Microvisual investigation of polymer retention on the homogeneouspore network of a micromodel

W. Yun, A.R. Kovscek n

Stanford University, Department of Energy Resources Engineering, Stanford, CA 94305, USA

a r t i c l e i n f o

Article history:Received 10 September 2014Accepted 3 February 2015Available online 19 February 2015

Keywords:Enhanced oil recoveryPolymer floodingMicrogelMicrofluidics

a b s t r a c t

A new experimental technique is reported for visualization of polymer retention on the solid surfaces ofporous media. Etched silicon micromodels with well-characterized pore networks were used duringsingle-phase flow to examine the retention of 0.2 wt% partially hydrolyzed polyacrylamide (HPAM)solution. Image analysis included an image subtraction and an RGB-based global thresholding techniquefor quantification of polymer retention/adsorption. Results are reported as the percentage of porosityoccupied by immobile polymer.

Three factors were investigated including the salinity of displacing water, the change of wettability ofthe micromodel surface, and mechanical degradation of polymer. With respect to salinity in displacingwater, the experiment confirmed that 5 wt% NaCl results in less polymer retention (6.370.3%) thanwithout NaCl (7.570.3%). The increase in Naþ concentration was sufficient to induce contraction of thesize of the flexible HPAM molecules and, therefore, decrease the thickness of polymer adsorption on thegrain. Two methods were used to alter the initially strongly water-wet surface of the micromodel.Wettability was changed by the deposition of crude oil and CTAB (cetyl trimethylammonium bromide) inaqueous solution. The micromodels treated using crude oil and CTAB showed polymer retention of15.070.3%% and 5.070.3%%, respectively. Oil-wet micromodels aged by crude oil showed largerpolymer retention than the polymer retention on water-wet micromodels (7.570.3%). Otherwise, thepolymer retention on the CTAB treated micromodel was lower than the polymer retention on the water-wet micromodel. Finally, polymer solution flowed through a 7 mm filter was tested. Average polymerretention was 4.370.3% and this is 3.270.3% point lower than the value of unfiltered polymer solution.Further investigation of microgels within pore networks provided a chance to demonstrate that a sizeand structural flexibility of microgel leading to a transition from mobile to immobile conditions.

& 2015 Elsevier B.V. All rights reserved.

1. Introduction

Mobility control is one of the most important factors in enhancedoil recovery (EOR) process design. As mobility-control agents for theaqueous phase, partially hydrolyzed polyacrylamide (HPAM) has beenused extensively in surfactant-polymer and alkali-surfactant-polymerEOR processes (Chang et al., 2006). With increasing use of polymers,improved polymer products have been rapidly manufactured withconsequent needs to characterize the behavior of EOR polymers ex-situ and in-situ. It is widely recognized that as polymer solution flowsthrough a porous medium, a portion of the polymer is retained. Inother words, both an additional resistance to flow and a loss ofpolymeric additive from the aqueous phase are driven by theretention of polymer layers on rock surfaces. Permeability reduction

during polymer flooding with a large quantity of polymer solution hasbeen reported (Treiber and Yang, 1986). Hence, analysis of polymerretention is of fundamental importance to EOR operations that involvethe flow of polymer solutions through porous media (Cohen andChrist, 1986). Most laboratory design studies of a polymer floodinvolve the measurement of the effective viscosity of polymer solu-tions in representative field cores.

Even though active research is underway to quantify accuratelypolymer molecules adsorbed at solid interfaces, such as the rocksurfaces found in petroleum reservoirs, there is little information inthe literature about qualitative and quantitative visualization ofpolymer adsorption. In other words, it is necessary to perform visualstudies in order to improve our understanding of polymer retention asan important mechanism in loss of mobility control. This experimentalstudy illustrates the successful application of silicon-wafer micromo-dels with interpretation using image processing techniques. Thus, thisstudy establishes methodology for the direct visualization of polymerretention that can be applied to more complicated pore networks.

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/petrol

Journal of Petroleum Science and Engineering

http://dx.doi.org/10.1016/j.petrol.2015.02.0040920-4105/& 2015 Elsevier B.V. All rights reserved.

n Corresponding author. Tel.: þ1 650 723 1218.E-mail address: Kovscek@stanford.edu (A.R. Kovscek).

Journal of Petroleum Science and Engineering 128 (2015) 115–127

2. Polymers for enhanced oil recovery (EOR)

There are several important characteristics that must be testedbefore injecting polymer solution into a formation. These include lowcost, good filtration properties, suitable viscosity, surfactant compat-ibility, salinity and calcium tolerance, thermal stability, mechanicalstability, good injectivity, good transport in reservoir rocks, and lowretention on rock surfaces. Among the characteristics, this studyfocused on the transport and retention of polymers. Polymers in thisstudy are polyacrylamides. They are water-soluble polymers andproduced for many different purposes. Water-soluble polymers areproduced by the hydration (hydrolysis) of the chain of monomeracrylamide. During the hydrolysis, some of the amide (CONH2)groups react and cause anionic carboxyl groups (COO�) to bescattered along the backbone chain. The properties of polyacrylamidein water solution are strongly related to the degree of hydrolysisdetermined by the amount of carboxyl group (COO�) in the moleculechain. The negatively charged carboxyl groups on the backbone chaincause anionic repulsion between polymer molecules and segmentson the same molecules. The repulsion is the main reason for theelongation of polymer molecules; hence, the mobility reduction(viscosity increase) takes place. Specifically, the HPAM macromole-cule has a structural flexibility that makes it very sensitive to theionic environment.

Cations, especially divalent cations, have screening effects on theelectrostatic repulsions between chargedmonomers, that decrease theexcluded volume and, as such, the macromolecular size. In otherwords, greater viscosity is obtained when the polymer molecule isuncoiled in a low salinity environment occupying the largest possiblevolume. In practice, the salt sensitivity of HPAM has drawn attentionfor EOR processes because the reservoir is an ion-rich environment. Interms of the rheological behavior of polymer solutions, HPAM showsviscoelastic behavior even when fairly dilute due to its flexible coilmolecular structure. On the other hand, the more rigid rod-typexanthan structure yields a purely pseudoplastic and inelastic solutionat lower concentrations. Such structural differences give rise todifferent aspects of polymer retention of xanthan and HPAM, asdiscussed in relation to Experiment IV.

2.1. Polymer degradation

Polymer degradation mechanisms include mechanical degrada-tion, chemical degradation, and biological degradation. Mechanicaldegradation occurs when the polymer solution is exposed to largeshear rates. Such conditions arise, during the mixing of polymersolution, flow through chokes, injection through perforations or thenear well bore area where flow velocity is very high. Seright (1983)showed a polyacrylamide that is very sensitive to shear degradation.In that study, the viscosity–shear rate curve of a given polyacrylamidesolution is shown before and after different levels of shearing througha consolidated sandstone core. After fairly modest levels of shearingfor the polyacrylamide solution (above 30 feet/day), the viscosity isconsiderably reduced. After extreme shearing at a very large flow ratethrough the sandstone core, the viscosity is only slightly above that ofthe brine. Later, Zechner et al. (2013) also investigated shear-thinning

behavior of polyacrylamide solutions flowing through fractures toexplain measured injectivity.

2.2. Salinity/hardness

As Sorbie (1991) pointed out, the relative viscosity of HPAM is afunction of both salinity (Naþ , C1� , etc.) and hardness (Ca2þ , Mg2þ ,etc.). The repulsion among the backbone charges is screened by thelocal double layer formed by the small electrolyte species. At greatersalt concentrations, the screening effect is more marked, and conse-quently the viscosity is lower. The effect of divalent ions, such as Ca2þ

and Mg2þ , is even more significant than that of monovalent species,such as Naþ and Kþ . The divalent ions bind even more tightly to thepolyelectrolyte because of their greater charge density and polariz-ability. Sorbie also reported the study of Sandvik and Maerker (1977)showing Ca2þ has a much greater effect on the reduction of intrinsicviscosity of HPAM than Naþ .

2.3. Polymer retention

Dominant polymer retention mechanisms were explained by Huhet al. (1990). Polymer retention occurs predominantly via twomechanisms including the adsorption of polymer molecules on thesurfaces of large pores, and mechanical entrapment in small pores.The first mechanism for mechanical entrapment is “straining” (Huhet al., 1990). Mechanical entrapment can be described by theaccumulation of polymer molecules in the pore channel whose radiusis comparable to or smaller than an average size of polymer molecule.Even though the fact that the size of the pore channel is smaller thanthat of polymer molecule, a flexibility of the polymer molecule allowsthe solvent to force the polymer molecules into the smaller channel.The polymer chain is then trapped via adsorption at opposite walls ofthe pore. Consequently, the number of trapped polymer chainincreases as the solvent flows through the channel and brings morepolymer into the channel.

Secondly, Maerker (1973) has proposed hydrodynamic retentionthat is associated with a “sticky” surface. As the solvent brings morepolymer molecules into a small pore, a large chemical potentialgradient is generated. Due to large chemical potential, the channelreleases some trapped polymer molecules without continuouslyflowing solvent. On the other hand, the “sticky” surface with largeadsorption energy prevents the chemical potential gradient fromcausing outward release of polymer when the flow of solvent stops.Alternative to these mechanisms, DeGennes (1979) describes that asflow continues, a macromolecule remains statistically adsorbed beca-use the desorbing “trains” of polymer are always replaced by adsorb-ing “loops”. Consequently, all retention mechanisms are interrelatedand contribute to the trapping of polymer in small pores and beingadsorbed on the surface of grains.

Earlier studies indicated how significantly the retention mechan-isms contribute to the polymer retention in different pore structures.Polymer retention in very permeable porous media should be duemainly to adsorption and the level of polymer retention is notsignificantly high. For instance, the xanthan retention levels obt-ained (Sorbie et al., 1987) in permeable Clashach sandstone cores

Nomenclature

M molecular weight, g/molA cross sectional area, cm2

Δp pressure dropμw viscosity of water, cPq volumetric flow rate, cm3/s

L length, cmk absolute permeability, DarcycP centipoises secondnpw_p the number of white pixels from polymer retention

image andnpw_b the number of white pixels from the base image

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127116

(850–1850md) range from 0.2 to 3 μg/g of rock; however, thepolymer retention levels in low permeability rock are usually muchgreater than the polymer retention in high permeability rock (Cohenand Christ, 1986). Therefore, the contribution only from adsorption isgenerally believed to be small in permeable rock. The mechanicalentrapment, which additionally occurs in small pores (low perme-ability), makes a significant contribution to polymer retention alongwith the adsorption.

2.4. Microgel

Although there is no precise and limiting definition of the termmicrogel, often referred to as a hydrogel, the literature has proposed adefinition of microgel as a material made when a water-insolublepolymer absorbs a large amount of water, or else it is simply a water-swollen polymer network (Gibas et al., 2010). Microgel can also bedefined as a permanent or chemical gel stabilized by covalently cross-linked networks. Cross-linking may be either physical (e.g. hydrogenbonding) or chemical (covalent, atomic, and ionic). Swelling of micr-ogel takes place through several steps including the hydration of thehydrophilic group with water in the micro-gel matrix followed by theinteraction between hydrophobic groups with water molecules. Anadditional amount of water is absorbed into the network because theosmotic driving force of the network is resisted by the covalent crosslinks. The temperature and the specific interaction between watermolecules and polymer chains have significant influence the totalamount of water absorbed by microgel (Pradas et al., 2001).

With such a structure, microgels are able to swell, absorbing a largeamount of water without the polymer dissolving, which gives thempolymer-like rheological properties that can affect EOR significantly. Arecent study (Dong et al., 2009) reported the influence of polymermicro-gel concentration, salinity of water, reservoir temperature andswelling time on the rheological properties of polymer microgeldispersions, that is a water/oil (W/O) microemulsion containing poly-mer microgel. The results showed that with increasing swelling timeand NaCl concentration, the polymer microgel dispersions changedfrom a shear-thickening fluid to a Newtonian fluid. To explain, thepolymer microgel tends to be more rigid with a smaller amount ofwater adsorbed than the gel with a larger amount of water adsorbed.By increasing the swelling time, more water entered the polymermicrogel particles, the hydration degree and apparent size of theparticles increased. Due to the relatively large size of microgel, thetrapping of polymer microgels can take place in the pore space. Thetrapped polymer microgel in the larger pore space becomes a drivingforce to direct the polymer solution to flow into the smaller orrestricted channel.

2.5. Summary

In surveying the literature on polymer adsorption in porousmedia as it pertains to oil recovery applications, it is evident thatthe level and nature of polymer adsorption depends mainly onthree factors. First, the polymer, of which the most importantfactors are the type of polymer (xanthan or HPAM) and the polymerproperties such as molecular weight, molecular size and chargedensity or degree of hydrolysis (for HPAM). Second, the solvent(only aqueous solutions are considered here) pH, salinity (Naþ ,C1� , etc) and hardness (Ca2þ , Mg2þ , etc.) are the most significantfactors. The presence of other species in solution, such as alcohols,may also affect solvent quality and hence the level of polymeradsorption. Third, the surface area and type of surface (silica,calcium carbonate, clay, etc.) are very significant factors. The surfacecharge is also an important factor in determining polymer adsorp-tion. There are usually clear differences between static adsorptionon powders and sands and adsorption in packed cores or

consolidated sandstones. Pretreatment by oil or other adsorbingspecies such as surfactants, or by fluids that change the wettabilityof the adsorbing surface, can also be important in this respect.

In this study, the experiments were designed to test the effect ofthe three factors discussed above. In particular, having used a constantconcentration of HPAM, filtrations with different filter size wereperformed in order to cause mechanical degradation. In turn, degra-dation caused the reduction of polymer solution viscoelastic property.Moreover, concentration of NaCl and wettability of the surface of themicromodel were modified to investigate the effect of the second andthird factors on polymer retention.

3. Experimental

Fig. 1 demonstrates the setup including a microscope, camera,pump, fluid transfer vessels, a micromodel, and a pressure gauge. Thedecane inside the pump was controlled by three-way values in orderto flow into the top of the one of three vessels containing isopropanol(IPA), water with a dye sensitive to UV light, and HPAM solution. Thedecane forced the fluid in each vessel to flow out of the bottom. Thefluid flowed into the micromodel and occupied all accessible porespace. The pressure gauge was used to monitor any pressure changes.The syringe pump was a Teledyne Isco Model 100 DM. Pump capacityis 103 mL. Decane was the only fluid in the pump and was used topump the other fluids in the vessel system throughout the experi-ment. A Nikon Eclipse ME 600 reflected-light microscope with a metalhalide lamp for illumination was used for direct observation of themicromodel. An adapter mounted on the microscope allowed acamera to be installed on top and to take the images of the porespaces. Three different magnifications were used: 40� , 100� and200� . 200� was used extensively.

A micromodel is a two-dimensional pore system that simplifiesand imitates the pore structure of real porous media. Micromodelsallow microscopic analysis of fluid flow. Our micromodel consists of asilicon wafer etched to a predetermined depth and bonded to a glasscover plate. In this research, a micromodel with a 5 cm by 5 cmsurface area was used; it has four ports at the corners and two flowchannels as shown in Fig. 2. A micromodel holder set was designed forthe purposes of protecting the micromodel and facilitating theconnection of the wafer to the vessel system. The micromodel holderconsists of two aluminum plates with four ports and O-rings to sealproperly the connection of ports between the holder and the wafer.The upper plates are placed on the wafer and secured with eightscrews carefully tightened in the way of applying balanced pressureon the micromodel to avoid a leak between an O-ring and micr-omodel.

Fig. 1. A schematic diagram of the setup.

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127 117

3.1. Micromodel fabrication and characterization

Micromodel fabrication takes place in the Stanford NanofabricationFacility (SNF). As a first step, the prime oven dehydrated the siliconwafers at 150 1C for 30 min and primed the wafers using HMDS(hexamethyldisilazane) allowing better coverage and adhesionbetween oxides and resists. The primed wafer was then coated withphotoresist using the SVG (Silicon Valley Group) coater, which is anautomated track system for dispensing photoresist on 4″ siliconwafers. After this coating process, the wafers were exposed toultraviolet light with the mask containing the pore network. Becauseit is a 1:1 exposure system, the resolution of the exposed patterns onthe wafer is determined by the resolution of the mask used.

Once the wafers were exposed, the wafers were processed throughthe SVG developer for developing and post-baking exposed photo-resist-coated 4″ wafers. In order to finalize the pore network on thedeveloped wafers, wafers were etched to about 20 μm using anInductive Charged Plasma Deep Reactive Ion etcher. The remainingphotoresist on the wafers was then removed by soaking the wafers ina chemical bath of “piranha” (90% sulfuric acid/hydrogen peroxide) for20 min. The etched wafers without any photoresist now could beprocessed in the Zygo White-Light 3D Surface Profiler for characteriz-ing surface topography of the wafers (Fig. 3). The Zygo provides fast,non-destructive, quantitative surface characterization of step heights,texture, roughness, and other surface topography parameters. TheZygo was utilized to characterize the micro-pore network, flow cha-nnels, and grain configurations. Fig. 3 demonstrates that the micro-models in this study had a simplified pore network structure thatconsisted of staggered uniform cylinders 40 μm in diameter. Thecylinders are separated by 10 mm fromwall to wall and the etch depthis about 22.0 mm. Porosity is measured as 42%.

Four ports (inlet and outlet) were drilled into the wafers. Oncethe ports had been drilled, the wafers went through a thoroughcleaning process again to avoid the possibility of remainingphotoresist or organic/inorganic precipitations that might blockthe microchannel on the wafers. Finally, a glass wafer wasanodically bonded to the top of the clean wafers at 1200 V and300 1C for 0.5–1 h (Buchgraber et al., 2012).

3.2. Procedure

An HPAM sample (FP 3630 SNF Floerger) with a molecular weightof 20 million Dalton and a concentration of 2000 ppm (0.2 wt%) in DIwater was studied. The usual amount of hydrolysis for EOR polymerproducts is between 25% and 30%. The degree of hydrolysis of thepolymer in this study falls in this range. The aqueous polymer solutionwas prepared by slowly adding the powder onto the shoulder of thevortex of DI water and, at the same time, stirring the mixture with afloating magnetic stir bar at 360–400 rpm. More importantly, foradequate dispersal of polymer powder, the addition process was

completed within 2–3min before the solution developed too muchviscosity (Levitt, 2009). A nitrogen gas blanket minimized contact withoxygen. Polymer solutionwas stirred continuously at 100–200 rpm forat least 48 h at room temperature to ensure full hydration of thepolymer powder. As the polymer hydrates, the solution becomes morehomogenous and clear. Filtration was carried out in order to ensurethat proper hydration of polymers had been achieved. Nitrogen wasinjected into the column of polymer solution at 40-psi pressure. Then,the nitrogen pushed approximately 250 mL of aqueous polymersolution into a 2-μm stainless steel precolumn filter. Polymer solutionswith a filtration of different size of filter or without any filtration werealso tested in order to investigate the effect of filtration on polymerretention. A Brookfield LVDV-II-Proþ viscometer was used for viscos-ity measurements.

In general, the four entry ports of the holder and the O-rings canbe freed from any contamination using a vacuum pump, toluene, IPA,and water. Due to the retention of polymer in the micromodel and itsirreversibility, however, the use of a micromodel with polymer insidewas not acceptable. Thus, a newmicromodel was used for every otherexperiment. A first step was to saturate fully the micromodel with UV-dyed water. As shown in Fig. 1, the decane inside of the pump wasdriven to flow with a constant flow rate of 0.001 ml/min. Once themicromodel was saturated with the UV-dyed water, a base image ofthe pore-network was taken. A base image was then used for imageanalysis. As a means of removing the UV-dyed water saturated in themicromodel, copious amounts (100 Pore Volume) of DI water wereflooded through the micromodel. Images were taken under UV blacklight after the DI-water flooding for removing dye molecules. Imagestaken after dye removal confirmed no measureable amount offluorescent materials left in micromodel. During the water flooding,the permeability of the micromodel was calculated using the incom-pressible form of Darcy's law as

k¼ � qμw

A

� � LΔp

� �ð1Þ

where A is the cross sectional area, Δp is the pressure drop, μ is theviscosity of water, q is the volumetric flow rate, L is the length, and k isthe absolute permeability. Several steady state pressure drops acrossthe micromodel were measured at specific flow rates. Using thedimensions of the micromodel, known constants, and measuredvariables, the plot (qμwL) versus (AΔP) was constructed to find aslope that is equal to permeability k.

Once the UV-dyed water inside of the micromodel was removedby the DI water flooding, polymer solution was injected into themicromodel at 0.001 ml/min until the micromodel was fully saturatedwith polymer solution and any retention was satisfied. After thesaturation with polymer solution, the UV-dyed solution was flooded

I-A

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

I-B

O-A

O-B

Flow distribution

channel/fracture

Flow distribution

channel/fracture

Injection port

Productionport

Fig. 2. Sketch of the positions where the images were taken during the experi-ments. Injection ports and production ports are labeled as I-A, I-B, O-A, and O-B,respectively.

Fig. 3. Surface profile of the micromodel.

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127118

into the micromodel at a volumetric flow, 0.005 ml/min (E7m/day).Low flow rates avoided rapid increases in inlet pressure caused by theresistance of flow through micromodel pore space saturated withviscous polymer. Hence, micromodel breakage from over pressuriza-tion was minimized. Once the UV dyed water broke through in theproduction channel and the inlet pressure stabilized, images of thegrains of the homogeneous micromodel were taken and visualinvestigation of polymer retention was performed by comparison ofthe base and the retention images.

4. Image analysis

An image analysis method was initiated due to the difficulty ofconducting precise material balance calculations. The pore volume ofthe micromodel is exceptionally small and calculated based on theporosity and the etching depth. In Fig. 4, pictures are taken at ninedifferent locations across the surface of the micromodel. In this paper,homogeneous patterned micromodels were used. Hence, experimen-tal conditions (grain size and pore structure) for all nine locations inthe micromodel are the same. Due to the high mobility ratio of waterdisplacing polymer solution, unstable displacement took place. Theimages at the locations on the micromodel allowed observation of anyeffect of unstable displacement of polymer on the polymer retention.The image after UV-dyed water flooding through a clean micromodelwas taken as a “base image” representing the grain and pore spacewith black and green color, respectively. Once the base image wasobtained, an image was taken after UV-dyed water flooding throughthe polymer-saturated micromodel. This image is called the “polymerretention image.” It was not necessary for both images to be takenfrom the same location of the micromodel due to the homogeneity ofthe pore network in the micromodel (Fig. 3). Images are analyzedusing ImageJ software and Matlab. A threshold tool in ImageJ utilized

an RGB-based global thresholding technique (Otsu, 1979) to convertcolor images into the binary (white and black) images.

In Fig. 4, the thresholding tool is activated and the windowappears. The histogram represents the distribution of pixel intensitiesin the image in RGB mode, black towards the left (0) and white at theright (255). In the case of ‘Green’ on the right in Fig. 4, all the pixelsbetween 71 and 255 were transformed from green to white. Binaryimages from the base and polymer retention images are subtractedfrom each other for the visualization of the polymer attached to thegrain wall. In the image subtraction process, it was very important toalign carefully two images in order to find the location where thepolymer retention took place. Once the subtracted image has beenobtained, the image was further analyzed by Matlab to calculate thevalue of percent polymer retained or adsorbed. Fig. 2 shows theapproximate positions at which the images were taken.

Having both base and polymer retention images after theimage subtraction process, the number of pixels of white or blackwere counted and used to calculate a percent polymer retention as

Retention%¼ npw_p

npw_b

� �ð100%Þ ð2Þ

where npw_p is the number of white pixels from the polymer retentionimage and npw_b is the number of white pixels from the base image.Eq. (2) yields the percentage of the pore space occupied by polymerand is somewhat different from other measures of retention based onthe mass retained. In this process, the 2D image is assumed to be therepresentation of 3D structure of the entire depth of micromodel.More details are discussed in Section 3.

Numerical experimentation revealed that the most sensitiveoperation during image analysis is determination of the value of‘Green’ during thresholding. Propagation of uncertainty caused by thedetermination of the value of ‘Green’ in the image analysis procedureresulted in 70.3% point in the calculation of percent retention. A

Fig. 4. Parameter setting using the threshold tool of ImageJ software showing the Otsu method based on RGB value (right) for a binary image transformation (left).

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127 119

sensitivity analysis of the image analysis process using representativeelementary volume (REV) is reported in the Appendix.

5. Results and discussion

The various experiments are discussed in turn beginning withan overview of the procedure.

5.1. Experiment I: overview of experimental procedure and results

Viscosity of FP 3630S at a concentration of 0.2 wt% in DI water wasmeasured and tabulated in Table 1. The filtered solution decreasedsubstantially in viscosity as the shear rate increased. The micromodelpermeability was measured during the process of water floodingbefore polymer solution was injected. The permeability was obtainedas 1.205 Darcy.

After saturation of the micromodel with UV-dyed water, baseimages were taken for use in porosity calculation and image subtrac-tion. The polymer injection process was followed by UV-dyed water

injection. Viscous fingering was observed at the beginning stage ofUV-dyed water injection. The diffusion of UV-dyed water into polymersolution, however, took place as the steady state approached. Fig. 5provides evidence of trappedmicrogel in the pores of the micromodel.Although the polymer was filtered through a 2 μm filter, someamount of microgel remained in polymer solution, blocked the porenetwork, and forced the polymer solution to flow into the otherchannel. Presumably, the microgel formed after the filter.

5.2. Experiment II: observation of microgel

In order to investigate the process of immobilization of polymermicrogel, an unfiltered 2000 ppm (0.2 wt%) polymer solution wasinjected into a micromodel followed by UV-dyed water injection at0.0008 ml/min which is equal to the superficial velocity at 1 m/day.Figs. 6–8, and 9 show that microgel was successfully observed usingimage analysis. As shown in Fig. 11, the large polymer microgel(410 μm) possessed structural flexibility during the polymer injec-tion at constant pressure drop (60 psi). Hence, it changed its shape asit flowed through the micropore channels. Eventually, the microgelbecame immobile at a pressure drop less than 60 psi where pressuregradient, frictional force, and internal energy of structural changeacted to stabilize the polymer microgel.

mportantly, this structural flexibility of microgel implies a possibi-lity for polymer solution to flow into so-called inaccessible porevolume. Image (3-A) has been successfully converted into a binaryimage (3-B) for more convenient visual observation. Even at a largerflow rate, 0.02 ml/min, 15 mm filtered polymer has rigid polymermicro-gel that blocks the pore channel in Fig. 10. In order to mimic theactual field operation of filtration, 2000 ppm polymer was filteredthrough a sand-pack. Filtered polymer was injected at constantpressure, 30 psi. As shown in Fig. 12, immobile microgel 1 holds sand

Table 1Viscosity of the filtered or unfiltered polymer solution.

0.2 wt% FP 3630S Viscosity Shear rate

(cP) (s�1)

No filtration 1690730 1.3234974 9.9

Filtration with 7 mm filter 1650730 1.3227074 9.9

Filtration with 2 mm filter 1620730 1.3218474 9.9

Fig. 5. Microgel trapped inside of micromodel. Right image is at greater magnification and shows a black micro-gel between 4 grains.

Fig. 6. Image of 2000 ppm FP 3630S after filtration with a 2 μm filter showing immobile microgel. 23.770.3% polymer retention including microgel. Far right image showsretained polymer as white.

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127120

particles that came from the sand pack and its interface is more rigid.Immobilization of microgel 2 is initiated by stationary sand particlesbetween the walls. Variability of its size has been observed in Fig. 12where the smaller size microgel (o10 μm) flows through the micro-channel. Any variability in pore size gives the resistance against itsflow (Figs. 11 and 12).

5.3. Experiment III: effect of salinity on polymer retention

The influence of salinity on polymer retention was tested. Polymerretention images after 0 wt% NaCl and 5 wt% NaCl dyed water

flooding are compared. The polymer retention on the walls of grainsat different locations of the micromodel was visualized. Figs. 13 and 14present the measured values of percentage polymer retention atlocations throughout the micomodel. As various studies proposed,brines containing high concentrations of divalent magnesium andcalcium ions may accelerate the degradation process by precipitationof petroleum sulfonates (Du and Guan, 2004). The most sensitivepolymer to brine and multivalent ions is hydrolyzed polyacrylamide.This sensitivity results in a loss of viscosity and thus increasingmobility (Kim et al., 2010). As the literature proposed, the polymerdisplaced by 5 wt% NaCl dyed-water has less retention on the porewalls as compared to the polymer retention of very low salinity water.Both figures show relatively low polymer retention at location (3) thatis the closet to the inlet (I-B) of dyed-water injection. This indicatesthat location (3) has the greatest induced shear stress by thedisplacing dyed-water and has the longest time to be displaced untilthe dyed water break through to the outlet flow distribution channel.

The increase in Naþ concentration was sufficient to decrease thethickness of polymer adsorption on the grainwall. The literature states

Fig. 7. Image of 2000 ppm FP 3630S after filtration with a 7 μm filter showing immobile microgel. Far right image shows retained polymer as white.

Fig. 8. Image of 2000 ppm FP 3630S after filtration with a 7 μm filter showing immobile microgel. 7.770.3% Polymer retention including microgel. Far right image showsretained polymer as white.

Fig. 9. Image of 2000 ppm FP 3630S with Sand-Pack filter without microgelobservation. 4.870.3% polymer retention. Retained polymer is white.

Fig. 10. Immobile microgel of 2000 ppm FP 3630 filtered through a 15 mm filterduring the UV-dyed water flooding at 0.02 ml/min.

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127 121

that the increase in ionic strength of the 5 wt% NaCl dyed solutioncauses two related effects (Sorbie, 1991). First, salinity contracts thesize of the flexible HPAM molecules because of the screening effect ofcations. This allows more polymer to fit sterically onto the surface andmeans that there will be less loss of conformational entropy of thepolymer chain on adsorption. A closely related second point is that thesize of the polymer chain in solution became smaller as saltconcentration increased and the shorter polymer molecules occupiedless volume of solution. The effects of the polymer–solid surfaceinteraction became more important that the interaction betweenpolymer and solvent, thus leading to greater adsorption levels.

The second effect seems opposite to the result from this studywhere increased salinity reduced the volume of the pore space filledby immobile polymer. Reduced volumetric occupancy of pore space bypolymer, however, is consistent with proposed contraction of HPAMmolecules. The method of calculating polymer retention in thismicrovisual study as the percentage of the pore space filled withimmobile polymer differs from mass balance measurements. In otherwords, the retained polymer density may increase with salinity eventhough thickness of the adsorbed polymer is smaller. This differenceseems to explain the trend of polymer retention as influenced bysalinity as measured here.

Fig. 11. Mobile microgel in unfiltered 2000 ppm FP 3630 in DI water becomes immobilized. Microgel labeled 1 transferred to the location labeled 2.

Fig. 12. Sand pack filtered polymer with microgel holding (1) or held (2) by sand particles. Mobile micro-gel due to its small size during UV dyed water injection (0.02 ml/min).

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127122

5.4. Experiment IV: effect of wettability change on polymer retention

Twomethods were used to change wettability of the surface of themicromodel. Micromodel surfaces were treated through exposure tocrude oil (Saudi Arabia) and CTAB (cetyl trimethylammonium bro-mide) cationic surfactant. The acid and base numbers of the crude oilwere measured using the technique described by Fan and Buckley(2006). The acid number was 1.76 mgKOH/g and the base number

was 1.77 mgKOH/g. No further characterization of the oil wasconducted.

In the first method, the wettability of a micromodel changed as aresult of being flooded with crude oil and aged at room temperaturewith no initial water saturation. Specifically, a micromodel wassaturated with crude oil for 48 h and the crude oil was then displacedby n-decane. Once the crude oil had been removed by n-decane, n-decane was removed by CO2 injection and dried for 12 h. In thesecond method, a cetyl trimethyl ammonium cation (CTAþ) and Br-anion were dissociated from CTAB in solution. CTAþ adsorbed on thenegatively charged SiO2 surface of the micromodel to form a CTAþadsorption layer. Based on a study done by Bi et al. (2003), theconcentration of CTAB in water must be carefully determined in orderto achieve a successful wettability alteration from water wet tomixed wet.

The concentration determined by compact adsorption mono-layer at 5�10�4 mol/dm3 (¼0.18 g of CTAB per 1000 ml of water)shows a larger contact angle indicating the wettability change toneutral wet from water wet (Bi et al., 2003). This is because thehydrophobic group of CTAþ surfactant is toward the water phase.A concentration greater than 5�10�4 mol/dm3 results in bilayeradsorption. The hydrophobic group of the surfactant is orientedtoward the first layer making the silicon dioxide surface hydro-philic. The determined concentration of CTAB aqueous solutionthen was injected into a micromodel and placed in a 50 1C vacuumoven for 48 h to deposit CTAþ and dry out the liquid phase. Once amicromodel has been treated by the two methods explainedabove, the observation of wettability alteration and polymerretention were carried out.

In Fig. 15, wettability alteration techniques were tested byinjection of green UV-dyed water into the dry crude oil treatedmicromodel. Presumably crude-oil components, such as asphal-tenes and maltenes, adsorb to the solid thereby changing wett-ability in the absence of an aqueous phase (Kovscek et al., 1993). Asindicated elsewhere (Yun, 2014), water saturation by spontaneousimbibition for both cases are compared and the water saturation inCTAB treated micromodel was smaller than a water-wet micro-model. This is good indication of a reduction in water wettability.Moreover, Yun (2014) reported that CTAB treated micromodelsshow a water–air contact angle near 831. Pore-level observationsof increased contact angle within micromodels confirmed wett-ability alteration.

The polymer retention at one specific location of the micromodelwith crude-oil deposition was measured as 15.070.3%. On the otherhand, polymer retention on nine different locations of CTAB treatedmicromodel is shown in Fig. 16. Polymer retention on the CTABtreated micromodel was, on average, 5.070.3%. Lakatos et al. (1979)studied the effect of adsorbent surface wettability on the adsorption ofHPAM in sandpacks, it was reported that chemically siliconised oil-wet surfaces had less retention of HPAM than the retention of water-wet surfaces. The result of lower polymer retention within the CTAB

Fig. 13. Percent polymer retention at nine locations in a micromodel/5 wt% UV-dyed water with 5 wt% NaCl was flooded after the injection of 2000 ppm unfilteredFP3630S polymer in DI water at Q¼0.005 ml/min with all outlets (O-A and O-B) areopen. UV-dyed water injection into I-B port/Average polymer retention is6.370.3%.

Fig. 14. Percent polymer retention at nine locations in a micromodel/5 wt% UV-dyed water with 0 wt% NaCl was flooded after the injection of 2000 ppm unfilteredFP3630S polymer in DI water at Q¼0.005 ml/min with all outlets (O-A and O-B)open. UV-dyed water injection into I-B port/Average polymer retention is7.570.3%.

Fig. 15. Wettability alteration contact angle showing oil-wet (solid circle) oil phase trapped in smaller pore (dashed circle).

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127 123

pretreated micromodel relative to the base case (7.570.3%) withwater wet surfaces is analogous to the previous study.

Interestingly, the micromodel treated with crude oil showedgreater polymer retention than the retention of the base case.Having greater retention of HPAM polymer on the crude oil treatedmicromodel is opposite to the study (Kolodziej, 1988) showing alower level of xanthan adsorption at residual oil saturation. This isbecause rigid rod-type xanthan molecule resulted in a layerdepleted in polymer near pore wall due to the molecular surfaceexclusion (Sorbie, 1991).

On the other hand, the flexible coil molecular structure of HPAMlowered the molecular surface exclusion (slip effect). That is, in thecrude-oil deposited micromodel, there are both chemical andmechanical aspects of polymer adsorption. The change of surfaceroughness of the micromodel by the deposition of crude oil resulted inthe mechanical entrapment (less slip effect) of flexible coil molecule ofHPAM. In addition, this greater adsorption is attributed to the increasein the interaction between polymeric carboxyl groups and organicbases in the crude oil.

5.5. Experiment V: effect of filtration on polymer retention

In field polymer treatments, the polymer solution is preparedon site from a broth, gel, or powdered solid, and this would

typically be filtered at a very high rate through membrane or sandfilters. In Sorbie (1991), the research summarized (Sandvik andMaerker, 1977; Willhite and Dominguez, 1977) demonstrates thatpreparation of the solution of synthetic polymers such as HPAMneeds additional care in terms of avoiding residual pore blockingafter adsorption. Specifically, small quantities of microgel maysignificantly affect the observed levels of polymer adsorption andretention. For these, it is usually necessary to prepare solutionsthat are free of all debris and microgel by filtration. During

Fig. 16. Percent polymer retention at nine locations in a CTAB treated micromodel/5 wt% UV-dyed water with 0 wt% NaCl was flooded after the injection of 2000 ppmunfiltered FP3630S polymer in DI water at Q¼0.005 ml/min with all outlets (O-Aand O-B) open. UV-dyed water injection into I-B port/Average polymer retention is5.070.3%.

Fig. 17. Percent polymer retention at nine locations in a micromodel/5 wt% UV-dyed water without salinity was flooded after 2000 ppm 7 μm filtered FP3630Spolymer in DI water at Q¼0.005 ml/min with all outlets (O-A and O-B) open. UV-dyed water injection into I-A port/Average polymer retention is 4.370.3%.

Fig. 18. Percent polymer retention at nine locations in a micromodel/5 wt% Greendyed water without salinity was flooded after the injection of the 2000 ppm 2 mmfiltered FP3630S polymer in DI water at Q¼0.005 ml/min with all outlets (O-A andO-B) open. UV-dyed water injection into I-A port/Average polymer retention is7.870.3%.

Table 2Summary of polymer retention.

Experimental scenarios Polymer retention

(%)

1. Base case 7.570.32. High salinity 6.370.33. Mechanical degradation 7 mm filter: 4.370.3

2 mm filter: 7.670.34. Wettability modification with crude oil (28)a 15.070.35. Wettability modification with CTAB 5.070.3

a Retention is observed in the single location of micromodel.

Fig. 19. Calculated porosity for different size of domain of micromodel showingthat a representative elemental volume (REV) is obtained using more than 6�106

pixels.

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127124

filtration, the HPAM polymer solution is sheared to degrademechanically the polymer backbone, thereby reducing the shearviscosity. In order to mimic the field polymer treatment forincreasing injectivity of polymer solution, the polymer solutionhas been very efficiently filtered by 2 mm and 7 mm filter. Thefiltered and unfiltered polymer solutions were compared for theeffect of degradation on polymer adsorption/retention. As shownin Fig. 17, the histogram of percent polymer retention at ninedifferent locations in the micromodel is demonstrated. The imagesubtraction technique shows that 7 μm filtered polymer has anaverage polymer retention of 4.370.3%. Fig. 18 presents thepercent polymer retention of 2-mm filtered polymer at nine diffe-rent locations of the micromodel. The average polymer retention is7.870.3%.

Compared to the value of polymer retention of unfiltered polymer,which was 7.570.3%, the amount of polymer retention of 7 mmfiltered polymer was 3.2% points lower. The lower value of polymerretention is explained by the mechanical degradation that takes placeduring the filtration process. When the polymer solution flowsthrough the series of porous media, 7 mm filter in this study, the flowresistance increases greatly as a result of the sharp increase in theextensional (elongational) viscosity within the porous structure. Thus,the sharp increase in elongational viscosity leads to very high stresses

on the polymer molecules that, in turn, mechanically pull themolecule apart. Larger molecules offer more resistance to flow andconsequently experience greater shearing or elongational stresses andare therefore more likely to break. As a result of molecular chainscission, the larger molecular weight species in the molecular weightdistribution (MWD) are broken down into some combination of thelower molecular weight fragments, leading to redistributed molecularweight after shearing.

In Sorbie (1991), the research of Gramain and Myard (1981)was discussed. They proposed a schematic diagram of the ads-orbed layer of HPAM molecules for different molecular weightspecies. Larger molecular weight species show a thicker adsorbedlayer and tend to displace previously adsorbed lower molecularweight fragments. Specifically, for the larger molecular weightpolymer, even though the shear induced by flow of dyed waterinitially increased, the polymer chain remains attached as opposedto the fact that shorter chains are detached from the surfaces.This occurs because the more entangled longer chains do not havesufficient mobility to escape the gap, as did the shorter chains.Hence, lower molecular weight fragments produced by the mole-cular chain scission during the filtration process, are more likely tohave a thinner adsorbed layer in comparison to unfiltered largermolecular weight polymer.

Fig. 20. RGB parameter setting for the base case (top left)/sensitivity test for binary image transformation changing green (top right), red (bottom left), and blue (bottomright) values in ImageJ thresholding tool/zero percent deviation indicates the base case. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127 125

The adsorption of 2 mm-filtered polymer solution is larger thanthe adsorption of 7 mm-filtered polymer and comparable to theresult of unfiltered polymer. This indicates that the reduction inthe size of molecular chain is not the only factor determining thedegree of polymer retention on the surface. The relationshipbetween the size of polymer molecular chain and the accessibilityof polymer molecules to the grain walls has to be considered. Thesize of polymer segments decreased by the molecular chainscission during the 2 mm-filtration and reached the optimum sizethat provided better accessibility to the grain surface than theaccessibility of 7 mm-filtered polymer chains. Moreover, due to thesmaller size of 2 mm-filtered polymer molecules, a better packingand sorting of polymer molecules on the surface was achievedthan that of 7 mm-filtered polymer. Moreover, molecular chainscission due to the filtration became a reason for the increase inthe number of molecular chains in solvent. The increase inaccessibility and the number of molecular chains in solvent hasbeen combined and result in the larger polymer retention for the2 mm-filtered polymer solution.

6. Summary and conclusion

A visual and quantitative examination of polymer retention wasconducted in two-dimensional micromodels with uniform and well-characterized pore networks. Through single-phase flow experimentsand image analysis, visualization of the retention of HPAM with aconcentration of 0.2 wt% was achieved. To our knowledge, this is thefirst study that directly visualized polymer retention in a porousmedium under flow conditions. Although the pore network used wasuniform, extension to pore networks of greater variability is anext step.

Specifically, three mechanisms, hydrodynamic retention, formationof immobile microgel, and mechanical entrapment in pore matrices,were investigated. In the high permeability, homogeneous pore net-work of micromodels, polymer retention/adsorption was by hydro-dynamic retention and the formation of immobile microgel. Microgel,the inhomogeneous portion of polymer, was found trapped in thepore network of the micromodels. It was shown that microgels have asize and structural flexibility leading to their mobile or immobile flowcharacteristics in porous networks as a function of pressure drop.Moreover, the salinity of displacing water, wettability of the micro-model surface, and mechanical degradation of polymer were studiedbecause these factors correlate with hydrodynamic retention. Table 2summarizes the percent polymer retention from all experiments.

Scenario 2, high salinity (5 wt% NaCl) displacing water, resulted inlower polymer retention (6.370.3%) than the polymer retention(7.570.3%) of scenario 1 without salinity and filtration. In addition,wettability of micromodel surfaces was changed by the deposition ofcrude oil (scenario 4) and CTAB (scenario 5). Scenarios 4 and 5 showedpolymer retention of 15.070.3% and 5.070.3%, respectively. Lastly,scenario 3, the effect of filtration of polymer solution through 7 mmfilter, was tested and the filtered polymer has an average polymerretention of 4.370.3%, which is 3.2% point lower than scenario 1. Theadsorption of 2 mm-filtered polymer solution, however, is larger thanthe adsorption of 7 mm-filtered polymer and comparable to the resultof scenario 1. The accessibility of polymer molecules to the pore wallsdepends, evidently, on the size of the polymer molecular chain.Therefore, a 2 mm-filtration reduced the size of the polymer chainand increased the accessibility of polymer molecules to the surfaceand the number of molecular chains in solvent. These changesresulted in the greater polymer retention for the 2 mm-filteredpolymer solution.

Based on the summary of results, it is concluded that themechanical degradation by 7-mm filtration had the most significanteffect on reducing polymer retention. The crude-oil deposited

micromodels exhibited mixed wettability and showed the greatestpercent polymer retention, which is twice as large as the base-caseretention.

Acknowledgments

We thank the SUPRI-A Industrial Affiliates for supportingthis work.

Appendix

Representative elementary volume

During the study of flow of fluid in channels and pore space, acontinuum assumption is required. Thus, the property includingporosity is described in terms of an equivalent continuum. Performinga porosity calculation averaged over a “representative elementalvolume” (REV) is a practical approach to measure the representativecharacteristics of the entire domain of porous structure. Fig. 19 showsthe calculation of porosity for different sizes of the local domain in themicromodel. The REV of the micromodel used was found to be morethan the area of 4�106 pixels, which is large in comparison with apore in micromodel, yet small in comparison with regional variationsin properties of the medium. The porosity of the REV is about 46%,which is greater than the design porosity of 42%. Even though weobtained a larger value of porosity from the image analysis, we usedthe same setting of ImageJ tool for the base case of the sensitivityanalysis.

Sensitivity analysis

Transformation of a color image into binary image is an importantprocess during image analysis. During image analysis, a threshold toolin ImageJ utilized an RGB-based global thresholding technique (Otsu,1979). As shown in Fig. 4, the selection of different regions within theRGB scale on the histogram (which represents the distribution of pixelintensities in the image in RGB mode) needs to be evaluated for anyimpact on the size of area of black (grain) and white (pore).2500�1875 pixels image was used as an REV. By changing separatelythe three parameters, red, green, and blue, on the histogram, the mosteffective parameters to the porosity calculation were determined.Fig. 20 shows the RGB settings for the base case in the ImageJthresholding tool, which has red (65), green (51), and blue (50). Basedin Fig. 20, the area calculation in the image process, the porositycalculation is most sensitive to the change of the green parameter.Both increasing and decreasing the value of green are highly affectiveto the calculation of porosity. For instance, 79.8% deviation of greenfrom the base case gave rise to the 4.5% and 3.5% change in porosity,respectively. On the other hand, in cases of changing red and blue,more than a �30% deviation is not enough to increase the porosity byþ2%. In conclusion, the sensitivity test informed that green is asignificant parameter determining the measurement of porosity,hence the selection of consistent green parameter was required foraccurate image processing.

References

Bi, Z., Liao, W., Qi, L., 2003. Wettability Alteration by CTAB Adsorption at Surfaces ofSiO2 Film or Silica Gel Powder and Mimic Oil Recovery. Key Laboratory ofColloid and Interface Science, Institute of Chemistry, Chinese Academy ofSciences, Beijing 100080, China.

Brookfield DV-IIþ Pro Programmable Viscometer Operating Instructions Manualno. M/03-165, Brookfield Engineering Laboratories, Inc., pp. 58–60.

Buchgraber, M., Al-Dossary, M., Ross, C.M., Kovscek, A.R., 2012. Creation of a dual-porosity micromodel for pore-level visualization of multiphase flow. J. Pet. Sci.Eng. 86–87, 27–38. http://dx.doi.org/10.1016/j.petrol.2012.03.012.

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127126

Chang, H,L., Guo, Z.D., et al., 2006. Advances in polymer flooding and alkaline/surfactant/polymer processes as developed and applied in the People's Repub-lic of China. J. Pet. Technol. 58 (2), 84–89. http://dx.doi.org/10.2118/89175-JPTSPE-89175-JPT.

Cohen, Y., Christ, F.R., 1986. Polymer retention and adsorption in the flow ofpolymer soltuions through porous media. SPE Res. Eng. 1 (2), 113–118. http://dx.doi.org/10.2118/12942-PA SPE-12942-PA.

Dong, Z., Li, Y., Lin, M., Li, M., 2009. Rheological properties of polymer micro-geldispersions. Pet. Sci. 6, 294–298. http://dx.doi.org/10.1007/s12182-009–0047-3.

Du, Y., Guan, L., 2004. Field-scale polymer flooding: lessons learned and experi-ences gained during past 40 years. Paper SPE 91787 Presented at the SPEInternational Petroleum Conference, Puebla Pue., Mexico, 7–9 November,http://dx.doi.org/10.2118/91787-MS.

Fan, T., Buckley, J. S., 2006. Acid number measurements revisited. SPE Paper 99884Presented at SPE/DOE Symposium on Improved Oil Recovery, Tulsa, OK, April22–26.

DeGennes, Pierre-Gilles, 1979. Scaling Concepts in Polymer Physics. 1. CornellUniversity Press, Ithaca.

Gramain, P., Myard, P., 1981. J. Coll. lnt. Sci. 84, 114.Gibas, Iwona, Janik, Helena, 2010. Synthetic polymer hydrogel for biomedical

application. Chem. Chem. Technol. 4 (4), 297–298.Huh, C., Lange, E.A., Cannella, W.J., 1990. Polymer retention in porous media. Paper

SPE 20235 Presented at the SPE/DOE Seventh Symposium on Enhanced OilRecovery, Tulsa, Oklahoma, 22–25 April, http://dx.doi.org/10.2118/20235-MS.

Kim, D.H., Lee, S.J., Huh, C., Pope, G. A., 2010. Development of a viscoelastic propertydatabase for EOR polymers. Paper SPE 129971 Presented at the SPE AnnualMeeting, Tulsa, Oklahoma, 24–28 April, http://dx.doi.org/10.2118/129971-MS.

Kovscek, A.R., Wong, H., Radke, C.J., 1993. A pore-level scenario for the develop-ment of mixed wettability in oil reservoirs. Am. Inst. Chem. Eng. J. 39 (6),1072–1085.

Kolodziej, E.J. 1988. Transport mechanisms of xanthan biopolymer solutions inporous media. Presented at the SPE Annual Technical Conference and Exhibi-tion, Houston, Texas, 2–5 October. SPE-18090-MS, http://dx.doi.org/10.2118/18090-MS.

Lakatos, I., Lakatos-Szabo, J. and Toth, J., 1979 Factors influencing polyacrylamideadsorption in porous media and their effect on flow behaviour. Presented at theSymposium on Surface Phenomena in EOR, Stockholm, Sweden, August 1979.Plenum Publishing Corp., New York.

Levitt, D.B., 2009. The Optimal Use of Enhanced Oil Recovery Polymers Under HostileConditions (Ph.D. dissertation). University of Texas at Austin, Austin, Texas ⟨http://www.pge.utexas.edu/images/pdfs/theses09/levitt.pdf⟩ (May 2009), (downloadedMay 2012).

Maerker, J.M., 1973. Dependence of polymer retention on flow rate. J. Pet. Technol.25 (11), 1307–1308. http://dx.doi.org/10.2118/4423-PA SPE-4423-PA.

Otsu, N., 1979. A threshold selection method from gray-level histogram. IEEE Trans.Syst. Man Cybern. 9, 62–66.

Pradas, M., Gomez Ribelles, J., Serrano, Aroca, A., et al., 2001. Colloid Polym. Sci. 279,323.

Sandvik, E.I., Maerker, 1977. Application of xanthan gum for enhanced oil recoveryin extracellular microbial polysaccharides, In: Sandford, P. et al (eds.) ACSSymposium Series. vol. 45, Ch. 19, pp. 242-264.

Seright, R. 1983. The Effects of Mechanical Degradation and Viscoelastic Behavioron Injectivity of Polyacrylamide Solutions, SPEJ 23(3): 475–85. SPE 9297. http://dx.doi.org/10.2118/9297-PA.

Sorbie, K.S., 1991. Polymer-improved Oil Recovery. Blackie, Glasgow and London.Sorbie, K.S., Parker, A., Clifford, P.J., 1987. Experimental and theoretical study of

polymer flow in porous media. SPE Res Eng. 2 (3), 281–304. http://dx.doi.org/10.2118/14231-PA SPE-14231-PA.

Willhite, G.P., Dominguez, J.G., 1977. Mechanisms of polymer retention in porousmedia. In: Shah, D.O., Schechter, R.S. (Eds.), Improved Oil Recovery bySurfactant and Polymer Flooding. Academic Press, New York, pp. 511–554.

Treiber, L.E., Yang, S.H., 1986. The nature of polymer plugging and a wellboretreatment to minimize it. Paper SPE-14948 Presented in the SPE Enhanced OilRecovery Symposium, 20–23 April, Tulsa, Oklahoma, http://dx.doi.org/10.2118/14948-MS.

Yun, W.J., 2014. Microvisual Investigation of Polymer Retention on HomogeneousPore Network in a Micromodel (Master of Science thesis). Energy ResourcesEngineering Department, Stanford University, USA ⟨https://pangea.stanford.edu/departments/ere/faculty-research/ms-and-phd-theses-search⟩.

Zechner, M., Buchgraber, M., Clemens, T., Gumpenberger, T., Castanier, L. M.,Kovscek, A. R., 2013. Flow of polyacrylamide polymers in the near-Wellbore-region, rheological behavior within induced fractures and near-Wellbore-area.Paper SPE-166085 Presented in the SPE Annual Technical Conference andExhibition of, 30 September–2 October, New Orleans, Louisiana, USA, http://dx.doi.org/10.2118/166085-MS.

W. Yun, A.R. Kovscek / Journal of Petroleum Science and Engineering 128 (2015) 115–127 127