15 pages LITERATURE REVIEW (Dissertation Literature review ) in APA style. this is related to petroleum or we can say oil reservoir
Favouritewriter
ITPC 13451 Innovative Integration of Seismic and Well Data to Characterize Tar Mat in Carbonate Reservoirs T.M. Matarid, C.T. Lehmann, K.I. Ibrahim, D.O. Cobb, Abu Dhabi Marine Operating Company; A. Smith, CGGVeritas Copyright 2009, International Petroleum Technology Conference This paper was prepared for presentation at the International Petroleum Technology Conference held in Doha, Qatar, 7–9 December 2009. This paper was selected for presentation by an IPTC Programme Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the International Petroleum Technology Conference and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the International Petroleum Technology Conference, its officers, or members. Papers presented at IPTC are subject to publication review by Sponsor Society Committees of IPTC. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the International Petroleum Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, IPTC, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax +1-972-952-9435.
Abstract
This paper presents an integrated approach using the 3D seismic and well data to enhance our understanding of the lateral and/or vertical distribution of the Tar Mat. The study was carried out utilizing a recent stat-of-the-art, high resolution and high quality 3D ocean-bottom seismic dataset (OBC) acquired offshore Abu Dhabi and several wells with an excellent suite of logs, thousands of feets of core data and geochemical studies. A Model Based Acoustic Impedance Inversion was conducted following the 3D seismic reservoir mapping. A comprehensive porosity prediction analysis and validation were conducted for each well. The observation of the abrupt destruction of porosity in the well data associated with Tar Mat presence in the core led to the idea of computing the porosity derivative cube from the seismically predicted porosity cube. This significant and dramatic change in porosity associated with the Tar presence suggested that this porosity destruction might be visible in the seismically predicted porosity cube. The derivative of the porosity volume after post-stack Impedance inversion was generated to visualize the rate of changes in porosities. The high negative porosity derivative in a highly porous section may represent the top of a Tar mat. The high positive porosity derivative values also can be used to indicate Tar free developed porosity. Good match was found between the generated porosity derivative volume and the top tar from wells. Cross-plots between the seismic acoustic impedance and porosity for all wells (including Tar wells) suggest difficulty to distinguish between Tar and lithology change for porosities less than 12.5%. The lateral Tar distribution was found to be predictable utilizing this approach, through blind test well validation. The seismic Tar mat prediction on the porosity volume has provided new and important interpretation of the top of the Tar in the inter-well region and for the static model. Different Tar prediction schemes from seismic have been evaluated for further refinement. Differentiating tight rocks from the porosity plugged with tar remains ambiguous in the lower reservoir tight rocks. Therefore, a detailed sampling and geochemical analysis of the tar is being performed on the core to determine its base.
Study area The subject structure is undeveloped and located about 120 km offshore North-West of Abu Dhabi (Figure 1). The undeveloped structure is a North-South elongated anticline, approximately 15 Km by 7 Km in size. The structure relief with 1200ft is considered one of the largest in the area. The structure developed as a result of periodic deep-seated salt plug associated with basement faulting. The field was first recognized in 1955 following the interpretation of the earliest seismic survey acquired in ADMA-OPCO concession in 1954. The first exploratory well-1 was drilled 1969 followed by 8 additional wells, between 1970 and 2007, with the objective to appraise the structure and evaluated the reserves.
2 IPTC 13451
Background and motivation Following the discovery of well-1, an intensive coring program for the Arab reservoir was implemented in the subsequent wells # 2, 3, 5, 6& 7 (see Figure 2 for core with tar from Well-2, 3 and 9). Top Tar mat had been identified from the core description and from the Open Hole log interpretation. The reserve estimates post drilling Well-9 were significantly different because the estimated Tar had been encountered much deeper as per production test, core and logs and interpretation. These results encouraged further study to understand the Tar distribution. The occurrence and distribution of Tar have been identified as one of the main subsurface uncertainties impacting oil in place characterization, the structure development plans and its production. The Tar mat plugs porosity in the reservoir section and acts as a vertical permeability barrier potentially separating the aquifer from the oil filled reservoir. Therefore, understanding the lateral distribution of tar is important in well placement for the subsurface development plan. In this field the carbonate, reservoir named Arab group, and lies immediately below 200’ thick cap rock of anhydrite and can be divided into upper and lower reservoirs. The upper Arab is thin and heterogeneous reservoirs 15’-25’, consisting of carbonate dolomite and interbedded anhydrite, and lies immediately below thick cap rock. The main hydrocarbon container Lower Arab and named Arab”D” is a thick 450’ parasequence coarsening carbonate reservoir and lies immediately below the upper Arab. The cap rock “Hith” and the upper Arab are together responsible for the data quality deterioration for the underlying seismic. The thick anhydrite and the thin intercalation of dolomite and anhydrite are very likely causing energy attenuation, multiple diffraction and wavefield deformation in the upper Arab and the underlying Arab”D” seismic image. ADMA-OPCO acquired a stat-of-the-art Ocean Bottom Seismic (OBC) with offset and azimuth diversity on Q1 2007. Well driven seismic processing and true relative amplitude with zero phase data were achieved. Distinct amplitude anomaly observed from the first seismic cross-section and associated only under the structure closure. The amplitude anomaly in the N-S seismic line is interpreted to reflect the reservoir properties after seismic modeling (Figure 3). Wells located at the structure apex have almost no Tar mat, while structure flank wells shows variable Tar mat (Figure 4). Tar mat occurs in this field, primarily in the Arab C & D. The Arab ”D” is the main reservoir container of that structure and the subject work considered only the D reservoir. Regarding core description side, the lower dense Arab”D” was difficult to be distinguished from the Tar mat. Generally the Tar mat occurs in Arab”D” reservoir and tar top been picked in most of the cored wells (Figure 4). The Arab D is a coarsening upward succession with a gradual increase in porosity toward the top of the reservoir section. Arab ”D” is divided into three main sub-zones (upper-D, middle-D and lower-D. The Tar mat interpretation results from the core and Open Hole log indicated difficulty to identify the base Tar-mat since as it coincide with the lower-D dense zone or dense Diyab. The good seismic data quality with its inversion led to reasonable interwell region porosity computation with reasonable level of confidence. The observation on the computed well porosity of abrupt porosity destruction at the Tar surface encourage us consider the computed porosity volume for tar prediction.
Tar mat interpretation from Core and Open-Hole log Several wells have penetrated tar at different depths in the main reservoir Arab section (Arab”D”) (Figure 5). Tar mat interpretation and zonation from log data was a challenging because Neutron and resistivity logs having similar response with heavy oil. In addition, poor vertical resolution and dated logs made for identification difficulties. The following resistivity log characteristics were used as indication of Tar mat presence. When both the deep and shallow resistivity logs tends to read high, this suggest that there is no indication for Tar, indicating there is no invasion. That criterion has been used with the Rxo indicating much reduced movable hydrocarbon in zone where Rt shows good hydrocarbon saturation (Figure 4). The core grain density versus log grain density data were used to identify Tar mat zone in the cored wells. The criteria of 2.7 gm/cc grain density reading versus 2.71 gm/cc in limestone were used to identify Tar mat zone. The Tar mat well pick interpretation from logs using the above critetion was found to be consistent with core sample interpretation.
Seismic Modeling Extensive seismic modeling and spectral analysis were performed in parallel with the 3D processing. The objective was to generate well driven seismic processing and to understand the seismic signature from the existing well data.
IPTC 13451 3
Several conclusions are drawn from the seismic modeling: Firstly, the acoustic impedance contrast at the target Arab group is dominated by density contrasts (Figure 6) due directly to porosity. In short, seismic amplitudes can be used to map porosity in this target. Secondly, temporal resolution measurements suggest that we need to measure frequencies up to 100Hz to be able to resolve the upper individual Arab reservoirs (see also Arab A, Arab B, Arab C in figure 5). Thirdly, there is variation of amplitude with offset at the Arab reflectors which will facilitate pre-stack and or angle stacks interpretation. Synthetic seismic were generated at all existing wells with inconsistent number of cycles for the Arab reservoir at only one well which might reflects heterogeneous reservoir or poor vertical resolution logs .The seismic reflections of Arab reservoir spectral analysis show a maximum frequency of 50-60Hz with dominant frequency of 35-40Hz. Because of the modeling and spectral analysis results, the focus on Tar mat prediction from seismic only considered the lower thick Arab”D” reservoir. Based on core and log interpretation, the Arab reservoirs are heterogeneous carbonate in this structure with fair to poor reservoir properties, with porosity ranging from 2-22%. The Well-3 synthetics show a peak in front of the top tar and tie with the same peak character in the seismic data. However, it was difficult to laterally follow that peak within the cube. The calculated maximum positive amplitude map (Figure 7) for Arab “D” suggests difficulty to map that peak. The maximum positive amplitude map (Figure 7) for Arab”D” did show some correlation with the later (Figure 12) computed porosity derivative map.
Seismic Inversion A Model Based Acoustic Impedance Inversion (MBI) was conducted following the 3D seismic reservoir mapping. The model based Impedance 3D volume was generated after inversion analysis and validated with blind wells. The seismic cube data were inverted into Acoustic Impedance (AI) cube utilizing color inversion (CI) techniques as well. Focus will be directed to MBI as its results would match the existing seismic amplitude cube accurately through an iterative process and showed better correlation to porosity. The computer CPU time, incidentally, was several times greater than the CI because of the iterative process of the MBI technique. The available 9 wells in SARB structure have sonic and density logs with reasonable quality. A complete recent suite of logs including DSI were acquired with drilling the latest appraisal well SR-9, but due to operational problem 9470- 9810ft interval were not recorded for Top Hith to lower Thammama. The missing interval is un-predictable from the other wells. Individual sonic log correction for individual wells was conducted with deterministic and statistical wavelet extraction. The extracted wavelets were zero phase, but with different amplitude spectrum. The best well tie and wavelet extraction results were found at well-1. Because we will deconvolve the seismic amplitude data with a single wavelet, the decision was to select the statistical wavelet at the structure apex well, well-1, with the best tie. The residual error (seismic amplitude data-derived synthetic) and the AI prediction error were significantly lowered. Some selected well logs (AI) and two surfaces (Thammama II and Hith) were used to establish the initial Geological model (Initial AI Model). Different well combinations were tested and others were left blind to compute the inversion error. Inversion analysis at individual wells was carried out to compare the original AI well log versus the inverted results from the initial model. Inversion constraints window were used to control and steer prediction computation within well log values and towards seismic results. The computed MBI AI and the Edge detection seismic attribute cubes were found to be very useful when superimposed on the Amplitude cube for the second final phase of interpretation. The computed AI data found useful to map the top reservoir (Arab”A0”) after inverting the seismic reflection amplitude to layers similar to geology (Figure 8).
Porosity Prediction Step wise regression was carried out to choose the best seismic multi-attribute list that best model the porosity. A probabilistic neural network was used to combine the selected attributes in a non-linear manner. The training results were validated at all wells. The ability to confidently model porosity and predict it from seismic was achieved. A comprehensive porosity prediction analysis and validation was conducted for each well. The cross-plots between the seismic acoustic impedance and porosity for all wells (including Tar wells) suggests difficulty to distinguish between Tar and lithology change for porosities less than 12% (Figure 13).
4 IPTC 13451
The computed Inversion cubes were followed by computing Porosity for inter-well-region using the well porosity as a control point and the computed seismic Inversion and amplitude attributes. Several seismic amplitude attribute calculations for the Arab”D”-Diyab layer were conducted to understand the lateral distribution of the High amplitude reflections which is related to reservoir porosity as per earlier seismic modeling and the AI versus well porosity (Figure 7) The seismic amplitude attribute maps (Figure 7) over the Arab”D”-Diyab reservoir interval show an abrupt change might be linked with the stratigraphic architecture. There is a possibility of amalgamated Arab”D” grainstones with improved porosity and that is shown with a positive seismic anomaly. The abrupt change in the amplitude (Figure 7) might be linked as well to the structure growth during deposition with improved porosity on structure apex. The deteriorated amplitude anomaly between well-3 &5 can be referred to data quality due to faulting and the tar which has left only 30’ porous reservoir (see Figure 5). In addition poor reservoir porosity to the South can be related to the interpreted poor reservoir facies in the South. Prediction analysis had been conducted for individual wells to compare the target well porosity logs with the seismic amplitude trace and the other inversion cubes. Training the data is required to learn the relationship between the log porosity (PHIE) and seismic attributes through the multi attributes analysis and neural network. It should always be true that adding more attributes will predict the data better. This does not always mean that adding attributes will predict the data more reliably. Eventually, adding more attributes will simply predict the details or “noise” in the log or in the attribute themselves. Adding more attributes is similar to fitting a higher order polynomial to a set of points. The average error plot for all wells with the number of seismic attributes to properly model and validate porosity was found between 7 to 8 seismic attributes. The number of seismic attribute (10 vs. 7 attributes) was tested with variable operator length (1 point to 9 points operator) and computed the prediction error. The computed average error percentage curve shows no better log prediction using more than 7 seismic attributes. Once we have the list of attributes that give a high correlation coefficient and a small error, neural network training can be performed on that list to find the “hidden” relationship/network for predicting porosities. The input for the porosity prediction is the log porosity, original post-stack seismic and impedance volumes. The red curve is the validation error and this can help us to decide when we have added too many attributes. Each point in the validation error has been calculated by “hiding” each of the wells and predicting its values using the operator calculated from the other wells. For examples the last red point correspond to 10 attributes has been calculated away and the 10 attribute has been arranged according to the table. The first well has been removed from the calculation. The weights for the eight ten attributes have been calculated using only wells 2 to 9. The derived operator is then used to predict the value at well 1. since we already know the exact value, the RMS error for well 1 has been stored. Then we hide well two and repeatthe computation, and so on. Cross plotting the actual log porosity versus the predicted porosity shows a cluster around the perfect correlation line with 0 intercept and 1 slop with 66% overall correlation. The average prediction error found was 2.6- 3.7% with and average of 3.3%. The extracted East West cross section from the computed reservoir porosity volume shows improved porosity for the main reservoir Arab”D” over structure apex relative to structure flanks (Figure 9). The average porosity map over the main reservoir Arab”D” shows a 15% overall with possible improving average porosity in the North to 17-18% (Figure 10). The computed porosity map can be related in the North with the interpreted high porosity high permeability capped Arab”D” with Stromatoporoids which were described in well-9 core. The computed average porosity map for the Arab”D” can be divided into three sectors with different level of confidence using the traffic signal color(Figure 10). The porosity prediction in the red sector might be impacted by the Tar mat presence in SR3 (only 30’ of calculated porous Arab”D” at the top) or due to relative data quality as a results of faulting.
Porosity Derivative The lateral distribution of Tar is a significant subsurface uncertainty for both the oil-in-place characterization and for implementing the full field development for SARB. Based on well data the tar plugs porosity in the reservoir section and acts as a vertical permeability barrier (Tar Mat) potentially separating the aquifer from the oil filled reservoir. Therefore, understanding the lateral distribution of tar is important in well placement for the subsurface development plan.
IPTC 13451 5
Several wells have penetrated the Tar at different depths in the main reservoir section (Arab D). The Arab D is a coarsening upward succession with a gradual increase in porosity toward the top of the reservoir section (Figure 4). The thick Tar mat column observation at the structural peripheral wells (SR2, SR4, SR5 and SR3) might be indicative as it has been generated with the paleo-OWC. The Cross plot of the logs P-Impedance versus computed porosity for the tar wells over the reservoir level shows possibility to separate Tar mat if it is located in the top highly porous zone above 12.5%porosity and below 45000 ft/s*g/cc Figure 13. Superimposing the Porosity-Impedance cross-plots of the tar wells (highlighted in Black) with the non tar wells indicates difficulty to separate the Tar mat in the lower Arab”D” dense zone. An abrupt change of porosity especially in the upper part of the reservoir section is presumed to be due to presence of Tar in this section. This is confirmed from the detailed core description. Based on the above observation the Porosity Gradient Cube was calculated in order ideally help in identifying the blocked porosity by tar mat. The porosity derivative cube was utilized to interpret a pseudo top tar mat surface in the Arab”D” reservoir. The derivative of the porosity volume after post-stack Impedance inversion was generated to visualize the rate of changes in porosities. The high negative porosity derivative in the high porous section may represent the top of a tar mat. The high positive porosity derivative values also can be used to indicate Tar free developed porosity. Good match was found between the generated porosity derivative volume and the top Tar mat from core and logs (Figure 11). The North South derivative cross-section show the truncation of the red reflector (Tar mat) and the interpreted to Arab”D” (Figure 11).
The Red reflector was considered for 3D interpretation, depth conversion and grided with 100m X 100m as a pseudo Tar mat surface for the input to the static model. However, the Arab”D” mapped horizon was used to compute the Maximum Negative Porosity Derivative map to understand the Tar mat lateral distribution (Figure 12). The derivative map shows possible Tar mat layer to the Northern structure half and patchy in the Southern structure have. There is some sort of correlation between the derivative map (Figure 10) and the maximum positive amplitude map (Figure 6). The maximum positive amplitude map might represents the identified peak (Figure 5) for top tar in well-3 which was difficult to map. Figure 13 shows the tar intersection with top Arab “D” reservoir work progress based on guesstimate (Red outline), drilling results and paleo-owc (Orange) and from seismic porosity (Green).
Conclusion The seismic Tar mat prediction on the porosity volume has provided new and important interpretation of the top of the Tar in the inter-well region and for the static model. Different Tar prediction schemes from seismic will be further evaluated and refined. Differentiating Tar in tight rocks and to recognize the remaining porosity plugged with tar remains ambiguous in the lower reservoir tight rocks. The lateral Tar distribution was found to be predictable utilizing post stack 3D seismic acoustic impedance inversion followed by porosity prediction and its derivative volume. The seismic Tar mat prediction on the porosity volume has provided new and important interpretation of the top of the Tar in the inter-well region and for the static model. Different Tar prediction schemes from seismic have been evaluated for further refinement. Differentiating in tight rocks and to recognize the remaining porosity plugged with tar remains ambiguous in the lower reservoir tight rocks. Therefore, a detailed sampling and geochemical analysis of the tar is being performed on the core to determine the base of the tar.
Acknowledgments The authors thank the management of the Abu Dhabi Marine Oil Company for the constructive comments and permission to publish this paper. We thank peer reviewers for their comments and helpful suggestion. The subject new field reservoir model has been built by team, the author also thank the team members.
6 IPTC 13451
Figure 1 : The study location map and structure map.
Figure 2: Tar filling porous reservoir Well-2 & 3. Tar filling lower dense reservoir Well-9.
IPTC 13451 7
Figure 3: N-S seismic line with simplified Tar thickness in pink
Figure 4 : Tar versus non Tar wells (Tar Wells found to be structure flank wells)
8 IPTC 13451
Figure 5 : Arab”D” cartoon with vertical Tar mat distribution from cores and logs
Figure 6: Well-3 with abrupt Φ destruction Due to Tar mat
IPTC 13451 9
Figure 7: Arab”D”-Diyab Maximum Negative (LHS) and Maximum positive (RHS) Amplitude maps
Figure 8: AI section with cartoon for Arab reservoir sequences
10 IPTC 13451
Figure 9: Φ log with tar (Top), computed Φ section (Bottom)
Figure 10: Arab”D” Average Φ map
IPTC 13451 11
Figure 11 : N-S porosity derivative Cross section
Figure 12: Arab D based Maximum Negative Porosity map superimposed With the oil column above Tar surface
12 IPTC 13451
Figure 13: AI versus Φ for Tar wells and all wells.