The potential use of NMR as a direct indicator of hydrocarbon saturation via techniques such as the Differential Spectrum Method (DSM) has generated significant interest in the petrophysical community in recent years. Although originally developed for applications involving natural gas, the DSM has also been used successfully in light hydrocarbon environments. However, success has been limited to the low end of the viscosity spectrum because of the T1 separation requirements between the brine and hydrocarbon phases. The T1 separation requirement in higher viscosity applications can be eliminated by using the Enhanced Diffusion Method (EDM), where diffusion is turned into the dominant relaxation mode for the wetting brine phase. Given that brine is more diffusive than the hydrocarbons, the longest apparent T2 from the brine phase can be made short enough to cause separation between the two phases in T2 space, thereby eliminating the need for T1 separation. Wait time manipulation can then be used to quantity hydrocarbon volumes when the two phases are separated in T2 domain. This paper focuses on the determination of residual oil saturation using EDM, while also providing guidelines for job screening and acquisition parameter selection. Several case histories provided are used to illustrate basic concepts and different methodologies available. P. 267

Emulsifiers are common to invert oil mud systems since they are an aid to drilling mud stability. When filtrate from such muds invade, they are likely to include some of these emulsifiers and result in altering the rocks wetting characteristics. If so, the NMR response will change since the relaxation rate of an oil's hydrogen protons, when contacting the rock, is typically much slower than the hydrogen protons of capillary bound water. Therefore, as the wetting fluid changes from water to oil, the relaxation rate, being measured by a MRI logging tools, changes. The consequence of this to the NMR interpreter, being unaware that a change in wetting has occurred, is that they are likely to under-estimate the irreducible (BVI) water volume, and, when using a calculation of permeability dependent upon the FFI/BVI ratio, will overcall permeability. When an under called BVI scenario was identified from the comparison of core-to-log permeability's in a series of wells drilled with invert emulsion mud's, it was speculated that the rocks wetness had been altered. To investigate, core samples were obtained from one of the suspect wells for a laboratory evaluation of the reservoir's NMR properties. A detailed laboratory protocol, focused on the influence of invert emulsion filtrates, was then developed based on the use of rock samples form the troubled area in order to see if, and how, the mud system alters the NMR characteristics. A clear demonstration of wetting alteration was found, and, the observed effect on BVI, and permeability, was confirmed, closely replicating the MRI BVI observations as given by the MRI log thus offering one explanation for the observed log behavior. Presented here are the procedures, methods and findings, as well as, a new process for determining reliable BVI and permeability values when MRI logs are run in invert oil mud systems. Introduction Several MRI logs performed in the North Sea area were called into question when core measurements showed the formations permeability to be generally of poorer quality than predicted by MRI log results. Figure 1 is an example log from the area. It shows that MR permeability (MPERM) is generally higher than the core's, meaning the MRI log's FFI/BVI ratio is too low. A comparison of core BVI to MBVI was made as a point of reference. CoreBVI* values were computed using the correlation shown in figure 2. This was accomplished using the samples from the troubled area. They were measured for air permeability (Ka) and desaturated to irreducible water saturation (Swi) using an air/brine displacement (100 psi). CoreBVI* values were then calculated from, (1) Comparing MBVI to the computed CoreBVI* tends to verity that, for a water wet rock, BVI is underestimated and a likely cause of the overestimation of MPERM. A closer inspection of fig. 1 shows MPERM to correctly reflect the relative amplitudes, but the magnitude is too high. The under-called BVI problem was conjecturally linked to these wells since they had all been drilled with invert oil mud systems. And other wells in the area, drilled with water base muds, did not appear to have the same difficulties. Potential Reasons for Anomalous Results There are three potential reasons for this set of anomalous responses. First, the relaxation time used to separate bound from free fluid may be incorrect for these formations. P. 203

No abstract

The presence of viscous oil in a reservoir greatly complicates the interpretation of NMR log data. Because the NMR signal from an oil with a viscosity greater than 1,000 cp in the sub-surface typically decays with a T2 time-constant that is comparable to that of capillary-bound or clay-bound water, it is impossible to distinguish the signal from the oil phase from that of the bound-water phases. Various petrophysical quantities, such as permeability and fluid volumes, that are normally derived relatively directly from NMR measurements, thus require a significant additional interpretation effort if they are to be determined in reservoirs containing viscous oils. Work published by other authors shows that signal (porosity) loss in viscous oils is a predictable function of the viscosity and the interecho spacing used in the NMR CPMG acquisition sequence. Therefore, a reasonable estimate of viscosity can be obtained by combining the NMR logs with conventional logs to estimate the NMR signal loss at a specific interecho spacing. When this method is combined with NMR diffusion measurements, the volume of movable water can be estimated. Further combinations of conventional and NMR log data provide the quantity of capillary-bound water and a good estimation of permeability. Log examples are available from areas where viscous oils are problematic to users of both conventional and NMR data. These examples are presented to introduce and demonstrate these new methods. Introduction Heavy oil and bitumen account for roughly 6 trillion barrels of the world's known oil reserves, the vast majority of which is found in Venezuela, Canada, and the Former Soviet Union.1 Most of the heavy oil produced in the United States comes from fields in California, Wyoming, Utah, Texas, Kentucky, and Mississippi. The term heavy oil in the context of this paper refers to oils having API gravity lower than 20° and viscosity at formation conditions above 100 cp. These hydrocarbons generally pose challenging production problems and are marketed at a discounted price, however, improved technology and higher oil prices have combined to make exploitation of heavy oil deposits more economically attractive. In addition to the multitude of production and processing problems associated with heavy oils, evaluation of petro-physical properties from NMR logging measurements can be complicated for a number of reasons. Heavy oil NMR responses are similar to signals from capillary-bound water. Furthermore, heavy-oil chemistry may be conducive to a wettability alteration2 that may contribute to a misinterpretation of water content from conventional and NMR logs. These phenomena make it difficult to quantify fluid volumes in heavy-oil reservoirs from NMR measurements alone. Present-day NMR logging instruments do not fully capture heavy-oil signals because they operate at interecho spacings that make them unable to adequately sample important rapid-decay components when viscosity exceeds ~ 1000 cp. This situation causes the indicated NMR porosity to be too small, as though the reservoir fluid had a hydrogen index (HI) smaller than one. LaTorraca, et al., have shown how the NMR signal loss in these situations is related to oil viscosity and interecho spacing.3 These factors make it necessary to apply additional interpretation methods to obtain indications of altered wettability and evaluate petrophysical quantities such as fluid volumes, permeability, and apparent in-situ oil viscosity. The methods outlined in this paper rely on combinations of NMR and conventional wireline logs to determine the signal loss and estimate the viscosity of oils whose in-situ viscosity is larger than a few hundred centipoise. Additional combinations with conventional logs can be formed with NMR diffusion measurements to infer movable and capillary-bound water volumes which can be used to refine interpretations of resistivity logs, indicate altered wettability, and provide an improved estimate of permeability in heavy-oil reservoirs.

No abstract