<<<< go back
This week our assignment is about designing and making a composite part that will be moulded. This means that also the mould has to be made. The dimension should not be bigger then a square foot. In this case I will make an additional part for the gripper that was moulded in week 12. The function of this additional part is giving the stability to the frame that is taken out of the hive by the H-bot. In the picture below the position for the composite part is shown with a red circle.
In week 12 the gripper was moulded in an elastic material. The position for the gripper is shown in the picture below. Only the gripper in the middle is not stable enough. The additional composite part in the shape of a bough with flaps on the side will stabilize the beehive frame during transport.
Below picture shows the interaction between the parts and the beehive frame.
View from above.
View from the side.
Composite part 3d above.
Composite part 3d below.
The part design was made in Solidworks and also the mould was modelled in Solidworks with the special mould tool. Normally this tool works good when the parting line is not too complicated. In this case the parting line at the end of the product goes down steeply. This usually causes trouble with modelling as I know it in Solidworks. With some smart modelling with surfaces this can be solved, this often is not easy. But in this case I discovered a new button in the feature. By selecting the manually button is suddenly is very possible to make complex mould split surfaces.
The mould tool menu starts with a draft check. The top plane is the direction of the mould. Red surfaces can be made in the direction to the ground and the green surfaces can be made in the direction of the sky. The blue line indicated in the picture below shows the parting line. In the central section of the part the parting line is rather smooth horizontally. Towards the ends it is going down steep.
From this blue parting line the tool now needs to create a nice parting surface, in the direction perpandiculat to the direction of the mould. Starting from the parting line. This is now much easier by selecting the Manual mode as shown in the picture below.
Below picture shows the creation of the parting surface starting from the parting line.
Actually now three main surfaces are created. First the green side surface from the part. Second the red side surface from the part and third the gray parting surface. All these surfaces should meet each other exactly on the parting line. Solidworks is allergic to leaks in surfaces. Than it will not work.
With these three main surfaces it is possible to split a main piece of solid mould material into three bodies. The upper side of the mould, The Lower side of the mould and the part it self.
The blue part is the part that will be moulded. The purple part is the upper mould and the yellow part is the lower mould.
The two main mould parts need some more features which will be explained in the next section. In the picture below the differences are shown globally.
Starting with the upper part of the mould. The material will be MDF 18 mm thick and to find out how the material will be oriented to the shape The mold half was cut into sections with a surface. These are 18 mm slices.
Now, because the material MDF needs to be coated to get a nice closed surface I decide to create a clearence of 0.4 mm between the mould halves.
Four alignment holes were added as shown below.
Below quite similar actions were taken to create the other part of the mould.
The lowest plate was made bigger to be able to clamp the mould better to the machine bed.
Two holes and gates were added, indicated in blue. The proces will be VARTM if it will be possible. One hole is for applying the vacuurm and one hole is for pressuring the material.
Alignment holes were added. Four pins indicated in dark blue.
About a year ago I was attending an introduction presentation of Solidworks 2015 and there a new feature was discussed. The name of this feature is "Surface flatten" and this feature unfolds a double curved surface. This is done by making a mesh of the surface and it will distribute the stress as even as possible when it creates a flat surface. This way it is possible to unfold for instance a piece of leather an the nose of a shoe. In this case I can use this feature to unfold the curved surface. Here for I start with 1/4th of the model. I deleted the fillets just to make it a little easier. It should also be possible with fillets.
From the bottom outside of the model I create a surface copy. This is the green surface in the picture below.
Below the Surface Flatten feature is shown.
The purple surface is the Flattened surface. As you can see in the picture below the flattened surface is mirrored two times. There is a gap between the 4 flat surfaces. This can be altered manually by creating a new surface.
The orange extrude shows the new adjusted surface on top.
When this surface is mirrored two times there is a closed top surface. This surface can be cut on a lasercutter by exporting a dxf-file from the top surface..
To determine the next surfaces it is necessary to determine the layer thickness first. This is done by..... and is ... mm thick.
An offset of this cross section to the inside can be calculated in excel....... With these dimensions xx new layers can be created. In the middle the bending goes inward, so the dimensions get smaller. And at each end the bending goes outward, so the dimensions get bigger.
I am very curious how accurate Solidworks is and how the behaviour of the material will be. By doing a test first I will try to determine the progress and accuracy.
The picture below shows how the dimensions for a cross section on each layer can be determined.
The milling was done on the Shopbot, in this case the 3D software was used. This software works on stl-files. The MDF material was prepared by glueing and screwing layers of 18 mm thick on top of each other. With additional pieces of wood the block was attached to the table. Also with screws to prevent it from coming lose when milling.
To prepare the settings in the software the tools need to be selected. One tool for rough cutting in this case a 1/4 inch end mill. And one fine tool for the finishing, in this case a 3 mm end mill initially. This one broke and was replaced with a 6 mm end mill.
The rough cut went really well and fast as you can see in the video below. The fine cut however had more problems. The rough cut did not make the holes, so the finish cut had to make the whole hole. This was too much for the fine mill so it broke. Also the shaft of the 3mm mill is 4 mm and it was already touching the straight MDF walls when milling. This probably damaged the mill as well.
The video below shows the milling of the mould.
After the rough milling from the first mould halve the fine milling started with a small diameter mill (diameter 3 mm and shaft 4 mm). This mill broke in the first hole because the rough milling did not pre mill the hole. This because the diameter of the hole is smaller than the diameter of the rough mill.
Just to make the documentation complete you will find screen dumps from the settings that were used in the milling software. For both mould halves the settings were the same.
The rough cut went really well and fast about half an hour per mould half. The fine milling takes much more time, more than 3 hour per mould. But the result of the milling is very smooth.
The Shopbot had a lot of errors when milling the first mould. The spindle stopped turning many times. We now probably know why this is happening. The dust collector is generating static electricity. This might give input signals to for instance the emergency stop pin. This we found out because we had several electric shocks when we were working on the Shopbot. At the second mould the dust collector was not used and there were no problems in the milling process. The solution might be grounding the Shopbot frame and parts better.
While milling the fabric was cut on the Trotec laser cutter. Each ply of fabric is different because it has an offset to the inside.
The settings that were used are for 3 mm Balsa wood. The power is on 100%? and the speed on 4.5 m/s?
Each layer is marked with a position code. The green dots shows what layer the ply of fabric is.
While making the unfold in Solidworks I already had doubt about the shape of the flattened fabric it created. The picture below shows that the fabric does not match to the shape of the mould. Probably the mesh of the flatten feature in Solidworks was not fine enough to be accurate. For some reason I could not make a fine mesh.
It was decided to alter the shape of the fabric manually.
The wire thickness of the jute fabric is rather big. This is good for infusion, but it is quite rough towards the fine details of this product.
The whole VARTM idea was also learned from this video. Here you will see it is possible to have a very long distance for the resin to go.
The milling from the second mould half left some material that should be billed away. The reason for this is not clear to me for this moment. Initially the mould could not close because of this issue.
By sawing this problem was solved.
The inside of the mould was painted with an MDF primer manually. Befor paiting it was sanded just a little by hand.
Ok, the idea here is to apply pressure to the liquid resin material on one side and pull with a manual created vacuum on the other side.
The resin enters the mould on the right side and is pushed / pulled through the mould to the far left side.
This vacuum pump is normally used for whine bottles to conserve an opened bottle of wine. For different reasons, which will be explained below, this did not work as is was supposed to.
The white surface of the mould was coated 3 times with a release agent from smooth on. This was not a good idea. It sticks to the resin and the connection between the mdf primer and the mdf is weaker.
The layers of jute fabric are assembled into the mould dry.
With silicon kit a gasket or seal was made between the mould halves. This to make the vacuum possible. Stupid as I was I did not take enough drying time for the silicon. This also because I was in a hurry to finish all in time.
Mould including jute fabric, release agent and silicon seal was closed with clamps.
Resin was mixed and prepared. It took about 15 minutes to inject 2/3 of the tube. This should be more the enough, however the silicon seal was not closed and much resin came out of the mould.
The vacuum system did not work and I decided to use the vacuum cleaner. After some more minutes the 1/3rd af material in the tube became very hot and solid, so I had to stop filling the mould.
The material did not arrive at the other side of the mould, which is a bad sign, but now nothing can be done but waiting for the resin to set.
Waiting for a few hours....
To prevent the resin from squeezing out painters tape was used to close the outside of the mould.
The resin set also between the two mould halves. They were coated with the Smoothon release agent but this was not working well. The mould halves were just glued together by the resin.
With wood carving tools the mould was released with caution in about 15 minutes. I was trying to save the mould of course....
Saving the mould did not work....
Part was filled with resin for about 40% maximum.
Rest of the fabric was completely dry.
This part is quite nice. Some air bubbles are clearly visible on the outside. The white paint indicates that the release agent did not work.
So, the result after all this work is quite disappointing. On for hand I knew that choosing for the VARTM- process was more risk, but I expected a better result of it. At least I expected to be able to reuse the mould. Now I need to make a new mould again. With the hand layup method, assembling pre wetted plies of fabric in the mould and then close the mould, I expect a much better result.
Unfortunately the Light weight construction department in our University (Saxion) did not want to share their vacuum pump. If they were cooperative the result would be much better....
....to be continued......
So now I decide to make a new mould and to mill it out of one block.
From Github: Bough files. Click "view raw" to download.
Here are the files for downloading.