Abstract
This work concerns logfacies modeling for an onshore oil field of Reconcavo Basin, Bahia state, Brazil. Data from geophysical logs and core description of only one well was used to model 33 wells that did not have core information, but have a suite of logging curves which includes neutron pulsed logs. Logfacies modeling was performed according to discriminant statistical rules, resulting in synthetics facies for all (34) wells. In order to check the efficiency of logfacies prediction on the 33 wells for which core description was nor available, the logfacies columns were compared with a hydrocarbon saturation curve derived from pulsed neutron data. A very good match was found for all wells, suggesting the used procedure as suitable for logfacies modeling, at least for this oil field. Introduction According to Soares (2005) the logfacies modeling, which uses statistical techniques, can be defined as the attempt to recognize the facies column of a well from its geophysical logs. This is an important activity that allows the construction of the geological model of the area, even in the absence of continuing cores. The geological model can be constructed in many ways. A component that differentiates each of these ways is the input data used in the construction of the model affecting the responses in terms of uncertainty and resolution. A model created from core’s description, for example, has desirable qualities, such as reliability and resolution. However, well core extraction is expensive and it is not always possible to recover the whole interval. Thus, the availability of cores is usually limited. On the other hand, the exploration area has, commonly, ample availability of geophysical logs data, which have variable degree of uncertainties and resolution. This work aims to apply the strategy of logfacies modeling described by Soares (2005) in an oil field in Reconcavo basin. Method This study uses a data package consisting of various geophysical logs from 34 wells and core description of just one well. Each well contains curves like gamma ray, caliper, resistivity, density, neutron, photoelectric factor, and several curves from pulsed neutron log. For the logfacies modeling done in this study we used statistical techniques of discriminant analysis and clustering, which are the main statistical techniques used to determine logfacies (Souza Jr., 1992; Soares, 2005). Using statistical techniques of supervised and nonsupervised classification (Soares, 2005; Albuquerque et al., 2004; Hair et al., 2005), through successively applying of clustering and discriminant rule, best results for stratigraphical refinement of this field were achieved. SAS statistical software was used to obtain these results. Initially, a step-by-step discriminant analysis was used in order to choice the logs for logfacies modeling. As a result of this step, the curves ROHB (density log) and YCA (calcium yielding from neutron pulsed log) showed the most discriminating power, and they were therefore chosen. Considering the diversity of original facies, 13 in total, it was decided to define an ideal number of synthetic facies for easy identification. Logs, in general, recognize smaller number of facies than those recognized by the geologist responsible for the core description. So, it is necessary to reduce the number of facies to be recognized through logfacies modeling. Clustering analysis was done by average linkage and centroid techniques, using statistical indicators of the optimal number of groups, such as pseudo-F, pseudo-t and CCC. For average linkage technique the pseudo-F statistics indicates good numbers of groups when this statistics shows high values. Thus, for our data, average linkage technique indicates 2 or 3 as a good number of facies to try identification (Figure 1). The pseudo-t indicator shows high values for the number of facies immediately prior to the ideal number, in this case, good numbers would be 2 or 3 (Figure 2). For CCC, positive values greater than 2 or 3 are indicative of good numbers of facies, values between 0 and 2 values indicate potential number of facies, whereas high negative values are indicative of the presence of outliers, so, according to Figure 3 a good number of facies to be recognized would be 2. In the case of centroid technique of clustering analysis the numbers 4 or 5 would be ideal for the pseudo-F statistics (Figure 4), 3 for pseudo-t statistics (Figure 5) and 1 for CCC indicator (Figure 6). Based on an intersection between the values from the three statistics above, it was decided to recognize 3 facies: a non-reservoir, a good-reservoir facies and a reservoir with oil potential. It is important to remember that now the recognized facies aren’t the same of the original facies from the core description. Thus, the main thirteen lithofacies from core description in the P08 well
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.