Every time I dive into an additive manufactured (AM) design project, the outcome of that design (the part) is always a bit of a surprise. I expect most of you have experienced something similar.
In my Design for No Assembly blog post, I identified the following four additive manufacturing production level process categories and then dug into an example of #2 the final part, or “end-use” (defined in Basics Of Design: Quality STL Files for Laser Sintering):
- Functional prototype
- Final part
Let’s keep in mind that a 3D printed part is the only product definition method that requires a 3D model. Model Based Definition (MBD) is first and foremost for additive manufactured parts.
I made up an example to illustrate the Design for No-Assembly concept using MBD and then my friends at GrabCAD printed it. In general, I am happy with the resulting part. However, the flexibility always surprises me because from day one I have been a designer of metal parts. So when I make components from plastic, I’m either too thick or too thin.
I also get annoyed with the surface finish of an FDM print, but because this part only moves air, the surface finish is not relevant to the design form, fit, or function. I just like to have pretty parts. Post-processing options are always available. I could sand it and/or paint it, but, again, the surface finish is not important. So, as a good engineer, I let it go because there is no value-add.
However, the geometry features that were created as a solid loop as cable mounting tie-downs came out hollow. Very strange, and I will need to dig more into the printing pre-processing to understand why this phenomenon occurred. As designers, these are details of using 3D printing that we need to understand. Design for manufacture has not gone away.
Manufacturing the air duct
Summary of how this air-duct was built from the design to manufacture process: designed, defined, and 3D printed.
1. Creating the Geometry
- I used SolidWorks 2014 to create the geometry.
- Then I used DimXpert to add the appropriate annotations. Because a 3D printed part will never need a presentation-based (human consumed) fully annotated model, I simply identified an overall profile tolerance and the mount holes for the fan interface and the air-duct mount tab holes. All other geometry is adequately tolerance using the overall profile tolerance.
2. Creating the Technical Data Package
- I generated a 3D PDF with the MBD, native SolidWorks file, and STL file into a technical data package (TDP). I could then pass the TDP to GrabCAD to print the part with all the pieces of the puzzle tightly and completely contained in the 3D PDF container.
3. Design to Manufacturing Hand-off
- So my friend Blake at GrabCAD pulls in the STL, and shoots me back an email…
Hey Jennifer, what about this goofy geometry I identified when I opened the STL file you sent me. Politely suggesting that the STL I generated from SolidWorks was riddled with a few geometric imperfections.
The following email discussion ensued.
From: Jennifer Herron
Subject: Print me a part?
To: Brian Neville-O'Neill
Created a 3DPDF that holds an STL and STEP model. I’m not an expert on STL files, so I made it fairly fine and thus, a very large file. I am sure you have someone there that can improve upon the STL as necessary.
Here is the dropbox link to the 3DPDF: https://www.dropbox.com/s/l7ifrltd027u81f/80065.pdf?dl=0
Headed out on vacation, let me know what you might need in addition by tomorrow – or wait 2 weeks.
Blake Courter wrote:
There are some geometry problems with the 3DPDF, although I don't know whether they occurred on export or import. A very large and fine STL would work better. Or the native CAD part.
Blake Courter wrote:
Could you please try sending the native part?
Blake Courter wrote:
Okay, I saw your note in the rev block and opened the PDF up in Acrobat Pro. There I saw the attachments.
How do normal people get the attachments? i.e. People who don't pay for Acrobat?
Blake Courter wrote:
The STEP and SLDPRT open without issue. I tessellated it myself from the SLDPRT and am printing it now.
Oh, the tangled web of communication we weave. Even though I am an excellent communicator and a 3D product definition, MBD, and MBE expert, Blake and I still tangled up a bit.
It's no surprise that current research trends show that the designer to manufacture communication loop is wrought with errors and significant time spent. Understanding and identifying these common inefficiencies, and applying value-added changes to your design to manufacture hand-over process is likely to achieve significant cost and quality savings for your products.
4. Setting-up the 3D Print
The work setup was straightforward and utilized default properties of the 3D printing build setup software.
This screenshot shows a view of the air duct part build direction and orientation, which was incidentally not identified in the 3D TDP and left to the manufacturer. Leaving off build direction and orientation is an oversight that needs correcting via a 3D product definition standard, such as the emerging ASME Y14.46 Digital Product Definition for AM.
This screenshot shows a cross section of the contours, infill, and supports that the dimension software automatically creates to build the part. Support and infill definition are also a topic of major interest in the field of product definition for AM parts.
Check it out live being printed.
5. Support Removal
Once printed, the part was put it in an alkali bath that dissolves the support structure while leaving the ABS (dark green plastic). The part sits in the alkali bath for a few hours until all the support material is removed. Finally, the resulting green part is soaked in two different water baths to rinse off the alkali.
The part is now complete.
Is this what happens with most 3D prints today? I don’t know, but I expect it's a fairly common scenario. Using this simple case study, there are clearly some time saving opportunities in the process that could be turned into real dollar savings.
Also, the final step of verification has not been complete. Verification is the requirement to dimensionally verify that the as-built part meets the as-design definition.
Additionally, I could have avoided the goofy fillet geometry translation errors by completing a validation step during the TDP creation process. In fact, I broke one of my cardinal MBE implementation rules. Bad Jennifer! Well, always room from improvement.