Spe 173867 Ms

SPE-173867-MS
Downloaded from http://onepetro.org/SPEBERG/proceedings-pdf/15BERG/All-15BERG/SPE-173867-MS/1405255/spe-173867-ms.pdf/1 by China University of Petroleum Beijing user on 19 March 2024
Case Study: Dynamic Visualization of Miscibility for EOR Design and
Implications for Field Planning
John Godlewski, Schlumberger; Emily Wu, Arc Resources
Copyright 2015, Society of Petroleum Engineers
This paper was prepared for presentation at the SPE Bergen One Day Seminar held in Bergen, Norway, 22 April 2015.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
CO2 flooding is a well-known enhanced oil recovery method which is made attractive both by the
opportunity to recover significant additional oil and the opportunity to sequester CO2 at abandonment.
However, as the acquisition and compression of CO2 is costly, the flood must be carefully designed to
maximize recovery from the injected volumes. Typically, design is done with a plan to maintain reservoir
pressure above the minimum miscibility pressure (MMP), but this study investigates if a far more detailed
analysis of miscibility is warranted.
A pilot CO2 flood in Western Canada and its existing development plan are investigated. Instead of a
relying on the single MMP from slim tube tests, miscibility is calculated on cell-by-cell basis with a tuned
equation of state (EOS) and history matched reservoir model. A ‘distance’ to miscibility is calculated, in
pressure terms, by comparing the saturation pressure to the cell pressure throughout the flood. The result
is then visualized to illuminate possible improvements.
Because it takes in account changes in pressure, composition and temperature in both space and time,
this method may model important reservoir phenomena that are not considered with the slim tube method
- including channeling, buoyancy, heterogeneity, multiple gradients with depth, and cross-flow diffusion.
The calculation was applied to the field development plan to evaluate the effectiveness of the current
strategy. Even though the MMP conditions were nominally met, the new parameter highlights areas where
miscibility was not occurring and oil was being by-passed. The measure also shows areas where the CO2
concentrations were excessive, and so the injectant was underutilized in sweeping oil towards producers.
Based on these results, changes to the field development plan were proposed. Well plans and operating
constraints were altered to improve downhole mixing and miscibility, leading to improvement in predicted
oil recovery and economic measures such as capital expenditure, operating expenditure and net present
value. These results demonstrate that MMP can be overly simplistic in the design of miscible flooding
strategy.
The case-study presented adds to the database of past experience when designing EOR floods, while
providing a simple visualization parameter to aid engineers in understanding and optimizing miscible
recovery design. It is particularly useful for multiple contact floods, for fields with limited CO2
availability, and where in-situ heterogeneity, gradients with depth, and diffusion phenomena complicate
recovery.
2 SPE-173867-MS
Introduction
CO2 has been known to be an effective enhanced oil recovery (EOR) agent for many years, with the first
field tests on pure CO2 slugs being conducted in 1964 and carbonated water floods being carried out as
early as the 1950s (Izgec et al. 2005. API). More recently, interest in the field has increased due to the
opportunity to leave injected volumes downhole and offset other carbon emissions.
These factors mean that today some jurisdictions benefit from favourable royalty terms, carbon credits
or evaded taxes for operations under CO2 EOR, and so in a properly managed flood an operator may
benefit both from increased recovery and direct project support.
Technically, CO2 is also attractive as a reservoir injectant for a number of reasons. Most notably, under
sufficient pressure CO2 can reduce gas-oil surface tension to zero, thus inducing miscibility. This
condition may occur a couple of thousand psi lower than for other typical injection gases (Izgec et al.
2005), and can result in microscopic sweep efficiencies of more than 90%. However, ensuring that CO2
floods benefit from this effect is not always straightforward. At moderate pressures, miscibility may
develop in complicated ways and can revert to two phases depending on the CO2 concentration.
Development of Miscibility and Traditionally Minimum Miscibility Pressure (MMP)

If the reservoir pressure is very high, miscibility can occur immediately upon injection (first contact
miscible) and be independent of CO2 concentration. Much more typically, however, a single phase is
reached only after a number of downhole mixing steps, in a process known as multiple contact miscibility.
Figure 1 shows how the two processes lead to a single hydrocarbon phase on a ternary diagram (further
explained in Appendix A).
Figure 1—Ternary diagram, first contact and multiple contact miscibility
To mimic this processes in the laboratory a slim tube experiment is used. This is a long, 1-dimensional
displacement of reservoir oil by CO2 under controlled temperature and a series of pressures. The
minimum miscibility pressure (MMP) is determined by a plot of oil recovery versus experiment pressure
[Figure 2], and then the MMP is used as a downhole target in field planning to “ensure” miscibility.
SPE-173867-MS 3
Figure 2—Slim tube tests and the interpretation of MMP, Wu and Batycky
There are a number of problems with this methodology. The first is that this method is imprecise. A
small number of slim-tube experiments is carried out, usually just 4 (the minimum needed to draw 2
straight lines). Then, there is no firm definition for the determination of MMP. Miscibility is sometimes
interpreted to occur where there is a change in slope in the pressure vs recovery curve. Also commonly
applied is 90% recovery at 1.2 PV of CO2 injection, or sometimes 95%. As noted in an example by Wu
and Batycky [1990], the difference between the slope method and 90% method can lead to differences of
50 bar in the MMP [see Figure 2]. Using 95% would give an even larger difference.
Another issue is that the solution is purely one dimensional in nature and is completed at constant
pressure. These conditions obviously do not hold true in a real reservoir. If the order of fluids contacting
each other in Figure 1 changes, the process is fundamentally altered. Likewise, the pressure changes
between the injector and the producer affects the shape of the phase envelope, and again the process is
fundamentally altered.
Therefore the following phenomena may limit the use of MMP in flood design:
– Change in pressure from injector to producer
– Temperature variation with depth
– Compositional variation with depth
– Buoyancy of CO2 and other gravitational effects
– Baffling caused by heterogeneity
– Channelling and diffusion, causing mass transfer between different layers
In short, in a real reservoir the variables of pressure and composition are constantly changing both
spatially and with respect to time and this is not reflected by the MMP parameter. Also, since we are
operating near the critical point of the mixture, small differences in pressures and temperature result in
more abrupt changes in fluid properties.
Finally, the MMP does not give us a good indication of when water injection may be more beneficial
than continued CO2 flooding, a practical consideration for many floods.
The proposed solution is to not use the MMP as a standalone design parameter, rather to use the
slimtube tests to improve the equation of state (EOS), and then dynamically calculate miscibility
conditions. By repeating the flash calculations to solve for saturation pressure and subtracting the
predicted cell pressure, a property can be created to visualize the ‘distance’ to miscibility across the
reservoir. The application of this concept is applied to a real pilot flood under design.
4 SPE-173867-MS
Method
A Western Canadian field being designed under the pilot phase for CO2 flooding is investigated. The
field is very well characterized in this case, containing 92 wells in the formation and over 55 years of
history. The field is supported by a simulation model, closely matching the long history of the wells to
(⬍5% difference in cumulative volumes for oil, water and gas at all times).
An 8 component Peng-Robinson compositional model is used to describe the phase behavior of the
system, and is tuned to match standard laboratory experiments - including constant composition expansion
(CCE), differential liberation (DL), swelling tests in the presence of CO2, and slim tube experiments. The
slim tube test did not show a clear break in the pressure versus recovery curves, and so there was some
uncertainty regarding the MMP value. It is estimated at 150 bar, although 90% recovery occurs closer to
105 bar.
The model consists of 550 000 cells, 350 000 of which are in the pilot area of interest inside of a large
local grid refinement. 9 local cells are used for every global cell (3x3) to provide additional mixing steps
between the injectors and producers of the CO2 flood.
Figure 3 shows the well locations in the pilot area. All of the vertical wells shown already exist, except
for two of the water injectors. The two horizontal gas injectors and a multilateral producer are planned
new wells for the flood.
Figure 3—Diagram area and wells in the CO2 Pilot Flood
Trapped oil saturation

Without an explicit trapped oil saturation, the simulation will let oil be swept to zero saturation under
miscible conditions. This is known not to be true when compared to actual CO2 core flooding experiments
(due to unswept micro regions and/or the precipitation of a heavier phase).
To mitigate this effect, the trapped saturation was set explicitly to the lab-measured value using the
“SOR” functionality (Schlumberger, 2014). SOR sets a saturation of oil that does not vaporize and is
effectively totally immobile. The residual oil composition was set to be equal to the composition of the
original oil. An improvement for future work, if more lab data comes available, would be to set the
immobile residual oil fraction to be heavier than the original oil.
SPE-173867-MS 5
Diffusion
Diffusion was included in the simulation model. In the absence of experimental data, the diffusivity
was estimated from Grogan et al. (1988) at 3.9E-4 m2/d in liquid and 3.9E-3 m2/d in gas.
Miscibility Calculations
The basic idea is to get the saturation pressure, i.e. the dew-point pressure or bubble point pressure,
based on the composition and temperature of the cell at various times. Obtaining the saturation pressure
is calculation-intensive, as it requires an iterative procedure through the equation to state, which itself
needs to be duplicated for every cell and timestep.
The simulator run a second time for this purpose. Where the saturation pressure is desired, a zero-time
simulation is initialized using the composition solution from the first simulation (the liquid and gas mole
fractions and temperature from a restart file). A high pressure is substituted as an initial guess so that the
upper boundary of the phase envelope is found, and not the lower (500 bar was used). The saturation
pressure is then requested as an output.
By subtracting the cell pressure from the saturation pressure, a 3-D property is created highlighting
where the flood is currently under miscible conditions, and by what amount of pressure. The basic process
is illustrated in Figure 4. Fortunately, the steps demonstrated have become easier to automate via a pre-
and post-processor.
Figure 4 —Calculation process to obtain pressure difference to cell miscibility
Original Design
The general strategy of the original flood design was to charge up the area of the pilot flood to the
MMP. Producers were set to continue production at the existing rates, whereas the gas injectors inject
aggressively into the pay zone, with small breaks to preform water-alternating gas mobility control, and
water injectors on the periphery. The injected volumes of the original and re-designed plans are shown in
Figure 9.
The permeability distribution with depth is unfavourable to gas flooding – with higher permeability
towards the top of the formation and lower permeability with depth. Injected CO2 volumes are expected
to naturally migrate to the top of the formation, and so injected CO2 is planned towards the bottom.
Results
The saturation pressure was determined on a cell by cell basis through the described process. A histogram
was created for the cells under active CO2 flood – that is, with CO2 mol fraction of between 5-55%, and
the remaining reservoir oil. The results are shown in Figure 5 and exhibit a roughly log-normal
distribution of saturation pressures, ranging from 100 to over 200 bar. The interpretations of MMP, both
from the 90% recovery in the slim tube test and the 150 bar estimate, do not seem to mark any natural
break in the distribution where most cells are miscible at lower pressure.
6 SPE-173867-MS
Figure 5—Equation of State Calculated Saturation Pressure for Cells with 5 to 55 mol%
The difference between the saturation pressure and the cell pressure was then visualized to see if it
could illuminate areas where design improvements could be made across the pilot. In one example, shown
in Figure 6, a well section window between an injector and producer is demonstrated where three possible
improvements can be seen. These are:
Figure 6 —Pressure ‘Distance’ to Miscibility (100 to -100 bar) and Total CO2 mol fraction (0 to 1), Original Strategy
1. An area of low miscibility but high CO2 fraction on the left-hand side of the plot. This shows us
how, despite being well above the MMP, extra CO2 can pull the reservoir fluid back into the two
phase region. In reservoirs where CO2 injection is expensive and sequestration has not become a
major goal (the reality for most CO2 floods today), this may be considered excessive injection.
2. An area of by-passed oil exists near the producer where CO2 is over-riding the oil in place.
3. Around the producers, the bottom hole pressure is too low, leading to an area of low miscibility
near the wellbore. The presence of two hydrocarbon phases in this zone would lower well
SPE-173867-MS 7
productivity.
It is noted that the above effects are occurring despite cell pressures that are above the MMP as
determined by slim tube. Cell pressures at this time step range from 175 bar at the injector to 165 bar at
the producer (Figure 7).
Figure 7—Reservoir Pressure, Original Strategy
Changes in Design
Based on the analysis of the miscibility parameter, the following design changes were suggested:
– Remove the multilateral well and increase bottom hole pressure limits of producers
– Lower the quantity of CO2 injected
– Inject CO2 into both the gas injectors and producers. Temporarily shut in the producers for
change-over. This allows the injectors to pressure-up the region and also gives time for the CO2
to diffuse slightly.
– Chase the miscible bank with water
For comparison, the same cross section at the same time is shown in Figure 8 for the new design. The CO2
concentrations are more moderate in this case, and develop into a large miscible bank as they are later
pushed into the reservoir and pressurized by chase water. Because the CO2 injection is split across
multiple wells and there is some time for diffusion, more oil is contacted with a smaller volume of CO2.
The differences in production and injection volumes are shown in Figure 9 and Figure 10.
8 SPE-173867-MS
Figure 8 —Pressure ‘Distance’ to Miscibility (100 to -100 bar) and Total CO2 mol fraction (0-1), Redesigned Strategy
Figure 9 —Cumulative Injected Gas and Water for the Original and Re-designed Pilot Flood
SPE-173867-MS 9
Figure 10 —Cumulative Oil for the Original and Re-designed Pilot Flood
Impact on Economics
A full economic analysis was carried out using West Texas Intermediate futures less an offset, and
including the special allowances allowed under Alberta royalties for CO2 EOR floods. A 12% discount
rate was used. To compare the two cases, an incremental cash flow diagram is shown in Figure 11, where
many of the design impacts are clear.
Figure 11—Re-designed Case Cash Flow minus Original Case Flow, After Tax and Royalties, 12% Discount Rate
In the first year, the economics are improved as the multilateral is not drilled, thereby saving capital
costs. In the second year, negative incremental cash flow results from the producers being shut in. This
is then more than offset by the increased production and lower injection costs associated with the
production of the miscible bank over the following 10 years. Past this date, the cash flow between the two
cases is roughly similar.
Conclusions
– The slim-tube interpreted MMP seems does not seem to show any natural break in the distribution
of saturation pressure for cells containing 5-55% CO2 mol fraction.
10 SPE-173867-MS
– Changes in composition, pressure and temperature in a 3-D flood add complexity and may lower
the usefulness of minimum miscibility pressure as a design parameter.
– The creation of a property that is the difference between the saturation pressure and the cell pressure
can illuminate areas of the flood that are not under miscible conditions and could possibly be
improved via field development plan changes.
– A re-design of a Western Canadian pilot flood was proposed using the calculated parameter,
showing improved recovery and economics.
Caveats and further work
Numerical dispersion is known to be an issue with compositional models and miscible floods, (Gholam-
reza and Johns 2009). Since the predicted saturation pressure so strongly depends on the cell composition,
the grid size, the flow rate, flow direction and real dispersion rates should be carefully examined, and is
the subject of further work. Ideally, numerical dispersion should be near to physical dispersion.
The diffusion constants used are estimated from literature, and may be optimistic particularly for the
gas phase. Near critical gas and liquid phases are not materially different, and lab testing and sensitivities
should be carried out on the diffusion constant.
The final design is subject to changing operational constraints.
Acknowledgements
The authors would like to thank our colleagues at Arc Resources and Schlumberger for all their support,
technical expertise, and for the use of ECLIPSE/Petrel software to complete this study.
References
Gholamreza, G. Johns, R 2009. Upscaling of Miscible Floods in Heterogeneous Reservoirs Consid-
ering Reservoir Mixing. SPE 124000
Grogan, A.T., Pinczewski V.W., Ruskauff G., and Orr, F.M. 1988. Diffusion of CO2 at Reservoir
Conditions: Models and Measurements. SPE 14897
Izgec, O. and Demiral, B. 2005. CO2 Injection in Carbonates. SPE 93773 Schlumberger, 2014.
ECLIPSE Reference Manual pg 2118.
Wu, R.S. and Batycky, JP. 1990. Evaluation of Miscibility from Slim Tube Tests. Journal of
Petroleum Technology, PETSOC 900606
SPE-173867-MS 11
Appendix A
For the first contact miscible case, any oil in the red region (Figure 1) will mix with the injection gas (G). Any combination
of G and oil falls outside the two phase envelope and thus will be single-phase.
For a multiple contact miscible oil, the process is much more complicated. Virgin oil (R), contacts a CO2-rich injection gas
(G) and creates the mixture (M1) which splits into phases G1 and L1. The enriched gas (G1) continues into the reservoir and
contacts fresh virgin reservoir oil (R), creating a new mixture (M2) and a further enriched gas (G2). This process repeats until
composition leaves the two phase region, and can occur when the virgin oil is on the right of the limiting tie-line (that is, in
the green region in Figure 1).

Spe 173867 Ms

Uploaded by

Copyright:

Available Formats

Spe 173867 Ms

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Spe 173867 Ms

Uploaded by

Copyright:

Available Formats

SPE-173867-MS

Copyright 2015, Society of Petroleum Engineers

Development of Miscibility and Traditionally Minimum Miscibility Pressure (MMP)

Figure 1—Ternary diagram, first contact and multiple contact miscibility

Figure 3—Diagram area and wells in the CO2 Pilot Flood

Trapped oil saturation

Figure 4 —Calculation process to obtain pressure difference to cell miscibility

You might also like