REVIEW OF LITERATUR.....

TAR MATS CHARACTERIZATION FROM NMR AND CONVENTIONAL LOGS, CASE STUDIES IN DEEPWATER

RESERVOIRS, OFFSHORE BRAZIL João de D. S. Nascimento and Ricardo M. R. Gomes - PETROBRAS

Copyright 2004, held jointly by the Society of Petrophysicists and Well Log Analysts (SPWLA) and the submitting authors. This paper was prepared for presentation at the SPWLA 45th Annual Logging Symposium held in Noordwijk, The Netherlands, June 6–9, 2004.

ABSTRACT

Tar mats can be defined as hydrocarbon horizons with high asphaltene concentrations (20% to 60% in weigh) and high viscosity – typically more than 10,000 cp at reservoir conditions. As a consequence of these characteristics, tar mats represent a volume of hydrocarbon in place that are very difficult or even impossible to be produced and frequently form vertical permeability barriers. The occurrence of these high viscosity hydrocarbon layers is generally at the bottom of the oil column. Therefore they can isolate the oil leg from the aquifer. In these cases, the producing drive mechanism will be by expansion in a volumetric reservoir, instead of water drive. So, a previous identification of tar mats will help to correctly quantify reserves and predict recovery with maximum efficiency.

Nuclear Magnetic Resonance logs in conjunction with conventional logs can provide accurate identification of tar mat levels and viscosity estimation, from empirical relationships. In this paper we present field examples of tar mat characterization from NMR and conventional logs, supported by formation pressure measurements in the aquifer and in the oil leg. Despite a very clear continuity of the reservoir all along the aquifer and oil leg, with an obvious oil/water contact, pressure data show evidence of depletion by production in the oil column, whereas in the water zone no pressure drop is noted.

In the studied field examples, tar mat levels are tens of meters thick and estimated viscosities are around 20,000 cp. The NMR responses (total porosity and T2 distribution) are very different in

the oil leg when compared to the tar mat horizons, as a result of the low hydrogen index levels and high viscosities in the tar mats, compared to the hydrogen indexes and viscosities of the medium/light oil. Also, because of the hydrogen index, the total porosity values measured by NMR and density logs are very different in the tar mat levels, but they have good agreement in the aquifer and oil zone. Neutron porosity is also affected, in minor intensity, by the low hydrogen index of tar mats. Additionally, resistivity logs show different responses due to the low mobility of tar mats when compared to the oil leg. The non-consolidated characteristic of the reservoirs in addition to the absence of mud cake along the tar mat intervals due to low filtrate invasion, result in caliper enlargement all along these high viscosity levels but not in the oil zone or in the aquifer.

INTRODUCTION

Tar mats in petroleum reservoirs are zones of variable thickness – less than 1 meter to over 100 meters – containing extra heavy oil or bitumen, typically with gravity under 10 °API and/or viscosity in situ above 10,000 cp, generally at the bottom of the oil column (Nascimento and Pinto, 2003). The high gravity and viscosity of tar mats stems from the high asphaltene content, normally 20 to 60% weight (Wilhelms and Larter, 1994).

Asphaltenes are considered the highest molecular weight hydrocarbon compounds in petroleum. The chemical structure of these compounds is mainly formed by carbon (100 to 300 atoms per molecule), hydrogen, sulfur, nitrogen, oxygen and minor proportions of nickel and vanadium (Pineda-Flores, 2001).

Gravitational segregation is the main process causing asphaltene enrichment and tar mat formation in crude oil. It is governed by different factors controlling the asphaltene stability in oil

FF

1

solution, like the in situ oil composition, pressure and temperature (Hirschberg, 1984; Boer, 1992). Tar mats at the bottom of oil reservoirs can be expressed as the extreme manifestation of oil compositional variation, caused by gravitational segregation of asphaltenes (Hirschberg, 1988).

Tar mats identification in exploration wells is crucial because this high viscous oil zones may contain important volumes in place that are very difficult or even impossible to be produced and therefore must be considered as non-reserves. Furthermore, tar mats may occur in large areas, with high thickness, forming vertical permeability barriers, isolating the oil leg from the aquifer and therefore, preventing water drive production mechanism.

TAR OR BITUMEN IDENTIFICATION FROM RESISTIVITY LOGS

Because of the very low mobility of high viscosity oil, such as tar or bitumen, resistivity logs have been the main wireline devices used for characterization of this type of hydrocarbon in reservoirs. Arab (1990) reports the use of deep (Rt) and shallow (Rxo) resistivity curves to identify bitumen occurrence in Upper Zakum Field (Abu Dhabi).

In Upper Zakum Field, with mud filtrate resistivity (Rmf) higher than connate water resistivity (Rw), the following typical responses were achieved, according to Arab (1990):

• In the oil bearing zones ⇒ Rxo reads less than Rt;

• In the water bearing zones ⇒ Rxo reads higher than Rt;

• In the bitumen occurrence zones ⇒ Rxo reads higher than Rt like in the water leg but with higher resistivity values.

Arab (1990) explained the resistivity responses in bitumen intervals by the mud filtrate ability to flush formation water from nearby hole, while not capable to flush the bitumen. Therefore, in the invaded zone, Rxo reads bitumen resistivity plus Rmf while in the virgin zone Rt reads bitumen resistivity plus Rw. Since Rmf is higher than Rw and bitumen resistivity is constant in both zones,

then Rxo will read higher than Rt in bitumen zones, as verified in field case.

Also, according to Kopper (2001), in the Orinoco Heavy Oil Belt in Venezuela, when Rxo reads higher than Rt, means that no movable oil (or tar) exists in the logged interval. A whole interval core indicated that the zone was oil-satured, however, it produced very little oil during the drill stem test.

Some authors such as Wilhelms, Carpentier and Huc (1994), report the comparison between deep and shallow resistivity curves plus the Sw and Sxo values, to recognize tar mats because of their very low mobility when compared with producible oil. These authors don’t use the Rxo higher than Rt condition to characterize the tars. The same values of resistivity curves, or same water saturations, are considered enough to identify tar levels.

TAR OR BITUMEN CHARACTERISTICS AND VISCOSITY ESTIMATION FROM NMR LOGS

The NMR porosity is derived from the signal amplitude, which is proportional to the hydrogen index (HI) of fluids in the porous rocks. The HI of pure water is defined as 1 and it is used to calibrate all the measurements. For alkanes, which are the major constituents of light crude oils, the HI is also equal to 1. So, light oils have the same signal amplitude of water and, consequently, same values of porosity are obtained either in a light oil or in water-bearing reservoir.

Because of the minor alkane constituents, higher aromatic contents and non-hydrocarbon components in heavy oils HI tends to decrease with increment of oil density. The API gravity is usually a good HI indicator in crude oil, with accentuated HI reduction when API gravity declines below 20. According to Kleinberg (1996), for a 10 API gravity oil HI is close to 0,7. Consequently, NMR measurements in heavy oil zones will exhibit porosity deficit proportional to the reduced hydrogen index.

Another characteristic of the NMR responses in heavy oils is the short T2 caused by high

2

viscosity. Morris (1997) empirically found out that viscosity (η) is a function of T2 log mean:

η0,9 = 1200/T2 log mean (1)

for η in centipoises and T2 log mean in milliseconds.

Additionally, he noted that along with the increment of oil viscosity, a tail of shorter relaxation times in T2 distributions also increases, representing the heavier components with minor oil mobility.

To determine in situ oil viscosity by NMR logs using equation (1) is necessary that oil and water T2 distributions are not overlapping. In cases of heavy oil viscosity near or greater than 100 centipoises, for example, the expected T2 log mean is near or minor than 15 milliseconds and, in addition, the tail originated from more restricted motion nucleus spans for very short relaxation times. In such cases, the oil and irreducible water signals overlap and consequently it is not possible to have direct viscosity estimation.

To determine in situ viscosity of extra heavy and high viscosity hydrocarbons, such as tar and bitumen, using NMR logs, LaTorraca (1999) proposed a empirical equation for indirect determination based on one of the characteristics of this type of hydrocarbon – the low hydrogen index (HI) and therefore, the NMR porosity deficit when compared with porosity measurements from others logs unrelated to the HI.

According to LaTarroca (1999), an apparent HI (HIapp) can be estimated using as inputs the porosity estimated from a log insensitive to the HI of the oil (∅ ), the NMR porosity (∅ NMR) and oil saturation (So) in the following equation:

HIapp = (So∅ _ ∆∅ )/So∅ (2)

where, ∆∅ is the difference between the porosity not related to HI and the NMR porosity.

However, the HIapp of heavy oils from NMR logs also depend on the echo spacing (TE) used in T2 measurements. Because T2 signal is obtained from samples at the echo peaks, TE is also the sampling interval and NMR logging tools don’t

have sampling rates fast enough to detect all the hydrogen in heavy oils (LaTorraca, 1999).

Correlations between HIapp and oil viscosity as a function of TE have been established leading to an equation for heavy oil viscosity (η) estimation:

ln(η)=(11+1.1/TE) _ (5.4+0.66/TE)∗ HIapp (3)

for η in centipoises and TE in milliseconds (LaTorraca, 1999).

CASE STUDIES

Two fields examples from deep-water reservoirs with tar mat occurrences at the bottom of oil column are presented. In the first example (figure 1), well “A”, a tar mat about 40 meters thick occurs above the aquifer. The top of tar in figure 1 is located approximately at xx52m and the base is close to xx92m, in the same depth of hydrocarbon/water contact.

One of the main tar mat characteristics from NMR logs in well “A” (figure 1) is the unimodal T2 distribution with shorter mean times, caused by the high hydrocarbon viscosity, when compared with T2 signal above xx52m, with bimodal T2 distribution in the medium/light oil leg, where capillary water and oil signals are separated. In the tar mat interval T2 distributions from hydrocarbon and water signals overlap, and a pronounced tail of shorter signals is evident below xx60 m, due to a more restricted motion tar components (track 5, figure 1).

A good agreement between “total” NMR porosity and density log porosity (track 4, figure 1) is evident in the water zone (below xx92 m) and in the oil leg (above xx52 m), whereas an obvious porosity difference occurs in the tar mat zone, where the NMR porosity shows about 8 p.u. deficit compared to the density log porosity. This feature is typical of hydrocarbons with short hydrogen index. Using the porosity deficit (∆∅ ) plus other parameters, as ∅ , So and the operational TE in equations 2 and 3 results in an estimated viscosity of approximate 20,000 centipoises.

The resistivity curves response (track 2, figure 1) corroborated another characteristic of very low mobility hydrocarbons, in the cases when Rmf is

FF

3

greater than Rw, as already mentioned in previous works. In the oil leg (above xx52 m) the Rxo reads less than deep and medium resistivity curves, while in the tar zone (xx52/xx92 m) the Rxo reads higher than deep and medium resistivity, similar to aquifer responses (below xx92 m) but with greater resistivity values. The differences in resistivity measurements along the tar mat interval are caused by the lack of bitumen displacement and probably because of the ionic exchange between less salty mud filtrate and more salty formation water.

The wash out in tar mat interval (track 1, figure 1) makes clear a particular feature caused by the insignificant bitumen mobility in this unconsolidated reservoir. In the oil leg and in the aquifer, where invasion is effective, mud cake is formed in the well bore, keeping the caliper near to the bit size diameter. Whereas, in the tar mat, where filtrate invasion is more difficult, no mud cake is formed and the well bore is enlarged by erosion from mud circulation.

An additional indication of tar mat low hydrogen index can be observed in neutron log porosity (track 4, figure 1). Although neutron tools are sensitive to all hydrogens, including that associated with minerals – instead of NMR tools that are only sensitive to hydrogen from fluids – a minor neutron porosity deficit can also be observed when compared to density porosity in tar mat. In contrast, no neutron porosity deficit is observed in the aquifer or in the oil leg.

Another interesting feature shown in figure 1 is a vertical viscosity variation along the oil column. The red flags in track 1 indicate that insufficient wait time for adequate polarizations occurs at the top of the oil column, caused by the largest T2 distribution in this zone, corresponding to the lesser viscous oil in the reservoir. So, light oil with low viscosity at the top of the reservoir grades to oil with medium viscosity (xx10/xx52 m), ending in a tar mat occurrence, at the bottom.

The evidence of tar mat occurrence in well “A” are validated from pressure measurements in the oil and water zones. Although a very clear continuity of the reservoir all along the aquifer and oil column, with an obvious

hydrocarbon/water contact, pressure data show evidence of depletion by production in the oil leg whereas in the water zone no pressure drop is noted.

Figure 2 shows a depth vs pressure crossplot from wireline tests along the oil and water intervals. The pressure gap between the oil column and water zone is evident. Because of the enlarged caliper and or very low fluid mobility, no pressure was obtained in the tar mat zone. The expected pressure in the hydrocarbon/water contact projected from the oil pressure gradient is 150 psi lesser than measured pressure at the top of aquifer interval, characterizing the hydraulic discontinuity between the aquifer and the oil leg.

The second field example (figure 3), well “B”, is in the same area of well “A”. A tar mat near 55 meters thick and identical well log characteristics also occurs below the oil column, but in this case, no aquifer is present, instead the tar mat lies directly on a shale sequence.

In this example it is not possible to confirm the tar mat as a hydraulic seal because of the obvious absence of a hydrocarbon/water contact. Nevertheless, all well log characteristics noted in xx52/xx92 m interval of well “A” are also present in xx20/xx75 m interval of well “B”, which is a relevant evidence of tar mat occurrence in the second well.

In well “B” (figure 3), the end of bimodal T2 distribution and the start of overlapping short time oil signals and water signals are around xx20 m (track 5). At this depth, initiates the apparent “total” NMR porosity deficit, compared to density log porosity (track 4); the neutron porosity deficit, compared to density log porosity (track 3); the crossover of resistivity curves with Rxo reading higher than Rt (track 2) and the washed out hole section (track 1).

All the described characteristics of well “B” are restricted to the interval xx20/xx75 m. Above and below this interval occur respectively medium viscous oil in a fine and laminated reservoir and a shale sequence; both with their peculiar log characteristics, very different from tar mat log responses.

4

CONCLUSIONS

The tar mat resistivity log responses from field examples presented in this paper are similar to the ones described in previous works, for the common condition of Rmf higher than Rw. Additionally, another tar or bitumen log characteristics derived from the low hydrogen index and high viscosity of this type of hydrocarbon were recognized in NMR logs and also discussed. A particular characteristic of non- invaded unconsolidated reservoirs was also evidenced from wash outs in the tar mat intervals.

Original pressure in the aquifer and depletion in oil column after production, confirmed from wireline pressure data in well “A”, enabled a validation of the well log indications and created high confidence log response patterns to a reliable tar mat identification, including for situations where the aquifer is absent.

ACKNOWLEDGMENTS

The authors would like to thank PETROBRAS for the support and permission to publish the data. We also thank the geologist Almério Barros França for his valuable help, revising the original text.

REFERENCES

Arab, H., 1990, “Bitumen Occurrence and Distribution in Upper Zakum Field”, Society of Petroleum Engineers, Paper Number 21323.

Boer, R. B. de, et al., 1992, “Screening of Crude Oils for Asphalt Precipitation: Theory, Practice and the Selection of Inhibitors”, Society of Petroleum Engineers, Paper Number 24987.

Hirschberg, A., et al., 1984, “Influence of Temperature and Pressure on Asphaltene Flocculation”, Society of Petroleum Engineers Journal, June 1984, 283-294.

Hirschberg, A., 1988, “Role of Asphaltenes in Compositional of Reservoir’s Fluid Column”, Journal of Petroleum Technology, January 1988, 89-94.

Kleinberg, R. L., et al., 1996, “NMR Properties of Reservoir Fluids”, The Log Analyst, November - December 1996.

Kopper, R., et al., 2001, “Reservoir Characterization of the Orinoco Heavy Oil Belt: Miocene Oficina Formation, Zuata Field, Eastern Venezuela Basin”, Society of Petroleum Engineers, Paper Number 69697.

LaTorraca, G. A., et al., 1999, “Heavy Oil Viscosity Determination Using NMR Logs”, SPWLA 40th Annual Logging Symposium, May 30 – June 3, 1999.

Morriss, C. E., et al., 1997, “Hydrocarbon Saturation and Viscosity Estimation from NMR Logging in the Belridge Diatomite”, The Log Analyst, March – April 1997.

Nascimento, J. de D. S., and Pinto, A. C. C., 2003, “Tar Mats – Gênese, Caracterização e Implicações em E&P”, Internal Report, PETROBRAS.

Pineda-Flores, G., et al., 2001, “Petroleum Asphaltenes: Generated Problematic and Possible Biodegradation Mechanisms”, Revista Latinoamericana de Microbiología, Volume 43, Number 3, pp. 143-150.

Wilhelms, A., and Later, S. R., 1994, “Origin of Tar Mats in Petroleum Reservoirs”, Marine and Petroleum Geology, Volume 11, Number 4, pp. 418-456.

Wilhelms, A., Carpentier, B., and Huc, A. Y., 1994, “New Methods to Detect Tar Mats in Petroleum Reservoirs”, Journal of Petroleum Science and Engineering 12, pp. 147-155.

ABOUT THE AUTHORS

João de D. S. Nascimento is a geologist at the Exploration Department of PETROBRAS. He has been working as log analyst ever since joining PETROBRAS in 1976.

Ricardo M. R. Gomes is a geologist and petrophysicist with PETROBRAS, where he has been working as an exploration geologist since 1977. He holds a B.Sc. degree in geology from the Universidade Federal do Rio de Janeiro, Brazil (1976) and a M.Sc. degree in geology from the Colorado School of Mines (1999).

FF

5

Figure 1 – Log responses in Well “A”. A tar mat was identified in the interval xx52/xx92 m, from T2 distribution (track 5), large total NMR porosity deficit (track 4), slight neutron porosity deficit (track 3), Rxo higher than Rt (track 2) and wash out (track 1).

xx00

xx50

xx00

6

Figure 2 – Depth vs pressure cross plot in well “A”. Pressure distribution shows the evidence of a permeability barrier between the oil column and the aquifer, given by the tar mat at the bottom of the oil column, as indicated from log responses along xx52/xx92 m interval.

PRESSURE x DEPTH

3450

3500

3550

3600

3650

3700

3750

3800

3850 4750 4800 4850 4900 4950 5000 5050 5100 5150 5200 5250 5300 5350

PRESSURE (psi)

D E

P T

H (

m )

Pressures in Depleted Oil Column

Original Pressures in Aquifer

150 psi Gap at HC/Water Contact

Tar Mat Zone (xx52/xx92 m)

xx

xxxxxxxx xxxx xx xxxx xxxxxxxx xx

xx

xx

xx

xx

xx

xx

xx

Well “A” - Pressure x Depth PRESSURE x DEPTH

3450

3500

3550

3600

3650

3700

3750

3800

3850 4750 4800 4850 4900 4950 5000 5050 5100 5150 5200 5250 5300 5350

PRESSURE (psi)

D E

P T

H (

m )

Pressures in Depleted Oil Column

Original Pressures in Aquifer

150 psi Gap at HC/Water Contact

Tar Mat Zone (xx52/xx92 m)

xx

xxxxxxxx xxxx xx xxxx xxxxxxxx xx

xx

xx

xx

xx

xx

xx

xx

Well “A” - Pressure x Depth

Pressures in Depleted Oil Column

Original Pressures in Aquifer

150 psi Gap at HC/Water Contact

Tar Mat Zone (xx52/xx92 m)

xx

xxxxxxxx xxxx xx xxxx xxxxxxxx xx

xx

xx

xx

xx

xx

xx

xx

Well “A” - Pressure x Depth FF

7

Figure 3 – Log responses in Well “B”. A tar mat was recognized in xx20/xx75 m interval, with identical log responses observed in well “A”. In this case no aquifer is present. The tar mat lies directly above a shale sequence.

xx50

xx00

xx50

8

1. odd_pg1: SPWLA 45th Annual Logging Symposium, June 6-9, 2004
2. even_pg2: SPWLA 45th Annual Logging Symposium, June 6-9, 2004
3. odd_pg3: SPWLA 45th Annual Logging Symposium, June 6-9, 2004
4. even_pg4: SPWLA 45th Annual Logging Symposium, June 6-9, 2004
5. odd_pg5: SPWLA 45th Annual Logging Symposium, June 6-9, 2004
6. even_pg6: SPWLA 45th Annual Logging Symposium, June 6-9, 2004
7. odd_pg7: SPWLA 45th Annual Logging Symposium, June 6-9, 2004
8. even_pg8: SPWLA 45th Annual Logging Symposium, June 6-9, 2004