SPE General Meeting

Topic:        Approaches to Shale Gas Log Evaluation

Speaker:    Robert Cluff,  The Discovery Group Inc.

Date:        Wednesday - December 21, 2011

Time:        11:30 am

 Place: Denver Athletic Club

 Cost: $25.00 ($30.00 at the door, if space is available)
Reservation Deadline: Noon, Monday December 19, 2011
Deadline for reservations has been changed to Noon on the Monday before the General Meetings. or Call 303-620-9080 for reservations.

Abstract:

Petrophysicists faced with the problem of evaluating shales for hydrocarbon storage and flow have adopted a broad range of possible solutions. No consensus opinion has been reached by the community at large, and most of us are adamant in defense of our particular workflows and assumptions. Nearly all methods utilize core data at some level as the “ground truth” for testing or calibrating models for porosity, total organic carbon, mineral contents, hydrocarbon saturation, and in the most optimistic cases permeability.

Broadly speaking, most of the approaches used can be divided into three major classes:

1. Global stochastic models: these utilize so-called stochastic solvers that forward model a set of mineral and fluid properties to predict the expected log response, and then via an error minimization technique adjust the component volumes until a match with the actual logs is attained. There are numerous proprietary software products that can accomplish this, each using a somewhat different approach to the inversion. The key question faced by an interpreter in using these models is which components should be included, and what are the precise petrophysical properties of those components for the inversion? This is especially problematic in the case of kerogen and many clay minerals, both of which are a significant part of the shale problem.

2. Global deterministic models: these are equations or models based on large datasets, often play or basin specific, to predict one or more critical parameters of interest. Examples include the Schmoker equations for TOC from the density and gamma ray logs; the deltaLogR model for TOC; and logging vendor models to determine lithology from geochemical log elemental abundances. These models sometimes work well, sometimes do not, but almost always require local “tweaks” to apply to a specific problem.

3. Local deterministic models: these are locally determined, by area, play, or even at the well level, for various properties calibrated to core or other data. Examples include linear regression or multiple regression models for mineralogy and TOC; play specific correlations for water saturation from Vclay or porosity; etc. These models have the advantage of being immediately applicable to the problem at hand, but may or may not reasonably extend beyond the calibration area or dataset, and almost always require local calibration data (i.e. cores) or a direct analogy (“if it looks and acts just like the Barnett, then……”).

Many of the approaches that have been proposed suffer from complete lack of transparency (they are “black boxes”) or from lack of transportability (they require specific software or proprietary knowledge
to apply them). Many others suffer from requirements for a particular log or specific logging suite, such that if one or more critical curves are unavailable or else the “wrong” vendor logged the well, the model cannot be applied. This may or may not be important to an operator evaluating their own wells, but is most certainly a problem for regional play evaluations using public domain and non-operated well data.
We use elements of all three approaches in our work depending on the data available and the requirements of the specific project, but it is not at all clear at this time which path yields the best solutions from a theoretical much less the practical perspective.

Biography:

Robert Cluff is a geologist/petrophysicist with 35 years petroleum exploration and development experience. Bob’s research interests include the integration of geology with petrophysics and the evaluation of non-conventional reservoirs.  He received his BS in Geology from the U. California – Riverside, an MS in Geology from the U. Wisconsin – Madison,  and also holds a BA in Mathematics from Metropolitan State College Denver.  Bob co-founded The Discovery Group in 1987.