Basic HTML Version
OH – ET – VA - LL: Analysis of Dynamic Data in Shale Gas Reservoirs – Part 1 – Version 2 (December 2010)
p 1/24
The Analysis of Dynamic Data
in Shale Gas Reservoirs – Part 1
(Version 2) Olivier Houzé – Eric Tauzin
Vincent Artus – Leif Larsen
0B
The original version of the “Part 1” document was dated July 2010. It only addresses
theoretical considerations and simulations, with no real example. This update is simultaneous
to the release of Part 2, where a real data set is used and a workflow is suggested.
1 - Introduction
Shale Gas (or unconventional gas to be a little more generic) is the latest buzzword in town.
For KAPPA, the issue is not whether shale gas is economically viable or not. Operating
Companies say yes, and we do not have all the elements to express a qualified opinion.
However, if some of the statements are based on reserves calculated from straight line
analyses, operating companies may be up to some bad surprises in the years to come.
Well whatever, what we are interested at KAPPA is the analysis, modelling and forecasting of
dynamic data, mainly production and pressure, in shale gas reservoirs. We tried to identify the
specific problems, the limitations of the current methodology and models, and what can be
done to properly address this. In this document we are going to (respectfully) object to what
some of our fellow competitors, and their followers, do today. We are also going to be positive
and suggest what can be done.
If one looks at the process of producing shale gas, we can see four specifics:
Permeability is extremely low
As a result shale gas is generally produced using fractured horizontal wells, though
historically there has been flow through single fractures.
Gas diffusion is somewhat complexified by the desorption of gas from the shale
Reservoir also exhibits unconsolidation, i.e. the permeability reduces when pressure drops
To address dynamic data (PTA, PA, simulation) a few methods are offered in the industry:
“Early time” straight line analysis of linear flow
Sometimes dual straight line analysis of compound linear flow is considered
Late time material balance calculation on a restricted volume (SRV)
Analytical models
Numerical models using pseudopressures (but we will not use these here)
Numerical models simulating the real gas diffusion (Saphir NL, Topaze NL and Rubis)
Naturally, only the nonlinear numerical model, if properly implemented, will address exactly all
our hypotheses. Other methods will carry approximations, and the real question is whether the
errors linked to such or such method are acceptable or not for our production forecast.
Pressure Transient Analyses (Saphir) can be seen here and then, the time scale of a build-up,
when combined with the value of the diffusivity, lead to a very small volume of investigation.
History matching (Rubis) could be used for multi-well simulation for complex geometries.
However interferences between wells are rare, and today we are still fighting to discriminate
permeability from fracture length, so we are still far from modelling complex cases.
So, for any practical purpose the tool of choice today is still Production Analysis (Topaze NL).