[Discussion] How do you scope your models?

[jdoran]

I have always found that a natural part of the design process is to sit down and draw out (maybe in another tool that is better suited for conceptual work) the entire system and interactions you are planning to simulate, before you get cracking at the task in the actual simulation tool (in this case Simio).

This allows me to decide which parts of the simulation need to be cut in or out of scope, and gives me a checklist to refer back to at the end to verify the model.

For me, this has primarily been using UML-like diagrams in tools like Enterprise Architect (http://www.sparxsystems.com.au/).

How do you guys do it? Do you jump straight in? And have any of you found any way (or made your own), for either bringing your conceptualizations into Simio, or back out to compare at the end?

I suspect this will apply more to the academic uses of the tool, as there can be a lot of time pressure on simulation generation.

I will be looking into this more in the near future, and would love to hear how my fellow Simio compatriots approach this issue in their projects.

Of direct interest to this topic would be the Arena functionality to export models in UML. I believe there were also 3rd party plugins for importing too. Attached is a quick overview of this functionality. I believe it is related, as it would be nice to see this kind of functionality in the future in Simio (the next generation incarnation of Arena, that even possesses some of the creative genius/developers that made Arena).

ExportToUML.pdf

[equaglia]

I started thinking more about this subject in the WSC in Berlin. Before it I was just jumping in the model and now getting some difficulties (think I´m the only one that can go deep on my models).

I found with one colleague the technique IDEF-SIM http://www.informs-sim.org/wsc11papers/138.pdf, which I intend to apply to my models.

Agree that Simio could facilitate model conceptualization.

Eduardo.

[tulach]

Hi, nice topic.

The use of IDEF-SIM seems promissing, especially for bigger projects with more than one modeler. Thanks for pointing this out.

I often use pencil and paper or whiteboard & whiteboard marker, if there are more people around.

As we often do VSM mapping in manufacturing prior to making models I often have VSM maps, that sometimes contain neccesary data, but not always. (And sometimes the are not all relationships shown).

But my models are often done on a tight schedule and even me myself have problem to look deep down my own models, which is a shame.

[jdoran]

@tulach

I can understand where you come from completely, as I also experience the time pressures sometimes associated with projects (especially smaller ones). I initially started by jumping head-first into Simio when modelling, but as I have gained more experience with the tool, have been afforded a little extra time now to get a more complete idea of the model before starting.

@equaglia

Thanks for the share. I will definitely look more into this. It is always nice to look at problems from different angles, and I can definitely see the benefits of using those specific techniques to describe a simulation.

[dsturrock]

I have a slightly different perspective. I absolutely do NOT just jump in, but it may appear that I do since I start modeling in Simio almost imediately.

I strongly believe in having a functional specification (FS) for each project. This is one of the first things I do. And part of that FS is doing a prototype. While you could use many different tools for prototyping/conceptualizing, I find that using Simio itself offers several advantages.

There are multiple goals to the prototype including:

--Understanding the big picture

--Understanding the potential gotcha's (not just technical ones)

--Exploring scope and completeness

--Communication between modelers and stakeholders

In the ideal situation, the prototype is a simple representation of the system that looks and behaves similar enough to the real system that it helps stakeholders understand the issues mentioned above. But it is not necessary that it actually run, and it sometimes won't look much like the real system. It will NEVER provide useable results (e.g. answers) beyond the goals mentioned above.

In most cases this prototype becomes a throwaway after it has met its goals, but if you are lucky, it may be close enough to the mark that you can use the prototype as the basis for the real model.

The FS process takes time (perhaps 5-10% of the total project time) but it is NOT EXTRA TIME! The FS process just moves some of the most important parts of the project into the early stages where it can dramatically improve the success of the project.

I have written several treatments of this approach. One can be found here: http://www.simio.com/publications/SASMAA/Simio_and_Simulation_PREVIEW_2e.pdf