Project

3D geological property modelling

Updates
0 new
0
Recommendations
0 new
0
Followers
0 new
10
Reads
0 new
18

Project log

Michel Garcia
added a research item
L’objet de cette étude était de réaliser une analyse bibliographique des méthodes géostatistiques adaptées à la caractérisation d’aquifères dans la perspective d’évaluer les temps de migration de radionucléides dans des aquifères situés à proximité de stockages souterrains de déchets radioactifs. Pour appréhender les enjeux liés à ce type d’application et pouvoir envisager une démarche géostatistique systématique, le site de l’Est a été choisi par l’ANDRA comme référence et deux scénarios d’impact d’un stockage sur un aquifère ont été définis. Le premier scénario envisage la mise en communication directe entre le stockage et l’aquifère par un puits qui traverse l’ensemble (cas d’une source ponctuelle de radionucléides). Le second porte sur la migration verticale de radionucléides depuis le stockage jusqu’à l’aquifère (cas d’une source diffuse de radionucléides). Une réflexion préalable, fondée sur les données disponibles de ce site et les scénarios d’impact proposés, a été menée pour établir les grandes lignes d’une démarche géostatistique qui répondrait aux objectifs recherchés et permettrait d’exploiter au mieux les informations disponibles sur un site. La démarche décrit l’ensemble des étapes à franchir depuis l’analyse des données jusqu’à la simulation des écoulements et du transport en passant par la modélisation structurale de l’aquifère, la simulation des types de faciès et des propriétés, et le changement d’échelle souvent nécessaire pour pouvoir faire tourner un simulateur d’écoulement et de transport à partir d’un modèle géologique de grande résolution.
Béatrice Yven
added 4 research items
The geological exploration of the Meuse/Haute-Marne area began in 1994. Several boreholes were drilled, and the Callovo-Oxfordian argillite, thought to become a potential storage formation, were cored and logged. 2D seismic surveys were completed, as well as geological field observations, and an underground research laboratory was created. A 250 km2-wide Transposition Zone was delimited, which was subject to further investigations in 2007 and 2008, including another series of coring and logging in four additional boreholes, and a tighter 2D seismic survey. From this work, a restricted 28 km2 wide zone of interest (ZIRA) was identified as possible future repository area (Figure 1). In 2010, a 3D seismic survey was carried out to derive from it depth-converted seismic horizons and a 3D high resolution impedance cube that is 37 km2 wide and covers the ZIRA. They were used to build a structural and geological model of the 3D seismic survey area that is suitable for safety assessment. The structural model is presented in the associated article of Yven et al. (2015) [1]. This article focuses on the geological model with the estimation of flow and transport rock properties.
The research conducted by the French Radioactive Waste Management Agency over the past 20 years has served to demonstrate the feasibility and safety of deep disposal of high-level wastes and intermediate-level long-lived wastes. Today, this research is helping prepare for industrial disposal centre's construction and operation. The Callovo-Oxfordian claystones (COX) have been selected as a potential host formation. Therefore, the geometry of this formation is required to position the repository, design its shape and numerically simulate its behaviour. This article will present the structural model of the COX formation, estimated from a depth-converted 3D-high resolution seismic cube acquired in the contemplated repository area (ZIRA). As no borehole is located inside this area, the calibration process was performed using three 2D-seismic lines passing through the 3D seismic cube and three boreholes located 2-3km away from the ZIRA. Three strong reflection events were picked on the seismic data within the COX. The deepest event is correlated with the base of the COX (LS0), whereas the two other events represent the limits of the third-order depositional sequences (LS1, LS2) (Figure 1). The primary aim of structural modelling was to estimate the top of the claystone formation (SNC). Furthermore, the limits of the main petrophysical units, defined in a previous study from borehole log data and cores [1], also had to be estimated within the ZIRA. The petrophysical units are the Clayey unit (UA), the Transitional unit (UT) and the silto-carbonated unit (USC) as shown in Figure 1. They are essential to better understand and characterize the spatial variability of rock properties within the COX formation. The associated article of Garcia et al. (2015) shows how the petrophysical units were taken into account to quantify the spatial variability of rocks properties within the seismic cube [2]. The estimation of the different key horizons relies on marker data from boreholes where all the horizons are identified. They were used to establish relationships between non-seismic and seismic (reflecting) horizons in terms of (smoothed) bivariate histograms between either unit thicknesses or horizon elevations (Figure 2). The bivariate histograms were then used to estimate and quantify the spatial uncertainty of the non-seismic horizons within the ZIRA. A seismic facies analysis was eventually carried out to verify that the estimated horizons were matching the facies changes. This analysis proved to be efficient to confirm the major transitions from the Dogger to the COX, and from the COX to the Oxfordian limestones platform. As expected, however, it showed that there is no clear transition between the two units UA2 and UA3 at the seismic scale within the ZIRA (Figure 1). Figure 3 represents the elevation map of the base (LS0) and thickness of the COX formation in the ZIRA. These results confirm that the thickness increases in the northeast direction and that the dip is low (1°) and oriented nord-west.
Introduction In the context of a deep geological repository of high-level radioactive wastes, the French National Radioactive Waste Management Agency (Andra) has conducted, over the past 20 years, an extensive characterization of the Callovo-Oxfordian argillaceous rock in the Eastern Paris Basin. This research has served to demonstrate the feasibility and safety of deep disposal of high-level wastes and intermediate-level long-lived wastes. Today, this research is helping prepare for industrial disposal centre's construction and operation Therefore, the geometry and physical rock properties of the Callovo-Oxfordian claystone formation (COx formation) are required to position the repository, design its shape and numerically simulate its behaviour. This article details the sedimentological background and the results of the geological modelling of the COx formation in the contemplated repository area. Geological exploration and associated data The geological exploration of the Meuse and Haute-Marne departments (eastern border of the sedimentary Paris Basin) began in 1994. Several boreholes were drilled, and the COx formation, thought to become a potential storage formation, were cored and logged. 2D seismic surveys were completed, as well as geological field observations, and an underground research laboratory was created at 500 m depth to investigate the in situ formation properties through long-term experiments. A 250 km2-wide Transposition Zone was delimited, which was subject to further investigations in 2007 and 2008, including another series of coring and logging in four additional boreholes, and a tighter 2D seismic survey (170 km). From this work, a restricted 28 km2 wide zone of interest (ZIRA) was identified as possible future repository area. In 2010, a 3D seismic survey was carried out to derive from it depth-converted seismic horizons and a 3D high resolution impedance cube that is 37 km2 wide and covers the ZIRA. They were used to build a structural and geological model of the 3D seismic survey area that is suitable for safety assessment. Callovo-Oxfordian formation The claystone formation is 160 ± 5 million years old (middle Callovian to middle Oxfordian p.p.), at least 130 m thick and found between 400 and 600 m below ground. From the base to the top of the layer, the COx formation is divided into three third-order depositional sequences and three main petrophysical units (the Clayey Unit (UA), the Transitional Unit (UT) and the Silto-Carbonated Unit (USC)) defined in a previous study from borehole log data and cores [1] (Figure 1). These petrophysical units allow the identification of fine variations in the mineral, physical and mechanical properties occurring within the COx formation. The UA unit represents the mostly clayey levels of the formation, whereas the UT and USC units are siltier and more carbonated compared to the UA units. At the scale of the study area, the mineral proportions vary according to depth mainly due to geological sedimentary cycles; the clay mineral content is roughly anti-correlated to the carbonate content. The rock physical–chemical and petrophysical properties of the COx mudstone are also relatively well known thanks to a large set of studies based on core samples extracted from several boreholes and well log data. The top of the formation consists of interbedded argillaceous layers, carbonate rock and more or less silty marls. The COx is more homogeneous in its central part, with an average value of 45-50 % clay-mineral content and a maximum value reaching 62 %, which corresponds to maximum flooding within the area at the time of deposition. Structural and THM properties modelling: methodology and results To preserve intact the target formation, no borehole is located inside the 3D seismic survey. Therefore, the calibration process was performed using three 2D-seismic lines passing through the 3D seismic cube and three boreholes located 2-3km away from the ZIRA. Three strong reflection events were picked on the seismic data within the COx. The deepest event is correlated with the base of the COx (LS0), whereas the two other events represent the limits of the third-order depositional sequences (LS1, LS2) (Figure 1). The primary aim of structural modelling was to estimate the top of the claystone formation (SNC). Furthermore, the limits of the main petrophysical units also had to be estimated within the seismic cube. These units are essential to better interpret and characterize the spatial variability of rock properties within the COx formation. We will present how the petrophysical units were taken into account to quantify the spatial variability of rocks properties within the seismic cube. The estimation of the different key horizons relies on marker data from boreholes where all the horizons are identified. They were used to establish relationships between non-seismic and seismic (reflecting) horizons in terms of (smoothed) bivariate histograms between either unit thicknesses or horizon elevations (Figure 2). The bivariate histograms were then used to estimate and quantify the spatial uncertainty of the non-seismic horizons within the ZIRA. A seismic facies analysis was eventually carried out to verify that the estimated horizons were matching the facies changes. This analysis proved to be efficient to confirm the major transitions from the Dogger to the COx, and from the COx to the Oxfordian limestones platform. Figure 3 represents the elevation map of the base (LS0) and thickness of the COx formation in the ZIRA. These results confirm that the thickness increases in the north-east direction and that the dip is low (1°) and oriented nord-west. Figure 1: Geological units and stratigraphic sequence of the COx formation: Clayey Unit (UA), Transitional Unit (UT), Silto-Carbonated Unit (USC), top and base of the formation (SNC, LS0 respectively), 3rd-order T-R sequences limits (LS1, LS2). Figure 2: Scatterplots and associated smoothed bivariate histograms (density maps), between horizons (top left) or thicknesses, allowing to estimate the top of the COx formation (SNC) and the top of the petrophysical units (UT, UA and UA2). Figure 3: Elevation map of the base (LS0) and thickness of the COx formation in the ZIRA. Safety assessment studies require rock properties of the Callovo-Oxfordian claystone formation as input to simulate Thermo-Hydro-Mechanical phenomena that should result from the implementation of a geological repository in the ZIRA. Up to eight types of petrophysical, flow and transport rock properties must be characterized and modelled in the COx: clay content, total porosity, bulk density, thermal conductivity, specific heat, diffusion coefficient, accessible porosity for different components and permeability. The intended repository being the source of all the physical phenomena to simulate, the geological model must be precise enough in and around the ZIRA to capture the spatial distribution and the heterogeneity of rock properties that are consequential for safety assessment. As no borehole are available within the 3D seismic survey area where the rock properties must be estimated, relationships had to be inferred, between the seismic impedances (P-wave and S-wave impedances IP and IS) and the rock properties, from faraway wells where both types of data are present. These relationships were then used to estimate the spatial trends and spatial uncertainty of rock properties from 3D seismic data in the COx. Not all rock properties were successfully estimated this way. Only clay content, total porosity, bulk density and thermal conductivity were correlated with IP and IS, the clay content being best correlated. The proposed approach required to address several issues. • Change of support issues to relate at best rock property data representative of a small scale (30 cm for well-logs data, less for lab data), to 3D seismic impedance data corresponding to 7.5 m high support. The way to proceed to upscale rock properties depends on the available data (regularly spaced logs data vs. sparse lab data on cores) and the additive nature or not of the property (arithmetic vs. nonlinear averaging). • Data integration issues to combine strongly correlated IP and IS data as spatial trend information on rock properties. Multivariate polynomial regression was used to derive and extrapolate “geophysical” estimates of the rock properties from IP and IS. The latter being known everywhere within and around the ZIRA, the “geophysical” estimates were used as spatial trend information about the corresponding rock properties. • Bivariate statistical issues to infer and extrapolate bivariate distribution models that fully and properly describe the nonlinear relationships between “geophysical” estimates and rock properties. An innovative regression-based bivariate histogram smoothing method was used (Figure 4). • Uncertainty estimation issues to quantify local uncertainty about rock properties from previously inferred bivariate distribution models. The smoothed bivariate histograms fully defining the relationships between “geophysical” estimates and rock properties, they were used to derive confidence intervals. This approach allowed to estimate rock properties of the COx everywhere within the 3D seismic survey area and to quantify local uncertainty (Figure 5). The article will briefly detail the approach and will present results obtained for different rock properties. Figure 4: Scatterplot and associated smoothed bivariate histogram (density map) of clay content vs. its “geophysical” estimate. Figure 5: 3D view of mean clay content estimated at the 3D seismic grid scale (6 m high). References [1] Garcia M.H., Rabaute A., Yven B. and Guillemot D., Multivariate and spatial statistical analysis of Callovo-Oxfordian physical properties from lab and borehole logs data: Towards a characterization of lateral and vertical spatial trends in the Meuse/Haute-Marne Transposition Zone, Physics and Chemistry of the Earth, 2011, Vol. 36, Issues 17–18, 1469–1485.