Simulation Data Management

Remember back in the day when a few gigs of disk space was a serious cost that had to be considered when planning a project? Simulation activities were carried out on standalone workstations with key input decks archived off for posterity. Those were the days. There was no opportunity to capture any work in progress. If a key member of the simulation team was off sick or got another job, frantic calls would be made. Was BracketFinalFinal9_ImNotKiddingThisTime.inp really the correct input deck to set running on Friday afternoon ready for that essential progress meeting with the customer on Monday morning?

Thankfully, disk space became cheaper. We could archive more than just the input deck, and we could even include restart files so that didn’t have to lose all the CPU time that had been invested when extending a simulation. Most simulation departments got or​ganised with central repositories for simulation files with some sort of database, or at the very least a spreadsheet so that we didn’t have to resort to guess work when finding the correct model. You can see the early NAFEMS templates that were recommended to capture the meta-data for a simulation here.

It wasn’t all good, the availability of cheap file storage freed the lazy simulation engineer from considering what was important and these central repositories often ended up as a huge digital landfills. I’m sure most of us have been chased by the IT department convinced (probably correctly) that the simulation department couldn’t possible need ALL the Terabytes of data that has been saved for a particular project. We are also reliant on the project engineers to appropriately document the work they are doing so that others can easily identify the files. Accessing and interpreting the simulation output was entirely reliant on the support of the simulation team. Sure it made the simulation team vital but it didn’t allow other parts of the organisation to see the value of simulation.

Step forwards two decades to today and the pace of the design process has increased exponentially making the need for an audit trail more relevant than ever. Designs are continually evolving, we work with colleagues across the globe, and we can’t just walk over to our colleague and ask them to put the latest version of the model on the J:\ drive. More now hangs off the results of our simulations than ever before as they are increasingly being used to certify products and make decisions on maintenance and safety issues.

We need to have confidence that we can put our hands on all the inputs that fed into a particular simulation. With each simulation we should be able to answer the 5 W’s…

• Who did the simulation?
• Why did they do it?
• What was the result?
• When was it executed?
• Where are the relevant files?

We need a clear audit trail, if you will excuse the jargon, we need a digital thread that runs through all simulation activities.

Creating functionality that can create this audit trail is challenging. Not only is there a technology challenge there is a cultural one. A Simulation Data Management capability is only going to work if the engineers running the simulation support it, and they need to be bought into the concept. This means that using a SDM capability shouldn’t come with a high time overhead for the analyst. They need to see the benefits of using the system. If an effective system is created it should save a simulation engineer time as they won’t have to go rooting across the organisation for data.

There is more to SDM than maintaining an audit trail. Once your organisation is effectively managing your simulation data effectively you will start to build up a digital archive of knowledge. We all know that those grumpy engineers who have seen it all are oh-so-valuable. Well, SDM gives others the ability to learn and exploit what has been done previously. This is an important point to make when building the business case for investing in SDM capability as each simulation that is run contributes to the total simulation asset of an organisation. With machine learning capabilities developing rapidly, organisations are only going to get better at tapping this gold mine.

The case for managing your simulation data effectively is easy to make. How your management of simulation data is implemented will depend on the industry you work in and the number of simulations you are running. The needs of your average post doctorate researcher differ from the needs of the stress team in an aircraft OEM. Developing an SDM capability can involve an eye watering amount of money so make sure you know exactly what your requirements are before you start scoping out systems, and make sure your simulation engineers are involved in the process.

Ian Symington, Technical Officer, NAFEMS

Events

• NAFEMS World Congress 2021 incorporating the Int. SPDM Conference (25-29 October, Salzburg & Online)
• Visit the "Accepted Abstracts" page and search "Data Management" for a listing of the 129 related presentation titles, presenters, and abstracts.

Communities

• Join the NAFEMS SDM Community, which allows NAFEMS members to engage with our technical experts on the SDM Working Group.