The Proactive Use Of Global H Outliers
Please contact Caltest by Email for free, finished copies of this article.
Home
The Proactive Use Of Global ‘H’
You will encounter, from time to time, products that are borderline-high Global ‘H’ producers.
A sample producing a Global ‘H’ value greater than 2.99 cannot be used in any of the Caltest Certification Procedures.
The results of a sample producing a Global ‘H’ value greater than 2.99 should be suspect, and cannot be trusted to be accurate enough to make value judgments on the batch, or load of product it represents.
When a sample produces a high Global ‘H’ on a Caltest Certified Calibration and System, there can be a number of possible reasons after laboratory cross-contamination, improper sample preparation, and using the wrong calibration have been ruled out:
1. There is a contaminant in the product, or load, or in the sample, that may have been placed there intentionally, or unintentionally, at the point of sampling.
2. The sample that arrived at the laboratory was mislabeled at the point of sampling and the sample does not represent the product for which it was intended to represent.
3. A wrong formula was used to make the product in production.
4. A bad, or wrong load of ingredient is about to, or has been, brought into the plant, or storage facility.
5. Ingredients added to a production formula were miss-weighed and/or the wrong ratios were used.
6. Ingredient storage is cross-contaminated, and/or the ingredient has degraded while in storage.
7. A new formula with new ingredients, or different ratios of the same ingredients has been authorized.
8. Natural growing, geographical and climatic conditions have altered the product significantly to not allow it to be recognized by the calibration model.
If one or all of the first six conditions have been found to be true, then the system you are using to monitor your production, or incoming loads has worked as designed. You have to decide whether you want to receive the ingredient, or load, rework the product, destroy it, or release it for consumption.
Numbers (7) and (8) above present a little more work.
When a calibration is developed for a particular product, that calibration should be as sensitive as possible to the minute changes that occur in that product due to changes in formulation and production. The calibration should not be too sensitive to the changes encountered in sample handling and preparation.
If a calibration is not sensitive to minute changes in a product, it would not be a good indicator of when the product, it’s formulation, it’s production, or it’s sample handling procedures have changed. This striving for sensitivity can be a burden when the product or sample handling is intentionally changed due to new formulation models, operators and equipment.
In the early days of NIR, an intentional, or unintentional change in a product, or ingredient, was difficult to detect by an NIR operator, until the constituents of the product being tested by the NIR reached noticeable, out-of-spec levels. This made the sensitivity of the NIR calibrations used in these early systems too low, and detection of production or ingredient problems impossible until it was too late. There were many instances where a change in production or sample handling significantly altered the level of, say protein in a product, but the alteration would not have been noticed by the operator of the NIR until the protein would have reached a level that would have set off the operator’s own internal alarm. Different, and new operators all had different alarms. Reliable production and sample handling controls using NIR were hampered.
In 1983, ISI marketed it’s first version of a software package that planned to improve the sensitivity of NIR calibrations. Using a concept called Global ‘H’, derived from a sample’s spectra, ISI was able to increase the sensitivity of it’s NIR calibrations where, for the first time, the NIR could be used as a valid production control tool. There was now a indicator that could trigger an alarm whenever the slightest change in the product was detected either through a formulation change, a production change, storage or growing condition change, or a sample handling change. Unfortunately, intentional changes in formulation, production an/or sample handling would also trigger the Global ‘H’ alarm. It became important to first identify the change that triggered the alarm as intentional, and then to desensitize the calibration to that change.
1. After determining that the high Global ‘H’ is due to an intentional change in formulation, production, and/or sample handling, the operator of the NIR must prepare a blind-duplicate sample ("repeat") of the product producing the high Global H, using the ‘Outliers - How To Treat Them’ instructions in this documentation.
2. If you cannot determine whether the high Global ‘H’ is due to an intentional or unintentional change, it would be prudent and safe to continue anyway with the ‘Outliers - How To Treat Them’ instructions in this documentation. Caltest does reserve the right to not expand an existing calibration with outlier samples representing changes that cannot be determined.
3. When sufficient numbers of high Global ‘H’, and/or ‘T’ Value samples have been reported in the enclosed Outlier Record, (usually 10% of the sample population of the calibration to be expanded), Caltest will expand the calibration as per terms ordered and described in your quotation.
In the past, feed mill managers, hay brokers, hay re-sellers and feed users, in an effort to maximize profits, saw advantages of purchasing ingredients from their suppliers only on the basis of price. After consulting with their nutritionists, the ingredients were purchased and formulas adjusted to allow for the new nutritional characteristics of the new cheaper ingredient. The only concerns then for the substitution would have been the effects on the plant’s production equipment and time, and any nutritional effects on the animals for which the feed was intended to be used. When NIR came on the scene, a rapid method of analysis became available. Today, consideration must also be given to the effects the new ingredients will have on the NIR’s accuracy.
Failure to consider whether an NIR calibration can handle new loads of ingredients and new formulations can result in lost time and money due to excessive wet laboratory analyses and costs. Prior to going into production with the new ingredient, or buying a new season’s crop, a pilot run of the new crop and/or formulation should be made on the NIR. The determination that the rapid analysis method can accurately analyze the new crop, ingredient, or product formulation must be completed before going into routine NIR analysis if you want to be proactive with your NIR. If a pilot run cannot be made due to time constraints, there is no choice but to use the first loads, or production batches containing the new ingredients as the pilot run. This gamble pays off only if the new load, ingredient, or formulation does not trigger the Global ‘H’ or ‘T’ Value alarms. If outliers are flagged, not being ready, or proactive with your NIR, and the excessive time and costs of wet laboratory analyses will have a compounding effect on your stress level and bottom line.
Please Email Caltest for a free copy of this article. Request document "Proactive use of Global 'H' Outliers.
Copyright Caltest, LLC 1999 All rights reserved.