QUANTITATIVE VISUALISATION OF PULP REFINING IN A PRODUCTION LINE REFINER

Taito Alahautala, Juha Vattulainen and Rolf HernbergTampere University of Technology, Physics Department, Plasma Technology Laboratory, P.O. Box 692, FIN-33101TAMPERE, FINLAND

ABSTRACT

In this work pulp flow in a production-line first-stage TMP-refiner operating at 10 MW has been visualised usingmultipulse stroboscope, endoscope optics, specialised illumination system and a CCD camera. Measurements havebeen performed at different radial locations and with different refiner operation parameters. The parametersmeasured from the plate gap were pulp velocity (magnitude and direction), pulp orientation, percentage area coveredwith pulp and the presence of fibre flocs .

1 INTRODUCTION

Thermomechanical pulping is one of the most important mechanical pulping methods. Many parameters of TMPprocess are unknown and it is believed that the process still involves a large potential for energy saving. Opticaltechniques are among the most promising in producing qualitative and quantitative information of the refiningprocess. Only few optical measurements of production scale refining phenomena have been made and reported inliterature. Visualisation work has been done mainly by Canadian researchers Atack et al. and Stationwala et. al. inthe 1980's.

The first measurements were carried out at two different size and type of refiners [1,2]. The smaller one was a singlerotating disk refiner with capacity of 1,9 MW and it was operating at atmospheric pressure. The larger one was also asingle rotating disk refiner with capacity of 5 MW and it was operating as a second stage of a two-stage tandempressurised refining line. Both refiners were visualised through plastic stator plate segment using high-speed camera.Pulp flow patterns were observed to be similar regardless of different throughput rates and the manner of operation.Considerable recirculation pattern was observed in these studies. Back flow velocity in the 5 MW refiner was in therange of 0-30 ms-1 depending on the observation location in the plate segment "V" pattern. Back flow was found tobe a significant process compared to average forward flow velocity of 10-20 ms-1 in the refining zone.

Studies of pulp fibre behaviour in refiner bar surface require very short effective exposure times. Effective exposuretime determines the degree of motion blur in the images. Effective exposure time is limited by camera shutter or bythe duration of the stroboscopic pulses. Motion blur was 0,4-0,8 mm in the first refiner visualisations [1,2] accordingto the velocities of 10-20 ms-1 and 40 µs camera exposure time.

Both stroboscopic illumination and a fast shutter high-speed camera were applied to study a1,9 MW single rotating disc refiner with high accuracy [3]. The refiner was operating in a second stage position andat atmospheric pressure. Stroboscopic flash production was synchronised to the rotor axis and single 1 µs flasheswere produced at 30 Hz rate. The per cent bar area covered with pulp was found to be lowest in the refinerintermediate section (50-70%) being 70-80% in the inner section and 80-87% in the outer section. A fast shutter andhigh frame rate IMACON camera was used to produce high magnification photo-series of eight images. Themovement of pulp floc over stator bar was observed with resolving power high enough for individual fibre detection.

High-speed camera techniques has also been used to study wood and pulp behaviour in laboratory scale refiner [4]and to study low consistency fibre refining mechanisms [5].

The reason why optical methods have not been very widely used to study pulp refining in production-line refiners isdue to the extremely hostile conditions in the plate gap. Optical window material, and also the size, shape andlocation of the window need to be designed to withstand the heat, pressure and mechanical forces of the process. Thewindow needs to be highly transparent and at the same time the material should be such, that it can be machined into

appropriate windows. The camera system has to be able to freeze movement. This requirement becomes even moredemanding when high magnification is used. Quantitative information like pulp coverage percentage in certain areacan be calculated from single blur free images. Quantitative information of velocity requires very high frame rates ormultiple stroboscopic pulses.

During this work a new visualisation system has been constructed and used to study refining phenomena in aproduction-line refiner. The first field test was performed in 1996 [6]. Quantitative information is produced usingmultipulse stroboscopic illumination. Optical access to the refiner has been realised through sapphire windows andhigh magnification is achieved using endoscopic optics. Advantage of stroboscopic method is its freedom for cameraselection and the fact that all temporal information is included into single images.

The method has been applied to a production-line first stage TMP refiner (SD-65) in UPM-Kymmene Kaipola mill.The parameters measured from the plate gap were pulp velocity (magnitude and direction), pulp orientation,percentage area covered with pulp and presence of fibre flocs. Also statistical values were calculated.

2 EXPERIMENTAL METHOD

2.1 Overview of the visualisation systemPulp refining phenomena were visualised through plate patterned sapphire windows using an endoscope andscientific grade CCD cameras. Stroboscopic illumination was produced by two high power xenon flash lamps andalternatively by fully programmable stroboscope constructed of semiconductor diode lasers. High magnificationendoscope was used to relay image from the plate gap to the CCD camera with resolving power of about 10 µm.Illumination and image frame grabbing were synchronised to the rotor rotation. Images were produced at maximum25 Hz rate. Visualisation studies were performed at four radial positions and at three refiner operation points.

Installation probes were screwed into each of the four holes located in different radial positions in the refiner frontcover. Imaging probe, consisting of endoscope and illumination pipe, was then inserted into each installation probein turn. The imaging probe was cooled below 100 °C using filtered air.

Fig. 1. Schematic diagram of the visualisation system.

2.2 Principle of multipulse stroboscopy and quantitative informationIn multipulse stroboscopy the target is illuminated using two or more light flashes in the same image. Target movesto a new position during the delay time between successive flashes. Velocity is obtained from the displacement whenknowing the delay time. Displacement was measured from floc border movement or fibre movement.

In this work pulp velocities were measured with accuracy of 5,3%-9,1% depending on the case [7]. Illumination wasproduced by driving two xenon flash lamps with adjustable delay time. The duration of single xenon flash was 2 µswith electrical energy of 0,8 J. It is estimated that the target was illuminated by pulses of 4 mJ optical energy whenthe collection and transmission efficiencies are included.

In the field measurements also a new type of stroboscope based on high power diode lasers was tested. Thesesemiconductor lasers were supplied by SemiLab at Tampere University of Technology (TUT). Diode lasers are ideallight sources for arbitrary pulse (train) generation. The stroboscope constructed in this work producedmaximum 30 W red light (680 nm) pulses. Laser pulses provided very sharply focused images and also showedwater-air interfaces very effectively. It is estimated that the optical pulse energy focused to the target was 0,1 mJ.The power levels of the prototype laser system were however not high enough with the highest magnifications of thelens system.

2.3 Window constructionsWindows were made from synthetic sapphire of 20 mm in diameter and thickness. Conical windows were installeddirectly into the stator plate segment using conical holders (Fig. 7). In order to achieve condense and durable fittingthe conical angle of sapphire window was manufactured to a precision of less than 1'. Thin aluminium layer wasadded between the window and its holder. Aluminium was used for tightening and sealing purpose and also forguiding illumination.

Fig. 2. Double xenon flashes. Scale of thehorizontal grid is 5 µs and vertical axis is A.U.Pulse FWHM is 2 µs (fixed) and delay time 20µs (adjustable).

Fig. 3. Triple laser pulses with 20 µs delay time.Arbitrary pulses or pulse trains could be producedby laser stroboscope. For example amount ofpulses, duration of individual pulses and delay timebetween pulses could be adjusted.

Fig. 5. Pulp velocity determinationby fibre movement. The twostroboscopic pulses were producedusing diode lasers.

Fig. 4. Pulp velocity determination byfloc border movement. The twostroboscopic pulses were producedusing xenon flash lamps.

Two windows were constructed for refiner crush zone at radial positions of 537 mm and 607 mm. One window waslocated into the intermediate zone (669 mm) and one for refining zone (797 mm). All windows had a plate patterncorresponding their locations manufactured onto the surface (Fig. 6). Effective optical aperture of all windows was12 mm. All windows remained unchanged during the field measurements.

2.4 Illumination optimisationLight was conducted from xenon flash lamps to the refiner plate gap using an optical fibre bundle and a lightconduction pipe made of quartz. Light conduction pipe was machined to provide uniform illumination in the targetarea and also to prevent reflections from the window surfaces. The same light conduction pipe was used also with thelaser stroboscope.

Fig. 6. Sapphire windows. Top windows aresimilar and they were located in radialpositions of 669 mm and 797 mm. Grooveis in the middle of the window. Bottomwindows were located in 537 mm and 608mm and the groove step is in the middle ofthe window. Conical surface of the windowsis aluminium coated for light conduction andprecise fitting and sealing.

Fig. 7. Window installation. Windows wereinstalled directly into the refiner stator platesegment. In this figure illumination and imagingdirection is from left to right.

Fig. 8. Sapphire window installed intothe stator plate (797 mm). Fig. 9. Window locations in the stator plate segment.

2.6 Refiner operation pointsField measurements were performed in UPM-Kymmene Kaipola mill during 11-12 February in 1998. Refiner wasoperating as a first stage refiner in a two-stage production line. Different refiner operation points were studied bychanging the amount of dilution water (Table 1). During two days measurements over total of one million imageswere recorded.

3 RESULTS

A brief discussion of results is presented here according to different refiner plate zones. Velocity, fibre orientationand area covered by pulp is presented in the Figures 10, 11 and 12. Numerical data can be found from Tables 2, 3and 4 (Appendixes). Each pulp flow parameter is an average value of approx. 50 single samples. Parameters werecalculated using image processing methods and visual verification.

Table 1. Refiner operation points (op). In op2 maximum power was 9,54 MWand minimum 9,18 MW. In op3 maximum power was 9,66 MW and minimum9,29 MW. In op4 maximum power was 10,77 MW and minimum 10,22 MW,correspondingly.

Operationpoint (op)

Dilution water(litre/s)

Power(MW)

Stdv of power(MW)

Production(1000 kg/hour)

2 2,1 9,37 0,08 8,8

3 0,2 9,47 0,08 9,2

4 1,0 10,45 0,09 8,6

Fig. 10. Pulp velocity and direction as a function of radius and refineroperation point

Fig. 11. Fibre orientation as a function of radius and refiner operation point.

3.1 Crush zonePulp back flow velocity in the crush zone was found to be surprisingly small, only a few centimetres per second. Inthis region the velocity values were calculated from sequential images taken at video frequency (25 images/s). Thecrush zone was full of pulp and no gaps were observed. Back-flow velocity above the bar was 130 % of that in thegroove. Increase of dilution water increased back flow velocities.

Pulp volume fraction in the crush zone was high and probably in this zone there exists also some free water. Thelatter observation is based on the observation that wood chips or coarse fibres were difficult to distinguish in theimages. Same type of image appearance was observed in the laboratory when pulp was immersed into water.Because of the high pulp coverage value, and thus low transparency, the results obtained from crush zone describesituation exactly at the stator plate surface.

Fig. 12. Percentage area covered by pulp.

3.2 Intermediate zonePulp flow in the intermediate zone groove was very turbulent. Average back flow velocities were about1 ms-1 showing no linear dependence on the amount of dilution water. Per cent covered by pulp in the groove was 60.This indicates the actual pulp volume fraction to be far less than 60 %. Pulp water consistency was high in theintermediate zone groove and water droplets were also observed.

3.3 Refining zoneAverage pulp velocity in the refining zone groove was about 30 ms-1. This reduced to about 20 ms-1 when the amountof dilution water was increased. Pulp velocities in the bar region could not be measured because fibres weretangentially oriented along the propagation direction. The per cent covered by pulp in the refining zone groove wasabout 6 indicating pulp volume fraction of only a few per cents. Flocks were observed in the groove about in every100th image which also indicates low pulp volume consistency.

4 SUMMARY

In this work processes in the plate gap of a first stage TMP refiner were visualised for the first time when running therefiner with completely realistic production-line power levels and other operating parameters. The quantitativeparameters determined from the images were pulp velocity, flow direction, pulp orientation, percentage area coveredwith the pulp and the presence of fibre flocs. These measurements were performed at four different radial locationsranging from the crushing zone out to the refining zone. The determined parameters can give information about theactual operation of the refiner, because these values were calculated as average values from images recorded over aconsiderable long period of time. Thus the values are not severely affected by any short term changes in the refineroperation. Standard deviations of the measured parameters are several times larger than the precision of thestroboscopic method itself, which also implies that the reliability of the measurement results is high.

To accomplish completely realistic conditions in the measurement volume, the optical window surface should havetotally identical properties compared to the corresponding original metallic plate surface. It is obvious that here themeasured flow parameters in windows located at 669 mm and 797 mm are in some degree affected by the fact thatthe neighbouring grooves in the window region are missing. How much this really affects the pulp flow values ishowever unknown. Also different friction coefficients of the window surface and the machined metallic plate surfaceshould be taken into consideration. The effect of different friction coefficients cannot be totally eliminated, but it canbe minimised by using as small windows as possible, as was done in this work.

In order to measure accurately the highest velocities in the refiner, very high temporal resolution is required. Whenindividual fibres to be distinguished are moving at 100 ms-1 the effective exposure time must be less than 1 µs.

The multipulse stroboscopic method seems to be a promising tool for high velocity target visualisation. According tothe experience gained in this work it is possible in the future to construct a diode laser stroboscope with power ofseveral hundreds of watts. This gives new possibilities for quantitative, high magnification visualisation of thethermomechanical refining phenomena.

5 REFERENCES

1 Atack D., Clayton D.L., Quinn A.E. and Stationwala M.I., SPIE High Speed Photography (Strasburg), "High Speed Photography of Wood Pulping in a Disc Refiner", 491:348-353 (1984).

2 Atack D., Stationwala M.I., Huusari E., Ahlqvist P., Fontebasso J. and Perkola M., Paper and Timber, "High- speed photography of pulp flow patterns in a 5 MW pressurized refiner", 6:689-695 (1989).

3 Stationwala M.I., Atack D. and Karnis A., J. Pulp Paper Sci., "Distribution and Motion of Pulp Fibres on Refiner Bar Surface", 18(4):J131-J137 (1992).

4 Demler C.L., TAPPI Proceedings 1994 Papermakers Conference, "Another Attempt at Refining Visualization", 101-114 (1994).

5 Oullet D., Bennington C.P.J. and Potkins D., J. Pulp Paper Sci., "Wood Comminution and Material Flow in a Laboratory Chip Refiner", 21(12):J415-J421 (1995).

6 Alahautala T., Vattulainen J. and Hernberg R., XIV IMEKO World Congress, "Visualisation of Pulp Refining in a Rotating Disk Refiner", XA:60-64, 1997.

7 Alahautala T., Vattulainen J., Lassila E. and Hernberg R., Tampere University of Technology: Physics Report 6-98, "Visualisation of Pulp Refining in a Rotating Disk Refiner", in Finnish, ISBN 952-15-0065-4, ISSN 0359-811X, pp. 36, 1998.

ACKNOWLEDGEMENTS

This research was funded by Finnish Technology Development Centre, TEKES (program SUSTAINABLE PAPER),UPM-Kymmene, Sunds Defibrator and Valmet Corp. Stock Reparations. The work was carried out during 1996-1998.

TABLES

Table 2. Measured parameters in refiner operation point 2 (power 9,37 MW, dilution water2,1 litre/s and production 8,800 kg/hour).

Explanations:v is velocity (ms-1)vdirection is the direction of velocity (degrees from radial direction, anticlockwise)o is fibre orientation (degrees from radial direction, anticlockwise)c is pulp coverage ( i.e.per cent area covered by pulp)f is flock frequency (i.e. frequency in images where flock is present)

537 mm 608 mm 669 mm 797 mm

Parameter Bar Groove Bar Groove Groove Groove

v(ave) 0,023 0,021 0,032 0,025 1,18 19,2v(stdev) 0,018 0,015 0,018 0,012 0,44 6,96v(max) 0,079 0,081 0,073 0,042 3,11 35,8v(min) 0,004 0,0016 0,012 0,0052 0,49 6,0

vdirection(ave) 186,6 188,8 181,8 167,9 169,0 -0,4vdirection(stdev) 7,8 7,8 7,3 12,9 16,2 13,0vdirection(max) 206,9 204,9 202,1 183,1 204,5 41,5vdirection(min) 167,9 171,9 166,1 153,1 141,5 -29,5

v(rad) -0,023 -0,021 -0,032 -0,024 -1,16 19,2v(tan) 0,0027 0,0032 0,001 -0,005 -0,22 0,13

o(ave) 25,7 -7,8 44,2 67,5 6,9o(stdev) 31,1 34,5 28,8 37,1 15,3o(max) 110,9 92,9 120,1 126,1 62,5o(min) -51,1 -63,1 -16,9 -16,9 -20,5

c(ave) 57,5 8,1c(stdev) 22,7 9,4c(max) 90,9 54,4c(min) 11,4 0,0

f(ave) 112

Table 3. Measured parameters in refiner operation point 3 (power 9,47 MW, dilution water0,2 litre/s and production 9,200 kg/hour).

Table 4. Measured parameters in refiner operation point 4 (power 10,45 MW, dilution water1,0 litre/s and production 8,600 kg/hour).

537 mm 608 mm 669 mm 797 mm

Parameter Bar Groove Bar Groove Groove Groove

v(ave) 0,037 0,029 0,054 0,033 1,07 33,2v(stdev) 0,019 0,019 0,031 count=2 0,34 8,8v(max) 0,088 0,098 0,129 0,041 1,79 60,0v(min) 0,006 0,006 0,011 0,040 0,47 31,4

vdirection(ave) 184,9 185,3 176,8 168,1 194,4 -0,7vdirection(stdev) 10,9 10,6 10,2 count=2 21,3 11,0vdirection(max) 226,9 214,9 189,1 172,1 237,5 14,5vdirection(min) 121,9 161,9 144,1 164,1 138,5 -31,5

v(rad) -0,036 -0,029 -0,053 -0,032 -1,04 33,2v(tan) 0,003 0,0027 -0,003 -0,007 0,27 0,4

o(ave) 39,1 0,2 43,1 98,5 8,1o(stdev) 34,2 32,3 31,6 40,2 12,6o(max) 131,9 110,9 119,1 146,1 50,5o(min) -10,1 -48,1 -13,9 -17,9 -15,5

c(ave) 58,4 4,8c(stdev) 19,2 2,8c(max) 93,6 11,9c(min) 17,0 0,0

f(ave) 165

537 mm 608 mm 669 mm 797 mm

Parameter Bar Groove Bar Groove Groove Groove

v(ave) 0,013 0,010 0,029 0,026 0,98 31,5v(stdev) 0,010 0,010 0,016 0,012 0,37 8,3v(max) 0,061 0,053 0,084 0,045 2,24 55,7v(min) 0,002 0,002 0,005 0,014 0,22 15,7

vdirection(ave) 191,3 193,6 183,2 176,7 165,3 5,6vdirection(stdev) 8,6 13,3 16,8 22,1 18,4 9,1vdirection(max) 224,9 267,9 233,0 220,1 214,5 33,5vdirection(min) 171,9 186,9 121,1 155,1 102,5 -13,5

v(rad) -0,012 -0,010 -0,029 -0,026 -0,95 31,3v(tan) 0,0025 0,0024 0,0016 -0,0015 -0,25 -3,1

o(ave) 40,6 -41,0 53,2 87,5 12,1o(stdev) 25,1 44,2 27,2 55,9 3,7o(max) 118,9 31,9 100,1 146,1 18,5o(min) 7,9 -156,1 7,1 -31,9 3,5

c(ave) 61,5 6,6c(stdev) 20,7 8,8c(max) 95,9 56,4c(min) 10,2 0,0

f(ave) 94