Prototype Inversion of Multi-Probe Chemical Sensing Data to Estimate Inter-Well Distributions

Abstract

Abstract For several years, there has been an interest in responsive "NanoProbes," which, when injected along with waterflooding could sense reservoir properties locally along the trajectories they follow from injector to producer wells, giving a low-cost and very deep formation evaluation upon being collected, evaluated, and interpreted with respect to injection point, arrival point, and timings. Here, we introduce these novel "dual-mode" NanoProbe tracers, which can undergo chemical transformations when encountering target analytes within the reservoirs. We first built the dual-mode chemical sensing tracer functionality into our reservoir simulator and performed forward simulations to acquire model transformed and untransformed tracer breakthroughs. Specifically, the original tracer chemical (denoted tracer-1) can transform into a different chemical (denoted tracer-2) when encountering specific analytes of interest within the reservoir; and the ratio of tracer-1 and tracer-2 from injector-producer pairs provides information about the inter-well analyte distributions. Furthermore, we developed a history matching algorithm based on the iterative ensemble smoother with a rectifier linear unit transformation (ES-MDA-ReLU) that can successfully interpret the inter-well analyte distributions from the chemical sensing tracer data. We found that traditional ES-MDA algorithm is ineffective for the history matching of the inter-well analyte distributions form the chemical sensing tracer data if the inter-well analyte distributions are discrete; nevertheless, applying a ReLU filter to the analyte distributions combining with ES-MDA algorithm results in greatly improved history matching results. We also studied the spatial and temporal resolution of the inter-well analyte distributions inverted from the barcoded chemical sensing tracer data, whereby we found that the spatial resolution is sensitive to well spacing as well as the tracer travel paths; and the temporal resolution is sensitive to the shapes of the tracer breakthrough curves (notably, good history matching can already be achieved if the early parts of the breakthrough curves are collected from all producers). Finally, we compared the application of chemical sensing tracers on synthetic reservoir models with homogeneous or heterogeneous permeability fields and found that better history matching can be achieved on heterogeneous fields due to the more diverse travel paths of the chemical sensing tracers. Even though the responsive NanoProbes concept has been found promising, the details of the NanoProbes’ working principles and data processing have yet to be fully developed. We believe this work will bridge these gaps and begin to demonstrate the NanoProbes’ potential as novel formation evaluation tools with direct-sensing, low-cost, and very deep reservoir characterization capabilities.