Introduction
Feed manufacturing consists of a series of steps, or processes, in which individual
ingredients are combined into a homogenous mix, further processed into customer
desired forms (coarse dairy rations, pellets, flakes, etc.), and packaged or delivered, again
to the customer specifications.
Some work has been done evaluating the further processing stages as these steps
can directly influence the nutritional performance of the animal. However, very little has
been done regarding the process of ingredient mixing.
Therefore, the purpose of this article will be to discuss the mixing process and the
variables that effect a successful mix. In addition, the measurement of mix uniformity
will be examined in detail.
vertical mixer and the TMR mixer both utilize different principles of action, they both
serve a common purpose: the efficient mixing of rations that contain high levels of
coarse forage. Traditional ribbon or paddle horizontal mixers are not that effective at
diets that contain roughage. This is evidenced by the hesitancy to include whole
cottonseed in the feed manufacturing process. Instead, horizontal ribbon and paddle
mixers are designed to mix ingredients with a fairly similar particle size, quickly and
efficiently.
On the other hand, vertical and TMR mixers are designed to blend coarse forage
particles with other coarse ingredients, such as beet pulp and whole cottonseed. Keeping
this concept in mind, it is very easy to see that the TMR or vertical mixer is not designed
to effectively mix ingredients with a very small particle size or are included in very small
proportions in the ration. Therefore, the concept of pre-mixing some ingredients provides
the necessary transition between horizontal feed mixing and on-farm ration mixing.
90°
Width, in. 6.500 6.000 6.375
Length, in. 18.625 18.500 18.375
Shell clearance, in. .25 .500 1.250
180°
Width, in. 6.500 6.500 6.375
Length, in. 18.625 16.750 18.250
Shell clearance, in. .25 .190 .281
270°
Width, in. 6.500 6.250 6.500
Length, in. 18.625 18.250 18.500
Shell clearance, in. .25 .500 .250
a
Blades were changed the first of June, 2003.
The shell clearance category will be dealt with at a later point, as there are other
concerns/solutions to this problem. The key to any analysis is to examine a new paddle
alongside the current paddles being examined. In this case, three paddles were examined
and two salient points emerge: the wear is occurring along the width transversal and the
length as well. Therefore, the take home point here is that even though the width is still
within specification, if the length is significantly shorter than the new paddle, then the
paddle should be changed. Again, the wear on this design of paddle occurs in two places
and the paddle should be analyzed in this context.
Analysis of this mixer revealed that the length dimension (horizontal, point-to-
point) was the cause of the most concern, especially in the paddle in the 180° position.
Consequently, based on the excessive build-up in the mixer and the dimensional analysis,
it was suggested that the entire paddle configuration be changed to improve mixer
performance.
Mixing Element/Mixer Shell Clearance – going hand-in-hand with the blade
analysis is the gap between the mixer shell and the mixing element. In this discussion,
the differentiation between paddle vs. ribbon can be ignored, as the only difference is the
method of adjustment. A good rule of thumb is that when the gap between the shell and
the mixing element exceeds ¼”, then the mixing element should be adjusted to decrease
this gap. For ribbon mixers, this is accomplished by removing individual sheaves in the
main shaft bearing mount. For paddle mixers, the individual paddle is moved closer to
the shell by adjusting the location of the paddle bolt in relation to the support shaft; all
paddles have an oblong bolt hole that allows for some degree of adjustment, closer or
further from the mixer shell. It is important to note that whenever adjustments are made
to the ribbon or paddle, care should be given to the uniformity of the mixer shell or any
deformations in the center shaft. Consequently, this adjustment should be undertaken by
competent maintenance personnel who know the physical condition of the entire mixer.
If this is ignored, severe damage to the mixer can occur as metal-to-metal contact will
ensue.
Mixer tub liner – the mixer tub liner should be inspected every 6 months for signs
of excessive wear. In general, the warning signs are ridges and valleys that occur with
improperly adjusted ribbons and paddles. This occurs primarily by ignoring the gap
between the mixing elements and the tub. When the gap becomes too excessive, the feed
flow pattern becomes layered. When this occurs, any movement of the bottom feed layer
is due to friction imparted by the mixing element. Over time, this excessive friction will
cause wear patterns to develop in the tub surface, which will increase the incidence of
wear. Similar to a pellet mill door, the uneven wear of a mixer liner is an expensive
problem that can be, to a large extent, minimized by routine preventative maintenance.
Liquid addition – in today’s feed
manufacturing environment, liquids are
routinely added to the mix. These
liquids can include: water, molasses, fat,
and liquid amino acid sources. Proper
maintenance/design of the liquid
addition system is key, as ignorance of
the issues can lead to excessive build-up,
ingredient carry-over, and increased
mixer liner wear. The attached picture
demonstrates this issue very well. The
liquid addition system for this mixer consisted of two copper pipes with a crimped
discharge end. This “arrangement” was responsible for excessive build-up along the
walls; furthermore, the actual amount of liquid that was added to the mixer could not be
verified, as it was strongly suspected that a portion of the liquid did not make it to the
feed, but was instead entrained in the build-up. Consequently, some changes to the
addition system were required to facilitate uniform liquid application. One change that
can be attempted would be to fit the addition line with a spray nozzle to facilitate some
degree of backpressure in an attempt to get the liquid away from the side of the mixer
walls. In addition, the use of the spray nozzles would allow for a pattern to be applied to
the stream, which would facilitate more uniform liquid distribution throughout the mix.
Another issue with this liquid addition system was the use of copper piping. The
use of copper piping with animal fat is strongly discouraged, as copper will react with the
fat and cause rapidly accelerated oxidation of the fat, which can lead to off-flavors. This
is especially important in swine diets, as pigs are very sensitive to these off-flavors.
Discharge gate – the most
overlooked and ignored component of the
mixer is the discharge. Proper maintenance
and function of the discharge is essential to
insuring that all ingredients have an equal
opportunity to co-mingle with one another.
Mixer discharges can be classified in two
categories: gate and drop bottom. Gate discharges are essentially comprised of a port
installed in the bottom of the mixer that typically has a movable slide installed over the
mouth of the port. When the mix cycle is complete, the gate is opened (via pneumatic or
hydraulic means, primarily) and the feed is discharged into a conveyor below the mixer.
In contrast, drop bottom discharges are composed of two doors that also function
as the mixer tub liner. When the mix cycle is complete, the doors are opened and the
entire batch of feed is dropped into a surge bin installed between the mixer and the take-
away conveyor.
From an efficiency stand-point, drop bottom discharges are the most efficient and
allow for maximization of the mix cycle. However, the drop bottom design is also more
prone to the one major concern with the discharge: ingredient leakage. One of the issues
that will be dealt with in more detail is density of the individual ingredients. For this
discussion, however, it is important to realize that the ingredients in the mix have a wide
range of bulk densities. As a result, heavier ingredients have a tendency to settle during
the mixing cycle, compared to the lighter ingredients (cobalt carbonate vs. wheat midds).
Because of the difference in ingredient densities, there is a propensity for greater density
ingredients to settle towards the bottom of the mix, especially if the mixing elements are
not functioning as designed. If the discharge gate is completely closed, or if gaps exist
along the run of the drop bottom discharge, ingredient leakage can occur. Therefore,
some attention must be paid to ensure that when the discharge gate/doors are closed, that
the seal is tight and ingredient leakage is not occurring.
As can be seen, the average particle size of the ingredients can vary tremendously.
Recall, 1 mm is approximately 1000 μm. Therefore, the particle size of the ingredients in
a common feed mix can vary tremendously.
But why is particle size important? The differences in particle size allow us to
mix much more efficiently. For example, if the goal was to mix 1 blue basketball into
5000 red basketballs, how much confidence would one have that the target recipient
would have an equal chance of finding the blue basketball. On the other hand, if the 1
blue basketball was equivalent to 100 blue baseballs, it is very easy to make the
connection that the probability of finding a blue baseball is significantly greater than
finding the blue basketball.
The above discussion lends well to the particle count issue. Examination of Table
two demonstrates a very salient point: significant differences in particle size also lead to
significant differences in particle count. This fundamental concept is very critical in
mixing theory, as significant differences in particle count can lead to fundamental
differences in the distribution of the ingredients throughout the mix.
One of the trends in animal nutrition is the push to include micro-nutrients into
the daily rations of animals. These micro-nutrients can be defined as any ingredient that
goes in at rates lower than .5% in the mix; however, it is not uncommon to find
ingredients that are being included at rates lower than 100 ppm. Because of this trend to
feed the animal on the micro-level, it is absolutely critical that the micro-ingredients are
uniformly distributed throughout the entire mix. For example, consider biotin; biotin is
typically included in the ration at very low levels. Consequently, it is very important that
the biotin is distributed throughout the mix, so that the animal has an equal opportunity to
receive the biotin it requires in each and every bite. Table Three illustrates this point
very nicely; with broiler chicks, considering that the daily feed consumption is
Table 3. A comparison of biotin intake by broiler chicks consuming two sources of biotin
Product
Category Rovimix Biotin Competitive Triturate
very low, it is absolutely essential that the micro-nutrient have a very large particle count,
so that the animal is assured that, in each and every bite, it has an equal opportunity to
ingest the proper amount of active nutrient.
Analogous
to the particle
count discussion is
uniformity. From a
practical stand-
point, the more
uniform an
ingredient is, the
greater the chance
the ingredient will
be distributed
uniformly
throughout the
mix. The attached
pictures compare
Rovimix H-2 (left
picture) to a competitive triturate product (right picture). Examination of the pictures
demonstrates an important point: using the product on the left insures that each particle
delivered to the animal contains biotin; in contrast, each particle, in the picture to the
right, delivered to the animal may be biotin, or it may be a carrier.
This area of discussion is not complete without covering the area of premixing.
Premixing is the practice of taking very small inclusion micro-ingredients and mixing
them with a nutritive carrier to “pre-disperse” the micro-ingredients in, with, and under a
carrier. The completed premix is then added to the feed at the desired inclusion rate.
The primary advantage of premixing is that the end user only has to mix the
premix carrier into the finished feed; the micro-nutrients have already been successfully
mixed into the carrier. Therefore, if the premix is uniformly distributed throughout the
mix, it can be assumed that the micro-ingredients are also uniformly distributed.
One of the primary reasons for the improvement of micro-ingredient delivery is
that the use of a premix can significantly alter the physical properties of the “ingredient”
added to the mixer, many times to benefit of the end-user. Table Four presents a
comparison of the physical properties of Rovimix A-500 and a common premix, an A-30
dilution.
Table Four is instructive for a couple of reasons. The first reason is that it clearly
shows that the use of a premix can improve some of the more important handling
properties of a micro-nutrient, such as bulk density and dustiness (Heubach dust index).
Secondly, Table Four demonstrates that the use of premix helps minimize the difference
in particle size between ingredients.
Previously, it was suggested that large differences in particle count and particle
size can make a tremendous difference in the mixability of an ingredient. Taking this
concept a step further, a premix allows the end-user the flexibility of uniformly pre-
distributing several micro-nutrients at once and then, only having to worry about
dispersing the premix into the desired finished product; the end-user no longer has to
worry about the micro-ingredients, as they have already been “pre-mixed” into the
finished product!
15-20 Fair Increase mixing time 50%; look for worn equipment,
overfilling, or ingredient sequencing
Many times, the equipment manufacturer provides solid advice regarding the appropriate
dry/wet mix times for the make and model of mixer installed. However, it is important to
note that these times may be adjusted up or down according to product mix and
production requirements. As a note of caution: the batch cycle needs to be considered
any time that the mix cycle is changed. Many times, the batch cycle is synchronized with
the mix cycle to insure that the ingredients are scaled in time for the mixer to complete its
mix cycle and discharge. A dramatic shift in mix time, in either direction, can have
additional ramifications to production efficiency, irrespective of the mix cycle.
Paddle 3 3
Twin shaft, double paddle .5 1
Ribbon 2 3
Double ribbon 1-2 2-3
Twin shaft, twin ribbon .75-1 2
Tip speed – if a mixer has a poor uniformity result and changing the mix times is
not an option, the speed of the mixing elements at their tips can be increased. There is
limited data in the public domain regarding this issue (Table 8); however, most mixer
manufacturers can assist regarding the proper balance between tip speed and mix
uniformity.
Coefficient of Variation
Salt, % 40.5 12.1 9.7
Red micro-tracer, % 53.4 16.6 11.3
Blue micro-tracer, % 53.9 17.0 10.6
Sodium, % 44.5 23.2 22.8
a
Adapted from McCoy et al., 1994.
2 normal (double-peaked)
40.00%
1
distribution. 20.00%
0
Distribution analysis is
.00%
76 82 88critical to the proper calculation of
More
Co, ppm
the uniformity. Pfost (1976)
proposed that the uniformity of a
mix can be best described by calculating the coefficient of variation (C.V.). The C.V.
relates the standard deviation to the
Binomial Distribution
mean in the form of a percentage.
7 120.00% Said another way, the C.V. is
6 Frequency
100.00%
essentially a method to express the
Cumulative %
5
80.00%
sample standard deviation as a
4
60.00%
percentage of the sample mean. In
3
40.00%
general, in properly categorized
2
distributions, the standard deviation
1 20.00%
and the mean change together.
0 .00%
0.11 0.14 0.17 More Therefore, the C.V. is an effective
Sodium, % tool for describing the variability
present in the sample population
(Snedcor and Cochran, 1967). The general equation for the C.V. is as follows:
⎛s⎞
C.V .,% = ⎜ − ⎟ × 100 (eq. 1)
⎜ ⎟
⎝ x⎠
where:
The general equation for the C.V., as can be seen above, is dependent on the accurate
calculation of the sample standard deviation and the sample average (or mean).
Consequently, these calculations are strongly dependent on the classification of the data
distribution. In general, two data distributions are commonly observed in mixer testing.
These include the normal and non-normal distributions.
Normal distributions – equations 2 and 3 can be used to calculate the sample
average and sample standard deviation for data that approximates a normal distribution.
−
x=
∑x i
(eq. 2)
n
∑ ⎛⎜⎝ x − x ⎞⎟⎠
−
i
s= (eq. 3)
n −1
These equations are dependent on the data assuming a normal distribution. Therefore,
the data must be categorized prior to making these calculations.
One area that deserves some mention is the determination of outliers in a data set.
The presence of outliers in a data set can skew a distribution and can be responsible for
faulty data analysis. One of the best tools to test for the presence of outliers in a data set
is the Kurtosis value. Kurtosis measures the degree of departure from normality present
in a given data set. The general equation for the kurtosis of a population is given in
equation 4:
k=
( xi − μ )
4
(eq. 4)
σ4
Distribution analysis, %
< .11% 13.33
.12-.15% 40.00
.16-17% 20.00
>.17% 26.66
Using this approach, it is apparent that 40% of the observations fall within an acceptable
range of variability. However, 60% fall outside of the range. As can be quickly noted,
the use of this approach requires some understanding of the acceptable ranges in a
system. Furthermore, the data must be categorized prior to analysis to allow for the
determination of the frequency distribution.
Sample size is a critical issue to consider, especially if the distribution is
suspected to be non-normal. The industry average in regard to the appropriate number of
samples is ten. However, if non-normality is suspected, that number should be increased
to at least 15; twenty is a more reliable value.
An alternate approach that can be used to determine a C.V., is to make a normal
approximation to a binomial distribution. Using the distribution analysis above, it is
apparent that 40% of the observations fell within the desired range of values. Therefore,
a C.V. can be calculated using equations 5 and 6 and applying the values to equation 1.
−
x = nπ (eq. 5)
where:
n = number of observations
π = probability that a sample will result in the desired value
s = nπ (1 − π ) (eq. 6)
Analyzed using equations 2 and 3, the C.V. for this set of data was found to be 22%.
When the binomial equations were applied, the C.V. was found to be 32%. This
difference is significant enough to change the entire interpretation of the data set.
The above example clearly illustrates the importance of quantifying mixer
performance correctly. Many times, 10 samples are obtained from the mixer, and are
tested for an analyte that has been added at a relatively high inclusion rate. Following
this approach and neglecting to test for ingredients at critical inclusion levels can result in
a misleading mixer profile. As a general rule of thumb, a mixer should be tested for
uniformity using a tracer at the lowest ingredient inclusion rate utilized in the plant.
Using this approach can help quality and production managers adjust appropriate mixing
variables to achieve a consistent mix.
Conclusion
Mixing is one of the most critical operations in a feed mill. Factors such as
equipment type/condition, ingredient size, and mix time can significantly alter the
uniformity of the mixed product. However, a recognition of these critical factors can
help maintain acceptable mix C.V’s, which will result in the delivery of nutritionally
uniform diet to the animal.
Literature Cited
American Society of Agricultural Engineers. 1976. Test Procedure for Solids – Mixing
Equipment for Animal Feeds. In: H.B. Pfost (Ed.), Feed Manufacturing
Technology. AFMA, Arlington, VA. pp. 521.
McCoy, R.A., K.C. Behnke, J.D. Hancock, and R.R. McEllhiney. 1994. Effect of
mixing uniformity on broiler chick performance. Poult. Sci. 73:443.
Pfost, H.B. 1976. Feed Mixing. In: H.B. Pfost (Ed.), Feed Manufacturing Technology.
AFMA, Arlington, VA. pp. 85.
Pfost, H.B. and V. Headley. 1976. Methods of determining and expressing particle size.
In: H.B. Pfost (Ed.), Feed Manufacturing Technology. AFMA, Arlington, VA.
pp. 512.
Snedecor, G.W. and W.G. Cochran. 1967. Statistical Methods. Iowa State University
Press, Ames, IA.