Validating mixing of formula feed
If a feedmill is mixing longer than needed to achieve a complete mix, it is reducing its capacity, wasting energy and labor, and causing unnecessary wear to its mixer.
By David Eisenberg, MicroTracersWhy is validating feed mixing important? If a feedmill is mixing longer than needed to achieve a Complete mix, it is reducing its capacity, wasting energy and labor, and causing unnecessary wear to its mixer. If feed is not mixed “Completely”, it will negatively impact the performance of certain animals, poultry and fish.
The “level of scrutiny” of the feed being manufactured is a critical issue. Research studies from Kansas State University evidenced significant negative impacts on baby chicks and baby swine if feed was not mixed Completely. No negative impact was found with adult swine.
A baby chick consumes 10 grams of feed per day and an adult cow 10 kgs. If one wants each day's feed to contain all nutrients at the formulated level then the feed consumed each day may constitute the “level of scrutiny”. Each day's feed should contain all nutrients at their formulated level.
Shrimp feed may demand an even higher standard. Certain vitamins must first be sprayed onto a fine powder carrier before adding the vitamin to a premix and then adding the premix into the final feed.
In validating feed mixing, five major issues must be considered:
#1. Choice of tracer
#2. Addition of tracer
#3. Sampling of feed
#4. Analysis of samples
#5. Interpretation of results
Choice of tracer
The tracer should come from only one source and be formulated into the feed at a low part per million level to reasonably assure therapeutic animal drugs, coccidiostats and other critical ingredients – many of which are added to feeds at low ppm levels – are mixed Completely.
It is almost meaningless to test for protein as it is contributed to the feed from several ingredients and is present in the feed at a very high level. Efforts to use NIR led to meaningless results (1).
Salt (sodium chloride) is widely used as a tracer in validating mixing but the American Society of Agricultural and Biological Engineers Standard (ASABE) (USA,2017) specifies it be added to a specially formulated feed at 2% (20,000 ppm) of the formula to assure results will not be confused by salt being contributed to the feed from fishmeal or other ingredients.
Salt is not considered an adequate tracer in Germany, the Netherlands, France and many other EU countries or by Brazil or Japan because it is added to the feed at far too high a level to reflect the dispersion of microingredients.
Amino acids lysine and methionine added to feeds at about 1-lb/2,000-lbs (500 ppm) have been used widely in the poultry feed industry especially and have yielded meaningful results (2).
Manganese and other minerals including zinc are used widely with the manganese added to the feeds at about 100 ppm and have yielded meaningful results. Manganese is included in the GMP+ Test Methods, (the Netherlands, 2021) used widely around the world to meet Feed Quality System requirements.
Microtracers F (colored uniformly sized iron particles, 25,000/gram) added to feeds at 50 ppm and other Microtracer variants are included in the ASABE Standard as well as in the GMP+ Methods as well as in Official publications of Spain, Brazil and Japan. They have yielded meaningful results as evidenced by their inclusion in both the ASABE and GMP+ Methods..
In Germany and other Northern European countries, a Special Microtracer FSS (colored stainless steel particles, 500,000/gram) is widely used because feed manufacturers want to add a tracer at 10 ppm and achieve a CV of 5% by counting 400 spots. This is included in the GMP+ Methods.
Addition of the tracer to the trial feed
This depends on the type of mixer being tested.
Horizontal paddle mixers often mix as two halves so when possible it is best to test one batch of feed adding one tracer (.e Microtracer F-Red) at one end of the mixer and another tracer (i.e. Microtracer F-Blue) at the opposite end and then after mixing to take one sample from each end.
Horizontal ribbon mixers usually achieve a Complete mix so the tracer is usually added at the same time and location as vitamin premixes or “hand add” ingredients. The tracer should be mixed into 1-lb per ton of diluent to simulate addition of a 1-lb vitamin or medicated additive.
Vertical mixers usually achieve a Complete mix and the tracer can be added as for horizontal ribbon mixers.
Sampling the trial feed
This depends on the purpose of the trial.
If the purpose is to validate performance of the mixer, one should ideally take samples from within the mixer. One mixer manufacturer took 300 “grab” samples from within their mixer using compartmentalized probes. After validating the performance of their mixer, they manufactured hundreds more over a period of many years.
Usually, it is difficult or impossible to take samples from within a mixer. In such instances, it is best to take samples from the screw conveyor - usually 10 or more “grab” samples from spaced/timed parts of the batch. Feed from earlier batches not formulated with the tracer may be left in the surge bin below the mixer and will commingle with feed containing the tracer. It is often best to formulate the tracer into two or more batches of feed and to not sample the first batch or to recognize the first and last samples taken for analysis may be unrepresentative and legitimately discarded.
If the purpose of the trial is to validate the mixing of the feed delivered to animals, poultry or fish, taking samples from truck loading or even after feed is delivered to the animals may be best. Results though will evidence not only performance of the mixer but also comingling of feed as it passes through the feedill- bucket elevators, surge bins, pellet mills and holding bins.
Samples must be “grab” and not composites and should be approximately 2 times the amount required for analysis.
Analysis of the feed samples
This depends on the tracer used. Amino acids are determined by high performance liquid chromatography (HPLC). These instruments are now widely available at commercial analytical laboratories and often at feed manufacturers own laboratories. The typical cost for analysis for both lysine and methionine is $75. The analytical error is about 5%.
Manganese, zinc and other metals are determined by atomic absorption spectroscopy (AA) or inductive coupled plasma emission mass spectrometry (ICP MS). These instruments are now widely available at commercial analytical laboratories and often at feed manufacturers own laboratories. The typical cost for analysis of manganese by AA is $60. The cost for an ICP MS scan including many other elements is $100-150. The analytical error in metals analysis is about 5%.
Microtracers F are determined by magnetic separation of the tracer using a “Rotary Detector” laboratory magnetic separator, spinking the tracer onto filter paper, spraying 40-80% alcohol over the paper, drying the paper and counting colored spots either manually or using automated spot counting/data transfer APPs. The cost per analysis for samples sent to Micro-Tracers,Inc. Is $30 though most feed manufacturers buy a Rotary Detector (cost about $1,260) and perform the analyses themselves with a throughput or about 10 samples/hour. The analytical error is about 2%.
A feed manufacturer can perform Microtracer analyses at the feedmill and thereby obtain results under one set of mixing conditions and if mixing results are unsatisfactory perform additional trials the same day under different conditions that may then yield acceptable results.Interpreting trial results
Amino acid and manganese and other metals analyses are interpreted by evaluating the Coefficient of variation (CV) obtained from analysis of the samples taken and analyzed, usually 10 or more. Arbitrary standards have been set by various organizations and governments. Usually a CV below 5% is an ideal objective, a CV below 8% is generally considered as evidencing a Complete mix, 8-12% as evidencing an Acceptable but Probably Incomplete mix and about 12% an Incomplete mix.
Microtracer particle counts are evaluated using the objective Poisson statistics and Chi-Square test. The critical property of the Poisson statistical distribution is that a count has an inherent standard deviation equal to its square root. If one counts 100 spots per sample, then analysis with no analytical error of an infinite series of samples of a Completely mixed feed would yield a standard deviation of 10 and then a CV =10% (10.100).
But one is not analyzing an infinite number of samples, usually only 10. One uses the Chi-Square test to determine objectively if a result from the trial reasonably evidences a Complete mix, or does not. Usually a Chance Probability higher than 5% is Considered as evidencing a Complete mix because any excess variability is not “statistically significant”. A Chance probability between 1% and %5 evidences a Probably Incomplete mix and a Chance Probability below 1% evidences an Incomplete mix.
The CV from tracer counts will depend on the number of particles counted per sample. If one counts 25 spots per sample, then the expected CV is 5/25 or 20%. If one counts 400 spots per sample, the expected CV is 20/400 or 5%.
It is better to take, analyze and interpret results from more samples than to achieve a low CV by counting a large number of particles per sample (3).
Other Issues
Different feeds may require different mixing conditions. Feeds with finer particle sized ingredients have proved to mix more quickly than feeds with larger particle sizes ingredients (4). Certain feeds may “puff up” during mixing leading to overfilled mixers that can yield Incompletely mixed feeds. Ribbons or paddles of a mixer.should extend 4 inches (10 centimeters) above the feed.
It is meaningless to use salt if one is testing pelleted feeds as the salt dissolves so even if the dry micro ingredients in the feed are not mixed, the salt analyses will evidence a Complete Mix. TMP cattle feeds with high moisture and high roughage pose a challenge in sampling or analysis and in the analyses performed.
Validating versus Verifying Mixing
It seems common sense to validate the performance of all mixers at the time they are installed to establish they meet the specifications of the mixer manufacturer. Specifications may include the mixing time, batch size and revolutions per minute of the mixer required to Completely mix one or more than one type of feed.
Mixer performance should also be verified periodically after the mixer has been operating.
The Province of Quebec (Canada) has required mixer performance verification twice each year, testing for a micro ingredient (salt accepted) twice each year for commercial feed mills manufacturing medicated feeds since 1987.
The European Union has required mixer performance verification since 1995, though the various member States set individual standards.
Brazil and Japan have enacted mixer performance verification in 2023.
The USA has no regulatory requirements.
Certain feed manufacturers set standards that are far more demanding. One major commercial feed manufacturer in Mexico verified performance of all its mixers at five feedmills every week. One major poultry integrator in Indonesia also verifies its mixing every week. They both advised that they wanted to detect manufacturing errors very quickly and the cost of testing was low enough they could test frequently.
Conclusion
Knowledge is power. Taking and analyzing even one sample is an infinite improvement over testing none.
References
“The Impact of marker selection, in-line near-infrared spectroscopy (NIR), and mix time, on the coefficient of variation (mix uniformity), body weight uniformity and broiler growth performance during the starter, grower, and finisher periods”, A.A. Rubio and More, Poultry Science, December 2023.
“How Well is Your Mixer Performing?”, Wicker and Poole, Feed Management, November 1991
“On the Use of Particulate Distributions for Determining the Degree of Homogeneity in a Feed Mixture”, Letter and Article, Professor David Bernotas, UC San Diego, April 2013 (at website: www.microtracers.com, Published Literature, Item B-1).
“Effects of Particle Size and Mixing Time on Uniformity and Segregation in Pig Diet”, N. Amornthewaphat, K.C. Behnke and J.D. Hancock, Kansas State University, Swine Day, 1998.
.