| Autor: |
Kurt Chambers, Patrick Turpin, Frederik Pivot |
| Rok vydání: |
2013 |
| Předmět: |
|
| Zdroj: |
All Days. |
| DOI: |
10.2523/iptc-16847-abstract |
| Popis: |
Abstract Reservoir modeling is performed by integrated teams of geologists, geophysicists and reservoir engineers. The final grid is a simplified 'best guess' of the field being modeled due to limitations in modeling software and interpretation choices. The main features controlling the fluid flow are preserved and all that is considered secondary by the integrated team may be over-simplified. This is risky as elements considered secondary by a reservoir engineer may be important to a geologist and oversights can occur. Interpretation and uncertainty are inherent in each of the steps used to model facies and petrophysical distributions within the grid which filter through to the final reservoir model. The need to verify the final model against some reference to ensure no gross errors have been made during its construction was the driver for developing the seismic backloop technology presented in this paper. Technology 3D seismic data is the only data that is commonly available with sufficient areal coverage to be taken as reference in the Interwell space. Deterministic seismic inversion is used to extract quantitative elastic attributes, for example P-impedance (Ip) and Poisson Ratio (PR) from the seismic data. There are some known flaws with using absolute elastic attributes derived from seismic to condition reservoir models as seismic data lacks both the lowest and highest frequencies required to estimate absolute impedances. The low frequency range part is embedded into the inversion results through a low frequency model which is typically an interpolation of well log data. An under constrained low frequency model can contain artefacts which will affect the final absolute impedances. The missing high frequencies result in the estimated absolute impedances being a smoothed average of the true absolute impedances over a thickness of approximately 15m which is case dependant and is a function of seismic bandwidth and the frequency content (Francis and Hicks, 2010). Inversion accuracy also suffers in the presence of seismic noise whether it is random noise, coherent multiples or uncorrected amplitude dimming due to fault shadowing for example. At the end of the day a reference dataset is required and these limitations have to be carefully addressed when interpreting the results of a seismic backloop study. The seismic backloop verifies the consistency of the static reservoir grid properties defined in the reservoir grid against P-impedance and Poisson ratio extracted from 3D seismic data using deterministic seismic inversion. If necessary anomalies identified in the grid can have their petrophysical properties updated in a trial and error manner until the misfit between modeled IP, applying the rock physics model (RPM) to the grid and the reference inverted IP is reduced to an acceptable level. To convert the petrophysical parameters defined in the grid to elastic attributes (Ip and PR) an RPM is required. The RPM needs to be able to reliably model elastic attributes (Ip and PR) from the reservoir properties populated in the grid. A large part of the study is dedicated to ensuring this requirement is met as the confidence in the RPM along with the confidence in the inverted seismic attributes are the two limiting factors on the confidence and precision in the study results. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
