PETROLEUM SOCIETY OF CIM PAPER NO. CIM 93-54 THIS IS A PREPRINT SUBJECT TO CORRECTION RESERVOIR FLUID SAMPLING AND RECOMBINATION TECHNIQUES FOR LABORATORY EXPERIMENTS BY Jeff Strong,Hycal Energy ResearchLaboratoriesLtd. F. Brent Thomas,Hycal EnergyResearchLaboratoriesLtd. D. Brant Bennion,Hycal EnergyResearchLaboratoriesLtd. PUBLICATION RIGHTS RESERVED THIS PAPER IS TO BE PRESENTED AT THE CIM 1993 ANNUAL TECHNICAL CONFERENCE IN CALGARY, MAY 9-12,1993. DISCUSSION OF THIS PAPER IS INVITED. SUCH DISCUSSION MAY BE PRESENTED AT THE TECHNICAL MEETING AND WILL BE CONSIDERED FOR PUBLICATION IN CIM JOURNALS IF FILED IN WRITING WITH THE TECHNICAL PROGRAM CHAIRMAN PRIOR TO THE CONCLUSION OF THE MEETING. ABSTRA~ The sampling of oil and gas condensate reservoirs require that representative fluid samples be removed by either surface or subsurface sampling techniques. This paper briefly reviews both of these techniques and discusses their relative merits. Several practical examples are provided that demonstrate the utility of an equation of state model to verify the quality of separator samples to be useC; in a recombination. In situations where free gas has been entrained with the separator samples, the equation of state model can frequently be used to synthesize an appropriate gas to be used in a recombination. Introduction Obtaining representative reservoir fluid samples has become of increasing importance in the development and exploitation of oil and gas condensate reservoirs. This is especially true of reservoirs where extensive computer simulations are usedto scopeout developmental strategiesor where , enhanced oil recovery options are investigated. Often times these decisions are based on properties measured on relatively small fluid volumes produced from the reservoir at one point in time. Therefore it is imperative that the fluid samples used to make these decisions closely match the characteristic properties of the reservoir fluids at actual reservoir conditions. Representative fluid samples can usually be obtained from producing reservoirs at either surface or subsurface locations. Surface samples are removed at either the separator or at the wellhead, with the associated gas and liquid subsequently recombined in proportions to represent the actual reservoir fluid. Subsurface samples are removed from within the wellbore at actual reservoir conditions using bottom hole sampling tools and techniques. The suitability of the particular sampling technique will depend on a large nun't>er of factors which may include economic considerations such as the cost of sampling and associated loss of production, the type of surface facilities that are available, the fluid volumes that will be required and the type of reservoir and fluid to be sampled. The sampling technique employed can be of particular importance in saturated oil or gas condensate reservoirs where the possibility of entrainment of disassociated phases decreases the likelihood of obtaining a truly representative fluid. In some of these situations, various techniques can be employed to compensate for entrainment of these dissociated phases. The focus of this paper is aimed at determining the techniques used to recombine the separator samples to represent oil and gas condensate systems along with many situations where they have been depleted into the two-phase region. A brief review of those situations where bottom hole sampling will more likely provide a representative sample will also be discussed. Sampling Techniques A thorough review of the equipment and techniques used to obtain these different types of fluid samples is outside the scope of this discussion and individuals who are interested in more extensive information on sampling procedures should refer to the cited literature (1,2.3) and information available from equipment vendors and service companies that specialize in sampling. However, a brief review of the most common sampling techniques will be useful to establish the fundamental principles that will discussed later in this paper. In general, surface samples obtained at the separator require collection of high stage separator gas and liquid which must be subsequently reconDined in a ratio that corresponds to the relative amounts of gas and liquid produced as the reservoir fluid travels up through the wellbore and on through the surface separation facilities. This type of sample is the most frequently used for several reasons: or asphaltenes) or emulsion formation Removing fluid samples from the wellhead itself is possible, although the presence of multi-phase flow will often result in heterogenous samples that will require modification of the gas phase in order to obtain representative reservoir fluid. Wellhead samples are usually taken only In those instances where chemicals are being added at the surface separator and there is no other location available for removing an uncontaminated fluid sample. In rare instances, If the reservoir is highly undersaturated with a bubble point pressure that is actually lower than the wellhead pressure, wellhead samples may provide fluids that are nearly equivalent to subsurface samples. However, in general, wellhead samples will not directly provide representative reservoir fluids without altering the gas phase to achieve the correct reservoir fluid. Subsurface samples are collected by lowering a special sampling tool through the wellhead into the bottom of the well near the perforations where live reservoir fluid can be captured and brought back to the surface. Prior to collecting subsurface samples the well Is typically conditioned by restricting the flowrate in order to level out pressure imbalances In the near wellbore region and then shutting In the well for a period of time (usually 24 to 72 hours) to allow fluids to collect and equilibrate in the well bore. The advantagesto subsurfacesampling of fluid reservoirsare: - in situations where nothing is known about the reservoir fluid, subsurface samples may provide a good indication of overall fluid properties such as composition, GOR and saturation pressure. These results can be useful in assessing the quality of any subsequent surface samples. - accessto the surface fluid samples is readily available - relatively large fluid volumes can be collected - the cost of collecting surface sa~les is - usually much lower than collecting bottom hole samples there is virtually no interruption of production during the sampling period (although conditioning of the well prior to sampling may require some alteration of the oroduction rate). Surfacesamples will typically yield representative samples from the well provided that the well is producing at a stable gas/oil ratio (GOR), disassociated phases are not entrained in the produced fluids and there is no solid {such as waxes 2 - the fluid when brought to surface represents the In situ fluid, and as such, does not needto be recombinedto a target saturationpressure or target GOR fluids that have cloud points greater than surface temperature, those with a propensity to precipitate solids (such as asphaltenes) with reduction in pressure or temperature or those with tendencies to form emulsions will be more representative from a subsurface sample than those fluids collected at the surface separator - entrainmentof disassociated phasescan be less severe depending on the effectiveness of the well conditioning. While many believe subsurface samples provide the best opportunity of achieving a representative reservoir fluid, these samples can be subject to relatively low capture rates and have a limited collection volume. Moreover, considerable disparities have been observed on some oils where multiple bottom hole samples have been taken. Consideration of which method to use for sampling will depend on the aforementioned variablesand particularlyon the type of well to be sampled. For the sake of clarity, the issue of sampling oil wells and condensate wells will bs treated separately although many of the sams principleswill apply to both. Sampling of 011Wells For undersaturated reservoirs, the recombination of surface separator samples will usually result in a representative reservoir fluid provided that the well is producing at a stabilized gas-oil ratio (GOR). For undersaturated reservoirs where the producing GOR is not stable then the possibility exists that the bottom hole flowing pressure (BHFP) may actuaJly be lower than the saturation pressure of the fluid. In this situation, solution gas may be liberated in the near wellbore area which then must first achieve a critical gas saturation before it will flow Into the wellbore and on to the separator. However, once a steady state equilibrium is established in the near wellbore region then the producing GOR will usually stabilize and the surface separator should yield fluids suitable for recombination. Sampling saturated oil reservoirs provide a special challenge since the production of any gas cap or previously liberated solution gas will usually result in a non-representative recombination. Reducing the flow rate of a well and observing if there is a corresponding drop in the measured separator GOR may reveal if gas coning or gas liberation effects are being observed in the well. Figure 1 shows the standard relationship between GOR and flow rate which one might expect for such a well. For the best opportunity to obtain representative samples, the surface separator should be operated at a condition below the threshold flowrate that will induce entrainment of disassociated gas phase. If operated above, then the likelihood of having excess gas (over and above the solution GOR) along with a leaner gas phase is much increased. 3 The recombination of surface separator samples is achieved by either recombining the gas and fluid to match the measured separator GOR or to match a specified saturation pressure at the reservoir temperature. Ideally, matching the recombination to one of these characteristics will result in a fluid that corresponds well in the other, although this is not always the case. Whether the saturation pressure or the gas-oil ratio is selected as the fluid characteristic to be matched will usually be determined by whether the separator is producing at a stable GOR and whether an accurate estimate of the saturation pressure is actually known. In cases where the reservoir fluid is known to be highly undersaturated, the target saturation pressure may be significantly lower than the actual reservoir pressure and therefore the separator GOR may be a better reservoir fluid characteristic to attempt to match. For saturated oil reservoirs where an existing gas cap is known to be in contact with the oil, the saturation pressure of the oil will generally be equal to the current reservoir pressure and therefore, the saturation pressure may be the better fluid characteristic to match. Sampling of Gas Condensate Wells Similarly, gas condensate systems may also exhibit extreme sensitivity with respect to pressure and temperature conditions and it is difficult to get representative samples from the surface separator unless one has access to a large separator, where large fluid volumes have been averaged, and where stable flow rates are observed. In the case of subsurface samples, these are also sometimes subject to temperature and pressure sensitivities and their results need to be quantified and scrutinized closely. Consequently, for gas condensate systems there is no guarantee that a bottom hole sample will be superior to surface samples and therefore all results need to be closely examined. Indeed, for gas condensate systems the most important factor may be when the sample was taken once production had been initiated. Simulation studies conducted by McCain and Alexander3) on gas condensate wells have suggested that gas samples should be obtained early during the first 30 days of production in order to obtain a representative reservoir sample. After that point the gradual buildup of a condensate ring around the wellbore will prevent the collection of representative samples regardless of the amount of well conditioning that is performed. Equation of State Modelling Equation of State modelling can be particularly usefulto evaluatethe qualityof the surfacesamples and provide a method of recombiningphases in orderto predictoverallphasebehaviourat reservoir conditions. In instances where the entrainmentof disassociatedphasesis suspected,the Equationof State can be a valuable tool to determine if the separator gas collected is representativeof the evolved reservoir solution gas. This can be illustratedwith the followingexample. EOS we can detennine whether the separator samples can be recombined to represent the present in situ reservoir liquid. The results of this comparison are provided in Table 2. Table .- 2 . Comperi80n Component Example 1. A saturated oil reservoir had an original pressure of 15,168 kPag (2200 psi) at 65°C (14eoF). Since that time the reservoir has been depleted to a current reservoir pressure of 11,032 kPag (1600 psig). In order to perform laboratory tests on the field It was desired to recombine separator oil and gas samples to represent the present in situ liquid phase. An original compositional analysis was available from a bottom hole sample taken in 1960 and co~ositional analyses of the current separator gas and liquid were also available.These are summarized in Table 1. Figure 2 shows a general schematic of a depleted reservoir as used in this exa~le. N. co. HaS c, c. c. I-c. n.c. .,c. n.c. ~ of ~.tor - EOS G.s 0.0$1 0.0271 0.1231 0.6808 0.0840 0.0277 0.0025 O.~ 0.0013 0.0011 0.0013 ~ Measured Gas 0.0056 0 .0264 0.1295 0.6790 0.0812 0.0351 0 .0034 0.0084 0.0016 0.0013 0.0008 This comparison shows that the methane content of the EOS-generated separator gas was slightly higher than that of the sampled gas. This suggested that there was no gas cap gas entrained in the separator gas since the EOS predicted a gas cap gas containing approximately 76% methane which would have contributed to a higher methane content in the separator gas. A subsequent recombination of the separator gas and liquid samples according to the measured separator GOR provided a recombined oil sample with a saturation pressure that was within 100 kPa of the present reservoir pressure. The compositional analysis of the final recombined oil is provided in Table 3 which shows relatively good agreement with the composition predicted by the EOS. Therefore, by using the EOS, the quality of the recombination has been evaluated and the confidence level is improved. In order to determine whether the separator gas and liquid could be used to recombine a representative sample of the in situ liquid an equation of state was used simulate the reservoir fluid. Based on the original compositional analysis of the bottom hole sample an EOS model was tuned to fit the original bubblepoint pressure of 15 168 kPag at 65°C. Once the EOS had been tuned to match the original reservoir conditions, the oil was then partially depleted to current reservoir conditions and then flashed to the present separator conditions. Figure 3 provides a schematic description of this EOS procedure. By co~ring the composition of the actual separator gas to that generated by the 4 As previously mentioned, the presence of a gas cap in a saturated oil reservoir can frequently result in the entrainment of the disassociated gas phase into the separator fluids thereby making a direct recombination of fluids invalid. However, based upon preliminary EOS analysis a suitable separator gas can be synthesized in the laboratory in order tc obtain a representative recombined oil sample. appropriateGOR. In order to reduce the saturation pressure to the target reservoir pressure, the GOR had to be decreased by almost 20% thereby altering the transport properties and composition of the oil. However, when a second recombination was performed using a blended synthetic separator gas based on the composition predicted by the EOS, the GOR obtained was 82.5 m3/m3 with a saturation pressure within 100 kPa of the current reservoir pressure. A comparison of the recombined oils is providedin Table 5. Example 2. This second example shows a situation where the separator oil and gas as sampled cannot be used in a direct recombination due to the entrainment of gas cap gas. As in the first example, an initial reservoir fluid composition was available from very early in the productive life of the reservoir which was used to input into the EOS model. The model was tuned to give an oil with a bubblepoint pressure of about 15 860 kPa (2300 psig) at 75°C (16~F). The schematic of Figure 4 shows the EOS process used for this situation. However, unlike the first example a comparison of the sampled separator gas and the EOS-generated separator gas showed considerable differences. Table 4 shows the comparison of these two gas compositions. Table 4 . Comparisonof SeparatorGases Component EOS N. co. 0.0059 0.0512 H.s 0.0064 Ct 0.6387 0.1362 0.0907 0.0554 0.0155 Example 3. The next example considers a situation where separator oil is not available and the recombination must be performed with dead stock tank oil. This can occur when an emulsion has formed in the separator and the oil must first be degassed and then centrifuged in order to remove the water. However in this example the oil actually came from an overseas well which was accidentally degassed during transport. A compositional analysis of the reservoir fluid was available along with the saturation pressure and single stage flash GOR which were used to tune the equation of state model. Tuning of the EOS was performed by adjusting the temperature and pressure of the EOS flash until the composition of the liquid from the EOS closely matched the actual analysis of the present separator liquid. Once the model had been tuned, the necessary gas was synthesized in the laboratory and then recombined with the remaining dead oil. Table 6 provides the properties of the resulting recombined oil sample compared with the original properties of the bottom hole sample. I c. c. c. ~ 0.7309 O.~ 0.0641 0.0392 0.0080 c. There is a significantamountof excessmethanein the sampled gas which is most likely the result of gas cap gas entrainment into the separator. Using the actual separator fluids in a recombination resulted in a fluid with a saturation pressure much higher than the maximum reservoir pressure for the 5 from a free liquid leg. The producing GOR was in the range of 1500 m3/m3and the liquid had a gravity of 43 API. An equation of state model was developed to detennine if the liquid could have possibly resulted from a gas-condensate system. This was perfonned by using the equation of state to model the gas and liquid phases produced through the separator and then recombining these two phases with the EOS in varying amounts to detennine the GOR that would be required to achieve a single phase at reservoir conditions. The results indicated that a GOR in excess of 100000 m3/m3 would be required at 63°C in order to achieve a saturation pressure of 15 MPa. Since this GOR is far in excess of what was observed in the field, this would appear to suggest that the production from this well was a combination of gas and entrained free liquid rather than resulting solely from a gas condensate system. The sampling of gas condensate wells has already been discussed in terms of the relative merits of subsurface versus surface sampling. One particular problem that is sometimes encountered is determining whether a sample collected at surface represents a gas condensate, a free liquid leg in the reservoir or a combination of both. The distinction between oils and condensates is usually evident from a compositional standpoint and as a general rule it has been suggested by Moses(4)that reservoir fluids which are less than 12.5 mole% heptanes plus are usually in the gas phase in the reservoir. This can also be confirmed with an equation of state model performed on separator samples taken early in the life of the reservoir. The key question to answer is if the separator gas at the reservoir temperature and pressure can vaporize the liquid corresponding to the measured separator GOR. If not, then the GOR needs to be increased until the liquid is vaporized. If the saturationpressure of that fluid Is within the realistic limits of the reservoir then the produced liquid may be derived from the vapour phase in situ. However, if the corresponding saturation pressure is much higher than the maximum pressure of the reservoir then there is good evidence that a free liquid leg exists in situ. This technique should be only employed early in the life of a suspected condensate well since the loss of liquids in the near wellbore region will distort the phase behaviour of the recombined fluids and give erroneous results. Example 4. An example of this previous situation is where a gas well at 15 MPa and 63°C was producing significant quantities of liquid at the surface separator although it was not known whether the liquidswerethe resultof a gas condensate or 6 When separator samples are recombined to represent very lean gas condensate systems with GOR's in excess of 5000 m3/m3, the resulting mixtures are very difficult to use in the laboratory tc measure volumetric properties due to the small volume of liquid. Constant volume depletion (CVD) tests require that exact dewpoint pressures and phase volumes be determined experimentally using relatively small volumes of overall sample. The resulting experimental error associated with these measurements can result in inaccurate estimates of the two phase formation volume factors. An alternative methodology that can be e~loyed when dealing with reconmined gas condensate systems is to recombine the samples to a GOR that is low enough to be able to accurately measure the phase behaviour of the mixture. Once this has been accomplished then an EOS model can be tuned with the experimental data and then subsequently used to predict phase behaviour at the actual field GOR. Recommendations The collection of representative samples from a reservoir can be accomplished through the use of surface separator samples or subsurface samples but results should be scrutinized carefully to ensure that the final reservoir fluid Is consistent with the properties of the reservoir. Equation of state models can be employed to great advantage to assist in the evaluation and synthesis of separator samples to be used in recombinations. As with all computer simulations, the quality of the input data should be evaluated before it is used to model prospective recombination fluids. Conclusions References 1. The selection of an appropriate sampling technique will generally be made based on factorssuch as overallsamplecosts,the surface facilitiesthat are available, the fluid volumesthat are required and the type of reservoir to be sampled. Regardlessof the methodemployedto collect the sample, the resulting reservoirfluid should be scrutinizedcarefully to ensure that It accuratelyrepresentsthe in situ fluid before It is used in any laboratorystudy. 1. Reudelhuber,F.O.,Sampling Proceduresfor Oil Reservoirs; Journal of Petroleum Technoloav. Dec. 1957.00. 15-18. 2. An equationof state model can be used to evaluatethe qualityof surfacesamplesespecially those where the entrainment of disassociated phases is suspected. 3. In situationswhere separator gas phases have been contaminated or are not available, an equation of state model can be used to synthesizethe appropriateseparatorgas to be used in the reco~ination. The accuracyof the recombinationwill depend on the quality of the originalfluid compositionused in the model. 4. For suspectedgas condensatesystems,an EOS modelcan be used to determineif the produced liquid is a condensate resulting from the productionof gas or a free liquid resultingfrom entrainmentwith the produced gas. The EOS model should only be based on samples producedearly in the life of the well since later samplesmay be nonrepresentative. Acknowledgments The authors would like to acknowledge the supportof Hycal EnergyResearchLaboratoriesLtd. in providingresearchfacilitiesand sampledata used to preparethis presentation. 7 2. API RP 44. API Recommended Practice for SamDlina Petroleum Reservoir Fluids; 1st addition, Jan. 1966 3. McCain, W.D. and Alexander, A.A., SalT1>lingGas Condensate Wells; SocietY of Petroleum Enaineers Reservoir Enaineerina. Aua. 1992 . DD. 358-362., 4. Moses, P.L., EngineeringApplicationsof Phase Behaviourof CrudeOil and CondensateSystems; Journal of PetroleumTechnoloay.July 1986. DD. 715-723. GAS ENTRAINMENT FIGURE 1 AS A FUNCTION OF FLOWRATE / ,/ 0 ~ ~ ~ Well Flowrate FIGURE 2 GENERAL SCHEMATIC OF DEPLETED RESERVOIR Separator Gas Separator Oil Gas Cap Saturated Oil System Depleted Oil FIGURE 3 EXAMPLE 1 - EOS SCHEMATIC FIGURE 4 EXAMPLE 2 - EOS SCHEMATIC