Computational Fluid Dynamics (CFD) Workflow

What is CFD Workflow?

CFD Workflow is the process of gathering results from tests or ‘runs’ of the car in simulations, and using these results to generate plots, pictures, graphs and videos to better understand our results and easily compare them.

We use Cp (Coefficient of Pressure) plots to describe the distribution of pressure under the car, allowing us to identify what areas of a specific part is making downforce or generating lift. This is also vital when used with streamlines as it allows us to analyse if the flow structures are behaving as we expected and hoped.

We also rely on Tp (Total Pressure) videos to show the total pressure of the flow, with the total pressure being a combination of static pressure and dynamic pressure. This helps us to identify the location of vortices, and most crucially, allows us to see where energy (or Tp) is being taken out of the flow. To generate downforce, energy needs to be used. Seeing where the Tp changes gives us an indication about energy losses, flow attachment, vortices bursting and more.

The Benefits and What We Wanted to Achieve.

The main goal behind starting the post-processing system was and is, to increase the efficiency of the operations as a whole. To achieve a higher level of efficiency, we wanted to be able to standardise the type of post-processing and reporting and save time on the system in general. We also wanted to be able to increase the overall quality of our post-processing, and finally, our goal was to save time on the setup and running of our CFD runs. Ideally, we also hoped to have the opportunity to be able to remotely perform CFD analysis while overseas in Europe.

The Beginning

To begin such a technological project, it is extremely important to form the foundations first. This is almost always done through research and, in this case, multiple programs were analysed and looked into, with CFD-Post being selected as the program of choice for the basis of our post-processing system, as it worked well with ANSYS Fluent, which we already had the scripts needed for the automation process. More research was conducted in the form of reviewing existing articles relevant to the project, such as Willem Toet’s article regarding good practices for post-processing.

The System Structure

As a result of the updated post-processing system, we no longer have to manually upload our CAD, instead we can now send our CAD via an email to our MMS server. The scripts are designed to be continually checked which, as a result, gives us the capability to send emails from anywhere in the world, without having to connect to the university VPN. This provides us with fast download speeds, no matter where we are.

The run data is sent off according to the parameters that are set by the designer. All of the information is stored on the MMS server to allow ease of access for the whole Aerodynamics design team. The run’s information is inputted into our run tracking spreadsheet, providing a history of all statistical information and iterations that were performed throughout the design period and all other runs individually.

Once the run is complete, we use a virtual machine to allow our post-processing to occur on Monash’s high performance computing (HPC) Monarch cluster. This enables us to cut post-processing time from multiple hours on a typical workstation computer, to a mere forty or so minutes on the cluster. After this, the results are all automatically downloaded and a standard post report is created. This allows us to compare the runs that have already been completed and helps us better analyse using a baseline or other runs of interest. This concept was a key point made by Willem Toet in his post-processing article.

The user is notified and informed throughout the whole process, ensuring they are aware of all the significant details regarding what is happening as the operation progresses. On top of this, an error email system has also been integrated for setup issues or any other problems involving data and runs. This allows the user to fix the issue and continue with their run relatively quickly.

The Challenges

A difficulty faced when attempting to design the post-processing system was working with Monarch’s lack of screen-rendered CFD Post. This had been a major hurdle in previous years, but was eventually overcome through porting the output to our MMS server.

As well as this, working with Gmail and Google Sheets has meant that access has to be refreshed to our dedicated Monash CFD Google accounts every two weeks or so, which must be done manually.

The Design Timeline

From the initial research and planning phases of the project to the primary objective and goal being reached (a highly efficient and reliable CFD workflow and post-processing system), the project took approximately seven months to complete. Excluding the three following months spent on initial updates, improvements and debugging.

Research for the project began in January of 2020 and was undertaken for a month. The implementation of the scripts took approximately three months (to April) and then work on getting the email and complete automation loop went until July. Since the project was completed, continuous work has been done to update graphical elements, add aero maps and continue on debugging.

Conclusions

So far, this project has allowed the team to perform well over 1300 simulation runs during the 2020 design period, matched with better efficiency and post-processing in general. This has enabled us to spend more time designing our aerodynamics package, and reduced the amount of time spent sitting at a computer waiting for CFD Post to load and run altogether.

It has also given us a base to expand the automation system to every aspect of the Dynamics department in the future, with aero map and CarMaker (an industry grade car simulation tool that we have started using for vehicle controllers and lap simulation) implementation possible. This new system has the ability to adapt for years to come, just by updating names and the locations of files alone. This has given us the opportunity to easily ensure all reporting is being completed to the quality and quantity we are happy with, and not falling off later into the design period, which is something that has been noted by previous Aerodynamics Leaders.