76th EAGE Conference & Exhibition 2014 Amsterdam RAI, The Netherlands, 16-19 June 2014

We P13 09Seismic Quantitative Interpretation of a GasBearing Reservoir - From Rock Physics toVolumetricP. Alvarez* (Rock Solid Images), F. Bolivar (Rock Solid Images), W. Marin(Rock Solid Images), M. Di Lucas (Pacific Rubiales Energy) & T. Salinas(Pacific Rubiales Energy)

SUMMARYA complete well-seismic integration project was successfully performed in a gas field located in the LowerMagdalena Basin, Colombia. The reservoir level corresponds to a laterally limited, mixed clastic turbidtesystem (laminated, shaly sands) which makes prospect identification and delineation difficult. The studycan be divided in three phases: well log analysis, seismic analysis and interpretation/integration. Well loganalysis included petrophysical volumetric estimation and rock physics modeling. Seismic analysesincluded gather conditioning and pre-stack simultaneous inversion. The interpretation/integration phasepulled together the results from the phases described above. Based on the rock physics modelling and therock physics template, there is a clear separation between gas shaly sands and other facies (wet shaly sandsand shales) in the Poisson’s ratio vs. P-impedance domain. During the interpretation phase, seismicinversion attributes guided by the rock physics information made it possible to separate the shaly gas sandsin order to identify new prospective areas in the field. New successful wells were drilled, validating theworkflow.

76th EAGE Conference & Exhibition 2014 Amsterdam RAI, The Netherlands, 16-19 June 2014

Introduction

One of the aims of quantitative interpretation is to predict inter-well reservoir properties leveraging information from rock physics models and seismic attributes. This methodology was successfully applied in fluid prediction of mud-rich deepwater turbidites. Three phases were used in order to complete this work. The first, petrophysics and rock physics analysis, which includes the volumetric estimation, well log conditioning and rock physics modelling, was carried out to understand the relationship between reservoir properties (clay volume, porosity and fluid saturation) and elastic rock properties (P- and S-impedance, Poisson ratio, etc.). The second phase includes seismic gather conditioning and pre-stack simultaneous inversion. Finally, rock physics and seismic inversion results were calibrated and integrated with geologic information in order to define zones with a higher probability of gas content. These results were subsequently used to guide a successful drilling campaign in the survey area.

Well log and rock physics analysis

One of the earliest and most important steps to integrate well data for a seismic reservoir characterization is an integrated analysis of the log curves over the full well-bore. It includes the standard petrophysical volumetric estimation (lithology, porosity and fluid content), and the well log conditioning based on rock physics diagnostics. Lithology estimation was done via linear spectral gamma ray and the water saturation using Simandoux equations. The soft sediment model (Dvorkin and Nur, 1996) provided the best match between measured and modelled logs in intervals with good quality log data, and, as a consequence, this model was used to correct density and sonic curve in intervals affected by bad borehole conditions. S-wave velocity edits were made using Greenberg-Castagna equation. The resulting conditioned logs and volumetric estimation were used to evaluate the relationship between the petrophysical and elastic properties in the cross-plot domain. Figure 1a, 1b & 1c show a multi-well cross-plot of Poisson’s ratio (PR) vs. P-impedance (Ip) at seismic scale colour coded by water saturation, clay volume and porosity respectively.

a) b)

c) d)

Figure 1. Multi-well cross-plot of PR vs. Ip at seismic scale colour coded by a) water saturation, b)

clay volume, c) porosity. d) Rock physics template showing the same relationship between the elastic and petrophysical properties observed at the multi-well cross-plot.

76th EAGE Conference & Exhibition 2014 Amsterdam RAI, The Netherlands, 16-19 June 2014

Gas-bearing shaly sands tend to have lower Ip and PR than wet sands and shales. Figure 1d shows a rock physics template (Avseth et. al. 2005) that is useful for validate and extrapolate the relations observed in the multi-well cross-plots. These relations, obtained from well-log data and rock physics models, provide a guide for the quantitative interpretation of the seismic inversion attributes.

Gather conditioning and pre-stack simultaneous inversion

Gather conditioning and pre-stack simultaneous seismic inversion were performed following the approach of Singleton (2009) and Tonellot et. al. (2001), respectively. Signal-to-noise ratio, offset-dependent frequency loss and gather alignment were the properties addressed in the gather conditioning. Figures 2a, 2b, 2c & 2d show a comparison between conditioned seismic and synthetic data allowed quality control of the results.

Figure 2. a-b) Comparison between pre-stack seismic and synthetic (right) data, c) post-stack well tie, d) comparison between real and synthetic AVA response, e) extracted wavelets per angle stack.

The data used as input for the seismic inversion were partial angle stacks, extracted wavelets per angle stack (Figure 2e) and Ip & Is low frequency models (LFM). The Ip LFM was built combining calibrated seismic interval velocities with up-scaled well data. To that effect, seismic velocity was calibrated to log data in order to compensate for weak anisotropic effects. Then, a density volume was created from seismic velocities through a transform estimated from upscaled well data, which made it possible to estimate an Ip volume from seismic velocities (Figure 3b). Next, the low frequency component provided by the up-scaled well-log data was propagated consistent with the constraint of conformable stratigraphy, by an ordinary kriging algorithm (Figure 3a). Finally, both models were combined at equal proportions obtaining a LFM that honored the well data while retaining the inter-well trends provided by the seismic velocities (Figure 3c).

a) b) c)

Figure 3. Horizon slices of Ip LFM based on a) well data, b) seismic data and, c) well and seismic data combined in the same proportions. Is LFM was estimated based on a linear relationship.

76th EAGE Conference & Exhibition 2014 Amsterdam RAI, The Netherlands, 16-19 June 2014

Quantitative interpretation

The quantitative interpretation starts with the rock physics analysis. This step, as shown in Figure 1, allowed us to identify the differences between pay and non-pay zones based on their elastic properties. After the completion of this analysis, it was necessary to perform a reservoir properties calibration. This step has two goals: the first is to make sure that both seismically derived and well-log derived elastic attributes have the same background trend and magnitude (Figure 4). The second goal is to establish the cut-off parameters for the isolation of the gas bearing sands using a polygon-based cross-plot approach.

Figure 4. Cross-plots of PR vs AI. a) Seismically derived elastic attributes, b) seismically derived elastic attributes color coded by point density, c) well-log derived elastic attributes color coded by

water saturation overlain on seismically derived elastic attributes

Figure 5a shows a cross-section through the PR cube along two of the wells drilled in the area, together with its logs of PR (seismic scale) and Sw. Notice the good match between the seismic and well-log derived PR attribute, as well as the expected relationship between zones of low PR and gas bearing sand. Black lines show the boundaries between eight stratigraphic intervals defined in the area. Figure 5b shows the resultant geobodies associated with zones of high probabilities of being gas-bearing sand in interval 5, obtained after the isolation of the gas bearing sands using a polygon-based approach. Most of the highlighted geobodies are conformable with structure and located up-dip, which add confidence to the results.

SwPR

Well-D Well-B

0.33

0.45

Well-CWell-B

Well-D

b)a)

Poisson’s ratio

(+) (-) (-) (+)

Figure 5. a) Cross-section through the PR cube along two of the wells drilled in the area. b) Geobodies associated with zones of high probabilities of being gas-bearing sand in the interval 5.

76th EAGE Conference & Exhibition 2014 Amsterdam RAI, The Netherlands, 16-19 June 2014

Net pay volume estimation

For each one of the eight stratigraphic intervals, different sets of polygons were defined using the rock physics analysis results, to highlight zones with higher probabilities of being gas-bearing sand. Next, the resultant geobodies were filtered, based on different criteria, to eliminate noise and keep only the larger and connected geobodies. Then, a thickness map in time was calculated, and converted to depth using a velocity model. As a result each stratigraphic level had a generated thickness map in depth, which, for the interval 5, is shown in Figure 6a. Figure 6b shows the map obtained after stacking the thickness maps obtained from the eight stratigraphic intervals. This final map, which represents a net pay map for the target formation, was used to estimate the net pay volume, and can be used to reduce risk in the drilling of new wells in the field.

X

X

X

X

X

X

X

X

Thickness Depth (ft)

Well-A

Well-B

Well-CWell-D

Well-A

Well-B

Well-C

Well-D

a) b)

Figure 6. a) Thickness map in depth of the geobodies identified in the interval 5. b) Total thickness map in depth of the stacked geobodies identified in the eight stratigraphic intervals

Conclusions

Calibration between rock physics results and seismic attributes permitted identifying the 3D distribution of zones with a higher probability of gas saturation. As a direct result of this reservoir characterization study, two new locations were drilled in the field, both of which penetrated producer levels and were since classified as discovery wells. This successful result allowed us to validate the current workflow and update the geological, geophysical and petrophysical model for the future exploration and exploitation plan in the area.

References

Avseth, P., Mukerji, T. and Mavko, G. [2005]. Quantitative seismic interpretation. Applying rock physics to reduce interpretation risk. Cambridge, NY, Melbourne: Cambridge U. Press.

Dvorkin, J., and Nur, A. [1996]: Elasticity of high-porosity sandstones. Theory for two North Sea data sets. Geophysics. 61. p. 559-564.

Singleton, S. [2009]. The effects of seismic data conditioning on prestack simultaneous impedance inversion. The Leading Edge, 28(7), 772–781.

Tonellot, T., Macé, D., and Richard, V. [2001]. Joint stratigraphic inversion of angle limited stacks. SEG Technical Program Expanded Abstracts 2001: pp. 227-230.