Development of a surrogate model for elemental analysis using a natural gamma ray spectroscopy tool.

Abstract A systematic computational method for obtaining accurate elemental standards efficiently for varying borehole conditions was developed based on Monte Carlo simulations, surrogate modeling, and data assimilation. Elemental standards are essential for spectral unfolding in formation evaluation applications commonly used for nuclear well logging tools. Typically, elemental standards are obtained by standardized measurements, but these experiments are expensive and lack the flexibility to address different logging conditions. In contrast, computer-based Monte Carlo simulations provide an accurate and more flexible approach to obtaining elemental standards for formation evaluation. The presented computational method recognizes that in contrast to typical neutron–photon simulations, where the source is typically artificial and well characterized ( Galford, 2009) , an accurate knowledge of the source is essential for matching the obtained Monte Carlo elemental standards with their experimental counterparts. Therefore, source distributions are adjusted to minimize the L2 difference of the Monte Carlo computed and experimental standards. Subsequently, an accurate surrogate model is developed accounting for different casing and cement thicknesses, and tool positions within the borehole. The adjusted source distributions are then utilized to generate and validate spectra for varying borehole conditions: tool position, casing and cement thickness. The effect of these conditions on the spectra are investigated and discussed in this work. Given that Monte Carlo modeling provides much lower cost and more flexibility, employing Monte Carlo could enhance the processing of nuclear tool logging data computed standards.