See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/44448746

Prediction of breakage during roller milling of mixtures of wheat

kernels, based on single kernel measurements

Article in Jurnal Teknologi · June 2008

DOI: 10.11113/jt.v48.235 · Source: OAI

CITATIONS READS

2 179

3 authors, including:

Ida Idayu Muhamad Grant M. Campbell

Universiti Teknologi Malaysia University of Huddersfield
239 PUBLICATIONS 1,436 CITATIONS 124 PUBLICATIONS 2,515 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Setting up chemical engineering programmes - new paper published in Education for Chemical Engineers View project

functional food and drug delivery system View project

All content following this page was uploaded by Ida Idayu Muhamad on 28 May 2014.

The user has requested enhancement of the downloaded file.

Jurnal Teknologi, 48(F) Jun 2008: 75-83
© Universiti Teknologi Malaysia

PREDICTION OF BREAKAGE DURING ROLLER MILLING OF

MIXTURES OF WHEAT KERNELS, BASED ON SINGLE KERNEL
MEASUREMENTS

IDA IDAYU MUHAMAD1*, CHAOYING FANG2 & GRANT M.

CAMPBELL3

Abstract Wheat flour milling involves repeated breakage (by roller milling) and separation
(by sifting) of flour particles to give efficient recovery of fine flour relatively free from bran
contamination. A starting point for modelling the behaviour of mixtures of wheat kernels is to be
able to measure the variation in kernel properties within the mixture. The Perten Single Kernel
Characterisation System (SKCS) gives the distributions of kernel mass, moisture content, diameter
and hardness within a mixture, from 300 kernels within 5 minutes. A challenge remains to relate
these measured distributions to predictions of milling performance. To this end, a breakage
function in terms of these measured parameters for individual kernels has been constructed,
and integrated over the distribution of kernel properties using the breakage equation. These
models allow prediction of the output particle size distribution delivered by First Break roller
milling of kernels varying in size, hardness and variety, based entirely on SKCS characteristics.
Predictions have been developed for both sharp-to-sharp and dull-to-dull roll dispositions, and
show encouraging agreement with independent data.

Keywords: Milling behaviour, particle size distribution, breakage, single kernel testing, wheat

Abstrak Pengilangan tepung gandum melibatkan pengisaran bijirin secara berulang-kali

(menggunakan roller mill) dan pemisahan (secara pengayakan) bagi perolehan semula hasilan
tepung yang berkesan yang secara bandingannya bersih dari pencemaran bran. Satu permulaan
bagi membina model perlakuan terhadap campuran bijirin gandum ialah keupayaan bagi
menentu-ukur kelainan ciri-ciri yang wujud dalam campuran tersebut. Alatan Perten Single Kernel
Characterisation System (SKCS) dapat memberikan taburan kuantitatif bagi jisim, kandungan
lembapan, garis pusat dan nilai kekerasan bagi campuran bijirin tersebut, berdasarkan
pengukuran 300 kernel dalam masa 5 minit. Cabarannya ialah menghubungkaitkan taburan
yang telah ditentukur kepada ramalan keupayaan pengilangan. Pada tahap ini, satu bentuk
kefungsian pengisaran (breakage function) dalam konteks parameter tentu-ukuran bagi setiap kernel
bijirin telah diterbitkan, dan diintegrasi kepada taburan sifat kernel menggunakan persamaan
proses pengisaran (breakage equation). Dengan menggunakan model tersebut, satu jangkaan boleh
dilakukan ke atas taburan saiz partikel hasil keluaran pengisaran siri pertama atau First Break roller
milling bagi kernel bijirin yang berbeza saiz, kekerasan dan kelainannya, berdasarkan hanya pada
ciri-ciri taburan yang diberikan oleh SKCS. Model tersebut telah diuji ke atas hasil keluaran

1
Faculty of Chemical & Natural Resources Engineering, Universiti Teknologi Malaysia, 81300
UTM Skudai, Johor, Malaysia
2&3
Satake Centre for Grain Process Engineering, School of Chemical Engineering and Analytical
Science, The University of Manchester, UK
* Corresponding author: Tel +607-5535541, Fax. +607-5581463, Email: [email protected]
76 IDA IDAYU, FANG & CAMPBELL
pengisaran pada disposisi (roll disposition) pengisaran sharp-to-sharp dan dull-to-dull, dan didapati
menunjukkan keseragaman dengan data bebas (independent data).

Kata kunci: Tingkah laku pengisaran, taburan saiz partikel, pemecahan, pengujian satu-kernel,
bijirin

1.0 INTRODUCTION
Wheat kernels are initially broken into a wide range of particle sizes during First
Break roller milling. These particles are separated by size using plansifting into several
fractions, which are sent to different parts of the mill for repeated grinding and
sifting operations to give efficient recovery of floury endosperm material, relatively
free of bran contamination. The wheat entering the First Break rollers contains
kernels varying in variety and geographical origin, and hence in size, hardness,
protein content, etc. Even within a single variety grown in the same field, or within
a single spike from the same wheat plant, wheat kernels vary in size and other
physico-chemical characteristics [1, 2] and in functionality [3, 4]. Millers therefore
need to be able to predict reliably of a heterogeneous mixture of wheat kernels
with widely varying properties that will mill. The particle size distribution from First
Break determines mills flow and balances throughout the rest of the mill; keeping
the particle size distribution at this point constant, in the face of a varying feedstock
of wheat, would greatly simplify the rest of the milling process.
The first step towards achieving this understanding of the behaviour of wheat
mixtures is to be able to measure the variation in kernel characteristics within a
batch of wheat. To this end, single kernel testing has become popular in recent
years, culminating thus far in the introduction of the Perten Single Kernel
Characterization System (SKCS), developed by the US Department of Agriculture
[4-8]. The SKCS separates individual kernels from a sample, weights each kernel
and then crushes each between a rotor and a crescent, while measuring the force-
deformation profile during crushing, along with the electrical conductivity of the
grain. From the fundamental measurement, the SKCS calculates the mass, moisture
content, diameter and hardness of each kernel, reports the distributions of these four
parameters, along with the mean and standard deviation. The SKCS is thus useful
to millers as a quality control instrument at wheat intake. However, the information
it delivers is potentially much more useful than merely indicating acceptance or
rejection of wheat deliveries [9, 10]. The difficulty is in interpreting distributions
of size, hardness, etc. in terms of milling behaviour. Milling experience is primarily
with bulk average measurements, but conceivably two wheat mixtures (grists) could
have, for example, different kernel shapes but the same average hardness, however
extremely different hardness profiles and hence milling behaviour [11].
The objective of the current work is to establish that the particle size distribution
 PREDICTION OF BREAKAGE DURING ROLLER MILLING OF MIXTURES OF WHEAT 77

resulting from First Break milling of mixtures of wheat kernels could be predicted
directly from SKCS characteristics. This work is part of a larger programme aimed
at modelling and optimising the flour milling process.

2.0 THEORY
Previous work has introduced the breakage equation for roller milling and established
the form of the breakage function in terms of kernel size, roll gap, moisture
content and roll disposition [12-16]. The breakage equation is based on the fact
that individual wheat kernels break independently during First Break milling, each
according to its own physico-chemical characteristics. In the current work, kernel
hardness was introduced into the breakage function. For reasons of practicality,
the effect of moisture was excluded from the current work and all wheat samples
conditioned to 16% moisture; in practice millers do not condition wheat’s to widely
varying moisture contents. The breakage equation in terms of kernel size, D, and
hardness, H, is then
H=∞ D=∞
P(x) = ∫ ∫ B(x, D, H) ρD (D) ρH (H)dDdH (1)
H=0 D=0

where x is particle size in the outlet stream from First Break milling. P2(x) is the
proportion of material in the outlet stream that is smaller than x, B(x, D, H) is the
breakage function, and ρD (D) and ρH (H) are the distributions of kernel size and
hardness in the feed, respectively. ρD (D) and ρH (H) are measured by the SKCS;
knowing the breakage function, the cumulative particle size distribution of the outlet
from First Break can be predicted from the SKCS data by numerical integration of
Equation (1).

3.0 MATERIALS AND METHODS

Table 1 shows the average physical properties for nineteen samples of wheat used in
this study as measured by the Single Kernel Characterization System (Model 4100,
Perten Instruments AB, Sweden) shown in Figure 1.
Before milling, 100 grams of each dry wheat sample was conditioned overnight
to reach 16% (optimum) moisture content, wet basis; this toughens the bran and
germ and softens the endosperm, making the separation of endosperm from germ
and bran easier. Then, the moist wheat was allowed to temper for at least 16 hours
to give an even distribution of moisture.
78 IDA IDAYU, FANG & CAMPBELL

Table 1 The SKCS properties of nineteen varieties of wheat used in the study

Wheat Weight Diameter Hardness Moisture

variety (mg) (mm) index content (%)
Consort 48.35 3.01 11.2 15.80
Claire 40.80 2.66 24.6 16.20
Riband 51.36 3.11 27.6 15.87
Drake 46.36 2.83 29.2 15.72
Crofter 51.06 2.98 42 15.89
Soissons 44.12 2.91 52.2 15.94
Raleigh 48.57 2.99 58.5 15.65
Charger 45.90 2.89 59.3 15.66
Abbot 45.32 2.95 61.1 15.65
Buster 52.01 3.33 62.7 15.65
Avalon 53.51 3.05 62.8 15.86
Malacca6 44.02 2.93 63.2 16.17
Hereward 43.54 2.92 65.3 15.56
Rialto 50.58 3.08 65.7 15.82
Brigadier 51.03 3.02 67.1 15.51
CWRS 31.11 2.39 71.5 15.91
Mercia 44.28 2.90 73.6 15.54
Cadenza 56.32 3.35 76.9 15.80
Spanish 45.19 2.81 80.4 16.72

Conditioned samples were milled on the Satake STR-100 test roller mill (Satake
Corporation, Japan) at roll gaps 0.3 – 0.8 mm under both sharp-to-sharp (S-S) and
dull-to-dull (D-D) roll dispositions. During milling, the breakage of wheat kernels
during first break roller milling depends on many factors, including the disposition
of the fluted rolls. Dull-to-dull gives slippage between the flutes and the kernel, and
thus imparts less shear strain than sharp-to-sharp (see Figures 2 and 3).
The resulting particle size distributions were measured on a Simon plansifter
using 200 mm diameter wire mesh sieves of aperture size: 2000, 1700, 1400, 1180,
850, 500 and 212 um, along with a bottom pan. From these, the breakage functions
incorporating kernel hardness, diameter and roll gap were constructed for each
milling disposition. A further four varieties, plus two 50:50 mixtures of varieties,
were milled to obtain independent data with which to validate the predictions.
SKCS characteristics for all 23 wheat varieties were measured using the SKCS 4100
(Perten Instruments AB, Sweden).
 PREDICTION OF BREAKAGE DURING ROLLER MILLING OF MIXTURES OF WHEAT 79

Figure 1 Single Kernel Figure 2 Satake STR-100AU test

Characterization System 4100 roller mill
(Perten Instruments, 1995)

Sharp-to-sharp Dull-to-dull

Figure 3 Roller mill at disposition sharp-to-sharp and dull to dull

4.0 RESULTS AND DISCUSSION

Figures 4 - 6 show the percentage smaller than x for various values of outlet particle
size, at a roll gap of 0.3 mm, 0.6 mm and 0.8 mm, for both D-D and S-S roll
dispositions, for the 19 wheat varieties. The fitted breakage function is also shown.
Clearly, as hardness increases, the proportion of flour stocks smaller than 212 µm
decreases under both dispositions. The proportion of material larger than 2000
µm also decreases, while the proportion of mid-range particles increases with
hardness. The effect of hardness is greater under D-D milling, which produces a
larger proportion of very large and very small particles, with fewer in the mid-size
range, compared with S-S. These results agree with previous findings [14]. Similar
results were obtained at other roll gaps, with larger roll gaps producing larger output
particle sizes. The results from different wheat varieties show consistent trends,
indicating that the SKCS hardness measurement is meaningful in terms of actual
breakage during roller milling, although some samples depart significantly from the
general trend.
80 IDA IDAYU, FANG & CAMPBELL

100 100
2000 2000
90 90
than xsmaller than x

than xsmaller than x

80 1700 80 1700

70 70
1400 1400
60
100 60
100
1180
2000 2000
1180
50
90 50
90
Percentage

Percentage
40
80 850
1700 40
80 1700
850
30
70 30
70
Percentage smaller

500
1400 1400

Percentage smaller
500
20
60 20
60
212
1180 1180
10
50 10
50 212

0
40 850 0
40 850
30 0 20 40 60 80 100 0 20 40 60 80 100
30
500 500
20 Hardness 20 Hardness
212
10 10 212
(a) (b)
0 0
Figure
0 4 Percentage
20 40 smaller
60 80 than x versus hardness under
100 0 20(a) dull-to-dull;
40 60 and
80 (b)100
sharp-to-
100
sharp roll dispositions at roll gap 0.3 mm
Hardness 100 Hardness
than xsmaller than x

90 2000 2000
90
than xsmaller than x

80
1700 80 1700
70
70
1400 1400
60
100 60
100
50
90 1180
2000 50 1180
2000
Percentage

90
40
80
Percentage

850 40
80
1700 850
1700
30
Percentage smaller

70 30
70
500
Percentage smaller

20 1400 500
1400
60 20
60
10
50 212
1180 10 212
50 1180
0
40 0
850 40
850
30 0 20 40 60 80 100
30 0 20 40 60 80 100
Hardness 500 Hardness 500
20 20
10 212 10 212
0 100 0
100
than xsmaller than x

0 20 40 60 80 100 90 0 20 40 60 80 100
than xsmaller than x

90 2000
Hardness 2000 80 Hardness
80 1700
70 (a) 1700 70 (b)
60 1400
60 1400 100
100
Figure 5 Percentage smaller than x versus hardness
50
90
under (a) dull-to-dull; and (b) sharp-to-
1180
Percentage

50 1180 2000
90 sharp roll dispositions at roll gap 0.640
mm
Percentage

2000 80
40 850
80 850 1700
Percentage smaller

1700 30
70
30
Percentage smaller

70 500
500 20 1400
20 1400 60
60 212
10 212 10
50 1180
50 1180
0 0
40
40 850
850
0 20 40 60 80 100 30 0 20 40 60 80 100
30
500
Hardness 500 20 Hardness
20
212
10 212 10

0 0

0 20 40 60 80 100 0 20 40 60 80 100

Hardness Hardness
 30 30

Perce

Percent
500 500
20 20
10 212 10 212
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Hardness
PREDICTION OF Hardness
BREAKAGE DURING ROLLER MILLING OF MIXTURES OF WHEAT 81

100
100

Percentage smaller than x

90
Percentage smaller than x

90 2000
2000 80
80 1700
1700 70
70
60 1400
60 1400
50 1180
50 1180
40
40 850
850
30
30
500
500 20
20
212
10 212 10

0 0

0 20 40 60 80 100 0 20 40 60 80 100

Hardness Hardness

(a) (b)

Figure 6 Percentage smaller than x versus hardness under (a) dull-to-dull; and (b) sharp-to-
sharp roll dispositions at roll gap 0.8 mm

Figure 7 shows the predicted and experimental cumulative particle size distributions
for Consort (a soft wheat, SKCS hardness = 23.7) and Spark (a hard wheat, SKCS
hardness = 74.4) at three roll gaps under D-D milling. The agreement between the
predictions and the independently-derived experimental data is excellent. Similar
agreements were obtained for a 50:50 mixture of these two wheats.
A similarly good result (result not shown) was obtained for Malacca (SKCS
hardness = 61.5) under both the S-S and D-D milling, but Soissons (SKCS hardness
= 54.4) and Malacca50/Soissons50 did not give quite such good agreement.
Clearly, the inclusion of hardness in the breakage function has gone a long way
towards making the breakage function more universal; however, there are still some
wheat varieties for which good predictions are not achieved. This may be due to
additional factors such as kernel shape. Work is ongoing to include the other SKCS
data to produce a more comprehensive breakage function, and to identify whether
other single kernel parameters such as density are needed to allow truly universal
predictions.
82 IDA IDAYU, FANG & CAMPBELL

100 100

90 90
Percentage smaller than x

Percentage smaller than x

80 80

70 70

60 60

50 50

40 40

30 30
0.3 mm 0.3 mm
20 20
0.5 mm 0.5 mm
10 10
0.7 mm 0.7 mm
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
Particle size x (µm) Particle size x (µm)

(a) (b)

100

90
Percentage smaller than x

80

70

60

50

40

30
0.3 mm
20
0.5 mm
10 0.7 mm
0
0 500 1000 1500 2000
Particle size x (µm)

Figure 7 Predicted (lines) and experimental(c)

(symbols) cumulative particle size distributions
for (a) Consort; (b) Spark and (c) Consort50/Spark50 wheats milled at different
roll gaps under a D-D disposition

5.0 CONCLUSIONS
Prediction of the particle size distribution resulting from First Break roller milling of
unknown wheat samples or mixtures under S-S or D-D milling has been demonstrated,
based solely on integration of SKCS size and hardness distribution data using the
breakage equation for roller milling. Additional single kernel information such as
mass, density, kernel shape and moisture content, and their effects on the particle
 PREDICTION OF BREAKAGE DURING ROLLER MILLING OF MIXTURES OF WHEAT 83

size distribution from First Break milling of wheat are needed to be incorporated to
allow universally applicable breakage functions to be established.

ACKNOWLEDGEMENTS
This work was funded by the Engineering and Physical Sciences Research Council
(EPSRC, grant no. GR/M49939). The authors are grateful to Universiti Teknologi
Malaysia and the Satake Corporation of Japan for supporting the research.

REFERENCES
[1] Evers, A. D. 1996. New Opportunities in Wheat Grading. The Nordic Cereal Industry Towards Year
2000. Proceedings of 26th Nordic Cereal Congress. 61-67.
[2] Doehlert, D. C., M. S. McMullen, and N. R. Riveland. 2002. Sources of Variation in Oat Kernel Size.
Cereal Chemistry. 79: 528-534.
[3] Sutton, K. H., R. L. Hay, and C. H. Mouat. 1992. The Effect of Kernel Weight on the Assessment of
Baking Performance of Wheat’s by RP-HPLC of Glutenin Subunits from Single Grains. J. Cereal Science.
15: 253-265.
[4] Ohm, J. B., O. K. Chung, and C. W. Deyoe. 1998. Single-Kernel Characteristics of Hard Winter Wheats
in Relation to Milling and Baking Quality. Cereal Chemistry. 75(1): 156-161.
[5] Martin, C. R., R. Rousser, and D. L. Brabec. 1993. Development of a Single-kernel Wheat
Characterization System. Trans. ASAE. 36: 1399-1404.
[6] Sissons, M. J., B. G. Osborne, R. A. Hare, S. A. Sissons, and R. Jackson. 2000. Application of the Single-
kernel Characterization System to Durum Wheat Testing and Quality Prediction. Cereal Chemistry. 77(1):
4-10.
[7] Osborne, B. G. 2001. Wheat Flour Milling, Part 2. In Cereals and Cereal Products: Chemistry and Technology.
D. A. V. Dendy and B. J. Dobraszczyk (Eds.). Maryland, USA: Aspen Publishers Inc. 172-181.
[8] Osborne, B. G., R. Jackson, and S. R. Delwiche. 2001. Rapid Prediction of Wheat Endosperm Compressive
Strength Properties Using the Single-Kernel Characterisation System. Cereal Chemistry. 78(2): 142-143.
[9] Muhamad, I. I. and G. M. Campbell. 2004. Effects of Kernel Hardness and Moisture Content on Wheat
Breakage in the Single Kernel Characterisation System. Innovative Food Sci Emerging Technol. 5: 119-125.
[10] Muhamad, I. I., C. Y. �������������������������������������������������������������������������������
Fang, and G. M. Campbell. 2006. Comparison of Grain Particle Size Distribution
in the Single Kernel Characterisation System and During First Break Roller Milling. ������������������
Jurnal Teknologi.
44(A): 41-52.
[11] Campbell, G. M. C. Y. Fang, and I. I. Muhamad. 2007. On Predicting Roller Milling Performance VI.
Effect of Kernel Hardness and Shape on the Particle Size Distribution from First Break Milling of Wheat.
IChemE Journal: Food and Bioproduct. 85(1): 7-23.
[12] Campbell, G. M. and C. Webb. 2001. On Predicting Roller Milling Performance, Part I: The Breakage
Equation. Powder Technology. 115(3): 234-242.
[13] Campbell, G. M., P. J. Bunn, C. Webb, and S. C. W. Hook. 2001. On Predicting Roller Milling
Performance. Part II. The Breakage Function. Powder Technology. 115(3): 243-255.
[14] Fang, C. Y. and G. M. Campbell. 2002. Effect of Roll Fluting Disposition and Roll Gap on the Breakage
of Wheat Kernels During First Break Roller Milling. Cereal Chemistry. 79(4): 518-522.
[15] Fang, C. Y. and G. M. Campbell. 2003a. On Predicting Roller Milling Performance. Part IV. Effect of Roll
Disposition on the Particle Size Distribution from First Break Milling of Wheat. J. Cereal Sci. 37: 21-29.
[16] Fang, C. Y. and G. M. Campbell. 2003b. On Predicting Roller Milling Performance. Part V. Effect of
Moisture Content on the Particle Size Distribution from First Break Milling of Wheat. J. Cereal Sci. 37:
31-41.

View publication stats