Grain-Chaff separation in a vertical air stream

Grain-Chaff

Separation

in a Vertical

I. G. FARRAN*:

R. H.

Air Stream

MACMlLLAYf

Theoretical and experimental studies were made of the separation of harvested materials in a vertical air stream. Grain and either polystyrene balls (simulating chaff) or actual sieve materials were injected at rates comparable to those occurring in combine harvesters. The results suggest that, not only are air velocity and feed rate important in determining the effectiveness of separation and the level of grain loss, so also are the speed and, direction (relative to the main air stream) ot the injected material.

1.

Introduction

The material presented to the sieves of a combine harvester for separation is a complex mixture of fractions (particle types) each with a range of aerodynamic and mechanical properties. The most appropriate property to be exploited in any such process is one which meets the criteria that all material containing grain should be retained and all non-grain material should be rejected. A previous study of the characteristics of typical sieve materials’ suggests that no single aer+ dynamic or mechanical property, which can meet this criteria, is available. However, at least two non-grain fractions, viz. chaff and threshed heads, which together constitute a large proportion of the total mass of material, have terminal velocities substantially less than grain. It therefore follows that it may be possible to make an initial separation of these non-grain particles using their aerodynamic characteristics. To do so would greatly reduce the material input to any subsequent separation process. As a result of their work on the conventional sieve system, Persson2 and Lee and his coworkers3,4 have already noted the importance of aerodynamic action in the separating process. The latter workers show that at low feed rates the chaff and straw mixture was carried in a dispersed state about the sieve by the airstream; grain loss was small and nearly constant. There was however a change in the mode of separation and a significant increase in grain loss when, due to increasing feed rate, the mixture came in contact with the sieve and was carried rearwards partly by mechanical and partly by aerodynamic action. When the mixture was allowed to fall through an air stream before reaching the vicinity of the sieve. thus increasing the aerodynamic component, the feed rate at which this change in mode occurred was twice that without the air stream. The complex form of the system and of the air stream used by Lee did not allow the influence of the various parameters controlling the aerodynamic component of separation to be evaluated. Persson considered only single particles, while the work of Kashayap and Pandya5 was limited to very low feed rates. There is therefore little published information on the process of injection of a two phase material at practical feed rates and the objective of the present work was to study this aspect of aerodynamic separation of materials in a duct of simple geometric form. In particular the influence of air velocity and of feed rate, speed and direction of the input materials on their separation in a vertical air stream were examined.

116 2.

Experimental 2.1.

layout and techniques

E.uperimenfal apparatus

The following constraints were important in the design of the apparatus. particularly the column : (a) simple geometrical form so that a theoretical analysis could be made and all results cxtrapolated to a wider column chosen according to the width of the combine to which it was fitted; (b) suitable for fitting (at least in a modified form) to a combine harvester without the addition of an air cleaning device before the fan to remove the lighter fraction: this implied a positive pressure open-circuit column; (c) to be as near full size as possible but to fit into a limited space (particularly height) in the laboratory and the combine harvester. The apparatus shown in Fig. I consisted of a variable speed centrifugal fan (A) which delivered air through a diffusing tunnel (B) in which three multiple screens (C) were placed. The air stream then entered a contracting section (D) and was turned through 90’ by circular arc vanes (E) placed near the base of the 0.46 m square vertical column. These vanes were designed with a gap/chord ratio of 0.5 which is somewhat greater than the recommended value (Pope and Harper6) due to the need to avoid a build up of material (mainly grain) falling through the air stream. These were adjusted to give the most uniform distribution at the height of injection.

1.E

4340

Fig. 1. Layout qf the experimentnl rig. All dimensions in mm

Crop materials were fed into the column at various injection velocities by passing them between a short conveyor belt (F) and a contra-rotating wooden roller (G). This injection system was in turn fed from a 7 m long conveyor (H) on which the crop materials had been previously placed by hand. The quantity of material and the speed of the belt were chosen to give the required feed rate. Material which passed down through the air stream was sufficiently dispersed not to build up on the turning vanes and was collected through access door (I) at the base of the column after each run was complete. That which was carried upward by the air stream was turned through 140 and was collected in chute (J) the upper section (K) and sides (L) of which were constructed of fine wire mesh to allow the air to exhaust to atmosphere. The free stream air velocity distribution across the column at the height of injection was determined on a 72 point grid by a pitot tube and inclined tube manometer. When measured over

1. G.

FARRAN:

R.

H.

11-7

MACMILLAN

the range of air velocities used in this work, the mean deviation ranged from I .O to I .2”,, of the mean air velocity. The air velocity was set before each run by varying fan speed in accordance with a previous calibration of fan speed versus mean air velocity.

2.2.1. Genertrl The experimental work was conducted in two stages. (i) Initial separation trials were conducted with the object of obtaining a preliminary understanding of the role of air velocity, and speed and direction of injection. Polystyrene balls were chosen to simulate the individual chaff particles because: (a) they have terminal velocities near that for chaff; (b) they are nearly spherical and enable a theoretical analysis to be made; (c) there is no deterioration of the balls with use and variability between runs is small. One disadvantage of the balls is that their particle density is much less than that of chaff. Hence the use of typical mass feed rates would have resulted in excessive numbers of balls in the column, For these initial trials the feed rate was therefore a somewhat arbitrary compromise between a typical mass feed rate and a value giving typical numbers of these lighter particles. (ii) Performance trials were then conducted with the object of extending the earlier work to harvested materials injected at typical and nominally constant feed rate and grain to non-grain weight ratio. Because feed rates were greater than in (i) above appreciably greater injection speeds and air velocities were necessary. As a consequence some grain was carried over with the chaff and straw, hence grain loss as well as rubbish content was required to specify the performance of the column. In accordance with standard testing procedures for combine harvesters’ the results of all esperiments were represented on the basis of the weight of mixture (grain plus balls or rubbish. i e. non-grain material) collected at the base of the column:

balls unseparated (or rubbish content) grain loss

mass of balls (or rubbish) in mixture at base of column = =

total mass of mixture

collected

mass of grain recovered total mass of mixture

collected

at base of column

at top of column

collected

at base of column

loo”,,.

loo”,,

2.2.2.

Tsiu1.y with a mixture of grain and bulls E#>cI of .speed of horizontal injection and of air velocit>. Separation trials were conducted using a mixture of grain and balls injected horizontally at 5 different air speeds and at each of 4 different air velocities. The grain/balls weight ratio and the total malerial feed rate were kept constant at 6.2 and 30.2 kg/min, respectively. iTfli3t oj direction of injection Separation trials were also conducted with a mixture of grain and balls injected into the airstream at angles above and below the horizontal. Grain/ball weight ratio and total material feed rate were 6.2 and 30.2 kg/min, respectively.

A mixture of grain and balls was injected horizontally into the column at different feed rates, An air velocity of 3.02 and injection velocities of 1.27 and 1.40 m/s, which gave near perfect separation at 30.2 kg/min (Fig. 2), were used. The mean grain/ball weight ratio was 5.9.

2.2.3.

Trials wlith grain, chajf’and straw

A typical Australian sieve mixture8 of grain, chaff and straw having a grain to chaff and straw weight ratio of 4.4 to I was fed into the column at a mean total feed rate of 15 I .5 kg/min. Various horizontal injection speeds were used at four different air velocities. Similar tests with grain and chaff only were conducted to remove the influence of straw on the results obtained above. The mean grain to chaff weight ratio and total feed rate were 4.2 and 153.4 kg/min, 2.3.

respectively.

Matrrials

The characteristics of the polystyrene balls as manufactured are shown in Table I. The balls were near spherical in shape and although each size fraction had a different particle density they all had similar terminal velocities. The values were close to the mean value of terminal velocity (1.83 m/s, standard deviation 0.33 m/s) determined previously for chaff.’ Hence the balls were suitable for this simulation and the trajectories calculated for them were based on an assumed sphere of diameter 5.55 mm and mass of 3.5 .’10ds g having a drag coefficient of 0.48 at a terminal velocity of 2.16 m/s. TABLE

I

Characteristics of polystyrene balls Percentage by weight 7.1 38.6 48.9 5.4

Mean diameter, mn,

Mean particle density, kg/n?

2.38 3.97 5.54 6.54

107.7 61.0 39.0 28.2

Calculated drag coejficient

Caludated teminrrl vel0cit.v. m/s

0.62 0.52 0.48 0.46

2.17 2.16 2.22 2.08

All the crop material was obtained from a conventional combine Summit) in Victoria. The characteristics of the grain are as shown in Table II.

harvesting

wheat

(variety

TABLE II

Characteristics of grain

Dimension-mean, mm -standard deviation, Mass-mean, g Moisture content, p; d.b.

The equal The had a which

mm

6.47 0.23 4.15 11.8

3.44 0.17

3.07 0.13

1 10-z

trajectory calculated for grain was based on an assumed spherical particle with diameter to the geometric mean of the above dimensions (4.08 mm) and the same mass.9 mixture of straw and chaff which was as collected from the sieves of the combine harvester moisture content of 12.5 % d.b. and was 86 % chaff by weight. The chaff was that material passed through a 12.5 mm round hole sieve.8. ’o

3. Theoretical prediction In order to provide a theoretical base for the experimental work, determine the trajectory of a single wheat grain and a single polystyrene

a computer program to ball was written. It used

1. G.

FARRAN;

R.

H.

I IO

MACMILLAN

the incremental technique developed by Lapple and Shepherd,” and is illustrated by the block diagram shown in Fig. 2. The calculations were based on the assumption that both particles were spherical, hence there were no lift effects, and the standard drag coefficient-Reynolds number relationship for spheres could be used. As the influence of acceleration on the drag coefficient is small in turbulent conditions, such as those existing in the duct during separation, this effect was neglected. The influence of turbulence on drag coefficient was also neglected owing to the difficulty of determining the intensity of turbulence in the column during the separation process.

Input: Inltlol condOwns Time mcrement AT Portlcle injectIon velocity Particle moss, dlometer Aw veloaty Drag coeffwent ; f (Reynold’s

number)

“$I

(-----4

JI

------+/7=!7+1

J

Compute mstontoneouf relotwe velocity of the portlcle to olr U

Compute

Reynold’s number

J De,termlne

4

Compute

drag

coeffuent

AU

I n=-

+-----I n=l

1st opproxlmotlon u “.?~+KOl “v fJHorllo”lol UH

AwzmgeAU, +Au,_, Averoge

Au,,+Au,,_,

nth approxlmotlon 4”.UH” I

Compute horuontol ond -vertical dlstonce in time

Fig. 2. Block diagram

ofthe computer program

AT

120

GRAIN-C‘HAFt

4. 4.1.1.

Stf’ARAlfON

Results and discussion

4. I. Prrfbrnmtc~r with mistwe EfSect of speed of horizontal injection

of grain and balls

The results from trials with grain and balls at various air velocities and injection speeds are shown in Fig. 3. An analysis of variance showed that at each air velocity there was a significant reduction in the sum of squares when a quadratic relation between 0, balls unseparated and injection speed was tested against a linear relation. It also showed that at each air velocity the “(, unseparated balls differed significantly from that at each other air velocity.

Il,ectlcn i.>CCJ !Ya’s) Fig. 3. Eflect of speed of horizontal

injection on separation of balls and grain at various air velocities: , 3.02 m/s V, 2.72 m/s; 0, 2.85 m/s; ??

, 2.58 m/s;

It was therefore concluded that the linear relation could be rejected in favour of a non-linear one having a minimum and thus there is, at each air velocity, an injection speed for best separation. Further, a near perfect separation can be achieved by appropriate choice of air velocity. In order to determine the conditions which will cause the best separation shown in Fig. 3 the computer program was used to calculate the trajectories of a ball and a grain when injected at the four optimum velocities shown in Table III. These are plotted in Fig. 4. To further illustrate the importance of speed in horizontal injection, trajectories for the two extreme conditions (high air velocity and low injection speed and vice versa) shown in Table Ill are also plotted in Fig. 4. TABLE 111

Data for trajectories shown in Fig. 4 Trajectory

no.

Horizontal

Air velocity, m/s

injection .speed. m/s

Ball

Graiiz

1

I*

2.58

2 3 4

2* 3* 4*

2.73 2.85

I .03 I.14 I .28 \ optimum

3.02

I.40

5 6

5* 6*

3.02 ?.58

--

I

0.81 7.00

extreme

1. G.

FARRAN:

R.

l-l. MACMILLAN 5

I234

6

/ IO

cm

Fig. 4. Ctrlcul~rtedtrajectories qf ball (I to 6) and grain (I* to 6*) for air velocities and horizontol injection .speeds shown in Table Ill

It appears from the trajectories that the best condition for separation requires the lighter material to be injected so as to move up the centre plane of the column. Poor separation occurs when an unfavourable combination of speed of injection and air velocity presents the material to only a part of the air stream. The reasons for this are presumably related to the actual air velocity conditions existing in the upper part of the column when the separation process is taking place. These may differ significantly from the ideal uniform velocity profile assumed in the analysis due to the presence of the boundary layer and the turbulence generated by the stream of grain through the air stream in the lower part of the column. High air velocity and low injection velocity causes little penetration of the column, hence a greater density of grain and balls in the air stream and boundary layer nearest the point of injection. This in turn causes greater turbulence, lower drag forces on the balls, and thus a greater likelihood that they will fall or be carried down with the grain stream. The effective width of the column is thus less than the actual width which is equivalent to having a higher feed rate, a variation which is later shown to cause a similar increase in the unseparated balls. If reducing the penetration causes an effective increase in the feed rate and a poorer performance, an increase in penetration associated with high injection speed and low air velocity might be expected to improve it. This may apply to a semi-infinite velocity field but the presence of the boundary layer and the extra turbulence associated with the rebound of grain from the column wall (evident in Fig. 4 for trajectory 6*) may be the cause of the smaller drag force on the balls and the poorer performance at higher injection speeds.

Fig. 5 shows the process at best separation and the corresponding trajectories. The lower velocity boundary layer and the more turbulent low-drag regions in the column can be seen from the presence of balls suspended in the air stream near the wall opposite the injection point and in the upper left corner of the photo, respectively. While the calculated trajectory for grain agreed well with the experimental results, that for the balls was less accurate particularly in the early stages. That may be expected because of the high concentration of balls and the influence of the grains in the region of the injection point.

Fig. 5. Tyniccrl separation

process

utxler optitnrrm conditions

with calculated

trajectories

marked

4.1.2. Effect of direction of injection The results for injection at angles of 22” upward and 30” downward, as well as for horizontal injection, are shown in Fig. 6. For the air velocity giving complete separation with horizontal injection (3.02 m/s), injection downward at 30” and upward at 22” gave inferior results. Although increasing the air velocity to 3.28 m/s improved the performance at these angles the use of lower air velocities are generally to be preferred in any application in an effort to minimise grain loss.

I.

G.

IARRAN:

R.

H.

MACMILLAN

Inpxtlon

speed (m/s)

I $, 6. iZft&/ (?/speed of injection on separation of balls and grain or wriou.~ air velocities and at t,vo injection directiotrs: 3.28 m/s; 30’ dorcv ?? , 22 up : 3.02 m/s; 30’ donw W, 22 up v. horizontal 0, from Fig. 3 fiw refkrence

In order to provide a basis for interpretation of these results the trajectories for a ball and for a grain were computed for the directions given above. The air velocity of 3.02 m/s was as used in the experimental trials and the injection speeds were those giving best separation at their respective angles. The data are shown in Table IV and the trajectories, together with those for injection at I .O m/s and 60” downwards, are shown in Fig. 7. Perssons’ basis for selecting the best direction of injection was that giving the maximum Applying displacement of the particles in the direction of the air stream for any given penetration. this to the trajectories for single particles in the vertical column suggests at first sight that, of the angles considered, downward injection at 60” is the best and horizontal is the worst. However, one aspect of the actual separation process which this idea1 analysis fails to illustrate is the interaction of the two types of particle within a stream of finite thickness. This interaction, which may be represented by the duration of the flight of the particles in the region of injection. is more likely to be significant where the lighter particles take more time to diverge from the heavier owing to the near reversal of their motion when they are injected against the air stream. A second point revealed by the calculated trajectories is the penetration of particles into the air stream and the resulting space in which separation can take place. The trajectories for horizontal TABI.E

IV

Data for trajectories shown in Fig. 7 Trajectory no. Ball

~_

Grain

Air velocity, m/s 3.02

I 2 3

1* 2* 3*

3.02 3.02

4

4*

3.02

Injection

_..__

Speed, m/s

Direction __-__

I .40 0.90 I .oo I .oo

Horizontal 22 up 30 down 60

down

124

GRAIN-(‘HAI-t

4

Sl:l’ARA

I IO\

23

/ / / 60" down 30’ down \

Hormnto

I/

4 ,/

-,:

220 up

L,V ;__\

AL 1-l

‘1

‘\\\ \\ \

\

\\\

‘\\

\ \

\

\

\

4f

\

\

+

\ \

\\\

’

\ \ :

\

\

’I

:

I*

2*

3’

I

I

IO cm

Fig. 7. Calculated trajectories of ball (1 to 4) andgrain (I * to 4*) for injection speed giving best separation at respective directions shown in Table IV

injection show a greater penetration of the air stream than for other angles and confirm the conclusions drawn in section 4.1.1 that the separation process is most efficient when the particles are most effectively presented to the air stream. 4.1.3.

EfSect of air velocity with horizontal injection

From preliminary trials it was apparent that, even when balls alone were injected, the ideal that they would all rise at air velocities near their terminal value (2.2 m/s) did not occur. Velocities well above this were required to cause all balls to rise and even higher values were required to separate them from grain when it was present in the material being injected. To illustrate this and the significant effect of air velocity shown by the analysis of variance, the data of Fig. 3 were replotted in Fig. 8. The influence of air velocity on separation at two constant horizontal injection speeds may be compared with the somewhat better separation which can be achieved by the choice of optimum injection speed at each air velocity. 4.1.4.

EfSect of feed rate

Fig. 9 shows the effect of different

total feed rates on separation using air velocity of 3.02 m/s and horizontal injection speed of 1.4 m/s (the corresponding optimum values from section 4. I. I). The mean grain-ball weight ratio was 5.9 to 1. The deterioration in performance of the column from the ideal is partly due to the presence of large numbers of particles which increase the turbulence in the air stream and cause a corresponding reduction in the particle drag coefficient. Increasing the material feed rate also increases the total head on the fan and this will cause a reduction in air velocity from the nominal value set before each trial run.

I.

G.

FARRAN:

R.

H.

MACMILLAN

5.0 -

s 4.Q0) = :: p 3.0 G B E

2.0-

I.0 -

I 2.4

I

I

2.5

2.6

2.7

2.9

2.0

Au velocity

3.0

3.1

h/s)

Fig. 8. Effect c?fair velocity on separation of balls and grain with horizontal injection at two speeds: , 1.53 m/s: and at speeds giving best separation at each air velocity (ml. Data as for Fig. 3

, 0.81 m/s:

Total feed rate (kg/min)

Fig. 9. Eflkct of feed rate on separation of balls and grain injected horizontal1.v at: W, I.40 m/s; velocity = 3.02 m/s 4.2.

Performance

, I.27 m/s. Air

lt*ithharvest materials

The work with grain and balls described above confirmed the importance of speed and directlon of injection, as well as of air velocity, and feed rate on effectiveness of separation. Similar tests with harvested materials, which are described below, were restricted to horizontal injection. 4.2.1. Performance with grain, chaff and stran The results in terms of grain loss shown in Fig. IO indicate that at any air velocity (and particularly for the higher values) grain loss varies inversely with injection velocity. Minimum losses of less than 0.2% were obtained with an air velocity of 5.25 m/s which is appreciably less than the terminal velocity of the grain (approximately 8.0 m/s).

126

GRAIN

Injection

Fig. IO. E&t

of horizontal

speed

(‘H/ii-i-

SEPARAl’IOh

(m/s)

injection speed yf’gruin, chaff’and straw on grain loss at various rrir velocities: 0, 6.33 m/s; 7, 5.85 m/s; I-‘, 5.25 717/s

?? , 6.612 m/s;

The corresponding performance of the column in terms of rubbish content of the grain is shown in Fig. II. At the lower air velocities there is no visible trend in the results. However, at the higher velocities the rubbish content reaches a minimum at a particular injection velocity and shows a variation similar to that with grain and balls described above. About half of the rubbish in all samples consisted of straws and threshed heads, the remainder being chaff and other small particles. It is considered that the scatter in the results for the lower air velocities could be attributed to the presence in the non-grain material of those larger particles. The difficulty in separating them arises from their orientation in such a way that the drag force on them is small and so they are able to fall with the grain to the base of the column. It appears from Fig. 10 that the injection speed for minimum grain loss is at least 3.5 m/s. a condition that causes the grain to be projected right across the column. Fig. 11 shows, however, minimum rubbish content is achieved with an injection velocity of about 2.5 m/s. The difficulty of choosing an air velocity to give both a minimum grain loss and a minimum rubbish content is illustrated by the results in Table V taken from Figs IO and I/ for an injection velocity of 3.0 m/s. 4.2.2. Performance with grain and chafl To test the suggestion that the variability in the results reported above for the separation process using grain, chaff and straw arises from the presence of the straw, the performance of the column using only grain and chaff was evaluated. The grain to non-grain ratio and the total feed rate were nominally the same as in section 4.2.1, the actual mean values being 4.2 and 153.4 kg/min, respectively. Only the lower two air velocities of 5.25 m/s and 5.85 m/s were used. Fig. 12 shows the grain loss and, for comparison, the results given previously (Fig. 10) for the grain, straw and chaff tests. Fig. 13 shows the rubbish content corresponding to the highly variable results given in Fig. Il.

I.

G.

FARRAN:

Fig. 11. W’ct

R.

H.

MACMILLAN

of horkontal

injection speed 0, 6.82 m/s;

qf grain, chafl and straw on rubbish Content at varioas air velocities: O,

6.30 m/s: V, 5.85

t77~s:

.

5.25 uis

0.4 -

e~ I 9 c e 0

0.3-

0.2 -

0.1 -

0

I.0

2.0

3.0

1

Injection speed (m/s)

Fig. 12. E/f&t

qf horizontal

injection speed on grain loss at two air velocities. Grain. chaff and straw: -, 5.25 m/s. Grain and chaff: ?? , 5.85 m/s; i‘. 5.25 m/s

. . 5.85 m/s;

128

GRAIN-(‘HAFF

TABLE

SEI’ARA-IIO\

v

Grain loss and rubbish content

Air velocity, m/s

Grain i0s.s. 00 Less than 0.1 1.0

5.25 6.82

Rubbish eontent. 00 I.5 0.2

0

Fig. 13. E#ect of horizontal injection speed of grain and chaJ”on rubbish content at two air velocities: ?? , 4.85 m/s; 1. 5.25 m/s

It will be seen that the reduction in the grain loss and in the variability of the rubbish content results is presumably due to the absence of the straws and their tendency to orientate themselves in the air stream or to form clumps with chaff and grain and to fall down through or to rise with the air stream.

5.

Conclusions

The effectiveness of the column is a complex function of the air velocity, of the injection speed and direction, of the feed rate and of the nature and composition of the mixture being separated. The theoretical and experimental studies using particles simulating harvested materials suggest that for best separation, there was a preferred region in the column into which materials should be injected. Appropriate values for both magnitude and direction of injection speed and for air velocity, chosen to place materials in that region, gave best performance. The performance of the column in separating actual harvested materials differed from that with a mixture simulating harvested materials and required consideration of both grain loss and rubbish (non-grain) content. The choice of air velocity was critical in attempting to reconcile the conflicting requirements for both minimum grain loss and minimum rubbish content. Appreciable improvements in both measures were achieved when the straw material was removed before separation. The results of this work extend previous studies by offering some understanding of the effect of injection speed and direction on separation at practical feed rates. It also provides a basis for further study of more complex arrangements.

I. G.

FARRAN;

R. H.

129

MACMILLAN

Acknowledgements The authors wish to acknowledge the help of Mr G. R. McIlroy with the experimental work, the advice of Mr G. J. Davies (Agricultural Research Council, Unit of Statistics, University of Edinburgh) in the statistical analysis of the results and the Victorian Wheat Industry Research Committee in the provision of funds. REFERENCES ’

Shellard, J. H.; Macmillan, R. H. Aerodynamic properties qf’rhreshed wheat muteriuls. J. agric. Engng

Res., 1978 23 273-281 2 Persson, S. The most fuvourable combination sf ,filctors ,fbr wind separation-A mathematical studs,. Landtech. Forsch., 1967 17 (2) 53-57 3 Lee, J. H. A.; MacAulay, J. T. Grain separation on oscillating combine sieves us ajfkcted bv mrrreritrl entrance conditions. Trans. Am. Sot. agric. Engrs, 1969 12 (5) 648-651, 654 4 Rumble, D. W.; Lee, J. H. A. Aerodynamic separation in N combine shoe. Trans. Am. Sot. agric. Engrs, 1970 13 (1) 6-8 5 Kasbayap, M. M.; Pandya, A. C. Air velocity requirements fbr winnowing operations. J. agric. Engng Res., 1966 11 (1) 24-32 6 Pope, A.; Harper, J. J. Low Speed Wind Tunnel Testing. New York: Wiley, 1966 ’ O.E.C.D. Standard Testing Procedure jbr Combine Harvesters. AGR/T(67)12, Paris, October 1967 8 Brown, W. T.; Vasey, G. H. Wheat Harvester Suryv. Dept. of Agricultural Engineering, University of Melbourne, 1967 9 Mohsenin, N. N. Phj’sical Properties ofPlant and Animal Muteriuls. New York: Gordon and Breach, 1970 lo Uhl, J. B.; Lamp, B. J. Pneumatic separation ofgrain andstrcrw mixtures. Trans. Am. Sot. agric. Engrs, 19669(2)244-246 ” Lapple, C. E.; Shepherd, C. B. C’ulculution qf’particle trujectorks. Ind. Engng Chem., 1940 32 ( 5) 605-6 I 7