Success story

How to improve processes and products with Design of Experiments (DoE)

If you want to improve a product or process, you can of course rely on your gut feeling. However, that is usually not a good idea, because no matter how much experience you have, the real-life situation is usually quite different from that gut feeling. Design of Experiments (DoE) is meant to set up an experiment to systematically determine which factors influence the process. Marc Schrijvers and Michel Werts will introduce you to the basic principles of DoE in a workshop. They tell us already a little bit about what to expect.

Marc Schrijvers and Michel Werts

design-of-experiments_2024-05-28-124737.jpg

If you want a production process to run optimally, it is necessary to know which factors influence the process. You can do a lot of experiments, but that takes a lot of time and money. Thanks to Design of Experiments (DoE), it is possible to establish the relationship between independent variables (factors that you can influence) and dependent variables (outcomes of processes) as reliably as possible, with as few experiments as possible.

As few experiments as possible

Senior Process Engineer Michel Werts explains further. "Let's take a simple example. Imagine that you are going to bake bread at home. There are many factors that determine the quality of your bread: the composition of the dough, the amount of salt or water, the time the bread is in the oven, the oven temperature... The tricky part is that you don't know in advance which factor has which effect. In addition, factors can influence each other. You can look at them one by one, but then you must make sure that the other factors remain constant. Moreover, you have to do a lot of different experiments to find out the effect of each factor."

This is where DoE or Design or Experiments comes into play. DoE is a smart method to statistically validate and analyze combinations of factors. You do this in the most efficient way possible with as few experiments as possible. You can apply it to all kinds of production processes. Think, for example, of a printing process, an adhesive process or a mechanical grinding process.

Ideally, you conduct the experiments before the production process has started, but it can also be done in an ongoing process. For example, to increase returns or if things have changed after a product has been launched: a different raw material or new equipment.

Minimal variation

Marc Schrijvers is a Senior Process Development Engineer. He explains how DoE contributes to the quality of a production process. "Suppose a product has to be 5 centimeters long but may vary between 4 and 6 centimeters. With DoE you can then determine exactly which values you need to set the factors to get a size of 5 centimeters with as little variation as possible. " In practice, no two products are one hundred percent alike. You are always dealing with a slight variation, but you want the variation to be as small as possible and not to deviate too much from a predetermined value.

Screening

Before getting started with DoE, it is important to investigate which factors are relevant and which are not. "You may not need a lot of factors and interactions between factors at all," Michel says. "At many companies, the process owners already know a lot. They have often already tried everything necessary. But often the list of factors is still too long. You then carry out a screening. There are very useful mathematical methods to eliminate those factors. After the screening, you are left with a limited number of factors, usually between two and five, which you will investigate further. In the bread example the color of the bread is not determined by the ingredients, but mainly by the oven temperature and the baking time. For these two factors you will then determine in detail how large the effects are, whether these effects are linear, and what the optimal value is to get exactly the right color.

Teamwork

No one conducts a DoE experiment on their own. Marc: "For example, specific knowledge is needed to determine which factors you are going to investigate in the first place. A process specialist or process owner can tell you more about this. When conducting an experiment, I also work together with operators who set up and operate the machine. When I am going to measure products, I may need someone to help me measure the quality of the products. There is a complete team around it. However, I often set up a DoE and perform statistically."

A DoE is preceded by thorough preparation. Marc: "You have to think very carefully about how you are going to implement it. Therefore, it sometimes initially seems as if nothing is happening. This is something I know from my own experience. But once I got started I was done much faster and didn't have to do six or seven more experiments. That's the power of DoE."

Six Sigma Green Belt

If you want to be able to apply DoE properly, it is important to work systematically and accurately, says Marc. "You don't have to be a process specialist, but it is useful if you can quickly understand a process. I became acquainted with DoE about thirty years ago through a theoretical course. I had to improve a process at a company and didn't know what to do. Since then I am very enthusiastic about it and think it should be used a lot more."

DoE is offered as part of the Lean Six Sigma Green Belt training. "That sufficient to be able to implement it. If you want more information and details, more theory is recommended."

Happy with unexpected results

Despite their years of experience, Marc and Michel are still equally enthusiastic about every experiment. Michel: "The great thing is that you can demonstrate very precisely what exactly plays a role based on input factors. If we go back to the bread: do you have the oven at exactly a certain temperature with the right amount of water, salt, flour and yeast, then you get exactly the bread you want with at most a little bit of variation, that has something magical."

Such an analysis is always an exciting moment, says Marc: "You always have a gut feeling beforehand about which factors have an influence and which do not. I like it best when at the end of the analysis the outcome appears a bit different than what I expected. I was wrong, but that's okay. I didn't do the experiment for nothing."

Hands-on workshop

What can participants expect from the workshop? Michel: "Marc and I have extensive experience with DoE, so we can tell a lot about it. We will certainly do that, but the workshop will mainly be hands-on. Baking bread will take too much time, but we're going to do something practical. An experiment where you have all kinds of different variables that can play a role in the outcome. At the end of the evening, we're going to have a competition to see who has optimized their process the best. Of course we have a prize for the winner."

Related vacancies Kick-start your career
design-of-experiments_2024-05-28-124737.jpg
Stories from our happy members

Meet our people and find out what they do, how their careers develop, and what inspires them.

Let's get in touch
Let's get in touch!

Reach out for opportunities, collaborations, or questions. We're here to connect.