OIL-FOAM INTERACTIONS

IN A MICROMODEL

SUPRI TR 110 Report

By

Neeraj S. Sagar
Louis M. Castanier

November 1997

Work Performed Under Contract No. DE-FG22-96BC14994

Prepared for
U.S. Department of Energy
Assistant Secretary for Fossil Energy

Thomas Reid, Project Manager

National Petroleum Technology Office
P.O. Box 3628
Tulsa, OK 74101
ii
 TABLE OF CONTENTS

Page

List of Figures v
List of Tables x
Acknowledgments x
Abstract xi

1. Introduction 1

2. Literature Review 4

2.1 What is Foam? 5

2.2 Early Work on Mobility and Propagation of Foam 6

2.3 Recent Work on the Mobility and Propagation of Foam 6

2.4 Work on Effect of Wettability of the Rock on Oil-Foam Interactions 11

3. Experimental Apparatus 13

3.1 Equipment: General Description 13

3.2 Pressure Transducer and Demodulator 16

3.3 Surface Tensiometer 17

3.4 Interfacial Tension Measurement Using Spinning Drop 19

iii
4. The Silicon Micromodel 27

4.1 Design of the Micromodels 27

4.2 Micromodel Characteristics 30

5. Experimental Procedure and Experiments Performed 34

5.1 Procedure 34

5.2 Experiments Performed 35

6. Results 37

6.1 Experiment 1 37

6.1.1 Observations 38

6.2 Experiment 2 39

6.2.1 Observations 39

6.3 Experiment 3 41

6.3.1 Observations 41

6.4 Interfacial Tension Measurements 42

6.5 Scanning Electron Microscopy 43

Experiment 1 44

Experiment 2 57

Experiment 3 78

References 96

Appendix A 100

Appendix B 101

iv
 LIST OF FIGURES

Page

1. Line diagram of the Experimental Apparatus 14

2 Schematic of spinning-drop apparatus 20

3. The video camera mounted on the microscope 21

4. The micromodel in the specially designed holder with the plastic connector
in between 21

5. Photograph of the experimental set-up in the lab 22

6. The high pressure CO2 cylinder used in the experiments 22

7. The micromodel holder. Refer Figs. 11(a) and (b) for the design 23

8. Photograph of the acid resistant plastic (Teflon) connector which is mounted

between the micromodel holder shown in Fig. 7 23

9. Surface Tensiometer 24

10. Spinning drop apparatus used for interfacial tension measurement 24

11(a) Design of the micromodel holder 25

11(b) Design of the micromodel holder 26

12. Flow channels in the two silicon micromodels 31

13. SEM of model A (old) 32

14. Photograph of the repeat etch pattern on the silicon wafer 32

v
15. SEM of model A (old) 33

16. SEM of model B (new) showing the melted glass on the etched pattern due
to bonding of the wafer with Pyrex at extreme conditions 33

17. CO2 bubbles flowing in the outlet channel during CO2 flood. Flow is top
to bottom diagonally 45

18. CO2 bubbles have just emerged from the porous network into the outlet
channel 45

19. Crude oil emulsions seen in the large pores in the model after 0.01%
surfactant flood 46

20. Water films in the outlet port during waterflooding. The water contains
dissolved CO2 gas 46

21. The CO2 bubbles are generated near the corner. Flow is from top to
bottom diagonally. The bubbles coalesce to form bigger bubbles almost
instantaneously 47

22. Crude oil invading the inlet channel. It can be seen that the oil does not fill
the model uniformly 47

23(a) 48

23(b) 48

23(c) 49

23(d) 49

23(e) 50

23(f) An emulsion forming sequence 50

24(a) 51

24(b) 51

24(c) 52

24(d) 52

24(e) A bubble coalescence sequence 53

vi
25(a) 54

25(b) 54

25(c) Emulsion breakdown sequence 55

26. A crude oil emulsion still 56

27. An oil film can be seen surrounding the gas 56

28. Emulsions as seen after 1% surfactant flood 58

29. Very stable emulsions seen at the end of the experiment (after 1%
Surfactant flood) 58

30(a) 59

30(b) Water snapping off into a kersosene bubble. No emulsions are formed here
as there is no surfactant present 59

31(a) 60

31(b) 60

31(c) 61

31(d)Emulsions seen after 0.001% flood. At this time there is only kerosene and
no gas is present 61

32. CO2 is discontinuous and does not flood the entire model. This is done
after 0.001% Nansa surfactant flood 62

33(a) 63

33(b) 63

33(c) 64

33(d) 64

33(e) 65

vii
33(f) A sequence of moving emulsions seen during CO2 flooding (after 0.001%
Nansa surfactant flood). Ther eis however not much CO2 present in this part
Of the model 65

34(a) 66

34(b) 66

34(c) 67

34(d)An emulsion moving sequence observed with the 40X magnification lens. As
can be seen there is no CO2 present in this part of the micromodel (after
0.001% surfactant flood) 67

35(a) 68

35(b) 68

35(c)An emulsion moving sequence observed after 0.01% Nansa surfactant flood,
just before starting CO2 flood 69

36. Both emulsions and gas at this time are flowing. This is observed after 0.1%
Nansa surfactant flood 70

37. Emulsions observed near the outlet channel after 0.1% surfactant flood 70

38. Different sizes of emulsions seen in different pore/throat sizes. Indicates that
Emulsions depend on pore/throat size 71

39. A very fine oil film seen in the pore (after 0.1% surfactant flood and CO2
Flood) 71

40(a) 72

40(b) 72

40(c) 73

40(d) 73

40(e) An emulsion forming and breaking sequence observed after 0.1% surfactant
flood and during CO2 flood (just before beginning 1% surfactant flood) 74

41(a) 75

viii
41(b) 75

41(c) 76

41(d) 76

41(e) Another emulsion forming and breaking sequence observed after 0.1%
surfactant flood and during CO2 flood (just before beginning 1% surfactant
flood) 77

42. Water can be seen residing in the narrow throats 79

43. A lot of small CO2 bubbles with surfactant film seen. The CO2 phase is not
continuous any more. Seen after 0.1% FC740 surfactant flood 79

44(a) 80

44(b) 80

44(c) 81

44(d)A CO2 bubble snap-off sequence 81

45(a) 82

45(b) 82

45(c) 83

45(d) 83

45(e) A lamella forming and breaking sequence observed after 0.1% FC740
surfactant flood 84

46. A drop of water in a lamella. It doesn’t seem to affect the lamella stability 85

47. A stable lens 85

48. A tripod like lamellae construction 86

49. Four unstable lamellae seen after 0.1% FC740 surfactant flood. 86

50. Stable lamella seen in the big pore after 0.1% surfactant flood 87

51. The foam lamella seen here seems to be thinner than usually seen in this
Pore geometry and is still stable 87

ix
52. Another stable foam lens 88

53. A similar pore geometry to some earlier figures with stable lamellae in a
Different configuration 88

54(a) 89

54(b) 89

54(c) 90

54(d) 90

54(e) Some different lamellae configurations seen throughout the micromodel after
1% surfactant (FC740 in kerosene) flood 91

LIST OF TABLES

Page

1. Properties of Objective Lenses 15

2. Experiments Conducted 36

3. Composition of Brine (Wetting Phase in Experiments 2 and 3) 36

4. Interfacial Tension Measurement 43

ACKNOWLEDGEMENTS

Financial aid for this work was provided by the Department of Energy under Contract
DE-FG22-96BC14994 and the SUPRI-A Industrial affiliates. This support is gratefully
acknowledged.

x
Abstract

This report presents results of a pore-level visualization study of foam stability in the
presence of oil. Many laboratory investigations have been carried out in the absence of oil,
but comparatively few have been carried out in the presence of oil. For a field application,
where the residual oil saturation may vary from as low as 0 to as high as 40% depending on
the recovery method applied, any effect of the oil on foam stability becomes a crucial
matter.

Sandstone patterns were used in this study. The micromodels used are two-dimensional
replicas of the flow path of Berea sandstone etched on to a silicon wafer to a prescribed
depth, adapting fabrication techniques from the computer chip industry. After flooding the
models up to connate water and residual oil saturations, surfactant flood followed by gas
injection to generate foam was done. Starting with lower concentrations of surfactant and
gas injection the procedure was followed up to higher concentrations of surfactant. Visual
observations were made using a high resolution microscope and pictures were recorded on
videotape before being processed as they appear in this report.

The single most important reason for this study on silicon micromodels compared to
previous micromodel work is pore dimensions. With glass micromodels, for example, the
reaction kinetics of acid etching makes it necessary to enlarge the pores by a factor of 5 to
50 thus providing a serious limitation for their use in studying processes that depend
critically on capillary forces or involve thin films as compared to real rock pores.

xi
Two different surfactants were used, a fluoro-surfactant (for generating an oil foam) and an
Alpha-Olefin Sulfonate. Oseberg crude was the non-wetting phase in the first set of
experiments, and Kerosene in the next two sets. While the fluoro-surfactant created a strong
static gas-blocking foam in the presence of oil, the Alpha-olefin Sulfonate (AOS) foamer did
not.

The fluoro-surfactant foam gave the oil-tolerant behavior expected from its non-entering,
non-spreading characteristics. The AOS on the other hand, did not behave in accordance
with its bulk observations and its behavior was seen to be controlled by formation of
oil/foam emulsions. Generation sites for both foam and the emulsions were seen to be
controlled by pore geometry and local saturation. For the foam, no obvious link could be
found with the number of films observed and the strength of gas blockage. A lot of other
interesting observations included snap-off, emulsion formation and breakdown sequences,
foam lamella formation and breakdown sequences, and static emulsion and foam in different
configurations within the model usually at higher concentrations of surfactant.

xii
Section 1

Introduction

Oil production from a petroleum reservoir involves primary and secondary production
modes, which may recover less than half of the original oil in place. To recover
additional oil, it is necessary to apply Enhanced Oil Recovery (EOR) techniques such as
miscible or immiscible gas displacement (CO2, hydrocarbon gases, etc.). However,
major problems occur in these EOR methods because the displacing agent has high
mobility and low density compared with that of reservoir fluids. Fingering (channeling)
and gravity override reduce the sweep efficiency, contribute to early breakthrough of
injected fluid, and thus reduce the amount of oil recovered.

The use of surfactant stabilized foams to counteract these kinds of problems was
suggested decades ago. The use of foam is advantageous compared with the use of a
simple fluid of the same nominal mobility because the foam, which has an apparent
viscosity greater than the displaced fluids, lowers the gas mobility in the swept or higher
permeability parts of the formation. This diverts at least some of the displacing gas into
other parts of the formation that were previously unswept or underswept. From these
underswept areas, the additional oil is recovered. Because foam mobility is reduced more
in higher permeability zones than in lower permeability zones, improvement in both
vertical and horizontal sweep efficiency can be achieved.

This report will focus on the stability of foams flowing in porous media in presence of
oil. Many laboratory investigations have been carried out in the absence of oil, but
comparatively few have been carried out in the presence of oil. For a field application,
where the residual oil saturation may vary from as low as 0 to as high as 40% depending
on the recovery method applied, any effect of the oil on foam stability becomes a crucial
matter.

Two different approaches have been used in an attempt to scale up foam properties to the
reservoir scale. They represent different levels of ambition in describing the physics of
confined foam. One approach is purely empirical and relies on calibrating a flexible
simulator by selected laboratory data, which are measured on conventional core-sample
scale and believed to represent those foam effects which are most critical. The empirical
foam model implemented in STARSTM, a pseudo-compositional reservoir simulator is
probably the best example of this approach (Coombe, 1992; Surguchev, 1995). It has the
advantage of relative simplicity, but depends heavily on obtaining “correct” values of the
lab parameters and on the assumption that these parameters can be measured in a manner
that is not too strongly dependent on the mode and conditions of the experiment. Recent
work by Hanssen and Dalland (1993, 1994) have shown that this critical assumption may
not hold.

The other approach to scaling up foam properties attempts to extract information about
the physical flow mechanisms as actually observed on the pore-level, and construct
mathematical relationships between experimental parameters that, if correct, can be used
to translate foam properties to different scales. The best known such foam simulator is
the one developed at the University of California at Berkeley. The physical mechanisms
in the Berkeley model rely on observations of foam flow in etched-glass micromodels,
especially a comprehensive study by Chambers and Radke (1991).

2
Micromodels, designed to represent the geometric structure of a rock pore network, while
allowing direct visual observation of the flow phenomena, have proven useful for
viewing pore-level events in several reservoir processes. Currently accepted theories on
foam flow are largely built on the basis of pore model observations by Mast (1972),
Owete et al. (1984), Huh et al. (1989), Schramm (1990) etc. Most micromodels have
been made by etching a flow pattern into a glass plate using. e.g., hydrofluoric acid (HF)
and are two-dimensional representations of the pore structure. With glass micromodels,
the reaction kinetics of acid etching makes it necessary to enlarge the pore sizes of typical
sandstone pores, usually by a factor of 5 to 50 compared to their natural size. Obviously,
this could be a serious limitation for their use in studying processes that depend critically
on capillary forces and involve thin films.

The silicon micromodel on the other hand, replicates a pore structure of any design on a
silicon wafer rather than a glass plate, adapting fabrication techniques from the computer
chip industry. The pore cavities are created photochemically, a process which is not rate-
limited such as acid etching. The micromodels used in this study offer 1:1 scaling of
typical sandstone pores and were made to represent Berea sandstone. This is a major step
forward for visual studies of pore-level events.

This section of the report is followed by a section on Literature survey, after which the
experimental apparatus is defined. A brief description of the fabrication techniques used
for the silicon micromodel comes next followed by the experimental procedure and the
experiments performed. To the end of the report we have a section on results and
discussion closely followed by the conclusions.

3
Section 2

Literature Survey

There has been a considerable amount of work published regarding foam research. The
literature provides valuable insight into the occurrence, properties, importance and
propagation mechanisms of foams based on colloid science. Many laboratory
investigations of foam flooding have been carried out in the absence of oil, and
comparatively few have been carried out with oil present. An argument for not including
oil was simply, that a need to understand the basic propagation mechanisms of foam
dictated such a move. Any additional components would only serve to complicate
matters. As research on foam and its propagation in porous media progressed, foam was
seen as a fascinating fluid, both because of its unique microstructure, and because of its
dramatic influence on the flow of gas and liquid.

When research began to include oil, it was seen that the effect of oil on foam stability was
indeed complicated. Oil saturation could alter foam stability and its propagation
mechanisms to a great extent. From a practical viewpoint for example, a field application
where the residual oil saturation may vary from as low as 0% to as high as 40%,
depending on the recovery method applied, any effect of oil on foam stability becomes a
crucial matter. Also the nature of the foam, foam injection procedures, reservoir

4
wettabilities and pore geometry were beginning to be seen as affecting the oil-foam
interactions making it even more complicated.

Clearly, more work is required in this area. This need provides a basis for our study.

2.1 What is Foam?

A foam is a dispersion of gas in liquid, usually with a surface-active agent present.

Foams are not thermo-dynamically stable and ultimately decay into their constituent
phases, but can be mechanically stable.

When a foam exists inside a confining medium, dimensions of this confining medium
relative to the average bubble size determine the texture and properties of the foam.
When the confining diameter is large relative to typical bubble size, such as in a pipe, the
foam is similar to bulk foam. Its flow can then be treated as that of a non-Newtonian,
compressible pseudofluid. Where the diameter of the confining body is comparable to, or
smaller than the minimum bubble size, the foam exists as a network of lamellae rather
than bubbles. Such a lamellar-structure foam can not be treated as a single fluid, because
liquid and gas flows by different mechanisms. Holm (1968) found liquid to flow as a
continuous phase, but gas flow to occur by a sometimes very slow process of continually
breaking and re-forming the liquid films. This explains why gas flow may be essentially
blocked by foam in pores, while the permeability to liquid is merely reduced in
proportion to the liquid saturation.

The ability of foam to reduce gas mobility has led to its application in a number of
processes, including gas flooding, steam flooding, and well treatment techniques. It has
been somewhat uncertain whether foam in these cases actually works by gas blockage
and flow diversion, or whether it is better described as a “viscosifying agent” for the gas.

5
2.2 Early Work on Mobility and Propagation of Foam

The use of surfactant stabilized foams to counteract problems such as Fingering and
gravity override were suggested several decades ago. Fried (1961) demonstrated that
aqueous surfactant stabilized foam could dramatically reduce the mobility of gases in
porous media. At that time foam was studied mainly from a phenomenological
perspective. In the intervening 35 years, foam has been recognized as a fluid with unique
rheological properties within porous media, and the scope of research has expanded to
include local pore scale phenomena and local microstructure. Considerable work was
published by Raza, Marsden, Bernard et al, Mast, Khan, Nahid in the 1960’s and 70’s
regarding the behaviour of foam. Details of these are not covered in this literature survey
as they have been exhaustively covered earlier by Marsden (1986).

However none of these proposed early mechanisms adequately accounted for all the
observed properties of foam in porous media. The only general agreement was that foam
generally hindered the flow of gas in porous media.

2.3 Recent Work on the Mobility and Propagation of Foam

Marsden (1986) reviewed previous findings, surmising that apparent gas viscosity was
indeed a function of the gas flow rate, flow history and surfactant concentration, and that
‘breaking and reforming’ processes are a dominant mode of foam propagation in
micromodels. Sanchez and Schechter (1986) suggested that gas permeability was a
function of wetting phase film thickness.

Owete et al (1984) suggested that air was propagated by displacement of lamellae in long
bubbles flowing and extending across several pore lengths while the liquid flowed
through the network of films. They also observed snap-off occurring at pore
constrictions and the resultant bubble ‘break and reform’ process. Radke and Ransohoff
(1986) categorized the various mechanisms of foam generation within glass bead packs

6
as ‘snap-off’, ‘lamella leave behind’, and ‘lamella division’, with snap-off being the
primary mechanism for strong foam propagation.

Roof (1970) and Mohanty et al. (1980) each studied and reported specific pore geometry
necessary for snap-off mechanisms. Roof noted that the front is always at least seven
pore radii from the throat of the constriction before snap-off can occur. Mohanty et al.
calculated that snap-off occurs when the ratio of pore body radius to adjacent throat
exceeds three. Falls et al (1988) and Kovscek et al. (1994) proposed that the rate of
generation of lamella must be equal to their collapse at steady state conditions.

Castanier and Hanssen (1995) noticed the sliding of lamellae (bubble train flow) over
liquid films covering the grains in silicon micromodels. Using atomic force microscopy
between the liquid/solid interface, a thick layer of surfactant was found. This layer they
state could explain the apparent bubble train effect observed.

Spreading and Entering Coefficient Theory for Oil-Foam Interactions :

The spreading coefficient, S, for an oil-foam system is given by:

S = γ ° f − γof − γ ° o

where : γ ° f = Foaming solution surface tension

γof = Initial foaming solution/oil interfacial tension

γ ° o = Surface Tension of the oil

When oil spreads over the gas-aqueous interface, a certain amount of both gas-oil and
aqueous-oil interface is created, and some gas-aqueous interface is eliminated. For a unit

7
surface area, the Gibbs free energy change (∆G ) is given by ‘-S’. The spreading is

predicted to be favored when ∆G is negative and S is positive. Rapid spreading of a

drop of oil that has a low surface tension over the lamella can cause rupture by providing
a weak spot. The spreading oil lowers the surface tension, increases the radius of
curvature of the bubbles, alters the original surface elasticity, and also changes the
surface viscosity. Thus the interfacial film loses its foam stabilizing capability. If S is
negative, then the oil should not spread at the interface.

Another mechanism of rupture involves the drawing up of a droplet of defoamer into the
lamella region between two adjacent bubbles so that it can bridge the two bubbles. The
breaching of the aqueous-gas interface by the oil from within the lamellar liquid is termed
‘entering’. The process is thermodynamically favorable if the ‘Entering Coefficient’ is
positive (originally defined by Robins and Woods (1948) as a rupture coefficient, R). For
a oil-foam system, the entering coefficient is given as :

E = γ ° f + γof − γ ° o

For unit surface area, the Gibbs free energy change is given by ‘-E’. The entering is
predicted to be favored and spontaneous when ∆G is negative and E is positive.

Work on “Effect of Oil on Foam Stability and Propagation”:

Oils such as Castor oil, are among the earliest chemicals to be used for foam inhibition
and foam breaking. Literature by Harkins (1941), Robinson et al (1948), Ross (1944) etc.
deals with it. Some theories for the mechanisms of anti-foaming action have emerged
from this work. In general matters are not simple. Apparently foams can be destabilized
by oils by several mechanisms, and more than one mechanism may be operating in a
given case.

8
Some possibilities for the mechanisms of foam destabilization by a given oil phase
include:

∗ Foam-forming surfactant may partition in the oil, especially if there is emulsification ,

causing depletion in the aqueous phase and hence from the gas-liquid interface.
∗ Surfactants from the oil may be adsorbed by the foam lamellae, from either a mixed
or replaced adsorption layer, and produce a less favorable state of foaming.
∗ Components from the oil may be adsorbed by the porous medium altering the
wettability of the solid phase, and this alteration makes it more difficult for foam to
be generated and regenerated.
∗ The oil may spread spontaneously on foam lamellae and displace the foam stabilizing
interface.
∗ The oil may emulsify spontaneously and allow the drops to break and rupture the
stabilizing interface.

Nikolov et al (1986), when studying bulk oil-foam interactions, reported the possible
occurrence of three distinct films during the process of three phase foam thinning. They
are ‘foam films’, which are water films between air bubbles; ‘emulsion films’, which
represent water films between oil droplets; and ‘pseudo-emulsion films’ where water film
resides between air and oil droplets. Their findings said that pseudo-emulsion films,
droplet size and droplet numbers may all contribute to foam destabilization.

Rendall et al. (1990) investigated the behavior of several commercial surfactant

destabilized foams, in the presence of crude oil. On the basis of dynamic bulk foaming
tests, gas mobility reduction factors measured in reservoir cores, and observations in a
microvisual apparatus, it was found that all but one of the foams studied were destroyed
when brought into contact with oil.

Similarly Kuhlman (1988) and Manlowe and Radke (1990) observed in micromodel
studies that oils were capable of destroying foams. Work done by Novosad et al. (1989),
Jensen and Friedmann (1987) and Holt (1991) suggested that foam sensitivity to oil is

9
also manifested as an increased difficulty of forming and in propagating foams through
porous media containing oil. Novosad and Ionescu (1987) found lower mobility reduction
factors in foam floods conducted in cores containing residual oil compared to the same
floods conducted in cores that were completely free of oil. The lower mobility reduction
factors were interpreted to be due to some kind of foam destruction by the oil.

Lau and O’Brien (1988) provided further insight into the effects of oil saturation on foam
propagation. Their work relies upon an understanding of oil spreading ability in which
spreading is determined from ‘spreading coefficients’ or free surface energies. Through
the use of sandpacks and both spreading and non-spreading oils, they were able to show
an increase in foam generation time, a reduction in foam propagation rate, and an
increased foam destruction rate with spreading oils. This finding is contrary to Manlowe
and Radke’s who argue that there is no correlation between oil spreading and foam
stability. Instead, they state that pseudo-emulsion film collapse is the general
destabilization mechanism.

Hanssen et al (1990), in a model system (glass bead pack) of a gas blocking foam in an
oil reservoir, observed clear trends in the mobility reduction as measured in a gas-
blockage test, with varying concentrations of oil and surfactant. They concluded that the
change in interfacial tension between gas and the aqueous phase, upon equilibration with
oil, appeared to be a good predictor of gas-blockage efficiency for the systems studied.

Oil interactions with foam have been studied by Dalland et al. (1993). On conducting
experiments with flowing and gas blocking foams in porous media, using a set of
conventional and fluorinated surfactants and various oils, they found varying oil tolerance
and sensitivities. In all cases it was found that the combination of non-spreading and non-
entering oil created a strong gas-blocking foam or a low mobility flowing foam. The
‘Lamella number’, L, calculated from interfacial tensions was found to be of no
predictive power. The results indicated that the mechanisms of oil-foam interactions are
the same for gas blocking and for flowing foams in porous media.

10
From unpublished work by Castanier on oil-foam interactions using a fluorosurfactant
foam and an AOS foam, the fluorosurfactant foam seems to give the oil tolerant behavior
expected from its non-entering, non-tolerant characteristics. AOS foam by contrast, is
seen to be controlled by formation of oil-water emulsions which appear to prevent the
formation of strong foam until a large excess of foaming agent is present.

2.4 Work on Effect of Wettability of the Rock on Oil-Foam

Interactions

Most of the research on foam sensitivity to oils in porous media, whether in microvisual
apparatus or core-fluid tests, has been concerned with water wetted pore and throat
surfaces. Because petroleum reservoirs are of intermediate, mixed or oil wettability,
some researchers have spread their work to include reservoir wettabilities.

Huh et al. (1988) found that micromodel wettability was altered to intermediate or oil-wet
as a result of crude oil saturation. The effects of various phase affinities to grain surfaces
was studied by Sanchez and Hazlett (1989) as well. An indication of surface reaction,
from hydrophobic to hydrophilic, was detected after the addition of surfactant. This fact,
along with a shift in liquid phase relative permeability of the oil-wet medium, suggested
wettability alteration.

Kuhlman (1990) commented on similar results of his own, remarking that the deleterious
effect of oil on foam was due, in part at least, to the high concentration of light
hydrocarbons in the oil, and the oil wetness of the medium itself. Work by Hornbrook et
al. (1992) noted that wettability alteration was due to the surfactant injection procedure.
They found that with surfactant slug injection, oil appeared to be the wetting phase, and
with foam injection, surfactant appeared to wet the medium.

Schramm and Mannhardt (1994) conducted oleophilic microvisual cell experiments with
several foams. They found that the foams were significantly less stable in the presence of
crude oil and oleophilic solid surfaces, compared with some crude oils and hydrophilic

11
surfaces. Suffridge et al. (1989) reached the same conclusions on the basis of core-flood
experiments. They found that the foams were more effective (stable) in water wet cores
than in oil wet cores. Holt and Kristiansen (1991) studied foams flowing in cores under
North Sea reservoir conditions that were either partially or completely oil wetted. They
found that foam effectiveness was favored by water-wet conditions. Any degree of oil-
wet character reduced the effectiveness of the flowing foam.

A complication is that the foam forming surfactants may adsorb onto the solid surfaces
and may alter the wettability. In the microvisual experiments of Schramm and
Mannhardt (1992), some of the foaming systems investigated appeared to change the
wettability back to water-wet, in which case the foam sensitivities to oil reverted back to
those appropriate to the water-wet cases. The result is consistent with the core-flood tests
of Sanchez and Hazlett (1992) in which foaming surfactants apparently caused
wettability reversal in oil wet porous media. This reversal, in turn, was postulated to be
the reason that stable foams could be generated and propagated in their oil-wetted cases.

12
Section 3
Experimental Apparatus

All the experiments were done at constant pressure. Fluid injection (Carbon Dioxide,
surfactant, oil, water etc.) was facilitated by using pressure bottles (connected by the
pressurized lines). A basic line diagram of the apparatus is shown in Figure 1.

A photograph of the experimental set-up along with the pressure cylinders used is shown
in Figures 5 and 6. The various equipment used are outlined below. All the tubing used
was made of Teflon with a diameter of 1/16”. The silicon micromodels used and the
fabrication techniques are discussed in Section 4.

3.1 Equipment: General Description

1. Microscope: A Nikon Optiphot-M with a photo tube allowing for video imaging was
used (Figure 3). The Objective lenses used along with their properties are as outlined
in Table 1. Working distance is the distance between the tip of the lens and the focal
plane of the objective. Numerical Aperture relates to the light gathering ability and is
a dimension less number between 0 and 1. Higher the value, better the light
gathering.

13
 Inlet
Pressure
Pressurized Gauge
Line VCR / Monitor
Microscope &
Video Camera
Inlet Valve

Micromodel

Filter
Outlet
Pressure Transducer Macintosh with
Capture Board

Demodulator Chart Recorder

Figure 1. Line diagram of the Experimental Apparatus.

14
 Table 1: Properties of Objective Lenses

Model Magnification Working Distance Numerical Viewable

(mm) Aperture Diameter (µm)

5X 5X 20.0 0.1 3000

0.4LWD 20X 6.0 0.4 800
0.5ELWD 40X 10.1 0.5 375

2. Video Camera: A Sanyo closed circuit television camera was used to convert the
images through the microscope into electric impulses. An attached photo tube
allowed image viewing (Figure 3).

3. Video Cassette Recorder: The image from the video camera was recorded on a
Mitsubishi NTSC format video cassette recorder. Its NTSC format allowed recording
and playback at a speed of thirty frames per second (Figure 5).

4. Television Monitor: In order to view the recorded pore-level events, the VCR was
connected to a high definition Sony Color Monitor (Figure 5).

5. Macintosh: The images from the VCR were captured with the aid of a Spigot II tape
video capture board in an Apple Macintosh computer. Image enhancement was done
using Adobe Photoshop (Figure 5).

6. Transducer: A Celesco differential pressure transducer with a fifty psi pressure plate
was used to measure fluid pressure differential across the silicon model. Refer
section 3.2 for details.

15
7. Demodulator: A Celesco carrier demodulator was used to convert the electrical
signals from the transducer into pressure units. Refer section 3.2 for details.

8. Chart Recorder: A Soltec strip chart recorder was used to plot the differential
pressure as recorded by the transducer (Figure 5).

9. Micromodel Holder: A holder was specially designed (Figures 11(a) and 11(b)) to
mount the micromodel for unobstructed viewing under the microscope (Figures 4,8).

10. Micromodel Connector: An acid-resistant plastic (teflon) holder was used to mount

the micromodel on the holder and was provided with an inlet and outlet port (Figure
4,8).

3.2 Pressure Transducer and Demodulator

A series CD 10D Celesco Carrier Demodulator was used with a variable reluctance type
transducer (for measuring the pressure drop across the micromodel) to produce a DC
signal output proportional to both steady-state and dynamic parameters sensed by the
transducer. The series CD10 D Carrier Demodulator operates on 95 to 125 VAC, 60 to
400 Hz at %W, and is used with a variable reluctance transducer to provide DC signal
output voltages. A 10V pk 4Hz Sine excitation is applied to a bridge circuit which
incorporates the two inductance arms of the transducer. Bridge unbalance proportional to
transducer actuation produces an AC signal which is amplified and demodulated for final
DC signal output. The linear output capability of the Demodulator was +/- 10VDC
which corresponds to a nominal +/- 25mV/V transducer output. Zero adjustability can
compensate for a transducer unbalance of +/- 15mv/V (+6V at nominal sensitivity).

Calibration and Operation

16
Although no calibration of the demodulator itself is required, it is necessary to calibrate
the system comprised of the demodulator, transducer and the readout printer. With the
demodulator connected to a voltmeter, set the zero and span potentiometers to their
respective midscales. After the power switch has been turned on, and with no force on
the transducer, adjust the zero potentiometer on the demodulator until zero is indicated on
the voltmeter.

Next, impose a full-scale calibrating force on the transducer. For a 0-50 psi range
transducer, 50 psi should be applied with a deadweight tester or monitored using a
precision gauge. Adjust the modulator span potentiometer until the voltmeter indicates
the desired full scale value. Exercise the transducer-demodulator system through at least
three successive cycles to ensure stable zero and span reading.

3.3 Surface Tensiometer

A FISHER Surface Tensiometer, Model 20 (Figure 9) was used to determine the apparent
surface tension and interfacial tension of the liquids used in the experiment. The
Tensiometer utilizes the principles of operation originally devised by Dr. Pierre Lecomte
du Nouy, a noted biochemist, for studies of blood serum and other biological fluids.
Essentially a torsion-type balance, it is a kind of instrument currently specified by the
American Society for Testing Materials in Methods D-971 (interfacial tension of oil
against water) and D-1331 (surface and interfacial tensions of detergents).

In the du Nuoy method, a platinum-iridium ring of precisely known dimensions is

suspended from a counter-balanced lever arm. The arm is held horizontal by torsion
applied to a taut stainless steel wire, to which it is clamped. Increasing the torsion in the
wire raises the arm and the ring, which carries with it a film of the liquid in which it is

17
immersed. The force necessary to pull the test ring free from this surface film is
measured.

The Surface Tensiometer shows this “apparent” surface or interfacial tension (the latter
measured at the interface between two immiscible liquids) on a calibrated dial. The dial
readings can be used directly for comparative studies or converted to “true” values by
using a correction factor chart.

Operation

The sample should be placed in a glass beaker or cylindrical vessel with a diameter of at
least 45 millimeters. For testing oil samples according to ASTM Method D-971, the
glassware should be cleaned according to a definite procedure. Any residual oil from the
previous sample is removed with petroleum naphtha or benzene followed by several
washes with methyl ethyl ketone and water, then the glassware is immersed in a hot
cleaning solution of chromic acid. The glassware also should be rinsed thoroughly with
tap water, then with distilled water.

The platinum-iridium ring should be cleaned by first dipping it in benzene (to remove
hydrocarbons), then squirting it with acetone (to remove the benzene) and allowing the
acetone to evaporate. Following this, flash the ring in a Bunsen burner flame to remove
residual hydrocarbons.

Measuring Surface Tension

The cleaned platinum-iridium ring should first be attached to the hook at the end of the
lever arm. The arrest mechanism should be holding the arm at this time.

18
The liquid to be measured is transferred to the clean glass vessel and placed on the
sample table. The sample table is moved around until it is directly beneath the platinum-
iridium ring. Raise the sample table until the ring is immersed in the test liquid. The ring
should be in the liquid, beneath the surface so that the entire liquid ring will be wetted.
About 1/8 inch immersion is generally considered sufficient.

The torsion arm is now released and the instrument adjusted to a zero reading. Adjust the
knob on the right side of the case until the index and its image are exactly in line with the
reference mark on the mirror. Be careful to keep the ring in the liquid during this
manipulation, raising or lowering the sample table (if necessary) by means of the knob
adjustment underneath the table. Now turn the knob beneath the main dial on the front of
the case until the vernier reads zero on the outer scale of the dial.

Lower the sample table until the ring is in the surface of the liquid, adjusting the knob on
the right side of the case to keep the index lined up with the reference mark on the mirror.
The surface of the liquid will become distended, but the index must be kept on the
reference. Continue the two simultaneous adjustments until the distended film at the
surface of the liquid breaks. The scale reading at the breaking point of the distended film
is the apparent surface tension.

3.4 Interfacial Tension Measurement Using Spinning Drop

The spinning drop apparatus used in this experiment was designed at the University of
Texas. A schematic of the apparatus used is shown in Figure 2. A hysteresis
synchronous motor was used. Its speed was controlled by varying the frequency from a
frequency generator. The rotational stability was 1 part in 105 as determined by a period
averaging counter stable to 1 part in 10 8. The range of speeds used was from 1,200 RPM
to 24,000 RPM. A Gaertner traveling microscope with a filar eyepiece was used to

19
measure the length and width of the drop and to calibrate the glass tubes for their
magnification of the drop diameter. This effect was determined to be a constant for the
experiment of y-measured/y-true=1.332 for all the aqueous phases studied. The tube
housing and assembly had a design precision ground 0.245” O.D. pyrex glass tube
rounded on one end and sealed against a rubber septum on the other. A glass cell
enclosed the apparatus (Figure 10).

POWER MOTOR SCOPE

CONTROL MOTOR
AMPLIFIER
BOX

FREQUENCY TEMPERATURE ROTOR ASSEMBLY

COUNTER CONTROLLER

LAMP

SIGNAL STROBE
GENERATOR SUPPLY

Figure 2: Schematic of spinning-drop apparatus.

The procedure for loading the cell is to fill the glass tube and metal cap completely with
the more dense phase. Holding the tube upside down (capillary pressure retains the more
dense phase in the tube), the less dense phase is then injected with a microlitre syringe,
the tube is then placed in the cap and the whole assembly is placed in the shaft and
secured by screwing the cap onto the shaft. A small hypodermic needle is inserted into
the cell through the septum to release any pressure build-up caused by screwing the cap
on.

A simple Fortran program to do the calculation is attached in Appendix A.

20
Figure 3: The Video Camera mounted on the microscope

Figure 4: The micromodel in the specially designed

holder with the plastic connector in between.

21
 Figure 5: Photograph of the experimental set-up in the lab.

Figure 6: The high pressure CO2 cylinder used in the experiments.

22
Figure 7: The micromodel holder. Refer figures 11(a) & (b)
for the design.

Figure 8: Photograph of the acid resistant plastic

(Teflon) connector which is mounted between the
micromodel holder shown in Figure 7.

23
 Figure 9: Surface Tensiometer

Figure 10: Spinning drop apparatus used for interfacial tension measurement.

24
 6 Pls.
Clearence for 10/32

5.000
Section A
0.300

0.400
4.500

6.000
PART # 1 : BOTTOM VIEW

Figure 11(a): Design of the Micromodel Holder

25
6 Pls.
Clearence for 10/32

0.500

0.100

0.700

A 4.5000
6.000
PART # 2 : TOP VIEW

Figure 11(b): Design of the micromodel Holder

26
Section 4
The Silicon Micromodel

Microvisual apparatus of several kinds have been used to visualize the interaction of
constrained foam lamella with oil. This section talks briefly of the design of the silicon
micromodels as opposed to the glass micromodels. It then goes on to define the
characteristics of the two different types of silicon models which were used for the
experiments.

4.1 Design of the Micromodels

Initially micromodels were etched in glass. The first type consisted of regular, straight or
constricted channels, made into an interconnected lattice. They are useful in comparing
experimental observation with theories of fluid flow in regular systems and provide a
useful conceptual link to other media such as capillary tubes. Later the designs were
improved upon to include images of actual rock pores. Three dimensional continuity was
a natural loss.

For glass micromodels, the reaction kinetics of acid etching makes it necessary to enlarge
the pore sizes of typical sandstone by a factor of 5 to 50 compared to the original size.

27
This turns out to be a limitation in studying processes that depend critically on capillary
forces. Acid etching also introduces a degree of surface roughness to the pore walls.

The micromodels used in our experiments comprise the next generation of pore-models.
The design replicates the pore structure of Berea sandstone on a silicon wafer. The
fabrication technique is a direct adaptation of the etching techniques used by the
computer industry. The pore cavities are created photochemically, a process which is not
rate-limited such as acid etching. These high resolution silicon micromodels offer 1:1
scaling of the Berea sandstone pores and throats. A detailed description of the fabrication
and construction process of the models used can be found in Hornbrook et al. (1992),
Woody et al. (1995) and Lolomari et al. (1996).

The construction involves four major steps.

1. Imaging
2. Etching
3. Bonding
4. Finishing

Imaging of the model onto the silicon wafer involves the digitization of a rock section
into a computer from a high quality photograph. The section, which is approximately
500µ m across is replicated to fill an area of 5 cm2. After assuring two dimensional
continuity, the image is reproduced on a chrome or glass mask. The grains being opaque
and the pores being transparent. A coating of photoresist is placed on the silicon wafer.
An ultravilolet ray is passed through the mask onto the wafer producing the image. After
the exposure, the flow path image exists as clean silicon for the pores and photoresist
coated silicon for the grains.

Once the pattern has been created using the photoresist, a dry etch technique is used to
create the network of pores and grains on the wafer.

28
Anodic bonding of the silicon wafer requires the placing of a glass plate (pyrex) on top of
the silicon wafer. At elevated temperatures (limited by the softening point of the pyrex),
the positive sodium ions in the glass become quite mobile. They are attracted to the
negative electrode on the glass surface where they are neutralized forming a bond. A
description of the anodic bonding process is provided by Terry (1975). Pyrex is ideally
suited for the above procedure because its coefficient of linear expansion (3.25 x 10-6 /oC)
is very close to that of silicon (2.56 x 10-6 / oC) and it is inexpensive and readily
available. Detailed properties of the pyrex are attached in Appendix B.

Port holes are drilled and aligned to match those on the wafer. Electrodes are attached to
the model and the bonding is done. An additional glass plate is epoxied to the bottom of
the wafer to provide support. The bond strength of the silicon to the glass due to the
anodic bonding is great but the strength of the silicon itself is not. Therefore large
pressure drops can cause the wafer to break. To strengthen the micromodel, this
additional glass plate is epoxied to the bottom of the silicon wafer.

A description of the pore structure of the model along with permeability calculations can
be found in Woody et al. (1995). A photograph of the repeat etch pattern can be seen in
Figure 14. Scanning Electron Microscopy (SEM) photographs of the model can be seen
in Figures 13, 15 & 16.

Models Currently Being Made

The new models currently under construction are being etched at the ‘Center for
Integrated Systems’ at Stanford. The etching will be done using the STS DRIE etcher.
The details of the silicon wafer being used are:

29
Diameter : 4”
Type/Dopant : P/Boron
Orientation : <100>
Grade : Prime
Resistivity : 10-20 ohm-m
Thickness : 500-550 µm
Flats : SEMI Standard
Frontside : Polished
Backside : Etched
Lot : # 4514
Manufacturer : Silicon Valley Microelectrics Inc., San Jose.

A positive photoresist (Type 1813) will be used up to a thickness of 1µm. Photoresist-

postbake will be based on the standard program, and the photoresist side wall profile will
be vertical.

4.2 Micromodel Characteristics

Two different types of micromodels were used in the study. The characteristics of these
models are outlined below:

• 500 micron repeat pattern repeated 10,000 times on a 5 sq. cm silicon wafer
• Etch depth 15 to 35 microns.
• Porosity 35%
• Permeability: 1 millidarcy.
• Pore Sizes : 1-150 µm
• Throats : 0.5-10 µm

30
Specific Characteristics of ‘Model A (old)’

• Bonding at 200-400oC & 600 Volts

• Etch depth of 15 µm
• Flow Channels only near the inlet and outlet ports.

Specific Characteristics of ‘Model B (new)’

• Bonding at 450oC & 1500 Volts.

• Etch depths of 15, 25 and 35 µm.
• Flow Channels extended to the length of the sides near the inlet and outlet ports.
• Flow channel depth was up to a maximum possible etch depth of 35 µm (for all
models).

Model B (new) was used for experiments one and two. Model A (old) was used for
experiment three.

Type A Type B
(old) (new)
Outlet Outlet

Inlet Inlet

Figure 12: Flow Channels in the two silicon Micromodels

31
 Figure 13: SEM of model A (old)

Figure 14: Photograph of the repeat etch pattern on the silicon wafer.

32
Figure 15: SEM of model A (old)

Figure 16: SEM of model B (new) showing the melted

glass on the etched pattern due to bonding of the wafer
with Pyrex at extreme conditions

33
Section 5
Experimental Procedure & Experiments Performed

5.1 Procedure
Before beginning the experiments, the model was first cleaned with water followed by
methanol to remove traces of foam or solvent. All the experiments were performed at
ambient temperatures using the following generic procedure:

1. CO2 Flood
2. Water / Brine flood to dissolve all CO2.
3. Oil flood to connate water saturation.
4. Water / Brine flood to residual oil saturation.
5. Surfactant flood.
6. Gas Injection to generate foam
7. Surfactant flood (Higher concentration)
8. Gas Injection to generate foam

Steps 5 and 6 were repeated (as shown in 7 and 8) for higher concentrations of surfactant.
Before beginning each experiment, the models were cleaned using water and methanol.
The outlet was at atmospheric pressure and the inlet pressure was measured. All fluids

34
entering the micromodel were filtered to 0.22µm. A soluble dye was used to color the
surfactant so that it could be visualized better. The dye used was an ethyl-oil soluble dye:
benzene-azo-2-Napthol. Note that interfacial properties were not affected by the dye.

5.2 Experiments Performed

All the experiments were recorded on videotape. The following procedure was adapted
to generate the pore level photographs you see in this report.

1. Image capturing on the Macintosh using ‘Spigot’ capture board from the tapes being
played on the Video Cassette Recorder.

2. Using ‘Adobe Photoshop, Version 4’ for enhancing the quality of the captured image.

3. Converting the image to JPEG format to enable viewing of the images with a web
browser and also for importing into Microsoft Applications.

4. Transferring the images in JPEG format from the Macintosh to our mainframe
computer from where they were transferred to ‘Microsoft Powerpoint’ for generating
the presentation material and ‘Microsoft Word’ for use in this report.

The SEM and equipment photographs were all scanned using an Apple Scanner and the
quality enhanced using ‘Adobe Photoshop’. Steps 3 and 4 were then repeated for the
scanned images.

The experiments conducted are tabulated in Table 2.

35
 Table 2: Experiments Conducted.
Experiment Surfactant Concentration Oil Model Type

1. AOS 1618 0.001% Oseberg Crude B

0.01%
2. Nansa (AOS) 0.001% Kerosene B
0.01%
0.1%
1%
3. FC 740 (In Kerosene) 0.01% Kerosene A
0.1%
1%

For Experiments 2 & 3, brine was used as the aqueous phase instead of water. The
following table lists the brine formulation. This formulation is based on actual North Sea
water. The brine was filtered with a 0.45 Milli-Pore filter.

Table 3: Composition of Brine (Wetting Phase in Experiments 2 & 3)

Chemical Grams/ Liter

NaCl 24.79
MgCl2.6H2O 11.79
KCl 0.80
CaCl2.2H2O ** 1.60 **
SrCl2.6H2O 0.021
Na2SO4 4.14
NaHCO3 0.206
Formaldehyde (1000ppm) 2.7
** CaCl2.2H2O was not added because of possible precipitation problems.

36
Section 6
Results

The results of the three sets of experiments conducted are described in this section. Each
set of experiments is discussed separately, and is accompanied by images captured during
the course of the experiment.

6.1 Experiment 1

Experiment 1 was the only experiment in which crude oil (Oseberg) was used as the non-
wetting phase. The surfactant used was an Alpha Olefin Sulfonate AOS 1618 and the gas
used for foam generation was carbon dioxide. The aqueous (wetting) phase was water
and the model used was ‘B’.

The pressure difference applied to the micromodel was usually 35psig. During this
experiment we did not have the fabricated micromodel holder (Figures 7, 11) and used
the holding technique described by Lolomari et al. (1996).

37
Only two different concentrations of the AOS surfactant were used in this experiment
after which the experiment was discontinued due to excessive clogging and asphaltene
deposition.

6.1.1 Observations

The flow through the micromodel was extremely slow as compared to the flow
mentioned by Hornbrook et al. (1992) and Woody et al. (1995) even though the etch
depth for the new models (Type B) was greater then those used by Hornbrook and
Woody. The time it took to flood the model with carbon dioxide or water was in the
region of a couple of days.

Carbon dioxide was seen to bubble out through the outlet channels rather slowly. The
bubbles would originate on the extreme corner of the channel (Figures 17, 18, 21) and
would start out as very small bubbles coalescing to form bigger bubbles before flowing
towards the outlet port. The water along with dissolved carbon dioxide (during water
flood) could be seen flowing in films in the outlet channel (Figure 20).

On injecting crude oil through the inlet channel it could be seen that the crude oil did not
fill the model uniformly (Figure 22). After injection of the surfactant stable crude oil
emulsions were observed in the model (Figures 19, 26). These emulsions were seen at
concentrations of 0.01% and were observed to be stable in certain pore geometries in the
micromodel. An oil film could also be seen encapsulating the injected carbon dioxide in
the inlet port (Figure 27).

An emulsion forming sequence with the crude oil was observed where crude oil bubbles
started forming in a narrow throat and were seen being pushed into a pore where these
bubbles conjoined to form an emulsion (Figure 23). This was in the presence of 0.01%
surfactant concentration.

38
A bubble coalescence sequence was also observed where carbon dioxide bubbles coated
with crude oil were generated from a very narrow throat and were pushed into a pore
where they coalesced to form a bigger carbon dioxide bubble (Figure 24).

An emulsion breakdown sequence was also seen to occur in a pore. A very visible crude
oil emulsion broke down in a matter of seconds and was captured in the stills shown
(Figure 25).

Emulsion sizes with crude oil varied from 10µm to 60µm in size. The emulsions were
usually composed of water surrounded by oil films. The color of the water seemed to be
darker (brownish) after contact with the crude oil. This shows some still unexplained
mass transfer mechanism between oil and water.

As can be seen from the above observations, the oseberg crude oil clearly formed
emulsions in the presence of the AOS surfactant indicating the non tolerant behavior of
the oseberg crude oil to the AOS surfactant. No foam lamella were observed during the
experiment. The displacements were according to the theory of immiscible displacement:
Displacement of water by gas caused viscous fingering and displacement of water by oil
was a stable front propagation.

6.2 Experiment 2

In Experiment 2, the surfactant used was also an Alpha Olefin Sulfonate (1618) and the
gas for foam generation was carbon dioxide. The wetting phase was brine and the non-
wetting phase (oil) was kerosene. The model type used was ‘B’.

6.2.1 Observations

Oil (kerosene) emulsions were observed at as low as 0.001% concentration of surfactant

(Figure 31). During carbon dioxide flood the gas phase was discontinuous and did not
fill the model completely (Figure 32). The emulsions were very mobile at low

39
concentrations of surfactant (< 1%) as can be seen in some of the captured stills (Figures
33, 34, 35).

Not more than one gas bubble (Carbon dioxide) was seen to in one pore. In fact in some
areas one gas bubble was seen to occupy more than one pore (Figure 36). Before
surfactant flood water was seen snapping off into a kerosene bubble. No emulsion was
seen to form because of the absence of any surfactant in the system (Figure 30). The
diameter of the emulsions were smallest at lower concentrations and were seen to
increase with higher concentrations. The stability of the emulsions also increased at
higher concentrations (Figures 28, 29, 37). Certain areas in the model which had less oil
were seen to have thinner emulsions (Figure 39). Although at increased concentrations
the oil present in the model had decreased the emulsions were still widely present and
were more stable.

The emulsion size also was seen to depend on the pore geometry (Figure 38). Different
sizes of emulsions seem to co-exist but varied in size with the size being the smallest in
the throats and the smaller pores.

The emulsions were all composed of water covered by oil films. The size of the
emulsions at 0.001% surfactant concentration was as low as 1µm (Figures 33, 34) and up
to 20µm. The emulsions at this stage flowed very rapidly through the narrow throats till
they clogged the flow path.

Emulsion sizes, as mentioned earlier, also increased with surfactant concentration.

Although the pore geometry had also some effect, emulsion sizes after 0.1% surfactant
flood varied from a few microns to about 60µm (Figures 38, 39, 40, 41). Emulsion
composition after 1% surfactant flood was very similar to that seen at 0.1% except that
they were more stable (Figures 28, 29). The emulsion were hardly seen to move, as was
observed with lower concentrations because most of the narrow throats had been blocked
by the smaller emulsions.

40
Two emulsion forming and breaking sequences were also observed providing a frame by
frame view of the process (Figure 40, 41).

No foam was observed during the experiment. Also, flow through the model was
observed to be extremely slow and was comparable to the flow in Experiment 1. This
observation prompted us to investigate further into the problem. The results of the
investigation are discussed after Experiment 3.

6.3 Experiment 3

In Experiment 3, we formed an oil-foam. The surfactant was a Fluoro-Carbon, FC 740,

diluted in kerosene. The wetting phase was brine and the non-wetting phase was
kerosene. Carbon dioxide was used to generate the foam in-situ. The model type used
was ‘A’.

The exploitation of the gas-blocking properties of foam by using it as a gas-coning barrier

is an old idea. An attractive means of achieving a favorable barrier geometry is by using
low density non-aqueous foams. This is the primary reason for our focus on the oil-foam
studied in this experiment.

6.3.1 Observations

Flow through this model was extremely fast as compared to the Experiments 1 and 2. It
took only about 20 minutes to flood the model completely with carbon dioxide.

After the water flood and the oil flood, water was seen residing in the narrow throats
indicating that the model was water wet (Figure 42). Before foam was generated in the
system, the carbon dioxide phase was observed to be very discontinuous (Figure 43).
Stable foam lamellae were seen beginning at 0.01% surfactant concentration. A snap off
sequence of a carbon dioxide bubble coated with surfactant was seen (Figure 44).

41
In certain areas at 0.1% concentration of the fluoro-surfactant, a lamella formation and
breakdown sequence was captured in a five frame sequence (Figure 45).

Foam was observed as lamella all over the model at the different concentrations of the
surfactant. Certain pore geometries were seen to be more conducive to forming stable
lamellae then others (Figure 50). Many different lamellae configurations were seen
(Figures 48, 49, 53, 54). Stable lenses were also observed to co-exist with the lamellae
(Figures 47, 52).

In one configuration, a water droplet in a foam lamella did not seem to affect the lamella
stability (Figure 46). Also, in a certain pore geometry a thinner stable lamella was seen
as compared to lamella thickness in similar geometries (Figures 50, 51).

No significant increase in the number of foam lamellae were seen from 0.1% to 1.0%
surfactant concentration, in spite of a noticeably higher gas blockage as seen from flow
rate reduction.

6.4 Interfacial Tension Measurements

Interfacial tension between the surfactant solution and the oil (kerosene) was measured
for Experiments 2 and 3 and is listed below. No significant difference in the interfacial
tension was observed because of the presence of an ethyl-oil soluble dye: benzene-azo-2-
Napthol.

42
 Table 4: Interfacial Tension Measurement.

Surfactant Concentration Interfacial Tension

(dynes/cm)
Nansa (AOS) 1% 1.4
0.1% 0.71
0.01% 1.63
0.001% 7.9
FC740 (in Kerosene) 1% 14.928
0.1% 8.868
0.01% 19.686

6.5 Scanning Electron Microscopy

The flow problems encountered in experiment sets 1 and 2 as opposed to the very good
flow observed in the old model (Type A) used in experiment 3 led us to investigate the
etching and bonding process for the two models. SEM’s were taken of both the types of
models (Figures 13, 15, 16). It was observed that although the etching was similar in the
two model types, melted glass could be spotted in the SEM for model B (Figure 16). It
was also very difficult to separate the etch pattern from the Pyrex glass for usage in the
SEM machine.

This led us to conclude that the extreme bonding conditions of the Pyrex glass in the
batch of the new micromodels ( 450oC, 1500 Volts) was the reason for the reduction in
flow rates in Experiments 1 and 2. The glass had melted in sections of the silicon
micromodel reducing its permeability.

43
EXPERIMENT 1: FIGURES

44
Figure 17: CO2 bubbles flowing in the outlet channel during CO2 flood. Flow is top to bottom
diagonally.

Figure 18: CO2 bubbles have just emerged from the porous network into the outlet channel

45
Figure 19: Crude oil emulsions seen in the large pores in the model after 0.01% surfactant
flood.

Figure 20: Water films in the outlet port during water flooding. The water contains dissolved
CO2 gas.

46
Figure 21: The CO2 bubbles are generated near the corner. Flow is from top to bottom
diagonally. The bubbles coalesce to form bigger bubbles almost instantly.

Figure 22: Crude oil invading the inlet channel. It can be seen that the oil does not fill the
model uniformly.

47
Figure 23(a)

Figure 23(b)

48
Figure 23(c)

Figure 23(d)

49
Figure 23(e)

Figure 23(f): An emulsion forming sequence

50
Figure 24(a)

Figure24 (b)

51
Figure 24(c)

Figure 24 (d)

52
Figure 24(e): A bubble coalescence sequence.

53
Figure 25(a)

Figure 25 (b)

54
Figure 25(c): Emulsion breakdown sequence

55
Figure 26: A crude oil emulsion still

Figure 27 : An oil film can be seen surrounding the gas.

56
EXPERIMENT 2: FIGURES

57
Figure 28: Emulsions as seen after 1% surfactant flood.

Figure 29: Very stable emulsions seen at the end of the experiment (after 1% surfactant
flood).

58
Figure 30(a)

Figure 30 (b): Water Snapping off into a kerosene bubble. No emulsions are formed here as
there is no surfactant present.

59
Figure 31(a)

Figure 31 (b)

60
Figure 31(c)

Figure 31 (d): Emulsions seen after 0.001% surfactant flood. At this time there is only
kerosene and surfactant in the system and no gas is present.

61
Figure 32 : CO2 is discontinuous and does not flood the entire model. This is done after
0.001% Nansa surfactant flood.

62
Figure 33(a)

Figure 33(b)

63
Figure 33(c)

Figure 33(d)

64
Figure 33(e)

Figure 33(f): A sequence of moving emulsions seen during CO2 flooding (after 0.001%
Nansa surfactant flood). There is however not much CO2 present in this part of the model.

65
Figure 34 (a)

Figure 34 (b)

66
Figure 34 (c)

Figure 34(d): An emulsion moving sequence observed with the 40X magnification lens. As
can be seen there is no CO2 present in this part of the micromodel. (after 0.001% Surfactant
flood)

67
Figure 35 (a)

Figure 35 (b)

68
Figure 35 (c): An emulsion moving sequence observed after 0.01 % Nansa surfactant flood,
just before starting CO2 flood.

69
Figure 36: Both emulsions and gas at this time are flowing. This is observed after 0.1%
Nansa surfactant flood.

Figure 37: Emulsions observed near the outlet channel after 0.1% surfactant flood.

70
Figure 38: Different sizes of emulsions seen in different pore/throat sizes. Indicates that
emulsions depend on pore/throat size.

Figure 39: A very fine oil film seen in the pore (after 0.1% surfactant flood and CO2 flood).

71
Figure 40 (a)

Figure 40 (b)

72
Figure 40 (c)

Figure 40 (d)

73
Figure 40 (e): An emulsion forming and breaking sequence observed after 0.1% surfactant
flood and during CO2 Flood. (just before beginning 1% surfactant flood)

74
Figure 41 (a)

Figure 41 (b)

75
Figure 41 (c)

Figure 41 (d)

76
Figure 41 (e): Another emulsion forming and breaking sequence observed after 0.1%
surfactant flood and during CO2 Flood. (just before beginning 1% surfactant flood)

77
EXPERIMENT 3: FIGURES

78
Figure 42: Water can be seen residing in the narrow throats.

Figure 43: A lot of small CO2 bubbles coated with surfactant film seen. The CO2 phase is not
continuous any more. Seen after 0.1% FC740 surfactant flood.

79
Figure 44 (a)

Figure 44 (b)

80
Figure 44 (c)

Figure 44 (d) : A CO2 bubble snap-off sequence.

81
Figure 45 (a)

Figure 45 (b)

82
Figure 45 (c)

Figure 45 (d)

83
Figure 45 (e) : A lamella forming and breaking sequence observed after 0.1% FC740
surfactant flood.

84
Figure 46 : A drop of water in a lamella. It doesn’t seem to affect the lamella stability.

Figure 47 : A stable lens.

85
Figure 48 : A tripod like lamellae construction.

Figure 49: Four unstable lamellae seen after 0.1% FC740 surfactant flood.

86
Figure 50 : Stable lamella seen in the big pore after 0.1% surfactant flood.

Figure 51: The foam lamella seen here seems to be thinner than usually seen in this pore
geometry and is still stable.

87
Figure 52: Another stable foam lens.

Figure 53: A similar pore geometry to some earlier figures with stable lamellae in a different
configuration.

88
Figure 54(a)

Figure 54(b)

89
Figure 54(c)

Figure 54(d)

90
Figure 54(e): Some different lamellae configurations seen throughout the micromodel after
1% surfactant (FC740 in kerosene) flood.

91
Section 7
Conclusions

The silicon micromodels used in these experiments give us a realistic feel of the pore
level mechanisms which would occur in reality. The three sets of experiments conducted
in this study which included different oils and foamers lead us to some interesting
conclusions which are outlined below.

7.1 The AOS Surfactant

With the AOS surfactant in the presence of different oils, no foam formation was seen at
the pore level. However emulsions were observed which varied in size and stability with
increased concentration of surfactant. The pore and throat sizes also seemed to influence
the emulsion size. Most of the emulsions seen were composed of water covered with oil
films. With the crude oil the water became discolored to brown. Very little oil was
required to form these emulsions as could be seen at higher concentrations of surfactant
after the model had been flooded with surfactant a few times.

92
It can be concluded that the emulsion formation with the AOS surfactant could be the
reason for the delayed foam forming observed by Kristiansen et al. (1992). These have
in the past often been ascribed to so-called “aging” or “incubation” effects. According to
a mechanism suggested by Lau (1988), the emulsion scavenges all surfactant and makes
it impossible to generate foam initially. Only when excess surfactant solution becomes
available -through reinjection or rearrangement - foam generation can take place. This,
however does not apply to our case where excess surfactant is always present.

The ability to form water-in-oil emulsions is governed by the balance of forces between
those tending to keep the oil in large drops (or continuous oil areas) and those favoring
formation of small emulsified oil droplets. It is well known from antifoam theory that
small emulsion droplets make better antifoam agents. An attempt to develop a theory for
foam destabilization by emulsified oil droplets has been published by Schramm (1990).
However the lamella number fails to correlate oil tolerance data for fluorosurfactant
foams; likewise it is not understood how it can explain the lack of emulsions formed with
the fluorosurfactant.

The presence of oil in the system resulted in emulsions and no foam was observed.
However the emulsions seen were seen to be somewhat effective in blocking the gas flow
path in the system. Whether foam would actually form on continuous surfactant and gas
flooding is yet to be seen in the micromodel and would be an interesting area for further
research. Currently no theory explains this phenomenon which seems different from the
conventional “aging” of cores, which is essentially a liquid/solid adsorption process.

7.2 The Fluoro-Surfactant

For the oil-foam with the fluoro-surfactant, a strong foam was generated both in the oil-
saturated and the oil-free zones. Lamellae were observed both in oil-saturated and oil-
free areas, indicating oil-tolerant behavior as expected from the negative entering
coefficient and the bulk behavior. Generation sites for the foam lamellae were seen to be

93
controlled by pore geometry and local saturation. The foam lamellae were stabilized by
the surfactant and therefore an increase in foamer concentration increased the lifetime of
the lamellae and thereby the strength of the foam. However, no obvious link could be
found with the number of films observed and strength of gas blockage. Although lenses
were observed to be present in the system, no lens-lamella transition was observed. We
can conclude from these observations that this oil-foam can be used for achieving a very
favorable barrier geometry for gas-coning problems.

7.3 SEM Study

From the SEM study we can draw a conclusion on the fabrication technique used for the
new micromodels (Type B). As discussed in the previous section, the extreme conditions
used in the bonding process of the Corning, type 7740 pyrex wafer to the etched silicon
wafer was the cause of the reduced permeability in the new models. Extreme care should
therefore be taken while specifying the bonding conditions for the micromodels.

7.4 Further Research

Further research suggestions include studying the behavior of emulsions and how they
effect foam properties in porous media with an emphasis on foam destabilization by
emulsified foam droplets. Using different thin sections (other than Berea) to compose a
mask and etching it onto the silicon wafer would help in studying flow behavior in other
types of rocks. The effect of “aging” as the reason for the delayed foam forming (if any?)
with the AOS surfactant also deserves more investigation. As mentioned in section 7.1,
whether foam would form on continuous surfactant and gas injection would be an
interesting area to be looked into to supplement the work done in this project.

94
The new micromodels which are currently under construction are being etched at the
‘Center for Integrated Systems’ at Stanford. The anodic bonding apparatus will be set up
in the laboratory which would enable us to complete the micromodels in house. Different
etch depths varying form 15µm to 35µm would be available. Effect of varying etch
depth on the flow behavior and foam formation and stability would further enhance the
understanding of oil-foam interaction.

95
References

1. Bikerman, J.J.: Foams ; Springer-Verlag : New York, 1973

2. Coombe, D. A.; Mohammadi, S. S.: “ Characteristics of steam-foam drive process in

massive multi-zone and single-zone reservoirs.’ Paper SPE 24030 presented at SPE
California Regional Meeting, Bakersfield, 1992

3. Castanier, L. M.; Hornbrook, J.W.: Observation of foam/oil interactions in a new,

high resolution micromodel” paper SPE 22631 resented at the 66th Annual Technical
Conference and Exhibition, Dallas, 1991

4. Dalland, M.; Hanssen, J.E.: Oil Interaction with foams at static and flowing
conditions in porous media. Paper at the 13th EOR workshop and symposium, Banff
1992

5. Falls, A. H.; Hirasaki, G. J.; Patzek, P. W.: “Development of a mechanistic foam

simulator: The population balance and generation by snap-off.” SPE Reservoir
Engineering, 3: 884-892, August 1988

6. Fried, A.N.: “The Foam-Drive Process for increasing the recovery of Oil”, Report of
Investigations 5866; U.S. Department of the Interior, Bureau of Mines, Washington,
D.C. 1961

7. Hanssen, J.E.; Dalland, M.: Proceedings of the SPE/DOE 7th Enhanced Oil Recovery
Symposium; Society Of Petroleum Engineers : Richardson, TX ; paper CIM 86-37-
01, 1990

96
8. Hanssen, J. E.: “ Foam as a gas-blocking agent in petroleum reservoirs II:
Mechanisms of gas blockage.” Journal of Petroleum Science and Engineering, 135-
156. 10: 2, 1993

9. Hanssen, J. E.: “ Foam as a gas-blocking agent in petroleum reservoirs I: Empirica

Observations and parametric study.” Journal of Petroleum Science and Engineering,
117-134. 10: 2, 1993

10. Harkins, W. D.: Journal of Chem. Phys. 9, 552-568, 1941

11. Holm, L.W.; “ The Mechanisms of Gas and Liquid Flow Through Porous Media in
the Presence of Foam” , SPEJ, 359-369, December 1968

12. Hornbrook, J. W.; Castanier L. M.; Pettit, P: “ Visualizaition of foam/oil in a new

high resolution sandstone replica micromodel.” Report SUPRI TR-86, Stanford
University, 1992

13. Huh, D. G.; Cochrane, T. S.; Kovarik, F. S.: “ The Effect of Microscopic
Heterogeneity on CO2/ Foam Mobility: Part 1- A mechanistic study” SPE 17359
paper presented at the SPE/ DOE EOR Symposium, Tulsa, April 1988

14. Huh, D. G.; Cochrane, T. D.; Kovarik, F.S.: “ The effect of microscopic heterogeneity
on CO2 foam mobility: Part 1- mechanistic study.” Journal of Petroleum Technology,
872-879, 41: August 1989

15. Jensen, J.A. ; Friedmann, F. : Proceedings of the California Regional Meeting of SPE;
Society Of Petroleum Engineers : Richardson, TX ; SPE/DOE 16375, 1987

16. Kovscek, A.R.; and Radke C.J.: “ Simulation of foam transport in porous media.”
Paper SPE 26402 presented at 68th SPE Annual Technical Conference and Exhibition,
Houston, 1993

17. Kovscek A. R.; Radke C. J.: “ Fundamentals of foam transport in porous media” in
Foams: Fundamentals and Applications in the Petroleum Industry. L. L. Schramm,
Editor. Advances in Chemistry Seies, 242. American Chemical Society, Washington,
DC, 1994

18. Kovscek, A. R.; Patzek, T. W.; Radke, C.J.: “ Mechanistic Prediction of Foam
Displacement in Multidimensions: A Population Balance Approach” SPE/DOE
27789, presented at the 9th Symposium on Improved OIl Recovery, Tulsa, April 1994

19. Kristiansen, T. S.; Holt, T.: “Proceedings of the 12th International Workshop and
Symposium of the IEA Collaborative Project on Enhanced Oil Recovery”; AEA
Technology: Winfrith, United Kingdom, 1991

97
20. Kristiansen, T. S.; Holt, T.: “ Properties of flowing foam in porous media containing
oil” Paper SPE 24182 presented at the 8th SPE Symposium on Enhanced Oil
Recovery, Tulsa 1992

21. Kuhlman, M.I.: Proceedings of the SPE/DOE Enhanced Oil Recovery Symposium;
Society Of Petroleum Engineers : Richardson, TX ; SPE/DOE 17356, 1988

22. Lau, H. C.; O’Brien, S. M.: “ Effects of spreading and non-spreading oils on foam
propagation through porus media.” SPE Reservoir Engineering, 893-896, 3: August
1988

23. Lolomari, T.O.; “ Micromodel Studies of Three Phase Flow in Porous Media”, MS
Thesis, Stanford University, 1996

24. Manlowe, J.D.; Radke, C. J.: “ A Pore-Level Investigation of Foam/Oil Interactions

in Porous Media.” SPE Reservoir Engineering, November 1990.

25. Marsden, S. S.: “Foams in Porous Media”, SUPRI TR-37, Stanford University, 1986

26. Mast, R. F.: : Microscoppic behavior of foam in porous media.” Paper SPE 3997
presented at the 47th SPE Annual meeting, San Antonio, 1972

27. Nikolov, A. D.; Wasan, D. T.; Huang, D.W.; Edwards, D.A.: “The Effect of OIl on
Foam Stability, Mechanisms and Implications for Oil Displacement by Foam in
Porous Media” SPE paper 15443 presented at the 61st Annual Technical Conference
and Exhibition, New Orleans, 1986

28. Novosad, J. J.; Ionescu, E. F.: “Proceedings of the 38th Annual Technical Meeting of
CIM”; Canadian Institute of Mining, Mettalurgy, and Petroleum: Calgary, Canada;
Paper CIM 87-38-80, 1987

29. Owete O. S.; Castanier, L.M; Brigham, W.E.: : Folow of foam through porous media”
Report SUPRI TR-37, Stanford University, 1984

30. Owete, O. S.; Castanier, L.M; Brigham, W.E.: “Flow Behaior of foam: A Porous
Micromodel Study.” SPE Reservoir Engineering. 315-323, 2: August 1987

31. Radke, C.J. ; Ransohoff, T. C.: “Mechanisms of Foam Generation in Glass Bead
Packs” SPE 15441, paper presented at the 61st Annual Technical Conference and
Exhibition, New Orleans, 1986

32. Radke, C. J.; Chambers, K. T.: “Capillary phenomenon in foam flow through porous
media.” in Interfacial Phenomenon in Petroleum Recovery, N.R. Morrow, Editor.
Surfactant Science Series. Marcel Dekker, Inc. New York, pp 191-255, 1991

33. Robinson, J.W.: Woods, W. W. ; J. Soc. Che. Ind. 67, 361-365, 1948

98
34. Roof, J. G.: “ Snap-off of oil droplets in Water-Wet pores.” Society of Petroleum
Engineers Journal, 85-90. 10: March 1970

35. Ross, S.; McBain, J. W.: Ind. Eng. Chem. 36(6), 570-573, 1944

36. Sanchez, J.M.; Schechter, R.S.: “ The effect of Trace Quantities of Surfactant on
Nitrogen/Water Relative Permeabilitie” SPE 15446. Paper presented at the 61st
Annual Technical Conference, New Orleans, 1986

37. Sanchez, J.M.; Hazlett, R.D.: “ Foam Flow through Oil-Wet Porous Medium: A
Laboratory Study”, SPE 19687, paper presented at the SPE Annual Technical
Conference and Exhibition, San Antonio, 1989

38. Sanchez, J. M.; Hazlett, R. D.: SPE Res. Eng. 7(1), 91-97, 1992

39. Schramm, L. L.; Novosad, J. J.: “Micro-visualization of foam interactions with a

crude oil.” Colloids and Surfaces, 21-43, 46: 1990

40. Schramm, L.L.: Emulsions: Fundamentals and Applications in the Petroleum

Industry; American Chemical Society, Washington D.C. 1992

41. Schramm, L.L.: Foams: Fundamentals and Applications in the Petroleum Industry ;
American Chemical Society, Washington D.C. 1994

42. Surguchev, L. M.; Coombe, D.; Hanssen, J. E.; Svorstøl, I.: “Simulation of WAG and
gas Injection with potential sweep improvement by application of foam”. Paper
presented at 8th European Symposium on Improved Oil Recovery, Vienna, 1995

43. Terry, C. S.: “ A Gas Chromatography System Fabricated on a Silicon Wafer Using
Integrated Circuit Technology” Integrated Circuits Laboratories Technical Report No.
SEL-75-011, Stanford University, 1975

44. Woody, F.; Blunt, M.; Castanier, L.M.: “Pore level visualization of foam flow in a
silicon Micromodel” SUPRI TR-100, Stanford University, 1995

99
APPENDIX A

c This program computes the Interfacial tension from the spinning

c drop apparatus output.

write(*,10)
10 format(’Input speed in msec/round’)
read(*,20)speed
20 format(f7.3)

write(*,30)
30 format(’Input top level of drop in mm’)
read(*,40)top
40 format(f8.5)

write(*,50)
50 format(’Input bottom level of drop in mm’)
read(*,60)bot
60 format(f8.5)

write(*,70)
70 format(’Input density of liquid 1 in g/cc’)
read(*,80)den1
80 format(f8.5)

write(*,90)
90 format(’Input density of liquid 2 in g/cc’)
read(*,100)den2
100 format(f8.5)

r=abs(top-bot)*1e-3/2.

c converting the speed to rad/s

s=2*3.14*1000/speed

deltad=abs(den1-den2)*1000

ten=deltad*s*s*r*r*r/4
ten=ten*1e3

write(*,1)ten
1 format(’The IT between the fluids in dynes/cm is ’,f5.3)
stop

end

100
APPENDIX B

Material Information
Corning Pyrex : 7740

Description:

Glass Type - Soda Borosilicate

Color- Clear

101
102
103
104
105
106