About This Item

Share This Item

The AAPG/Datapages Combined Publications Database

AAPG Special Volumes

Abstract


Pub. Id: A081 (1987)

First Page: 489

Last Page: 506

Book Title: SG 25: Exploration for Heavy Crude Oil and Natural Bitumen

Article/Chapter: Problems Frequently Encountered in Evaluating Tar Sand Resources--The South Texas San Miguel Deposit: Section V. Exploration Histories

Subject Group: Oil--Methodology and Concepts

Spec. Pub. Type: Studies in Geology

Pub. Year: 1987

Author(s): M. W. Britton

Abstract:

Good porosity and fluid saturation values are essential for accurately delineating original oil in place (OOIP) and evaluating the effectiveness of in situ recovery methods. Whereas the techniques for measuring and calculating these parameters from cores or downhole logs are well established for most conventional oil and gas resources, significant errors and inconsistencies can result if the same procedures are used to evaluate unconventional resources, such as tar sands and super-heavy oil deposits, thus adding more fuel to the age-old controversy that exists between core and log experts. For these resources, both evaluation techniques are not only useful but absolutely necessary, because neither alone will consistently yield good results.

This paper discusses some of the problems frequently encountered in evaluating unconventional resources where the hydrocarbon density and viscosity, and the formation matrix properties, are significantly different from those used in standardizing traditional core and log analysis procedures. It uses, for example purposes only, a -2° API (1093 kg/m3) gravity resource located in Texas and known as the San Miguel tar sand deposit.

One concludes that in dealing with unconventional deposits it is best to anticipate evaluation problems right from the start. Resolution of the problems is an evolutionary process that will likely not be perfected until a good number of wells have been drilled. Recognizing this ahead of time is important because it requires that the resource delineation and formation evaluation programs be properly integrated. Early in these programs all emphasis should be placed upon maximum data collection and retention in a form in which it can be reanalyzed at a later date with modified techniques. Resolution is further expedited by working closely with the core analysis and logging companies in the area. This assures that any improve evaluation techniques will be properly implemented.

Text:

INTRODUCTION

Whereas formation evaluation involves the determination of many different reservoir parameters, industry spends most of its time and effort trying to obtain good, accurate fluid saturation (S0, Sw, Sg) and porosity (^phgr) information for use in volumetric calculations. In no way is this meant to infer that physical and chemical properties of the oil (viscosity, density, composition, interfacial tension, etc.), the water (salinity, hardness, resistivity, etc.), the reservoir rock (compressibility, grain size, pore structure, mineral and clay content, cementation, etc.), and those parameters that depend on fluid-rock interactions, such as capillary-pressure characteristics, relative permeabil ties, wetability, and the critical fluid saturations (Swc and Sor) are not equally important to the reservoir engineer and to optimum development of any given resource. In comparison, formation evaluation begins with attempting to quantify the amount of oil present in the reservoir, and this depends solely upon the oil saturation and porosity values. Fortunately, procedures for measuring and calculating these parameters in cores and from downhole logs are well established and through repetitive application have become very reliable for most conventional oil and gas resources.

A problem begins to develop, however, when one attempts to routinely apply these standard measurement techniques, which have been empirically perfected on conventional resources, to unconventional hydrocarbon deposits such as tar sands or super viscous heavy oils. Here, tar sands or super-viscous heavy oils are loosely defined as those less than 10° API (1000 kg/m3) and/or having an in situ viscosity greater than 10,000 cp. Frequently, these materials contain excessive amounts of sulfur (>4 wt%), are generally deficient in hydrogen (H/C <1.75), and are found in unconsolidated friable sands. At some point, each of these characteristics begins to significantly impact the accuracy of standard core and log analysis procedures. All too frequently the tendency is to for et how the physical and chemical properties of the reservoir fluids and rock really affect core and log analysis procedures because these factors have, over

End_Page 489------------------------

time, been reduced to simplified correction factors that are often automatically incorporated into the final calculations, either in the form of a constant or through use of a nomograph.

This paper highlights some of the potential problems and errors that may be encountered when standard core and log analysis procedures are used to determine saturation and porosity values in tar sand and super-heavy oil resources. For the most part, it represents first-hand experience that Conoco gained in trying to exploit the -2° API (1093 kg/m3) San Miguel tar sand deposit, located in south Texas. Much of the quantitative discussion is therefore oriented to this deposit. The facts are that the specific problems encountered and the absolute magnitude of any analytical errors tend to be very resource-dependent. On a relative basis, the severity of any inaccuracies increases as the API gravity of the resource decreases, as the reservoir matrix becomes less consolidated, and as the geology gets more complex.

Although the concepts discussed in this paper are neither new nor unknown to those who have worked with unconventional hydrocarbon resources, they are frequently overlooked by individuals and small companies who get involved with tar sands development and lack extensive research and analytical facilities. The author sincerely hopes that, by highlighting some of the potential problems, future tar-sand developers will be forewarned of pitfalls that others have learned about the hard way.

SAN MIGUEL TAR SAND DEPOSIT

Since many of the core and log analysis problems discussed later are quantitatively described specifically for the San Miguel tar sand deposit, a general description of the resource will provide pertinent background information.

Location

The San Miguel tar sand deposit is located about 201 km (125 mi) southwest of San Antonio and about 48.3 km (30 mi) northeast of Eagle Pass, Texas (Fig. 1). The tar resource is contained within the San Miguel-4 Formation at depths ranging from 366 to 701 m (1200-2300 ft) under some 23,320 ha (90 mi2) of ranchland along the Maverick-Zavala county lines. Estimates indicate that the San Miguel-4 sand in this vicinity may contain 2-3 billion barrrels (0.32-0.48 km3) of -2° API (1093 kg/m3) gravity tar.

Conoco's Development Activities

Conoco Inc. currently owns or leases about 12,084 ha (29,860 acres) in the area (Fig. 2) and has systematically worked toward commercial development since early 1974. Undoubtedly, the most significant accomplishments to date have been the successful operation of two steamflood pilots at the locations indicated in Figure 2. The initial Street Ranch pilot (Britton et al., 1982) produced 169,040 barrels (26,875 m3) of tar during the 31-month period from December 1977 to June 1981 while the Saner Ranch pilot (Stang and Soni, 1984) recovered 133,260 barrels (21,200 m3) during a 23-month period ending January 1983. Although highly encouraging on a technical basis, pilot testing was recently terminated in view of current economic constraints and problems in upgrading th tar to readily salable energy products (Britton, 1983).

Over the years, Conoco has drilled, cored, and logged more than 50 wells penetrating the San Miguel-4 Formation. In an effort to accurately characterize and evaluate the tar sand deposit on both a macro- and microscopic level, well spacing has been varied from a maximum of one well per section for global-resource definition to one well every 0.19 ha (0.38 acres) for the pilot tests and recovery evaluations. It is this wealth of reservoir-evaluation experience that provides the background for this paper and the following resource description.

Tar Properties

The tar present in the San Miguel-4 sand is one of the most viscous, dense, sulfur-laden hydrocarbons known to exist anywhere in the world. It has an API gravity of -2° (1093 kg/m3) as compared to a gravity of 8-10° API (1014-1000 kg/m3) for tar in the Athabasca deposits of Canada and the 10-20° API (1000-934 kg/m3) range for most heavy crudes. It is essentially a solid at the 33°C (95°F) reservoir temperature. The immobile nature of the tar at reservoir conditions is illustrated by its 82°C (180°F) pour point and its extrapolated 20 million cp (20 kPa.s) viscosity (see Fig. 3).

From a refining or upgrading viewpoint, the tar is almost completely devoid of low molecular weight, normal paraffins, has an initial boiling point around 260°C (500°F), contains 37 wt % asphaltenes, and has an atomic H/C ratio of only 1.34. To make matters worse, the tar contains in excess of 10 wt % sulfur.

Obviously, the properties of the San Miguel-4 tar represent a major hurdle for most conventional production and refining methods. Two of its redeeming qualities are the fact that (1) the viscosity responds favorably to large increases in temperature, and (2) its metals content is relatively low (see Table 1).

Geology

The San Miguel-4 is the fourth in a series of nine sands deposited within the Taylor shale sequence of the Upper Cretaceous Montana Group. The northwest, updip location of each successive sand unit indicates that the San Miguel series (Weis, 1979) is a transgressive marine unit.

The San Miguel-4 deposit is predominantly sandstone with numerous irregular limestone intercalations. Its widespread geographical distribution suggests that it represents a highly reworked deltaic deposit on a shallow-marine shelf with occasional influence by an upper shelf, barrier bar, strand plain environment. The high degree of reworking was most likely caused by long-shore currents during a fairly

End_Page 490------------------------

steady, moderate energy period, as indicated by the very low clay content, good sorting, and the presence of thin-shelled pelecypods. A fairly high content of feldspars, ranging from 10% to 45% indicates short exposure to weathering agents and infers rapid transport and burial.

Probably the most striking characteristics of the San Miguel-4 sand are its uniformity of texture and its extremely high degree of sorting. The sand is generally very fine to fine grained, firm to hard, mostly subrounded, unconsolidated to occasionally friable, and quite clean. Bedding is mostly massive with horizontal burrowing fairly common. Scattered fragments of thin-shelled pelecypods are present through the sand section; however, they are locally much more abundant in the limestone intercalations.

The limestone sections are generally very hard, dense, and medium to coarsely crystalline. For the most part, they are thin, randomly dispersed throughout the gross sand body, and not very extensive areally, which suggests that they represent intertwining distributary channels. Occasionally, these channels became isolated and marine life flourished until the resulting calcareous growth was buried under new deposits of sand. Thus, the limestone streaks have a sharp upper contact and an undulating lower contact with the contiguous sand.

On at least two different occasions during deposition of the San Miguel-4 body, sand inflow stopped. Both times were evidently during a period of rising sea level. The resulting widespread calcareous growth resulted in the formation of a thin, semicontinuous, dense limestone streak about midway in the gross sand body and a massive, very continuous, limestone deposit on top of the sand. Therefore, at most locations the San Miguel-4 sand appears to be divided into two equally thick zones, each with thick, well-developed limestone caprocks. Heterogeneity within the sand zones is largely the result of the numerous discontinuous limestone intercalations.

Reservoir Conditions

The San Miguel-4 sand crops out just north of the Maverick-Kinney county lines and dips to the south-southeast at a slope of 2°. At the southern edge of Conoco's property (Fig. 2) the formation is about 762 m (2500 ft) deep. Gross thickness ranges from 6.1-24.4 m (20-80 ft) with an average of about 15.2 m (50 ft). Except for the interspersed limestone streaks, the sand is very clean with a mean particle size of 110 microns (µm) and a Schwartz uniformity coefficient less than 2.0. Its cleanliness and uniformity result in porosities of 26-30% and permeabilities of 250-1000 md (0.25-0.99 µm2).

Estimates of the original tar in place in the San Miguel-4 sand vary from 2 to 3 billion barrels (0.32-0.48 km3). This is a large deposit, but tar saturations range from a low of 20% to a high of only 60% and seldom average above 55% over any continuous 7.62-m (25-ft) interval. To make matters worse, the tar saturations vary considerably on an areal and vertical basis as may be seen by comparing the average properties for the two pilot areas (Figs. 4, 5). These saturation changes and reversals are further complicated by the preponderance of laterally discontinuous dense limestone streaks that run through the vertical section. Such conditions dramatically affect the performance of in situ recovery processes, particularly steamfloods, and they increase significantly the amoun of formation evaluation work necessary to optimize commercial development. In addition, formation evaluation is made difficult by the fact that the viscous tar provides the bulk of the cementation holding adjacent sand grains together.

CORE ANALYSIS PROCEDURES AND PROBLEMS

The following brief review of the conventional retort and Dean Stark core analysis procedures will help set the stage for the subsequent discussion

Fig. 1. Location of the South Texas San Miguel Tar Sand deposit.

Fig. 2. Conoco acreage and the location of various pilot tests.

End_Page 491------------------------

regarding how and why evaluation errors can occur when working with core specimens obtained from tar sand and super-viscous heavy-oil resources.

Conventional Retort Method

The conventional retort core analysis procedure is by far the most frequently utilized in the industry. It is a standard technique used by almost all core service companies and permits a large number of samples to be run in a short time period. As compared to more exotic core analysis procedures, it is quick and cheap.

In the conventional retort method, sealed metal extraction thimbles containing core plugs of known bulk weight (Wb) and volume (Vb) are placed in a thermostatically controlled oven. The temperature is first increased to about 204°C (400°F) to vaporize and remove any water present in the cores. The expanded vapor leaves the thimble via a small tube, condenses when it encounters a cold trap, and is collected in vessels for subsequent measurement (Ww). After a short period the temperature is increased gradually in steps to about 649°C (1200°F) in an effort to drive out all hydrocarbon material. Whether the hydrocarbon is removed in vapor or liquid form makes no difference, and no special effort is made to collect and measure all the hydroc rbon material (Whc) because this quantity is calculated by the following material-balance relationship:

[EQUATION (1)]

where Wrm is the weight of the core plug after it has been removed from the retort device; in theory, it is equivalent to the weight of only the rock matrix (rm).

The hydrocarbon saturation (S0) is calculated from the following equation, using a porosity value (^phgr) that has either been subsequently determined using an appropriate technique for the same core plug from which the hydrocarbons were extracted or another plug deemed representative of the formation:

[EQUATION (2)]

where ^rgrhc and ^rgrrm are the densities of the hydrocarbon and rock matrix, respectively. To obtain a true in situ value for the oil saturation it is also necessary in the above equation to divide by the appropriate formation volume factor (B0); however, for tar sands and super-heavy oils this value is very near unity and thus can be disregarded without introducing serious error.

Water saturation is then calculated from the equation:

Fig. 3. Viscosity comparison of Street Ranch, Athabasca, and Midway-Sunset hydrocarbons.

Table 1. Physical properties of south Texas tar.

End_Page 492------------------------

[EQUATION (3)]

which is an acceptable simplification for resources with very little to no initial gas saturation (Sg).

It is important to note that the above calculation automatically takes into account the probable loss of water from the core sample after it was removed from the formation, due to evaporation or whatever. Therefore, it is very likely that a summation-of-fluids analysis will lead to incorrect saturation and porosity values. In fact, for the types of resources being discussed, determination of ^phgr from the following equation:

[EQUATION (4)]

will almost always lead to pessimistic porosity values. If these values are then used to calculate the water saturation from the equation:

[EQUATION (5)]

and then the oil saturation is calculated as:

[EQUATION (6)]

the oil saturation can be overstated by as much as 50%. Thus, it is very important that the calculations be done correctly by someone who understands the implicit assumptions.

Other assumptions required to obtain accurate saturation values are:

1. That there has been little or no loss of hydrocarbon material from the core sample due either to flushing during the drilling operation or to seepage during subsequent core handling procedures;

2. That water removal and collection during the retort procedure is 100% efficient;

3. That all the hydrocarbon is removed from the core plug during the cleaning procedure, even if it is not collected and measured; and

4. That the porosity determined on the core plug is indeed representative of the true in situ porosity.

Assumptions 3 and 4 will be discussed in greater detail in the following paragraphs.

The initial developers of the retort method recognized that assumption 3 is not entirely true. Some hydrocarbon remains in the core, and therefore most core service companies apply a standard correction factor that was originally determined for core containing 30° (876 kg/m3) API crude, to get a true value for Whc.

Dean Stark Method

The Dean Stark core analysis procedure is a solvent extraction process that is much slower and therefore more expensive to run than the conventional retort method. In addition, not all core service companies are set up to offer this procedure.

In the Dean Stark method, core plugs of known weight and volume are placed in ceramic Soxhlets or extraction thimbles, covered with cotton, and placed in the recycle leg of a glass distillation column that is designed to circulate a suitable boiling-point solvent such as toluene. Condensed toluene drips into the top of the Soxhlet, drains through the core, flushes out both the hydrocarbon and water, and is collected in a heated vessel that acts as a reboiler. The toluene and water are vaporized in the reboiler and rise through a condenser, where the two are separated. The water is collected for measurement and the condensed toluene is returned to the Soxhlets. All removed hydrocarbon remains in the reboiler, and no attempt is made to measure it directly. The entire process is driven b gravity/density differences and takes place at the boiling

Fig. 4. Average reservoir properties at the Street Ranch pilot site.

Fig. 5. Average reservoir properties at the Saner Ranch pilot site.

End_Page 493------------------------

point of the solvent, which is about 110°C (231°F) for toluene. The cleaning process is usually deemed complete when water collection terminates.

Tar and water saturations are again calculated, using equations 1, 2, and 3 above, as in the case of the conventional retort method. For all practical purposes the same underlying assumptions prevail, but with the Dean Stark method the core is sufficiently clean to eliminate the need for a correction factor to obtain a true value for Whc.

Core Analysis Problems

It is very important to remember that these procedures do not measure oil saturation directly but rather, back-calculate it based upon a number of weight and density values. Any errors in making these measurements therefore directly affect the accuracy of the final saturation values.

With respect to the conventional retort and Dean Stark methods, no problems are involved concerning assumptions 1 and 2 above that detract from the results. Assumption 3 may lead to major error, as described next.

Retort Method

Removal of tar and heavy oil during the retort process is very inefficient. These materials tend to drain from the core plugs much more slowly than lighter, less viscous materials and to form coke deposits much more rapidly, thus leaving some residual hydrocarbon material that is never removed. Although the core service companies apply a standard correction that is supposed to account for some ineffectiveness in the core cleaning process, the correction can be grossly inadequate for viscous deposits. Even the correction applied in the "modified" or "heavy oil" retort procedure may not be sufficient for some deposits. Failure to adequately clean the core will consistently lead to oil saturations that understate the true in situ values.

This was found to be the case for the San Miguel tar sand; saturation values determined by the retort method consistently were less than those determined by the Dean Stark and other, more sophisticated analytical techniques. Investigation into the exact retort procedures being used disclosed that the core service company had the foresight to use a correction curve that had been determined for a 13° API (979 kg/m3) gravity crude but even this amount of correction was inadequate.

Tests with crushed core plugs showed that some carbon residue remained in the retort samples while those cleaned by the modified Dean Stark procedure were devoid of any residual hydrocarbon.

The amount of tar removed by the Dean Stark method was plotted against the amount of tar removed by the retort method (Fig. 6) in an effort to determine an applicable correction curve for the San Miguel deposit. Three attempts resulted in three different correction curves, the magnitude of the adjustment depending on:

-- the amount of tar originally present in the sample;
-- the retort thimbles that were utilized for the test;
-- the number of samples being processed in the retort oven; and
-- the rate at which the temperature is increased from 400° to 1200°F.

Although regression analysis eventually led to the correction curve in Figure 6, the retort core analysis procedure was deemed unreliable for critical saturation determinations in the San Miguel deposit because of the greater chance for significant error.

Dean Stark Method

Compared to the retort method, the Dean Stark technique is much better at removing viscous tar and heavy oil from core samples, thus yielding more reliable results. Even so, Conoco found that the standard cleaning procedure was still not 100% efficient for the San Miguel tar sand without a slight modification to increase the time that the core plugs were left in the extraction Soxhlets. Tests showed that the cleaning process needed to continue for at least 8 hours after water collection ceased but not longer than 16 hours. Tar removal was also accelerated and improved when the core plugs were crushed prior to extraction; this, of course, precludes subsequent porosity measurements on the clean core samples.

Core Porosities

Assumption 4 was that core-plug porosity represented true in situ porosity. This is a major problem, particularly when the rock matrix is friable to unconsolidated and basically held together by the hydrocarbon material. The problem stems from the fact that the core tends to expand upon removal from the reservoir (Zwicky and Eade, 1979; Dusseault, 1980).

Core-derived porosity and saturation values are generally assumed to be representative of in situ conditions, thereby underestimating the impact that expansion can have on these parameters. Actually, core-derived porosities can be 15-25% higher than those determined by in situ techniques. Overstating the core porosity results in tar saturations that are understated by a comparable amount (see equation 2), regardless of whether the retort or Dean Stark procedure is used. Core expansion also causes permeability, transmissibility, and compressibility measurements to be off by several orders of magnitude. Erroneous core-derived porosity and saturation values have little to no impact on original oil in place determinations because the inaccuracies cancel out in the standard volumetric equa ion:

[EQUATION (7)]

where A is the reservoir area in acres, H is the net pay thickness of the reservoir in feet, B0 is the formation

End_Page 494------------------------

volume factor of the oil, and the oil saturation, S0, has been determined from equation 2. Although the errors may have little impact on OOIP calculations, they do have a significant and generally negative impact on commercial recovery projections and pilot evaluations.

For the San Miguel deposit, core expansion was confirmed by the following:

-- 22.9 m (75 ft) of core measured by the geolograph and tubing strapping almost always amounted to 23.5-23.8 m (77-78 ft) of recovered core.

-- Comparison of gamma measurements on the core showed an elongation when compared to the downhole gamma logs.

-- Despite every effort to prevent water evaporation from freshly cut cores, summation of fluid analysis generally accounted for only 65-75% of the pore space.

Although expansion from 22.8 to 23.7 m (75-78 ft) may appear to be only a 4% increase, the true impact on porosity is given by the following:

[EQUATION (8)]

where: ^Dgr^phgr = fractional change in porosity caused by expansion
^DgrPV = resultant change in pore volume
^DgrL = observed change in core length
PVi = the initial or real pore volume of the core sample
R = the radius of the core sample
Li = the initial or real length of the core sample
^phgri = the initial or real porosity of the core sample

The inherent assumptions in the above equation are that only the pore space can expand and that expansion of the rock matrix and pore fluids is negligible. For the South Texas resource, these assumptions are reasonably valid. Using an in situ porosity value of 27.5% for the San Miguel deposit, the above equation indicates that core expansion could result in at least a 14.5% increase in core-derived porosities. This magnitude of error agreed well with the 30-33% porosity values frequently reported by the core service companies.

Unfortunately, no procedural changes can be effected to improve the accuracy of core-derived porosities; the only solution is to find a better, more reliable way to determine a true in situ porosity value. If this can be done, then this porosity value should be used in conjunction with the Dean Stark method and equation 2 to calculate representative saturation values.

LOG ANALYSIS PROCEDURES AND PROBLEMS

Logging Techniques

After running a variety of resistivity, induction, and porosity logs Conoco standardized its logging program by running Schlumberger's Triple Combo tool as the base log. This gives a shallow (SFLM), medium (ILM), and deep (ILD) induction signal, as well as a gamma ray (GR), spontaneous potential (SP), apparent water resistivity (RWA), hole caliber (DCAL), formation density (DPHI), and compensated neutron porosity (NPHI) response in a single log-presentation format (Fig. 7). It was chosen because:

-- It maximized the amount of data gathered with a single logging run;

-- It eliminated the need to correct for depth the data obtained from different tools and logging runs; and

-- It provided a log that was available from computerized logging trucks which automatically digitized the raw logging signals on tape.

The last point is extremely important, for the raw data are stored in a form whereby new log interpretations can be readily produced if and when better base-logging parameters are determined.

Other tools were occasionally run in combination with the Triple Combo log to examine special formation properties. These included the electromagnetic propagation tool (EPT), nuclear magnetism log (NML), the borehole-compensated sonic (BHC), and the carbon-oxygen (CO) log.

Fig. 6. Retort correction curve.

End_Page 495------------------------

Log Evaluation Problems

Log evaluation is both an art and a science, and it is no secret that interpretation is more difficult than normal in highly heterogeneous tar sand and heavy-oil deposits like the San Miguel (Freedman and Studlick, 1981). For discussion purposes, the problems will be couched in terms of how they either affect the porosity or saturation determinations.

Porosity

The basic problem with porosity indicators is that they do not directly measure porosity but rather, physical properties of the formation that can be related to porosity if the properties of the matrix and pore fluid are known. For tar sands especially, some of the critical fluid and matrix properties can be seriously in question.

The sonic log (BHC) measures the interval transit time, ^Dgrt, in microseconds, for an acoustic wave to travel through 1 ft of formation along a path parallel to the borehole. Porosity is then calculated from the Wyllie time-average equation:

[EQUATION (9)]

where: ^Dgrt = reading on the sonic log in µsec/ft.
^Dgrtrm = transit time of the rock matrix, µsec/ft.
^Dgrtf = transit time of the formation fluid, µsec/ft.

This equation is normally evaluated with ^Dgrtrm = 55.5 and ^Dgrtf = 190. For the San Miguel Formation, ^Dgrt values between 95 and 105 µsec/ft consistently gave porosities ranging from 29.4 to 36.8%. These values on average were even higher than those determined on the expanded core samples, indicating that either the travel time of the rubber-like tar was much slower than 190 µsec/ft or that the travel time of the partially consolidated SM-4 sand was slower than 55.5 µsec/ft. In any case, the sonic log was deemed to be of little value in this particular a plication.

Another porosity tool that seemed to give consistently high values was the neutron (CNL) log, which responds chiefly to the presence of hydrogen atoms and is a relative measure of the hydrogen density per unit volume. The high values are completely contrary to form, because, with a hydrogen index (Hh) of around 0.712, the neutron density log should read low by the following factor:

[EQUATION (10)]

where: Hh = hydrogen index of the tar
Hw = hydrogen index of water = 1.0
Shr = residual tar saturation in the zone of investigation

For the San Miguel Formation, Shr is approximately equal to the initial tar saturation of 0.50-0.55, which means that the neutron porosity should have been low by about 15%, or 4.1 porosity units (p.u.), compared to a true in situ value of 27.5%. Because of this inconsistency, the CNL log was also deemed of little value for the SM-4 sand.

The compensated formation density (FDC) porosity indicator seemed to work best for Conoco, once the appropriate shale and fluid density adjustments were made. For tar sands the fluid density needs to be increased to account for the fact that the tool normally expects to see nearly 100% water (^rgrf = 1.0) in the flushed zone; however, in this case the tar doesn't flush and therefore the formation fluid density in effect becomes a function of the tar saturation. True in situ (i.e., corrected) FDC porosities were therefore calculated from the equation:

[EQUATION (11)]

where: ^phgrufdc = the uncorrected porosity from the FDC log
^rgrh = tar density, g/cc
^rgrb = the bulk density value from the FDC log, g/cc

Fig. 7. Triple Combo log presentation, MSR #8.

End_Page 496------------------------

^rgrf = density of the formation fluid, g/cc
^Rgr = total dissolved solids content of the water (ppm) ^times 10-6

With S0 ranging from 0.3 to 0.6, ^rgrh = 1.09, P = 0.015, and ^rgrf = 1.06, the porosity varied from 26% to 29% through the best pay intervals and averaged about 27.5%.

Saturations

Saturation calculations in tar sands and super-heavy oils also can be very difficult and subject to substantial uncertainty owing to the fact that the formation properties vary so greatly from the norm.

Most interpretations from open-hole logs begin with a simplified version of the Archie formula for water saturation, Sw:

[EQUATION (12)]

where F is the formation resistivity factor, Rw is the resistivity of the formation water, and Rt is the true resistivity of the entire system in its natural state. Normally, Rw is either known or calculated, Rt is obtained from the induction log, and F is calculated from one of the following equations:

For soft formations:

[EQUATION (13)]

For hard formations:

[EQUATION (14)]

where m is an appropriate cementation exponent for the rock matrix. The above equations are really simplifications for the following basic relationship:

[EQUATION (15)]

where a is an empirically derived constant and n is the saturation exponent.

For friable, unconsolidated tar sands that are basically held together by a viscous, dense fluid, in contrast with normal cementation materials, the true values of a, m, and n can deviate substantially from the traditional values of 0.62-1.0, 2.0-2.5, and 2, respectively. The fact that different values of a, m, and n can have a significant impact upon the calculated Sw and S0 values can be readily seen in Figure 8.

This was indeed found to be true for the San Miguel deposit when use of the simplified Archie equations continually led to S0 and ^phgr combinations that disagreed with the bulk tar (S0 · ^phgr) values determined by the Dean Stark core-analysis procedure. Attempts to back-calculate equivalent a, m, and n values using an Rw of 0.3096 ohm-m led to values of 0.28, 2.9, and 2.9, respectively, but because of the tremendous scatter in the data, these values were not considered reliable enough for critical saturation work.

Combined Core and Log Analysis

Given the above core and log analysis problems for the San Miguel tar sand, the decision was made to use a combined log and core-analysis approach for evaluation purposes. This had the advantage of involving the strengths of each technique and eliminating their weaknesses. The synergistic effects of this combined approach have often been recognized by others (Zwicky and Eade, 1979; Granberry et al., 1964).

This approach relied upon the Dean Stark core-analysis procedure to give the true point values for the bulk tar saturation, Stb:

[EQUATION (16)]

where Vt is the volume of tar in a given core sample and Vb is the bulk volume of the core sample. Conoco requested that all Dean Stark data be reported according to this volume-fraction format. True in situ tar saturations (St) were then calculated by dividing the bulk-saturation values by representative porosity values determined from the FDC logs:

[EQUATION (17)]

Old Dean Stark tar saturations were corrected to this new basis using the following equation:

[EQUATION (18)]

Since the old Dean Stark tar saturations had been calculated using core-derived porosities, multiplying by the core-derived porosities converted the data back to a bulk basis.

POST-PILOT CORE AND LOG ANALYSIS

Unfortunately, the discussion above pertains only to core and log evaluation work that is conducted on the native formation prior to the initiation of any in situ recovery tests, particularly those involving any of the thermal EOR processes, such as steamflooding or fireflooding. These recovery techniques complicate formation evaluation because they alter fluid and rock properties, rendering the above analytical procedures inadequate.

The main problem is caused by coke deposition resulting from the high in situ temperatures of the steam and fireflooding processes. The fact that tar and heavy oil are thermally unstable and will undergo severe physical and chemical change during the deployment of thermal methods is widely recognized (Britton et al., 1982; Henderson and Weber, 1965);

End_Page 497------------------------

however, this same knowledge is not always shared by those involved with post-pilot log and core evaluation studies.

POST-STEAM LOG ANALYSIS

Microscopic variations in water salinity and formation temperature, neither of which can be precisely determined and both of which have a significant effect on water resistivity, Rw, caused by steamflooding, almost totally eliminate the utility of conventional resistivity and porosity logs for post-pilot evaluation. The problem is much more severe when coke deposition has occurred in parts of the reservoir. Although some of the problems and errors can be overcome by more sophisticated logs and analysis, the additional costs begin to shift post-steam evaluation in the direction of core analysis.

Conoco found this to be true for the Street (Britton et al., 1982) and Saner (Stang and Soni, 1984) Ranch post-pilot evaluation studies, where core and log results were compared directly. Figure 9 shows cropped log sections for four of the Street Ranch post-pilot core wells; the problem caused by injecting freshwater steam is indicated by the irregular and unpredictable SP response. The only accurate use for post-pilot logs has been for determination of net pay and logs for correlation. Therefore, if cost is a primary concern and a choice has to be made between cores or logs in post-steam wells, it is recommended that the time and money be allocated for core analysis.

POST-STEAM CORE ANALYSIS

Varying water resistivities, of course, have no adverse effect on core analysis procedures so the main concern is coke deposition. The fact that the temperatures associated with steamflooding can result in significant in situ coke formation has been observed and confirmed in several instances (Britton et al., 1982; Stang and Soni, 1984; Whitebay, 1984). Coke formation should be suspect any time that temperatures exceed 260°C (500°F) for a substantial time. In reality, the visbreaking and hydrocracking reactions that are responsible for the coke by-product are kinetically limited, and therefore they are time and temperature dependent. Coke formation is more severe as the tar density, viscosity, and molecular weight increase.

Fig. 8. Formation factor vs. porosity.

End_Page 498------------------------

Fig. 9. Cropped log sections following steamflooding in the Street Ranch pilot.

End_Page 499------------------------

Conoco first observed significant coke deposition during the Street Ranch post-pilot evaluation work when the Dean Stark core analysis procedure indicated that the tar saturation had been reduced to zero over a considerable vertical interval in many of the eight post-pilot core wells. Volumetric calculations and recovery estimates based upon the indicated tar saturation change were totally inconsistent with the production volumes and literature correlations. As a result, the cleaned core plugs were subjected to a number of inhouse tests (Whitebay, 1984), including additional extraction with stronger solvents, such as tetrahydrofuran (THF), differential thermal analysis (DTA), thermal gravimetric analysis (TGA), X-ray diffraction, and basic elemental analysis, including organic carbon. Although each test qualitatively indicated that there was indeed substantial toluene-insoluble material present in the core, basic elemental analysis was particularly useful in identifying it as carbon. At this time, scanning electron microscope (SEM) pictures, such as those shown in Figure 10, were used to confirm visually the presence of a thin amorphous substance on the sand grains and in pore throats.

Having established that the residual material was basically carbon, the next problem became one of developing a post-pilot core-analysis approach that would lead to an accurate recovery evaluation. A second series of tests was therefore initiated using both pre- and post-pilot core to establish background levels for any critical materials. The following analytical procedure was identified as yielding the most consistent and accurate results:

1. Run a normal Dean Stark analysis on each post-pilot core plug to determine a toluene-soluble, bulk-tar saturation. Since only the bulk tar value is needed, the analysis can be conducted on crushed core samples to improve the cleaning process.

2. Next, digest the samples in excess HCl to remove any inorganic carbon.

3. Determine the amount of organic carbon in each plug by subjecting it to a 1500°C (2732°F) oxygen environment, measuring the amount of evolved CO2, and calculating a wt% organic carbon value.

4. Subtract from the values obtained in step #3 a background level of 0.2 wt% organic carbon determined by this same procedure on presteamed core samples.

The final evaluation step involved converting both the Dean Stark bulk-tar data and the organic-carbon values into equivalent residual tar saturations so that accurate recovery estimates and projections could be made. The key is to be consistent and not violate any applicable material balance relationships.

Conversion of the bulk tar saturations was no particular challenge, because it required only dividing by an appropriate porosity value in accordance with equation 17. Since porosities could not be determined on crushed core samples, the only value that did not violate a material balance constraint was the average porosity for the pilot area:

[EQUATION (19)]

where OTIP is the volume of tar originally present in the pilot area, in barrels, Bt is the tar formation volume factor, A is the pilot area in acres, H is the average net pay in feet, and St is the average initial tar saturation. For Street Ranch, ^phgr was about 27.5%.

One key to relating the residual coke to an equivalent tar saturation was the use of a density value equal to that used for the toluene-soluble tar fraction and the initial tar, which in this case was 1.09 g/cc. Although efforts to determine empirically the true density of the amorphous coke material yielded values ranging from 1.3 to 1.5 g/cc, their use would have led to an invalid comparison. A second key was recognition that the coke-like residue contained elements other than carbon that were not represented by the carbon-weight fractions. Although an exact composition could not be determined, the small apparent density difference suggested that it was not appreciably different from that of the native tar. With these basic assumptions, the following equation was then applied:

[EQUATION (20)]

where: C0 = Ct - Cr
Ste = equivalent tar saturation, %
Co = corrected organic carbon, wt %
Ct = total organic carbon measured, wt %
Cr = reference organic carbon, assumed 0.2 wt %
^phgr = porosity
^rgrrm = grain density, assumed 2.65 g/cc
^rgrt = density of residual hydrocarbon, assumed 1.09 g/cc
^agr = weight ratio of carbon to tar, assumed 80%

The coke-equivalent tar saturation values were then added to the toluene-soluble saturations to get the total residual. This was done for each foot of core cut in the eight Street Ranch post-pilot core wells drilled at the locations shown in Figure 11.

As an example of the application of the above technique to the post-pilot analysis, the data for JSR-11 are presented in Table 2, with the Dean Stark saturations in column 2 and the organic carbon analyses in column 3. These same data are displayed graphically in Figures 12 and 13. Note how the intervals with the least amount of toluene-soluble material coincide with the highest coke concentrations. Correlation of the coke intervals with temperature profiles obtained during the pilot indicated that they most likely occurred adjacent to the live steam zone. Figure 14 shows the final saturation profile for JSR-11 and highlights the potential error that could have been made had the coke correction not been included.

End_Page 500------------------------

Click to view image in JPEG format. Fig. 10. [Grey Scale] Scanning electron micrographs of Street Ranch reservoir rock.

End_Page 501------------------------

Fig. 11. (opposite) Street Ranch pilot pattern.

Table 2. Post-steamflood tar saturations well: Joe C. Street No. 11.

End_Page 502------------------------

As mentioned, this same analysis procedure was used to determine the residual tar saturation at each of the post-pilot core-well locations. Before and after comparisons were then generated in accordance with the format shown in Figure 15 for JSR-11. This sort of pictorial presentation was very useful in helping the engineers visualize how well the steamflood had performed both areally and vertically throughout the pilot. Similar data for JSR-12-15, where significant saturation changes were observed, are presented in Figures 16-19, respectively, for comparison. The final step in the Street Ranch pilot evaluation involved using the point data provided by the post-pilot core wells to generate the 3-D recovery interpretation shown in Figure 20. In the end, knowledge of the existence of re idual coke prevented Conoco from overestimating by almost 12% the recovery efficiency of the steamflood.

SUMMARY AND CONCLUSIONS

Given Conoco's experience in the San Miguel tar sand deposit, the following conclusions seem warranted:

1. In dealing with unconventional resources, such as tar sands and super-viscous heavy-oil deposits, it is prudent to anticipate that formation-evaluation

Fig. 12. Organic-carbon profile for JSR-11.

Fig. 13. Comparison of Dean Stark and organic-carbon profiles for JSR-11.

Fig. 14. Coke-adjusted residual tar saturation profile for JSR-11.

End_Page 503------------------------

Fig. 15. Post-steam tar saturations at JSR-11.

Fig. 16. Post-steam tar saturations at JSR-12.

Fig. 17. Post-steam tar saturations at JSR-13.

End_Page 504------------------------

problems will persist, particularly when the formation is unconsolidated.

2. Early exploration and formation-evaluation priorities should be placed on identifying the source and magnitude of any problems.

3. If core- and log-analysis inconsistencies do exist, discussing the situation openly with the core and log service companies very often leads to the most expedient resolution.

4. Early in the evaluation process, it is best to retain all data in a form whereby improved analytical procedures can be readily used to generate a new interpretation.

5. During exploration, delineation, and initial pilot-testing, emphasis should be placed upon collecting as much data as possible from each well, even if it is not clear how some of the data may be used.

6. Cores cut in unconsolidated tar sands can expand, thereby causing porosity values to be too high and tar saturations to be too low. This may not affect OOIP calculations, but it can have a negative impact on recovery estimates.

7. Core and logs can often be simultaneously used to resolve formation-evaluation problems when neither approach alone will yield consistent results. In these instances, core will usually give reliable bulk-saturation values that can be converted to representative in situ values by using porosities obtained from an appropriate downhole logging tool.

8. Cores yield precise point values while logs look at an average response over an interval of several feet. Cores, if sampled properly, therefore can give a much truer impression of reservoir heterogeneity than logs, which are more useful for correlation and reservoir mapping purposes.

Fig. 18. Post-steam tar saturations at JSR-14.

Fig. 19. Post-steam tar saturations at JSR-15.

End_Page 505------------------------

9. The high formation temperatures associated with thermal recovery processes can significantly alter fluid and rock properties, thus further complicating formation evaluation procedures.

10. Variable water salinities and formation temperatures, affecting Rw, for all practical purposes preclude using induction logs for post-pilot analysis following steamflooding. This shifts the evaluation emphasis in favor of cores.

11. Coke deposition can occur during steamflooding, thus necessitating modified core analysis procedures in order to obtain accurate residual saturation and recovery data. In Conoco's case, a combined Dean Stark-organic carbon analysis procedure worked best on the -2° API (1093 kg/m3) South Texas tar sand deposit.

Fig. 20. Final 3-D post-steam recovery picture for the Street Ranch pilot.

References:

Britton, M. W., 1983, Tar the FAST way: Prepared for presentation at the Heavy Oil/Tar Sands: Producing, Refining, and Financing Conference held in Los Angeles, California, November 17-18.

Britton, M. W., W. L. Martin, R. J. Leibrecht, and R. A. Harmon, 1982, The Street Ranch pilot test of fracture-assisted steamflood technology: SPE Paper 10707 prepared for presentation at the 1982 California Regional Meeting held in San Francisco, March 24-26.

Dusseault, M. B., 1980, Sample disturbance in Athabasca Oil Sand: Canadian Petroleum Technology Journal, v. 19, n. 2, p. 85-92.

Freedman, R., and J. R. J. Studlick, 1981, How a Texas heavy oil prospect was evaluated: Oil and Gas Journal, November 30, p. 63-76.

Granberry, R. J., R. C. Wilshusen, and E. H. Koepf, 1964, Core analysis and electric-log data gang up on formation-evaluation problems: Oil and Gas Journal, August 3.

Henderson, J. H., and L. Weber, 1965, Physical upgrading of heavy oil by application of heat: Journal of Canadian Petroleum Technology, October-December, p. 206-212.

Stang, H. R., and Y. Soni, 1984, The Saner Ranch pilot test of fracture-assisted steamflood technology: SPE Paper 13036 prepared for presentation at the 59th Annual Technical Conference and Exhibition held in Houston, Texas, September 16-19.

Weis, B. R., 1979, Wave-dominated deltaic systems of the Upper Cretaceous San Miguel Formation, Maverick basin, South Texas: M. S. thesis, The University of Texas at Austin.

Whitebay, L. E., 1984, Tar sand analysis: Prepared for presentation at the Western Research Institute-DOE Tar Sand Symposium held in Vail, Colorado, June 26-29.

Zwicky, R. W., and J. R. Eade, 1979, The tar sands core analysis versus log analysis controversy--does it really matter?: UNITAR Proceedings--The Future of Heavy Crude Oils and Tar Sands, Edmonton, Alberta, Canada, June 4-12, p. 256-259.

End_of_Record - Last_Page 506-------

Acknowledgments:

The author wishes to thank the management of Conoco Inc. for permission to prepare and present this paper. He also wishes to acknowledge the tremendous effort put forth by all those individuals who worked on the South Texas Tar Sands Project over the years, and in particular to the engineers, research scientists, and technicians who spent their time trying to standardize and perfect both the core and log analysis procedures discussed in this document.

Copyright 1997 American Association of Petroleum Geologists

Pay-Per-View Purchase Options

The article is available through a document delivery service. Explain these Purchase Options.

Watermarked PDF Document: $14
Open PDF Document: $24