurations were tested:

  1. Receiver chain with 7 receivers in a radial borehole and a sledge hammer as the source. The shot points were placed along the tunnel with a spacing of 0.7 m.
  2. Receiver chain with 7 receivers in a radial borehole and small explosives as the source. The shot points were placed along the tunnel with a spacing of 0.7 m. The shots were fired in small holes (l= 400 mm, Φ=14 mm).
  3. Explosives in the radial borehole and receivers along the tunnel. The receiver spacing was 0.7 m.

Additionally to the VSP-type measurements we measured a conventional reflection seismic profile along the tunnel by shooting through the array. Shotpoint and receiver spacings were 0.7 m with an offset between shot and receiver of 0.35 m, leading to a CDP (Common Depth Point) spacing of 0.35 m.

The raw sections of all records showed only very weak reflections. This was caused by the very low reflectivity of the fracture zones at GTS, if compared with surveys conducted in layered sediments. Accordingly, the processing consisted of a relatively large number of steps which are not discussed in this paper in detail. In general, the direct P- and S-waves as well as the reflected S-waves were suppressed by using a median filter while the remaining P- reflections were enhanced by Image Point filtering, a method based on the generalized Radon transform developed by C. Cosma and his collaborators at Vibrometric.

A large effort has been put for the completion of this project in the development of routines for translating the geophysical results into a reliable and unambiguous geological picture. Due to the many experiments carried out previously at Grimsel Test Site (GTS), we had the possibility to compare in different phases the geophysical predictions expressed in terms of geological structure with the exhaustive information available. The agreement between predicted and real location was excellent. We were able to detect discontinuities up to distances of about 150 m.

The conventional reflection section was processed using standard seismic processing routines. Due to the weak source (sledge hammer) the investigation depth was in the order of about 40 m. Within this area a number of reflectors (mainly lamprophyre dykes) were detected. Additionally we got good reflections from a nearby tunnel. Although this test was successful we consider the VSP configuration as the better layout for the prediction measurements.

As a trial of the method in a realistic environment, VSP measurements were conducted in August 1990 in the Leissigen tunnel (a road tunnel in the canton of Bern), which was under construction at the time of the test. The processing sequences tested in earlier experiments were used. Although the circumstances were more complicated and the field conditions more difficult than in GTS, the predictions coincided well with the geological structures known or supposed to exist at the tunnel site.