Enhanced recovery

Use of Partitioning Tracers To Estimate Oil-Saturation Distribution

Partitioning interwell tracer tests (PITTs) have been used to estimate remaining oil saturations (ROSs) during waterflooding.

Abstract image
Getty Images

Partitioning interwell tracer tests (PITTs) have been used to estimate remaining oil saturations (ROSs) during waterflooding. Compared with core tests, well logs, and single-well tracer tests, PITTs sample a much larger representative elemental volume (REV) and provide interwell estimates of remaining oil saturation. In this paper, the authors present the information gained from conducting a polymer PITT and the saturation estimated during the PITT. The polymer PITT allows characterization of polymer-flood efficiency and is a useful tool in polymer-flood evaluations in heterogeneous reservoirs.

Introduction

The difference in arrival time between conservative tracers has long been used to estimate the volume of oil remaining in a reservoir after tracer tests. The method-of-moments solution to calculate ROS from mean arrival times of conservative and partitioning tracers is based upon single-phase flow. When large changes of saturations occur in between the initial conservative and partitioning tracers, the volume calculated is hard to interpret. Use of equations with ­single-phase-flow assumptions makes it a challenge to understand the correlation between the volume calculated and its corresponding time.

The method of moments has been used to calculate oil volumes and saturations. With the method of moments, the mean residence time of the tracer for a given slug duration is calculated from the concentration of tracer in the produced effluent.

Oil volume is normally related to the difference in mean residence volume between a partitioning tracer and a conservative tracer, although two partitioning tracers can be related. The remaining oil volume can be calculated from the mean residence volumes of the conservative and partitioning tracers. When oil cut and saturation change are large, the volume becomes difficult to correct. Often, correcting for production on the basis of  the mean residence time of the first tracer gives the right value if oil production is small, but this does not give the correct answer if oil cut is high.

Conducting a PITT with a favorable mobility ratio helps alleviate the problem of high oil cut during tracer production because it allows the partitioning tracer to contact oil at residual saturation.

In order to calculate the volume of oil and saturation more accurately during an active waterflood with mobile oil, a technique is required that estimates oil remaining behind the waterflood. To achieve that objective, modifications to the method of moments, known as the residence-time-distribution-­analysis (RTDA) method, were made in order to account for two-phase flow. In this new derivation, oil and water can be produced simultaneously and are accounted for by the fractional flow of each phase.

Simulation Results

Several scenarios (described in detail in the complete paper) were simulated with an established simulator. Most parameters remained the same throughout all simulations. The one changing variable was the endpoint mobility ratio (M) of the tracer slug and chase. Each simulation case is performed in a 1D homogeneous, isotropic reservoir with a permeability of 7 darcies and a porosity of 0.4. Longitudinal dispersivity is 0.6, and the transverse dispersivity is 0.06. The residual water saturation and residual oil saturation to water are both 0.10. Endpoint relative permeabilities are 0.9 and 0.25 for oil and water, respectively.

The results indicate that the modified RTDA technique correctly estimates remaining oil saturation for all the partitioning tracers. For the tracers with the higher partition coefficients, the correct value of ROS is achieved at a later time. The higher-­partition-coefficient tracers are the most retarded and see minimal saturation changes and, hence, provide the most-­accurate estimates of remaining oil. However, it must be noted that high partition coefficients translate into longer tracer-test durations and have to be balanced with field costs.

Laboratory Results

In addition to simulations, corefloods were performed with partitioning tracers in sandpacks. Oil and water were both present in order to see the effects of different mobility ratios. Tracer data were obtained from the effluent by gas-chromatography analysis. Effort was made to ensure accurate material balance by measuring sandpack pore volume and keeping track of produced-oil and -water volumes. The oil volume and average ROS  were calculated by the RTDA method with fractional flow.

A total of four partitioning tracer corefloods were performed at different mobility ratios, with and without mobile oil. Isopropanol was always used as the conservative tracer. Ethylene glycol monobutyl ether (EGBE); 2,2,3,3,3–pentafluoropropan-1-OL (PFP); 1H,1H–heptafluorobutan-1-OL (HPFB); 2,2,3,4,4-hexafluorobutan-1-OL (HXFB); and 1H,1H,5H–­octafluoro-pentan-1-OL (OCFP) were used in various experiments as partitioning tracers.

Conservative and partitioning tracers were injected in slugs of approximately 0.15 pore volumes. The first coreflood has a poor mobility ratio (M=30) and a high amount of mobile oil (0.41). The second coreflood has a poor mobility ratio (M=30) and a low amount of mobile oil (0.26). The third coreflood has a favorable mobility ratio (M=1) and low ­mobile-oil saturation (0.17). The fourth coreflood has a favorable mobility ratio (M=1) and a high amount of mobile oil (0.50).

After being oilflooded to an initial oil saturation of 0.89, the sandpack for the first coreflood was waterflooded to an oil saturation at the start of tracer of 0.41. Two partitioning tracers, EGBE and OCFP, were used and have dynamic partition coefficients (K-values) of 0.48 and 1.67, respectively, for this oil/brine combination. Analyzing the first coreflood is difficult with the method of moments; when the RTDA method with fractional flow was used, accurate pore volumes and ROSs were calculated. The calculated pore volumes from EGBE and OCFP are within 1% of the pore volume from material balance. The values of ROS from EGBE and OCFP tracers were within 1 and 2.5% of the material-­balance ROS, respectively.

The second coreflood was simple to interpret because of no noticeable change in oil saturation. After being oilflooded to an initial oil saturation of 0.88, the sandpack was waterflooded to an oil saturation at the start of tracer of 0.26. At this point, even with poor mobility, oil cut was less than 1%. PFP, HXFB, and OCFP were used as partitioning tracers. Their dynamic K-values with this oil/brine combination were 0.36, 0.63, and 1.54, respectively. Pore-­volume estimation was within 1% of material balance. ROSs calculated using PFP, HXFB, and OCFP were within 2.5, 2, and 1% of ­material balance, respectively.

The third coreflood had both a low ROS and favorable mobility ratio. After being oilflooded to an initial oil saturation of 0.88, the sandpack was water­flooded to an oil saturation at the start of tracer of 0.13. Polymer was used to achieve favorable mobility at the start of tracer injection. The same tracers were used in the experiment as partitioning tracers. The K-values were 0.41, 0.77, and 1.67, respectively. Pore volumes from all three tracer analyses were within 1% of the material-balance values. ROS ­errors were within 3.5, 2, and 1% for PFP, HXFB, and OCFP, respectively.

The fourth coreflood closely matched material-balance data even with a high initial oil saturation because of the favorable mobility ratio. After being oilflooded to an initial oil saturation of 0.84, the sandpack was waterflooded to an oil saturation at the start of tracer of 0.50. Polymer was once again used to achieve favorable mobility at the start of tracer injection. EGBE and OCFP were used as partitioning tracers. Their dynamic K-values were 0.48 and 1.67, respectively. EGBE and OCFP tracer analy­sis showed that the values calculated for pore volume were within 1% of the values from material balance for both tracers. The error for ROS was 6 and 11%, respectively, but this is solely because of the low residual oil saturation of this sandpack. It should be noted that the values calculated are still within 1% (oil-saturation units) of the material balance.

Laboratory results clearly illustrate that a favorable mobility ratio makes analysis of partitioning tracer tests easier. There are fewer problems arising from multiphase flow because the partitioning tracer does not see significant saturation changes. For partitioning tracer tests with significant oil mobilization and, thus, changes in swept volume, the modified RTDA method detailed in the complete paper provides accurate estimates of ROS. The authors have validated the approach by testing the modified RTDA method with synthetic simulated data as well as carefully controlled laboratory experiments.

An approach to reduce errors from PITTs is described in the complete paper.

Conclusions

The authors present a modified RTDA approach to calculate ROS accurately from PITTs when mobile oil is present. Given that reducing error in PITTs requires smooth fits of measured data, a modified log-normal approach to fitting both conservative and partitioning tracer data is presented. The partitioning tracer data are predictive when the conservative tracer is matched and estimates of ROS for each flow path and the tracer partition coefficient are ­entered.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 179655, “Use of Partitioning Tracers To Estimate Oil-Saturation Distribution in Heterogeneous Reservoirs,” by R.M. Dean, D.L. Walker, V. Dwarakanath, T. Malik, and K. Spilker, Chevron, prepared for the 2016 SPE Improved Oil Recovery Conference, Tulsa, 11–13 April. The paper has not been peer reviewed.