Fishbone Workflow - A combine approach for improving the classification of RRT and the estimation of Permeability/Porosity: A Case Study by Ana de Sousa

Fishbone Workflow - A combine approach for improving the classification of RRT and the estimation of Permeability/Porosity: A Case Study by Ana de Sousa

Introduction

The definition of Reservoir Rock Types (RRT) is a key element of permeability and saturation modelling in complex clastic reservoirs, where detailed characterization is required. A field case in Central Asia is hereby presented, where the main reservoir belongs to the Lower Cretaceous (Aptian), and is highly heterogeneous:  it is composed of stacked, thinly interbedded and partially amalgamated marine sandstones and heteroliths/shales forming tortuous and complex architectures.

In such a complex reservoir, a combined approach was carried out to improve the classification of RRT and the estimation of permeability and porosity, by using an integrated Fishbone Workflow (Figure 1).  The first stage used the core data to define the Rock Quality Index (RQI) and Flow Zone Indicator (FZI), to arrive at the characterization of Hydraulic Flow Units (HFU). These HFU were quality controlled and confirmed, before leading to the RRT definition.  A Neural Network (NNW) was applied in order to obtain porosity and permeability for the cored and non-cored wells. On the case study, this approach has resulted in a robust RRT classification and a more confident porosity and permeability estimation.

2- Proposed Methodology

All the data were obtained from the conventional core analysis (12 wells) and from wireline logs (98 wells). Before any data analysis was done, a tight quality control was carried out, to remove non-representative data points (e.g. fractured and broken plugs). Porosity and permeability underwent corrections (stress and Klinkenberg corrections) to simulate reservoir conditions.

2.1    Definition of FZI

Initially, a relationship between core porosity and horizontal core permeability in a semi-log plot of each core sample was established (Figure 2). This figure shows that it is not feasible to use a linear trend, as the scatter is significant. This scatter can be attributed to lithology and facies differences, which indicates the presence of several rock types in the reservoir, each one with different fluid flow properties.

It is clear that porosity alone is not enough to explain permeability variations. To overcome this limitation, a combined approach was implemented by using the following equations (Amaefule et al, 1993):

RQI= 0.0314√k/ɸz

where RQI – Rock Quality Index, k – Permeability, ɸz – Porosity Normalization Index

ɸz=(ɸ/(1-ɸ))

where ɸ – Porosity of the core

FZI=RQI/ɸz

where FZI – Flow Zone Indicator

This approach consisted in using the FZI, together with the RQI and Normalized Porosity, in a histogram representation, to identify the number of HFU.

Figure 3 provides an indication of the FZI classes. Such classification is confirmed in a log-log plot by considering the steep slopes changes as being different HFU (Figure 4). The cumulative plot allowed the identification of 5 HFU with their respective FZI boundaries. After the validation of the script for the cored wells, the same was applied for the non-cored wells to propagate the HFU to all the wells in the field.

2.2    Neural Networks for Porosity and Permeability Estimation

For the porosity and permeability estimation, a NNW methodology was applied. Initially, a careful selection of the logs to be used was done by applying PCA (Principal Components Analysis). For porosity, the set of logs Bulk Density, Gamma Ray, Micro Resistivity and Photoelectric factor were chosen. Whereas for permeability, the chosen set of logs comprised Bulk Density, Gamma Ray, Micro Resistivity and Porosity. After this step, all wells (cored and non-cored) had porosity and permeability ready to be used in in property modelling after running an appropriate NNW. An example of NNW can be seen on Figure 5. This approach was proven to be a suitable one, when tackling this type of challenging reservoirs.

3- Results and Conclusion

The application of this methodology on the case study has significantly improved permeability and prediction and modelling. The HFU methodology was applied to estimate the appropriate number of rock types in the field, by using a combination of data analysis techniques (core analysis in scattered plots, histograms, and probabilities).

With careful application of NNW, porosity-permeability relationships seen in the cored wells were propagated for the entire field: this greatly improved the reliability of reservoir models, which in turn resulted in a more reliable production forecast. It also helped in daily operations, by assessing formation damage potential, and optimization of well completions.

The use of Mercury Injection Capillary Pressure (MICP) data quantifies pore distribution and capillary pressure functions and can be used to model saturation in the absence of SCAL data when caution is taken in the functions transformations.

The implementation of the above methodology resulted in the optimization of a phased field development plan, initially by relocating development wells into better reservoir areas, and for a future flank development, where data is scarce. This case study demonstrates the importance of integrating the geological knowledge into the definition of RRT with a future improvement towards Dynamic Rock Types. Furthermore, it also shows that a detailed core data analysis is essential to achieve a robust reservoir characterization and reliable reservoir models that together will lead to field development optimization and recovery maximisation.

4-References

Adolfo D. Winds [2007], Reservoir Zonation and Permeability Estimation: A Bayesian Approach. SPWLA 48th Annual Logging Symposium held in Austin, Texas, USA, June 3-6.

Genliang Guo, Marion A. Diaz, Francisco Paz, Joe Smalley and Eric A. Waninger, SPE, Occidental Exploration & Production Company [2007], Rock Typing as an Effective Tool for Permeability and Water-Saturation Modelling: A casa Study in a Clastic Reservoir in the Oriente Basin. SPE 97033.

Jude O. Amaefule and Mehmet Akltunbay, Core Laboratories: Djebbar Tiab, U.of Oklahoma; David G. Kersey and Dare k. Keelan, Core Laboratories [1993], Enhance Reservoir Description: Using Core and Log Data to Identify Hydraulic (Flow) Units and Predict Permeability in Uncored Intervals/Wells. SPE 26436.

O.D. Orodu, Z. Tang and Q. Fei [2009], Hydraulic (Flow) Unit Determination and Permeability Prediction: A Case Studt of Block Shen-95, Liaohe Oilfield, North-East China, and Journal of of Applied Sciences 9 (10): 1801-1816.

By: Ms. Ana de Sousa