Abstract Numerical reservoir modelling is the most reliable geothermal resource assessment tool for quantifying power production at a sustainable level, but it does not mesh well with the Monte Carlo method for quantifying uncertainties. Building proxy numerical models using Experimental Design (ED) and Response Surface Methodology (RSM) workflows may provide a solution. The present study attempts to improve existing approaches through the use of the two-level and three-level Full Factorial and Box-Behnken designs. The improved workflow yields consistent results for both the Ohaaki and Rotorua geothermal fields suggesting that probabilistic MWe output distribution is between P90 to P10 of the widely used volumetric method.

中文翻译：

---

使用代理数值模型进行改进的实验设计和响应面方法工作流进行概率地热资源评估

摘要 数值储层建模是量化可持续水平发电量的最可靠的地热资源评估工具，但它与量化不确定性的蒙特卡罗方法不太吻合。使用实验设计 (ED) 和响应面方法学 (RSM) 工作流程构建代理数值模型可能会提供解决方案。本研究试图通过使用两级和三级全因子和 Box-Behnken 设计来改进现有方法。改进的工作流程为 Ohaaki 和 Rotorua 地热田产生一致的结果，表明概率 MWe 输出分布在广泛使用的体积方法的 P90 到 P10 之间。