Summary
We report a case study of using nuclear magnetic resonance (NMR) multiple-te, dual wait-time (tw) log acquisitions for quantitative characterization of San Jorge Basin reservoir oil viscosity. Previously, dual-tw logs have been used to discern gas and oil from water, while dual-te logs have been used as a qualitative light oil indicator. Although theoretically simple, quantitative determination of viscosity from dual-te logs is complicated by several factors, including poor signal-to-noise ratio, difficulties in separating oil from water, and the uncertainty of internal gradient strength. In the present study, multiple-te acquisitions of dual-tq logs were used to isolate the oil from the water signal. The values of viscosity of the reservoir fluids can be estimated from either intrinsic T2 or T1. In estimation of the apparent T2, we used a model that does not explicitly require knowledge of the internal gradient, thereby minimizing the effects arising from the uncertainty of the internal and tool gradient strengths. Because T1 and intrinsic T2 are estimated independently, the degree of agreement between the two values provides an indication of the reliability of the two estimates. The main example in the study of four pay zones was thought to contain viscous oil. However, our analysis indicated that the viscosity values of the oil are less than 5 cP. The predictions have been substantiated by production of light hydrocarbons from the three zones that have been perforated. Further, a good agreement is obtained for the viscosity estimates based on NMR log data and laboratory pressure/volume/temperature (PVT) analysis.
Introduction
Hydrocarbon viscosity is an important reservoir fluid parameter that significantly affects oil recovery and economics. Fluid flow is inversely proportional to viscosity and the higher the viscosity the lower the flow rate and the slower the recovery. Further, when two or more fluids are flowing, the ratio of the viscosities, the mobility ratio, is one of the key parameters that affects sweep efficiency and ultimate recovery. In many reservoirs, it is uneconomical to produce heavy, viscous oil, and thus it is crucial to determine oil viscosity before completing the well. The problem is even more pressing when oil viscosities vary within a hydrocarbon column or from zone to zone when attempting to commingle multiple zones.
Many laboratory procedures can determine viscosity. Samples for viscosity determination may be obtained from reservoir fluid samples recovered from well tests or drill stem tests, downhole fluid samplers, or reconstituted from separator samples. Sampling procedures are generally limited to a few depths; and, of course, samples reconstituted from separators are associated with the entire producing interval and may not be associated with a single depth. Further, there is always the concern that the fluid samples may not be representative of the in-situ reservoir fluids. Nuclear magnetic resonance (NMR) logging measurements have the potential to provide in-situ viscosity measurement because the NMR relaxation times, T1 and T2, correlate strongly with fluid viscosity. The difficulty with NMR-determined viscosity is that the measurement is relatively shallow. Thus, the hydrocarbon saturations may be significantly reduced (So or Sor), and hence, the sought NMR oil signal is small. Further, the interpretation may be complicated by NMR signals originating from the invading fluids. NMR response is controlled by both rock and fluid properties. In fact, NMR log interpretation is complicated because it is not always clear whether the T2 decay reflects hydrocarbon and/or rock properties. However, we are fortunate because two NMR experiment parameters, te and tw (Fig. 1), can be used to tailor the NMR data acquisition to separate the hydrocarbon and rock property effects.
One of the first specialized NMR applications that took advantage of the ability to control NMR log acquisition parameters was hydrocarbon typing.1 NMR hydrocarbon typing in porous media relies on the difference in NMR response in either or both of the relaxation times (T1 and T2) or diffusivity of oil, water, and gas. By carefully designing the logging program and using combinations of pulse sequences, one can enhance the relaxation and diffusivity contrasts between the different fluid phases. The two commonly used approaches for magnetic resonance image log (MRIL®)** based hydrocarbon typing are dual-tq logging1,2 and dual-te logging.3,4
Dual-tw logging utilizes the T1 contrast between nonwetting light hydrocarbons and the wetting water for quantitative light hydrocarbon typing, while dual-te logging utilizes the viscosity (and thus diffusivity) contrast between reservoir fluids. The latter, to our knowledge, previously has been used mainly as a qualitative, or, at best semiquantitative, hydrocarbon indicator. Quantitative estimation of oil viscosity and saturation require solving problems related to uncertainty of internal magnetic-field gradient and separating oil from water signals.
We combined the dual-tw and dual-te approaches to maximize the advantages of both T1 and T2 contrasts. With multiple-te passes of dual-tw logs, we are able to eliminate a majority of the water signal from dual-tw logs; the remaining signal is predominantly an oil signal. The characterization of oil viscosity is achieved by analyzing the relaxation times and diffusion effect on the isolated oil signal with the multiple-te data acquisition. This reduces uncertainties due to the interfering water signal originating from either the irreducible and bound water or from the invading mud filtrate present on conventional dual-te logs.