Home‎ > ‎

Briefing note

Field data indicative of a general involvement of geomechanics in reservoir behaviour

 

The data are of 2 completely independent types:

1)     speeds of breakthrough in different directions of injected fluids to producer wells.  Relative speeds of flood front progress are conventionally considered to be determined by the sedimentary, stratigraphic or faulting architecture of the reservoir; possibly natural fracture trends in those reservoirs deemed to be ‘naturally fractured’: those are the geological factors that are commonly used to populate the heterogeneous properties in reservoir models.  However, the data tell us that modern-day stress state is a determining factor in most of the cases: the directions ‘preferred’ by injected fluid are those close to the maximum horizontal stress state (Shmax).  As seen in the figure this preference holds for ostensibly ‘unfractured’ reservoirs as well as for ‘fractured’ ones:


Plots from Heffer & Lean, 1993). Each plot shows how many fields showed ‘preferred’ flooding directions in each sector, all rotated to a common orientation of Shmax up and down the page.

 

2)     Correlations in the fluctuations of flowrates at pairs of wells.  Flowrates out of producers or into injectors are rarely constant in time: fluctuations are often a significant proportion of the average.  Causes of these fluctuations can be induced by the operator of course: e.g.deliberate choke changes (say, to balance Voidage Replacement Ratios); downhole well conditions change; well stimulations; changes in surface facility conditions (e.g. separator pressures); conducting flow tests etc.  But, some of the fluctuations are the result of signals passing from well to well  through the subsurface.  This would conventionally be labelled ‘well interference’ and usually ascribed to hydraulic reservoir flow alone. However, a series of projects has again found that geomechanics is also instrumental in the correlations.  The following figure shows the aggregation of correlations in flowrate fluctuations from over 500,000 injector-producer well pairs from 8 different field areas of flooding operations.  In each sector relative to a common direction of Shmax up and down the figure, the curves show the average strength of correlation coefficients for well pairs whose separation is in that direction.  The green curve is for raw flowrate data (including all frequency of fluctuations); the red plot is for flowrate data that has been filtered to remove low frequency trends, leaving only the high frequency fluctuations for correlation.  Either curve shows that flowrate fluctuations correlate much more for well pairs aligned sub-parallel to Shmax.  In the case of raw data the average correlations are in fact negative along the orthogonal axis (Shmin).

(figure after Heffer et al, 1995)

A further recent development along this line has examined the rates of change of interwell correlations in flowrate fluctuations, to extract what has been termed ‘rate diffusivities’.  Application of this technique to 6 North Sea fields (2 ‘fractured’ reservoirs: Valhall + 1 other; and 4 ‘unfractured’ reservoirs: Scott, Gullfaks, Magnus, Miller) has resulted in the following aggregated plot:

The plot (from Heffer, 2012a) is of the average major diffusivity axis in each sector relative to the common direction of Shmax.  In other words, rate correlations diffuse in space most strongly in directions that are about 30o to Shmax

 A proposed explanation for the trends in both types of data (flooding directionalities and rate fluctuation correlations) is now available; it also ties in with manifold observations of shear-wave splitting from most formations reported by Stuart Crampin of The University of Edinburgh and the British Geological Survey and co-workers.   Crampin (1994) is a good starting point for those new to these concepts.  Essentially the interpretation of the characteristics of the observed shear-wave splitting involves the presence of compliant, stress-aligned (micro-)cracks, whose stress fields interfere constructively in shearing directions.  What is the ‘shearing direction’ for vertical cracks aligned with Shmax?  Why, about 30o to Shmax.  The figure opposite shows the tensile stress field around a central dilated crack: tensile stresses are greatest at directions about 30o to Shmax; the stresses are less tensile parallel to Shmax; the stresses are compressive sub-parallel to Shmin.  This pattern of tensile stresses has been invoked by several authors as the basis for the genesis of faults. 


So the concept is that crack dilations or compressions induced locally by rate changes at one well propagate via interacting (micro-)cracks with a pattern that follows the pattern for a single crack over large distances, giving rise to corresponding rate changes at remote well locations.  It is possible that the propagation mechanism is mainly an elastic one, even over long distances (Heffer, in preparation); the density required for this is a percolation threshold, equivalent to a 3D crack density ~0.035.  That value compares well with crack densities interpreted by shear-wave splitting: the range is only 0.015 to 0.045 Crampin (1994).

How does this tie in with the flooding directionality trends?   Let’s see the consequences for a single injector well around which (micro-)cracks, all aligned with Shmax, are dilated (opened) with the pattern similar to that of a single crack (see green pattern on the left below).  The isobars of the pressure field for constant injection would take the form shown in coloured contours on the right-hand figure (calculated with a minimum path technique and with a moderate matrix permeability assumed).

    Pattern of dilations of microcracks around injector               Pressure contours corresponding to microcrack pattern

                                     

 





                                                                                                                                                                

Those isobars, which represent approximate progression of a floodfront from the injector, resemble the preferred flooding breakthrough directionalities for reservoirs not naturally fractured from the very first figure above (as shown by superimposition).  Similarly, if the matrix permeability is reduced relative to the crack permeability, the isobars trend more parallel to Shmax, resembling the pattern of preferred flood directionalities observed in naturally fractured reservoirs:

So, we have consistency between the following 3 field observations, all tied with a common mechanism, viz. compliance of stress-aligned microcracks at a percolation density:

1)     flooding directionalities

2)     correlations of rate fluctuations

3)     shear-wave splitting characteristics

 

In other words, it looks as if geomechanics is a common mechanism in many, if not most, reservoirs.  But it is not taken into account in most reservoir models, implying that it is a risk that should be evaluated in most cases.  What is the value associated with the risk?  Potentially large.  It has long been known that anisotropic permeabilities can have a large effect on recovery factors, at least until fluid breakthrough.  The following figure summarises a finding from 50 years ago:


Reference: Caudle, BH & Lonaric, IG ‘Oil Recovery in Five-Spot Pilot Floods’, Trans AIME (1960) v219, 132-136Results reproduced in ‘The Reservoir Engineering Aspects of Waterflooding’ ed. FF Craig, SPE Monograph vol. 3 (1971) Figs 5.14 & 5.15]

Getting the alignment of injectors and producers relative to permeability axes therefore makes a big difference to oil recovery to breakthrough; and some difference up to high water-cuts.   To this potential additional recovery can be added the saving in cost of treatment for produced injection fluid and the well configuration.

 

Stress state orientation can be interpreted borehole image logs of course, as one piece of information towards enabling a better well configuration.  But what about heterogeneities in the stress state and/or (micro-)crack characteristics?  Here’s where the technique of analysing correlations in flowrates can be used for individual fields: they enable a form of tomogram of the reservoir between wells. Principal components and rate diffusivity axes extracted from the correlations can highlight the faults and fracture corridors that are prominent in passing geomechanical signals between wells.  Here is an example from a carbonate reservoir, showing the 4 most important principal components (or modes) of the rate correlations:


The maps of important fracture trends across the field extracted from rate correlations agree well with independent observations of structural lineaments, porosity and fracture trends from log and core data.  Geomechanical changes had never been considered as influential in this field before: these findings provide a strong indication that they are acting here, as they probably are in most reservoirs undergoing flooding (and even under natural depletion as implied by some field data).

ċ
filelist.xml
(1k)
Kes Heffer,
20 Jun 2013, 08:44
ċ
header.htm
(2k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image002.emz
(9k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image004.emz
(4k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image006.emz
(2k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image008.emz
(19k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image012.emz
(2k)
Kes Heffer,
20 Jun 2013, 08:44
ċ
image013.emz
(56k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image015.emz
(58k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image017.emz
(85k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image019.emz
(15k)
Kes Heffer,
20 Jun 2013, 08:44
ą
Kes Heffer,
20 Jun 2013, 08:44
ċ
image021.emz
(250k)
Kes Heffer,
20 Jun 2013, 08:45
ą
Kes Heffer,
20 Jun 2013, 08:43
Comments