3D and 4D Seismic Data Integration for Geomodel Infilling: A Deep Offshore Turbiditic Field Case Study.

Transcription

1 IPTC MS 3D and 4D Seismic Data Integration for Geomodel Infilling: A Deep Offshore Turbiditic Field Case Study. V. Silva, T. Cadoret, L.Bergamo, and R.Brahmantio, TOTAL E&P France Copyright 2015, International Petroleum Technology Conference This paper was prepared for presentation at the International Petroleum Technology Conference held in Doha, Qatar, 7-9 December 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 , Richardson, TX , U.S.A., fax Abstract When seismic data is of good quality and that it can be related to useful geological properties it can become a useful driver to distribute spatial heterogeneity within the geomodel. This paper describes a workflow to incorporate efficiently seismic data during the geomodel infilling process. We propose to combine the seismic facies probability attribute (obtained after 3D pre-stack inversion) with 4D attributes. This is done by increasing the seismic facies probability where 4D information highlights the presence of permeable reservoir facies. Afterward the obtained attribute is combined with well information using Total in-house workflow to generate the final facies proportion cubes. Finally, Associated Facies (AF) simulation is performed using SIS (Sequence Indicator Simulation) algorithms. Comparison is performed between the AF obtained using 3D seismic data only and AF obtained using combined 3D and 4D seismic data. The integration of 4D information early in the geomodeling workflow (i.e. in facies modelling stage) improves the efficiency of the dynamic back-loop by allowing an early combination of Static and dynamic information. Thanks to this, history matching of dynamic reservoir simulation is facilitated. Introduction While defining a geological model, different pieces of information must be integrated to obtain a reliable spatial organization of geological facies and petrophysical properties. In the field case presented below, a new reservoir model is designed to improve the representation of geological facies heterogeneities for a turbiditic deep offshore field. Total in-house Seismic Reservoir Characterization tool (CARESS) enables conversion of inverted seismic attributes (such as acoustic impedance and poisson s ratio) into attributes describing the probability of occurrence of selected geological facies. It must be pointed out that these attributes have the same resolution as seismic and can possibly be influenced by pore fluid content. In the field considered for this study 4D seismic data has proven to be a key piece of information to understand the dynamic behaviour of the reservoirs. The main 4D attribute used in this study is the fractional P-wave velocity changes between two different seismic monitors (DV/V). If no subsidence effect occurs in the reservoir interval, it is assumed that strong DV/V response corresponds to some production or injection related effect which can only occur in the presence of permeable sand.

2 IPTC MS 2 Until now, 4D seismic data was integrated in geomodel workflow following facies simulation to assign deterministically reservoir facies in zones where strong 4D response is visible. In this work we propose to combine the seismic facies probability attributes and 4D attribute (DV/V) earlier on in the infilling process to better constrain the facies property infilling in the model. Seismic data QC and seismic quality map definition Before the use of the seismic data to constraint the geomodel infilling, the quality of the different seismic data were assessed to estimate the relative confidence we can have on it. The computations are performed at reservoir interval and the final product of this analysis is the seismic quality map. The seismic stacks used for seismic inversion derives from the Broadband survey. Computation of several attributes was performed for these stacks as RMS amplitudes, signal to noise analysis, standard AVO QC maps between near and far stacks after common bandwidth filtering. The stretch between near and far stacks was performed on raw data, without common bandwidth filtering. Seismic inversion products are the input for the Caress study, so its qualities have a direct impact on the potential seismic constraint. Over the reservoir interval the residuals amplitude are as expected weak and without lateral or vertical organization. In order to analyze in more detail its distribution the residuals are normalized with respect to the amplitude of the input sub-stacks. This is performed by dividing the RMS of the residuals calculated in the interval of interest by the RMS amplitudes of the reference seismic. The Caress quality assessment is performed at well scale by comparing the facies found at well with the caress prediction, section view by analyzing the continuity of the events and at base map view by analysing the Geological coherence. Jointly with the Caress cubes, DV/V is another source of seismic constraint for model infilling. The assessment of its quality is performed by analyzing the residual time shift between base and monitor survey below reservoir after warping operation using such attribute. We then look for zones showing higher time shift which would be an indication of less reliable DV/V. 4 different attributes have been selected to analyze the seismic quality. Figure 1 presents their spatial variation within a layer defined around the reservoir interval. To highlight zones of relatively poorer AVO behaviour the correlation and stretch of the seismic signal between Near and Far traces have been measured. Such attributes allows getting some hint about lateral variation in the quality of NMO correction. The correlation map exhibits higher correlation values between sub-stacks within the channels. The stretch attribute shows globally reasonable values except in some small localized zones. Signal to Noise ratio is another obvious criteria to understand the seismic quality. It has been computed using the near stack. It indicates that the seismic is noisier in the central panel area because of a highly faulted zone. It highlights also areas of poor seismic signal in the west flank of the survey where a salt diaper induces very steep seismic reflectors. A last seismic QC taken into account to define a quality indicator is a normalized RMS inversion residual map computed for the near substack. It allows to assess, at least partially, if the pre-stack inversion process has been successful to find an impedance solution explaining the seismic response. This map shows higher residuals, and therefore lower impedance reliability, in the faulted zone within the sand channels. We interpret these areas as being less reliable in term of inversion products and therefore less able to have an important weight in the reservoir model infilling. In order to define a qualitative indicator reflecting the seismic quality over our interval of interest a map is defined using some of attributes described in the previous paragraph. For each single attribute, a quality map was defined, as can be seen in Figure 1. To obtain a single attributes quality map a superposition of each maps is achieved to delineate zones where low reliability seismic can be defined. Figure 2 displays the final seismic quality map. In overall the seismic quality is high to medium, except in the salt zone and in a small zone to the west, possibly affected by shallower faulted zone. This quality map is afterward integrated to weight the seismic information put in the Geomodel.

3 IPTC MS 3 3D and 4D Seismic attributes time alignment Prior to the combination of 3D and 4D seismic attributes, a time alignment had to be performed between dataset to compensate for time shifts introduced by 4D effects and different seismic acquisitions and processing. While the Caress has been computed using a broadband seismic survey acquired while the field was in production, the 4D attribute DV/V is obtained from 3D high resolution survey referenced to the baseline time. In order to be used together the Caress and 4D DV/V attributes have therefore to be put in the same time and geographical referential as they have been acquired and migrated differently. In order to do so the time shift has been computed by measuring the amount of shift necessary to maximize the correlation between the Full stack volume coming from both dataset. 3D and 4D attributes combination and geomodel infilling The dataset for seismic constraint definition is composed of four CARESS lithoseismic probability cubes (non reservoir, laminated sand, massive soft sand and massive hard sand) related to each lithology occurrence. Three 4D seismic monitors (M1, M2 and M3) are also available derived from different vintages of high resolution seismic surveys. Figure 3 illustrates a schematic workflow for 3D and 4D seismic data combination. Most probable facies cube was computed using as input the facies probability cubes by assigning in the output cube the facies with highest probability value. Moreover, cumulative 4D attribute is calculated by adding to M3 the response of M1 and M2 above one percent cut-off of DV/V in zones where M3 does not presents a response above the cut-off. This step was performed to account for possible attenuation/compensation in 4D response in M3 monitor. Afterward, a cross validation of the most probable facies and cumulative DV/V is performed by analyzing the zones where the 4D signal presents a clear response (DV/V above 1%) and the CARESS facies probabilities does not correspond to a reservoir facies. This allows the definition of a Facies Analysis Cross-Validation attribute (FAC) highlighting the incoherent zones which require some modification of the facies probabilities. Such update is performed by increasing the initial reservoir facies probability as shown in Figure 4. This step was performed at seismic scale and afterword upscalled into the grid to be used during the model infilling workflow. Final facies proportion cubes for the model were then built using an in-house workflow (Figure 5) allowing to combine two proportion cubes per facies: a quantitative cube (or reference cube) providing the target facies proportions (mainly guided by facies proportions at wells), and a qualitative cube guiding spatial distribution of the facies (derived from lithoseismic probability cubes providing 3D trends per facies). A seismic quality map derived from 3D seismic attributes QC has also been used to vary spatially the weight given to the seismic constraint introduced as qualitative cube. One final proportion cube is defined for each facies (5 cubes, one per AF). These cubes were then used to distribute the facies in the model using Sequence Indicator Simulation (SIS). AF facies were modelled stochastically (Figure 5) in the reservoir grid using the following inputs: Well AF upscaled (1D) at wells Facies proportion cubes Local Varying Azimuth (LVA) property computed in channelized Architectural Elements (AE4, AE5 & AE6) Vertical and Horizontal Variography The result of facies modelling was used as basis for Petrophysical modelling in the reservoir grid. The petrophysical simulations were performed using the Sequential Gaussian Simulation (SGS) algorithm. First of all, the Effective Porosity ( e) was simulated. Given the good relationship of e with Net Effective Porosity net, Net To Gross (NTG) and Permeability (K), these parameters were co-simulated with e.

4 IPTC MS 4 Two model infilling using seismic constraint were performed : one using only Caress lithoseismic probability cubes and a second one using combined Caress lithoseismic probability cubes and 4D DV/V attribute. Results Discussion and Conclusions Assessment of the quality of the input seismic data is a key step for an optimum integration of the seismic data in the geomodel infilling workflow. Even though the analysis based on various seismic attributes remains qualitative, this assessment has allowed defining a 2D seismic confidence map which was later integrated in the compromise workflow to balance the seismic weight in the output facies proportion cubes. The spatial distribution of the geological facies within the reservoir grid obtained with both 3D only or integrated 3D and 4D attributes exhibits a good coherency with the geologic context. Nonetheless, the combination of 3D attributes (CARESS) and 4D attributes (DV/V) has increased the sand content of the geomodel in comparison to the model infilling using the CARESS attribute alone. Statistics in AF for both models show that reservoir facies proportion within the main channel area is higher when 4D seismic data is integrated in the infilling workflow. More importantly, this integration is improving the reservoir facies continuity as illustrated in Figure 6. QC was also performed on the NTG per AE. It was observed a maximum NTG error (in comparison to well input) of 3%,. This difference was considered as acceptable having to take into account the multiple constraints used by the simulation (e.g. Seismic, Variograms with local variation, well constraint). 3D petrophysical properties coherence were also validated by a seismic back-loop study aiming to check the compatibility between actual inverted seismic and synthetic elastic properties (IP, PR). These synthetic attributes are obtained thanks to a petroelastic model using as input the geological model petrophysical properties. This exercise has allowed demonstrating that modelled P-impedance and Poisson s Ratio are globally less than 6% different from the inverted values. This has been considered has a sufficient level of coherency to validate the model infilling from this point of view. To conclude it is important to stress the benefit of this integration of 4D information early on in the geomodel infilling workflow. Indeed, it makes the dynamic modelling more reliable by allowing an early combination of static and dynamic information. This has improved the dynamic loop by reducing the time needed to history match the reservoir model with production data. Acknowledgments The authors would like to thank TOTAL for permission to publish this paper. The authors would like also to thank Hildebrando Vicente-Pedro for assistance in the Facies modelling workflow. References Sengupta S., Cadoret T., Pivot F., (2014). Semi-Automatic Facies Up-scaling Techinique for Litho-Seismic Classification Application to a field located in Western Offshore Africa. IPTC Paper presented at International Petroleum Technology Conference in Doha Hubans C., Cauquil E.C., Brechet E. (2014). 4D (time lapse) seismic: an emerging tool for underwater monitoring. OTC paper presented at Offshore Technology Conference held in Houston 2014.

5 IPTC MS 5 Correlation Near/Far Near/Far Stretch Seismic Confidence Low Medium High Mean Relative Inversion Residual Near Substack MeanSignal/Noise Near Substack Fig. 1: Seismic QC attributes and confidence contours. Top left - correlation between near and far sub-stack, top right - stretch between near and far sub-stack. Bottom left mean relative inversion residual for near sub-stack, bottom right mean signal to noise ratio for near sub-stack.

6 IPTC MS 6 Confidence Low Medium High Fig. 2: Seismic Quality Map highlighting zones with low, medium and high seismic confidence. CARESS (Facies Probabilities) 3D + 4D seismic data Combination Time Alignment to Baseline reference 4D DV/V (M1, M2, M3) Seismic Quality Map UPDATED CARESS Compromise workflow Fig. 3: Schematic workflow for 3D and 4D seismic data combination.

7 Confidence Low Medium High IPTC MS 7 Fig. 4: Caress lithoseismic probability update using 4D data. Red ellipse highlights a zone where 4D anomaly is visible but no reservoir facies exists. Ref. cube Context / applicability of unchanged the method (2) Aux. Cube modified Step 1: proportions of the auxiliary cube are modified to honor the proportions of the reference cube Objective of the proposed method: given proportion targets (the reference cube), add to it the shapes, the spatial organization contained Ref. cube Aux. Cube in the auxiliary cube or, ain x other unchanged words, modify+b the global x modified proportions of the auxiliary cube in order to honorthose of the reference cube The workflow uses the second option: With a + b =1 To keep the proportions of the reference cube 10 - Références, date, lieu Step 2: Combination. In Case of VPC or global proportions, only the auxiliary cube is used (a=0, b=1) Ref. cube Aux. cube Ref. cube unchanged Aux. Cube modified Step 1: proportions of the auxiliary cube are modified to honor the proportions of the reference cube a x 10 - Références, date, lieu Ref. cube unchanged +b x x Aux. Cube modified With a + b =1 To keep the proportions of the reference cube Step 2: Combination. In Case of VPC or global proportions, only the auxiliary cube is used (a=0, b=1) Fig. 5: Compromise and Facies modeling workflow. Reference proportions from well data and auxiliary proportions from seismic data. Seismic quality map used to weight the seismic information (alpha parameter).

8 IPTC MS 8 AF CARESS No 4D 27.1% Statistics AF without 4D AE6 AF No 4D 23.2% 15.8% 17.7% 16.2% AF CARESS + 4D Statistics AF with 4D AE6 AF_4D 30.3% 22.2% 12.9% 17.6% 17% Fig. 6: Comparison between AF property generated using only 3D as seismic constraint and AF generated using combined 3D and 4D seismic.

9 Paper No D and 4D Seismic Data Integration for Geomodel Infilling: A Deep Offshore Turbiditic Field Case Study V.A. DA SILVA*, T. CADORET, L. BERGAMO, R. BRAHMANTIO

10 CONTEXT Motivations to perform the study OUTLINE Slide 2 1 SEISMIC DATA QC AND CONFIDENCE MAP DEFINITION Assessing the spatial variability of the seismic quality 3D AND 4D SEISMIC DATA COMBINATION How to integrate 3D and 4D attributes in the geomodel infilling workflow? GEOMODEL INFILLING Impact of the 4D integration in the outcome? SEISMIC BACKLOOP Validating the petrophysical infilling CONCLUSIONS Key messages

11 CONTEXT Slide 3 This study was performed in a deep offshore turbiditic field Study motivated by the need of an updated model integrating new data 4D attribute (dvp/vp) has proven to be useful as an additional lithological indicator This presentation will show a workflow to integrate efficiently the seismic information in the geomodel infilling workflow Seismic has also been used to quality control the geomodel infilling

12 CONTEXT Motivations to perform the study OUTLINE Slide 4 1 SEISMIC DATA QC AND CONFIDENCE MAP DEFINITION Assessing the spatial variability of the seismic quality 3D AND 4D SEISMIC DATA COMBINATION How to integrate 3D and 4D attributes in the geomodel infilling workflow? GEOMODEL INFILLING Impact of the 4D integration in the outcome? SEISMIC BACKLOOP Validating the petrophysical infilling CONCLUSIONS Key messages

13 SEISMIC DATA QC AND CONFIDENCE MAP DEFINITION Slide 5 Seismic Attributes QC Bad Bad Seismic Quality Map Good Bad Good Bad Boundary compilation Good Good

14 CONTEXT Motivations to perform the study OUTLINE Slide 6 1 SEISMIC DATA QC AND CONFIDENCE MAP DEFINITION Assessing the spatial variability of the seismic quality 3D AND 4D SEISMIC DATA COMBINATION How to integrate 3D and 4D attributes in the geomodel infilling workflow? GEOMODEL INFILLING Impact of the 4D integration in the outcome? SEISMIC BACKLOOP Validating the petrophysical infilling CONCLUSIONS Key messages

15 Initial Most Probable Facies 3D AND 4D SEISMIC DATA COMBINATION 4D attribute Cumulative DV/V Slide 7-3% -1% LithoSeismic Facies Check the Consistency +1% +3% Most Probable Facies Post Update Zones to be updated Abs(DV/V)>1% & MPF=0 Update the inconsistent zones

16 3D AND 4D SEISMIC DATA COMBINATION: Spatial Continuity QC Slide 8 Most Probable Facies before 4D Update: Most Representative in Layer 4D Signed Absolute Maximum in Layer Most Probable Facies after 4D Update: Most Representative in Layer Attributes Combination LithoSeismic Facies

17 3D AND 4D SEISMIC DATA COMBINATION: QC at well Slide 9 GR AF Sand Probability before update 4D DV/V B00M2 Sand Probability After update Associated Facies (AF) LithoSeismic Facies WOC Cut off used for facies update Increased Water Sand probability

18 CONTEXT Motivations to perform the study OUTLINE Slide 10 1 SEISMIC DATA QC AND CONFIDENCE MAP DEFINITION Assessing the spatial variability of the seismic quality 3D AND 4D SEISMIC DATA COMBINATION How to integrate 3D and 4D attributes in the geomodel infilling workflow? GEOMODEL INFILLING Impact of the 4D integration in the outcome? SEISMIC BACKLOOP Validating the petrophysical infilling CONCLUSIONS Key messages

19 Confidence Low Medium High From seismic Litho-Cubes + 4D The workflow uses the second option: a x Architectural Elements AF/AE on wells 10 - Références, date, lieu Ref. cube Ref. cube unchanged FACIES MODELLING WORKFLOW Proportion of AF per AE (CARESS) Ref. cube unchanged AF Target Proportions a x 10 - Références, date, lieu +bx With a + b =1 Ref. cube unchanged Aux. cube Proportion cube from CARESS, Ref. cube one cube per AF unchanged «spatial Aux. organisation Cube» Constant per AE, one cube per GAF «target proportion» To keep the proportions of the reference cube Aux. Cube modified Slide 11 Step 1: proportions Ref. cube of cube are modified to proportions of the ref Ref. cube unchanged Cubes Aux Step SIS2: Combination. Ref. cube In C Aux. Cube or global a x unchanged proportion + MODELLING +b O Step x 1: modified proportions of the auxiliary Wit Cubes PC auxiliary cube is use To keep the propor M cube are modified to honor the 10 - Références, date, lieu Final modified With Proportion cube P a + proportions b =1 Facies of the reference cube with spatial To keep the proportions of the reference organisation cube R Proportion respecting the target proportion, + O one Cubes cube per AF M Step 2: Combination. In Case of VPC Cubes Aux. Cube Ref I or global proportions, only the modified S auxiliary cube is used (a=0, b=1) E

20 AF Lithocubes No 4D Geomodelling Results Statistics AF without 4D In AE6 27.1% 23.2% Slide % 17.7% 16.2% AF2 AF3 AF4 AF5 AF6 AF Lithocubes + 4D Statistics AF with 4D In AE6 Observation: 30.3% Sand proportion more in line with wells results 22.2% 12.9% 17.6% 17% AF2 AF3 AF4 AF5 AF6

21 CONTEXT Motivations to perform the study OUTLINE Slide 13 1 SEISMIC DATA QC AND CONFIDENCE MAP DEFINITION Assessing the spatial variability of the seismic quality 3D AND 4D SEISMIC DATA COMBINATION How to integrate 3D and 4D attributes in the geomodel infilling workflow? GEOMODEL INFILLING Impact of the 4D integration in the outcome? SEISMIC BACKLOOP Validating the petrophysical infilling CONCLUSIONS Key messages

22 3D Geomodel SEISMIC BACKLOOP (SBL) WORKFLOW 3D Inverted cubes Slide 14 Update Petrophysics Comparison is done in the Geomodel space

23 Reservoirs Non-Reservoirs SEISMIC BACKLOOP AT WHOLE GRID Slide 15 Initial mismatch analysis on the whole 3D grid DIP = RPM - INVERSION DPR= RPM - INVERSION Non-Reservoirs +DPR 0 -DIP Vcl +DIP PR IP 0.6 Rock Physics Model Display Strong Mismatch Moderate Mismatch Small Mismatch Evaluate the impact -DPR of petrophysical properties (VCL, PHI, SW) on the Elastic IP-PR) attributes by color code to obtain the Summary X-plot

24 Slide 16 SUMMARY X-PLOT AND STATISTICAL ANALYSIS OF ANOMALY BODY DIP = RPM - INVERSION DPR= RPM - INVERSION IV +DPR I IP Overestimation Possible interpretations: IP Percentage Difference 1. Overestimation of SW 2. Underestimation of PHI 12,0% 10,0% 8,0% 6,0% 4,0% 2,0% 0,0% SLAP_Initial SLAP_Final Anomaly Threshold -DIP +DIP 3. Combination of options 1 and 2 But High Stakes FACIES III -DPR II 1. Bodies located below WOC (options 1 and 3 are unlikely to happen ) 2. Most likely a porosity effect Properties Update SLAP_Inial: Initial infilling SLAP_Final: porosity increased by 3pu 0 PHIE 0.5

25 CONTEXT Motivations to perform the study OUTLINE Slide 17 1 SEISMIC DATA QC AND CONFIDENCE MAP DEFINITION Assessing the spatial variability of the seismic quality 3D AND 4D SEISMIC DATA COMBINATION How to integrate 3D and 4D attributes in the geomodel infilling workflow? GEOMODEL INFILLING Impact of the 4D integration in the outcome? SEISMIC BACKLOOP Validating the petrophysical infilling CONCLUSIONS Key messages

26 CONCLUSIONS Slide 18 Seismic constraint allows to distribute realistic lateral heterogeneities and increases the reservoir facies continuity Seismic quality map permit weighting spatially the seismic constraint during model infilling The integration of 4D information has reduced the time needed to history match the reservoir model with production data Seismic Backloop shows that model in-filling allow to globally honor the inverted 3D seismic

27 Slide 19 We thanks Total and its partners for the authorization to present this work Thank You / Questions