A Local Computerized Multi-Screening of Vast Amount of Data to Select Hydraulic Fracturing Candidates in Iranian Carbonate Oil Fields

Abolfazl Hashemi, Seyed Reza Shadizadeh, Mansoor ZoveidavianpoorInternational Journal of Computer Applications 2012 · DOI

Abstract

Iranian oil companies are increasingly turning to hydraulic fracturing (HF) to enhance hydrocarbon recovery from deep carbonate formations. However, the lack of a standardized computerized framework for candidate selection in these complex reservoirs has led to inconsistent results. This paper presents a locally developed MATLAB-based program designed to automate the selection of specific zones for HF. The tool integrates vast amounts of multidisciplinary data and utilizes Artificial Intelligence (AI) techniques, such as Neural Networks and Fuzzy Logic, to compensate for missing geomechanical data. Candidates are screened based on user-defined parameters, cut-offs, and weight factors, with results prioritized in visual stacked bars. The tool was successfully applied to the "M" oil field in southern Iran, analyzing 585 zones across 42 wells to identify optimal stimulation targets.

Introduction

Hydraulic fracturing is a critical technology for revitalizing mature oil and gas fields. In Iran, where recovery rates in carbonate reservoirs often stagnate between 20% and 25%, HF offers a path to significant production increases. Despite its potential, historical HF operations in the region have frequently underperformed. The primary cause of these failures is often poor candidate selection—the "frontline" of any successful stimulation project.

Engineers face two significant challenges when selecting candidates in large Iranian fields: data overload and data gaps. A single reservoir can contain hundreds of wells and thousands of potential zones, each described by dozens of parameters. Furthermore, essential geomechanical data, such as Dipole Sonic Information (DSI) logs required for stress profiling, are often missing due to high acquisition costs. This research addresses these hurdles by providing a computerized decision-support tool that bridges data gaps with AI and streamlines the selection process through multi-criteria screening.

Methodology

The developed tool follows a rigorous three-step workflow to transform raw field data into actionable insights.

1. AI-Driven Data Completion

To address the scarcity of DSI logs, the program employs soft computing techniques. Using MATLAB’s Fuzzy Logic and Artificial Neural Network toolboxes (specifically genfis2, anfis, and newfit), the system is trained on "key wells" where complete data exists. It then predicts critical rock mechanics properties, such as Poisson’s ratio and Young’s modulus, for the remaining wells in the field.

2. Stress Profile Determination

Accurate in-situ stress profiles are vital for predicting fracture geometry. The tool calculates these profiles using various correlations. For the Iranian carbonate formations studied, the Eaton correlation was identified as the most reliable, showing the best alignment with actual field leak-off tests.

3. Mechanical Screening and Prioritization

The core of the tool is a Graphical User Interface (GUI) that allows engineers to input specific criteria, such as net thickness, oil saturation, and stress contrast. Users define "cut-off" values and assign weight factors to each parameter based on their relative importance. The system then performs a dual screening of both wells and individual zones.

Figure 1: Flowchart of the computerized candidate selection methodology, integrating zone and well data through AI and mechanical screening.

Case Study: The "M" Oilfield

The methodology was validated using data from the "M" Oilfield in Southern Iran, focusing on the Bangestan reservoir. The dataset included 42 wells, each containing 13 distinct zones, resulting in a total of 585 potential candidates. Over 30 parameters per zone—ranging from petrophysical logs and production history to geomechanical properties—were processed by the tool.

Key Insights and Results

A critical finding of the study is the importance of stress contrast. For effective fracture containment, the difference in stress between the target zone and its bounding layers is more significant than the absolute stress value. A target zone with relatively lower stress "sandwiched" between high-stress barriers is ideal for ensuring the fracture remains within the pay zone.

Sensitivity analysis conducted using FracPORO software revealed that Zone Stress, Zone Thickness, and Young’s Modulus are the most influential parameters affecting fracture half-length. By applying these insights through the screening tool, the researchers were able to rank the 585 zones effectively.

Figure 2: The final ranking output of the tool, showing top candidates for Well and Zone selection using prioritized stacked bars.

In the "M" field case study, the tool identified Zone 6 in Well 6 as the primary candidate for hydraulic fracturing. The visual output (Figure 2) uses stacked bars where the length of each colored segment represents how well a candidate meets a specific weighted criterion (e.g., Net Thickness, Oil Saturation, or Stress Contrast). This transparency allows decision-makers to understand exactly why a particular zone was ranked highly.

Conclusion

The MATLAB-based multi-screening tool provides a robust, data-driven framework for HF candidate selection in complex carbonate reservoirs. By integrating multidisciplinary data and leveraging AI to fill information gaps, it reduces the technical risks associated with stimulation. While the tool offers a trustworthy preliminary selection, the authors emphasize that a physical field calibration test (minifrac) remains necessary to finalize the HF design. Future work involves expanding this framework to other major Iranian fields to establish a comprehensive regional stimulation strategy.

Summary updated

Abstract

Iranian oil companies are developing the technique of Hydraulic Fracturing (HF) operation to enhance the hydrocarbon recovery of deep carbonate formations. However, there is not a computerized tool or well defined framework for Iranian carbonate oil fields to select candidates. The ineffective HF experiences in the past emphasized that candidate selection is the frontline of a victorious HF operation. This paper presents the development of a local programme to automatically select specific zones for special purposes like HF. The program is written in MATLAB in such a way to integrate large amount of data from different disciplines. In addition, the missing data are compensated with Neural Network and Fuzzy Logic techniques. In the end data are mechanically screened based on the user selected parameters, cut-offs and weight factors. Results of screening within the limitations are prioritized in stacked bars to make decision easier. This tool is applied for a purpose of candidate selection for HF in M oil field located in south of Iran. This field has 585 zones which each zone has more than 30 parameters form different disciplines. The result of this programming is printed schematically and it is conclusive to our clients.
BibTeX
@article{hashemi2012local,
  author = {Abolfazl Hashemi and Seyed Reza Shadizadeh and Mansoor Zoveidavianpoor},
  title = {{A Local Computerized Multi-Screening of Vast Amount of Data to Select Hydraulic Fracturing Candidates in Iranian Carbonate Oil Fields}},
  journal = {International Journal of Computer Applications},
  year = {2012},
  doi = {10.5120/4842-7106},
}
Figure