A thought experiment in military logistics and visualization

I am an instructor of logistics and resource operations,  and a curriculum designer at the US Army Command & General Staff College at Ft Leavenworth. My 25 yrs of service were as an infantryman and a practitioner of brute force military logistics.

having retired, I am trying to add rigor to my own teaching, our curriculum and our field grade officer students by introducing and demonstrating the cumulative effects of uncertainty in large transportation networks in a theater of operations. I am a fan of your work and have reasonable practical excel skills.

I need to model the following “story problem” and provide a visual tool that allows students to make assumptions about the nature of the complex movement, to turn those assumptions into a visual distribution, and then monte carlo the results to see the performance of the transportation network contained within the set of their assumptions. The next step would be for them to vary their assumptions to see the effects, in order to reveal which of their assumptions are important and sensitive, so they can make decisions about where to deploy units in order to affect the plan.

For example:

1. should we employ the Military Police brigade forward to guard an assembly area at the end of the network and thereby accept risk along the lines of communication, or should we guard the roads and accept risk forward near the battle area?

2. What are the Intelligence officers estimate of how many days out of 100 that the weather will degrade road movements by 1 of perhaps 5 categories (0 -.2)(.2 -.4)(.4 -.6)(.6 -.8)(.8 -1.) or to pick a distribution from among a set of choices that represents their estimate.

3. If we move a civilian population away from the battle pre-emptively, what will that do to our performance (ie no displaced civilians around to clog the roads in the heat of combat) vs waiting until the last minute

4. etc

The transportation challenge in our scenario: sitting on the ground at 2 seaports waiting to be moved forward are 750 Army units, and 100,000 short tons of cargo, with “x” number of trucks, over poor roads, a distance of 600km and about 2000km of total road network, , with an uncertain enemy, an uncertain civlian population, with a variable degree of road march discipline among the 750 units, and a choice of parameters about how fast and how densely we may organize our movements, mission to be completed in less than 100 days.  The commander needs our estimate of how soon we’ll be done. we need to express that reasonably and with proper confidence, and give him choices

there are so many variables and combinations, that we cannot rely on our intuitions to describe our capabilities and expected performance in a meaningful way to a commander. we cant describe the impact of the choices he has

My problem: the military ORSA community at TRAC (TRADOC Research and Analysis Center) are interested but without much funding, I cannot jump to the head of the line; I believe either XLsim 3 or Risk Solver will do everything I need, but i’d appreciate your insight into which platform is sufficient. I ‘ve found a student (no modelling skills) here with an interest in exploring this in his thesis next year at the school for advanced military studies (SAMS) the “Jedi Knights” and i may have another, who is an Excel wizard who I might convince to play along too.

The potential of this model is enormous, as the curriculum I write will be taught to EVERY major in the US Army (active and reserve) and many Navy, Air force, Marine offices, and several hundred international officers. This is about 5000 officers a year

I am hoping that this visualization and simple model will do for their appreciation and respect for risk and modelling what your book has done for mine.

I am hoping that this model will help them learn to operationalize the implications of their assumptions, and demonstrate a way to manage, appreciate and explore complexity beyond the limits of their intuition

I am envisioning a model of the network which defines a complex of nodes and pipes, each of whose performance would be modeled in a standard way with the variables above, and which would permit insight into how variable performance oalong each node serves to degrade performance more than a simple avergage of all their performances in the time period

I’ve designed a small world simulation that we are already using to let students sample these ideas simply thru game play in a small way, but we need more depth toctake it to the next level

I say all that to simply say: is XLSim3 sufficiently robust to explore this model, or would you recommend Risk Solver? In a perfect world I’d have something that I could deploy in January 2011, and I am prepared to spend my own money on this if I cant convince our leadership to provide some money, because i believe this is crucial to our profession.

I’d be grateful for your thoughts. I’d be even more grateful if you had a spare ORSA grad student with an inquiring mind and some modelling chops 😛

my work email, if you’d like to comment to me directly is long-kenneth@conus.army.mil