Oleum Khaos

Jeremy’s primary technical expertise is Petrophysical interpretation from a geological foundation. This foundation can be significant because, without knowledge of depositional systems, bed-forms and lithology; understanding what logging tools are providing and what valid information can be gleaned from them, is the first critical step in evaluation.

Jeremy was fortunate to work with Dr Paul F Worthington who sadly died in May 2020. One of Dr Worthington’s many strong points is the number of high quality papers he’s published, relating to pragmatic petrophysical evaluation.

The following papers, published by Dr Worthington, are highly recommended:

• Conjunctive Interpretation of Core & Log Data Through Association of the Effective & Total Porosity Models, 1998
• The Evolution of Shaly-Sand Concepts in Reservoir Evaluation, 1985
• Scale Effects on the Application of Saturation-Height Functions to Reservoir Petro-facies Units, 2001
• Application of Saturation-Height Functions in Integrated Reservoir Description, 2002
• The Effect of Scale on the Petrophysical Estimation of Intergranular Permeability, 2004
• The Role of Cut-Offs in Integrated Reservoir Studies, 2005
• Optimizing the Value of Reservoir Simulation Through Quality-Assured Initialization, 2014

All Petrophysicists plus Geoscientists and Reservoir Engineers can benefit significantly from the topics covered in the papers above.

A Petrophysicist is not defined by the software he or she can use. It doesn’t matter whether an Operator uses Techlog™, Interactive Petrophysics™, Geolog™ or any other package. None of these applications will give rise to better results because it is the expertise of the Petrophysicist who undertakes proper data quality assurance and then cerebral, iterative analysis. The software is simply a tool; however, many people now give the software far too much weighting. If you can drive Techlog™ or Petrel™ you are now an experienced Petrophysicist or Geologist / Modeller. I beg to differ.

Jeremy can provide Operators with the expertise to achieve the lowest uncertainty data, including whole core acquisition, handling and analysis, encompassing conventional and special core analysis programmes.

Understanding uncertainty is fundamental. Many people consider stochastic methods provide the best way to understand uncertainty. In theory this is the case; however, in practice we consistently under-estimate the range of uncertainty in input parameters, assume probability density functions which may be incorrect, fail to assign, or assign, incorrect dependencies. This leads to a range of results which staff and Management tend to believe is accurate. It generally isn’t. Plus, the range of results may not be realistic in terms of what each scenario means. I think Shell started to realise their stochastic reservoir model-derived in-place volume estimates were flawed because many of the reservoir models were simply untenable.

A further nail in the coffin of stochastically derived estimates is the significant difficulty in auditing them. If it can’t be audited then really, it shouldn’t be used. Don’t get me started on auditing 3D reservoir models!

Jeremy’s preferred approach to petrophysical uncertainty is not to derive a single set of reservoir properties. In theory the objective of all technical evaluations is to derive the Most Likely properties, not the Mean or P50. In anything other than a uniform clean sand, estimating in isolation, the Most Likely properties, tends to be difficult, if not impossible. However, this is what the majority of Petrophysicists try to do, on a day-to-day basis.

There is a more rigorous way to describe some of the uncertainty. The validated raw data provides low and high values of each input parameter. The key words here are valid data. Thorough data quality assurance is absolutely essential before any evaluation is undertaken, otherwise invalid data gets adopted. Once the valid minima & maxima are known these provide boundaries for uncertainty scenarios. They are unlikely to describe the full range of uncertainty because the data probably doesn’t reflect the total population. But they are, at least, auditable data-driven parameters.

Two petrophysical scenarios can easily be constructed using the Low & High input properties of the porosity model. This is easy and very auditable. The same approach can be used for Shale Volume, mineral proportions, permeability and water saturation. Now I hear you say; using all the Low (or High) estimates together is unrealistic; which I agree.

The probability of all the low parameters combining in nature is very low. Which is actually very useful because we can adopt these results as sort of equivalent to a P01 in the stochastic world, suggesting to Management that it’s unlikely to be less than this, based on the data we have to work with. So, the Low (nominal P01) and High (nominal P99) reservoir properties can be derived in a straight-forward manner.

How does the Most-Likely estimate get derived? Therein is the problem. If the data does not provide an indication of a uni-modal value, then it cannot be derived. In this situation the fall-back is to derive a Mid Case between the Low and High parameters and ensure Management understand this is not the Most-Likely scenario.

There are generally two explanations for inaccurate Petrophysical interpretations; Firstly poor data quality assurance and secondly; the lack of partitioning of different petro-facies. In other words the Petrophysicist fails to separate out the different petro-facies and tries to characterise a mixture of log & core responses with one category. The classic core-plug poro-perm cross-plot is a common expression of this. I don’t recommend anyone uses such a cross-plot without classifying all of the core plugs first. In fact don’t try to derive poro-perm correlations using such cross-plots, at all.

Why do we have to deal with such uncertainty? The root cause is a misunderstanding, by Operators, of the value of information.

Operators perceive that obtaining whole core is too expensive and that it’s more cost-effective to use other means such as logs. This is actually a fallacy.

Consider the levels of uncertainty in reservoir & fluid properties when estimated from (particularly uncalibrated) logs versus, for example, whole core. In theory, if the whole core is obtained correctly and the correct core plug handling and analyses are undertaken; then the sedimentological, lithological and reservoir property, plus fluid uncertainty, is minimal or even close to zero.

In other words, we know the answer at the well.

Without whole core of the reservoir, we do not know the answer at the well, we typically make a guess, or with a suitably experienced Petrophysicist; a range of guesses.

The moral of the story is spend the money to get the answer with minimal uncertainty because this allows confident investment or divestment decisions. Get whole core with a minimal suite of logs, perhaps just basic LWD, plus an RFT to confirm free-water levels.

Introduction

As with most things, when you learn something you tend to automatically rely on established approaches and workflows. You’re taught, or read, how to do things and tend to accept these are the best practices and therefore should be utilised. This isn’t always the case.

Petrophysical evaluation is a case in point, where there are a few legacies which, when scrutinised, do not have very solid technical foundations. The Effective Porosity System (EPS) is one of these; however, the vast majority of evaluations of conventional clastic reservoirs utilises this particular system.

As most experienced Petrophysicists know, the first fundamental flaw in the EPS is the estimate of Shale Volume (VSH), from which Effective Porosity (PHIE) and Effective Water Saturation (SWE) are subsequently derived. If (as is almost always the case) VSH is inaccurate; then PHIE and SWE are also inaccurate. This paper is not going to expand on, or discuss in great detail, the technical pros and cons of the EPS; this can be found in the literature. The Total Porosity System (TPS) is technically more robust than the EPS and would be preferred so long as the necessary data has been acquired.

This paper summarises a workflow that derives technically defensible Low-Mid-High estimates of VSH, which can then feed into Low-Mid-High estimates of PHIE, SWE and permeability. The workflows for PHIE, SWE and permeability will be addressed in separate papers; however, in the meantime the experienced Petrophysicist should be able to adapt the approach, described below, to derive a range of PHIE, SWE and permeability estimates.

This “EPS approach” can be adapted to derive three sets of results via the TPS.

The vast majority of Petrophysicists’ “clients” tend not to specify a requirement for a range of petrophysical results. However, all subsurface technical disciplines should, by default, be deriving a technically substantiated range of estimates, in order to account for uncertainty. Sadly, this is not routinely undertaken. Of course, to do this requires more time and more understanding. We should not be producing single, so-called “Base Case” or “Best Technical Estimate” results, without properly describing the upside & downside scenarios.

If your client asks you for a petrophysical evaluation without specifying the porosity system, or the scenarios required; the likelihood is that the client doesn’t understand petrophysics or uncertainty; or maybe they do, but their prepared to wing it, which isn’t really top decile professionalism. In this situation, it’s incumbent on the Petrophysicist to discuss the options with the client (and, if necessary, diplomatically with the client’s Technical Manager); to ensure there’s full understanding of the best porosity system(s) to use, their implications and the need for a range of estimates. If the client (and TM) choose not to follow the Petrophysicist’s recommendations, ensure you’ve made your views clear (for example in an email) and summarise what you’ve been asked to undertake, together with any limitations these give rise to.

Paul Worthington has an excellent summary of using the TPS and EPS together in order to derive what he calls “ground-truthed” results (Conjunctive Interpretation of Core & Log Data Through Association of the Effective & Total Porosity Models, 1998). His approach is highly recommended; however, a range of estimates, is still required.

Shale Volume Estimation

As mentioned above, the estimation of a VSH log is fraught with issues plus, whichever VSH model is adopted, there’s significant uncertainty around this estimate. This uncertainty is “inherited” into the estimate of PHIE and then again into SWE, because VSH and PHIE are inputs to SWE. So, you can see why undertaking a single EPS evaluation is a risky business.

In order to mitigate the issues with estimating a VSH log; below is a relatively simple workflow designed to derive Low-Mid-High estimates of VSH using the Gamma-Ray (GR) as the input log.

This approach is not restricted to solely VSH from the GR; the approach can be adapted to other input logs used to estimate VSH, as well as the input parameters to porosity, permeability and saturation estimates.

The key aspect is identifying realistic Low-Mid-High input values for the Shale Volume model(s), plus Porosity & Saturation model(s).

VSH-GR Workflow

1. Ensure all borehole environmental corrections have been undertaken and bad log & core data either removed or replaced (I prefer to null bad data rather than replace it).

2. The available GR data controls the best technical approach. The best quality wireline GR data is acquired at the slowest logging speed, which tends to be with the (pad-based) bulk density tool. So, identify the best GR data over the Zone Of Interest (ZOI). If Spectral GR is available, investigate the CGR and SGR over the ZOI. The Thorium or Potassium logs may be useful, for example, if there are Feldspars, or not (see Crain for more details: https://www.spec2000.net/11-vshgr.htm and leave a donation).

3. For the reasons above, ideally evaluate the ZOI on a well by well basis, but if this is not feasible, for example, due to the large number of wells; generate normalised input log(s). Typically the GR database will be a mixture of standard wireline (acquired at different logging speeds), spectral GR (often over limited intervals) plus LWD data. These data can be quite disparate and not really suitable for normalisation or multi-well evaluation; hence keeping evaluation to a well by well basis is recommended.

4. In conjunction with other logs, core analyses, sidewall core information, cuttings, formation pressure plus DST data; interrogate the GR response over the ZOI to understand what the radio-activity represents. As part of this, do other quick-look VSH estimates using logs other than the GR.

5. Once you have selected the ZOI(s) generate a histogram and cumulative frequency plot of the best GR log, over the ZOI, such as the one shown below:

   

6. Decide on what cumulative frequency percentiles to use to describe the “Clean,” low GR intervals and the “Shaly”, high GR intervals. Illustrative values, taken from the distribution above, are shown in the table below:

   Cumulative %	GR Value	Class
   P05	20	“Clean”
   P10	22	“Clean”
   P15	24	“Clean”
   P85	81	“Shaly”
   P90	92	“Shaly”
   P95	125	“Shaly”

7. The selection of the ZOI, plus the cumulative percentage values, has a direct impact on the resulting estimates of VSH; so some iteration is recommended before selecting the final parameters. This is discussed in more detail later on.

8. For each ZOI derive raw VSH GR-Hi, VSH GR-Mid and VSH GR-Lo, using the equations below:

   VSH GR Hi =

   (GR – GR _P05)

   (GR _P85 – GR _P05)

   VSH GR Mid =

   (GR – GR _P10)

   (GR _P90 – GR _P10)

   VSH GR Lo =

   (GR – GR _P15)

   (GR _P95 – GR _P15)

9. Create final constrained VSH GR Hi, Mid and Lo logs, from the raw logs.

10. Plot the constrained VSH GR logs in a layout together with the input log(s) and check for suitability. It may be necessary to iterate on the ZOI and/or the “Clean” plus “Shaly” end-points. The three VSH GR logs, shown below, were derived from the GR data in the histogram shown above. The Net-To-Gross (NTG) flags and NTG percentages, based on application of VSH GR < 50%, are also shown. The Gross and “Net” interval thicknesses, for each VSH GR scenario, after application of the VSH GR < 50% cut-off are also summarised:

    

    Illustration of VSH GR-Hi, -Mid & -Lo Estimates Together With “Net” Flags

     

     

    	VSH GR-HI	VSH GR-MID	VSH GR-LO
    Gross Interval (M)	106.38	106.38	106.38
    “Net” Interval (M)	81.24	84.13	88.92
    NTG (%)	76%	79%	84%

    Table of Gross, “Net” & NTG After Applying VSH<50%

Summary

The range in “Net” thickness from 81.24 m to 88.92 m will tend to outweigh the uncertainty range attributed to porosity plus, where valid calibrating data for saturations exist and vertical fluid distribution(s) are well-constrained; NTG uncertainty will outweigh the saturation uncertainty range.

But, if fluid contacts and/or fluid types are poorly constrained, then saturations and/or the definition of Net Pay, can have significant uncertainty ranges; potentially a similar order of magnitude as NTG. This is why it’s essential to incorporate fluid contact uncertainty into all petrophysical evaluations.

The example illustrated above highlights how sensitive the inputs to the VSH GR estimate are, in terms of the resulting NTG. Looking at the input GR log, you could make an argument that the ZOI could be changed; introducing three zones for example: 3080-3095m, 3095-3155m and 3155-3190m. If you have other data to support such a zonal breakdown, then subdivision should be done.

This also demonstrates how inappropriate it is to undertake a single deterministic estimate.

With a range of approximately 8m of “Net” reservoir resulting from the same input GR log but different “Clean” and “Shaly” end-points; it’s essential to derive a range of VSH estimates.

Discussion

I am not recommending the workflow above is the be-all and end-all of EPS petrophysics; it’s not. Worthington’s 1998 paper, referenced earlier, provides an excellent all-encompassing EPS and TPS approach, albeit leading to a single deterministic estimate in each system.

I am recommending that all petrophysics moves away from the single estimate scenario, to an industry-wide default of deriving: Low, Mid (or Most-Likely, if this can be achieved) and High estimates of lithology, porosity, permeability and water saturation. All technical professionals agree with this, but then don’t tend to implement it!

The workflow, described above, provides a good starting point for VSH. The approach can be adapted and expanded to estimate three scenarios of PHIE and SWE, linking with the three estimates of VSH. The approach can be adapted to estimate three cases of VSH from other input logs such as the SP, RHOB, NPHI, etc.

As I’ve discussed previously, I recommend you combine, deterministically, all the pessimistic estimates of VSH, PHIE and SWE to provide what may be close to a P99 scenario; where there’s a 99% chance the actual values are larger. This pessimistic scenario is very useful in terms of estimating a “downside” case, which geomodellers, reservoir engineers and Subsurface Managers really ought to be considering before investing millions of Dollars in appraisal or development.

It’s recognised that combining all pessimistic values is unlikely to occur in reality; however, this is deliberate in order to extend the range beyond the incomplete data actually used to establish the Low, Mid and High scenarios.

The same argument applies to the combination of all optimistic VSH, PHIE and SWE parameters to provide what may be close to a P01 scenario, where there’s a 99% chance the actual values are smaller. This can be considered an “upside” case, which should also be modelled for in-place and recoverable volumes.

Lastly, the long legacy of the EPS and its adoption as the petrophysical model of choice, probably means the industry will continue using it, even though it is technically unsupportable!

Instead of following the petrophysical herd, think about alternative ways of describing potential reservoir intervals, the inter-connected pore volume and permeability within the reservoir intervals, plus the fluid saturations.

Is a single VSH log a good enough parameter to start this process?

No.

Is a range of VSH estimates good enough?

Not really, but it’s significantly better than a single estimate.

Can I estimate (connected) porosity and permeability with reasonable accuracy, without using a VSH?

If so, then you can define reservoir and non-reservoir using these parameters and VSH becomes redundant; wonderful!

The TPS is technically more robust than the EPS; so planning data acquisition for the TPS is highly recommended, whereas reliance on the EPS, is not recommended.

You are welcome to contact me for additional clarification or to share your own applications, or adaptations, of this approach.

© Oleum Khaos Limited (July 2017)