ACCURATE SUBSURFACE CHARACTERIZATION FOR HIGHWAY APPLICATIONS USING RESISTIVITY INVERSION METHODS

Transcription

1 ACCURATE SUBSURFACE CHARACTERIZATION FOR HIGHWAY APPLICATIONS USING RESISTIVITY INVERSION METHODS Ioannis F. Louis 1, Filippos I. Louis 2 and Melanie Bastou 3 1 Geophysics & Geothermic Division, Geology Department, University of Athens, Panepistimiopolis, Ilissia, Athens 15784, Greece. jlouis@geol.uoa.gr 2 Geophysics & Geothermic Division, Geology Department, University of Athens, Panepistimiopolis, Ilissia, Athens 15784, Greece. flouis@geol.uoa.gr 3 AKTOR Constructing Group, Philellinon 18, Chalandri, Athens 15232, Greece ABSTRACT In cooperation with the AKTOR Constructing Group of Engineers the Geophysics and Geothermic Department of the University of Athens performed an electrical resistivity survey at a section of Ymittos highway in Athens. The survey aimed to image the subsurface structure including karstic voids and to evaluate their extent that can result in potentially dangerous collapse of the road segment overlying these features. A synthetic simulation study preceded the field survey was performed aiming to determine whether the electrical resistivity method could identify such features and to choose the acquisition parameters for optimum detect ability and resolution. The field survey results indicated the presence both of conductive and resistive anomalies along the proposed stretch of the highway. The interpretation of the resistive features indicates that the affected section of the highway overlies five prominent voids attributed mostly to the karstic activity. The conductive anomalies indicate an area where the host limestone rock has been lowered by faulting and is associated with the subsurface continuation of the fault zone observed in the slope of the road during the excavations. Geophysical data interpretation indicated air-filled voids in six drill locations. Drilling results supported our interpretations and a remedial action plan with consolidation grouting works was implemented prior to the construction. INTRODUCTION Karstic voids can affect many highway locations in areas underlain in carbonate rocks causing construction delays, stability problems, and may result in a significant increase of cost. Road and highway subsidence, building foundation collapse, and dam leakage are a few of the problems associated with karstic voids. Structural instability associated with voids can occur as a sudden collapse of the ground surface or as a less catastrophic, but recurring drainage problem. Within karst regions, either scenario can be expensive to design and implement control for present and future structures. Frequently, borings drilled within karst regions do not intersect areas of concern in the subsurface. Misplaced borings can provide inadequate subsurface data, and could misrepresent the subsurface system leading to additional cost for remedial design or additional investigation. Rapid reconnaissance surveys using remote sensing and surface geophysical techniques integrated with a boring plan are best used to aid in the proper location of test borings to identify subsurface features related to karst development. Journal of Electrical & Electronics Engineering, Special Issue October 2002,

2 There are a number of geophysical techniques that can be used to detect the presence of voids below the surface. The application of 2D resistivity imaging methods to such geotechnical problems and site characterization is well documented in literatures (El- Hussein et al., 2000; Tsourlos et al., 1998; Dahlin, 1996; Noel, Xu, 1992; Louis et al., 2002). The development of resistivity imaging software have allowed for more costeffective resistivity surveys and better representation of the subsurface. During the construction of Western Ymittos Peripheral Highway in Athens, the AKTOR Constructing Group of Engineers encountered underground karstic voids at location This became a major concern to engineers, causing a temporary stop of the contract work at this site. In support of the subsurface investigation the Department of Geophysics and Geothermic of the University of Athens (Louis, 2001) conducted a high-resolution shallow geophysical survey to identify and map the extent of the underground voids, or weak permeable or highly fractured zones that may contribute to stability problems in the specific portion of the under construction highway. Due to the presence of cultural noise, the 2D resistivity imaging method was selected for the subsurface investigations instead of other geophysical methods. Moreover, the resistivity contrast of air or water filled voids compared to the surrounding limestone bedrock combined with the advantage that the method is not time consuming for small scale projects in both pre- and post- acquisition processing steps, makes it most appropriate for prospecting. GEOLOGY OF THE AREA The greater area is mainly composed of the geotectonic sub unit of Attica or the zone of Ymittos South Attica and is the extension to the south of the northeast Attica- Almyropotamos zone (Katsikatsos, 1992). The rock formations, oldest to newest, which are detected in Ymittos South Attica zone are Vari Schist (Attica), Pirnari Dolomite (Attica), lower formation marble, Kaisariani Schist (Attica), and upper formation marble. Above these formations overthrust the formations of the neohellenic tectonic cover, a section of which constitutes the sub zone of Lavrio or the phyllitic tectonic cover zone as it called today (Katsikatsos, 1992). The area under investigation is located on the northeast side of Ymittos Mountain and constitutes recrystallized limestones of the phyllitic cover. THE METHOD A non-destructive high-resolution resistivity imaging survey was carried out with the primary objective to locate evidence for near surface karstic features responsible for possible ground failures at the specific section of the highway. 44

3 In electrical resistivity imaging applications current is introduced into the ground through one pair of electrodes. A second pair of electrodes is then used to quantitatively measure the voltage pattern on the surface resulting from the current flow pattern of the first set of electrodes. If multiple electrodes are used and the results recorded automatically the area to be examined can be searched more efficiently, and also probed at various depths at the same time. A fast numerical approach is then used to optimise an initial multi-layer model constructed usually directly from the observed apparent resistivity values. A finite difference or finite element technique is usually used to calculate the 2-D forward response of the model. By subsequent iterations the model is upgraded until a minimum (or an acceptable) rms misfit between the observed and model pseudosection is achieved. Resistivity differences typically correspond to changes in the lithologic composition of subsurface materials or the chemistry of pore fluids. The applied current, the resulting voltage potentials and the electrode geometry are recorded and in general the greater the distance between electrodes the greater the depth of penetration. The potential difference values are plotted in the field as apparent resistivities in pseudo-sections, to verify proper collection of data. These resistivity values are averages over the total current path length and are plotted at one depth point for each source-receiver combination. SYNTHETIC SIMULATION Prior to field-data acquisition a geophysical survey of the same cross-section was simulated to choose the acquisition parameters for optimum detect ability and resolution. This effort had two goals; the first was to test the capability of the method to reconstruct images expected to be present in the subsurface. The second benefit was to use the information obtained from the synthetic-data inversions as constrains helping the interpretation of the field-data inversions. A finite element algorithm (Loke, 1994) was used to calculate the direct current response of a set of resistivity models representing real geological and environmental conditions of the local area. Taking into account the borehole observation log files, made available from the constructing group, concerning information of the revealed karstic voids (Fig. 2), a group of synthetic resistivity models was composed. Several situations representing a major void or a system of karstic features were examined while varying the spatial distribution and geometries of these structures. A resistivity value of ohm-m was chosen to represent the air filed cavity placed on a limestone environment of 5000 ohm-m (Fig. 1). 45

4 Figure 1. Input model for the synthetic simulation consisting of a void feature 3 metres width in a limestone environment. Figure 2. Karstic void revealed during excavations at the site of the highway. The imaging abilities of three different electrode arrays (1, 3 and 5 meters) were examined using the least square inversion technique. Gaussian noise was also added both to background and target models to demonstrate that the inversion scheme is reasonably robust and will work in an environment with unsystematic geologic or instrumental noise. RES2DINV software used for the inversions is based on the smoothness constrain least squares method and basically tries to reduce the difference between the calculated and measured apparent resistivity values with respect to some smoothness constraints such as the complexity of a model. The resulting inversions were compared with the original input models for the three different electrode spacings. In general the inversions gave relatively highresolution images and revealed that the geometries of the resistivity anomalous areas were sufficiently recovered. However it was found that, by increasing the electrode spacing (5m), the inverted resistivities diverged from their initial model values. The 46

5 response of the karstic void of figure 1, for a 3m Wenner-Schlumbereger array, is shown in the upper part of Figure 3. The resistivity image of the inverted response is shown in the lower part of the same figure. In Concluding, synthetic simulation indicated an optimal electrode spacing of 3 meters to delineate the specific targets up to 15m depths with no significant loss in resolution. For a more detailed imaging of the top 5m depth zone an electrode spacing of 1m was decided to be used. If cavity targets are filled or partially filed, their response is expected to overlap with the limestone environment making in that way the target discrimination a more difficult task. Figure 3. Reconstructed resistivity image for a 3m Wenner-Schlumberger array. DATA ACQUISITION AND PROCESSING Three resistivity traverses, namely LINE-1, LINE-2 and LINE-3 were conducted along the survey area. The orientation and extent of those lines are shown in the location map of Figure 4. Each survey line was double scanned using both 1 and 3 meters electrode spacing in order to get a more detailed image of the top zone. The Wenner- Schlumberger configuration was deployed with a maximum N separation (ratio of maximum and minimum electrode spacing) equal to 13. Data were collected with the SYSCAL R1 Plus resistivity meter (IRIS). The apparent resistivity pseudosection produces a distorted image of the subsurface resistivity. Inversion of the field observations is the standard procedure to obtain an estimate of the true resistivity. The true resistivity structure was interpreted using 2DINVS software (Tsourlos, 1995; Tsourlos et al., 1998). This algorithm is based 47

6 on a 2.5D smoothness constrained inversion to invert the apparent resistivity data by employing a quasi-newton technique to reduce the numerical calculations (Loke and Barker 1994). The subsurface is divided into a grid of nodes, thus the inversion is not affected by the geometry of subsurface resistivity anomalies. The purpose of the program is to determine a resistivity of each node such that the apparent resistivity pseudo-sections agree with the actual measurements. The algorithm is iterative and fully automated. The inversion estimates a resistivity model by minimizing the difference between the observed and the calculated data. The smoothness constrained inversion method imposes another condition, namely that the roughness of the resistivity model is minimum. Figure 4. Map showing the study area and the location, orientation of the resistivity survey traverses LINE-1, LINE-2 and LINE-3. RESULTS AND INTERPRETATION Figures 5, 7 and 8 display the outputs for the three parallel traverses. The plots represent parallel slices through the ground to a depth of 20 m below the ground surface. A large resistive anomaly can be clearly seen starting at approximately 13 m below ground surface in all traverses. Separate smaller resistive or conductive anomalies can also be seen at shallow depths in all traverses. 48

7 The high resistive elongated feature appeared in the resistivity section LINE-1 (Fig. 5) at 14 m depth was attributed to a series of two or more different karstic voids separated by limestone mass. The conductive feature marked as F shown in the top 5 meters of the section is attributed to an intensely fracture zone filled with clay and is probably the continuation of the fault (Fig. 6) revealed in the slope of the road during the excavation works. The conductive feature marked as C at 7 m depth is probably attributed to an intensely fracture zone associated with the fault zone. Figure 5. Tomographic image of traverse LINE-1. Figure 6. Fault observed in the slope of the road during the excavation works. The same high resistive elongated feature appeared in the resistivity image of Figure 5 is also observed in the resistivity image of the parallel traverse LINE-2 (Fig. 7) at the same depth and is also attributed to the same source of anomaly. The conductive feature marked as F is also attributed to the continuation of the same fault observed in the slope of the road during the excavation works. The small shallow resistive feature marked as E may be a small void, which was not encountered from drilling D4 since it passed adjacent to the anomaly without to intersect it. 49

8 Figure 7. Tomographic image of traverse LINE-2. Figure 8. Tomographic image of traverse LINE-3. The results obtained from traverse LINE-3 (Fig. 8) were interpreted under suspiciousness having in mind that the field measurements were conducted under difficult weather conditions (high presence of noise due to drizzling). However the conductive feature marked as F, observed in traverses LINE-1 and LINE-2, is also present here indicating the continuation of the fault zone under the road surface. The resistive structure marked as H is also attributed to a possible void. Figures 9, 10 and 11 show a set of horizontal plane sections computed from the resistivity images of traverses LINE-1, LINE-2 and LINE-3 by averaging the data in depth intervals. These clearly illustrate the lateral and in depth extent of the resistive and conductive anomalies observed in the resistivity traverses. 50

9 Figure 9. Horizontal plane section of resistivity distribution at 2.25 m depth. Figure 10. Horizontal plane section of resistivity distribution at 4.25 m depth. 51

10 Figure 11. Horizontal plane section of resistivity distribution at m depth. Figure 12 illustrates how these horizontal plane sections can be arranged vertically as a volume viewed from the south. The picture clearly depicts that the intense fracturing in the limestone mass is reduced to the first 5 to 6 meters depth. In greater depths a more compact limestone mass is prevailing. Figure 12. Diagram showing how depth slices can be stacked vertically to give a volume representation. This plot is viewed from the south. 52

11 Figure 13 illustrates how the resistivity depth sections LINE-1, LINE-2 and LINE-3 can be arranged horizontally as a volume viewed from the south. The picture clearly depicts the trace of the fault zone under the road surface. It also illustrates the horizontal extent of the resistive feature delineated at 14 m depth. Figure 13. Diagram showing how resistivity traverses LINE-1, LINE-2 and LINE-3 can be stacked horizontally to give a volume representation. This plot is viewed from the south. RESULTS OF CORE DRILLING AND REMEDIATION ACTIONS After the interpretation of the data was concluded, we recommended some areas for exploratory drilling. The areas were given a ranking based on the strength of the anomalies, and the confidence in their interpretations. Six locations were chosen for coring and they were all drilled in areas that demonstrated anomalous signatures in the geophysical data. The cores showed that the subsurface consisted of a zone of intensely fracture limestone associated with the presence of air or clay filled karstic voids. This fractured zone is associated with the subsurface continuation of the fault zone observed in the slope of the road during the excavation works. Three of the cores encountered voids, while the other three encountered heavily fractured bedrock. Drilling D1 intersected a void between 14.5 and 18 meters deep, which is in a very good agreement with the location of the resistive feature marked as A in the resistivity section LINE-1. The same drilling encountered a void between 7.5 and 9 meters deep, which is also in a 53

12 very good agreement with the resistive feature marked as C in the same resistivity section. Drilling D2 penetrated compact limestone up to a depth of 18 meters without to intersect voids. Drilling D3 encountered the roof of a karstic void between 11.5 and 12 where drilling was completed. The location of this void is also in a very good agreement with the resistive feature B in the resistivity section LINE-1. The results of borings D1, D2 and D3 reinforce the aspect that the deep elongated resistive feature observed in traverses LINE-1 and LINE-2 consists of two or more separate karstic voids. Drillings D4 and D5 were completed at the depth of 10 meters without to give a chance to verify the nature of the subsurface resistive feature D observed in section LINE-2. Drilling D6 intersected a karstic void between 8.6 and 9.8 meters and then limestone up to the depth of 10 meters where it was completed. The resistive feature H in section LINE-3 is partly confirmed by the drilling results. This observed discrepancy is mainly attributed to the poor quality of the specific geophysical dataset as it was stated before. The geophysical data indicated problems at all six drill locations. Three of these cores indicated problems that were repaired with remedial grouting works prior to construction of the highway in this area. The remaining cores did not reach the subsurface resistivity anomalies to verify their nature. CONCLUSIONS Because of the high resistivity response of air-filled karstic voids or intensely fractured rock within compact limestone environment, high resolution electrical resistivity imaging can be a very effective tool for locating underground voids or fractured zones. These features can affect many highway locations in areas underlain in carbonate rocks causing construction delays, stability problems, and may result in a significant increase of cost. In cooperation with the AKTOR Constructing Group of Engineers The Geophysics and Geothermic Department of the University of Athens performed an electrical resistivity survey at a section of the Ymittos highway in Athens to image the subsurface geologic structure including karstic voids and intense fractured zones and to evaluate the extent of the subsidence zone that may occur due to void collapse. Prior to field data acquisition a synthetic simulation study was performed to choose the acquisition parameters for optimum detect ability and resolution. The field survey results indicated both conductive and very high resistivity anomalies along the proposed stretch of the highway. The high resistivity anomalies, attributed to the presence of karstic voids, were verified by the drilling results in the majority of the cases. The low resistivity (conductive) anomalies indicated areas where the host limestone rock has been lowered by faulting and thus containing water, making the area less resistive to electrical current. This zone is associated with the subsurface continuation of the fault zone observed in the slope of the road during the excavation works. The interpretation of geophysical data indicated air-filled karstic voids in six drill locations. The core control supported our interpretations and a remedial action plan 54

13 with consolidation grouting works was implemented prior to the construction of the highway in this area. REFERENCES Dahlin, T., D resistivity surveying for environmental and engineering applications. First Break, 14(7), Hussein, I. E., Kraemer, G. and Myers, R Geophysical characterization of a proposed street extension in Cape Girardeau, Missouri. Proceedings of the First International Conference on the Application of Geophysical Methodologies & NDT to Transportation Facilities and Infrastructure Katsikatsos, G., Geology of Greece, lecture notes. University of Patras. Loke, M. H. and Barker, R. D., Rapid least-squares inversion of apparent resistivity pseudo-sections. Extended Abstracts of Papers 56 th EAGE Meeting Vienna, Austria 6-10 June 1994,1002. Louis, I. F., Geophysical site assessment at the Ymittos Western Peripheral Highway (Attiki Odos). Technical report, Athens December Louis, I. F., Karastathis, C. V., Vafidis, P. A. and Louis, F. I., Resistivity modelling and imaging methods for mapping near-surface features: Application to a site characterization at the Ancient Temple of Olympian Zeus in Athens. Journal of the Balkan Geophysical Society (In Press). Noel, M. and Xu, B., Cave detection using electrical resistivity tomography. Cave Science 19, 91Ð94. Sensors and Software, Inc (1996), Pulse Ekko Tools User Guide Version Technical Manual 22. Sasaki, Yutaka, Resolution of resistivity tomography inferred from numerical simulation. Geophysical Prospecting, V. 40, pp Tsourlos P., Modeling, interpretation and inversion of multi-electrode resistivity survey data. Ph.D. Thesis, University of York. Tsourlos P., Szymanski J., and Tsokas G., A smoothness constrained algorithm for the fast 2D inversion of DC resistivity and induced polarization data. Journal of the Balkan Geophysical Society, 1,