[{"data":1,"prerenderedAt":42},["ShallowReactive",2],{"blog-besin-matrisi-nir-laboratuvar-en":3},[4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40],{"p":5},"New-crop corn started arriving in October. In the mill's nutrient matrix, corn has carried the same values for years: 8.0 percent crude protein, 14 percent moisture and an energy value taken from a textbook. Nobody has touched those numbers, because the formulas seem to work fine. A few weeks later an integrated customer calls: live weight in the grower phase is running a little under target. Nothing is wrong with the look of the feed or with the production records. The problem is in the matrix.",{"p":7},"Least-cost formulation finds the best answer for the numbers you give it. If the numbers do not reflect reality, the answer is right on paper and wrong in the barn. That is why a nutrient matrix is not a table you fill in once and leave. It is a data set that needs constant care.",{"h2":9},"A small gap with a large effect",{"p":11},"Take an example, with illustrative numbers. In our broiler grower diet, corn is included at 58 percent and the formula is solved for 21.0 percent crude protein. The average of samples from the new-crop corn, however, comes in at 7.3 percent crude protein, not 8.0. The gap is 0.7 points. Multiply by corn's share in the formula and the feed's real crude protein is 0.58 × 0.7, about 0.4 points, too low. The feed labelled at 21.0 percent is really around 20.6. Digestible amino acids fall by a similar proportion.",{"p":13},"When we update the matrix and solve again, the formula shifts from corn to soybean meal to close the gap. Each point of 46 percent soybean meal that replaces corn adds about 0.39 points of protein to the feed, so the gap needs a shift of about one point. With soybean meal at 18,000 TRY a tonne and corn at 9,500, that shift raises the feed cost by about 90 TRY a tonne.",{"p":15},"At first sight this looks like bad news. But those 90 TRY are a cost that was already due. If the matrix is not updated, the cost does not go away. It shows up in animal performance and in customer satisfaction. The reverse happens too. When analyses come in above the table, the mill that does not update its matrix pays for protein it does not need.",{"h2":17},"Do not forget moisture",{"p":19},"Nutrient values are usually held on an as-fed basis, but moisture changes from lot to lot. When corn defined at 14 percent moisture arrives at 15.5 percent, its dry matter drops from 86 to 84.5 percent. Not only protein but energy, fat and starch are diluted in the same proportion. If your NIR measures moisture, correcting the other values for dry matter is the cleanest way. Holding the matrix on a dry matter basis and working out as-fed values from each lot's moisture cuts the error considerably.",{"h2":21},"NIR is fast, but not enough on its own",{"p":23},"NIR lets you sample a truck before it reaches the weighbridge and see estimates of protein, moisture, fat, fibre and starch within minutes. But NIR is an estimate, not a measurement. The instrument relies on a calibration built from reference lab results. If that calibration does not represent your ingredients, your region and this year's harvest well, the results can drift in a systematic way.",{"p":25},"A method that works well in practice is this. Every month, a few samples of each main ingredient are read on the NIR and also sent to the lab for wet chemistry. The average difference between the two is tracked. If the difference stays in the same direction for several months, that is a bias and the calibration needs correcting. Scatter between single samples is normal and no reason to panic.",{"h2":27},"The sample matters more than the analysis",{"p":29},"Even the best lab cannot fix a badly taken sample. A handful of corn from the top of a truck does not represent the load. You need sub-samples taken with a probe from different points of the truck, mixed together, and the analysis sample drawn from that. The lot, supplier and silo each sample belongs to should be recorded as well. Otherwise you have a number, but you do not know what it stands for.",{"h2":31},"Which value goes into the matrix?",{"p":33},"It is tempting to write the latest lot's analysis straight into the matrix, but what is in the silo is usually a blend of several lots. In practice two approaches stand out. The first is to use a weighted average of the last few lots by quantity. The second is to define the ingredient by silo and keep an average for each silo. The second takes more work, but it gives a far more accurate answer for ingredients from different suppliers whose quality differs clearly.",{"p":35},"Variability needs to be considered too. If one of two soybean meals with the same average gives much more scattered results from lot to lot, it is wise to leave a safety margin in formulas that use it. Rather than adding the same margin to every nutrient, set it by the more variable ingredient and the critical nutrients.",{"h2":37},"Getting results into the formula",{"p":39},"The part of analysis that wastes the most time is copying results by hand from one place to another. A value typed into the wrong row can take weeks to notice. In OptiFormula, NIR and lab results can be imported straight into ingredient analyses, and the new values can be reviewed before the formulas are solved again. Seeing the difference between old and new values also makes it easier to tell a real change in quality from a data error.",{"p":41},"The nutrient matrix is the most concrete form of what a mill knows. Feeding it regularly with real analyses is the shortest way to make formulas work in the barn the way they do on paper.",1794234580159]