Cooling, the hidden engine of product performance

Cooling is one of the most complex and underestimated technologies in product development. While heating is relatively straightforward, removing heat efficiently, affordably and sustainably requires deep scientific insight and system-wide thinking.

Success lies not in isolated choices but in orchestrating every element; from selecting the right cooling technology (vapor compression, thermoelectric, passive) to balancing temperature control, component choices and environmental impact. When viewed as a whole, you turn cooling into an enabler of performance, precision and user delight.

This system-level approach has delivered successful products across a variety of industries, including home appliances like ice-cream makers or beer taps, wearable air-conditioning and high-precision medical devices. Developed in close partnership with forward-thinking companies like AB InBev, Nestlé, Coca-Cola and imec, more than 80.000 functional devices now demonstrate what’s possible when cooling is seen as a strategic asset.

New demands on innovation: comfort, precision, sustainability

Consumers and industries alike are setting higher expectations: greater comfort, higher precision and more responsibility toward the planet. Cooling technology sits at the heart of meeting these ambitions, whether it’s through creating energy-efficient appliances, improving portability or reducing environmental impact.

The challenge lies in finding the optimal solution for each specific context. Cooling is never a one-size-fits-all solution. Each innovation journey requires carefully managing trade-offs: choosing between vapor compression or Peltier cooling, selecting the most suitable compressor, and designing an effective thermal management strategy. Aligning these technical choices with broader product and business goals is key to creating meaningful and lasting innovation. As a vital part of a product’s DNA, cooling reveals its true potential.

Driving cooling innovation through system thinking

Breakthrough innovation rarely comes from focusing on individual parts. It happens when you see the whole picture; when technology, design and market understanding come together in harmony.

Cooling is no exception. A holistic, system-wide approach allows businesses to transform complex ideas into scalable, manufacturable products that meet real-world needs. This mindset opens the door to new product families, greener solutions and entirely new market segments.

Ultimately, true innovation isn’t just about controlling temperature; it’s about pushing boundaries, solving problems that matter, and shaping the future through better products.

The post Beyond temperature control: How cooling shapes the future of product innovation appeared first on Verhaert Masters in Innovation.

The post Beyond temperature control: How cooling shapes the future of product innovation appeared first on Verhaert Masters in Innovation.

However, the process isn’t easy and requires a serious and dedicated approach with accredited laboratories and notifying bodies. In addition to analyzing benefits, risks, strengths and weaknesses, we can lead your emissions filter project through each phase and with multiple international partners, focusing on the overall product objectives set.

Emission regulations in FMCG

Safety & perception

Emission management cannot be taken lightly. In countries around the world emission limits are rising as awareness of health and safety is rising. Regulatory limits for both work and public spaces are increasingly more strict, and make no mistake thinking western countries set the trend! It could become an expensive exercise if you think you can to tackle countries outside of Europe in a later stage. Even for in-home units or products that have no regulatory limits but do emit odorous compounds, a screening test campaign is advised. The wrong combination of odorous compounds could mean the difference between the main act or the side-kick of the show.

Emission groups

Air pollution emissions can be divided into different groups depending on their characteristics. Each type has a different impact on health and pollution. A good understanding of the various compound groups is needed to analyze the root cause of an emitted compound, and subsequently filter the emissions.

Volatile organic compounds
Volatile Organic Compounds (VOC) are organic gases that exist in the gas phase at room temperature. Most components that have an odour are VOCs, yet not all VOCs are odorous. Some have a negative health impact, while others are polluting the environment. A lot of VOCs are biologically generated by plants, animals or micro-organisms.

The most important, human-generated, sources of these VOCs are solvents (paint, ink, etc.), bio- and fossil fuels and biomass combustion. Total Volatile Organic Compounds (TVOC) is a metric often used to describe the general VOC performance of a process or device. A selection of the VOCs interact with the measurement device and give an indication of the general presence of VOCs.

TVOC is expressed as an equivalent of a certain compound such as propane or isobutylene. Since not all VOCs are taken into account with TVOC measurements, a verification on compound-level must always be made when interacting with compound-level regulations. It is, however, a good metric to make relative comparisons and performance assessments, and is itself a regulated value in certain countries.

Inorganic compounds
In general, inorganic compounds are odourless. Examples of volatile inorganic compounds include NO, NO₂, ozone, SO₂, CO, CO₂, etc. Some of these compounds impact health and environment and are thus regulated.

Particulate matter
Particulate matter (PM) consists of microscopic solid or liquid particles suspended in the air. Particulate matter can be found in various forms such as smoke, pollen, smog, dust, vapour and – topically – bacteria and viruses. Depending on the size of the particles, they can be harmful when inhaled. Small particles have the ability to penetrate deep into the lungs and bloodstream, and can cause respiratory diseases, cancer and heart attacks. Aside from entering the lungs of living creatures, particulate matter suspended in the air can create extremely hazardous situations if the particulate matter is flammable. A mixture of air and oil droplets or fine wood dust only needs a minor spark to create a huge explosion.

PM is categorized in different size levels, indicated by their maximum diameter in µm. PM10 is known as coarse particles and consists of all particles with a diameter lower than 10 µm. PM2.5 is fine dust and PM0.1 ultrafine dust. Note that PM10 includes both fine and ultrafine dust.

Strict regulations exist on PM10 and PM2.5 and PM levels are continuously monitored, especially in urban or industrial locations. When particulate matter limits are exceeded, smog alarms often go into effect and people are encouraged to stay indoors, wear masks and vehicle speed limits are activated to reduce fine dust emissions. The European Air Quality Index includes PM10 and PM2.5 to assess the air quality throughout the EU and can be consulted online in real-time.

Download the perspective to continue reading about the regulations, emission analysis & know-how of air pollutants.

Looking for solutions to innovate?

Leave us your email and get in contact with Paul Poelmans, Business Development Manager Solutions ‘Smart FMCG & Consumer’, to help you with your innovation process.

The post Regulations on emissions & filter technology in FMCG appeared first on Verhaert Masters in Innovation.

The post Regulations on emissions & filter technology in FMCG appeared first on Verhaert Masters in Innovation.

Monitoring air quality

Factory workers are exposed to several sources of work-related physical stress factors: loud and intense noise, heat and air pollution. Some of them, such as noise and heat can be easily measured and predicted. Other sources of pollution, like Volatile Organic Components, cannot be readily sensed and predicted. Nevertheless, they present a real danger for possible health problems.

The physical phenomenon of diffusion is theoretically well understood. Mathematically, it’s described by an equation known as Fick’s law:

Fick’s law

Where:

• c is the concentration mol/m³ (or equivalent units)
• t is time [s]
• D is the diffusion coefficient [m²/s]
• x is the spatial dimension [m]
• s is the source term [mol/m³s]

The equation describes the transport of mass through diffusive means. It describes that there will be a solute flux along the concentration gradient. The solute will move from a position with high concentration to a neighboring position with lower concentration. The speed of this process is defined by the diffusion coefficient D. Using this equation, it’s possible to calculate the concentration in every position at any moment in time, assuming you know the starting condition, diffusion coefficient, source terms and boundary conditions. However, the latter is harder than it sounds.

In the context of modeling the air concentration of volatile chemical compounds in a chemical factory, the source terms are the chemical processes and production machines. Accurately measuring the emissions of each machine is possible, but it would require shutting down the complete factory and performing a high number of repetitive experiments and measurements for each machine operating under each possible setting. This would be a very costly procedure, both due to time consumption for the tests and the opportunity cost of stopping production.

Verhaert Masters in Innovation developed a tool and procedure to solve this problem based on real-time data from air quality monitoring called Smart Spot, provided by HOPU. The HOPU SmartSpot device, equipped with an ION Sensor MiniPID for volatile organic components, allows you to gather a large dataset of observational data. From this dataset, the unknown parameters can be inferred.

This ‘IoT – Big Data’ approach permits to gather data for the production of a large number of materials. The advantages?

• It combines monitoring with prediction: permanent temperature, noise and air pollution measurement allow to evaluate the current situation, independent from predictions.
• Data can be gathered during the normal operation of the factory.
• When the product portfolio changes, the new parameters can be learned by the system without further interference.

Using IoT & Big Data analysis methods to build a diffusion physics model

Discretization of Fick’s law

Fick’s law can be discretized. This allows to numerically solve the diffusion problem. This can be discretized on a rectangular m × n grid:

Discretization of Fick’s law

• Where ℭ = {(i +/- 1, j),(i, j +/- 1)} is the set neighboring nodes to node (i,j);
• A is the diffusion surface between nodes k and (i,j);
• L is the distance between these nodes;
• g is a possible adjustment for the diffusion coefficient to take into account walls, doors and screens. Diffusion is not possible through a wall, the diffusion coefficient between nodes k and (i, j) at either side of a wall is there for 0.

This results in a matrix multiplication model. In this work, this model could be restricted to the steady-state case:

Basic matrix multiplication model of the discretized Fick’s law (steady-state case)

Steady-state case solution

The result is c, a vector containing p = m.n-p concentrations that need to be determined. This is calculated based on the influence matrix G, a sparse p × p matrix containing the influence terms.

The influence of node i on node j is given by:

Influence term of node i on node j

This term is 0 for all pairs (i,j) that do not share a diffusion border. The diagonal elements (i,i) of the influence matrix contain the sum of all the terms of influence on node i.

Sum of influence terms on node i

This influence matrix is combined with:

• s, the sparse vector containing the source terms
• The sink terms, listed in the vector f and their influence terms.

This is the result of a calculation with input values:

• Square map: 20 m × 20 m on a grid of 20 × 20 nodes
• 1 wall (g=0) in the middle, 15 m long
• 1 source of value 1 mol/m³s on position (2 m,5 m)
• 1 sink of value 0 mol/m³ on position (2 m,15 m)
• Diffusion coefficient D=0.1 s/m² (in a problem like this, the steady-state result is not influenced by the value of D).

The example below shows the steady-state IoT solution for the given (hypothetical) problem. The concentration value tag color has no meaning. It changes from black to white merely to optimize readability.

Result of the steady-state solution of Fick’s Law

Model calibration

Having an IoT tool to solve the diffusion problem with known source and sink terms is not a complete solution to our problem! The source and sink variables are unknown. In the context of modeling the air concentration of volatile chemical compounds in a chemical factory, these terms represent the amount of airborne chemicals produced by the chemical processes and production machines. These terms could be determined from individually testing each machine in controlled conditions.

For the IoT4Industry ‘Equality project’, we developed a proof of concept of a technique to infer these parameters from distributed observational measurements.

First, we need to be able to optimize a single diffusion map. Based on n measurements, the variables of a diffusion map can be estimated by minimizing the Mean Sample Error (MSE):

Mean Squared Error loss term

The MSE expresses the difference between the calculated concentrations c(θ) and the measured concentrations c. By minimizing this term concerning the unknown parameter, these unknown parameters can be estimated. Usually, it’s not possible to find a solution for θ without residual error. In the illustration below, such a case is illustrated. Using the same geometry as in the previous example, we suppose to have two measurements, not that far from the source.

For the sake of the example, the source term is supposed to be a known parameter in this example. The unknown variable is the diffusion coefficient D. If a value of 2.7 mol/m3 is measured in two different points such as in this problem, there is no perfect analytical solution to this problem. The solution with minimized MSE is displayed on the right: the measured values of 2.7 mol/m3
are approximated with [2.6 ; 2.8]mol/m3 by estimating the diffusion coefficient to be 1.42 m²/s.

Best fit solution

Performing this optimization for a complete factory with a large product portfolio and complex production schedule is another challenge. A factory doesn’t produce the same product on the same production line every day. The challenge is to infer the contribution of each independent process from measurements of the combinations of these processes.

In practical settings, IoT sensor data isn’t always available for each sensing location at every moment. Sensors can break, but it’s also possible that sensors are rotated between measurement locations to increase covered location by the study while keeping the investment in individual sensors low. Consider the following simplified example on a (hypothetical) factory geometry as below:

Table example case

Hypothetical factory floor plan

Assume that in this factory, production line 1 and 2 are identical production units. When the same product is produced from the same material, these machines can be assumed to exhaust the same pollution levels.

Knowing the problem geometry, there are 3 unknown parameters in this problem:

• The production line exhaust during the production of recycled PVC;
• The line exhaust during the production of PE;
• The diffusion factor D.

This problem can be solved by minimizing the squared error for the complete factory problem consisting of the combination of S situations:

Loss function over all situations

For the example discussed above, the result of this optimization is shown in the following figures:

Approximation for situation 1

Approximation for situation 2

The developed model is applicable for a wide range of two-dimensional diffusion models, whether it is diffusion in air or liquids or other problems such as heat diffusion.

Application development

As production plant, Lisanplast has been the testing environment, a plastic extrusion company. Smart Spots were installed in different zones of the plant to collect the IoT data. These data are analyzed through a software platform based on Grafana, an Open Source technology that allows us to analyze all the data to obtain conclusions and export them to other tools and to create reports.

Verhaert’s AILab developed the app ‘Emission Predict Equality’ allowing the production manager of Lisanplast to estimate the zone-specific air contamination in his factory. So it’s possible to adapt the tasks and areas of work for a pregnant and nursing woman.

The final app is easy to use, the screen shows a simplified illustration of the Lisanplast floor plan where light green squares indicate the production line positions. Based on the production planning for this week, one can select the product, the type of plastic that will be processed for each of the 13 production lines, and the material mixer. The image below shows that on a typical day, the exposure limit of 50 µg/m³ PM10 is not expected to be breached. The contours indicate that around the extrusion heads of the production line one can expect that the PM10 level will reach 10% of this value. This zone is safe, but one might choose to avoid unnecessary presence in this zone.

Screenshot of the developed application

Acknowledgment

The development of the calculator prototype and the approximation method was made possible thanks to the funding of IoT4Industry.

Watch the movie

Download the perspective

Looking for solutions to innovate?

Leave us your email and get in contact with Lieven Claeys, Business Development Manager Services ‘Digital Innovation’, to help you with your innovation process.

The post IoT data for air quality mapping appeared first on Verhaert Masters in Innovation.

The post IoT data for air quality mapping appeared first on Verhaert Masters in Innovation.

The last decade we’ve seen these boundaries degrade to the point that CFD tools are applicable to everyday product design. The improvement of user interfaces, integration with CAD tools and automation of various modeling tasks have brought down the entry costs. There’s still a steep learning curve, however CFD has become an invaluable tool for more and more industries.

Let’s take a look at an example in the Fast Moving Consumer Goods business. Maybe you’re familiar with at-home beer dispensers or coffee vending machines with a multitude of products. Most of these appliances contain some kind of cooling system to cool down at least some of the contained products.

Home beer dispenser development

These appliances are governed by requirements of low cost, minimal material usage, space allocation constraints and ever more stringent power requirements. Squeezing out the last bit of performance has become a non-trivial task which can no longer be done with empirical formulas and hand calculations only.

Generic coolbox design (left) and CFD simulation of internal air circulation (right)

Here we show a generic coolbox design with internal air circulation which must cool down and keep cool 2 cardboard BIB containers. Hand calculations are fine to determine steady state conditions, but CFD can be used for so much more.

• Optimizing the placement of BIBs, air inlets and outlets for optimal airflow and cooling performance.
• Assessing the effect of cooling strategy on cool down time.
• Assessing the ideal control loop feedback for sensor locations.
• Calculating power requirements across various operating regimes and environmental conditions.
• Assessing the product temperature uniformity and prevention of product freezing conditions.
• Localizing cold spots which may cause condensation problems.

Building a single concept breadboard takes weeks and requires hundreds of hours designing, manufacturing and assembling. A CFD model however can provide answers in days. While a single cool down test takes hours, it can be simulated in minutes once a model has been created.

As such, CFD speeds up development cycles and delivers insights which are hard to obtain with empirical formulas and testing only.

Interested in finding out more on how Verhaert can apply simulations to your product designs or engineering problems? Get in touch.

Download the perspective

Looking for solutions to innovate?

Leave us your email and get in contact with Max van der Weyden, Manager Innovation Acceleration Services – Product Innovation, to help you with your innovation process.

The post Transient conjugate heat transfer simulation of a cooling solution appeared first on Verhaert Masters in Innovation.

The post Transient conjugate heat transfer simulation of a cooling solution appeared first on Verhaert Masters in Innovation.