Scholarlink

<p>Evaluating oil field exploration and management requires investigating numerous realizations and strategy scenarios. The computational cost for simulations limits decision-making processes, hindering the field´s contribution to the economy or environmental impact analysis. This study presents a novel approach for robust optimization of oil production to tackle the time/accuracy compromise between large geostatistical realization ensembles or a reduced number of representative models (RM). This work introduces the concept of evolutionary representativeness: initially exploring production schemes with a small ensemble and gradually expanding the evaluation of geological scenarios as the strategy demonstrates promise. The concept is integrated with a targeted optimization algorithm and tested in a real case heterogeneous reservoir from Brazilian pre-salt. After implementing our technique as an engine in commercial software, a candidate well optimization process was conducted to maximize economic factors such as Expected Field Economic Indicator (<i>EFEI</i>) and Expected Monetary Value (<i>EMV</i>). The method was benchmarked against a traditional approach utilizing nine RMs, achieving up to an 83% reduction in optimization time while maintaining similar optimized strategies. Ultimately, this optimization process led to a remarkable increase of 650 million USD in <i>EMV</i> compared to the base case. A gradient-descent based linearizing proxy was developed to mitigate deterministic biases from comparisons among divergent ensemble sizes. The proxy improved the correlation between estimated and calculated objective functions from 0.189 to 0.899, at the best case, thus aiding the optimization process. This work methodology helps achieving faster and more efficient decision-making concerning reservoir oil production management, sparing resources for deeper analyses or supplementary evaluation stages.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A concept for accelerating robust oil production strategy optimizations: evolutionary representativeness

  • 期刊论文

  • 发表日期：

    2025年11月20日

查看全文
  • Leandro H. Danes,
  • Guilherme D. Avansi,
  • Marx V Miranda,
  • Igor V. L. Silva,
  • Denis J. Schiozer

摘要

Evaluating oil field exploration and management requires investigating numerous realizations and strategy scenarios. The computational cost for simulations limits decision-making processes, hindering the field´s contribution to the economy or environmental impact analysis. This study presents a novel approach for robust optimization of oil production to tackle the time/accuracy compromise between large geostatistical realization ensembles or a reduced number of representative models (RM). This work introduces the concept of evolutionary representativeness: initially exploring production schemes with a small ensemble and gradually expanding the evaluation of geological scenarios as the strategy demonstrates promise. The concept is integrated with a targeted optimization algorithm and tested in a real case heterogeneous reservoir from Brazilian pre-salt. After implementing our technique as an engine in commercial software, a candidate well optimization process was conducted to maximize economic factors such as Expected Field Economic Indicator (EFEI) and Expected Monetary Value (EMV). The method was benchmarked against a traditional approach utilizing nine RMs, achieving up to an 83% reduction in optimization time while maintaining similar optimized strategies. Ultimately, this optimization process led to a remarkable increase of 650 million USD in EMV compared to the base case. A gradient-descent based linearizing proxy was developed to mitigate deterministic biases from comparisons among divergent ensemble sizes. The proxy improved the correlation between estimated and calculated objective functions from 0.189 to 0.899, at the best case, thus aiding the optimization process. This work methodology helps achieving faster and more efficient decision-making concerning reservoir oil production management, sparing resources for deeper analyses or supplementary evaluation stages.