1887

Abstract

Summary

This study explores the potential of agent-based frameworks powered by Large Language Models (LLMs) to integrate and interpret multimodal data in well drilling operations. The research leverages autonomous agents to process diverse datasets, including images and mechanical parameters, to assess wellbore stability and drilling efficiency. Using specialized agents, tasks such as caving detection from shale shaker images and the analysis of Mechanical Specific Energy (MSE) and Drilling Strength (DS) are performed.

Two language models—Phi-3 Mini and GPT-3.5—are compared to evaluate the trade-offs between performance and cost-effectiveness. The Phi-3 Mini model, despite being cost-efficient, exhibited occasional hallucinations, whereas GPT-3.5 consistently provided more precise and reliable outputs. Fine-tuned models like Llava were used for image-based analyses, enhancing multimodal integration.

The agent-based framework demonstrates its capability to autonomously reason, plan, and interact with external tools, producing actionable insights to guide operational decisions. Results indicate that the system effectively identifies risks, such as significant cavings and unusual MSE/DS ratios, offering recommendations to mitigate potential challenges. The proposed use of agent-based framework highlights the scalability and flexibility of LLM-powered agent systems, paving the way for advanced applications in the energy sector while addressing cost considerations through smaller, efficient models.

© EAGE Publications BV
Loading

Article metrics loading...

/content/papers/10.3997/2214-4609.2025101389
2025-06-02
2026-02-17

  • From This Site

    /content/papers/10.3997/2214-4609.2025101389
    dcterms_title,dcterms_subject,pub_keyword
    -contentType:Journal -contentType:Contributor -contentType:Concept -contentType:Institution
    10
    5
Loading full text...

Full text loading...

References

  1. Abdin, M., Aneja, J., Awadalla, H., Awadallah, A., Awan, A.A., Bach, N., Bahree, A., Bakhtiari, A., Bao, J., Behl, H. et al. [2024] Phi-3 Technical Report: A highly capable language model locally on your phone. arXiv, 2404.14219. https://arxiv.org/abs/2404.14219.
     [Google Scholar]
  2. Hadid, A., Chakraborty, T. and Busby, D. [2024] When geoscience meets generative AI and large language models: Foundations, trends, and future challenges. arXiv, 2402.03349.
     [Google Scholar]
  3. Kumar, D., Ansari, S., Wang, S., YiMing, J., Ahmed, S., Povstyanova, M. and Tichelaar, B. [2012] PS real-time wellbore stability analysis: An observation from cavings at shale shakers.
     [Google Scholar]
  4. Liu, H., Li, C., Wu, Q. and Lee, Y.J. [2023] Visual instruction tuning. Presented at NeurIPS.
     [Google Scholar]
  5. Menand, S. and Mills, K. [2017] Use of mechanical specific energy calculation in real-time to better detect vibrations and bit wear while drilling. AADE National Technical Conference and Exhibition, AADE-17-NTCE-033, Hilton Houston North Hotel, Houston, Texas, April 11–12, 2017.
     [Google Scholar]
  6. Nmegbu, C.J. and Ohazuruike, L. [2014] Wellbore instability in oil well drilling: A review. International Journal of Engineering Research and Development, 10(5), 11–20.
     [Google Scholar]
  7. Wang, L., Ma, C., Feng, X.et al. [2024] A survey on large language model-based autonomous agents. Frontiers of Computer Science, 18, 186345. https://doi.org/10.1007/s11704-024-40231-1.
     [Google Scholar]
  8. Zhang, J., Clairmont, C., Que, X., Li, W., Chen, W., Li, C. and Ma, X. [2025] Streamlining geoscience data analysis with an LLM-driven workflow. Applied Computing and Geosciences, 25, 100218. https://doi.org/10.1016/j.acags.2024.100218.
     [Google Scholar]
/content/papers/10.3997/2214-4609.2025101389
Loading
Leveraging Agent-Based Frameworks and LLMs for Multimodal Analysis in Drilling and Geological Operations
Conference Proceedings 2025, 1 (2025); https://doi.org/10.3997/2214-4609.2025101389
/content/papers/10.3997/2214-4609.2025101389
/content/papers/10.3997/2214-4609.2025101389
Loading

Data & Media loading...

Most Cited This Month Most Cited RSS feed

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error