An Investigation of Sand Reyention Test

SPE 151768

An Investigation of Sand Retention Testing With a View To Developing Better Guidelines for Screen Selection Tracey Ballard and Steve Beare, Weatherford

Copyright 2012, Society of Petroleum Engineers This paper was prepared for presentation at the SPE International Symposium and Exhibition on Formation Damage Control held in Lafayette, Louisiana, USA, 15–17 February 2012. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract Sand screens for specific applications are often selected by reference to the results obtained from laboratory sand retention testing. Some recent publications have highlighted the problems of running some types of sand retention tests (slurry tests) at high flow rates, such that the differences between wire wraps screens and metal mesh screens may be exaggerated. With these in mind and also to address some general concerns of the authors ways to reduce flow rates in laboratory slurry tests to more realistic levels have been investigated. This has created some unforeseen effects which are discussed; video has proved invaluable in understanding these unforeseen effects. In addition, an attempt has been made to better define plugging within sand retention tests by relating the pressure build-up gradient from slurry tests to characteristics of the sand itself. Although the pressure gradient generally correlates to certain sand size and sorting parameters the spread in data suggests another factor is important. The purpose of this work is to try and better define the differences in performance between different screen designs, primarily wire wrap and metal mesh screens, in order to better define their application envelopes in terms of sand quality and hence develop more definitive guidelines for screen selection.

Introduction The paper describes an attempt to understand the pressure data for slurry-type sand retention tests with a view to either extending the application envelope for wire wrap screens or justifying our current guidelines. In addition, the relatively recent development of ICDs and the subsequent reduction in annular flow has given more focus on the possibility of better defining screen application envelopes based on laboratory testing. All lab methods utilize a single sand in each test; no attempt is made (or could be meaningfully made) to represent mixing of the sands present along a production zone as could occur in a completion with an open annulus. Slurry tests therefore may be said to represent the compartmentalised inflow obtained with ICDs. During slurry tests the pressure is logged and these pressure data are often used to justify screen selection. However, as has been pointed out by Chanpura et al1, the pressure is a feature of the flow regime in the test and not representative of downhole conditions. Furthermore, it can be unfavourable to wire wrap screens because of their limited open area. Up until now the authors have avoided reading too much into pressure data and concentrated on the sand retention aspect of tests. However, questions about plugging persist and those not familiar with testing have the habit of over-interpreting the pressure data. With this in mind and possessing a large database of retention test results from the last 8 years it was time to try to find out how often plugging really occurs. This work has taken the form of the following:

• The database of test results has been examined to find out if there is a correlation between the pressure gradient and sand size distribution characteristics.

• The test method has been changed (reduced flow rate) to find out if the trend in results is the same as in our standard test

• Tests using a dense brine (caesium formate) have been performed to find out if changing the suspending medium yields similar results.

2 SPE 151768

Slurry test database analysis The authors have built up an extensive database of slurry test results from several years of sand screen development and selection. All the data have been obtained using a single test method with reservoir sands on both wire wrap (WWS) and metal mesh screens (MMS); in all, well over 1000 tests have been performed. Details of the test method are outlined on page 6 (the standard test). In slurry tests the pressure rise usually becomes steeper as the sand becomes finer and more poorly sorted, and is often steeper on WWS than MMS; but steep pressure rises do not necessarily indicate plugging. A steep pressure rise can be due solely to the permeability of the sand itself, which is influenced by its size and sorting. One approach to try to define plugging is to see if gradients can be predicted from a property of the sand distribution, if so then the gradients are purely a function of the sand itself and do not indicate plugging. The permeability of a sand is related to both the size and the sorting of the grains. The uniformity coefficient (Uc = D40/D90) gives an indication of the sorting of a sand but no information on the size of the sand and is therefore not a good predictor of gradient. Figure 1 shows the poor correlation between the pressure gradient observed and the sand Uc. It should be noted that these tests were performed for specific applications and each sand was not necessarily tested on both WWS and MMS, This explains why there are less WWS results as the fines content and uniformity coefficient increase.

0

200

400

600

800

1000

0 5 10 15 20 25 30 35 40 45 50

gradient (psi/g)

uniformity coefficient (D40/D90)

metal mesh

wire wrap

Figure 1. Plot of Pressure gradient vs Uc

Better correlations were obtained where Uc was ratioed to the grainsize (for example, D50/Uc as suggested by Constien and Skidmore(2) for predicting sand production), but these ratios were found to offer no advantage over a simple correlation with the fines content (defined as the <45 micron fraction) for predicting gradient. The fines content gives some indication of size and sorting and also has the advantage of always being measureable by both sieve and laser diffraction, and is easy to see from the particle size distribution. Figures 2 and 3 show the correlation of pressure gradient to fines content on wire wrap and metal mesh media respectively. For both screen types there is a definite trend upwards as the fines content increases. However, the spread in data also increases with fines content, especially for metal mesh screens. This suggests that fines content is not a perfect predictor of gradient and other factors are involved. Furthermore, at fines contents over 40% the correlation worsens; one possible reason for this is that at very high fines content the sorting of the sand improves. Where a sand has 50% fines the sand will necessarily be fine overall and therefore the Uc will decrease as the fines content increases beyond a certain point.

0

200

400

600

800

1000

0 10 20 30 40 50 60

0

200

400

600

800

1000

0 10 20 30 40 50 6

gradient (p

si/g)

%<45um LPSA0

metal mesh

wire wrap

gradient (p

si/g)

%<45um LPSA

Figure 2. Pressure gradient correlation to fines content for WWS Figure 3. Pressure gradient correlation to fines content for MMS

SPE 151768 3

Pressure gradients and screen selection When Figures 2 and 3 are superimposed to show both screen types in the same plot (Figure 4) it can be seen that at low fines content there is considerable overlap with the metal mesh and wire wrap screens where the gradients are very similar and low. Up to around 10% fines (measured by LPSA) there is very little difference in general between the wire wrap and the metal mesh screen. Figure 5 shows the same data but this time plotted against the fines fraction from sieve analysis, here up to 5% fines there is no appreciable difference between the screen types. As the fines content increases the spread in the data increases and there is more separation between the gradients from wire wrap tests compared to metal mesh. These results support the current Weatherford criteria for selection between WWS and MMS based on sand fines content. However, in a lot of cases even where the fines content is low (<5% by sieve analysis) there can be a relatively large difference between wire wrap and metal mesh gradients even though both gradients are low. This is often stated to be due to the lower flow area and different flow regime of WWS compared to MMS. Furthermore, from analysis of the database results it also appears that the gradients on WWS are more sensitive to aperture size than those on MMS. Some simple calculations have been performed using parameters specific to the standard test to investigate these effects.

0

200

400

600

800

1000

0 10 20 30 40 50 60 7

gradient (p

si/g)

%<45um LPSA0

metal mesh

wire wrap

Figure 4 Gradient correlation to LPSA fines content.

0

200

400

600

800

1000

0 5 10 15 20 25 30 35 40

gradient (p

si/g)

%<45um sieve

metal meshwire wrap

Figure 5 Gradient correlation to sieve analysis fines content.

Test fluid flow rate effects As shown above, wire wrap screens tend to produce higher pressure drops than metal mesh screens under the same test conditions, and this may bias data interpretation against wire wrap screens. It has been recognized(1) that the pressure drop through a screen with a sand layer present can contain a contribution due to non-Darcy flow (non-linear flow), and the effect may be larger in laboratory tests than in the field due to the high flow rates used creating an increased contribution from turbulent flow. Furthermore, the contribution from non-linear flow will be higher in WWS than MMS because the reduced flow area gives a much higher velocity through the apertures for a given volumetric flow rate. The magnitude of the effect may

4 SPE 151768

potentially be estimated using the Forchheimer equation, but in order to perform the calculations it is necessary to know the sand permeability and porosity, which are difficult to quantify accurately in slurry retention tests. It is also important to note that this process could not be used to determine likely field pressure drops or even predict if screen plugging is occurring in laboratory tests since there too many unknown parameters, but it indicates that laboratory test data interpretation needs to be done with awareness of the mechanisms involved. Since it is difficult to apply a turbulent flow correction with accuracy, a simpler approach was taken in which the measured pressure gradients were normalized for the screen open area to adjust for the higher aperture velocity in WWS. The results are shown in Figure 6, from which it can be seen that although the WWS values may trend slightly higher they are largely brought into a similar range to the MMS.

0

100

200

300

400

500

0 10 20 30 40 50 60

grad

ient from

 slurry test (psi/g)

<45 micron fraction

metal mesh

wire wrap

Figure 6. Data from Figure 4 adjusted to take into account the reduced open flow area of wire wrap screen

Wire wrap screen aperture selection The gradient correlations in Figure 4 refer to tests in which the sand was adequately retained; gradients, particularly on wire wrap screens, can be significantly lower if the sand is not properly retained. Previous work by the authors on Dutch twill weaves2 indicated that sizing the weave aperture on the D10 of the sand gave good retention, and this criteria has subsequently been found to also apply to other weave designs. However, analysis of the WWS test results suggested that sizing on the sand D10 for wire wraps does not necessarily result in good retention. Figures 7 to 10 show the trends in the amount of sand produced in retention tests on a 200 micron wire wrap screen, and similar results were obtained with other aperture WWS. From these plots it would appear that to be confident of good retention (equivalent to that given by a metal mesh) sizing the aperture in the sand D30 to D50 range would be necessary.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 100 200 300 400 500 600 700 800

sand

 produ

ced (g)

d10 of sand

200um wws screen aperture

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 100 200 300 400 500 600 700 800

sand

 produ

ced (g)

d20 of sand

200um wws screen aperture

Figure 7. Amount of sand produced vs D10 for 200μm WWS Figure 8. Amount of sand produced vs D20 for 200μm WWS.

SPE 151768 5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 100 200 300 400 500 600 700 800

sand

 produ

ced (g)

d30 of sand

200um wws screen aperture

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 100 200 300 400 500 600 700 800

sand

 produ

ced (g)

d40 of sand

200um wws screen aperture

Figure 9. Amount of sand produced vs D30 for 200μm WWS Figure 10. Amount of sand produced vs D40 for 200μm WWS.

Summary of database analysis The examination of the database of slurry test results indicates that Weatherford’s criteria for utilizing WWS in sands up to 5% fines measured by sieve analysis or 10% fines measured by LPSA appear valid, and also that the uniformity coefficient is not necessarily a good guide to screen application envelope. In addition, although MMS apertures can be reliably selected using the sand D10, for WWS more reliable retention is obtained if the aperture is selected on the sand D30.

However, it is apparent that artefacts due to the flow regime in the tests may influence the conclusions drawn from slurry retention test results and bias selection towards MMS if the data interpretation is not done carefully. It has also been suggested that it would be preferable for retention tests to better simulate the field situation. Therefore a laboratory study was performed to establish whether benefit would be gained from performing tests at lower flow rates to remove/reduce the flow regime effects. As part of the study the effect of using a denser, more viscous carrier fluid was also briefly examined. In addition, tests were performed to determine whether sand injection rate influenced retention.

Investigation into sand retention slurry test method Conclusions reached from examination of the test database are only valid if the tests give meaningful results and any problems with the tests are recognised and taken into account when interpreting the data. Recently there has been some criticism of retention testing because it does not represent downhole conditions; in particular the flow rates are orders of magnitude higher. Previous research(4) has shown the importance of using reservoir sands rather than attempting to simulate the particle size distribution using outcrop sands, but when using reservoir sands a test method is required that uses as little sand as possible and for practical reasons measurement of the amount of sand passing through the screen should be quick and easy. The authors are aware that there are drawbacks to retention testing and were interested to find out if the screen selection would change if the flow rate is reduced. For example would the gradient on a wire wrap be closer to a mesh screen at a lower flow rate. There were constraints on testing in that it is desirable to use a transparent test cell, which meant that the pressure limitations were still low since the same test cell was used in the modified tests, but the results were interesting and give some insight into the mechanisms involved in retention tests.

Test Procedures Table 1 gives a comparison of flow rates used in this research and some typical field values. A brief description of each method is given below.

Table 1 Summary of test conditions and velocities

Parameter Std test Reduced flow rate

Reduced flow rate and sand rate Dense brine field

Inflow area (cm2) 5.06 5.06 5.06 5.06 Depends on well parameters

Flow rate (ml/min) 500 100 100 50 104 - 106 Sand concentration

in xanthan (g/l) 100 200 200 n/a n/a

Sand rate (g/min) 0.5 0.5 0.1 0.25 unknown Sand concentration

in test cell (g/l) 0.01 0.05 0.01 5 unknown

Velocity (cm/s) 1.7 0.33 0.33 0.16 0.01 to 0.1

6 SPE 151768

The standard polymer test. The standard test used by Weatherford from which the database of test results is derived is described by Ballard and Beare3. It comprises suspending the sand in a concentrated xanthan solution, (the yield point of the xanthan is sufficient to suspend the sand without the need for agitation), and feeding the slurry into a water stream which disperses the sand, dilutes the polymer to negligible amounts, and carries the sand to the screen. Sand passing through the screen is collected, dried, and weighed, and the pressure increase as the sand builds up on the screen is logged. The standard test uses a water flow rate of 500ml/min which gives a high velocity and was chosen to give good dispersion of the sand and xanthan polymer. Figure 12 illustrates the test set-up.

Reduced flow rate polymer test. Reducing the water flow rate in our current test set-up runs the risk of poor sand dispersion. In order to overcome this problem the following steps were taken; • An inline static mixer was introduced just after the point where the sand suspension and water flow converge to aid the

water mixing with the polymer slurry • The sand concentration in the slurry was doubled so that less polymer per unit mass of sand was introduced to the cell • The actual polymer concentration was reduced. These changes allowed the flow rate to be reduced from 500ml/min to 100ml/min.

Dense brine tests: In addition to reducing the flow rate of our current method testing was also conducted using a dense brine, caesium formate, as a suspension agent. Although caesium formate has a high density (around 2200 kg/m3), this is not sufficient to suspend the sand without the use of agitation (i.e. a stirrer). In these tests (Figure 11) the sand was placed into the caesium formate reservoir directly above the test cell and the contents of this reservoir were flowed into the test cell and through the screen. This means unlike the previous tests there is no dilution of the sand slurry, and the fluid flowing in the test cell is caesium formate which has a viscosity of around 4cP. The sand passing through the screen was collected, washed, dried, and weighed, and the pressure was logged throughout the test. Details of the test conditions are summarised in Table 1.

screensample

Pressuretransducer

pump

Sand slurry

stirrer

water

Peristaltic pump

screensample

Pressure transducer

Sand slurry

Syringe pump

Figure 11. Set-up for dense brine tests Figure 12. Set-up for the polymer tests

The use of a transparent test cell allowed video recordings to be made of the tests, which were invaluable in understanding the results. There were some obvious differences between the standard high rate test and those performed at lower flow rates, these are described below.

Observations from Tests Details of the particle size of the sands used in this study are given in Table 2. Plots of the full particle size distributions are given in Appendix 1.

SPE 151768 7

Table 2 Details of sand tested. D values are in microns, Sc =D10/D95 sand D5 D10 D40 D50 D90 D95 % <45um Uc Sc comment B-48 445 360 180 142 15 7 23 12 51 Used in commissioning tests

K 526 464 217 188 40 10 11 12 39 Similar to a N Sea sand

P-SB 549 463 235 197 33 12 13 12 44

SK 370 318 200 168 10 4.4 25 15 88 Potentially plugging sand

YE 380 317 188 150 9 3.7 26 22 99

Particle dropout at low flow rates The initial commissioning tests for the reduced flow rate method and the dense brine method used a reservoir sand (B-48) with a fines content of around 20%. It was noticed that at 100ml/min that the larger sand grains fell faster than the finer particles, such that large particles reach the screen first followed by a cloud of finer grains. In terms of time in these tests there is a 20 second delay between the large particles and the cloud of fines reaching the screen. To try to overcome this problem the dilution water in the polymer tests was changed to a denser brine, in the hope that the density would reduce the drop-out of the larger particles. However, the higher density only exacerbated the problem by holding up the cloud of fines resulting in a longer delay between coarse particles and the cloud of fines reaching the screen.

The same phenomenon happened in the caesium formate tests performed at 50mls/min. The flow was less turbulent but as the sand approached the screen the coarser particles separated out from the main sand body and reached the screen 50 seconds before the main cloud of sand. Figure 13 shows the effect in a series of photos from a polymer test, and Figure 14 shows a caesium formate test. The reduced turbulence in the caesium formate test is also apparent in these Figures since the sand ‘front’ is much sharper. The arrival of coarse particles to the screen before finer material was not noticeable in the standard test.

5 sec 30 sec 45 sec

Large grainsdropping out

‘cloud’ of fines

Figure 13. Dropout of larger grains in a polymer test Figure 14. Dropout of larger grains in a formate test at 100mls/min at 50mls/min

Patterns of sand deposition Changes to the test method also affected the pattern of sand build-up on WWS. These effects are not so apparent on metal meshes, presumably because the open area is both higher and uniformly distributed. These effects only became apparent by studying the video recordings of the sand build-up on the screens.

At high flow rates (or where the carrying power of the fluid is sufficient) the sand is entrained in the flow and initially only deposited in the slots between the wires. The sand builds up in the slots and directly over the slots forming ridges of sand. Even where particles initially land on the wires they are swept sideways into the nearest slot. The sand ridges grow and expand eventually covering the wires. This pattern of deposition is evident in the 500ml/min polymer tests and the caesium formate tests, and is shown in the left-hand photo series in Figure 15.

The opposite happens where the flow rate is low in a polymer test. The sand falls onto the screen randomly, and where the particles land on the wires they stay in place. In this instance the sand builds up preferentially on the wires rather than in the slots. As the sand builds up ridges form on the wires resulting in the reverse pattern of deposition compared to the entrained pattern described above. This pattern of deposition was observed in the 100ml/min polymer/water tests, and is shown in the

8 SPE 151768

right-hand photo series in Figure 15. Each pair of photographs is taken from video recordings of the tests and represents the same amount of sand reaching the screen

0.1g sand 0.1g sand

0.22g sand 0.22g sand

0.5g sand 0.5g sand

0.71g sand 0.71g sand

Figure 15. Preferential sand deposition in slots (left) and on wires (right)

Figure 15. Preferential sand deposition in slots (left) and on wires (right)

Comparison of results from 500ml/min and 100ml/min polymer tests and dense brine tests. Tests were performed with reservoir sands with distributions typical of North Sea sands. Particle size details of the sands are given in Table 2 and Appendix 1. A pair of sands with very similar particle size distributions (sand K and sand P-SB) was tested on wire wrap and metal mesh screens with appropriate apertures. A summary of the results is given in Table 3 below.

Table 3. Comparison of results from different test methods

sand screen

std test 500ml/min 0.5g/min 100ml/min 0.5g/min 100ml/min 0.1g/min Dense Brine

(Cs Formate) sand

produced gradient sand produced gradient sand

produced gradient sand produced gradient

K 300μm wws 0.2105 107 0.4842 31.2 Not tested Not tested 0.1528 31 250μm wws 0.1374 134 0.1255 27 0.5696 21 0.0562 30 230μm mms 0.1115 34 0.1096 21 0.1506 30 0.0635 3.4

P-SB 300μm wws 0.3451 48 0.5332 7.2 0.379 n/a 0.1332 8.3 250μm wws 0.1123 80 0.1766 10.6 0.3872 11 0.0867 13.7 230μm mms 0.1058 11.5 0.0843 10.5 0.1474 24 0.0514 4.5

SPE 151768 9

The following observations can be made from these test results.

1. Where the sand is on the borderline for retention reducing the flow rate increases the amount of sand passing through the screen (300μm WWS tests). This is most likely because at the lower flow rate used the sand falls randomly over the area of the screen and more can fall through the slots before enough larger particles have reached the slots and bridged off.

2. The 500ml/min polymer test and the dense brine test give similar relative results for the WWS relative to the MMS for both gradient and sand produced. This is probably because both tests have a similar pattern of sand deposition – the sand is carried with the flow into the slots in the wire wraps.

3. When the flow rate is reduced to 100ml/min in the polymer tests the gradients on the WWS are reduced and become equivalent to, or even less than, those on the MMS. The gradient is measured in terms of psi per gram of sand reaching the screen, and when the flow rate is 100 ml/min much of the sand is resting on the wires of the wire wrap and not contributing to the pressure rise. This means more sand can reach the screen for each psi rise in pressure. In some respects this removes the difference in open area because the sand is equally dispersed over the whole area of the screen not just in the slots.

4. The gradients and sand produced through MMS are much less affected by the change in flow rates in the polymer tests. This is probably because the apertures through the screen are equally distributed throughout the area of the screen.

5. Reducing the sand concentration at the lower flow rates increases the amount of sand produced before bridging occurs. Figure 16 below shows a photomicrograph of the underside of a filtercake built up on a wire wrap screen with the P-SB sand. The slot pattern of the screen is evident from the lines of larger grains without any fine material. The part of the filtercake that was resting on the screen wires is composed of finer grains. This photograph illustrates how the large grains are essential for the bridging process to start on a wire wrap screen, and furthermore the grains need to be too large to fit through the slot. The dominant mechanism is size exclusion rather than bridges formed by a number of smaller particles.

Large grains wedgedinto a slot

Figure 16. A photomicrograph of the underside of the sand filtercake on a wire wrap screen.

Other sands were also tested, and gave similar results. None of these sands gave an indication of plugging the screens tested, and so to establish whether plugging is influenced by test conditions two further sands were selected for testing (SK and YE). One of the sands had shown plugging tendencies in previous standard tests, and the second sand was selected to have a similar particle size distribution (Table 2 and Appendix 1).

Identification of plugging in slurry tests As previously stated a high gradient in a slurry test need not necessarily indicate plugging, the gradient could be a consequence of the nature of the sand itself rather than the sand plugging the screen. In an attempt to differentiate between steep gradients due to sand permeability and actual plugging, at the end of the slurry tests the sand injection is stopped and the water flow is

10 SPE 151768

continued but is stopped and restarted several times in order to find out if the pressure dissipates (indicating unimpeded flow) or if the screen is genuinely plugged. Figures 17 shows a pressure profile from this part of the test where the pressure falls to zero when flow is stopped suggested that the screen is not impaired. The pressure during flow also usually decreases initially as some additional sand is washed through the screen. Figure 18 shows a profile where the pressure is slow to fall and does not return to zero, suggesting that there is some plugging of the screen. In the process of reviewing all the data from the database there were only a few instances where the pressure during this part of the test did not quickly dissipate. It could therefore be inferred that for a given screen type in most cases the observed pressure gradients result from variations in sand characteristics rather than plugging.

0

10

20

30

40

50

60

0 10 20 30 40 50 60 70 80 90 100

pressure (psi)

time (sec)

0

10

20

30

40

50

60

0 10 20 30 40 50 60 70 80 90 100pressure (psi)

times (seconds) Figure 17. Stop/start pressure profile - no plugging Figure 18. Stop/start pressure profile - with plugging

The effect of test method on observed plugging Tests were performed on a sand which had shown plugging tendencies in the standard tests (sand SK) and compared with the results from a sand with a similar distribution (sand YE). Table 4 below summarises the results. Both sands showed tendencies to plug, though the SK sand seems more prone to plugging. It should be noted that the screen apertures tested are comparatively small compared to the size of these sands; for the SK sands these small screens were tested in routine tests because this sand represented the largest sand in the well and a fine screen was required to retain the smaller sands.

From these data it appears that plugging can occur on both wire wraps and mesh independent of the test conditions. It is possible that particle shape influences the pressure gradient and plugging behavior observed, although the limited image analysis performed on the sands to date has shown no obvious difference between them. This aspect is being further investigated.

Table 3. Comparison of results from different test method. Tests displaying signs of plugging shown in red

sand screen std test 500ml/min 100ml/min 0.5g/min 100ml/min 0.1g/min Cs Formate sand

produced gradient sand produced gradient sand

produced gradient sand produced gradient

SK

125μm wws 0.0298 1419 0.0102 614 0.0242 2476 0.0108 605

200μm wws 0.0520 2598* 0.2566 598 0.3608 1782 0.0547 1176

120μm mms 0.0403 400 0.0199 249 0.0479 311 0.0123 137

YE

125μm wws 0.0116 1240* 0.012 693 .0100 1032 0.0142 1549

200μm wws 0.0561 929 0.0898 518 0.0112 925

120μm mms 0.0581 355 0.0398 227 0.0429 194 *probably plugging but insufficient sand was retained to be certain

Discussion Running tests at realistic reservoir conditions is very appealing but the short study presented here shows how the low velocities required to match field rates result in the large particles dropping out. Increasing the viscosity of the flowing fluid might overcome this to some extent – though it could also exacerbate it as well. The high viscosity required for the lower velocity will almost certainly result in entrainment of the sand into the slots of wire wrap screens giving the same deposition pattern as the high flow rate standard test.

SPE 151768 11

The flow regime in the test cell is noticeable; eddies and currents are very apparent. In a one-off polymer test where the flow rate was reduced to 50ml/min and a viscosified water (4cP) was used the flow was more laminar, but this created edge effects in that the flow at the edge of the test cell was slowed and the sand was deposited on the screen in the form of a cone.

The test results presented here do not look very promising for creating a test to mimic reservoir conditions - whatever they may be? More informed data interpretation may be a better approach, in which case a test which generates reliable, reproducible data would be required. The focus would then be on improving understanding of how laboratory results can best be transferred to the field, and hence improve screen selection criteria. To this end a direct comparison of field performance to laboratory tests would be very beneficial; this would require cooperation between operators and a screen supplier to perform a dedicated study.

There is currently no ‘standard’ retention test, and different laboratories all have their own method. The test data presented here shows that laboratory slurry test data requires informed interpretation (to avoid over-interpretation), especially if results from different test methods are being compared. Changes in flow rate, test cell configuration, and sand concentration can all influence the results obtained, and it is preferable to consider relative results from the same test method. However, the overall trends in results from each method are similar.

Conclusions Based on a study of data obtained over a number of years from the standard test method it appears that the correlation of slurry test pressure gradient to sand fines content (<45 micron fraction) appears to be the best that can be obtained with a simple sand size distribution parameter. Fines content also has the advantage that it is a parameter that can always be measured; for example Uc depends on the sand D90 which is not always known from sieve analysis. In addition, the differences between pressure gradients observed with WWS and MMS can be attributed at least in part to the effect of screen flow area. The study on retention test methods has highlighted that slurry test pressure data needs careful interpretation with an awareness of the mechanisms involved. The use of a transparent test cell coupled with video observations was essential in understanding the phenomena which can occur in the retention tests. Different test methods introduce different influences into the data obtained depending on the flow regime and test cell parameters. These include effects such as particle entrainment and dropout, both of which can influence the pressure profile obtained. Furthermore, changing the sand injection rate can affect the amount of sand produced prior to the screen bridging-off. Because of these processes test method development requires reservoir sands; phenomena such as particle dropout are unlikely to be detected using an outcrop sand. MMS screens are more tolerant to different sands and test parameters than WWS, but if a sand is prone to plugging it will do so irrespective of the test method used. Furthermore, pressure gradient is not necessarily a good guide to whether a screen plugs; in reality plugging is not often observed in laboratory tests.

Laboratory testing for screen selection needs to be relatively straightforward to perform and not time-consuming. Since attempting to simulate the reservoir situation appears problematic, it is preferable to perform a test which gives reproducible, well-understood results and couple it with careful data interpretation. From the large number tests performed Weatherford’s criteria for utilizing WWS in sands up to 5% fines measured by sieve analysis or 10% fines measured by LPSA appears valid, although WWS aperture should be selected on the sand D30 for reliable control. Uniformity coefficient is not a necessarily a good guide to screen application envelope, and current selection criteria based on particle size data are probably conservative. Possibly more of an issue for the application of wire wrap screens is the size variation of the sands in a completed zone. A slot sized for the finest sand may mean that the aperture is small compared to the larger (better producing) sands, which could lead to plugging. In laboratory tests meshes are more tolerant to variations in sand size, and may be preferable in heterogeneous formations. Comparison of screen behaviour in laboratory tests and subsequent field performance would enable a better foundation for development of selection criteria based on sand characteristics; with a couple of notable exceptions5,6 few studies of this nature were found in the literature. Future work It is probable that sand morphology will have an influence on pressure gradients, and may also influence screen plugging behavior. It is intended to perform a study utilising image analysis to examine these effects. However, although a better understanding of the phenomena associated with laboratory tests and screen/sand interactions will help to improve screen selection, relating laboratory results to the field situation is essential to optimize selection criteria. This requires cooperation between operators and a screen supplier to perform a dedicated study. Acknowledgements The authors would like to thank Nicola Spencer for much of the lab work and Weatherford for permission to publish this work.

12 SPE 151768

References

1. Chanpura, R.A., Hodge, R.M., Andrews, J.A., Toffianin, E.P., Moen, T., and Parlar, M. “A Review of Screen Selection for Standalone Applications and a New Methodolgy” 2011. SPE Drill& Compl 26 (1)

2. Constien, V., and Skidmore V. “ Standalone Screen Selection Using Performance Mastercurves” SPE 98363, presented at SPE International Symposium and Exhibition on Formation Damage Control held in Lafayette, L.A., 15–17 February 2006

3. Ballard, T. and Beare, S. “Media Sizing for Premium Sand Screens: Dutch Twill Weaves”, SPE 82244, presented at SPE European Formation Damage Conference, The Hague, The Netherlands, 13-14 May 2003

4. Ballard, T. and Beare, S “Sand retention testing; the more you do the worse it gets” SPE 98308 presented at SPE International Symposium and Exhibition on Formation Damage Control held in Lafayette, L.A., 15–17 February 2006

5. Hodge, R., Burton, R., Constien, V., and Skidmore V. “An Evaluation Method for Screen-Only and Gravel-Pack Completions” SPE 73772 presented at the 1995 SPE international Symposium on formation Damage Control, Lafayette, Louisiana, Feb 20-21.

6. Mathisen, A.M., Aastveit, G.L., and Alteras, E. “Successful Installation of Standalone Screen in More Than 200 Wells: The Importance of Screen Selection Process and Fluid Qualification” SPE 107539 presented at the Eurpean Formation Damage Conference, Scheveningen, The Netherlands, 30 May-1 June, 2007.

SPE 151768 13

Appendix 1. Test sand particle size distributions

Sand used for commissioning tests: B-48

0

10

20

30

40

50

60

70

80

90

100

1101001000

cumulative %

particle size

B‐48 LPSA B‐48 sieve

North Sea type sands: K and P-SB

0

10

20

30

40

50

60

70

80

90

100

1101001000

cumulative %

particle size

P‐SB LPSA P‐SB sieve K LPSA K sieve

Potential plugging sands: SK and YE

0

10

20

30

40

50

60

70

80

90

100

1101001000

cumulative %

particle size

SK LPSA SK sieve YE LPSA YE sieve