Better fat measurement provides better margins
Last autumn, Rendalen Kjøtt in Østerdalen used an NIR scanner to measure the fat content of pieces of meat.
MeatAutoSort is a user-controlled innovation project owned by Nortura SA. Active Norwegian participants are Nofima, Sintef ICT, Animalia and QVision. The project is financed by the Research Council of Norway, Nortura and QVision.
Foreign countries involved are VION (Netherlands), FACCSA (Spain) and Marel (Iceland).
This has given them totally new opportunities to check and control the proportion of fat in production. Savings come in the form of better utilisation of raw materials, better meat quality and more efficient production.
“The tests we have done have been positive, and when the system is correctly calibrated, there will be a great deal to be gained from using it,” says factory manager Ola Kværnes of Rendalen Kjøtt.
Fat content controls prices
All over the world, the price of meat is determined by fat content – the more fat, the lower the price. At the same time, many processing companies now choose to buy in pieces of meat and do the mincing themselves. This increases the need for being able to make an exact measurement of the fat percentage, even when the meat is in large joints.
A specially developed NIR scanner that is placed on the conveyor belt makes it possible to measure the fat content of the meat pieces in batches of up to 200 kg while they go past on the conveyor. This gives far better control, because the people cutting the meat get immediate feedback of whether the fat percentage is as it should be. If the fat content is too low, they add in fatter pieces and vice versa until the batch has the desired fat content.
“At Rendalen Kjøtt, they have connected up a monitor that continuously shows the fat content in the batch in production,” explains senior researcher Jens Petter Wold of Nofima. He has helped to develop the system, together with Sintef ICT and QVision.
Traditional measurement methods are much more cumbersome. They need the meat to be minced first and the fat content adjusted afterwards. If the customer wants the meat in whole pieces, a safety margin has to be included, which means the meat contains more fat and the price is lower.
Finding the calibration algorithms
The algorithms that are needed in order to estimate the exact fat content in a whole piece of meat have been developed by researchers at Nofima. Some of the main challenges in this work are connected with the heterogeneity of the meat, height differences in the pieces, temperature variations, the effect of light scatter in the surroundings and the need for constant measurement.
Nofima’s research team is also working on developing algorithms for other products.
“One example, which makes use of the possibilities of both imaging and depth measurement, is being used by Hitramat AS to measure the meat content of living crabs while they whizz past on the conveyor belt,” says Wold.
Another example is algorithms that scan herring so as to separate those that have roe or milt. Both roe and milt contain sought after health components, so demand for these raw materials is expected to increase.
Savings running into millions
NIR scanning of the newly delivered pieces of meat makes it much easier to optimise production, so that the fat content in the batches is precisely within defined limits. That means better utilisation of raw materials and more efficient production. The earlier in the process such sorting can be done, the better.
Buying in pieces of meat rather than minced meat has other advantages too.
“I see no problem in selling whole pieces of meat to the customers. Even though it means more work for them, there are clear advantages. The meat lasts longer, there is less waste and the risk of spreading bacteria is reduced. At the same time, there are savings for us because we can avoid the actual mincing process,” says factory manager Kværnes.
The next step is further selection
At the moment, the meat cutters are able to measure the fat content in intact pieces of meat and adjust fat content as they go. But the researchers at Nofima, equipment manufacturer QVision and Rendalen Kjøtt want to go a step further; the aim is automatic selection based on fat content.