Introduction

Traditionally, resistivity and nuclear logs are used to estimate porosity, F and water saturation S_w prior to WFT operations.  However, what are missing are continuous hydrocarbon type and properties logs. For example, it is impossible to detect compositionally graded oils from F, S_w, and k logs and therefore insufficient sampling might result. On the other hand, excessive pretests and rig time might have been spent to acquire fluid information in difficult environments such as thin beds, washouts, near wellbore alterations, tight formations etc. where it may be more efficient to utilize NMR data.

Hydrocarbon type is usually inferred from well logs or from prior field knowledge. Sometimes, large density-neutron separation allows us to distinguish gas from oil but, in other cases, lithological effects could mask it.  Hydrocarbon type can also be interpreted from pressure gradient plots. However, a reliable pressure gradient needs sufficient and well-spaced points that might be difficult in thinly laminated beds, or where there is a thin gas cap above the oil zone or a thin oil ring below the gas zone.  Also, Jackson et al. (2007) have shown that the technique might not be reliable in case of a compositional gradient and or compartmentalized reservoirs.

It is known from laboratory measurements that NMR can estimate oil viscosity (Kleinberg and Vinegar, 1996) and GOR (Lo et al., 2000). In oilfield applications, multi-dimensional NMR measurements are used to investigate fluid type and properties via Diffusion-Relaxation (D-T) maps (Freedman et al. 2001, Cao Minh et al. 2003, Heaton et al. 2004). However, applications to well log data pose the challenge of how to separate OBMF from native oil – since NMR will see both oils in the flushed zone

Identification of Oil from OBMF

Theoretically, both OBM filtrate and native oil are stable chemical compounds that are in thermodynamic equilibrium.  In a closed thermodynamic system, an external work must be exerted to perturb the equilibrium and change the state of the fluids (i.e. for the fluids to mix). One can argue whether this can happen during the downhole spurt invasion process. Whether the fluids mix or not, OBMF might be distinguished from oil using NMR radial profiling where the evolution of the two fluids with invasion distance can often be seen.

Most OBM filtrates encountered in deepwater West Africa have a T2 between 500 ms and 1500 ms at downhole conditions. We use the equation:  (where T2 is in seconds, μ is viscosity in centipoise and Tc is the temperature in degrees Celsius) to compute a viscosity of about 1 cp. When the native oil T2 is outside that range one can easily distinguishes the two oils at any DOI as seen in Figure 1 .

Downhole Fluid Analysis with WFT Tools

Cao Minh et al 2008 give a good summary of using spectroscopic based measurements for hydrocarbon differentiations.  Additionally, a new sensor recently introduced for WFT tools is a vibrating rod density-viscosity sensor. (O’Keefe et al. 2007)  This sensor measures the density of the fluid by the vibration of a mechanical resonator in the flowline. The resonant frequency of the vibrating element decreases as the fluid density increases and the quality factor will decreases as viscosity increases. Sensor characterization is performed using standard reference fluids to cover a wide range of viscosities. The outputs of the sensor are a fluid density in g/cc and a fluid viscosity in cp. Both of these measurements have quality flags associated with them that are driven by  how well the output of the sensor fits the response model. Field experiences indicate that the density measurement responds well to a variety of fluid types but the viscosity measurement works best in a single phase environment, i.e. in a well drilled with OBM sampling oil or in a well drilled with WBM sampling water. The presence of oil and water emulsion has proven problematic for the viscosity output. The DV sensor example that we discuss here is acquired sampling oil in a well drilled with OBM and therefore, the viscosity answer proves to be quite reliable.

Example 1: Reservoir compartmentalization

The first example is shown in Figure 2. The gamma ray and resistivity curves in tracks 1 and track 2 show several hydrocarbon-bearing and water-bearing sands. Track 3 and track 4 show viscosity and GOR from NMR (black) and from the WFT DV sensor (green) respectively. Although the NMR viscosity and GOR curves are continuous, their computation is blanked out over the intervals where the hydrocarbon volume is less than the noise level; typically about 1 pu. The three tracks to the right show the NMR fluids volume at 1.5, 2.7 and 4 inches DOI respectively. Combining WFT and NMR results in one picture leads to several observations:

1.      The large viscosity, GOR variations imply that the sands have different oils and therefore, compartmentalization is possible.

2.      The oils appear to divide into three general types:

a.       Darker oils above ~950 m with viscosities in the 20 cp or higher range and GOR ~90 m3/m3.

b.      Slightly lighter oils from ~950 m to 1200 m with viscosities in the 5 cp range and GOR ~100 m3/m3.

c.       Lighter oils below ~1200 m with viscosities in the 1 cp range and GOR ~150 m3/m3.

3.      The thin-bedded sands above 750 m are oil-bearing with the same viscosity ~20 cp as the oil in the thick sand below at 750 m. The top of the oil column is at 700 m. It would be difficult to determine the hydrocarbon type and properties in the thin-bedded section without a priori knowledge.

4.      The thin-bedded sands below 1300 m are oil-bearing with the same viscosity ~1 cp and GOR ~150 m3/m3 as the oil measured by WFT below at 1480 m and 1590 m.

5.      The viscosity and GOR profiles imply at least 3 distinct hydrocarbon-charging phases have occurred in the reservoirs.

In the case of this example considerable operational flexibility was realized. The viscosity mapping provided by the NMR measurements was able to guide the MDT sampling operations. Additionally the viscosity contrast between the upper and lower zones implied a significant economic consequence and early recognition of this was critical for optimizing the subsequent DST evaluation.

Example 2: Hydrocarbon ID in Tight Formations

Figure 3 shows in track 1 well-defined pressure gradients in the water and oil zones. Additionally, an OWC determined by the intersection of the oil and water gradients fits nicely where one would pick an OWC from the resistivity log shown in Track 4. However, above ~575 m TVD reservoir quality deteriorates.  The gamma ray in track 3 is seen to increase and the neutron and density curves in track 5 are seen to separate. As a result, obtaining WFT pretest data in this section proved difficult. As the plot shows in track 1, three dry tests were obtained and two supercharged pressures were measured.  The data could not be used to extend the gradient, and fluid typing from the pretest data alone in the upper part of the reservoir is not possible.

An NMR log was run in this well and the results are presented in the D-T2 maps to the right in Figure 3. The WFT results are corroborated by NMR results for the water and oil zones. In the lower two D-T2 maps, where the fluid type is known, it is possible to qualify and interpret the oil signal from the OBMF filtrate. We then extend the interpretation to the upper part of the reservoir, where pretest acquisition was not possible and fluid typing from neutron-density is difficult due to a change in lithology. The NMR map, however, clearly shows that the reservoir fluid is gas.

Example 3: Heavy oil

Figure 4 shows an example of a heavy oil application. The high viscosity oils can be seen from the short relaxation time highlighted by the yellow ellipses in T2 time (track 1), T1 time (track 2), and an absence of diffusion in track 3. The diffusion log in track 3 shows multiple OWCs. Since the heavy oil components overlap with the bound water components we use a special technique (Cao Minh et al. 2006) to compute the oil volume (green area in track 4), viscosity (black curve in track 5) and GOR (cyan curve in track 6) assuming that the water saturation estimated from the deep resistivity is at irreducible level in the heavy oil zones. Two WFT oil samples were recovered. At 213 m, the measured viscosity is 92 cp. At 221 m, the measured viscosity is 13 cp. These are plotted in track 5 as green dots.

Conclusions

We have shown that modern NMR logs and WFT go hand in hand to provide critical reservoir fluids information. The NMR multiple DOI and continuous logs can be used to assist WFT for maximum efficiency. NMR relies on models to derive reservoir properties such as irreducible water saturation and permeability, and correlations to derive fluids properties such as viscosity and GOR. As such it can not replace WFT. Its role is to help WFT in unknown or difficult situations that can result in unnecessary tests or insufficient tests. If left unanswered, fluid questions can lead to costly well testing.

NMR is best run before WFT to determine the most suitable points for pretesting and sampling and as well the points to avoid. It also gives a look-ahead picture of the degree of complexity of the fluid column. Knowing in advance if the fluid column is generally homogenous or heterogeneous can ensure that the fluids are neither over-sampled nor under-sampled. When run after WFT, the continuous logs of permeability, viscosity and GOR can be calibrated with WFT data to provide fluid information where the tool did not sample. Inflow curves can then be built to predict reservoir performance.

We conclude that the addition of the continuous NMR fluid properties log to WFT sampling and Downhole Fluid Analysis adds significant value in terms of both the efficiency of the operations and the quality and completeness of the acquired data.

Acknowledgements

The authors wish to express appreciation to Total, ENI and Schlumberger for permission to publish this paper. We would like to also thank the many anonymous reviewers.

Selected References

Cao Minh, C. et al., 2003, Planning and Interpreting NMR Fluid-Characterization Logs:

SPE 84478, 2003 SPE Annual Technical Conference, Denver, Colorado, USA.

Cao Minh, C. et al., 2006, Evaluation of Congo Heavy Oil Reservoir with Novel NMR

logging: Abstract from 2006 SPWLA Annual Logging Symposium, Veracruz, Mexico.

Cao Minh, C. et al, 2008, Using the Continuous NMR Fluid Properties Scan to Optimize Sampling with Wireline Formation Testers: SPE 115822, 2008 SPE ATCE, Denver CO, USA.

Castelijns, K. et al., 1999, Combining NMR and Formation Tester Data for Optimum

Hydrocarbon Typing, Permeability and Producibility Estimation: 1999 SPWLA Annual

Logging Symposium, Oslo, Norway.

Freedman, R. et al., 2001, A New NMR Method of Fluid Characterization in Reservoir

Rocks: Experimental Confirmation and Simulation Results: SPE Journal December 2001.

Heaton, N. J. et al., 2004, Saturation and Viscosity from Multidimensional Nuclear

Magnetic Resonance Logging: SPE 90564, presented at the 2004 SPE Annual Technical

Conference, Houston, Texas, USA.

Hürlimann, M. et al., 2008, Hydrocarbon Composition from NMR Diffusion and

Relaxation Data: 2008 SPWLA Annual Logging Symposium, Edinburgh, U.K.

Jackson, R. et al., 2007, Pressure Measurements and Pressure Gradient Analysis: How

Reliable for Determining Fluid Density and Compositional Gradients? SPE 111911,

2007 SPE Nigeria Annual International Conference and Exhibition, Abuja, Nigeria.

Kleinberg, R.L. and H.J. Vinegar, 1996, NMR Properties of Reservoir Fluids: The Log Analyst, v. 37/6, p.20-32.

Lo, S.W. et al., 2000, Mixing Rules and Correlations of NMR Relaxation Time with

Viscosity, Diffusivity, and Gas/Oil Ratio of Methane/Hydrocarbon Mixtures: SPE 63217,

2000 SPE ATCE, Dallas, Texas, USA.

Mullins, O. et al., 2001, Gas-Oil Ratio of Live Crude Oils Determined by Near-Infrared Spectroscopy: Applied Spectroscopy, v. 55/2.

O’Keefe, M. et al., 2007, In-situ Density and Viscosity Measured by Wireline Formation

Testers: SPE 110364, 2007 SPE Asia Pacific Oil & Gas Conference and Exhibition,

Jakarta, Indonesia.

Return to top.