Pipelining of coarse coal as a stabilized slurry - Another viewpoint

Page 1

41_+ Tech. Ef Mag Trampertak

Moch 29-70, 1979 baske

PIPELINING OF COARSE COAL AS A

STABILIZED SLURRY-ANOTHER VIEWPOINT

Dr. A.D. Thomas M.D. Research Company Pty Ltd. North Ryde, N.S.W., Australia

1. INTRODUCTION requirements are necessary for this to occur. Firstly some Increasing interest has recently been shown in the particles must be smaller than about 10 microns and possibility of transporting coarse coal as a "stabilized" secondly these particles must be flocculated. Because of slurry. (e.g. Pertuit et al, 1978). Such a slurry would be the first requirement it is essential for static stability that formed by grinding a small proportion of the coal very there be some fines present. The second requirement is fine to form a viscous "vehicle" usually not a practical problem as flocculation tends to be would support the coarse particles enabling transport the rule rather than the exception in industrial processes. under laminar flow. Furthermore such slurries would be Provided the yield stress is sufficiently large to support stable under shutdown conditions, the vehicle being suffiunder static conditions indefinitely. (e.g. Halvorsen, the largest particles such particles will remain suspended ciently "viscous" to prevent the coarse particles settling. 1976). In a previous paper (A.D. Thomas, 1977) the concept which they referred to as "stab-flo." The major Pertuit et al (1978) discussed the advantages of this author derived a simple criterion to determine the yield advantages over conventional fine particle long distance stress required to ensure stability of a certain size particle slurries (e.g. Black Mesa) are the greater product acceptance of coarse coal and lower size reduction and de- Tyu ≥0.092gd (pp-p) watering costs. These advantages are obtained at the expense of higher pressure gradients but in spite of this the where Tyv is the yield stress of the vehicle, d is the maxioverall transport costs look promising compared to conmum particle size, g is the gravitational acceleration, and ventional slurry systems. Pp and p are the particle and vehicle densities respec- In this paper the stabilized slurry concept is investigated tively. Unknown to the author, Russian workers had prefirstly by pipeloop testing, under laminar flow conditions, viously arrived at an equation of identical form with the simulated slurries consisting of coarse particles in highly constant given as between.083 to 0.10 (Traynis, 1977) viscous Newtonian fluids. These results are compared so that equation 1 is now fairly well established. with a theory based on a sliding bed concept. Next, actual stabilized slurries are tested in the same pipeloops and a modification to the above theory is proposed to explain 3. LAMINAR FLOW AND STATIC STABILITY their behaviour. If a slurry is made up totally of particles less than about

10 microns in size, (i.e. all particles affected by flocculation), laminar flow in both vertical and horizontal pipes is

2. STABILITY UNDER STATIC CONDITIONS readily attainable. Such would be the case for the vehicle

portion of a stabilized slurry. However, once coarser par-

For obvious reasons the ability to restart a pipeline full ticles are added laminar flow in a horizontal pipe may not of slurry on shutdown becomes of paramount importance be possible due to settling effects. Obviously in any particonce the pipeline length exceeds a few kilometres. Thus ular situation a certain minimum vehicle consistency is rethe static stability of the proposed stabilized slurries is a quired before laminar flow is possible. Traynis (1977) major advantage. However, it must be realized that exstates that if the slurry is statically stable then laminar flow isting long distance slurries already have this ability to be is possible and this was also inferred by Pertuit et al restarted (Wasp, 1969). In both cases it is achieved in the (1978). However, it has been shown by the author. same way-by having a fine particle flocculated vehicle • (A.D. Thomas, 1978), that in some cases this is not true. sufficiently "viscous" to trap the coarser particles and pre- Further evidence of this is suggested by the fact that most vent them settling preferentially. The physical mechaexisting "long distance" slurries, whilst statically stable. nisms whereby this occurs have received little attention in are known to deposit out under turbulent conditions withthe literature but it is clear from the limited works out laminar flow being possible. Thus although the satisavailable (e.g. Ansley & Smith, 1967; A.D. Thomas, fying of equation 1 should ensure static stability it does 1977; Traynis, 1977) that static stability occurs when the not necessarily follow that laminar flow is possible. It is yield stress of the slurry is sufficient to support the now evident that although both the existing long distance coarsest particles. The yield stress arises from the strength slurries and the proposed stabilized slurries are statically of the floc structure which forms under static conditions stable the latter have the additional property of being able

with flocculated suspensions (D.G. Thomas, 1961). Two to flow under laminar conditions. This will generally re-

196

Page 2

Kelative bed huight; h/p

-4 1.0

0+5

0•4

20 50

- In-situ Cance

Page 3

contact occurring between this sliding layer and the pipe

Relative bed height, h/D wall. At this stage the situation is analogous to the turbulent flow transport of very coarse particles where all of the

particles are being transported as a sliding or saltating bed

0 0-2 0-4 0-6 0•8 1•0 with no turbulent particle support. This turbulent flow 1•0 case has recently been analysed very successfully by

Wilson (1974), and his theory should apply equally as

0•9 well to the laminar flow case.

Wilson's theory is based on a simple force balance between the resisting force acting on the bed due to the

0-8 solid friction and the driving force due to the fluid pres-

sure gradient. Incipient deposition occurs when these two

0•7 forces exactly balance each other. His turbulent flow

theory allows for the effect of the different surface roughness of the top of the sliding bed and the pipe wall. For

0•6 laminar flow, fluid resistance is independent of surface

roughness and for this situation it can be shown that

ф 0•5 Wilson's theory indicates incipient deposition occurs when the slurry pressure gradient equals Ja given by

0•4- Ja u,Copg(S-1) Ø (2)

where u, is the co-efficient of sliding friction between the bed and the pipe wall (Wilson takes as 0.4).

0-3

C, is the maximum packing concentration by volume (= 0.6 for narrow size distributions)

0•2 p is the fluid density

S is relative solids density

0.1 and Ø is a geometrical function which depends on the

height of the sliding bed (i.e. the in-situ concentration of

0 solids in the bed). If C, is taken as 0.6 it can be shown, 0 10 20 30 40 50 60 using Wilson's equations, that Ø is as given by Figure 1

either in terms of the in-situ concentration of sliding solids

In-situ concentration or the ratio of bed height to pipe diameter (h/D).

The most important consequence of equation 2 is that

(volume %) deposition occurs at a constant value of slurry pressure gradient regardless of pipe size. This means that the pres-

sure gradient becomes increasingly uneconomic as the

FIGURE 1 pipe size increases. For example equation 2 indicates

VARIATION OF FUNCTION Ø WITH CONCENTRATION AND h/D. pressure gradients of the order of 1000 Pa m- 1 (0.1 ft

water/ft) for coal at 40% concentration, which although

quire a more viscous vehicle than that required to obtain not unreasonable in small size pipes, is a considerable disstatic stability alone. advantage in larger pipes (c.f. the pressure gradient for

the 450 mm (18 inch) Black Mesa pipeline which, from the information supplied by Love (1969) is around 100

4. LAMINAR FLOW OF NEWTONIAN SLURRIES Pa m- 1 (.01 ft water/ft). 4.1 Physical Analysis of Laminar Slurry Flow with Newtonian Vehicle 4.2 Pipe Length Required for Particles to Settle with Consider a suspension of coarse particles in a viscous Newtonian Vehicle Newtonian fluid. In a vertical pipe homogeneous laminar Consider a pipeloop through which a Newtonian slurry flow would be obtainable. Similarly in a horizontal pipe is being pumped under laminar flow. At the entrance to the same would apply provided the length of the pipe the line, immediately following the pump, the particles was not long enough to allow significant settling. Howcan be assumed homogeneously distributed. As the ever, given sufficient pipe length, it would be expected slurry travels along the pipeline the particles slowly settle that the particles would gravitate to the bottom since, unso that, as a first approximation, after a distance L the like the turbulent flow situation, there are no radial velocparticles have settled a height H given by ity components to support the particles. Once the particles reach the bottom they will be transported as a sliding H = WL • (3) or saltating layer of particles with solid/solid frictional

197

Page 4

where W is the free settling velocity of the particles and V 18.9mm (3/4 inch) and 9.41 mm (3/8 inch) is the mean flow velocity. (Throughout this paper W will be calculated as for spheres). All particles will have fully above loop. They has a 4.3 m (14 ft) long outward leg These two loops were schematically similar to the settled when H = D, i.e. after a pipe length, L,, given by with the middle of the 2.3 m (7.5 ft) long pressure mea-

L, = DV (4) ing section 4 m (13 ft) from the pump. Flowrates were suring section 2.9 m (9.5 ft) from the pump and the view-

measured by diverting the flow and timing a sample.

Of course the actual situation is far more complicated than this due to the influence of the velocity profile, the 4.4.2 Experimental Results for Various Size Sands pipe shape and hindered settling effects at higher concen- Figure 3 shows the results for 12% concentration of tration but a more detailed analysis is not warranted at 0.18 mm sand (narrow size distribution with 90% of par-

ticles between 0.12 and 0.30 mm) at two fluid viscosities of 60 cP and 95 cP in the 105 mm loop. It can be seen that laminar flow was only obtainable at the higher vis-

4.3 Pressure Drop in "Entrance" Region cosity. In this case the slurry pressure gradient almost under laminar conditions. At distances from the entrance Consider a pipeloop through which a slurry is flowing ing near homogeneous flow with little stratification. At parallels the fluid only laminar pressure gradient suggestless than L, the particles are still settling and so a non- 0.89 ms ' a mm high stationary bed was observed. equilibrium situation exists. L, is in effect the entry length. Figure 4 shows results for the same pipeloop with 12% Thus, just as single phase laminar or turbulent flow reconcentration of 0.82 mm sand (90% between 0.60 mm quires an entry length before fully developed flow is and 1.05 mm) in sugar solutions of three different viscosireached, so the laminar flow of a slurry requires a certain ties, 115, 155 and 270 cP. At the lowest viscosity laminar entry length before the equilibrium sliding bed flow is atflow was not obtainable before deposition occurred as tained. In some cases this latter entry length can theoretiwas evidenced by the laminar/turbulent transition bursts cally be many kilometres in length. As a consequence clearly visible at deposition. developed flow. data obtained in a test loop may not pertain to fully concentration of 12% in the 18.9 mm (3/4 inch) pipe The fine, 0.18 mm, sand was next tested at the same In this entry region the pressure gradient can be estiloop in a fluid of viscosity 22 cP. The results are shown in mated by the following procedure. At any axial position Figure 5. Also shown are the results for a slightly finer the vertical height the particles have fallen is given by (0.13 mm) sand at the same concentration in the 9.41 • therefore be calculated. equation 3. The proportion of solids in the sliding bed can mm (3/8 inch) pipeloop in a fluid of viscosity of 5.3 cP. Once this is known Ø can be obtained from Figure 1 4.4.3 Comparison With Theory should be noted that equation 2 strictly only applies at inand the pressure gradient obtained from equation 2. It How do these results compare with the previously outcipient deposition but it can be used to give a rough mm pipe loop. Although this sand could be termed of lined theory? Consider first the 0.18 mm sand in the 105 estimate of the pressure gradient in other cases. narrow size distribution, because of the strong depen-

dence of W on d in equation 4 it is thought most appropriate to use the particle size pertaining to 5 cumulative

4.4 Comparison With Pipe Loop Results - Newtonian. per cent greater than (i.e. 0.30 mm) rather than the me- Fluids dian particle size. For such size particles settling in the 95 Equations 2 to 4 provide means of estimating the becP fluid (p = 1310 kg m 3) W = 7x10-4 ms-' haviour of Newtonian based slurries. This predicted beobserved deposit velocity of 0.9 ms-1 equation 4 indihaviour will now be compared with experimental results cates an entry length of 135 m meaning that everywhere for some sands in viscous Newtonian fluids, the latter within the pipe loop the flow is still developing. Equation being sugar/water solutions at various concentrations. 3 indicates that H = 4.8 mm at station A and 10 mm in 4.4.1 Description of Pipe Loops area fraction originally occupied by the now settled parthe middle of the pressure measuring section so that the 105mm mm inch) ticles was 0.0165 and 0.049 respectively. But these parmm (4 inch) pipe loop in the configuration shown in The initial experiments were conducted with the 105 ticles settle down to occupy only 0.60 of their original Figure 2(a). At this stage pipe length was not known to be area so that h/D =.016 and.033, i.e. h = 1.7 and 3.5 mm respectively. Using Figure 1 and equation 2 this latter tions most of the observations were made at A. The slurry of importance and although there were two viéwing secvalue indicates a pressure gradient at deposition of 280 Pam"! beds upstream of A so that the length to A is taken as 6m. would be expected to be remixed in the two right angled Similar calculations have been performed for the other Flowrate was measured by a venturi and occasionally gether with the measured values. It can be seen that experiments and the results are tabulated in Table 1 tochecked by timing a sample, whilst concentration was debroad agreement is obtained. The first slurry is far from termined by sampling. fully developed and so a low pressure drop and bed

198

Page 5

13 - 4-5-

B

AP (2 a)

17 10

AP NEAR APFAR

(2b)

FIGURE 2

DESCRIPTION OF ORIGINAL (a) AND ExTENDED (b) 105 MM (4 INCH) PIPELOOPS. ALL DIMENSIONS IN METRES.

height is both predicted and observed. This also explains Figures 6(a) to (d) show the results obtained at various

the near homogeneous flow noted previously. The succoncentrations and viscosities. The sand of medium par-

cessive slurries are progressively more developed resultticle size 0.31mm had a size distribution such that 90% ing in increasingly larger pressure drops and bed heights. of the particles lay between 0.19 and 0.49 mm. In every The observed bed heights are consistently higher than the case the measured pressure gradient was highest at the predictions but this is to be expected since the latter refer downstream pressure tappings. Further evidence of the to the height at incipient deposition whereas the observed "length effect" is provided by the observations of sliding values are necessarily obtained at a lower velocity. and stationary bed heights at the three viewing sections 1, Note that, whereas laminar flow was possible with the 2 & 3. These heights are given in Figures 6(a) to 6(d). All 0.18 mm sand in a 22 cP fluid in the 18.9 mm pipe, a of these results can be shown to be consistent with the viscosity of 60 cP was insufficient to obtain laminar flow previously presented theory. with the same sand in the 105 mm loop (Figure 3). Fur- It should be noted that tests with the same sand in thermore a marginally finer sand in the 9.41 mm loop rewater (i.e. turbulent flow) revealed no differences in quired a viscosity of only 5.3 cP. either the pressure gradients along the pipe length or the

bed heights at viewing stations 1, 2 & 3. This therefore

4.4.4 Testing of 0.31 mm Sand in Extended 105 mm confirms that the length effects observed are a phenome-

non peculiar to laminar flow.

Although the experiments just discussed supported the proposed laminar flow theory, to prove it more conclusively it was desirable that simultaneous observations and 4.5 Summarized Conclusions for Newtonian Vehicle pressure measurements be made at different axial It has been fairly conclusively shown that the laminar tions along the pipe loop. For this purpose the 105 mm flow of coarse particles in a viscous Newtonian fluid can pipe loop was extended to roughly twice the length with be analysed by assuming that the particles slowly settle as additional pressure tappings and viewing sections incorthey travel along the pipeline. When they reach the botporated. The dimensions are given in Figure 2(b). tom they form a sliding bed the height of which deter-

199

Page 6

4 104

2

Deposition, 95 cp Pressure gradient ( Pa m-1) 2 6 4 0•6 0-8 Velocity (ms-', 1 • Deposition, 60 cp • 95 cp 2 • 60 cp › Reynolds N° 2000 4 6 Pressure gradient (Pa m-1) RESULTS FOR 0.82 MM SAND IN NEWTONIAN FLUIDS OF 6 8 0•6 0-8 1 8 Velocity (ms -1, FIGURE 4 2 × 270 cp. • 155 cp • 115 cp 2000 Reynolds N° 4 6

FIGURE 3 VISCOSITIES 115, 155 AND 270 CP. IN 105 MM PIPELOOP.

RESULTS FOR 0.18 MM SAND IN NEWTONIAN FLUIDS OF CONCENTRATION 12% BY VOLUME. FULL LINES REPRESENT

CONCENTRATION 12% BY VOLUME. FULL LINES REPRESENT VISCOSITIES 60 AND 95 CP. TESTED IN 105MM PIPELOOP. LAMINAR AND TURBULENT BEHAVIOUR FOR FLUID ALONE.

LAMINAR AND TURBULENT BEHAVIOUR FOR FLUID ALONE.

5. LAMINAR FLOW OF STABILIZED SLURRIES

mines the pressure gradient and the deposition criteria. (NON-NEWTONIAN) The major consequences of this theory are: 5.1 Tests in Extended 105 mm (4 inch) Pipe Loop a) The obtaining of laminar flow without deposition in tests in a small diameter pipe loop does not neceswith Newtonian fluids the next step was to test a stabilized The length effect having been reasonably confirmed sarily mean that the same slurry can flow in a laminar manner without deposition (stationary bed) in a slurry in the extended 105mm (4 inch) pipe loop. The slurry chosen was a mixture of nominal - 8mm + 1 mm larger size pipe. coal (median size 2.3 mm with 90% between 0.6 mm b) The obtaining of laminar flow without deposition in and 5.8 mm) in a China clay vehicle. The rheological properties of the clay had previously been determined in tests in a short pipe loop does not necessarily mean that laminar flow without deposition can be oba tube viscometer and various pipe loops (A.D. Thomas, pipe of longer length. tained with the same slurry in the same diameter behaviour having a yield stress of 3 Pa and a plastic vis- 1978) and it was found to follow typical Bingham plastic c) Given sufficient pipe length for full settling to occur cosity of 4 cP. The solids density of the coal was 1310 kg m-3, giving it a relative density in the vehicle of 1.22. tion under laminar conditions is approximately conthe pressure gradient required to prevent deposishould be statically stable. This was confirmed by the Application of equation 1 indicates that this slurry stant for a particular commodity for all pipe sizes. This pressure gradient will be similar to that needed absence of preferential settling after having previously stood, for 15 months in drums. Even after this time the if the same coarse particles were transported in water under turbulent conditions (since equation 2 slurry. coarsest particles were still suspended throughout the in using a high viscosity Newtonian vehicle. describes both situations) indicating little advantage Tests were performed at two concentrations of coal in

the clay suspension, namely 23% and 48% by volume. 200

Page 7

C = 6% C = 6 %

• 0-18mm sand in 18-9 mm pipe, 22 cp Pressure gradient (Pa m-1) • 0-13 mm sand in 9-41 mm pipe, 5-3 cp. 9-41 mm. 5-3 cp, Pressure gradient ( Pa m-1, 10° 8 2H -- 7. 17, 25 0. 2M. 5, 15 (a) 11009 co+-2M. 8M, 12M 10° 2- •_]-3.5.12 • 0 - 0,2,5 • 0-- 2м. 5M, 8M 10049 (b) 4- 4

0•6 0-8 2 0-6 0-8 1 Velocity (ms-', Velocity (ms-')

Reynolds N° 2000:

10 C = 10 %: (c) 2- C = 14% 25,25. 5M, 10м, 20м •(d)

0-1 0-2 0-4 0-6 0-8 2 10. 14, 18 •- 2 М. 8М. • - 1 М, 2М.

VOLUME. FULL AND DASHED LINES REPRESENT LAMINAR WITH NEWTONIAN FLUIDS. CONCENTRATION 12% BY RESULTS FOR FINE SANDS IN SMALL PIPE LOOPS AND TURBULENT BEHAVIOUR FOR FLUID ALONE BE O Velocity (ms-', FIGURE 5 Pressure gradient (Pa m-1) 103 8 6 0-6 0-8 Velocity (ms -1, • - 1 • • • 2 10 6- 0-6 0-81 Velocity (ms -1, 2 The results are presented in Figure 7. In both cases sta- FIGURE 6 tionary beds were evident at all three visual sections at RESULTS FOR 0.31 MM SAND IN ExTENDED 105 MM velocities below 0.15 ms-1. Some observations from PIPELOOP AT VARIOUS CONCENTRATIONS AND FLUID

VISCOSITIES. OPEN DATA POINTS DENOTE MEASURED

these tests were: PRESSURE GRADIENTS AT FAR TAPPINGS (SEE FiGUrE 2(b)); a) Once again the measured pressure gradient was CLOSED DATA POINTS DENOTE NEAR TAPPINGS. THE higher at the downstream tappings. However an in- GROUPS OF THREE NUMBERS WITH ARROWS INDICATE teresting feature was that this difference in pressure THE OBSERVED BED HEIGHTS IN MM AT STATIONS 1, 2 & 3 gradient between the upstream and downstream RESPECTIVELY. M INDICATES A MOVING BED. tappings was more evident in the laminar/turbulent transition region (0.8 <V<1.6 ms-1). Once pure Pa m-1 for the 23 and 48% concentration respectively. laminar flow was obtained the difference became The observed values are about one half of these predicmuch less. This effect has also been noticed in other tions. This could indicate that complete settling has not tests not presented here. occurred in the pipe length available, but, the use of b) The exact heights of sliding beds were not easy to equation 3 and 4 to check whether this is the case is determine. This was particularly so at low velocities made impossible because of the uncertainty regarding the where the slurries tended to move en bloc even settling rates of coarse particles in a flowing Bingham though a vertical concentration gradient was clearly plastic. evident. Furthermore the beds did not appear to be The settling of solid spheres in stationary Bingham so compact as with Newtonian fluids. plastics is far from satisfactorily understood at present c) In spite of the difficulty in observing the actual bed (e.g., Ansley & Smith, 1967) so that an analysis of the heights there did not appear to be any significant settling of particles in a horizontally sheared Bingham differences in bed height between the three viewing plastic is a very difficult problem. The settling velocity, W, stations. would no longer be constant due to the different resis-

tance of the sheared layer near the walls compared with

5.2 Analysis of Results the less sheared core region. At high shear rates (high If it is tentatively assumed that all of the coal has settled flow velocities and small pipe sizes) shearing will occur to form a sliding bed then application of equation 2 indiover the whole pipe area and the floc structure will be cates pressure gradients at deposition of 400 and 750 considerably broken down resulting in relatively high par-

201

Page 8

behaviour could be explained by the previous theory but

2 there are a number of factors which suggest that a differ-

Near tappings 48% shortly in Section 5.4. gradient (Pa m-1, • For tappings, 23% • Near tappings? gate the concept of laminar flow transport of coarse parent explanation exists. This matter will be discussed Elliot & Glidden (1970) were among the first to investi- 10' a Far tappings J Transition region

5.3 The Work of Elliot & Glidden (1970)

ticles. They tested various coal slurries in pipeloops up to

Pressure Deposition: 1.6 km (1 mile) in length and 250 mm (10 inch) in diamadvanced here as slurries were successfully pumped over

eter. Their results would seem to disprove the theory long distances apparently without settling occurring in the 100 mm (4 inch) pipe loop, although, it should be noted that the pressure gradient in that pipe loop was almost

0-1 0-2 0-4 0-6 0•8 1 2 4 what would be expected from a sliding bed approach. Of greater significance are their tests in the larger (250 Velocity (ms-", mm) pipe loop. In this loop they obtained laminar flow

without deposition at pressure gradients as low as 110 Pa

FIGURE 7 m-1. The only way that such low pressure gradients RESULTS FOR 8x1 MM COAL IN CLAY SUSPENSION IN could be explained by the previous sliding bed theory is if EXTENDED 105 MM PIPELOOP. FULL LINES REPRESENT incomplete settling had occurred in the relatively short BEHAVIOUR OF CLAY SUSPENSION ALONE. length loop. (It was 365 m (1200 ft) long resulting in a

similar L/D ratio as the author's 105 mm (4 inch) loop). However an alternative, and it is believed, more plausible

ticle settling rates. In contrast, at low shear rates (low flow explanation, is provided by the theory advanced in the

velocities and large pipe sizes), most shearing will occur following section.

near the walls with the floc structure in the core region remaining largely intact. In true Bingham plastic flow the 5.4 Proposed Theory for Stabilized Slurries core region could be completely unsheared so that the Although the results for Newtonian based slurries are floc structure would be completely unbroken and so satisfactorily explained by the previous theory the results theoretically be able to support some particles indefinitely obtained with stabilized slurries both here and by Elliot & just as under static conditions. To what extent this occurs Glidden (1970) suggest some additional phenomenon. It in practice is at present unknown. The author's experis now proposed that with stabilized slurries, although setience has been that with coarse particles present in flowtling does occur as the slurry flows along the pipeline, the ing Bingham plastics there is generally observed a vertical settled bed is less compacted than with Newtonian vehivelocity and concentration profile indicating that no comcles. This results in less particle/particle interaction and pletely unsheared region exists. This would suggest that, hence lower pressure gradients. The lower degree of given sufficient pipe length, the coarse particles would compaction could be due to the coarse particles being always eventually settle. kept separated by the compressible flocs. This would be Because of the uncertainty as to the degree of settling more likely to occur at low shear rates when the flocs are which has occurred in the pipe loop length these stabinot broken down to any great extent and would explain lized slurry tests are not easy to interpret. The observed the en bloc movement observed at low velocities in Sec-

• TABLE 1

COMPARISON BETWEEN THEORY AND EXPERIMENT FOR SANDS IN NEWTONIAN FLUIDS

ALL AT A CONCENTRATION OF 12% BY VOLUME

Particle Fluid Pipe Pressure Drop Bed Height Size Viscosity Size (Pa m - 1) (mm) Comments (mm) (cp) (mm) Predicted Measured Predicted Measured 0.18 95 105 280 500 1.7 3 Still developing (L, = 135 m) 0.82 155 105 1020 1060 8 25 Partly developed (L = 26 m) 0.82 270 105 900 1400 7 30 Party developed (L, = 26 m) 0.18 22 18.9 1700 1600 5 Fully developed (L, = 1.8 m) 0.13 5.3 9.41 1800 2200 2.5 3 Fully developed (L, = 2.2 m)

202

Page 9

70

60 - Bingham curve Ty= 26. Mpt =0-020 Loam

39%% 2

50 m-1,

30 20% 320% Transition scale Predicted by up BED

to turbulence 32%

PIPELOOP. FULL LINES REPRESENT FITTED BINGHAM CURVES. RESULTS FOR TWO CONCENTRATIONS OF LOAM IN 18.9 MM DJ/4 (Pa) 10 200 400 Bingham curve Ty =10. 7p=0-009 FIGURE 8 600 3V/D (sec - 1) 800 1000 1200 1400 1600 / Pressure gradient (Pa 102 0•4 1 0-6 0•8 1 39% Waler 6

Velocity (ms-1)

tion 5.1. In the laminar/turbulent transition regime the FIGURE 9 flocs are disrupted by the turbulent bursts so that the RESULTS FOR TWO CONCENTRATIONS OF LOAM IN 105 MM situation would be closer to the Newtonian case. This PIPELOOP. FULL LINES REPRESENT BEHAVIOUR PREDICTED could explain the greater differences between the BY SCALING UP FROM THE 18.9 MM PIPELOOP RESULTS. upstream and downstream pressure gradient observed in the transition regime, (see Section 5.1). The degree of compaction of the bed will depend on the floc strength. Under static conditions the yield stress is 5.5 Consequences for Stabilized Slurry Concept a measure of the strength of the flocs but in a flowing pipe The major consequence as regards the pumping of their strength will be reduced by the shearing action and stabilized slurries is that the design of such systems is will probably be a function of the shear rate 8V/D. (V is complicated both by test loop length effects and by the velocity, D is pipe diameter). For a given velocity the uncertainty as to the exact influence of floc support. The shear rate is less in a larger pipe meaning that the floc problem is best illustrated by an example. Figure 8 shows strength will be higher and the settled bed less compact the results of tests by the author in the 18.9 mm (3/4 than in a smaller pipe. The result is that instead of the inch) pipeloop of loam (a mixture of coarse (0.35 mm) pressure gradient at deposition being constant for all pipe sand and clay) at two concentrations, 32 and 39% by sizes, as the sliding bed theory predicts for Newtonian vevolume. (These tests have previously been referred to. hicles, it will decrease with larger pipe sizes. The size of 'A.D. Thomas (1978), but no details were presented at the decrease will depend on the relative strength of the that time). The results are plotted as wall shear stress verfloc structure and the size of the particles. sus apparent shear rate. In both cases velocities as low as In some cases the effect may be sufficiently great to 0.009 ms- 1 (.029 ft/sec) were obtained with no deposicause Ja to vary inversely with pipe diameter meaning that tion observed. Both slurries were statically stable. The full deposition occurs at a constant value of wall shear stress, lines drawn through the data represent the theoretical DJ./4, in agreement with the previous findings of the curves for Bingham plastics having the yield stresses and author (A.D. Thomas, 1978). This would also mean that plastic viscosities indicated. The laminar flow data are the usual scale up methods for pressure gradient would seen to fit these curves reasonably well especially for the apply so that the statement by Pertuit et al (1978) that; 32% concentration. "the stab-flo headlosses decrease in proportion to the in- Having obtained such data one would be tempted to crease in pipe diameter" would be correct. However this assume that the behaviour in a larger size pipe could be is not true in general. The behaviour of a particular stabicalculated by direct scale-up from this plot following the lized slurry could range anywhere from this ideal case to method of Bowen, (1961). Figure 9 shows the resulting the worst case approaching the Newtonian vehicle situalaminar flow predictions for a 105mm (4 inch) pipe, tion where Ja is constant for all pipe sizes. Unfortunately along with experimental data obtained at the two concenno means are presently available to quantify this effect. trations in the pipe loop described in Figure 2(a). The

203

Page 10

behaviour of the 39% concentration slurry is predicted c) Even if laminar flow is possible in a larger pipe the quite closely and no stationary deposit was observed pressure gradient may be considerably higher than even at the lowest velocity tested. Thus the scale-up techindicated from normal scale up relations. Dependnique can be considered successful. However, the 32% ing on the particular slurry properties, the variation concentration case was not so successful. Laminar flow of pressure gradient with pipe size can lie anywhere was barely possible before a stationary 10mm (0.4 inch) between the ideal situation, where J varies inhigh bed appeared at 1.2 ms-1. Thus, supposing for exversely with pipe diameter, to the worst situation, ample that the design was based on an operating velocity given by the Newtonian vehicle theory, where J is in the 105 mm (4 inch) pipe of 1 ms-!, it can be seen that constant for all pipe sizes. this would not be obtainable and blockage would probably result. This example has served to show that, just as in the Newtonian vehicle case, successful laminar flow opera- 7. ACKNOWLEDGEMENTS tion in a particular test loop does not necessarily mean The author wishes to thank M.D. Research Co. Pty that the same slurry can flow laminarly in a larger size Ltd. for permission to publish this paper. In addition pipe. Presumably, even though laminar flow was suc: special acknowledgements are due to Mr. P.N. Mitchell cessfully obtained in the 105 mm (4 inch) pipeloop at who first suggested to the author the possibility of slow 39% concentration, if that slurry were pumped in a larger settling of particles along the pipe length in laminar flow. size pipe laminar flow without deposition might not be possible.

Editors Note: Dr. Thomas was unable to deliver the

6. CONCLUSIONS paper personally. Norman T. Cowper, an independent

slurry consultant from Sydney delivered the paper in his

The results of tests with coarse particles transported stead. Mr. Cowper also answered questions from the under laminar flow conditions in viscous Newtonian fluids floor. can be explained by assuming that the particles slowly settle as they travel along the pipeline. When they reach the bottom they form a sliding bed the height of which 8. REFERENCES deposition. The major consequences were summarized determines the pressure gradient required to prevent 1. Ansley, R. W. and Smith, T. N. (1967). Motion of in Section 4.5, the net result being that such slurries spherical particles in a Bingham plastic, A. I. Ch. would not be attractive propositions due to the high E. Jnl, Vol. 13, No. 6, pp. 1193-1196. pressure gradients required especially in large pipes. 2. Bowen, R. L. (1961). Designing laminar flow sys- Available results obtained with stabilized slurries, i.e., tems, Chemical Engineering, June 12, pp. 243coarse particles suspended in flocculated non-Newtonian 248. vehicles, are less conclusive and further work is required. 3. Elliot, D. E. and Gliddon, B. J. (1970). Hydraulic Nevertheless in this paper a qualitative theory has been Transport of coal at high concentration, Proceedadvanced for such slurries which could explain their beings 1st Int. Conference on the Hydraulic Transhaviour. As in the Newtonian case this theory assumes port of Solids in Pipes, Organised by Brit. Hydrothat slow settling takes place to form a sliding bed. Howmech. Res. Assoc., Cranfield, paper G2. ever, in this case the sliding bed is less compacted, the. 4. Halvorsen, W. J. (1976). Slurry pipeline hydrauparticles being kept apart by the compressible floc struclies improved, The Oil and Gas Jnl, March 22, ture. This results in less particle/particle interaction and pp. 62-66. lower pressure gradients. Although this improves the sit- 5. Love, F. H. (1969). The Black Mesa story, Pipeuation and makes the stabilized slurry concept commerline Engineer, Nov. pp. 38-42 cially feasible the design of such systems needs to be han- 6. Pertuit, P., Tennant, J. D., Lawler, H. L. and dled with care. Scale up on pipe diameter is difficult and Cowper, N. T. (1978). Application of stabilised ideally tests would need to be carried out on the full size slurry concepts of pipeline transportation of large unless a pipeloop of sufficient length is employed. Transportation, March 29-31, Las Vegas, pp.

164-176.

concept are summarized below: pipe. Even then "length" effects can obscure the issue plead a eat tempted dum, 7. Thomas, A. D. (1978). Coarse particles in a heavy medium-turbulent pressure drop reduction and particle coal., Proc. 3rd Int. Tech. Conf. on Slurry

deposition under laminar flow. Proc. 5th Int. Conf. on the Hydraulic Transport of Solids in

flow is possible in all pipe sizes. Pipes, May 8-11, Hannover, Paper D5. b) Even if pipe loop tests reveal that laminar flow is 8. Thomas, A. D. (1977). A rational design philosopossible in a particular size pipe it does not necesphy for long distance slurry pipelines, Chemical sarily follow that laminar flow is possible in a larger Engineering in Australia, the Trans. of the Instn. of pipe with the same slurry. Engineers, Aust. pp. 22-33.

204

Page 11

9. Thomas, D. G. (1961). Laminar flow properties of sus sheer rate basis, so there's validity in that scaleup if it's flocculated suspensions, A. I. Ch. E. Jnl, Vol. 7, stabilized. The question Dr. Thomas presents is when a No. 3, Sept., pp. 431-437. slurry that may look stabilized in static condition is sub- 10. Traynis, V. V. (1977). Parameters and flow rejected to shear in a pipeline, it changes its characteristic, gimes for hydraulic transport of coal by pipeline, Translated from Russian by Terraspace Inc., Rockparticularly if the yield stress of the slurry if exceeded and

it tends to break up in deposit.

11. Wasp, E. J. (1969). What slurry pipelining is all ville, Md., U.S.A based on what Glidden and Eliot presented. I don't know But the scale up, if it's truly stable, then the scale up is 12. Wilson, K. C. (1974). Co-ordinates for the limit of about, Pipeline Engineer, Nov. pp. 30-35. if that answers your question, but this is the subtle point deposition in pipeline flow, Proc. 3rd Int. Conf. on • that he was looking at. I am sure that it's achievable, a stabilized slurry that would actually be a transport the Hydraulic Transport of Solids in Pipes, May 15thstabilizer. 17th, Golden, Colorado, U.S.A. Paper E1. MR. McDOWELL: Bob McDowell from Kilbourn,

Ltd. In regard to the statement that as pipeline sizes and lengths increase and thinking of the pipeline lengths and sizes we have been hearing about here for the last few

QUESTIONS AND ANSWERS days, what do you see as the sort of practical advantage DR. STRIPLING: My name is Travis Stripling. with you are talking about versus simply pumping with water in trying to commercially develop a stabilized type slurry the work by Eliot and Glidden? It was published at Brown and Root. I have a question. Are you familiar with for this that would be commercially beneficial? in a turbulent flow? Can you see any actual application Hydro-Transport. One concerning pressure loss calcula- MR. COWPER: Yes, there is a market for this particutions for coal slurry, stabilized coal slurry. They did experilarly in metallurgical coal. The coal is coarser than that of inches in diameter. mental studies on pipes from about one and a half to ten conventional slurry pipeline system. So particularly in MR. COWPER: Yes. I am very familiar with that. to handling three-quarter inch top size material or one Australia where we face an export market which is used concept you can scale to larger pipe sizes using a gen- DR. STRIPLING: My question is: they put forth the and a half inch top size material, there's a need to develop a system to transport that material. It's a market eralized Reynolds number type correlation. How do your situation. results affect this concept? Is that still valid according to your results? There are other approaches. We could stick with the MR. COWPER: If it's a truly stabilized slurry they're conventional coal slurry system and use agglomeration or some other technique to produce a coarser product in the valid. Really only scaling up on a straight shear stress verend of the system.

205