I thought, "Harrington- he's sharp. Let's see what he says about simulation." Nice article, but a little disappointing from the standpoint of missing the greatest pitfall. Mentioned that most simualtion produces reliable results. That's true, but these software programs still run on computers and the simulation output is only as good as the input. Very old catchphrase was "Garbage in, garbage out." This is still true, no matter how sophisticated the programming and it remains the grain of salt that must accompany all simulation reports. Second pitfall - the more eye candy in the report/presentation the more the uncertainty regarding the results. Trust, but verify.

As always, a well considered column and a topic more people need to be aware of. Yet, like most such pieces on simulation, it lacks the necessary details on implementation, and ultimately falls short in terms of its applicability to real-world problems. What exactly is the next step for the reader, who is now impressed with the efficacy of simulation? And where should the reader look to learn more? A stronger column would have delved into the software, which is a critical component of (and often barrier to) simulation, and would have guided readers to additional resources and case studies.

I agree with the others. More meat needed. Surprisingly light on value for a Harrington piece.

I was using process simularion models (visually based, not text) ~20 years ago. Not sure why this article seems to make it sound like some new technology.

I think using flowcharts in the title and lead paragraph is a bit of a straw man. Most flowcharts are not used for as-is vs. to-be process modeling. Most are used just to document existing processes.

I've been tangentially involved in one simulation study 15 years ago, and as I recall, the it took a lot of work to characterize the process steps in order to build the model. This article makes it seems as though the computer is magic and simulation tools act as an oracle.

I would much rather read a real world case study and an evaluation of which types of processes lend themselves well to simulation. I imagine you would get better results with low-mix, high-volume, or highly automated processes than you would with a high-mix, low-volume, or highly manual processes.