Altair has submitted this post.

Written by: Drew Buchanan, Engineering Manager, TrueInsight

I was born well after the Apollo 11 mission, but I was always deeply fascinated with the moon landing and space. As a kid I can recall building model rockets, and reenacting famous scenes from the Apollo 11 mission, as well as various other space moments. My fascination with space led me to the engineering field, and even though I do not directly work in the space industry, I still follow it with a close eye.

When NASA announced the Artemis Moon Missions, I felt my child-like excitement about space exploration come back in full force. For those unaware of the Artemis Missions, they will be a series of launches focused on human-led space missions to the moon. The Artemis missions’ end goal will focus on establishing a permanent base camp on the moon with hopes of facilitating future human missions to Mars.

Me at Kennedy Space Center in 1992.

The Artemis mission plan is set to evolve in stages, starting with an uncrewed test of the space launch system (SLS), all the way through a crewed launch to the moon and an eventual landing of astronauts on the lunar surface. As a result of the tiered mission process, there are numerous things the Artemis team must engineer and design, and the timeliness of the missions are very quick (Artemis I is expected to launch at the end of September 2022).

NASA Artemis Rocket Prior on launchpad (Image Source: NASA/Kim Shiflett.)

With so many engineering design challenges, and a speedy timeline, NASA and its partners will have to make engineering decisions quickly and efficiently for the Artemis Missions. To maintain their speedy timeline, NASA routinely uses Design of Experiments (DOE) during its design process. DOE is a common design technique in mechanical engineering design, and it will be fundamental in the success of the Artemis missions.

What is Design of Experiments (DOE)?

Design of experiments has its roots in statistics and has been implemented across various engineering disciplines. Ultimately, DOE is a methodology that utilizes a set of design or simulation criteria, and then evaluates that criteria through statistics to determine if an engineering design is acceptable.

An example of a DOE could be analyzing a new product across various dimensional variables. The engineer may have a set of dimensions they can modify their design to, and they can then run a DOE to determine the trends as they adjust the dimensions of the design. Ultimately, through this DOE process the engineer will be able to evaluate many designs efficiently and come to the right decision on what design to utilize.

In the past, the drawback of DOE was that it required longer lead times, because engineers or designers would have to come up with physical prototypes to run the DOE. Additionally, statistical methods had to be traditionally done by hand as well. As a result of this, DOEs would take substantial time prior to selecting the right part design. If you look at the initial Apollo 11 mission, over 400,000 scientists, engineers and designers were working on the project[1]. One reason is because engineers had to build, test and evaluate the majority of their designs with physical prototypes.

However, with advancements in technology, traditional engineering processes like DOE have been aided by Computer Aided Engineering (CAE), which has substantially decreased the number of engineers and engineering hours required for designs. Currently, the entire NASA workforce only has about 20,000 workers[2]. Therefore, you can imagine the expectations of Artemis engineers to evaluate and look at all designs, and how DOE could be very vital to their decision-making process.

Solving DOEs Quicker with Computer Aided Engineering

Altair’s HyperStudy has become an important DOE tool because it implements CAE Integration with advanced statistics. Users can import various FEA/CFD/CAE files and then run their designs directly through the software interface to make design decisions. In this case, I decided to run an example of a robot arm, which could be comparable to a part that Artemis Engineers are designing.

The robot arm that I wish to run a DOE on needs to meet numerous stress and displacement criteria, while remaining easy to manufacture. The DOE tool will allow me to set up one FEA study, then run all the remaining studies within its interface. In this case, I set up the original study in Altair HyperWorks and Altair OptiStruct to solve for a structural FEA. I then take the FEA study and feed it into my DOE tool to solve the additional studies. Note, HyperStudy also has capabilities to bring numerous CAE files into the tool, including third party CAE files such as those from Ansys, Adams, ABAQUS and more.

Robot arm as Modeled in HyperWorks.

Our objective in a DOE for this robotic arm is to reduce the mass and material cost, while maintaining the structural requirements (i.e. not exceeding the critical stress or displacement values). As you can see in the above image, there are numerous locations where we can adjust dimensions; something that would be very time consuming to iterate in traditional FEA. Through HyperStudy, I can utilize its automated interface to select the dimensional variables by adjusting the lower and upper bounds. The tool also enables users to select simulation-based variables, but in this case, I will stick to adjusting the dimensions only.

Adjusting bounds for dimensions of robot arm.

After keying in the dimensions, I can run the DOE study and then obtain the results for the various scenarios. The great thing about this study is the ability to utilize advanced statistics which are built into the tool to understand the direct impact of each variable.

Looking at this case, we have numerous variables we are adjusting, and we want to see what the greatest contributor is to increasing stress. Through looking at the Pareto Plot we can see that the greatest stress contributor is our Length_5 dimension (this dimension corresponds with the arm pocket length dimension).

HyperStudy robot arm stress Pareto Plot.

HyperStudy’s ability to collate large portions of data and then run advanced statistics to show trends is a key advantage. In this case it was able to run 512 studies. Had we utilized a traditional approach of trying to run all these studies directly in our FEA tool, it would have been significantly more laborious and more prone to human error. It also gave me the tools to understand my design in a fraction of the time. CAE-enabled DOE tools are pushing the future of space exploration and innovation. As I await the Artemis launches with an eager eye, I can’t help but be excited to see where the Artemis team will go with all the valuable CAE tools engineers have at their fingertips.

To check out the workflow for setting up this robot arm in HyperStudy, check out this video or visit Altair.com.

Sources: