Which core Sw measurement is correct?
Andy Brickell
Senior Petrophysicist | Upstream Oil and Gas Industry | Geothermal | Subject Matter Expert | Formation Evaluation | Logging | Static and Dynamic Core Analysis
When cores are cut in oil-based mud, petrophysicists often ask for Dean-Stark measurements of water and oil saturation. If the cores were cut high in the hydrocarbon column, this measurement is expected to give a good estimate of irreducible Sw. The data can then be used to validate standard resistivity log-derived Sw calculations.
Subsequent testing on core plugs may include capillary pressure, formation resistivity index, and relative permeability, which all require the core plugs to be desaturated from Sw=100% to irreducible Sw, or as close as can be obtained. This is done either with a porous plate or in a centrifuge.
While the two different measurements of Sw are expected to give comparable results, a recent review of 15 datasets from cores cut in good quality sandstone reservoirs found that the Dean-Stark Sw was often (but not always) 5-10 saturation units higher than the minimum Sw from porous plate desaturation.
When this discrepancy is present, petrophysicists and reservoir engineers are presented with a significant problem - which Sw is correct?
- Petrophysicists prefer to use Dean-Stark data to validate log-derived Sw curves, because it is part of the routine core analysis, and so is available on a foot-by-foot depth basis. In addition, Dean-Stark data often agree very well with log-derived Sw, and irreducible Sw estimated from NMR logs However, they use data from Porous Plate tests to determine the Archie saturation exponent, which is a parameter in the log-derived Sw calculation. Porous Plate data also provides capillary pressure curves, which are used to define the saturation/height model.
- Reservoir Engineers use relative permeability tests to determine critical two-phase flow parameters. These floods are generally performed on core plugs that have been prepared with either the porous plate or centrifuge methods, and so may have lower Sw than that from the Dean-Stark data used for log validation.
In discussion with colleagues, four viable explanations for the discrepancy were identified:
· Invasion of the core by the water phase of the mud during drilling and recovery. This would increase the volume of water in the cores, and lead to Dean-Stark Sw values that are too high.
· Incomplete cleaning of the core plugs prior to Porous Plate or centrifuge testing. Invasion by the oil phase of the mud is known to make cores more oil-wet, so thorough cleaning is needed to remove this effect. If the crude is asphaltenic, recovering the cores to surface conditions will lead to deposition of solid hydrocarbon in the pore space, also leading to more oil-wet rock. Oil-wet rock has been shown to yield lower Sw values on desaturation than water-wet rock. Dean-Stark data are not affected and should be representative of the reservoir.
· Damage to fibrous clay structures within the pore space by cleaning and drying. This has been reported in the literature (eg, Luffel et al., SPE 1983), and was also observed in one of the data sets reviewed for this study. This artificially increases the permeability of the rock, leading to lower Sw values after porous plate or centrifuge desaturation. Dean-Stark data is not affected.
· Saturation hysteresis resulting from multiple charging events. Published experimental studies (Masalmeh, SCA 2001-23) have shown that secondary drainage following wettability alteration leads to higher Sw values. Standard porous plate and centrifuge tests represent a primary drainage cycle, so they much give a lower Sw than Dean-Stark data. In this case, both measurements are correct but the Dean-Stark data is more representative of reservoir saturations.
How do we decide which data set to use? Here are some ideas -
- Invasion by the water phase. Wireline logs run after coring may be able to detect this, because the shallow reading resistivity logs will be lower than the deep reading logs. However, in oil-based mud the available logs do not provide very shallow readings. No convincing evidence of conductive invasion was seen in the data we studied.
- Chemical and radioactive tracers. These can be added to either the oil or water phase of the mud. Tracers were used in some of the data sets we examined, and they showed evidence of oil phase invasion, but did not detect any water phase invasion.
- Inefficient cleaning leaving the core oil-wet. Some SCA tests are wettability-sensitive, and may show oil-wet characteristics in cores that have not been fully cleaned. This includes relative permeability, saturation exponent and Amott/USBM wettability. The data set we reviewed has several examples of this behaviour. In one data set, wettability was measured after cleaning, and the cores were found to be partially oil-wet. Additional cleaning made them more water-wet.
- Estimation of Formation Volume Factor (FVF) from Dean-Stark data (see URTEC-2461642-MS, Capsan and Sanchez-Ramirez). Dean-Stark data provides both water and oil saturation, and these data can be used to estimate FVF. If the FVF from this method is significantly lower than the known FVF of the reservoir crude (determined from PVT tests), it indicates invasion by the oil phase of the mud (which has zero GOR). This was commonly seen in the subject data sets.
- SEM examination before and after cleaning and drying. If delicate, pore blocking/bridging clay fibres are present, they can be identified in SEM images. Images after cleaning and drying may show damage to these structures, which will lead to permeability enhancement. This was demonstrated to be happening in one of our data sets.
- Basin modelling. This may show that multiple charging events occurred. This has not been done for the fields we studied.
So what did we decide? More work is needed, but in the author's opinion, the most likely explanation is that the cores were not properly cleaned, so the Dean-Stark Sw is more representative of reservoir water saturation than the porous plate and centrifuge data.
My thanks to Jorge Sanchez, Navid Rafatian and Matt Honarpour for their insights and comments.
Operations Manager (Country Manager)
10 个月Thank you for the very useful information on this forever challenging topic. I would like to contribute on OBM invasion into cores and its effect on Sw from DS. Most OBMs used recently have some saline water at roughly 30% and are quite stable in emulsion, provided that this emulsion stability is not disturbed by water present in the near wellbore zone. When the emulsion stability is disturbed, some of the oil and water phases in the OBM may be free, penetrate into cores seperately and may impact Sw/Swirr measurements by DS. This phenonema actually results in overestimation in Sw due to water phase invasion of the OBM, which may not be easy to detect even by the use of tracers, as mentioned by many colleauges. We proved this in a quite simple mixing experiment. In short, it should not be assumed that the OBM is always stable. If there is some OBM and/or OBM filtrate invasion, it is not only the oil phase, but also the water phase in the OBM. If we assume only oil phase penetrates into cores, then we should also assume that this oil phase has somewhat seperated from the water phase, which may only happen if emulsion stability is destroyed! Therefore, OBM invasion correction should be applied to Sw obtained from DS.
Petrophysics Manager
3 年Andy, Thanks so very much for sharing the insights and still prevalent challenges we face in data validation and integration. I just wanted to offer food for thought that echoes previous comments by colleagues as well as a possible innovative alternative to consider on the project. My team and I are executing core analysis on unconventional and conventional plugs via an integrated laboratory work flow. The methodology involves accounting for fluid loss in the laboratory measurement environment. In particular we are breaking down the extraction approach to samples by executing a thermally based extraction process in a specially designed closed retort setup first. This allows for the investigation and quantification of the most efficiently mobile fluids to be extracted first and compared against in situ NMR logs and produced fluid volumes. We then extract by chemical exposure, such as toluene, and quantify volumetrics by understanding any change related to asphaltene, waxy crude, or bitumen presence. When we combine that approach to measuring the PHIT and SW on samples, we increase our ability to break down the pore systems and what they are filled with and how they are behaving when interacting. Something to think about. ? ? ? ? ? ? ? ? ?*additionally we have developed a way to take the BVO and BVW measured in the lab and go backwards to account for fluid loss in investigated cores utilizing produced fluids and investigating the laboratory extracted fluids via HRGC analysis. The comparison and understanding is powerful to then apply corrections to calibrated petrophysical models and to close the gap in discrepancies where the engineering rel. perm curves and modeled recovery factors may not agree with the assessed volumetrics. It’d be really interesting to compliment the study you present with the approach to see what, if anything, the additional results might show or validate or introduce? ? ?Last note as a thought is that we also physically measure both the extracted water and hydrocarbon phases in this alternative measurement known as TruSat. So rather than only measuring the extracted water and then having to assume remaining gas versus oil fractions based on density assumptions, let’s try and physically measure the oil and water in the sample so there are no assumptions? Then to also cross validate with additional tools? Great, great discussion and forum! Look forward to additional feedback and thoughts.?
Petroleum Reservoir Engineer
3 年Thanks for that easy to understand assessment
Sr. Petrophysicist at Home
3 年Andy, I’m not expert on this but does the lab provide the Sw uncertainty of either the Dean Stark or Porous Plate method? I knew the burette used by the lab I used to work with has an uncertainty reading of 0.05 cc. It matters in low porosity rocks but it should be less in porous rocks. However, it’s good to know the Sw and porosity uncertainty from any core measurement method. I’m not sure how possible it is in your situation but having oil saturation measurement from C/O logging or water saturation from dielectric or NMR might give alternates looking at saturation too. I don’t have much experience looking Sw from Karl Fischer method but I heard that its lowest water detection limit is quite good so it will be great if someone can share the light here. I’ve seen less water loses when the sample were taken in the field than in the lab. Also, less water loses in well-preserved sidewall cores than in whole cores. I don’t have experience in porous plate method or after centrifuged Sw but I wonder why many Petrophysicists chose Dean Stark despite of higher Sw. I thought I might have chosen the lowest Sw as irreducible Sw. Unless the lowest Sw requires impossible a,m,n to achieve. Anyway, thanks for the post, Andy!
Asset Growth Thru Uncertainty Quantification
3 年Andy, may I propose an alternative way of verification. I do understand that you’re comparing core to core, but it’s worth looking at measurements made in situ as well when looking for a reference. I would steer clear of any resistivity based calculated Sw for the reasons every petrophysicist is aware of and look at NMR Swirr. Best if it’s LWD as you’ll be indeed looking at the log response to the rock with native fluid properties. Wireline is also usable unless it’s the last run in the Wireline program with large overbalance and lots of invasion and potential alteration has already happened. There are then two scenarios in clastics: (1) massive sands with vertical variation in BVI (well, geologists will confirm any rock has some grain size related stratification) or (2) truly laminated sands where you’d be looking for the double peak on T2 distribution. Usually it’s pretty straightforward to pick the cutoff. This is assuming you’re above the transition zone, no heavy oil/tar in sands, good hole quality. In my experience, comparison of NMR Bound Water volume with Dean-Stark Sw*Phie_core and any other BVW core-based source, provides insights into the reasons why there are discrepancies.