The finish line

After 5 intense and instructive weeks we finished the project Augmented Prototyping with a satisfied feeling. We created the oscillation Belousov-Zhabotinsky reaction and tried to manipulate it in many different and new ways. We broadened our knowledge about 3D printing and the software needed to create a model. We had a taste of using reaction diffusion in 3D printing and saw it change from an untouchable concept to a physical bicycle saddle. The possibilities are even broader than we expected and we are thrilled to see how 3D-printing and the concept of printing chemical/natural processes will evolve in the future.

We are from different faculties and our specific skills could not always be used to the fullest in this project. We all had to experiment and learn as we went along. This was intense but also created a sense of equality. Most of us were not really familiar with groupwork but we functioned quite well as a group; We learned to exchange information and opinions in an efficient way (even when talking about really intangible concepts), we had a clear division of responsibilities and even if someone was late or could not finish something in time, the other group members were understanding and forgiving. It created a pleasant work environment. The mix of different fields of expertise helped to come up with some unexpected new ideas. Our different personalities complemented eachother in such a way that we had a good mix of time management and proper/deepened research.

We had some struggles too. The theme of the project was quite global and at the start it was hard to choose the path we wanted to take. Because of this it took a while before we were able to make a kickstart. Whenever we did narrow down our focus, we had to simultaneously keep in mind the bigger picture to assure that we were not forgetting important aspects. This was challenging. Also, the software was harder to work with than we thought. We spent quite some lost time on Monolith and during the last days we made lots of extra hours to get a valid and printable STL file.

All in all this was a great learning experience and we are thankful for this opportunity.

– Max, Heleen, Isabelle and Lemin


Science fair

After 5 weeks we succeeded in make a functioning 3D printed bicycle saddle using natural processes. We exhibited our product and research findings at the science fair today, the 1st of November at the faculty of Industrial Design. Below are some pictures of the final product and our stall at the science fair.

3D printing natural process


thumbnail_IMG-20161101-WA0022 thumbnail_IMG-20161101-WA0023












All in all it was a really nice day. We had time to look at the projects and great end results of our fellow students. The visitors were really enthusiastic and curious. The most commonly asked questions were: “ What exactly is the benefit of making a 3D printed product based on a natural process?” and “ How do you get a 3D printed saddle from your reaction?” The benefit of our product is that we let nature make the blueprint for our product. This is less time consuming than regular modeling processes on the computer. Moreover, the essence of this product is formed by nature without much human interference. We only created the boundaries for the natural process, which was in this case the saddle. This principal idea of our project really seemed to come across.

Introduction to Rhino, Nettfab and Meshmixer

First stage

The first stage was to experiment with low bit files. There were a lot of shapes created, like a sphere, a box and finally an easy-shaped-saddle. All of these shapes were created with Rhinoceros. After this section the 3D reaction models from Avizo were transferred to Rhinoceros. In this section it was important to type WRL in the avizo title so that Rhinoceros automatically convert a WRL file to a STL file and generate a mesh. It was also important to downgrade this files, because this made the experimenting process more fast. Finally, two downgraded Avizo files, one for the hard material and one for the soft material, and one Rhinoceros shaped model were loaded in Rhinoceros. This is done by the function ‘import’.

Schermafbeelding 2016-10-27 om 11.44.06The three meshes in a rendered top view.


How to cut out shapes of natural meshes?

This is done with the program Meshmixer from Autodesk. The function Boolean intersection is used in Meshmixer. Meshmixer is used instead of Rhinoceros, because Meshmixer recognizes the patterns as a volume, instead of Rhinoceros. So when two volumes intersect there is created a new closed volume, or a valid mesh/closed double precision polygon mesh. If Rhinoceros does this, there could be a chance that 4 million of triangles need to be flipped by hand and closed again. This would be less secure, and there will be a bigger chance of failure.

Important notice: When the parts were cut out of natural meshes, it was important that the saddle (2 files in this case) was coordinated at the same reference point.

Schermafbeelding 2016-10-31 om 15.30.54The soft part, created with the function Boolean intersection from Meshmixer.


Finetuning with Netfabb

After creating two shapes, the soft part and hard part, were the shapes controlled by Netfabb. Why Netfabb? Netfabb has a simple repair mode. This repair mode finds holes and repairs them easily and automatic. After this section, there is created a valid mesh, which is also a closed double precision polygon mesh.

Schermafbeelding 2016-10-31 om 20.14.40The soft part opened in Netfabb, the red sign shows us there is an error.

Creating the product

When the two parts were repaired, they were transferred to Rhinoceros again. And checked if they were both closed and valid meshes. The soft and hard part were seperated and grouped. Now they could be exported both, by export selected, and send as a STL-file to the Connex.

 Schermafbeelding 2016-10-31 om 20.27.06The final product in Rhinoceros, black part is hard, cyan part is soft.


Zwick testing our prototypes

Today, Monday the 31st of October, we did a last minute test on our prototype cubes (2 x 2 x 2 cm) with a Zwickmachine: a drawbench that tests materials on their tensile strength. By applying a pressure of 500 N in ~16 seconds on each of the test cubes and measuring the compression, we obtained data of the 5 cubes. Then we compared the data in order to draw a conclusion on which material and pattern will suit the saddle the best. The 5 test cubes:

The first test cube was a pink cube made of 100% hard material:

This cube compressed with 0,159 mm.

The second test cube was a blue cube made of 50% hard material:


This cube compressed with 8,045 mm.

The third test cube was a soft cube with a hard spiral pattern in it:

This cube compressed with  0,271 mm.

The fourth test cube was a soft cube with a hard circular pattern in it:

This cube compressed with  0,399 mm.

The fifth test cube was a soft cube with a 50% hard circular pattern in it:IMG_8467

This cube compressed with  9,460 mm.

From least to most compressable:
mm                what
0,159             pink cube 100% hard material
0,271             spiral pattern 100% hard material
0,399             circular pattern 100% hard material
8,045             blue cube 50% hard material
9,460             circular pattern 50% hard material

The results were mostly as we expected and with this information we can conclude that a circular pattern consisting of 50% hard material in a soft cube is best to use for suspension. Sadly, we don’t know how this pattern will work on a big scale (we are afraid it will rip easily).

This experiment was conducted on a short notice and we were unable to use the findings in our final saddle. It did give us an idea about all the variables that have to be taken into account when it comes to pressure distribution. Further research is needed to find the best combination of materials and patterns for optimal suspension. For our final saddle we decided to use a mix of hard and soft material (in a circular pattern) because we know from previous prototypes that this allows suspension to happen and is also strong enough to bear a person.


Week 5: Heading for the science fair

This is the final week of our project and we still have quite some things to do before we are ready for the science fair. Only half way through the project we discovered and really understood all the possiblities of using reaction diffusion in 3D printing. Since we are almost done, our goal is to lead the way for our successors and to also create some visual and scientific material about reaction diffusion in 3D printing to present next Tuesday.BZ Uitgesneden Mooi

Last week we created a lot of video material of the Belousov-Zhabotinsky reaction. One reaction showed really nice symmetrical spirals that also corresponds with the wanted pressure distribution of the saddle. We decided to use this pattern for our final 3D print. It is shown on the right.

Max has been working with Rhino and Meshmixer for the last two weeks and he created a nice saddle shape which we use to cut out the Avizo files. It is a standard racer bike model but we adapted the 2upside so that if someone sits on the saddle and the cones are pushed down, the surface will be approximitely flat. We printed a small sample of the saddle to get a better understanding of our final product (shown on the right). The blue parts are solid and the transparent parts are flexibel. We are really happy with this first print but there are two problems we noticed:
– the model can break easily if two solid parts are connected by only flexibel material
– there is no suspension if the hard material from different cones is not seperated completely by flexibel material.
Our first conclusion is that we need a plate underneath the 3D print to support the weaker parts of the saddle. We spend Tuesday morning building this part and added a construction that can fix the saddle onto a bicycle. We want to solve the second problem by magnifying the spiral patterns so that the width of the cones is always big enough to seperate the layers.thumbnail_IMG_20161027_134924

We printed a second test model of a sphere. It contains the spiral pattern which we talked about earlier, only this time enlarged. We started testing it and we noticed that there again is very little suspension. This is due to the fact that the hard spiral structure consists of one piece which cannot bend. We want to have a functioning saddle next week so we decided to test some different structures/materials to see if anything works better than our current model. The findings can be found in the research folder.

The rest of our work this week consisted of creating visual material for the science fair and a lot of hours in Rhino and Meshmixer.

Pressure distribution

This week we found some really helpful research results about pressure distribution for bikers. It shows us exactly how the pressure zones change when the cyclist in a different posture.

Afbeeldingsresultaat voor pressure point saddle posture

Then we found some different video’s which cover some more details about the comfort of the cyclists. For example the video shown below:

pelvic-bone-diagram-bicycle-saddle-contact-pointsIn this video Josh Cohen (physical therapist) shows where the pressure hits the pelvis area, what the difference is between the male and female pelvis and what effect this has when cycling. The sitbones (1 on the picture) makes contact with the bike saddle. When the cyclist leans forward the pressure is more distributed towards the pubic rami (2). This information made clear to us how to create a comfortable saddle for performance cyclists (men and women).

The pattern we are using consists of cones from hard and flexible material alternately. When this is densily packed, there will be more suspension and when it is less compact there will be little to no suspension. We will use the Belousov-Zhabotinksy reaction in such a way that the densily packed areas are placed at the intersection of the sitbones for a sportive cyclist.

Week 4: Reaching the final product

Week 4 was all about conceptualising and perfecting our product idea. On Monday we decided to continue with the bike saddle and we ditched the chair seating and shoe sole ideas (but maybe in the future..?). In the picture shown below (which is of a chair seating instead of a saddle) you can see our thought process:

Pressure, Reaction, Product


1. 2D pressure distribution


2. Belousov-Zhabotinsky reaction


3. 3D printed bicycle saddle



It starts with the identification of high and low pressure points on the saddle. A lot of research has been done in this field so we decided to use measurements from specialized researchers. The next step is to feed this information to the Belousov-Zhabotinsky reaction. We know from previous weeks that this can be achieved in different ways. Due to time pressure we are forced to use one of the less accurate methods which is controlling the starting point of the reactions by adding solution A in the right proportions. The final shape is then very dependent on nature’s random way of developing a pattern. Hopefully this week we can crack the method a bit more to come up with a shape that fits the pressure distribution well enough.

We film the reaction and create a stack of 2D images that represents the process over time. This information is fed to Avizo and we create a 3D model of the oscillating rings (where the height of the model is the passage of time). Once this has been done, we save it as an STL file and we move on to the next and final step. The pattern has to be measured and scaled to the real size and the the periphery of the saddle has to be cut out. One of the big problems is that Avizo turns the file into a mesh which is really difficult to work with in Rhino. Therefor we are now working with meshmixer software which makes this step a lot easier.thumbnail_IMG_20161026_131959

This week we also printed our first prototype made of two materials. The first thing we learned is the minimal thickness that the Connex3 can print. The blue part of the cilinder is the Belousov-Zhabotinsky reaction printed in hard material. The structure was lost because we scaled it too small. With this experiment we know all the scaling values we have to use for a proper pattern. The transparent part of the cilinder is flexibel material. We noticed that it bends really easily but there is not a lot of supsension when you push it in. We are going to research this further next week.

Avizo and the Connex3

Afbeelding van Lemin ChenThis week we tried to use different software to build a 3D model from 2D images. The program is called Avizo and it is mainly used in hospitals and medical research facilities. It is easy to handle and gives us all the needed parameters to create a smooth 3D surface. The interface of monolith was way harder to work with so not a tear was shed when we said our goodbyes today.


When we filmed the reaction we wanted to amplify the contrast between the two colours of the oscillating reaction. We tried using red and blue paper as background, but this did not give good results. White paper gives a good contrast but the footage has some light pollution because of the environment. In Avizo we found a filter called channel 3 that eliminates most of the light pollution and that gives great contrast between red and blue. By using this filter we have more tolerance during filming and we have one less problem to worry about. In Avizo we start with uploading the set of images and choose a scaling for all the axes. Knowing that a bigger z-direction scaling makes the conical structures narrower, we create the wanted 3D model.We then choose a greyscale value that separates the model into two parts (above and below the given value). Avizo creates two STL files that we send to Rhino. Here we scale it correctly and cut out the wanted saddle shape. We assign different colour and material properties to the different STL files and from that point on, the Connex3 printer does the rest.


Interacting with the Belousov-Zhabotinsky reaction


Today, Thursday the 20th of October, we worked in two groups: Lemin and I worked in the lab and started about 8 reactions and filmed them and Max and Heleen prepared the video footage of the reaction and made 3D-models with it.

We interacted with the reactions in multiple ways:

  • We added more of solution B
  • We added more of solution C
  • We added drops of solution A

When we added more of solution B, we found that the solution in the petridish became very grainy. Because this precipitation made it very difficult to obtain clear footage, we quickly decided to stop the reaction and throw it away.

Adding more of solution C to the reaction slowed the process down a lot. We had to swish the solution for a long time before it turned blue and red and the typical blue rings only started to form after about 10 minutes.

Adding drops of solution A to the reaction formed blue rings in that spot: this is exactly what we had tried to achieve multiple times before. With this information, from now on we can start reactions in every place we want and influence the reaction in such a way that we can make more or less rings appear in the wanted locations. In the pictures below you can see very clearly where we added drops of solution A to the reaction.


Tuesday the 18th we also did some small experiments. For instance, we wanted to see if the depth of the petridish had any influence on the reaction. We thermoformed a ‘dish’ with multiple bumps with different depths and started a reaction in it. The results were not of any significance, as the solution became too dark to see through and too chaotic. In addition to this, the acid started to bite through the plastic of the dish. In the picture below it is visible that this experiment was not successfull.


In the table below you can see our other findings:

What did we change? What kind of pattern forms? Why does it work or not? Are we going to use this?
Warmth Too difficult to realize No
Light The difference in warmth has little to no effect on the speed of reaction. No
Touching By touching it one time spirals are formed. By making a line the reaction becomes red without any blue circles. Too difficult to control No
Depth of dish Not visible Can not be filmed as no light travels through the dish. No
Add indicator  The added indicator forms a red drop. It turns blue and red respectively and slowly takes over the underlying pattern. Too much contrast between indicator and underlying pattern. No
Add solution A An oscillating blue pattern starts at the spot A was added. We can decide where a reaction starts. Yes
Add solution B The existing solution is separated by a green/yellow liquid that sticks to the petri dish The reaction stops. No
Adding coloured reaction (with indicator) to see-through reaction (without indicator) The oscillating pattern slowly spreads from the red dot. The contrast is worse than what is normally the case. A small part of the indicator is added to a reaction that has already started. This is already mixed and thus reacts faster with the see-through liquid. Maybe


Belousov-Zhabotinsky reaction

The Belousov-Zhabotinsky reaction is a so called oscillating diffusing reaction. Summarized there are three processes: A, B and C. In process A bromate turns into bromide, in process B bromide turns back into bromate and in process C the products from the reaction will react with each other. The critical concentration of bromide causes the oscillating pattern as it fluctuates from process A to process B. This reaction is made visible by using the ferroin indicator, which turns red when it is in its reduced form in process A and turns blue when it is oxidised in process B as shown in the left picture.



We use 5 solutions to create the reaction:

Solution A: 2 ml sulphuric acid and 5g sodium bromate (NaBr03) in 67 ml water                   Solution B: 1g sodium bromide (NaBr) in 10 ml water                                                             Solution C: 1g malonic acid in 10 ml water                                                                             Solution D: 1 ml ferroin (25 mM phenanthroline ferrous sulphate)                                         Solution E: 1g Triton X-100 (a kind of detergent) in 1 litre of water


We put 6 ml of solution A into a petri dish, add 1-2 ml of solution B and 1 ml of solution C. The solution turns a brownish colour. After a minute or so the brown colour will disappear. Once this has happened, we add 1 ml of solution D and a drop of solution E and the liquid tIMG_7880urns red. We swirl the petri dish gently to mix the solutions. It will turn blue and then quickly reverts to red again. Gradually, blue spots will appear randomly and the reaction will start. Below there is a video of our first experiment. In real time the reaction lasts about 45 minutes as it evolves chaotically, in the video it has been speed up 8 times.