On-line estimation of dry matter and fat content in production blocks of cheese by NIR spectroscopy

Modern dairy factories produce thousands of cheese blocks per day. We evaluated NIR spectroscopy for on-line determination of fat and dry matter in such cheeses. The motivation is true time control of the production process.

Dairy production involves a range of complex processes where quality of the end product depends on raw material variation as well as process settings. Achieving the target end quality is important for profit and to prevent waste and low grade products. In order to control such processes it is crucial to measure the relevant quality properties in real-time during processing. This gives the opportunity to detect deviations and adjust the process.

A modern dairy factory can produce thousands of cheese blocks every day, but there are so far no tools for on-line determination of chemical composition of such blocks.

On line determination

In this study we evaluated NIR spectroscopy for on-line determination of fat and dry matter in cheese blocks of approximate size 40 cm ×30 cm ×12 cm. Three different modes of sampling were tested: Scanning reflection, scanning interaction and imaging interaction measurements. The NIR measurements were collected in a pilot plant at three different production stages: 1) On fresh cheese block before pressing, 2) after pressing and 3) after salting.

A total of 160 cheeses from 10 different production batches were measured. Fat and dry matter content were determined in the local lab. Partial least squares regression (PLSR) was used to make calibrations between NIR and chemistry.

Quality properties

In dairy production, one of the most important products is large blocks of cheese. Of the most important end quality properties are fat and dry matter content. During cheese making both the raw material (the milk) and the different processing steps affect this final chemical composition. The amount of dry matter is also decisive for the profitability of the process, so it is important to control the process towards target quality.


With NIR scanning reflection and interaction instruments, it was possible to determine fat and dry matter in cheese blocks with an accuracy of about ± 0.53% and ±0.63%, respectively. This is comparable with what you can achieve with a lab NIR instrument on homogenized cheese. This means that the surface chemistry of the blocks was representative for the average chemistry of the blocks.

We also observed that it is possible to predict fat and dry matter in pressed and salted cheeses based on NIR measurements on non-pressed cheese earlier in the process. This is possible as long as the pressing and salting process is the same every day.

Improved control

The fact that NIR spectroscopy can determine fat and dry matter in cheese blocks at an early process stage, enables improved control of the cheese making process. Quality deviations can be detected, meaning that the process can be adjusted and cheeses outside specifications can be used for other products.

Industry 4.0

There are on-the-shelf NIR instruments that can be used for this application, so it is possible for dairy companies to implement this solution today. The use of such process analytical technology is expected to increase as food companies are now moving toward Industry 4.0 standards.

By: Jens Petter Wold