The range of equipment used to mix feed is at least as diverse as the ingredients. There have been
many attempts to reduce mixing concepts to a series of engineering equations thus facilitating
equipment design from a theoretical approach. The fact is most contemporary mixing equipment,
including horizontal ribbon mixers, vertical screw mixers, and drum mixers, have simply evolved from
historically successful designs without benefit of mixing theory input. For example, most horizontal
mixers have a length approximately three times their diameter and have a rotational speed of 75-100
meters/minute regardless of the diameter. The inside ribbon is usually 2.5 times the thickness of the
outside ribbon to balance the directional forces applied due to ribbon diameter.
Given this discussion, it is easy to appreciate the complexity of the mixing operation in a production
facility. Yet, this seems to be an area of little concern to most feed manufacturers - commercial or
private. As regulatory pressures for additive uniformity increases and as the need for providing uniform
nutrient density to genetically superior livestock becomes necessary, it will be in the best interest of
feed manufacturers to ensure uniformity through testing.
Current and Future Aspects
To focus on the regulatory aspect of uniformity, the following excerpt is taken from the 1990 FDA
Regulatory Guidelines (FDA, 1990):
Equipment (225.30) All equipment used in the manufacture of medicated feed shall have the
capacity and capability to produce a homogeneous medicated feed of the intended potency. The
capability of the mixing equipment should be demonstrated upon installation and periodically as
needed to ensure proper adjustments during operation. Written documentation of the adequacy
of the equipment should be available for FDA review.
In January of 1990, the Degussa Corporation introduced a program to monitor uniformity of feeds
manufactured by customers using their amino acid and other products (Wicker and Poole, 1991).
Their results would indicate that only about half of the feeds tested would be of satisfactory uniformity
(C.V. < 10%). About 30% had a C.V. of 10-20% and the remaining 20% of the feed samples had a
C.V. of > 30%. It is not known precisely at what level of uniformity animal performance will be affected,
but one can certainly assume that at a C.V. of greater than 20%, performance would be decreased.
The samples tested in this study were generally from large, centrally controlled feed mills. To gain a
perspective on how well on-farm feed manufacturers do, Stark et al. (1991) conducted a study similar
to that of Wicker and Poole except using salt as the tracer rather than synthetic amino acids. The
results tended to parallel the Degussa report with about 42% of the samples having a C.V. of < 10%,
46% between a C.V. of 10% and 20%, and 12% having a C.V. of > 20%.
It is apparent that, at least in a significant portion of feed produced, nutrient uniformity criteria is not
being met. As regulatory authorities move toward required equipment validation, it is imperative that
the feed and livestock industries come to agreement as to what levels of nutrient uniformity is needed and how that uniformity is to be measured. There currently exists a standard (ASAE Standards, 1990)
for testing solids-mixing equipment for animal feeds; however, the procedure is complicated and a
great deal of the data required is meaningless to either regulators or animal performance.
The objectives of mixer testing can be many. The most common reason for testing a mixer is to
determine the mixing time at which an adequate or satisfactory blend is obtained. The procedures
are relatively simple and involve taking samples at specific time intervals. The assay used and statistical
treatment is relatively straightforward.
Result Interpretation And Statistical Evaluation Of Mixing Tests
The standard deviation and coefficient of variation can be used to measure the results of a mixing
test. In very simple terms, they help measure the distribution of values and express the value as one
number. In order to interpret the results of a test, the variation of the procedure itself should be known.
For instance, the coefficient of variation of the Quantab7 method is about 10%; therefore, if the result
of the mixer analysis is 10% or less, we assume that a Agood@ mix has been achieved. The same
situation applies for other procedures.
One can also use these procedures to isolate points of segregation in the feed mill. If one has a C.V.
of 8% at the mixer and a C.V. of 18% after a transfer conveyor, there is a problem between the mixer
and that point.
Numerous assay methods have been used for mixer evaluation. One criteria for selection should be
to assay for an ingredient, nutrient, or chemical that comes from a single source. Salt, therefore, is
a good selection while protein or nitrogen would be a poor selection.
Sampling is also very important to any mixer evaluation. At least 10 samples should be taken, either
from several locations in the mixer or at timed intervals during the mixer discharge. Enough of each
sample should be taken each time (approximately 2 lb) to allow for a good test sample. In order to help
clarify the requirements for a test procedure to be used for uniformity, the following criteria are offered:
1. The assay principle should be based upon a common ingredient or nutrient that is usually in the
formula or can be added without risk.
2. The cost for each assay should be minimal (< $2.00 each).
3. The assay procedure should be simple, fast, accurate, precise, and able to be done on-site.
4. The assay should present no safety hazard to personnel or animals.
5. The assay principle should be supplied from a single source.
6. Sample size required should be reasonable but large enough to reduce or eliminate sampling error.
7. The target mix uniformity (C.V.) should be approximately two (2) times the proven analytical variation
for the assay selected but in no case exceed 10%.
8. The statistical procedures required should be easily understood and performed.
As the reader can imagine, there is no "perfect" procedure available.