Headquartered in Viby, Denmark, Arla Foods is the fifth-largest dairy company in the world, and remains a cooperative owned by more than 12,500 dairy farmers. Manufacturing milk and dairy product brands that are household names in 100 countries worldwide, Arla Foods is committed to developing responsible, resource-efficient and environmentally focused processes for creating healthy, nutritious dairy products.
Sensor technology monitors fluid properties during homogenization
In late 2024, GEA Area Sales Manager and homogenization technology specialist Preben Kristensen knocked on Arla’s Innovation Center door in Aarhus and introduced the Arla team to GEA NiSoMate® Quality Analyzer. This innovative, user-friendly sensor technology provides real-time insights into fluid properties and performance – without the need for direct sampling – that can then be used to fine tune the homogenization process in real time. When installed into a homogenizer plant, GEA NiSoMate® can provide critical insights that can translate directly into improved process reproducibility and stability, and potentially significant savings in energy use, emissions and waste.
GEA and Arla have a longstanding partnership, and the innovation center has previously purchased a GEA TriplexPanda laboratory homogenizer. This plug-and-play, trolley-mounted unit developed to support for basic R&D, process evaluation, new product development and small-scale production.
Ultrasound-based technology measures milk composition in real time
Homogenization is a key process that creates smooth-textured pasteurized milk with a uniform fat distribution. But the interplay of a whole host of factors, from the fat content of the milk to temperature and pressure, can all impact on process reproducibility. NiSoMate® comprises an array of dedicated sensors that accurately and continuously measure the composition of the milk and then pass the sensor-derived data to the homogenizer’s process control system. The control system can then make real-time adjustments in process parameters automatically to maintain product consistency and quality.
Talking technically, the sensor system employs an ultrasound-based “beamforming array” technology, developed by GEA homogenization experts, which precisely measures the fluid’s physical properties, such as density, changes in consistency and dilution. Employing this ultrasonic strategy means that no part of the probe comes into contact with the product, reducing contamination risk and making cleaning and maintenance much easier.
Improving reproducibility, while reducing waste and energy use
When installed in a production setting, NiSoMate® generates key data for process optimization, stability and reproducibility, which can ultimately cut waste, flag up when processing may be heading out of specification, and potentially help reduce energy use that can result from unnecessary over-processing.
Since its launch in 2024, GEA customers in the food, beverage and dairy sectors have been using the NiSoMate® in both production and R&D settings to help better understand fluid behavior, optimize existing processes and product consistency, troubleshoot, and develop new formulations and recipes.
«Used in a research environment, NiSoMate® provides vital data on product behavior and give users confidence in absolute batch consistency, with no quality loss due to process variation.»
Preben Kristensen
Area Sales Manager at GEA
Supporting ongoing R&D
Arla scientists introduced to the NiSoMate® were quick to appreciate the potential benefits of using the system to support ongoing R&D and product development. Importantly, the NiSoMate® technology was a perfect fit with the Innovation Center’s existing GEA TriplexPanda homogenizer.
GEA offered Arla a one-year try-before-buy agreement and the NiSoMate® was installed on the latter’s TriplexPanda unit. The configuration allows Arla to monitor fluids being delivered to the homogenizer, and fluids exiting the homogenizer after processing. Both sets of data are informative with respect to understanding and identifying more precisely the most appropriate homogenizer settings, alongside investigating how different milk properties impact on the process.
«Sitting in an innovation center we are curious about a lot of new things, and Preben already had a great collaboration with our facility, because we have other GEA equipment here,» noted Arla Senior Process Developer Kristina Schmidt Bejder. «GEA saw the potential opportunity for us to use the NiSoMate® technology within the R&D setting and offered us the try-before-you-buy collaboration.»
Digging into process data
The setup gave the Innovation Center scientists the freedom to dig into process data that will support optimizing the reproducibility, stability and efficiency of producing homogenized milk, Kristina noted. «Once the NiSoMate® was installed we could design a range of trials to reflect and mimic some of the variations in milk that we might see during homogenization in a production environment. We can then take samples at different points within the process and analyse the different milk components, for example, fat, protein, dry matter, and particle size, and then correlate what we have measured using our reference methods with the NiSoMate® signal output.»
Importantly, NiSoMate® can be installed to monitor the milk as it arrives for homogenization, as well as monitor the homogenized milk post process. The properties of milk arriving can be used to inform on the optimum homogenizer settings in advance of the process, but similarly, understanding the characteristics of the homogenized milk post process can also inform when the process settings may need to be adjusted. All this data can help better ensure a stable and reproducible process and output product quality.
Helping to ensure reproducible, efficient homogenization
By setting up a range of trials and tests Arla is working to assess ‘where’ in its production process the NiSoMate® may provide the most insightful data to help ensure homogenization is reproducible, efficient and stable. The technology can also be used to better understand the homogenization process itself. «We’ve used the system to monitor the milk pre and post homogenization, because we can see the potential benefits of knowing exactly what goes in and measuring what comes out,» Kristina added. There are a huge number of variables, Kristina noted, even when milk is sourced from the same farm and processed in the same way. «Not only do cows produce different milk in the winter and the summer, but the milk may be left standing in tanks for variable amount of time, and factors such as temperature can all impact on the homogenization process.»
Arla has already seen promising results with its tests using the NiSoMate®. «We haven't yet expanded our tests to assess what would be the full complexity of a production site, that will come in future. But so far, we are confident that it's a very valid technology, with some great potential. We need to continue our assessments to identify the level of sensitivity we need and can achieve, so we know the best places in production to apply the system. The next steps will include further trials just to introduce some more variations, and then probably investigate in more detail specifically what happens at production sites and then trigger some of the changes in the matrix that may be seen as an issue in a production setting.»
«We’ve used the system to monitor the milk pre and post homogenization − so far, we are confident that it's a very valid technology, with great potential»
Kristina Schmidt Bejder
Senior Process Developer at Arla Foods
Contextualizing sensor data and physical analyses
Ultimately the scientists will need to be able to understand the NiSoMate® readings in context with the results of their physical analyses and data from the homogenizer itself. The sensor data can then more accurately be used to understand fluid properties, composition and behavior, and to fine tune the homogenizer settings in real time. «We need to make sure that we can also filter out any unnecessary noise,» Kristina continued. «We need to correlate data from the physical analyses with the NiSoMate® sensor output, and the homogenizer settings.»
The NiSoMate® technology itself is inherently flexible and can be installed into a new homogenizer plant or retrofitted into an existing setup, whatever the capacity. The system also features a user-friendly human machine interface (HMI) that makes it easy to get started and familiar with the system right from initial installation.
And in industrial settings, as well as helping to ensure more reproducible, stable homogenization, NiSoMate® can also help reduce energy use by allowing optimization of parameters such as homogenization pressure. Over-homogenization can be relatively common in industrial processes, leading to greater power consumption than is actually needed, as well as causing more wear and tear on the material itself. By monitoring fluid characteristics and enabling process adjustment in real time, NiSoMate® can help cut unnecessary energy use.
Collaboration and training
Leonardo Maggi, Digital Product Manager at GEA Homogenizers, further explained the importance of supporting Arla scientists with ongoing training and insight as they started to use the system. «Our partnership with Arla is really collaborative, » Leonardo explained. «We visited the Arla Innovation Center to get to know the scientists and better understand their needs, expectations and setup prior to configuring the system for their try-out period. We also arranged online meetings for training, and in mid-2025 welcomed the Arla team to the GEA site in Parma for additional in-person training.»
The follow-on training in Parma gave Arla scientists a great boost in confidence and additional understanding of how to use the system to its best capacity, Kristina suggested. «Everything was new, and after some hands-on experience there may be areas that are a bit more complicated or need more focus. It was great to have a second round of training, and guidance, where we could ask the next level of nerdy questions, which the GEA team was able to answer. This is the first time collaborating on a project like this. Everyone has been very open and honest, and the GEA team has really gone the extra mile to help us out.»
After a successful period of tests and trials, towards the end of 2025 Arla purchased the NiSoMate® unit, together with a bespoke 9-sensor array on top of the standard 5-sensor arrays. Arla’s aim is to continue to use the system at the innovation center to generate insight into fluid properties and behaviors during homogenization, understand how to improve existing process reproducibility and consistency, and support product and process development and evaluation.«Our partnership with Arla is really collaborative. We visited their Innovation Center to better understand their needs, and then welcomed the Arla team to Parma for in-person training.»
Leonardo Maggi
Digital Product Manager at GEA Homogenizers
Compatibility with production settings
In the longer term, the goal may be to investigate how best to exploit the system in more than one production setting, Kristina noted. «Preben came in at a very early stage of our future strategy – focused on improving efficiency, reproducibility and sustainability – which was completely fine because he read my mind!» Importantly for Arla, evaluating the NiSoMate® in an R&D setting should make it easier to migrate the system into a production setting. «We’ve already evaluated compatibility with our existing Arla IT infrastructure and are confident that we won’t need a lot of work in terms of network configuration that might otherwise be tedious and time consuming,» Kristina noted. Meanwhile, GEA continues to develop software updates to the NiSoMate® system and will roll these out to Arla as they become available. «Such updates mean that the system can now be calibrated in an even smarter way, and it allows a more immediate comparison between inlet and outlet measurements», Preben noted. «We continue to update and optimize creation of reference curves relating to homogenization efficiency and will introduce additional features as part of our ongoing support to Arla.»
