Fluid Sampling in Tight Unconventionals

Abstract

In this paper we emphasize the duality of fluid sampling: (1) fluid characterization; to collect samples and measure pressure/volume/temperature (PVT) data that can be used to build and tune an equation of state (EOS) model, and (2) fluid initialization; to collect samples to estimate in-situ fluid compositions. It is hard, if not impossible, to obtain truly in-situ representative fluid samples in multi-fractured horizontal wells (MFHW). This paper explains why fluids measured in the lab may be significantly different from in-situ representative fluid samples, even if the fluid samples are collected shortly after the well is put online. The paper also suggests that practically all samples, in-situ representative or not, can and should be used to build a reliable EOS model. To make a comprehensive assessment of fluid sampling in tight unconventionals, reservoir fluids ranging from black oils to gas condensates have been studied. For a wide range of fluid systems, a compositional reservoir simulator has been used to assess two main scenarios: (1) an initially undersaturated (single-phase) fluid system, and (2) initially saturated (two-phase) fluid system. To quantify how collected surface samples change with time, three properties are studied as functions of time: (1) saturation pressure and type (dewpoint | bubblepoint), (2) producing gas/oil ratio (GOR), and (3) stock-tank oil (STO) API. Observations of how these three properties change with time is used to help explain why elevated saturation pressures, greater than the initial reservoir pressure, often can be observed. Rapid decline of the flowing bottomhole pressure (BHP | pwf), together with shut-in periods, makes it difficult to obtain in-situ representative samples in MFHW. For slightly undersaturated reservoirs, and saturated reservoirs, it may be impossible to obtain in-situ representative fluid samples because of the near-wellbore multiphase behavior. However, samples which are not in-situ representative can still be used to estimate original in-situ fluids using equilibrium contact mixing (ECM) procedures. In this paper, we propose two ECM methods that can either be carried out by physical measurements in a PVT lab or can be computed with a properly tuned EOS model.