Low Moisture Extrusion: Effects of Cooking

Moisture on Product Characteristics

ROBERT C. MILLER

ABSTRACT per, 1979; Ziminiski, 1981). The overall economic impact

A commercial Creusot-Loire BC-105 twin-screw expanded dry pet of low moisture extrusion is controversial, however (ANON.,
food extruder was operated to provide extruded moistures between 1983), and must be evaluated on a case by case basis
7.9% and 12.4% at constant product density. Low moisture extru- (Harper and Holay, 1979).
sion decreased product uniformity 45% and product shape round- The present investigation was accomplished with a twin-
ness by 70%. It increased air ceil size 41% and internal fissuring by screw low moisture extrusion process for expanded dry dog
25%. After drying all samples to low moisture, the samples extruded food. Dry pet food products consist primarily of farina-
at a low initial moisture were more fragile with 15% lower ultimate ceous ingredients, which provide sufficient starch, that
strength as measured by crushing with an Instron testing machine. upon gelatinization by the cooking extruder, becomes the
matrix of the product (Mason et al., 1982).
In the traditional dry pet food process, the product
INTRODUCTION moisture after extrusion is about 22%, which is reduced to
MANY PUFFED PRODUCTS such as cereals, snacks and a final moisture of about 10% by drying. The low moisture
dry pet foods are manufactured by direct-expansion cook- twin-screw process provides this final moisture at the ex-
ing extrusion. The porous structure of these products is truder discharge, with no drying required. This has not
obtained by creating a colloidal gel at high temperature and resulted in a substantial energy cost saving (Miller, 1984),
pressure within the extruder. When the product emerges but has provided a simple, compact and flexible plant
through the extrusion dies, the sudden drop in pressure operation. Product quality, and the associated cost of prod-
permits the superheated water to vaporize. If the gel is of uct loss must also be considered along with plant operating
the proper strength, it inflates with many small cells caused costs. In the low moisture process, higher temperatures
by the expansion of the water vapor (Williams et al., 1977). and pressures are required for comparable gelatinization
Water obviously plays a critical role in the expansion and expansion (Lawton et al., 1972).
process. It has been found to strongly affect the degree of Our objective is to determine the effects of the low
gelatinization as a reactant in complex interactions with moisture cooking environment on the physical characteris-
other components (Holay and Harper, 1982; Mercier and tics of the extruded product. These characteristics include
Feillet, 1975). Extrusion thermodynamics is also affected product irregularities caused by lateral movement of the
by water in its role as a heat sink and lubricant in high shear extruder screws under load, a problem peculiar to twin-
heat generation (Lawton et al., 1972). These roles often screw extruders (Miller, 1984).
conflict, making generalizations about the effect of water
content on expansion rate difficult. Empirical studies have MATERIALS& METHODS
shown, however, that water content and extrusion tempera- THE EXTRUDER used in this study is a commercial Creusot-Loire
ture are the most important variables influencing expansion BC-105 twin-screw machine (CL-Simon, Inc., Kansas City, KS)
in any particular system (Lawton et al., 1972; Hauck, 198 1; manufacturing a range of dry pet food products. Samples and data
Holay and Harper, 1982; Merceir and Feillet, 1975; Park, were collected during a normal production run of a regular “chunk
1976). style” dry expanded dog food. The proprietary formula for this
Water content has been found to affect the cellular product consists mainly of corn, meat and bone meal, soy bean
structure (Harper, 1981) and mechanical properties (Mer- meal, wheat, and wheat middlings blended to provide a finished
tier and Feillet, 1975) of extruded products in addition to product with a protein content of at least 21%. A total fat content
their overall densities. It has also been found that expansion of about 3% results in the extrudate. The entire run was made from
the same batch of feed mix.
is not always uniform in all directions. Longitudinal and To generate sufficient shear within the extruder for proper
transverse expansion can be selectively influenced by extru- mixing and gelatinization, reverse pitch screws were installed in the
sion moisture (Park, 1976). cooking zone (Miller, 1984). Four die orifices were placed sym-
High shear cooking extruders are commonly used for metrically in the circular die plate. As in most manufacturing situa-
making expanded products. In addition to developing pres- tions, the screws and barrel were somewhat worn at the time of the
sure, they provide the mixing and heat generation necessary experiment.
to properly gelatinize and condition the feed material for The dry feed rate to the extruder was kept constant at 2.5 T/hr
expansion (Rossen and Miller, 1973). In recent years, the at 7.2% moisture. Water was injected into the mixing section at OS-
single-screw extruders traditionally used for this purpose 1.0 gpm to vary the extrusion moisture. To compensate for the
have been supplanted to a degree by the newer twin-screw varying moisture on the expansion rate, the barrel temperatures
were adjusted, resulting in a constant bulk density of 22 lb/ft3. No
machines, particularly of the fully-intermeshing co-rotating other changes in operating conditions or feed rate were made.
type. The twin-screw extruders can accept a wider range of Product moisture and density are controlled by water feed rate and
dough rheology and are usually run at a lower moisture barrel temperature (Miller, 1984). Since two dependent variables are
than that for single-screw extrusion. under the control of two independent variables, there are no remain-
Low moisture extrusion requires 1eSsdrying of the prod- ing degressof freedom. By maintaining a constant density, all of the
uct, which results in a savings in energy consumption (Har- conditions become dependent on the single variable of extrusion
moisture.
The first sample was taken under normal operating conditions.
The water rate was then reduced to the minimum conducive to
Author Miller is a Consulting Engineer, RD 2, Box 4 13, Auburn, NY stable operation, and then raised to the maximum. These changes
13021. were made incrementally to minimize disturbances in the operation.

Volume 50 (1985)-JOURNAL OF FOOD SCIENCE-249

LOW MOISTURE EXTRUSION. ..
At each incremental change, about ten minutes were required to two distinctly different populations of equal numbers of
reach equilibrium, judged by watching the four zone temperature different sizes. The photograph in Fig. 8B illustrates the
indicators. Using the above iterative procedure, about 1 hr was re- range of differences found. To quantify the difference, the
quired for each major change in moisture level at constant density. “die flow differential” was defined as the ratio of the
Extrudate moistures were measured with an Apollo microwave
average piece weight difference to the smaller piece weight:
moisture balance.
The normal and high moisture samples were divided into several
subsamples which were dried in the microwave oven to reduce the DFD = (W, - WJW,
moistures for texture studies. The resulting six samples provided a
range of final moisture independent of extruded product moisture. where: DFD = die flow differential, dimensionless; W 1 =
Each sample was also classified according to size. The extruder often average weight of larger pieces; W, = average weight of
produces two distinctly different product sizes simultaneously, smaller pieces.
depending on the degree of screw and barrel wear (Miller, 1984). As shown in Fig. 1, the DFD increased from 0.12 to
These were segregated by visual inspection into two equal popula- 0.57 as the extrusion moisture was reduced from 12.4 to
tions. The pieces were then weighed for comparison and correlation 7.9%. Since, with die face cutting, the number of pieces
with extrusion moisture.
Overall product morphology, illustrated in the three view photo- from each die orifice is the same in any time period, this
graph (Fig. 8A), was determined by measuring several dimensions of difference reflects the difference in actual flow rates from
the extruded pieces. Internal product structure was evaluated by the dies. At the low moisture, therefore, half of the dies
50X microscopic examination of axial cross sections. The number were conveying nearly 60% more product than the other
and sizes of cells in the calibrated field were estimated. The cumula- half. At the high moisture, the flows became nearly equal.
tive void volume (proportional to the cell diameter cubed), ex-
pressed as percent of total void volume, was plotted against cell Productmorphology
diameter on log-probability paper. This allowed linear correlation of
the cell size distribution. The median cell size was the point at In an expanded die face cut product, the pieces undergo
which the resulting graph crossed the 50% line. Another measure of plastic and elastic deformation during and shortly after
internal structure was made by counting the number of large voids emergence from the die. The plastic deformation is a result
and fissures in longitudinal cross sections. of the velocity profile in the orifice, which has a maximum
An Instron testing machine was used to determine the mechani- in the center of flow. This leads to a cut product which is
cal properties of the product. Ten individual pieces from each sam- longer along its central axis than along its sides. (In highly
ple were crushed between flat plates at a cross-head speed of 1 cm/ expanded products such as cereals and snacks, all product
min, chart speed of 50 cm/mm and full scale calibration of 500N.
surfaces are greatly distorted by the balloon-like inflation.
RESULTS& DISCUSSION Pet foods are less expanded and retain much of the con-
figuration of the die. We therefore have a better oppor-
Extruder operation tunity to observe the flow properties of the extrudate from
the distortions which do occur and are not overwhelmed
The moisture range within the extruder was calculated to by great expansion.) To quantify the degree of plastic
be 11.4-15.3%. The extruded product moisture was mea- deformation, the ratio of the difference between edge and
sured in the 7.9-12.4% range, indicating that about 3% central lengths to the edge length were calculated (see Fig.
moisture was flashed off upon emergence from the die and 2).
expansion. To compensate for the moisture changes, it was
necessaq to vary the discharge barrel temperature from P = (B - C)/C
220-260 C. Stable operation was observed over the full
range of moistures with no evidence of surging or other where: P = plastic deformation, dimensionless; B = piece
instabilities. length along central axis; C = piece length along edge.
As shown in Fig. 3, the plastic deformation was found to
Die flow differential increase by 228% with increasing moisture. This indicates
As found in the past when working with worn screws that at low moisture, the stiffer extrudate resists developing
and barrel, the product was not uniform, but consisted of a laminar flow velocity profile in the die, and possibly

-3
y-c-

- Flow

OL ’ I I
8 10 12
Product Moisture, %
Fig. l-Asymmetrical die flow: die flow Fig. P-Product morphology: critical dimen- Fig. 3-Plastic deformation vs. extruded
differential lDFD1 vs. extruded product Cons of die face cut extruded pieces. product moisture, %.
moisture, %.

250-JOURNAL OF FOOD SCIENCE-Volume 50 (1985)

breaks free from the die surfaces before emerging. The modulus of elasticity was also calculated from the linear
resulting product is less rounded than that produced at initial stress-strain data.
higher moisture, having relatively flat leading and trailing The modulus of elasticity was found to decrease with
edges. final product moisture, meaning that the dry samples were
After emerging from the die, the shear stresses imposed stiffer. The stiffness was not affected by extrusion mois-
by the die flow velocity profile are relieved. In an elastic ture, only final moisture, so that all data fit on the same
product, the sudden release of internal stresses allows the curve regardless of how the final moisture was reached, by
cut piece to rebound in shape. For highly elastic materials, drying or directly at the extruder. This is an indication that
the flat cut trailing edge forms a bulge, sometimes matching throughout the range of extrusion moistures, the degree of
the leading edge bulge. To quantify the elastic rebound, the gelatinization, the product composition and other chemical
ratio of the trailing edge bulge to the leading edge bulge was and physical properties did not vary significantly.
calculated for each sample by (see Fig. 2): In contrast to the modulus of elasticity, the yield stress
and ultimate strength were affected by both the extrusion
R = (B - A - C)/A moisture and the final moisure. For each extrusion mois-
ture level, the yield stress decreased with increasing final
where: R = elastic rebound, dimensionless; A = leading edge moisture. The products extruded at a low moisture, how-
bulge thickness. ever, were found to yield at a higher stress than those
By this definition, a completely elastic product with extruded at a high moisture and later dried to the same low
equal leading edge (A) and trailing edge (B - A - C) bulges moisture. The ultimate strength increased with final prod-
would have an elastic rebound value (R) of + 1. In our tests, uct moisture at each extrusion moisture level, and also
the trailing edges were all concave, or had negative trailing increased with extrustion moisture.
edge bulges. This led to negative values for R, and indicated Therefore, products extruded at low moisture yield at a
that the elasticity of the products was not sufficient to higher stress, but fracture at a lower stress than products
overcome the residual momentum distribution from the extruded at a higher moisture and then dried to the same
velocity profiles. As the moisture was increased, however, final moisture. This is illustrated in Fig. 5, where the
the trailing edges became less concave and the values of R stresses for yield and fracture are shown for a final product
increased as shown in Fig. 4. This indicates that the prod- moisture of 7.9% over the range of extrusion moistures.
ucts were more elastic at the higher moistures. The higher moisture extrusion provides a tougher, more
resilient product which tends to yield earlier and to with-
Mechanical
properties stand higher stress before fracturing.
The data from the Instron testing machine showed that
upon compression, all samples exhibited an initial linear Internalproductstructure
stress-strain relationship indicating an elastic response. This The median cell size found in axial cross-sections of the
elastic region was terminated by a yield point indicated by product samples was found to decrease with increasing
an inflection in the curve. The high moisture samples then moisture as shown in Fig. 6. Correspondingly, the number
continued to compress in plastic flow, with ever increasing of cells per unit cross-section area increased with increasing
resistance. The low moisture samples, on the other hand, moisture to maintain the same degree of expansion or
fractured at a maximum compressive force. The peaks were density. Low moisture extrusion, therefore, expands the
not as sharp as one would expect from true brittle fracture product with a coarser cell structure, or has fewer cells of
because the products tended to crumble in a combination larger size.
of plastic flow and brittle fracture. As illustrated in the photographs in Fig. 8C and 8D,
The yield stresses were calculated from the yield point close examination of the internal structure revealed that it
and the cross-section area of the product pieces. The ulti- is not uniform in all directions, but shows a definite linear
mate strength was similarly calculated from the peak com- structure, or grain, in the axial direction of flow within the
pressive force for those samples which showed a peak. The

.25L-
8 10 12 8 10 12
-.8 I I Product Moisture, %
a 10 12 Product Moisture, %
Fig. 5-Yield stress andstrass at fracture (ul-
Product Moisture, % timata strength) in compression for products Fig. 6-Median cell size of expanded prod-
Fig. 4-Elastic rebound vs. extrudedproduct dried to 7.9% final moisture vs. initial ex- uct vs. extruded product moisture, %, at
moisture, %. truded product moisture, %. constant expansion rate (density).

Volume 50 (1985)-JOURNAL OF FOOD SCIENCE-251

 LOW MOISTURE EXTRUSION.. .

extrusion die. This is caused by shear strain in the die

orifices. Within this grain were large fissures or fracture
lines. As shown in Fig. 7, the number of these fissures was
greater at low extrusion moisture.

CONCLUSIONS
THE RESULTS SHOW that extrusion at low moisture is
responsible for reduction in plasticity and elasticity of the
extrudate. The most obvious manifestation of these changes
is in the shape of the extruded product, which is less
rounded at the lower moistures.
Product extruded at low moisture is less resilient than
product extruded at a higher moisture and then dried to the
same low moisture. This results in a more fragile product,
subject to more breakage and generation of fines in hand-
I I I ling.
8 The product resiliency seems to be related to the cellular
8 10 12 structure developed during expansion. The high moisture
extrudate has a finer cell structure with less longitudinal
Product Moisture, % fissuring. The larger cells and fissures in the low moisture
product may cause stress concentrations and increased
susceptibility to brittle fracture.
Fig. 7-Longitudinal fissuring: fissure count, n/cm, vs. extruded The structural changes are probably related to the
product moisture, %. changes in plasticity and elasticity revealed in the product

Fig. 8-Photomacrographs of product samples: (A) Three external views of extruded product; (6) Product size ranges for high and low mojsture
extrusion; (C) three sectional views of extruded product showing longitudinal grain of internal structure; (0) Longitudinal section of product
with fissures stained for enhanced contrast.

252-JOURNAL OF FOOD SCIENCE-Volume 50 (1985)

morphology. At low moisture, the relatively inelastic ma- Harper. J.M. 1981. “Extrusion of Foods.” CRC Press, Inc., Boca
trix can allow the developing cells to burst and coalesce Raton. FL.
Hawk: B.W. 1981. Control of extrusion variables in extrusion
into larger voids. Longitudinal fissuring is caused by failure cookmg. Cereal Foods World 26(4): 170.
under shear within the die orifices, and is aggravated by Holay, S.H. and Harper, J.M. 1982. Influence of the extrusion shear
environment on plant protein texturization. J. Food Sci. 47:
reductions in plasticity and cohesiveness. 1 Ilfio
A.,““.

Asymmetrical die flow is a phenomenon peculiar to Lawton, B.T.. Henderson. G.A.. and Derlatka. E.J. 1972. The
effects of extruder variables on the aelatinization of corn starch.
twin-screw extrusion, and is caused by nonuniform pressure Can. J. Chem. Eng. 50(4): 168.
distribution at the screw discharge, especially in the Mason, M.. Gleason, B.. and Rokey, G. 1982. A new method for the
presence of screw and barrel wear. This effect was found to determination of degree of cook. Yellow Splings Instrument Co..
Inc., Yellow Springs. OH.
be more pronounced at lower moisture due to either: in- Mercier, C. and FeiIlet. P. 1975. Modification of carbohydrate
creased lateral forces on the adjacent extruder screws at components by extrusion-cooking of cereal products. Cereal
Chem. 52: 283.
lower moisture; or reduced ability of the flow elements to Miller. R.C. 1984. Effect of wear on twin-screw extruder oerform-
equilibrate before entering the dies because of the changes an&. Food Technol. 38(2): 56.
Park. K.H. 1976. Elucidation of the extrusion puffing process. Diss.
in rheological properties at low moisture. Abst. Int. B 37(l): 148.
Rossen, J.L. and Miller, R.C. 1973. Food extrusion. Food Technol.
27(8): 46.
REFERENCES Williams, M.A.. Horn. R.E., and Rugala, R.P. 1977. Extrusion: an in
depth look at a versatile process. 1. Food Eng. 49(g): 99.
Anonymous. 1983. Extruders: single vs. twin-screw. Food Eng. Ziminski, R.D. 1981. Economics of twin-screw processing of food.
June, 103. 1981 Ann. Meeting AICHE, New Orleans.
Harper. J.M. 1979. Food extrusion. CRC Crit. Rev. in Food Sci. Ms received 6/11/84; revised 8124184; accepted 10/l/84.
Nutr. 11: 155. Presented at the 44th Annual Meeting of the Institute of Food
Harper, J.M. and Holay. S.H. 1979. Optimal energy usage in food Technologists in Anaheim. CA, June 11.1984.
extrusion. ASAE Paper 79-6508.

HEATING MILK-BASED FORMULAS. . . From page 214

REFERENCES Bigelow, W.D.. Bohart, G.S., Richardson, A.C.. and Ball, C.O. 1920.
Heat penetration in processing canned foods. National Canners
Anonymous. 1977. “Carrageenan. ” Monograph no. 1, FMC Corp., Assoc, Bull. No. 16-L; Washington. DC.
Marine Colloids Div., Springfield, NJ. Hunziker, O.F. 1949. “Condensed Milk and Milk Powder,” 7th ed.,
Barabas, J. 1981. The influence of combined action of heating tem- published by the author, La Grange. IL.
perature, active acidity and of CaCl supplements on milk protein Jones, A.T., Pflug, I.J.. and Blanchett, R. 1980. Effect of fill weight
coagulation. Milchwissenschaft 36(2 P: 98. on the F-value delivered to two styles of green beans processed in
Berry, M.R. Jr. and Bradshaw. J.G. 1980. Heating characteristics of a Sterilmatic retort. J. Food Sci. 45: 217.
condensed cream of celery soup in a Steritort. Heat penetration Naveh, D. and I.J. Kopelman. 1980. Effect of some Processing
and spore count reduction. J. Food Sci. 45: 869. parameters on the heat transfer coefficients in a rotating auto-
Berry, M.R. Jr. and Bradshaw, J.G. 1982. Heat penetration for clave. J. Food Process. Preserv. 4: 67.
sliced mushrooms in brine processed in still and agitating retorts Packard, V.S. 1982. “Human Milk and Infant Formula.” Academic
with comparisons to spore count reduction. J. Food Sci. 47: 1698. Press, New York, NY.
Berry, M.R. Jr. and Dickerson. R.W., Jr. 1981. Heating characteris- Stumbo C.R. 1973. “Thermobacteriology in Food Processing,”
tics of whole kernel corn processed in a Steritort. J. Food Sci 2nd ei. Academic Press, New York, NY.
46: 889.
Berry, M.R. Jr., Savage, R.A.. and Pflug, I.J. 1979. Heating char M S received 4/30/84; revised 7/30/84: accepted 8/14/84.
acteristics of cream-style corn processed in a Steritort: Effects The authors acknowledge Susan Grause for secretarial assistance
of headspace, reel speed and consistency. J. Food Sci. 44: 831. during the preparation of the manuscript.

Volume 50 (1985)-JOURNAL OF FOOD SCIENCE-253 I