Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
special topic first break volume 26, June 2008 Leveraging Technology How EM survey analysis validates current technology, processing and interpretation methodology Ståle Johansen,* Ketil Brauti, Stein Fanavoll, Hans Amundsen, Tor Atle Wicklund, Jens Danielsen, Pål T. Gabrielsen, Lars Lorentz, Michael Frenkel, Benjamin Dubois, Ole Christensen, Kathrine Elshaug and Stig A. Karlsen of Electromagnetic Geoservices (EMGS)** provide the most comprehensive study to date of how the company’s offshore electromagnetic survey method (seabed logging) for the E&P oil and gas industry successfully compares with well evaluation results. T he value of subsurface resistivity data has never been in doubt in the exploration community. In the past, this information only came from well logs. Now, electromagnetic survey methods can acquire subsurface resistivity data without drilling. EMGS has offered controlled-source electromagnetic (CSEM) services commercially since November 2002. Since then, the company and the oil and gas industry have gained experience with the method. The EM surveying technique used by EMGS, which is known as seabed logging, and the company’s EM imaging methods are now used routinely by major operators in many offshore basins of the world. Some seabed logging surveys have been performed before drilling to predict the presence of hydrocarbons (Eidesmo et al., 2002). Others have been performed after drilling to calibrate the technique. The desire of the industry to benchmark the performance of the method, as a stand-alone exploration tool or integrated with conventional geophysical techniques, is understandable. However, the experience of individual operators cannot easily be extrapolated to validate the technique in all applications. The number of surveys for a particular company is often too small to form a representative sample, and each survey differs in many respects: geological setting, survey geometry, available data, local knowledge and the experience of those interpreting the data. In late 2007, with a solid body of data collected from more than 300 seabed logging surveys, EMGS set out to demonstrate that its measurements consistently, within physical limitations, live up to expectations – that EM methods can reliably describe the resistivity of bodies in the subsurface. The analysis, summarized here, looked at the surveys to determine whether the presence, lateral extent, and depth of known resistive bodies in the subsurface could be estimated from seabed logging data. The analysis was confined to those survey areas where wells had been drilled and well data were available from the operator. * Definition of success The study aimed to define the success of the method, not the success of individual surveys in finding hydrocarbons. A survey that, when processed and interpreted, is consistent with the physical reality of the target formation is defined here as being a seabed logging technical success. Therefore, a survey is deemed a seabed logging technical success if the well result (in terms of the resistivity distribution) is correctly predicted by the interpreted EM response. Surveys that showed no response are also included in the statistics. For example, a survey without an EM response is a success in these terms if drilling confirmed the absence of anomalous resistive bodies in the survey area. A seabed logging technical success is therefore distinct from a commercial success and can occur regardless of the actual features predicted. About the dataset Data from 52 selected wells drilled in areas surveyed using seabed logging methods was made available to EMGS and used for the study, which commenced in December 2007. The seabed logging technical success of the surveys over all the areas penetrated by these wells has been analyzed. These wells are a significant proportion of those drilled in areas that have been surveyed, but EMGS is not privy to information about the total number and the operators had their own reasons for selecting well data to be made available for the study. The sample cannot, therefore, be considered completely random. That said, the analysis covered surveys with water depths from 250-2500 m and target depths from 200-2500 m. The wells were also geographically spread, from the Far East through India to the Mediterranean, West Africa, and Norway, and included a well in Brazil. However, no well data was made available for the Gulf of Mexico, despite several EM surveys in the region, or for surveys conducted in Australia or Canada (Figure 1). Correspondent E-mail: [email protected]. www.emgs.com. ** © 2008 EAGE www.firstbreak.org 83 special topic first break volume 26, June 2008 Leveraging Technology Figure 1 Geographical location of the surveys used in the EMGS study. Study methodology In some cases, stand-alone EM survey data can be sufficient for identifying the presence of resistive hydrocarbons, but integration with seismic data is required to reduce risk. To determine whether EM surveying is a reliable method for mapping subsurface resistivity, the 52 datasets were re-evaluated by experienced geology and geophysics professionals using all the available data, including well data. Many were comprehensively reprocessed. It is worth noting that, for analysis of seabed logging technical success rates, it is irrelevant whether a survey was recorded before or after the well was drilled. The evaluation steps included attribute analysis and modelling using all the geo-information made available by the operator. Inversion schemes or depth migration (Mittet et al., 2005) were also used for some of the surveys. The depths to resistors were estimated by either interpretation of attributes or using EM depth conversion techniques, such as depth migration or inversion. The attributes computed and analyzed included the normalized magnitude versus offset (NMVO); the phase difference versus offset (PDVO); the phase slope versus offset (PSVO) for single receivers; the line response through simultaneous evaluation of all offsets; and, where several lines were recorded, the areal response displayed over bathymetry maps. By including forward EM modelling (Maaø, 2006) in the interpretation workflow, an initial total background response based on all the available data was computed. In selected cases, this included applying inversion techniques 84 to constrain the initial model. The model was then refined by minimizing the error function between the modelled (synthetic) and the acquired data. In this process, as with all geophysical interpretations, the experience and knowledge of those performing the work were essential to creating realistic models that were consistent with the data. Inversion methods of varying levels of sophistication were also used to interpret the survey data. The methods use a systematic, iterative approach to optimize the resistivity model by comparing synthetic and acquired data. The post-survey availability of well logs, however, helps to calibrate the inversion results once a well has been drilled. In the study, inversion generally provided an EM-driven resistivity model of the subsurface that was a good starting point for interpretation and integration. Common midpoint (CMP) inversion is a quick, computationally efficient and robust multi-frequency method of obtaining a subsurface resistivity section that uses planelayer modelling of data (Mittet et al., 2007). It was used in the study to provide estimates of the background resistivities and of the resistivities of hydrocarbon-filled reservoirs. Owing to the plane-layer assumption in the method, the final resistivity-depth models were verified using 3D forward modelling. Three-dimensional inversion accommodates detailed 3D resistivity variations in the subsurface and is well-suited for analysis of 3D EM datasets. Rigorous 3D inversion of the EM data was applied on some datasets acquired by one EMGS client: since the study was performed, EMGS has made 3D inversion available to all clients. This method www.firstbreak.org © 2008 EAGE first break volume 26, June 2008 special topic Leveraging Technology Figure 2 Interpreted EM responses were classified according to the nature of the target. (a) Hydrocarbon discovery. (b) Hydrocarbon discovery with anti-models. (c) Dry formations. (d) Dry formations with anti-models. (e) Hydrocarbon discoveries modelled sub-detection. will therefore be applied to surveys that are added as the study continues. Inversion is a processing step, albeit an advanced one, that yields non-unique results. Even a carefully executed inversion yields several different results, only one of which best reflects the true subsurface resistivity. With sufficient information and experience, it should, in principle, be possible to identify which is the most plausible, but as in all geophysical interpretations, this depends on the ability of the interpreter. EMGS attributes its ability to correctly interpret a high proportion of the 52 surveys analyzed to its experience of more than 300 surveys. Lessons learned include the value of thorough basic processing of data, which is a pre-requisite for successful completion of more advanced processing and interpretation steps. Seabed logging technical success rates The interpreted EM responses were classified into five groups according to the nature of the measured target (Figure 2): N Hydrocarbon discoveries – wells revealed hydrocarbonfilled reservoirs N Hydrocarbon discoveries with anti-models – wells revealed hydrocarbons in addition to a subsurface formation with an anomalous resistivity not due to hydrocarbons (e.g. salt, volcanic, hydrates) N Dry formations – no hydrocarbons detected N Dry formations with anti-models – no hydrocarbons were detected from well data, but the well penetrated a subsurface formation with anomalous resistivity N Hydrocarbon discoveries modelled sub-detection – well data revealed hydrocarbon-filled reservoirs with properties shown by modelling to be unfavourable (too small, too deep or with too little resistivity contrast) for detection by the method. Classification as a seabed logging technical success, in the case of discovery wells, means that the reservoir identified from the well data was correctly interpreted from the EM data. Rigorous EM depth conversion was not performed on the earliest survey data; for these cases, the criteria for a © 2008 EAGE www.firstbreak.org seabed logging technical success did not include the depth to the discovered hydrocarbon reservoir. When migration and inversion were introduced by EMGS, the criteria for success were gradually expanded to include depth to target. Surveys were also categorized as a seabed logging technical successes when the interpretations showed no resistive anomalies and the well was dry. Formations with anomalously high resistivity unexplained by trapped hydrocarbons, which are described as anti-models, were present in some wells in the study. In these, a seabed logging technical success was recorded when the target and the anti-model signal were correctly interpreted from the EM data. The few sub-detection cases in the study arose most often when the subsurface information described a target that, based on forward modelling, should be undetectable on an EM survey with the acquisition parameters used. Despite the absence of a modelled survey response, the operator in these cases elected to acquire seabed logging data and to drill a well. Figure 3 Statistical distribution of seabed logging technical success for the different target classifications. Green is a seabed logging technical success, while yellow and red are ambiguous or not fully understood. 85 special topic first break volume 26, June 2008 Leveraging Technology In the case of a discovery, the survey was not considered to be a failure if post-drilling information confirmed that the drilling target in fact was below the seabed logging detection threshold. These surveys were classified as seabed logging technical successes with discoveries modelled subdetection, because the EM response was as expected in the context of the information about the actual reservoir. Such cases have given valuable information about the thresholds for the critical input parameters used for calibration of 3D forward-modelling software. Moreover, re-modelling has also shown that some cases that were sub-detection with the acquisition parameters used during the survey might have been detected using different acquisition parameters that are more common today or by using more recent survey strategies. For example, using survey grids rather than lines; adopting a different survey layout; acquiring data at larger offsets; using a stronger source; or operating the source in different frequency ranges might have improved the dataset. In most cases considered in the study it was clear, after completion of processing and interpretation, whether a survey was a seabed logging technical success. However, for a few surveys, the results of the interpretation did not conclusively explain the situation found in the well. In the analysis, these cases have been classified as ‘not fully understood.’ Results The seabed logging technical success rate was shown to be high for the 52 surveys analyzed (Figure 3). However, it was impossible to determine whether the sample was truly representative of the overall level of success for the more than 300 EM surveys recorded because most survey targets will not be drilled. The analysis found that more than 85% of the surveys considered could be identified as seabed logging technical successes. Of these 45 seabed logging technical successes, 25 were discoveries (six with and 19 without anti-models), and 15 were dry (eight with and seven without anti-models). A further five seabed logging technical successes reflected surveys over discoveries that, as expected, could not be resolved by the method - sub-detection successes. Further efforts are being made to understand the survey responses in the remaining seven cases that have been classified as not fully understood. Work on categorizing these definitively is expected to be possible with additional input or when techniques are further refined. And, as more well data becomes available from other survey areas, this will be incorporated into the analysis. The following examples illustrate the study methodology and the classification of seabed logging technical successes. Figure 4 When the inverted seabed logging data was overlaid on the seismic section an anomaly that closely aligned with the seismic prospect was seen. In the CMP inversion red indicates high resistivity formations. 86 www.firstbreak.org © 2008 EAGE special topic first break volume 26, June 2008 Leveraging Technology They include two surveys that agree with discovery well data, one with and one without an anti-model, and two surveys with anti-models in which the wells were dry. Hydrocarbon discovery Murphy Oil Corporation wanted to evaluate prospects identified from seismic data in a deepwater turbidite and shale sequence offshore Borneo, Malaysia. In nearby wells, the equivalent turbidite sandstones were 50-100 m thick and had good reservoir quality (Mittet et al., 2007). To reduce the risk of drilling into non-productive sands, Murphy decided to integrate EM imaging into its workflow. Three EM survey lines were acquired during October and November 2006. On one of the lines, two seabed logging anomalies were identified. A simple box model with a deep resistor produced a synthetic EM response that matched the measured data. However, attribute analysis indicated the presence of a second, shallower anomaly. So, once the first resistor had been modelled, a second resistor was added to account for the remaining error between modelled and measured data. Further confirmation of the deeper resistor came from CMP inversion of the EM data, which showed an anomaly that closely aligned with the seismic prospect (Figure 4). A pre-drilling seabed logging technical success was recorded when Murphy drilled the prospect and made a significant gas discovery at the predicted depth. Hydrocarbon discovery with anti-model Shell drilled a well in deep water South East Asia (Smit et al., 2006). The well targeted a flatspot at the flank of an anticline but was dry with minor hydrocarbon shows. Seismic amplitudes indicated shallow gas at the anticline crest which suggested that the trap might have expelled its content: a scenario consistent with the prospect location in a thrust play scenario. The shallow gas reduced the quality of seismic data at the anticline crest, which made further investigation of up-dip prospects challenging using seismic methods. After the first well, Shell acquired a seabed logging survey over the prospect. A significant EM anomaly just up-dip of the original well was mapped. The interpretation concluded that the anomaly indicated a potential high-saturation reservoir below the shallow gas. A second exploration well was drilled. This contained commercial hydrocarbons as predicted by the EM interpretation. In this case, the seabed logging data challenged the existing perception of the seal effectiveness on the top of the structure, as it was used to correctly image a high-resistivity anomaly below the shallow gas. Dry formation with anti-model A seabed logging survey was acquired for Oil and Natural Gas Corporation in the Krishna-Godavari basin, offshore © 2008 EAGE www.firstbreak.org Eastern India, in June 2006 (Engenes et al., 2008). The seabed logging survey showed a strong response that indicated increased subsurface resistivities. 3D modelling (before and after drilling) and depth migration linked the EM response to a thick (>300 m) resistive body buried approximately 600-1000 m below the seabed. The well drilled through a thick sequence of limestone and calcareous sediments with relatively high resistivities (20-30 ohm.m), representing a seabed-logging anti-model. This survey demonstrated that seabed logging data can detect subsurface high-resistivity layers; in this case the anomaly was very clear. Moreover, it illustrated the importance of 3D modelling and geological data in differentiating between hydrocarbon reservoirs and anti-models. Dry formation anti-model In 2006, PETRONAS Carigali conducted a seabed logging survey in a block in Southeast Asia. The prospect area was in relatively shallow water (200-500 m) with rugged seabed topography and complex geology dominated by a faulted anticlinal structure, shallow carbonates, and a shallow basement. The primary target was expected to be near the crest of the basement anticline. These factors combined to make the dataset challenging to interpret. Initial processing revealed a well-defined EM anomaly. The anomaly was still present after more advanced processing removing the effects of shallow water and rugged sea floor. Several 3D geological models were simulated, but a satisfactory match between real and synthetic data was not achieved with the original model. By introducing a significantly increased basement resistivity Figure 5 The top figure shows the match between the survey response and the computed response to the model shown beneath; green indicates a satisfactory match between the real and the modelled data, red denotes that the resistivity put into the model is too high, and blue that the resistivity is too low. The error computation indicated that an increased basement resistivity could on its own explain the response over the prospect area. 87 special topic first break volume 26, June 2008 Leveraging Technology a better match between the modelled response and the real data was achieved (Figure 5). A pre-committed well drilled in the prospect area in 2007 encountered only minor hydrocarbon shows. However, the well log showed a high-resistivity basement (50-200 ohm.m), which was in accordance with the interpreted EM response. the technique have believed for several years: EM methods can, within certain physical limitations of the method, reliably map the resistivity distribution in the subsurface. When the industry accepts this, EM surveying will become a standard tool in the exploration toolbox and operators will routinely include it as part of their exploration workflows. Conclusions Using current interpretation methods, and with the benefit of experience gained over more than 300 surveys, EMGS has demonstrated that EM surveying and imaging methods can determine the presence or absence, the lateral extent and, for the latest surveys, the depth of resistive subsurface targets within the present resolution and capabilities of the method. The analysis shows that if comprehensive pre-drill data is available and is integrated with EM survey data by experienced interpreters using the most recent processing methods, EM surveys can be expected to deliver reliable resistivity images of the subsurface. It is clear from the work to date that, as seabed logging techniques mature and experience is gained, a higher level of understanding follows. This will open the way for EM surveying to become a standard exploration technique which will be integrated naturally into the exploration work flow. For this to happen, it will be necessary for operators to build expertise and experience themselves and to have close co-operation with the service companies offering EM services. The exploration industry has long sought a measurement that would enable the presence of hydrocarbons to be detected before drilling. The hope of the EM surveying industry is that, in the near future, this and similar studies will establish beyond doubt what the proponents of References Chandola, S.K., Karim, R., Mawarni, A., Ismail, R., Shahud, N., Rahman, R., Bernabe, P. and Brauti, K. [2007] Challenges in shallow water CSEM surveying: A case history from Southeast Asia. International Petroleum Technology Conference. Eidesmo, T., Ellingsrud, S., MacGregor, L.M., Constable, S., Sinha, M.C., Johansen, S., Kong, F.N. and Westerdahl, H. [2002] Sea Bed Logging (SBL), a new method for remote and direct identification of hydrocarbon filled layers in deepwater areas. First Break, 20(3), 144–152. Engenes K., Josyulu and Amundsen H.E.F [2008] Calibration of Sea Bed Logging (SBL) survey results and well data in block KGDWN-2002/1, Krishna-Godawari Basin. SPG 7th International Conference & Exposition on Petroleum Geophysics. Maaø, F. [2006] Fast finite difference time domain modeling for subsurface electromagnetic problems. 76thSEG Annual International Meeting. Mittet, R, Maaø, F, Aakervik, O.M. and Ellingsrud, S. [2005] Depth migration of SBL data. 67th EAGE Annual Conference and Exhibition. Mittet R., Maulana H., Brauti K. and Wicklund T. A. [2007] CMP inversion of marine EM data. 69th EAGE Annual Conference and Exhibition. Smit D. and Wood P.R [2006] Experience is crucial to expanding CSEM use. World Oil, September, 37-43. (FUUJOHOPUJDFE "EWFSUJTJOHJOPVSKPVSOBMTXPSLT O 'JSTU#SFBL O /FBS4VSGBDF(FPQIZTJDT O (FPQIZTJDBM1SPTQFDUJOH P QFDUJOH PT H 7JTJUXXXFBHFPSHBEWFSUJTJOH PSDPOUBDUPVSBEWFSUJTJOHEFQBSUNFOU BUBEWFSUJTJOH!FBHFPSH 88 www.firstbreak.org © 2008 EAGE