As well as manufacturing advanced machines and software for Additive Manufacturing, Renishaw runs its own Dental services business focused on the production of crowns and bridges.
We spoke with David Turner, Software Development Manager, about how his team leverages Polygonica for more reliable automation of their dental production.
Where does Additive Manufacturing fit into the digital dentistry workflow?
“Traditionally custom crowns copings and bridges have been manufactured using lost wax casting and more recently CNC milling. CNC milling helps to bring dental manufacturing into the digital workflow but to mill small features requires correspondingly small tools. These suffer from push-off and if undercut features are required, more complex CAM programs can be required. This process is less efficient for complex frameworks although it is possible combine the two processes to manufacture complex geometry with the accuracy of milling where required.”
What problems do you encounter with mesh integrity?
“We commonly receive meshes with gaps, self-intersections, voids, islands and incorrectly facing polygons."
Small holes and gaps
For example a mesh with a big hole, such as in the image on the left, will actually go through the printer and it will print and it will come out the other side and look like a complete part. The problem is it is like an eggshell with each skin being the thickness of the melt pool. When someone takes it off the plate then hopefully, it cracks, and it breaks. If it doesn’t it could be dressed and end up going to the patient – where it cracks and breaks. Which is a lot worse. These parts absolutely have to be fixed.”
“The image on the right shows the same model that has gone through Polygonica and all the holes have been removed from the part – it’s now guaranteed to be watertight.”
Non-manifold bodies
“On the left we have a good example of a coping with a non-manifold body. This causes problems all over the place, especially when you have bridges. Bridges are where a number of crowns have been effectively splinted together by connectors and a pontic, a floating part which replaces a missing tooth. Instead of going for an implant which is pretty extreme dentists can use this more wallet friendly option.”
“We normally see the problem where they created the connectors. The CAD system has tried to splice the two bodies together, it’s done a bad job of it and its put triangles where it really shouldn’t. Often those end up in the cavity. “
“If we make it like that it won’t fit in the patient’s mouth. You create a part, it goes out to the end user, the dentist tries to fit it and it doesn’t fit. Wasted surgery time and added costs for all parties, which we really don’t want. Getting it right first time is good for everybody…. and business“
“In this example, Polygonica has just ripped out the non-manifold triangles and created a proper solid body.”
Floating Shells and Voids
“Another common problem is floating shells – multiple shells within a body. This bridge, to all intents and purposes, if you looked at it in a viewer application, would look fine. The problem is when you look inside where the connector is there is a second body.”
“This is a terrible scenario because it potentially creates a void in that connector, makes it understrength and it could crack in the patient’s mouth. In this case the internal shell is clearly visible, but often they are very small and not obvious. From Polygonica’s point of view we call one function and this is fixed.”
Incorrect orientation of Polygons
“Last, and probably least, are incorrectly oriented triangles.”
“You can see on the left we have a triangle that is wound in the wrong direction and it’s fixed by Polygonica on the right. This typically doesn’t cause huge problems but it can be an issue.”
How does Renishaw make use of Polygonica to tackle complex meshes?
“The amount of mesh detail is really key to us because we’ve got tens of thousands of parts coming in monthly. That takes up a lot of server space plus the amount of bandwidth it takes to shift these parts around our manufacturing system is huge. Also, the amount of time to process those parts and create laser paths that we can actually then build the parts with goes up massively.”
“Polygonica helps us reduce the amount of data in the mesh whilst guaranteeing not to lose important detail.”
“Another thing to also factor in is that not only does this process remove very small triangles, it helps to remove valid but very thin triangles with acute angles in them, which potentially cause downstream problems for processing algorithms.”
“The other point worth touching upon is that Polygonica’s algorithm for simplifying a body doesn’t have to work on a percentage basis but can work on a distance tolerance of the simplified body with respect to the original body. You specify a valid real world tolerance and Polygonica guarantees the result is within that tolerance. Areas of high rate of curvature change and high detail are still maintained and you only lose things in the very flat areas, which is really useful.”
How does Polygonica help with product identification and tracking?
“When we’ve got these hundreds of parts, thousands of parts, coming in every day, they all look very similar. We need to be able to identify them. So as you can see in the image, we actually create a physical tag on the part. “
“We take the identifier that comes out of our database, encode it to base 36, put that onto a tag using the Polygonica deboss functionality to embed the text within the tag, orient it onto the part and create two legs that fit onto the body. We then union this tag onto the part, again using the Polygonica Boolean engine, ending up with a single closed body that contains the identifier.”
“The key thing here is to recognise the marginal area. If we were to place those legs anywhere near the marginal area the part would be rejected by the customer. If they have to dress it back they may dress it back too far and bacteria can get in, it’s an error to the patient.”
“The previous solution we had before we had Polygonica would do all this stuff for us automatically. However it wasn’t margin line aware although it made a pretty good guess of where the margin lines were.”
“The problem was it got it wrong in probably around 5% of cases. We had to have someone sitting there, day in, day out, checking every single part manually. With Polygonica we just do all of this automatically now.”
In fact the whole stream of fixing, checking parts and ensuring supports were in the correct location was taking someone three hours every single day. That’s now completely automated.”
You’ve written your own nesting algorithms. How did Polygonica help with that?
“Yes, this was a quite cool thing. The body in the image represents a plate that will go in the machine that we would build from, and we’ve taken a good number of parts and nested them into that area.”
“Our nesting algorithm is not trying to optimise the usage of the plate. The powder is going to be reused so we’re not losing material that way. We’re just trying to ensure we get enough parts on the plate to meet our daily build time for that machine and that we’re achieving certain aims; namely that we segregate out the units with single supports from those with multiple supports.”
“One of the things to recognise is that with parts that come off a metal AM printer you have internal stresses. You are taking a laser beam, melting that area, putting another layer of powder on top, which is cooler than the area you have just melted. You get stress building up.”
“Once printed we shear off the parts that don’t need to be heat treated whilst the other parts go straight into the furnace.”
“To achieve this custom nesting we use a number of Polygonica functions: the silhouette function gives us a 2D profile of the part on the print tray. We break that down into a convex partitionand then use Polygonica’s 2D Boolean algorithm to union the results into containment bodies and collision bodies – and from that we nest the whole plate.”
How else does Polygonica support your mesh modelling requirements?
“When creating abutments we need to use CNC machining to accurately create the interface to the corresponding implants. The image on the left shows this modelled in a CAD system but we need to create a model of the initial stock material prior to machining and union that to the part. Again Polygonica provides all the functions we need to do that.”
“If we put the original model into our printer it would come out the correct size. The issue is that in a metal printer you can get spatter and you get material printed on the outside of the part that you really don’t want. To remove that we use an electro-chemical polishing process
“We provide a spherical offset of the whole model using Polygonica to ensure that after the hour or so of electro-chemical polishing it comes out exactly as it needs to.”
We basically turn the whole part into an anode and dip it into a very acidic electrolyte for a couple of hours which removes all of the defect from the surface.”
Anything else you found to do with Polygonica?
“Well, last but definitely not least is hub placement. To CNC machine these parts we place them on a kinematic locator body. We try to fit as many parts around it as possible in known locations. However, they potentially clash. We have symmetry by which we can machine those parts and depending on the body we have six to eight rotations we can put them through.”
“We don’t want these things to clash as that would completely kill the part so we use Polygonica for looking at potential clashes as we place these parts on the hub.”
After using Polygonica for your dental workflow did you use it on any other projects?
“Yes. Polygonica has been invaluable in the development of our ADEPT craniomaxillofacial design tool for 3D printing. This new software constructs a 3D model of a patient’s skull from CT scans and then uses the Polygonica engine to help the surgeon model implants and plan operations. You can find an overview on youtube.
For more resources from Renishaw about Additive Manufacturing please visit their handy Guide to Additive Manufacturing.
