I was born in Arlington, TX and spent most of my childhood in and around the Fort Worth area. When I was in high school, my family moved to Chattanooga, TN. I then went on to UTK for my undergraduate degree. I didn’t really know what I wanted to do with my life and ended up pursuing many different interests. I always enjoyed Latin, so I kept taking Latin classes and eventually added Greek to the mix, all while trying to figure out what my primary major should be. It took five years, but I finally graduated with a double major in Mathematics and Classical Languages (a major in Latin language and a minor in Greek language) in 2011.
I had become really interested in biology and biomedical applications. Dr. Steve Karman, my dad, pointed out that there is a field of computational biology and suggested I do my graduate work at the UTC SimCenter. My Master’s thesis involved 2-D CFD design optimization of the cross-sectional shape of arterial stents. I spent the summer of 2014 working at Pointwise as an intern, primarily stress testing Pointwise for the next release. The next fall, I returned to the SimCenter and shifted my Ph.D. dissertation to focus on high order mesh curving using analytic definitions derived from the geometry.
I have been around meshing my entire life. I remember watching my dad examining mesh quality and viewing CFD solutions from a very early age, but it wasn’t until graduate school that I finally understood what I had been looking at all these years. I became enamored with geometry, the meshing process, and the constant pursuit of higher quality and more automated meshing.
After my Ph.D., I accepted a position with as a Computational Geometry Specialist at EndoVantage, a small company in Scottsdale, AZ that has built a suite of tools to help surgeons assess, monitor, and plan treatment of cerebral aneurysms (including simulating the deployment of flow diverting stents). Although I was doing very interesting work with aneurysms, I wanted to redirect my career towards meshing, so this past July I joined the Engineering Services team at Pointwise.
- Location: Fort Worth, TX
- Current position: Senior Engineer
- Current computer: Windows 64-bit, Intel® Xeon® W-2135 CPU @3.70 GHz, 128 GB RAM, NVIDIA GeForce GTX 1070 Ti
- One word that best describes how you work: Thorough
What software or tools do you use every day?
Everyone has their preferred editor for coding/scripting. I learned to use Vim/gVim in my first computer science class in college and have stuck with it since. Most of my experience has been on Linux or MacOS, now I’m working primarily in Windows. Vim works the same across all platforms and you can install plugins to add additional functionality. Some of my favorites are UltiSnips and YouCompleteMe. With these you can define snippets to automatically insert and fill out functions, classes, etc. without worrying about syntax or formatting. If you find yourself typing the same thing over and over again, it is probably a good candidate for a snippet.
I use Corel PaintShop Pro to edit pictures for documentation and presentations. Finally, I spend an extensive amount of time using the Microsoft Office Suite, whether it is checking emails or my calendar in Outlook, analyzing data in Excel, or creating a presentation in PowerPoint.
What does your workspace look like?
What do you see are the biggest challenges facing CFD in the next 5 years?
As the CFD problems become bigger and more complex, so does the meshing, and automation becomes increasingly necessary. The technology has become advanced enough that fully automated meshing is possible from a technical standpoint. However, there are many decisions that a person makes (both consciously and subconsciously) that affect the quality of the resultant mesh, and thus the quality of the CFD solution. The challenge will be to maintain this quality in the automation process while also keeping it extensible to a wide variety of applications.
One of the unfortunate side effects of automation is that the process is no longer in the forefront of your mind. I think there will always be a need for a person to supervise the meshing and make changes if necessary.
What are you currently working on?
I am currently working on a couple things. First, I am working on generating meshes for the 5th AIAA Propulsion Aerodynamics Workshop. There are two geometries for the workshop, a nozzle and an intake. Secondly, I feel that there is a need to statistically analyze and compare meshes, so I have been working on a Python script that uses the Python Glyph API to collect data (initialization parameters, quality metrics, etc.) about the mesh and then uses Python visualization libraries (matplotlib, seaborn, etc.) to generate histograms, boxplots, and possibly other forms of data representation to compare the statistics of multiple meshes at once.
What would you say is your meshing specialty?
I would say my meshing specialty is high order meshing, which was the topic of my Ph.D. dissertation. There has been increased interest in finite element solvers because they can achieve higher accuracy with fewer degrees of freedom. However, it can be a challenge to maintain a high level of element quality while also resolving the high curvature inherent in CFD geometries. I am excited that this is a very active area of research and that accurate visualization of high order meshes is becoming a priority (when I was working on my dissertation the visualization options were extremely limited, especially above P2 elements).
Any tips for our users?
I will share two pieces of advice given to me over the years. First, if you are seeing a problem with element quality, check the connectors for problems. Most problems can be fixed by fixing the underlying connectors. Second, learn how to script. Scripting can improve, if not replace, the most tedious parts of meshing. You will thank yourself later.
What project are you most proud of and why?
I am probably most proud of my Ph.D. dissertation research. Whereas most people were approaching the problem of high order mesh curving from the standpoint of solving a system of equations (linear/non-linear elastic or some sort of optimization smoothing), I chose to pursue an analytic definition for the deformation informed by the geometry itself. I really had to learn how to rely on myself and effective brainstorming with peers and mentors to solve my problems since there was no similar research I could turn to for ideas.
What CFD solver and postprocessor do you use most often?
As most of my work has been on the meshing side, I don’t often run solvers. For my graduate research (both Master’s and Ph.D.) I used finite volume and finite element solvers I wrote myself. For postprocessing, I mostly use ParaView.
Are you reading any interesting technical papers we should know about?
Although this isn’t directly related to CFD, data visualization and especially statistical visualization has become a bit of a hobby of mine. After all, if you can’t properly see the data, how can you draw accurate conclusions? I seem to always be in search of a good color palette that is invariant to colorblindness and other visual imparities. Through my research I found a few interesting reads on data visualization:
- A paper from the Journal of Statistical Software about Polychrome, a package for constructing qualitative color palettes with many colors.
- An article on Data Novia about the top rated R color palettes
- An article on cran.r-project.org comparing the viridis color palettes to other standard color palettes, as seen through different colorblindness filters
Do you plan on attending any conferences or workshops this year?
I am not currently planning on attending any conferences.
What do you do when you’re not generating meshes?
Mostly I spend time with my husband and our animals. We enjoy taking the dogs on hikes if it isn’t too hot or cold, and the cats enjoy bothering us when we play video games or watch movies. I also usually have some sort of needlework that I’m working on (currently it is a cross-stitch collection of the solar system).
What is some of the best CFD advice you’ve ever received?
One of my professors in graduate school, Tim Swafford, used to always say, “There’s nothing more dangerous than a solution that looks about right.” I have found that even something as simple as how you view data (the type of plot used, the colorscheme, etc) can have an effect on the conclusions drawn. Over the years this has really sunk in to the point where I always double (or even triple) check my work and I try to analyze data by viewing it from as many perspectives as possible.
If you had to pick a place to have dinner, where would you go?