Suitability of Ground Penetrating Radar for Locating Large ... · Images were placed on measurement position in 3D presentation ... The work in Pöyry Finland has been carried out

P O S I V A O Y

O l k i l u o t o

F I -27160 EURAJOKI , F INLAND

Tel +358-2-8372 31

Fax +358-2-8372 3809

Eero He ikk inen

Pekka Kant ia

December 2011

Work ing Repor t 2011 -92

Suitability of Ground Penetrating Radarfor Locating Large Fractures

December 2011

Working Reports contain information on work in progress

or pending completion.

Eero He ikk inen

Pöyry F in l and Oy

Pekka Kant ia

Roadscanne rs Oy

Work ing Report 2011 -92

Suitability of Ground Penetrating Radarfor Locating Large Fractures

Base maps: ©National Land Survey, permission 41/MML/11

ABSTRACT

Posiva Oy is responsible for preparation of final disposal of spent nuclear fuel in Olkiluoto. The knowledge about existing network of fractures is important for the safety and feasibility of the final repository. The bedrock properties essential for safety case are analysed in investigations of Rock Suitability Criteria (RSC). One subtask in RSC is avoidance of large (long) fractures adjacent to disposal holes. The long fractures have been defined in tunnel mapping to indicate tunnel cross-cutting features (TCF) or full perimeter intersections (FPI). Suitability of ground penetrating radar (GPR) method for locating large fractures was assessed. The assessment used data measured with 100 MHz and 270 MHz radar tool on ONKALO access tunnel right-hand wall, chainage 3344 – 3578 and on TKU-3 niche floor chainage 15 – 55 and 25 – 67 m. GPR images were processed to enhance reflections and suppress interference and diffractions. Images were placed on measurement position in 3D presentation software. The tunnel wall and floor mapping data was presented along with GPR images. A review of observed GPR reflections, and assessment of visibility of large fractures, was drawn on basis of 3D view examination. The GPR tool can detect reflections from cleaned and dry rock floor and wall. Depth of penetration is 8-12 m for 270 MHz antenna. The antenna has high resolution. Coupling on rock surface is good, which suppresses ringing and interference. Penetration is 20-24 m for 100 MHz antenna, which has a trade off of higher interference due to weaker contact to surface caused by large antenna. There are observed many kind of reflecting surfaces and diffractors in the images, like for example lithological contacts and high grade shearing, and also fractures. Proper manner to apply the method is to use raw and processed images during geological mapping to confirm the origin of reflections. Reflections deemed to be caused by fractures are useful to be compiled to 3D model objects. The orientation of observed reflection trace (hyperbola) is forming an apparent angle with the tunnel axis. The hyperbola can be interpreted to intersection (alpha) angle of the fracture plane with the tunnel axis. Care shall be taken to compute the orientation of the reflector correctly. Migration will image the reflection to correct position if intersection is perpendicular. Migration will suppress, however, vertical reflectors. GPR method is suitable for reflector detection in the rock mass from tunnel surfaces. Reflectors can be further correlated to various geological features in mapping and modelling. Method is time and cost efficient to apply and it can provide imaging of reflectors to distances of 8-20 m from tunnel. Reflectors which are confirmed to be large fractures, can be traced for their visible length depending on observation geometry. Keywords: Ground penetrating radar, spent nuclear fuel disposal, rock suitability criteria, large fractures, crystalline bedrock, tunnel mapping.

Maatutkamenetelmän soveltuvuus jatkuvien rakojen paikantamiseen TIIVISTELMÄ

Posiva Oy huolehtii käytetyn ydinpolttoaineen loppusijoituksesta Olkiluodossa. Loppu-sijoituksen turvallisuusperustelujen kannalta keskeisiä kallioperäominaisuuksia selvite-tään kalliotilojen soveltuvuuden (Rock Suitability Criteria, RSC) tutkimushankkeessa. Yksi RSC osa-alueita on jatkuvien rakojen välttäminen loppusijoitusreikien välittö-mässä läheisyydessä. Jatkuvat raot on tunnelin kartoituksessa määritetty tarkoittamaan koko tunnelin leikkaavaa rakojälkeä, Tunnel Cross-Cutting Feature (TCF) tai Full Perimeter Intersection (FPI). Maatutkamenetelmän soveltuvuutta jatkuvien rakojen paikantamiseen selvitettiin. Selvi-tyksessä käytettiin 100 MHz ja 270 MHz kalustolla ONKALOn ajotunnelin oikealla seinällä paaluvälillä 3344 – 3578 sekä kuprikan TKU-3 lattiassa paaluväleillä 15- 55 ja 25 – 67 m mitattua dataa. Tutkakuvat käsiteltiin heijastusten tuomiseksi esille sekä häiriöiden ja sironnan vähentä-miseksi. Kuvat sijoitettiin mittausten sijainteihin 3D visualisointiohjelmistossa. Tunne-lin seinän ja lattian kartoitustiedot esitettiin tutkakuvien rinnalla. Tutkaheijastuksia sekä jatkuvien rakojen näkyvyyttä arvioitiin 3D esityksen tarkastelun avulla. Tuloksena voidaan todeta että maatutkamenetelmä pystyy havaitsemaan heijas-tuksia puhtaalta ja kuivalta lattialta tai seinältä tehdyllä mittauksella. Menetelmän syvyysulottuvuus on 8-12 m käyttäen 270 MHz antennia, jonka erotuskyky on hyvä ja häiriötaso matala hyvän kontaktin ansiosta. Matalamman 100 MHz taajuuden kalustolla tutkimussyvyys on 20-24 m, mutta häiriötaso on suuremman antennin heikomman suo-jauksen ja huonomman kontaktin vuoksi korkeampi. Maatutkan heijastuskuvissa havaitaan erilaisia heijastuspintoja ja sirontaa. Nämä voivat aiheutua esimerkiksi raoista, kivilajirajoista ja hiertovyöhykkeistä. Tehokas menetelmän käyttötapa on tarkastella käsiteltyjä tuloksia geologisen kartoituksen yhteydessä, jotta heijastusten alkuperä voidaan varmentaa. Jatkuvien rakojen aiheuttamiksi todetut heijas-tajat voidaan esittää 3D mallissa. Havaitut heijastusjäljet (hyperbeli) muodostavat tunnelin pituusakselin kanssa näennäisen kulman, josta on mahdollista laskea rakotason leikkauskulma (alpha) tunneliin nähden. Heijastajan todellinen asento on hyödyllistä varmentaa laskemalla, Migraation avulla kohtisuoraan leikkaavasta kohteesta saapunut heijastus voidaan kuvata oikeaan asemaan. Migraatio kuitenkin myös vaimentaa pysty-jen rakenteiden heijastuksia. Maatutkamenetelmä soveltuu heijastusten havaitsemiseen kalliomassasta tunnelin pin-noilta käsin. Heijastajia voidaan tarkemmin selvittää kartoituksen ja mallinnuksen avul-la. Menetelmä on nopea ja kustannustehokas käyttää. Sen avulla rakojen jatkuvuutta voi seurata 8-20 m etäisyydelle tunnelista. Niitä heijastajia jotka voidaan yhdistää laajoihin rakoihin, voidaan seurata niiden näkyvältä osuudelta havaintogeometriasta riippuen. Avainsanat: Maatutka, käytetyn ydinpolttoaineen loppusijoitus, kallion soveltuvuus, jatkuvat raot, kiteinen peruskallio, tunnelikartoitus.

1

TABLE OF CONTENTS

ABSTRACT TIIVISTELMÄ

PREFACE ....................................................................................................................... 31 INTRODUCTION .................................................................................................... 5

1.1 Background ................................................................................................... 51.2 Olkiluoto site.................................................................................................. 51.3 RSC programme ........................................................................................... 81.4 SKB-Posiva co-operation project .................................................................. 81.5 Objectives and scope of the report ............................................................. 10

2 GROUND PENETRATING RADAR REFLECTION SURVEY .............................. 132.1 Basics.......................................................................................................... 132.2 Uncertainties ............................................................................................... 132.3 Surveys ....................................................................................................... 152.4 Data............................................................................................................. 20

3 PROCESSING AND PRESENTATION ................................................................ 254 GEOLOGICAL PROPERTIES AND REFERENCE DATA OF TARGET

AREA .................................................................................................................... 335 COMPARISONS ................................................................................................... 35

5.1 Review ........................................................................................................ 355.2 GPR on access tunnel wall ......................................................................... 365.3 GPR on TKU-3 floor .................................................................................... 52

6 DISCUSSION AND CONCLUSIONS .................................................................... 57REFERENCES ............................................................................................................. 61

2

3

PREFACE

This work is a part of spent nuclear fuel disposal project of Posiva Oy. The work was co-operated with Swedish Nuclear Fuel and Waste Management Company (SKB), Sweden in joint research project for rock suitability criteria (RSC). Reporting was carried out in Pöyry Finland Oy. Field work was conducted by Roadscanners Oy.

Process and the results have been communicated with geophysicists Mari Lahti and Sonja Sireni, and geologists Paula Kosunen and Antti Joutsen in Posiva, geologist Jussi Mattila in Geological Survey of Finland and geophysicist Tomas Lehtimäki in SKB. Antti Joutsen reviewed the geological features from tunnel mapping data.

The work in Pöyry Finland has been carried out by geophysicist Eero Heikkinen, and the field work by Pekka Kantia. Mari Lahti and Tomas Lehtimäki participated in organising and conducting the field work in ONKALO.

We like to thank above mentioned contact persons and Turo Ahokas (Pöyry Finland Oy) for their important suggestions.

4

5

1 INTRODUCTION

Posiva Oy in Finland and SKB, the Swedish Nuclear Fuel and Waste Management Company, in Sweden, are both conducting a comprehensive project for the final disposal of spent nuclear fuel. The geologic disposal using a multiple barrier concept (KBS-3) has been developed in Sweden. SKB and Posiva have co-operation related to the research of the three barriers (host rock, bentonite buffer and copper canister). This work has been carried out within the co-operation project "Rock Suitability Criteria – strategies and methodology" that studies the geological, hydrological and geophysical data obtained from the underground research facility ONKALO at Olkiluoto in Eurajoki, Finland. The Swedish experience and expertise has been utilised in the work and in reviewing the results.

1.1 Background

Posiva Oy, jointly owned by Teollisuuden Voima Oyj, Fortum Power and Heat Oy, is responsible for implementing the programme for geological disposal of spent nuclear fuel in Finland. The programme consists of research, technical design and development activities, as well as construction and operation of the disposal facility. In 2000, Government made the Decision in Principle (DiP) for the disposal of spent fuel from TVO and Fortum's reactors on Olkiluoto Island in Eurajoki (Figure 1). Posiva plans to construct a KBS-3 type repository, designed to be situated at a depth of 400 m to 600 m in the bedrock at Olkiluoto. By decision of the Ministry of Trade and Industry (KTM, at present Ministry of Employment and Economy, TEM) made in 2003, Posiva is to submit an application to obtain a construction license for the disposal facility by the end of 2012.

Nuclear power companies in Sweden jointly established the Swedish Nuclear Fuel and Waste Management Company (SKB) in the 1970s. SKB’s assignment is to manage and dispose of all radioactive waste from Swedish nuclear power plants in a way to secure maximum safety for human beings and the environment. The facilities SKB is responsible for include a central interim storage facility for spent nuclear fuel (Clab) near Oskarshamn, and a final repository for short-lived radioactive waste (SFR) in Forsmark. SKB has been conducting advanced research for the final disposal of spent nuclear fuel for thirty years. A siting process was initiated 20 years ago in order to locate a potential repository for the final disposal of spent nuclear fuel. Subsequent analyses and site investigations resulted in the selection of Forsmark in Östhammar municipality in 2009. In March 2011 the applications were submitted to the Swedish Radiation Safety Authority (SSM) and to the Environmental Court to build the Spent Fuel Repository in Forsmark.

1.2 Olkiluoto site

During the past two decades, various investigations have taken place at the Olkiluoto site. Currently, construction of the underground research facility ONKALO provides opportunities to carry out investigations underground. These investigations enable the collection of more detailed information of the rock at repository depth and confirmation of site understanding obtained during previous surface-based investigations, as well as

6

making testing and demonstration of the disposal technology possible. The avoidance of brittle deformation zones, large fractures and transmissive fractures minimise the risks for hampering the long-term safety of the KBS-3 concept.

Figure 1. a) Location of Eurajoki in western Finland. b) Olkiluoto Island and location of the ONKALO area. c) Illustration of the planned repository layout including the access tunnel, shafts and deposition tunnels at -420 m depth. The bedrock of Olkiluoto site area consists of Palaeoproterozoic variably migmatised gneisses and migmatites together with pegmatitic granites and diabase dikes that have been subjected to multi-phase ductile and brittle deformation during the geological history (Aaltonen et al. 2010; Lahti et al.. 2009). The rocks of Olkiluoto area can be divided into four groups 1) migmatitic gneisses, 2) tonalitic-granodioritic-granitic gneisses (TGG) gneisses, 3) other gneisses including mica gneisses, quartz gneisses and mafic gneisses, and 4) pegmatitic granites (Kärki and Paulamäki, 2006). Diatexitic and veined gneiss types are the most prevailing rock types in the Olkiluoto area (Fig. 2). The rock types have also been subjected to extensive hydrothermal alteration, and alteration products such as clay minerals, sulphides and illite show spatial variation in the Olkiluoto area (Aaltonen et al. 2010). A total of 179 brittle deformation zones have been modelled in the Olkiluoto area (Aaltonen et al. 2010). These deformation zones are variable in size, but most of these zones are SE dipping. Ten hydrological zones have been modelled for the site area (Vaittinen et al. 2009). Most of these zones are prominently NE-SW orientated and some of them coincide with the modelled brittle deformation zones known in the area (Posiva Oy 2011) (Figure 3).

Olkiluoto Island in Eurajoki

ONKALO area

Planned repository layout

a) b)

c)

7

1524000

1524000

1526000

1526000

1528000

1528000

6792

000

6792

000

6794

000

6794

000

TGG gneisses

Pegmatitic granites

Diatexitic gneisses

Veined gneiss

Diabase dike

0̄ 1 20.5Kilometers

Figure 2. A bedrock map of the Olkiluoto area.

Figure 3. Showing modelled hydrological (blue) and brittle deformation zones (red) the depth of -420 m ± 15 m. First panel together with technical rooms are also shown.

8

1.3 RSC programme

In the research and development programme TKS-2006 (Posiva 2006), Posiva acknowledged the need for further development of requirements on the host rock and for ensuring their applicability and reliability of these requirements during the actual repository implementation. The Rock Suitability Criteria (RSC) programme has been set up for this purpose (Hellä et al 2009). The RSC-programme aims at developing a host rock classification process to be applied to the repository design and construction. This process includes determination of rock volumes suitable for the repository panels, assessment of whether deposition tunnels or tunnel sections are suitable for locating deposition holes and, finally, acceptance of each deposition hole for disposal. For this purpose, rock suitability criteria are being developed. Together with interpretation, modelling and general understanding of site properties, the criteria can be used to avoid those natural features of the host rock that may be detrimental for the safety of the repository, such as deformation zones, large fractures and fractures with high hydraulic conductivity. The functionality of the RSC classification process in locating suitable rock volumes for deposition holes will be demonstrated in facilities that will be constructed in ONKALO at about -425 m level. The demonstration facilities comprise a central access tunnel, two tunnels (52 and 120 m) and a set of canister holes, which will be constructed simulating the methods and dimensions of the real deposition tunnels and holes. The demonstration will also show how the RSC-process is aligned with design and construction activities and how it functions as a part of the whole deposition process. A detailed-scale model of the rock volume containing the planned RSC-demonstration facilities in ONKALO and parts of the planned first deposition panel will be developed simultaneously with the demonstration. The model will be constructed for the needs of the RSC and will, thus, aim at describing and predicting the rock characteristics relevant in the RSC classification process: brittle deformation zones, large fractures and inflow. The ultimate aim is to develop strategy and methodology for detailed-scale modelling of rock properties during the deposition process.

1.4 SKB-Posiva co-operation project

The ability to predict the occurrence of brittle deformation zones and single large fractures that are able to slip more than 5 cm is of great importance for the application of the RSC rock characterization process. Such displacements are most likely to occur in post-glacial stress regime. Posiva and SKB established a co-operation project called "Rock Suitability Criteria – strategies and methodology" to develop techniques for this purpose. The co-operation project has two objectives related to the RSC. The first objective is to develop and assess geological and geophysical investigation methods for identification of the brittle deformation zones and, especially, the large fractures. This will be carried out by analysing and correlating existing geological and geophysical data acquired from pilot holes drilled in the ONKALO access tunnel and from investigations carried out in the excavated tunnel. The second objective is to develop strategy and methodology for

9

creating a detailed-scale model of the rock conditions relevant to RSC, including, for example, deformation zones and large fractures. The project comprises five activities, which are carried out using data from the ONKALO access tunnel (Figure 4): Activity 1. The use of geological and hydrogeological data from pilot holes for predicting

large fractures Activity 2. The use of geophysical data from pilot holes for predicting large fractures Activity 3. Suitability of seismic investigations for locating large fractures, brittle

deformation zones and hydrogeological zones Activity 4. Suitability of ground penetrating radar for locating large fractures Activity 5. Development of strategy and methodology for detailed-scale modelling work

Figure 4. Illustration of the areal coverage of activities 1-5. Locations of pilot holes OL-PH8 to ONK-PH14 in the ONKALO access tunnel are marked. Activities 1 to 4 deal with data acquisition methodology. Pilot holes are the first source of direct, detailed data from a specific rock volume and it is, thus, of importance to be able to use pilot hole data in locating the relevant geologic structures. Identifying the single large fractures from a pilot hole drill core can be problematic; therefore, Activities 1 and 2 concentrate on pilot hole data and identification of large fractures. In Activity 1, data from the geological logging of several pilot holes from the ONKALO access tunnel are evaluated against tunnel mapping data of corresponding tunnel intervals. The purpose is to correlate tunnel cross-cutting fractures (TCF) or full perimeter intersecting fractures (FPI), (Hellä et al. 2009) with pilot hole fracture data and to determine if they have common geological characteristics. TCF is a fracture that can be observed in both walls and on the ceiling during the geological mapping, and FPI is a synonym for such a feature used in SKB literature (Munier 2006). It is noteworthy that most TCFs (or FPIs) are probably not large fractures but they are considered as possible representatives of large features. Activity 2 concentrates on interpretation of

10

geophysical data obtained from pilot holes and on creating a method for using the data to predict the occurrence of the large fractures. Data on the tunnel cross-cutting fractures are used to evaluate the created method. In Activities 3 and 4, two different geophysical methods are evaluated against the needs of the RSC-program. Activity 3 comprises interpretation, geological correlation and evaluation of data obtained from a seismic investigation carried out in the ONKALO access tunnel with the purpose of testing the suitability of seismic investigation methods for locating local-scale geological structures, for example the large fractures. In Activity 4, interpretation, geological correlation and evaluation of data from ground penetrating radar investigations carried out in a research niche in ONKALO are carried out for the same purpose. Activity 5 concentrates on issues related to modelling of the acquired data and aims at developing a strategy and methodology for detailed-scale modelling of rock properties during the deposition process. This is carried out by constructing a detailed-scale model of the rock volume containing the planned RSC-demonstration facilities.

1.5 Objectives and scope of the report

This report presents the results of Activity 4: Suitability of Ground Penetrating Radar (GPR) for locating large fractures. Different ways in detection of deformation zones and significant or large fractures may be in different phases:

- remote sensing (for example seismic reflection), extrapolation and correlations from nearby survey results, and application of existing models before tunnel pilot hole drilling

- core logging, analysis, and borehole radar reflection survey from pilot hole - tunnel wall mapping (tunnel crosscutting fractures, TCF) and investigations

from tunnel surfaces (like GPR) - detection of fractures from pilot of the disposal hole

Actual fracture length is not easily observable. When tunnel excavation is completed, tunnel mapping can detect fracture trace length on the wall. Fractures oriented perpendicular to tunnel can be avoided in disposal hole placement. Fractures having nearly perpendicular strike to tunnel may be observed in parallel tunnels, and when correlated, their length can be estimated. Fractures which reside below tunnel floor undetected on surface, and intersecting several disposal holes, are more difficult to measure with their continuity. On the other hand, GPR geometry is favourable in detection of these features. Previous radar investigations in ONKALO have included radar survey on tunnel floor covered by macadam. The images were correlated to TCF indications on tunnel wall, but range is limited due to filling (Sipola & Tarvainen 2007). Radar survey was carried out on rock wall in low and intermediate level waste disposal site of Olkiluoto. Feasibility of radar method in Olkiluoto conditions was assessed in Saksa et al. (2001, 2005) and generally in crystalline rock conditions in Olsson et al. (1987).

11

GPR measurements which were used in this report were carried out in three campaigns. On the right-hand wall of the ONKALO access tunnel (when viewing downwards in tunnel) at chainages 3344 – 3578, radar profiles were measured using 100 MHz and 270 MHz antenna frequencies. In a 67 m long niche named TKU-3 excavated north of access tunnel at chainage 3620 m GPR was run during the surveys for excavation damage zone (EDZ) characterization (Mustonen et al. 2010). Measurements were run both on cleaned floor (270 MHz, 500 MHz and 1.5 GHz) and on right-hand wall (270 MHz, 1.5 GHz and multichannel). Surveys and primary processing were carried out during 2009 - 2010 by Pekka Kantia, Tuukka Saikka and Mika Silvast of Roadscanners Oy. Pöyry Finland Oy (Eero Heikkinen) made further processing and review of the data. Pöyry Finland also carried out together with Posiva (Antti Joutsen) the comparisons to tunnel mapping data of ONKALO. Analysis of data was advised and reviewed by Tomas Lehtimäki of SKB. Work was reviewed by Mari Lahti, Sonja Sireni and Paula Kosunen of Posiva Oy. This report consists of description of method, tools, surveys and uncertainties in Chapter 2, description of processing in Chapter 3, description of geological environment in Chapter 4, comparisons to reference data in Chapter 5, and conclusions in Chapter 6.

12

13

2 GROUND PENETRATING RADAR REFLECTION SURVEY

2.1 Basics

Ground penetrating radar (GPR) is a range of geophysical survey methods and tools, which operate with 10 – 3000 MHz radio wave transmission. Radar antenna is used to transmit pulsed or continuous wave train. Antenna can be applied on ground response or from air (detached from the surface). Radiowave is coupled with the ground surface and propagates in the subsurface medium. Wavefield attenuates (decays) according to distance, specific attenuation of medium, scattering and reflections (Zhdanov & Keller 1994). Returning energy amplitude is recorded with similar antenna with respect to time. Time sampling is dense, and recorded time window is adjusted to contain the useful signal. Compilation of radar image requires densely repeated measurements on profile, placed along the surface. Measurements may be triggered on basis of time, or by distance.

2.2 Uncertainties

GRP reflecting events are always linked to physical origin. Sources of uncertainty can be divided to detection and resolution related issues, and conceptual matters on what features, and on which extent, are observed. Detectability of reflecting events in GPR signal is related to attenuation (depth of penetration) which is frequency dependent. Higher frequency has smaller range. Range is limited more in higher conductivity terrain. Also texture of rock mass may limit the range. Energy is scattering more in foliated rock. Detectability of reflecting events is related firstly to the feature’s physical dielectric and conductivity contrast to the background rock mass. Average relative dielectric value is

r = 6 – 6.7 in Olkiluoto migmatite. Considerably higher dielectric value in water ( r = 81), saturated fracture filling ( r = 20-30) or in mafic rock mass ( r = 8) will serve as a good contrast. Electrical resistivity (inverse of conductivity) of the background rock is over 10.000 Ohm.m. Increased conductivity in conductive mineral veins (graphite, pyrrhotite, pyrite) or fracture filling (water, clay, sulphide, graphite) is also causing good contrast. Layers with resistivities lower than 100 Ohm.m (conductivity > 0.01 S/m) are potential causes for reflections. Significant increase in magnetic susceptibility will amplify these effects. A combination of these contrasts will also provide origin for reflection. Detectability is also related to size and distance from tool of reflecting feature, compared to the wavelength of the field. The radius of reflecting surface has to be larger than distance multiplied with wavelength (that is, first Fresnel zone). Smaller objects may be seen as diffraction events, or remain undetected. Also orientation of the reflector is affecting to detectability. With reference to the radiation beam angle width (depending on critical refraction index defined from dielectric contrast), normal of reflecting plane is required to point towards the measurement station within this radiation beam angle, to be observed as reflector (Sellman et al. 1992). This holds

14

especially for transverse electric field (TE polarization) along the dipole antenna axis (towards the short ends of antenna). On floor, this means horizontal structures within dip range 40-60 degrees. The reflection beam width is differing according to antenna direction. Transverse magnetic field (TM polarization) perpendicular to antenna axis (long side) has also side lobes in radiation beam, enabling reflections from structures with steeper inclinations. Steeply dipping structures are seen on raw radar images as scattering, or diffractor events. Radar resolution is related to what kind of structures are detected with respect to dominant wavelength. Two reflecting layers or diffractor objects can be detected as separate when their distance is greater than the wavelength. The thickness of a layer to be detected as reflector, can be as thin as 1/10 – 1/20 of wavelength. For 270 MHz central frequency and 120 m/microsecond velocity the wavelength is 0.44 m, and thickness limit for detectability 2 cm. This includes both fracture filling or aperture and possible alteration halo. Evidences are found that even thinner layers can be observed as reflector. Even very thin layers with a high contrast may serve as mirror like reflector. Geometrically it is difficult to resolve the actual orientation of reflector. Reflector is arriving along normal to the reflecting plane, not directly below the tool. Reflectors crossing the survey line in perpendicular angle are showing apparent inclination, which is depending on the true inclination and radar velocity, and can be computed from the results. Off line arriving reflectors (oblique intersection or parallel to line) have gentler apparent inclination than for perpendicular strike, and correct computing would require either information from several lines, or on strike intersection angle with survey line. Correct subsurface positioning of the reflector would require knowledge of correct velocity. In schistose environment the GPR phase velocity is varying according to direction (Tillard & Dubois 1995). In Olkiluoto the velocity is varying also due to lithology, being highest in pegmatite granite and lowest in migmatitic gneiss. Survey is useful to be carried out from cleaned rock surface. Concrete liner on wall and macadam filling on floor are substantially attenuating the signal. Ground response antenna and survey mode require good contact on surface. Any air gap between surface and antenna will cause ringing (clutter, Benter et al. 2011). Variation of air gap due to irregularities on surface will reduce quality of image. Apart of geometrical aspects (size, distance, thickness, orientation) and contrast, uncertainty is related to origin of reflectors. Survey conducted on rock surfaces; floor or wall of the tunnel, can detect features residing in rock mass. Survey can detect also side reflections of ceiling and other walls of tunnel, end of tunnel, and irregular form of tunnel. As well metallic objects like bolts, installation, lighting components, transformer casing and wirelines will cause side reflections. Means to avoid reflections from artificial sources are correlation of the object on the images. Air reflections are met at apparent velocity of 150 m/microsecond (half of speed of light in air). Most critical aspect of uncertainty is the origin of reflectors arriving from rock mass. Reflectors can be initiated both in fractures and from lithological or banding (layering) surfaces. Only way to distinguish between different origin is to detect the cause of

15

reflector from drillhole logging or rock surface mapping. It is necessary to also compare whether the orientation of structure and suggested orientation from radar image are matching. Non matching orientations mean the features are not the same. Data processing can be used to suppress interference in images, and to enhance reflecting events. Purpose is to make images more clear, though processing can also delete some of features, and needs to be applied carefully. Processing techniques involved can be for example: 1) horizontal filtering (subtracting average trace) which suppresses ground coupling (first arrival) but can reduce also reflections parallel to surface, 2) deconvolution which suppresses ringing, 3) migration (different techniques) which is suppressing diffractors and imaging in 2D a perpendicularly striking reflector to correct inclination or in 3D any reflector to correct orientation, but also reducing visibility of vertical structures, and in 2D mode causing ring like shapes in presence of 3D objects, or if applied radar velocity is not correct. 4) Frequency filtering to enhance reflecting events and 5) Gain to enhance later events. 6) Stacking of traces to amplify events and to suppress noise. 7) Selection of colour or gray scale and limit of amplitude range to emphasise different features.

Proper application of filtering and processing will require testing. Several phases in filtering are useful to display in order to emphasise different kind of features.

2.3 Surveys

Field surveys were carried out using GSSI SIR-3000 GPR unit and ground coupled 100 MHz, 270 MHz and 500 MHz antennas. Field surveys on floor were carried out on cleaned and dry rock surface. Field surveys on the wall were run with antenna manually held on rock face. Positions were staked by Prismarit Oy. The surveys carried out are listed in Table 1 below. Location map of the surveys is shown in Figure 5.

16

Figure 5. Location of the GPR surveys for RSC in ONKALO access tunnel and niche TKU-3 (brownish colour). View from above, north on top. Ground penetrating radar (GPR) survey at access tunnel chainage 3340 – 3575 was intended to check the visibility of large fractures and deformation zones. Profile was placed at 50 cm elevation from the right-hand wall when viewing downwards along the tunnel. Survey was directed along the seismic 2D/3D reflection line and it applied the coordinates staked for seismic receiver stations on the wall. The 270 MHz survey covered the line entirely. The 100 MHz measurement encountered obstacles due to large size of antenna and was measured partially in five sections. Antenna is heavy to hold which made the survey also difficult to complete.

N

GPR survey in TKU-.3 25-67 m

17

Table 1. GPR surveys conducted in ONKALO access tunnel and niche TKU-3. Survey Date Location Parameters Purpose 270 MHz ground response

21.09.2009 ONKALO Access tunnel 3344 – 3578 (234 m), right hand wall, 50 cm elevation from wall

Single line 50 scans/m 249.6 ns time window 1024 samples 0.244 ns time incr. Direct wave 22 ns

Reference for seismic 2D/3D reflection survey line, RSC

100 MHZ ground response

21.09.2009 ONKALO access tunnel (150 m total) 3383 – 3428 (45 m) 3443 – 3455 (12 m) 3461 – 3479 (18 m) 3494 – 3542 (48 m) 3548 – 3575 (27 m)

Single line 50 scans / m 500 ns time window 512 samples 0.997 ns time incr. Direct wave 43 ns.

Reference for seismic 2D/3D reflection survey line, RSC

270 MHz ground response

24.03.2010 TKU-3, ten lines on floor at 25 – 65.8 m

Lines 0.5 – 1.0 m interval. 100 scans/m 180 ns time window 1024 samples/scan 0.1758 time incr Direct wave at 6 ns

RSC survey

270 MHz ground response

20.10.2009 TKU-3, 5 lines on floor and one on wall at 15 – 55 m

Lines 0.8 – 1 m interval. 100 scans/m 150 ns time window 1024 samples/scan 0.1465 ns time incr. Direct wave at 4 ns

EDZ survey deeper seated structures, RSC

500 MHz ground response

20.10.2009 TKU-3, 3 lines on floor at 15 – 54 m

Lines 0.8 – 1 m interval 100 scans/m 100/150 ns time window 1024 samples/scan (0.0977/0.1465 ns time incr) Direct wave at 8.5 ns.

EDZ survey deeper seated structures, RSC

18

Figure 6. Location of the GPR surveys for RSC in TKU-3 during the EDZ campaign (chainage 15 - 55), view to the north. Location of niche is shown with blue and green circles on Figure 5 above. White dots on right-hand wall indicate level of GPR survey on the wall. Lines -1,8 m, -1,0 m and 0 m were measured with 500 MHz, and all five with 270 MHz antenna. Surveys in TKU-3 investigation niche were carried out in two phases as excavation proceeded. Initially the niche chainage interval 15 – 55 m was measured on five lines centred in the tunnel (-1.8 m, -1m, 0 m, 1 m and 1.8 m) with 270 MHz antenna and three lines (-1.8 m, -1 m and 0 m) with 500 MHz antenna (Figure 5 and 6). The same area with 10 cm line interval was covered with 1.5 GHz survey in EDZ characterization.

19

The right-hand wall of the niche was measured with 270 MHz antenna on the same chainage interval, on line marked on the wall at 2.2 …0.5 m level from floor. The line on the wall follows path of pilot hole ONK-PP200. EDZ characterization covered 1 m wide area on the wall (11 lines) using 1.5 GHz antenna. Positioning was performed with measuring the end coordinates of the central line and lines on the edges of niche. Measurement was triggered with pulse encoder and tape attached on a screw, mounted on wooden plank in the end of line. This was intended to help later presentation of the results in 3D. The niche chainage interval 25 – 67 m was measured with 270 MHz antenna on the floor on ten lines placed asymmetrically at 0.5 – 1.0 m spacing, -4.0 m, -3.0 m, -2.0 m, -1.0 m, 0 m, 0.5 m, 1.0 m, 1.5 m, 2 m and 2.5 m. Lines are partly overlapping with previous measurements on 10 m length section. Offset with 0 m lines is -70 cm. The same area with 10 cm line interval was covered with 1.5 GHz survey in EDZ characterization.

Figure 7. Location of the GPR surveys for RSC in ONKALO.

20

2.4 Data

Primary data contains strong first arrival at 10-15 ns and some ringing at later 15 – 40 ns times. The air reflections from tunnel ceiling at 60-70 ns and side wall (at fairly constant arrival times), as well as from some roughness in the wall and the installations (winding paths), are clearly seen (Figure 8). The reflections from ceiling and side wall arrives cause some ringing, and can reflect the variation in shape of the surface. The airwaves and reflections from installations and back wall are causing hyperbola, showing for direct waves velocities of 0.3 m/ns and for two way reflections apparent velocities of 0.15 m/ns. The velocities can be used to distinguish air reflections from bedrock originating ones. The irregularities on the surface have caused some areas where the ground response antenna is detached from wall or floor. These points clearly encounter severe interference (Figure 9). Reflectors and diffractors are seen in the image. These are caused by fractures, lithology contacts and foliation (banding). Reflectors which are caused by interfaces aligned within about 0 - 50 degrees from parallel to the surface, are seen well (Figure 10). The termination points of reflectors may cause clear diffractions. Reflectors which are caused by near vertical interfaces, are seen as paths of diffractor hyperbolae (Figure 11). The texture of rock is apparently causing plenty of diffracting energy. The signal is much cleaner in areas which consist of pegmatite, and exhibit diffractions at gneissic rock mass. Velocity is varying. Velocity determinations are based on migration velocity and estimates of average velocity from borehole radar survey (direct wave arrival time, 0.117 m/ ns), and former VRP and other radar surveys in the area. Velocity seems to vary in different rock types at range 0.11 – 0.115 m/ns (mafic gneiss…migmatite) …0.125 m/ns (pegmatite). The different antennas and frequencies have slightly different characteristics:

- 500 MHz has highest detection capability and very clean signal, however the signal is decaying at 90-100 ns two-way time (5-6 m)

- 270 MHz can detect slightly less events, and encountered more interference from roughness of surface; the depth of investigation is minimum 7 m but possibly 10-12 m at 150-200 ns two-way time. Data quality is best obtainable for 1 – 7 m depth interval.

- 100 MHz can detect less, but more continuous events, but it is suffering from severe ringing (possibly caused by poor connection to surface). Tool is fairly large, for which reason it cannot be used on places with installations on the wall, and the manual operation is difficult to complete due to weight. Depth of investigation is 20-22 m at 400 – 450 ns two-way time for large objects.

Data resolution and depth extent may be enhanced by stacking, either by GSSI unit or by external software.

21

Figure 8. First arrival, ringing and air reflections in the radar image from floor, 270 MHz in TKU-3 at 15 – 55 m. North is on the right (tunnel end).

22

Figure 9. Area where ground response antenna is not properly attached onto the rock surface. ONKALO access tunnel right-hand wall, 270 MHz. West is on the right.

23

Figure 10. Reflection from a favourably positioned planar interface. TKU-3 Floor Line 0 at 25 – 67 m, 270 MHz. North is on the right.

24

Figure 11. Diffractions from discontinuous or vertically oriented interface. TKU-3 Floor Line 0 at 25 – 67 m, 270 MHz. North is on the right.

25

3 PROCESSING AND PRESENTATION

Standard GPR processing was performed by Roadscanners Oy during the field survey and data delivery. Primary processing and analysis applied GeoDoctor 2.2 software. Processing included horizontal filtering, vertical frequency filtering, arithmetic operations and signal amplification. Data analysis included detection of individual cracks and fractures, and delimitation of fractured areas. Analysis indicated both dipping fracture structures causing a distinct reflector, and several fractured areas where no clear reflectors are distinguished, but rather a group of reflections. For further processing in ReflexW (2003) a fairly uniform set of steps was applied to the data to make the weak reflections more visible in the images. Steps are listed in Table 2 below. Profiles were flipped to increasing direction of chainage (west), or to north, when necessary. Table 2. Further GPR processing steps Processing step 500 MHz 270 MHz 100 MHz Purpose 1 DC Level shift yes yes yes adjust DC levels

to enable later processing and avoid truncation errors

2 Subtract average traces

11 5…11 11…21 Remove first arrival and ringing

3 Band pass filter 100..150 – 800…1000 MHz

20..50 – 400…500 MHz

10…30 – 250..350 MHz

High frequency noise suppression

4 Deconvolution lag 3 ns, 2% noise

lag 5 ns, 2% noise

12 ns, 5% noise Remove multiples and ringing, spectral whitening

5 Band pass filter 100..150 – 800…1000 MHz

20..50 – 400…400 MHz

5…20 – 250..350 MHz

High frequency noise suppression

6 F-k migration (Stolt, 1978)

Velocity 0.125 m/ns

Velocity 0.125 m/n

Velocity 0.125 m/n

Suppression of diffracted energy

7 Gain 0.01 * 100.05, max 10

0.08 * 100.25, max 8

0.05 * 100.1, max 30

Emphasis of later events

8 Move start time

- 6 ns -6 ns or -12 ns - 44 ns Place first arrival to 0

9 Time clip if required

100 ns 145 – 170 ns 450 ns Suppress times after decay of signal

10 Time to depth conversion

0.125 m/ns 0.125 m/ns 0.125 m/ns Depth axis

26

The raw data before any processing exhibit strong first arrival and ringing. Subtracting short filter average traces (horizontal filtering) removes early horizontal banding (Figure 12). Band pass filter will gain visibility of low and mid frequency reflections above high frequency noise. It is performed first time before deconvolution. Deconvolution suppresses further prominent ringing and multiple reflections (Figure 13). Deconvolution will introduce high frequencies, which are suppressed by bandpass filtering again at later stage. Deconvolution is useful to be applied before migration, which is clearly reducing ring-shaped artefacts. F-k migration (Stolt 1978) suppresses diffracted energy (Figure 14). Velocity seems to be variable, so that there are some diffractions remaining in the data set after migration. Gain emphasises later reflection events (Figure 15). The signal starts to exhibit increase of noise level at 160-170 ns time. Moving start time to first arrival and when necessary, clipping time after decay of signal level allows meaningful presentation of data in depth image (Figure 16). Trace start times were not adjusted with elevation variation, a step which can be performed when found useful. After processing steps the reflections are clear and limited, and most of the sources of interference are suppressed. The migration has caused into the image some ring shaped artefacts.

Figure 12. Removal of first arrival and horizontal banding, subtract 11 trace moving average (270 MHz, TKU-3 at 25 – 67 m, Floor Line 0). North is on the right.

27

Figure 13. Deconvolution with 5 ns lag and 2 % noise (270 MHz; TKU-3 at 25 – 67 m, Floor Line 0). North is on the right.

28

Figure 14. F-k (Stolt) migration using 0.125 m/ns velocity. Same profile (Floor Line 0 in TKU-3, 25 – 67 m) as in previous figures. North is on the right.

29

Figure 15. Gain function 0.08 * 10 -0.25, max gain 8 (exponent dB/m, linear 1/pulsewidth). Same profile TKU-3 Floor line 0, 25-67 m. North is on the right.

30

Figure 16. Start time move -6 ns and bandpass filter 20…60 – 400…500 MHz. TKU-3 Floor line 0 at 25-67 m, north is on the right. Image end points (X, Y, Z in KKJ1) on surface were defined from staked coordinates of seismic line on access tunnel, and GPR line end coordinates on niché. The corresponding XYZ end coordinates in depth below floor or off the wall were computed from the coordinates and the time to depth range. Four corner coordinates of the images were used to present them in 3D with models and other data.

31

Figure 17. Complete profile of TKU-3 floor line 0 (25-67 m) after processing steps.

32

33

4 GEOLOGICAL PROPERTIES AND REFERENCE DATA OF TARGET AREA

The pilot holes were core drilled on the centre line of access tunnel. Lithology and deformation were logged from core samples. Geological data is reported along length axis, including lithology, fractures with their filling and orientation, and foliation with orientation. The niche was drill and blast excavated with smaller profile than access tunnel. Apart of pilot holes, several short characterization holes were drilled on the floor and walls of the niche. Pilot hole and other drillhole investigations include different kind of geophysical logging, and borehole radar reflection survey. The core logging data and borehole radar reflections of pilot holes ONK-PH09 (Karttunen et al. 2009) and ONK-PH10 (Mancini et al. 2010) were used in assessment. Tunnel wall and ceiling was mapped for lithology and deformation. Niche was mapped entirely, also from floor. Mapping data was presented in 3D system. Lithology in access tunnel at 3345 – 3575 m consists of diatexitic gneiss and veined gneiss. Some intersections of pegmatitic granite, mafic gneiss and potassium feldspar porphyry and quartz gneiss are also met. Fracture frequency increases > 10 pcs/m at 3345 – 3364 m and 3540 – 3550 m. On deformation indications, brittle fault intersections (BFI) and high grade deformation intersections (HGI) were mapped with relation to access tunnel length and named according to location. Two site scale brittle fault zones (BFZ) were correlated to intersect the access tunnel: BFZ045 at 3340 m and BFZ084 at 3550-3560 m. Large fractures (tunnel crosscutting fractures, TCF) were mapped separately of other fracture traces, and indicated with running label P338 - P352. The BFI, BFZ and HGI intersections as well as TCF fractures have also indicated with their intersection location (chainage) and orientation. The TCF fractures and zone intersections are listed in Table 3 below. The niche TKU-3 contains abundance of comparison data. Niche mapping data is available in 3D format for fractures, foliation, and rock type contacts. The drillholes ONK-PP199, PP200, PP202 – PP205, PP207- PP209, PP223, PP254 and PP259 – PP261 locate within or near the niche. The full profile disposal holes are located below the floor of the niche. Lithology and fractures were presented in 3D view. Also the borehole radar reflectors were presented in 3D view as disks. TCF fractures are not reported from investigation niche TKU-3. Some fractures having continuity are found in tunnel wall mapping.

34

Table 3. Deformation intersections and TCF fracture intersections compared with GPR images at access tunnel chainage 3345 – 3578 m (Joutsen 2011, Lahti et al. 2009). Intersection Chainage Orientation

core (tunnel) HGI-3342 3342 60/029 P338 3350-3352 86/082 BFI-3350 P339 3450-3454 86/228 HGI-3374 3374 28/105 HGI-3385 3385 16/210 P340 3399 82/104 P341 3407 87/293 P342 (left) 3409 24/023 P343 3420 87/068 P344 3430 85/154 P345 3436 P346 3468 89/070 (68/267) P347 3484 84/274 (81/286) P348 3492 84/277 (79/287) P349 3498 84/294 (83/286) P350 3510 57/149 (53/151) HGI-3515 3515 48/108 P351 3535 50/173 (51/180) BFI-3450, BFZ084 51/180 P352 3560 34/035 (25/108) HGI-3593 40/109 Contact of PGR and MGN 3570-3580 32/58

35

5 COMPARISONS

5.1 Review

The processed GPR reflection images were presented in 3D system for comparison with different data sets. The access tunnel chainage 3344 – 3578 GPR images were reviewed on tunnel mapping fractures and lithology. Specific attention was placed to tunnel cross cutting fractures (TCF). An attempt was made to characterize the reflectors from data. Reflector traces were picked for line location and time in ReflexW software and saved to text file. Line coordinate was converted to 3D location, and time to elevation. The traces of reflectors were presented in 3D (Figure 18). For parallel lines, some of the reflectors which could be recognised and correlated between lines, were picked to traces on plane, then joined on text file and converted to XYZ coordinates. The obtained cloud of coordinates was used to create planar model (Figure 19). Picked lines and objects seem to correspond in general view the rock type contacts, fractures, or banding; and may match also with borehole radar reflections.

Figure 18. Picking of reflector in a single image line (TKU-3620, PL 15 – 55, 270 MHz; Line +1.8 m). Lines are reflecting events. View to the west, north is on the right. Length of image is 42 m and hight 10 m. Procedure of reflector picking, joining and coordinate conversion is fairly quick. However the picked reflectors may represent almost anything geologically relevant from fractures to rock type contacts and intense foliation. Correlation of traces to reflectors is not straightforward unless data is acquired very densely, and offline reflecting events are not imaged initially to correct position (Figure 15). For these reasons, all reflectors were not explained easily even by reviewing with tunnel mapping and drillhole data. Reverse attempt was adopted. The significant fractures in tunnel mapping were taken as basis, and their appearance was reviewed from the radar images. Then the remaining clearly visible reflectors were checked from tunnel mapping data and photographs, to arrive to possible explanation for them as well.

36

Figure 19. Combining picks from several lines to a DTM object. TKU-3 at 25 – 67 m GPR line 0 shown with DTM of seven deduced reflecting bodies. View to the west, north is on the right. Length of image is 42 m and hight 10 m.

5.2 GPR on access tunnel wall

On access tunnel GPR lines, tunnel crosscutting fractures P338 – P352 and deformation intersections (BFI, BFZ and HGI) were checked from the GPR images. Results of comparison are listed on Table 4 below. Table 4. Deformation intersections and TCF fracture intersections (Joutsen 2011; Lahti et al. 2009) compared with GPR images at access tunnel chainage 3345 – 3578 m. Intersection Chainage Orienta

tion Match Continuity Figure

N:o HGI-3317 3317 possible reflection Borehole radar Figure 20 HGI-3342 3342 60/029 Foliation. Fractures at

60/353 do not fit with angle.

Figure 21

P338 3350 86/082 Not geometrically favourable

Figure 20 BFI-3350 P339 3450 Limited traces in

images. Figure 26

HGI-3385 3385 16/210 Possibly HGI. Not TCF.

Reflectors directed to NW (down)

Figure 22

HGI-3374 3374 28/105 Possibly HGI. If yes, seen from below floor.

Figure 23

P340 3399 3407 3409

82/104 87/293 24/23

not seen in radar images

Figure 24 P341 P342 (left) P343 3420 87/068 Seen in GPR image 20 m downwards, 6

m out of tunnel Figure 25

P344 85/154 Seen in image upwards, 4 m out of tunnel

37

P345 3435 Weak echoes. reflector 3460 N/A reflector trace similar

as with P343 seen 10 m out of wall

Figure 27

reflector 3465 N/A reflector upwards in tunnel

seen 10 m out of tunnel, until 3440

Figure 27

P346 3470 68/257 seen as a reflector down to 3490 m, 8 m out of tunnel

Figure 28

P347 3484 81/286 not seen in radar Figure 29 P348 3492 79/287 reflector 3495 N/A no explanation directed upwards in

tunnel

P349 3498 83/286 P350 3510 53/151 seen in image, may

be HGI upwards to 3480 and 9-10 m out of tunnel

Figure 30

HGI-3515 48/108 P351 3535 51/180 seen as reflector 20 m upwards in

tunnel, 5-6 m out Figure 31

BFI-3450, BFZ084 reflectors 3540 N/A possibly mica gneiss

or mafic gneiss inclusions

directed down in tunnel

Figure 32

P352 3560 25/108 reflector 20 m upwards and 6 m out

Figure 33

HGI-3593 3593 40/109 seen in image 40 m upwards and 10 m out

Figure 34

Contact of PGR and MGN

3570-3580 32/58 seen in image 10 m down and 8 m out

Out of 15 reported TFC fractures, eight are seen as reflectors, among of these one BFZ intersection (BFZ084). Seven are not seen, mainly due to unfavourable position (perpendicular to wall). One of these is BFI intersection. Traces of reflection may be seen in non-processed images. Five reflectors may be explained with HGI intersection at its location, and being favourably oriented. One reflector can be explained with lithological contact. Four reflectors remain without explanation. The high grade ductile deformation exhibit intense lithological banding of leucosome (pegmatite granite) and melanosome layers, which have contrast on dielectricity and may have contrast also on electrical conductivity (for example graphite or sulphides). This makes them a likely origin for reflections. The case is supported by borehole radar reflections associated with these zones in places (Döse 2011). In any comparisons between reflector orientations and their traces on the images must be borne in mind, that the reflector angles are always apparent (related to so called alpha angle on tunnel axis), and actual correlation needs to be computed by forward modelling. At 3350 m, a local zone BFI-3350 and TCF P338 are intersecting the tunnel perpendicularly. Reflections cannot be seen (Figure 20), but there are instead reflections near tunnel wall (directed in narrow angle upwards in tunnel) which probably would be

38

associated with high grade shear intersections (HGI-3317) dipping gently eastward, and reflecting surfaces would need to be observed from below of tunnel. The borehole radar reflections from ONK-PH09 pilot hole at this location are supporting the same idea. Reflectors at 3358 m indicate steep angle with tunnel and may be explained by foliation (Figure 21). Fractures could not produce reflection in this projection angle. Possible explanation is the HGI-3342 intersection orientation. At 3370 – 3390 reflectors are not explained by TCF fractures. There is a high grade shear intersection HGI-3385 which probably could explain the reflector (Figure 22). Reflectors are directed to NW. Strong foliation is associated to the intersection. At 3383 – 3400 a HGI-3374 zone could explain corresponding reflector (Figure 23). In this case the reflection is not arriving on perpendicular direction to tunnel face, but from below of the tunnel. This calls for radial symmetry for reflections, similar to borehole radar. The TCF P340 and P342 at PL3400-3410 are perpendicular to the tunnel and are not seen directly in the radar images (Figure 24). Fracture P343 at PL3420 is seen in image (Figure 25). Reflector continues 20 m downwards (to the west) and 6 m out of tunnel wall, and may consist of several parts. There are also several other reflections with similar apparent intersection angle. Fracture P344 is seen upwards to 3400 and c. 4 m out of tunnel. Reflection from P345 at PL3435 is uncertain, being perpendicular to tunnel is not favourably oriented, and is showing possibly some weak echoes. The TCF fracture P339 at 3450 m is dipping to the west, and may cause some limited traces of reflections in the images, specifically in 100 MHz (Figure 26). At PL3460 there is in radar image a reflection which is directed downwards in tunnel: The apparent angle is similar as for the reflector caused by tunnel cross cutting fracture P343. Trace is seen at 10 m distance from wall. At PL3465 there is a linear reflector directed upwards in tunnel, and seen to PL3440 and 10 m out of tunnel. This feature is not yet explained (Figure 27). Feature is better seen in 270 MHz. It is amplified by multiples in section where antenna is off the wall. Similarly oriented feature is seen in 100 MHz at slightly later time. Position of intersection is hosting a borehole radar reflection in ONK-PH09. The TCF fracture P346 at PL3470 which has orientation 68/257 is seen as a reflector (Figure 28). It may be encountered down the tunnel until 3490 and c. 8 m out of tunnel wall. Feature is better seen in 270 MHz image. The tunnel cross cutting fractures P347 (81/286), P348 (79/287) and P349 are not in orientation which could geometrically be seen as GPR reflection (Figure 29). There is a reflector directed upwards in the tunnel, which may coincide with borehole radar reflector at PL3495, without explanation yet.

39

The TCF fracture P350 at PL3515 is seen up to 3480 upwards and 9-10 m out of tunnel wall (even further in 100 MHz image). There is a HGI-3515 associated with this location, with slightly differing dip direction, which could also explain the reflections (Figure 30). The TCF fracture P351 which is associated with BFI-3540 and BFZ084, is seen 20 m upwards in tunnel, and 5-6 m out of tunnel wall (Figure 31). The continuous reflections directed to NW at 3540 are not explained by TCF in tunnel mapping (Figure 32). Their apparent projection angle is similar as the reflections caused by structures oriented to 80/250 or 80/50. Viewed from their intersection area, these reflections may be explained by mica gneiss or mafic gneiss inclusions, in case these would have similar orientation of continuity. The TCF fracture P352 at PL3570 is seen 20 m upwards in tunnel and 6 m out of tunnel wall (Figure 33). At 3593 there is HGI-3593 intersection which may act as reflecting body. It may be seen up to 40 m upwards in tunnel and 10 m out of tunnel wall. At 3570 – 3580 there is a contact of granite and mica gneiss seen as reflector 10 m downwards and 8 m out of tunnel (Figure 34).

Figure 20. Fracture zone BFI3350 (P338) and high grade shear zones HGI-3317 at tunnel chainage PL3350. Disks with large radius are borehole radar reflections and small disks are fractures in pilot hole ONK-PH09. View from above, north upwards. Tunnel diameter is 6 m.

40

Figure 21. Reflectors at tunnel chainage 3358 (red lines). Possible cause the HGI zone. Disks with large radius are borehole radar reflections and small disks are fractures in pilot hole ONK-PH09. View from above, north upwards. Tunnel diameter is 6 m.

Figure 22. Reflectors and HGI intersection at tunnel chainage 3370 - 3390. Disks with small radius are fractures in pilot hole ONK-PH09. View from above, north upwards. Tunnel diameter is 10 m.

41

Figure 23. Reflectors and HGI intersection at tunnel chainage 3383 - 3400. Above: 100 MHz, below, 270 MHz. Fractures P340 and P342 are not favourably oriented, whereas the HGI’s are. Disks with small radius are fractures in pilot hole ONK-PH09. View from above, north upwards. Tunnel diameter is 10 m.

42

Figure 24. Reflectors and TCF intersections at tunnel chainage 3400 – 3420. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH09. View from above, north upwards. Tunnel diameter is 6m.

43

Figure 25. Reflectors and TCF intersections at tunnel chainage 3440 - 3460. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH09. Blue disk is a borehole radar reflector. View from above, north upwards. Tunnel diameter is 6 m.

44

Figure 26. Reflector at tunnel chainage 3465 - 3440. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH09. Blue disk is a borehole radar reflector. View from above, north upwards. Tunnel diameter is 6 m.

45

Figure 27. Reflectors and TCF P346 intersection at tunnel chainage 3460 - 3480. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH09. Blue disks are borehole radar reflectors. View from above, north upwards. Tunnel diameter is 6 m.

46

Figure 28. Reflectors and TCF intersections at tunnel chainage 3480 - 3500. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH09. Blue disk is a borehole radar reflector. View from above, north upwards. Tunnel diameter is 6 m.

47

Figure 29. Reflectors and TCF and HGI intersection at tunnel chainage 3500 - 3530. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH09. Blue disk is a borehole radar reflector. View from above, north upwards. Tunnel diameter is 6 m.

48

Figure 30. Reflectors and BFZ intersection at tunnel chainage 3520 - 3560. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH09. Blue disk is a borehole radar reflector. View from above, north upwards. Tunnel diameter is 6 m.

49

Figure 31. Reflectors at tunnel chainage 3540 - 3580. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH10. Blue disks are borehole radar reflectors. View from above, north upwards. Tunnel diameter is 6 m.

50

Figure 32. Reflectors and TCF intersection P352 at tunnel chainage 3540 - 3580. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH10. Blue disk is a borehole radar reflector. View from above, north upwards. Tunnel diameter is 6 m.

51

Figure 33. Reflectors and HGI intersection at tunnel chainage 3560 - 3600. Above: 100 MHz, below, 270 MHz. Disks with small radius are fractures in pilot hole ONK-PH10. View from above, north upwards. Tunnel diameter is 6 m.

52

5.3 GPR on TKU-3 floor

The GPR results on TKU-3 niche floor are providing information which can be correlated both with tunnel mapping and with drillhole data. In TKU-3 floor at PL15 – 55 chainage the 270 MHz and 500 MHz GPR images show plenty of reflectors. Mostly these were associated with rock type contacts and foliation in closer examination (Figures 34 and 35). These can be partly joined with borehole radar reflections shown as large blue disks on Figures. There are also some reflections which can be associated with pilot hole ONK-PP199, ONK-PP200 and ONK-PP226 fractures shown on small disks (and part of tunnel mapping data, shown on lines). At PL40 – 45, depth 8-10 m there are found fractures in ONK-PP223 which match with orientation to a reflector, met on the surface at PL57, near vertical drillholes ONK-PP259-261 which were drilled on the floor of the tunnel. In TKU-3 at PL25 - 65.8 m chainage the 270 MHz images show several reflections which are again linked to lithological contacts. In drillholes ONK-PP199, PP200, PP226, and in vertical holes PP254-PP261 there are fractures which clearly are linked with a reflector. At least one tunnel mapping fracture is matching with its location to a reflector (Figure 36 and 37). Apart of the reflectors which are explained either by recognisable lithology or fractures, there are several others which are not explained. Either these are not intersecting with mapped surface or drillholes, or the explanation cannot be judged on available information. These may be checked, if found necessary, by drilling holes from tunnel surface.

53

Figure 34. The 270 MHz radar image from niche TKU-3, PL15 – 55, and the niche mapping data. Green: lithological contacts (pegmatite), yellow: foliation observations, white: fracture traces on the wall. Green disks (above) indicate fractures in drillhole. Blue disks (larger, below) indicate oriented borehole radar reflectors.

54

Figure 35. The 270 MHz radar image from profile 0, niche TKU-3, PL15 – 55, and the drillhole mapping data. Lithology: red = pegmatite, light blue = veined gneiss, medium blue = diatexitic gneiss, yellow=TGG gneiss, dark blue = mica gneiss. Green disks (above) indicate fractures in drillhole. Blue disks (larger, below) indicate oriented borehole radar reflectors.

55

Figure 36. The 270 MHz radar image from profile 0, niche TKU-3, PL25 – 67, and the niche mapping data. Green: lithological contacts (pegmatite), yellow: foliation observations, white: fracture traces on the wall. Green disks (above) indicate fractures in drillhole. Blue disks (larger, below) indicate oriented borehole radar reflectors.

56

Figure 37. The 270 MHz radar image on profile -200 from niche TKU-3, PL25 – 67, and the drillhole mapping data. Lithology: red = pegmatite, light blue = veined gneiss, medium blue = diatexitic gneiss, yellow=TGG gneiss, dark blue = mica gneiss. Green disks (above) indicate fractures in drillhole. Blue disks (larger, below) indicate oriented borehole radar reflectors.

57

6 DISCUSSION AND CONCLUSIONS

The GPR measurements provide detailed information on reflecting bodies below the tunnel floor or behind the wall up to distance of 10 – 20 m. The results will benefit if the survey is carried out on cleaned and dry rock surface. Water and mud hamper the results significantly, but gravel and especially shotcrete containing metal wires limit the penetration almost completely. The best applicable antenna and frequency is found to be the 270 MHz having 7-12 m depth penetration. Owing to its small dimensions the antenna has good contact to rock surface which minimises interference. Any continuous or irregular air gap between antenna and the rock surface tends to cause ringing in the image. Soil or water between antenna and rock surface has similar effect. The comparably light weight of antenna enables good manual operability. Reflector detectability is good, except for features perpendicular to the survey line. Depth penetration may be enhanced in limited extent with using stacking in the survey or processing. The 100 MHz antenna is clearly heavier and this way more difficult to operate. On tunnel floor this is not a significant limitation. Antenna size is proportional to nominal frequency. Larger size of antenna causes unavoidably more frequent occurrences of air gap which is adding ringing as the antenna is less perfectly seated on the surface at irregularities. Ringing can be attempted to be partly removed with deconvolution. Depth penetration is 20 m at its best. Somewhat less reflectors are detected than with 270 MHz tool. The 500 MHz antenna has higher detectability and low noise level, but the penetration of signal is limited to 5-6 m. Reflectivity requires a contrast in dielectricity or electrical conductivity, provided either by fracture with its coating, water content, alteration halo or zone of influence, or lithology (foliation). It seems useful to apply two antennas (frequencies) in the survey to provide more reliability in the results. The 270 MHz tool has the best performance if the limited range is considered adequate. Continuous reflections are also visible in the lower frequency results. If those are located close to the floor their detailed location can be checked using the higher frequency. According to correlation with the geological observations and their orientation, it seems that in some cases the survey geometry may be slightly radially symmetric providing reflections from below the tunnel floor with survey directed on the tunnel wall (90 degrees offset). These reflections are met with 100 MHz antenna in places where no recognised explanation is found from behind the wall, but are found from below the floor. Theoretically electromagnetic wave field tends to turn towards the normal of the surface, which would emphasise the reflections from normal to the surface rather than from offset. Also the antennas are shielded and the radiation pattern designed so that

58

reflections from off-line location should be suppressed. The width of the beam is depending on the antenna design. The case is dependent on antenna direction during survey. This is more likely with 100 MHz tool, which is less well shielded on the long edge of the antenna and is having wider beam. The long sides were directed upwards and downwards on the ONKALO access tunnel wall (antenna was directed horizontally), which seem to have enabled detection of near horizontal reflectors from below. Alignment was necessary due to installations on the wall which were preventing the up-right position. For the 270 MHz antenna the beam is narrower, antenna shielding is better and the antenna was oriented vertically (viewing front and back). Consequently these potential below the floor reflectors are not seen in 270 MHz images. Reflection angles in the image are not true in case the reflection is arriving off-line. The apparent (alpha) angles of reflector with the tunnel axis are necessary to use in correlation to geology and interpretation. Field operation will benefit of marking and measuring the line end coordinates with total-station already before survey. Optimally the survey would be carried out with several parallel lines along floor to detect continuity of reflectors. If the line density is adequately high, reflections can be reliably tracked between the lines. Otherwise frequent crosslines would be useful to assist recognition and to define offline dip of the reflectors. Short crosslines are of little use, though. The images are useful to present in 3D, possibly including basic processing and depth conversion. Coordinates and image draping can be prepared in advance. Further processing to suppress direct arrival, ringing, and diffractions, and to gain later events, is useful for the final assessment. Images can be applied into the 3D system using same draping as the raw images (geocoding). A suggested work flow using GPR in the disposal tunnels for the characterisation needed for the rock suitability assessment is summarised in Table 5 below.

59

Table 5. Suggested work flow in application of GPR for host rock characterisation and suitability assessment (RSC). Nr Activity Purpose Result Status/

Phase of information

1 Pilot hole investigations including geophysical loggingand borehole radar

Preliminary characterisation and tunnel suitability assessment

Prior knowledge on intersections

Preceding

2 Excavation of tunnel Preceding 3 Geological mapping of

tunnel Characterization, suitability assessment for deposition hole placement

Lithology, intersections, fracturing, TCF fractures

Preceding

4 Cleaning the floor Staking coordinates

Layout Enabling

5 GPR survey 270 and 100 MHz

Detection of reflecting surfaces

Raw images

6 Review of images as such Quick look of performance and possible features in data

Plan for further processing

Preliminary

7 Review of images with geological data, no coordinates

Explanation of reflectors with fractures and other origin

Plan for final processing

Preliminary

8 Final processing Draping images on 3D

Review of results with geology in 3D, documentation

Model, View Final

9 Geological review in 3D, common/shared view:

1) is TCF causing reflection?

2) other reflectors explained?

3) other reflectors not explained?

Explanation of reflectors to fractures and non-fractures

Fractures and their continuity presented

Final

10 Analysis of reflectors not explained by observations

Disclosing unknown features below the floor

Targets for possible checking

Final

11 Interpretation Measurement and presentation of interpreted structures

Fractures as objects in 3D system (DTM), model

Final

12 Suggestions Supporting data for characterisation and deposition hole placement.

Knowledge on intersection and large fracture continuation and location below tunnel floor

Final

Downhole radar from the horizontal pilot hole and tentative tunnel mapping results are available during the GPR survey in tunnel. Interpretation of the GPR results together with other available data would be a first step supporting the tunnel mapping, carried out as team work with mapping geologist(s) and surveying/processing geophysicist(s):

60

- check for each TCF or other large fracture, would it be visible in GPR data and if yes, how large the continuity is

- detect reflectors from the GPR images and try to find explanation from the tunnel mapping or pilot hole data (lithology and deformation)

- if the reflector is not met in the tunnel or drillholes geometrically, and it is considered large, it is useful to check the candidate disposal hole location by a short probing hole

- if a fracture is met in the disposal hole pilot, it is useful to check if it is visible in the GPR results and how large it is

- a reflector which is confirmed to represent a fracture, can be digitised as a 3D object for the detail-scale model.

According to the presented review, the GPR survey in tunnel surfaces (floor, possibly wall) is efficient and useful to detect reflectors. The field survey can be implemented within hours, soon after completion of excavation. Some attention has to be paid to suppress interference, and to get adequate coupling with the targets. Different processing phases of images are useful to maintain, as processing will suppress also some of useful information. Results can be correlated with their explanation to geology, and to estimate continuity of detected fractures. Results can be used for guidance in placement of the disposal holes. Results give some indication of fractures continuing further from the tunnel surface, up to ten metres distance, thus giving some indications to be used in further investigations. Locations lacking of reflectors may be set as preferred suitable locations, in case these are favourable also in other terms. GPR may not detect all types of fractures, so the final judgment is left after drilling and core mapping of the pilot hole, and mapping of the disposal hole. The GPR method alone cannot be used for rejection of disposal hole locations without checking the origin of the reflector.

61

REFERENCES

Aaltonen Ismo (ed.), Lahti Mari, Posiva Oy, Engström Jon, Mattila Jussi, Paananen Markku, Paulamäki Seppo, Gehör Seppo, Kärki Aulis, Ahokas Turo, Torvela Taija, Front Kai. 2010. Geological Model of the Olkiluoto Site - version 2.0. Posiva Working Report 2010-70. Annan, A.P., Waller, W.M, Strangway, D.W, Rossiter, J.R, Redman, J.D. & Watts, R.D. 1975. The electromagnetic response of a low-loss, 2-layer, dielectric earth for horizontal electric dipole excitation. Geophysics, 40(2): 285-298. Arcone, S.A. 1995. Numerical studies of the radiation patterns of resistively loaded dipoles. Journal of Applied Geophysics, 33: 39-52. Arcone, S.A., Tobiasson, W. & Delaney, A.J. 1995. Ground Penetrating Radar Investigation of the Proposed Dome-CARA tunnel route, and utilities at South Pole Station, Antarctica. U.S. Army Corps of Engineers, Cold Regions Research & Engineering Laboratory (CRREL) Report 95-24. Benter, A., Moore, W. & Antolovich, M. 2011. Focusing Ground Penetrating Radar Images with vertical offset filtering. Progress In Electromagnetics research M, Vol 19, 183-195. Döse, C. 2011. The geological signatures of radar reflectors in ONKALO. Posiva Working Report 2011-xx, (in prep).

Hellä, P. (ed.), Ikonen, A., Mattila, J., Torvela, T. & Wikström, L. 2009. RSC-Programme-Interim Report, Approach and basis for RSC development, layout determining features and preliminary criteria for tunnel and deposition hole scale, Working report 2009-29. Posiva Oy, Eurajoki. 118 p. Joutsen, A. 2011. The use of geological data from pilot holes for predicting large fractures. Posiva Working Report 2011-xx, (in prep).

Karttunen, P., Kosunen, P., Lamminmäki, T., Pekkanen, J., Pöllänen, J., Tarvainen, A-M. and Toropainen, 2010. Drilling and Associated Drillhole Measurements of the Pilot Hole ONK-PH9. Posiva Oy, Olkiluoto, Finland. Working Report 2010-09.

Keller, G. V. & Frischknecht, F. C. 1966. Electrical Methods in Geophysical Prospecting. Oxford, United Kingdom: Pergamon Press, 517 s. (International Series in Electromagnetic Waves Vol. 10). ISBN 0 08 011525 X. Lahti, M. (ed.), Ahokas, T., Nordbäck, N., Paananen M., Paulamäki, S., Vaittinen, T. 2009. The ONKALO Area Model. Version 1.1. Posiva working report 2009-113. Kärki, A., Paulamäki, S. 2006. Petrology of Olkiluoto. Posiva report 2006-02. Posiva Oy, Eurajoki. 77 p.

62

Mancini, P., Karttunen, P., Lokkila, M., Lamminmäki, T., Pekkanen, J., Pöllänen, J., Tarvainen, A-M., Toropainen, V., Kosunen, P. & Pere, T. 2010. Drilling and the associated drillhole measurements of the pilot hole ONK-PH10. Posiva Oy, Olkiluoto, Finland. Working Report 2010-21, 176 p.

Millard, S. G., Shaari, A. and Bungey, J. H. 2002. Field pattern characteristics of GPR antennas. NDT & E International Volume 35, Issue 7, October 2002, Pages 473-482 Munier, R. 2006. Using observations in deposition tunnels to avoid intersections with critical fractures in deposition holes. SKB R-06-54, Svensk Kärnbränslehantering AB (SKB) Stockholm, Sweden. Mustonen, S., Norokallio, J., Mellanen, S., Lehtimäki, T. & Heikkinen, E. 2010. EDZ09 project and related EDZ studies in ONKALO 2008-2010. Posiva Oy, Working Report 2010-27, 404 p. Olsson, O., Falk, L., Forslund, O., Lundmark, L. & Sandberg, E. 1987. Crosshole Investigations – Results from Borehole Radar Investigations. Tukholma, Ruotsi: SKB. 188 s. Stripa Project Technical Report 87-11. Posiva Oy 2006, TKS-2006, Nuclear Waste Management of the Olkiluoto and Loviisa Power Plants: Programme for Research, Development and Technical Design for 2007-2009. Posiva Oy, Eurajoki. 285 p. Posiva Oy 2011. Olkiluoto Site Description 2011. Posiva Oy, Eurajoki. In Print. Vaittinen, T., Ahokas, H. & Nummela, J., 2009. Hydrogeological structure model of the Olkiluoto Site – update in 2008. Posiva working report 2009-15. ReflexW 2003. Version 3.0. Karlsruhe, Germany, K-J. Sandmeier. 341 p. Saksa, P. 1986. Tutkamittaukset kallion rakennetutkimuksissa. Espoo: Valtion teknillinen tutkimuskeskus, 48 s. VTT Tiedotteita 634. Saksa, P., Hellä, P., Lehtimäki, T., Heikkinen, E. & Karanko, A. 2001. Reikätutkan toimivuusselvitys (On the performance of borehole radar method). Posiva, Working Report 2001-35, 134 p. Saksa, P. Heikkinen, E. & Lehtimäki, T. 2005. Geophysical radar method for safeguards application at Olkiluoto spent fuel disposal site in Finland. STUK-YTO-TR213. Sellman, P. Delaney, A. & Arcone, S. 1992. Sub-bottom surveying in Lakes with ground penetrating radar. U.S. Army Corps of Engineers, Cold Regions Research & Engineering Laboratory Report 92-8. Sipola, V. & Tarvainen, A-M. 2007. Ground penetrating radar measurements at the ONKALO research tunnel and eastern part of the Olkiluoto investigation area at July 2006.

63

Stolt, R. H. 1978. Migration by Fourier transform. Geophysics. Tillard, S. & Dubois, J-C. 1995. Analysis of GPR data: wave propagation velocity determination. Journal of Applied Geophysics 33, 77-91. Zhdanov, M. S. & Keller, G.V. 1994. The Geoelectrical Methods in Geophysical Exploration. Amsterdam, Netherlands: Elsevier. 873 s. (Methods in Geochemistry and Geophysics. Vol. 31). ISBN 0-444-89678-3.

64