Evaluation of Using Hand Held Near Infrared Spectroscopy for Pellet Fines and Moisture

Recently the use of Near Infra-red Reflectance spectroscopy (NIR) to predict nutritional information has grown. The ability to predict chemical information around protein, oil, and carbohydrates in real time has enabled nutritionists to make decisions around feeding values and to better predict animal performance. What is generally overlooked is that NIR can be used to predict physical as well as chemical properties. In addition new NIR equipment particularly at the portable end of the market has been introduced which allows the user to get closer to the production line and to make decisions whilst the product is still in process. One such physical attribute, pellet fines, is well established as resulting in poorer FCR especially if feeder adjustment is not closely controlled leading to an increased cost of pig production. The aim of the study was to determine if using a hand held (HH) NIR instrument it is possible to determine the percent pellet fines as measured using a standard sieve methodology. This trial was set-up to measure pellet fines on 89 swine feed samples using the 2 pan sieve method and developing a calibration to determine if it is feasible to measure particle size by HH-NIR. In addition 29 samples were sent for wet chemistry analyses for moisture; this data was used to determine if the level of moisture can be predicted by NIR. Using a HH-NIR instrument (wavelength between 920nm to 1650nm) approximately 1.2 to 2.0 kg of feed sample was placed in a specialized box and scanned at 5 different locations collecting 200 spectra per scan. The resulting 1000 spectra were averaged to produce 1 spectrum per sample. The same sample was then measured by being placed on to the sieve (#6) with the pan attached to the bottom .The sieve was then shaken for 45 seconds and repeated for 15 seconds until the sample was completely sieved. Pellets and fines were then weighed and the % fines determined. Chemometric analyses showed that there was a strong calibration between pellet fines analysis and the NIR spectra (r² = 0.93, SEP 3.1). Furthermore using a standard feed calibration provided by Aunir for this HH-NIR it showed that there is a good prediction of pellet moisture against the laboratory values (r² = 0.93, SEP 0.708). Overall, these results confirm that HH-NIR can be used at the mill to determine pellet fines and pellet moisture.