We P2 04 Rock Property Volume Estimation Using the Multiattribute Rotation Scheme (MARS) - Case Study in the South Falkland Basin P.K. Alvarez* (Rock Solid Images), B. Farrer (Borders & Southern Petroleum), M. Suda (Rock Solid Images) & D. Oyetunji (Rock Solid Images) SUMMARY This paper shows a case study where a seismic reservoir characterization was carried out by integrating well-log and seismic inversion data through the application of the multi-attribute rotation scheme (MARS) methodology (Alvarez et al., 2015). MARS is a hybrid rock-physics/statistical approach designed to yield the optimum seismic inversion attribute correlation to target reservoir properties. The method performs a comprehensive assessment and selection of all possible attribute combinations, ensuring the optimum rock-property calibration for each geologic condition within a given data-set of seismic and well information. This workflow was applied on the Darwin Field, located in the South Falkland Basin. The Darwin structure comprises two adjacent tilted fault blocks: Darwin East, which contains the discovery well 61/17-1, and Darwin West, which remains untested (Farrer and Rudling, 2015). From this workflow customized transforms were found from the well-log data to estimate reservoir properties from seismically-derived elastic attributes. The resultant rock property volumes (Sw, Vclay and total porosity.) characterize the reservoir s heterogeneity, and can be used as inputs for static model generation, reserve estimation, as well as to optimize the exploration, and exploitation plan in the area.
Introduction Estimating information about reservoir properties from seismic data is a key challenge in exploration, appraisal and production of hydrocarbons. We show how to perform quantitative reservoir characterization by integrating well-log and seismic inversion data through the application of the multiattribute rotation scheme (Alvarez et al., 2015). We demonstrate our methodology on the Darwin structure located in the South Falkland Basin along the southern margin of the South American plate. The goal of our workflow was to estimate seismically-derived volumes of reservoir properties (fluid, lithology and porosity) to characterize and delineate the proven and potential reservoirs in the area. Geological Setting The Darwin structure comprises two adjacent tilted fault blocks: Darwin East, which contains the discovery well 61/17-1, and Darwin West, which remain untested (Farrer and Rudling, 2015) (Figure 1a). Darwin has a good quality, quartz rich sandstone reservoir. Net pay in the discovery well was determined as 67.8m, with porosity up to 30%, averaging 22% (Figure 1b). The reservoir consists of one major early Cretaceous sand unit that extends across the two fault blocks and is clearly represented by amplitude anomalies on 3D seismic. The reservoir encountered in the Darwin discovery well (61/17-1) is interpreted to be shallow marine sandstone comprising predominantly quartz, but with some feldspar, lithic fragments and clays. (Farrer and Rudling, 2015). Figure 1 (a) 3D view of the Darwin structure. (b) Petrophysical evaluation of the well 61/17-1. Multi-attribute rotation scheme (MARS): Method and Theory The multi-attribute rotation scheme (MARS) is a hybrid rock-physics/statistical approach that uses a numerical solution to estimate a transform to predict petrophysical properties from elastic attributes (Alvarez et al., 2015). The transform is computed from well-log-derived elastic attributes and petrophysical properties, and posteriorly applied to seismically-derived elastic attributes. Figure 2 shows a sketch illustrating the methodology. Figure 2 Sketch of a cross-plot of two attributes colour coded by a target property. Dashed grey lines represent new attributes estimated via axis rotation, and the blue line represents the attribute that shows the maximum correlation coefficient with the target petrophysical property. MARS uses an exhaustive evaluation of all possible n-dimensional spaces (formed by n attributes) and angles to find an attribute τ that exhibit the global maximum correlation with the target petrophysical property.
Equation 1 shows the mathematical expression for the attribute for the specific case of two attribute dimensions used for the prediction τ = A1 SF A1 sin(θ i ) + A2 SF A2 cos (θ i ) (1) Where A1, A2, are elastic attributes; SF A1, SF A2, are scale factors, which are applied to equalize the order of magnitude of the attributes; and i, is the angle where the maximum correlation is reached. The final step in this workflow is to scale the attribute to units of the target property. This is done using equation 2, where, the coefficients m and c, can be estimated by fitting a line between and the actual petrophysical property. Target Property = m τ + c (2) Reservoir characterization and delineation using the MARS approach Figure 3a show the well-log data used as an input for the MARS application. This data consist of both, fundamental elastic properties (Vp, Vs and density) and target petrophysical properties, such as water saturation (Sw), total porosity ( t) and clay volume (Vclay) logs. Since the final goal of MARS is to predict petrophysical properties using seismically-derived elastic attributes, the first step applied consists of filtering the input logs to seismic resolution (black logs in Figure 3a). Figure 3 (a) Input well-log data for the MARS analysis. From left to right: Vp, Vs, density, Vclay, Sw and. t Red curves show the original logs, while black curves show upscaled versions to seismic resolution. (b) Set of attributes used in the MARS run. Each number represents a single attribute, which is obtained after applying the mathematical operation shown in the leftmost column to the uppermost row. For example, the number 21 represents the attribute 1 λρ. MARS was applied to estimate seismically-derived volumes of Sw, Vclay and t. For each case MARS was run evaluating all the possible 2D combination of the 64 elastic attributes shown in Figure 3b, which can be derived from Ip and Is, resulting in the assessment of 2016 independent bidimensional spaces. Figures 4a to 4e shows the results obtained after running MARS on the well 61/17-1 for Sw estimation. In this case the global maximum correlation between the attribute τ and Sw was found in the attribute space 1 λρ versus 1 Kρ at 61, with a correlation of 0.9671. Figure 4a shows the resultant parameters that were used in equations 1 and 2 to estimate an Sw transform from elastic attributes. Figure 4b shows a crossplot of θ versus the correlation coefficients between the derived set of attributes (estimated via axis rotation) and the Sw log showing the maximum correlation for θ=-61. Figures 4c, 4d, and 4e show a comparison between the actual and predicted Sw in the crossplot space 1 λρ versus 1 Kρ and in the spatial domain, showing an excellent match between the actual Sw log, estimated though a petrophysical analysis and the Sw log estimated from elastic attributes using the MARS analysis. Figures 4f and 4g show a comparison between the actual and predicted Vclay and t respectively, obtained after run MARS using these logs as target property. Notice that for these cases it was also obtained a very good correlation coefficient between the actual and predicted reservoir property log of 0.9602 and 0.9574 respectively.
Figure 4 (a) Parameters used in equations 1 and 2 to estimate Sw from elastic attributes at seismic resolution scale. (b) Crossplot between θ versus the correlation coefficients between the Sw log and the set of attributes estimated via axis rotation, (c and d). Comparison between the actual and predicted Sw log in the crossplot space 1 λρ versus 1 kρ. Grey arrows, orthogonal to the blue lines, indicate the maximum direction of change of Sw in this attribute space. (e, f and g) comparison between the actual and predicted Sw, Vclay and logs respectively. t Next, the resultant transforms were applied to seismically-derived elastic volumes to obtain a volume of reservoir properties. Cross-section of the resultant volumes of Sw, Vclay and t. along the Darwin West and Darwin East structure are shown in Figure 5. In this figure, it is possible to see a good match between the seismic and well-log-derived reservoir property. Notice that in the Sw volume it was possible to identify the presence of fluid contacts in the Darwin East and West structures, and in the Vclay volume the good lateral continuity of the shallow marine reservoir rock and the cap rock can be seen. In the porosity volume a decrease of porosity with depth in the reservoir rock which can be reproduced by a compaction trend can be observed. Finally, the spatial distribution of the reservoirs were mapped by cross-plotting the seismically-derived Sw and t volumes and backpropagating the areas with the best petrophysical properties (Figure 6a). From this, geobodies were created, and from these thickness was computed with the goal of creating a net pay thickness map of the reservoir (Figure 6b). The resultant map has a very good agreement with the structure and can be used for reserve estimation and to optimize future well locations. Conclusions For the Darwin field, reservoir characterization and delineation was carried out by applying the MARS methodology. From this workflow customized transforms were found from the well-log data to estimate reservoir properties from seismically-derived elastic attributes. The resultant reservoir property volumes (Sw, Vclay and t.) allow us to characterize the reservoir s heterogeneity, and can be used as inputs for static model generation, reserve estimation and to optimize the exploration, appraisal and exploitation plan in the area. Reference Alvarez, P., Bolivar, F., Di Luca, M. & Salinas, T. [2015] Multi-attribute rotation scheme: A tool for reservoir property prediction from seismic inversion attributes, Interpretation, 3, SAE9-SAE18. Farrer, B. & Rudling, C. [2015] South Falkland Basin: Darwinian Evolution, Geoexpro, 12(1).
X Y Figure 5 Cross-section of the resultants volume of Sw, Vclay and. t along the Darwin West and Darwin East structure, together with the log information of the well 61/17-1: Vclay (left) & Sw (right). Y X Figure 6 (a) Cross-plot of the seismically-derived Sw and t volume in the interval A-B (see Figure 5). The polygons shown were used to create geobodies related to the best reservoir properties in terms of hydrocarbon saturation and porosity. (b) Net pay map of the reservoir in two-way time. Black thick lines show the main faults in the area. The red dashed line indicates the location of the cross-sections shown in Figure 5.