What is Democratization of Simulation and Model Deployment?
First, let’s start with some definitions. What is “democratization” (in this context anyway)? What is simulation? Is it different to modeling? Is democratization the same as deployment? And what on Earth is “appification”? Once we’ve covered the “what”, then maybe the “why” will have more context…
“Democratization” – at least to my mind – is making a tool or process accessible to people who did not previously have access. This of course leads to the question of why they would want or need access, which we will come to…
“Democratization” is a bit more than just “deployment”. We can “deploy” a spreadsheet calculator or MATLAB model (or even a NASTRAN model) simply by emailing it to a group of recipients, or putting it on a network drive or a portal such as SharePoint. That is NOT the same as “democratizing” – ensuring SAFE and SECURE access, plus the ability for users to run the model, WITHOUT deep knowledge of the underlying technology, and WITHOUT making mistakes.
For that, we need appification – creating a simplified web-based GUI while removing the requirement for users to run the underlying solver locally. Yes, I know that in the case of Excel, everyone has it – but not always the version for which the model has been validated, not always with macros enabled – and what about a mobile user on a tablet? As for non-Excel models, they will almost certainly need to be run somewhere other than the user’s machine.
Democratization also implies we need to think about version control of the models that are being democratized, as well as security of the intellectual property embedded in those models, and often, capturing of results to a central database to enable reporting and analytics.
What is the difference between a “model” and a “simulation”? The two words are often used interchangeably, but I think simulation is actually a subset of modeling. “Simulation” implies an attempt to simulate reality, to determine how a design will perform. It often implies 3D simulation tools (FEA and CFD), although we see many, very valuable simulations performed with MATLAB, in-house codes, and, of course, Excel (arguably the most widely used engineering software in the world).
So how is a model different from a simulation? Well, while a simulation is almost by definition a model, a model is not always a simulation. I wouldn’t really use the term “simulation” when referring to a pricing model, a configuration tool, or a profitability model.
OK, so why would anyone want to democratize a model?
Well, let’s first clear up a HUGE misconception: while we here at EASA are in the business of helping our customers deploy and democratize their models and simulations, we do NOT wish to imply that ALL of an organization’s models can (or should) be deployed. In fact, there might be only 1 model in 10 (or less) that should be deployed.
Identifying which models will provide more value if democratized is paramount – which models need to be run again and again? Which models have an impact on revenue (we mentioned pricing and configuration tools, which often sit right where engineering meets sales)?
So, suppose we identify those “unicorn” models – what happens to their value to the business once democratized? Our experience is that the value can go up by orders of magnitude.
So, yes, it may require some effort to identify those models within your company. But if you do, you can increase the value of simulation and modeling by 10-, 100-fold, or more.