Investigative Paper - Accelerate Medical Devices Prototyping and Go-To-Market in a Crisis
By Hicham Farid, PhD
CAE/FEA Engineer at Aventec
During the COVID-19 pandemic, the need for certain medical equipment such as ventilators, masks, and gloves have outpaced the supply chain capabilities due to the overwhelming demand. Creative solutions have emerged to help the health workers during this fight; however, the challenge is to provide up-to-standard components and equipment while accelerating the production volume without compromising quality.
Additive Manufacturing (AM) is the technology that everyone is turning to because of its fast-paced production and ability to help manufacturing complex designs with high prototyping and production volumes. Still with this kind of technology, engineers and scientists need to understand how their products behave during and right after the additive manufacturing process is completed, thus the need to perform simulations come into play.
Let us dive into a little bit more technical description of the problem. From a classical manufacturing process standpoint, material is usually removed from the areas of interest to obtain the final products. Then, the structural and thermal integrity of the part or assembly is analyzed using FEA for validation purposes. However, during the additive manufacturing process, material is continuously added to the design space, which changes the problem definition altogether; the heated material that has just been added generates thermal stresses and that intern translates into mechanical stresses, leading to distortions in the final product. Predicting these local distortions help get the 3D printed prototype right the first time without wasting time or material.
I used SIMULIA Abaqus to model the additive manufacturing process of a ventilator splitter design shown is Figure 1. The simulation helps analyze design solutions before production in order to reduce the cost of prototyping and get reliable devices to patients and hospitals in a record time.
Figure 1: Pipe Splitter design
Since material extrusion 3D printing technology or Fused Filament Fabrication (FFF) is widespread among the engineering community due to its easy implementation and cost effectiveness, we chose to model this process using Abaqus/CAE combined with the AM plugin.
First, lets have few words about the FFF process. Material Extrusion 3D printing technology uses a continuous filament of a thermoplastic material as a base material. The filament is fed from a coil and through a moving heated printer extruder head, often abbreviated as an extruder. The molten material is forced out of the extruder's nozzle and is deposited first onto a 3D printing platform, which can be heated for extra adhesion. Once the first layer is completed, the extruder and the platform are parted away in one step, and the second layer can then be directly deposited onto the growing workpiece. The extruder head is moved under computer control. One layer is deposited on top of a previous layer until the object’s fabrication is complete.
Figure 2: Fused Filament Fabrication process
Additive manufacturing is becoming a more mainstream technology that is getting easier to implement in Abaqus using a combination of Python scripts, Fortran subroutines, and the AM Plugin. Let’s dive into the how:
Two separate models need to be prepared in Abaqus, one transient heat transfer model, and one general static. I used ULTEM™ 9085 (also known as PEI, or Polyetherimide) material, which is a versatile high-performance thermoplastic used in 3D printing which offers excellent mechanical strength and temperature resistance.
Preprocessing of the two models are fully prepared on Abaqus/CAE, including the material properties and sections assignments, step definition, mesh (Figure 3), boundary conditions and predefined fields. The transient thermal analysis time period was chosen based on the time series generated during the part slicing. I will discuss this in more details in the sections to follow.
Figure 3: Pump Splitter mesh
Abaqus/CAE Plug-in Utility for Additive Manufacturing Process Simulation
The AM plug-in (or AM Modeler) can be used to set up the thermal and mechanical models for the part-level sequential simulation of various AM processes, including SLM (Selective Laser Melting), LDED (Laser Direct Energy Deposition), and FFF (Fused Filament Fabrication). As mentioned above, the latter process will be simulated using the AM Modeler.
The plugin is available on knowledgebase article QA Article QA00000057533, and it is compatible with Abaqus R2018x FP.CFA.1838 and upwards. I used Abaqus R2019 HF3 (FP.CFA.1923) for this simulation. It is worth mentioning that a nice amount of information is also available on SIMULIA AM microsite:
Readers can also download the plugin directly from the above-mentioned website.
The QA article also contains a PDF file that gives an overview of the AM Modeler, however I’d like to give a more detailed instruction to perform the AM simulation here.
Extracting the STL file and generating the G-Code
The AM Modeler does not generate tool path data. The user needs to obtain the tool path data from other sources such as the 3DEXPERIENCE DELMIA Powder Bed Fabrication app or an open source 3D printing program like ReplicatorG, and convert it into the event series file format. I used ReplicatorG to generate my event series for the Pipe Splitter simulation.
Starting by generating the .stl file, as we don’t always have access to the part .stl file, I used the STL Export tool from the tool menu bar (Figure 4). The .stl file can be generated at the Part level or the Assembly level.
Figure 4: STL generation using Abaqus/CAE
The exported .stl file is used to generate the G-Code using ReplicatorG as mentioned (Figure 5).
Figure 5: ReplicatorG interface
Tip: Pay attention to the part orientation as the printing direction is z-axis.
ReplicatorG is open source and can be downloaded at:
Note that you may need to update your python to 3.8 for this.
Once the G-Code is been generated, I used the python script generateEventSeries.py that is attached in the Knowledge Base article QA00000057533. The script will convert the G-Code data into two Abaqus input files ready to use with the AM Modeler. The input files are written in a generic format of time, space (X, Y, Z), and amplitude.
Using the Plugin
After setting up the two thermal and mechanical simulations within Abaqus/CAE, generate the event series for both Power and Material. Now it’s time to go over the additive manufacturing process simulation set up using the AM Modeler. Once installed, the AM modeler will show up in the Plug-ins drop down menu bar (Figure 6).
Figure 6: A first look at the AM Modeler
An AM Modeler tree window will also appear next to the Model and Results tabs. The AM Modeler tab contains three main setup sections as part of the model tree definition; Data Setup, Model Setup, and Simulation Setup as shown in Figure 7 below.
Figure 7: AM Model tree
A simple double click on the top of the model tree will prompt the model creation window (Figure 8). The user is requested to choose the model source and definition. For this simulation I used a Thermo-Structural Analysis type using the Abaqus built-ins process type, I did point the two models previously created to the AM Modeler as well using this window.
Figure 8: Model Creation Window
Running the AM job
Now the job is all properly setup, the AM will include the proper Keywords into the main input files to run the Additive Manufacturing simulation properly. In order to run the whole simulation, run the heat transfer job first, then follow by running the structural job.
Results and Discussion
Figure 9 illustrates the temperature distribution on the left as well as the residual stresses due to the process. Material is gradually added during the simulation while we can see at the same time the temperature changes due to the heat as well as the residual stresses due to the heat gradient.
Figure 9: Final Results
We can see how the use of the AM Modeler has made the additive manufacturing process simulation much easier as if offers lots of control over the model.