Avoiding a $40M system that would not have helped

The problem

Where the time was going

When a roadside bomb goes off in a war zone, the fragments are collected, scanned, and shipped to an advanced biometrics group in the United States, where the remains are examined for evidence of origin, tactics, and trends. That intelligence shapes how the military counters future attacks, so a timely answer can save lives.

The existing system looked effective, but too much time was being lost in the analysis itself. The delays came from several places at once: the time-zone gap between onsite investigators and US scientists, the many different kinds of information sent for a single incident, the way that data was routed through different departments, and the effort to consolidate everything back together before a report could go out.

The hard part of a system this tangled is judging how much any single improvement helps the whole. That is exactly the situation when a vendor proposed a $40 million solution promising to cut turnaround time by 30 percent. The presentation was excellent, and with soldiers' lives at stake the proposal was on a fast track to approval, until a simulation study was authorized first.

What we modeled

Mapping the process, then testing the fix

Spreadsheets are strong tools, but they fall down on time-dependent work like overlapping schedules and processing that varies with the mix of inputs, where a turnaround estimate can be off by 50 percent or more. Simulation links those moving parts together in a way a static sheet cannot.

ProcessModel simulated the full analysis workflow end to end: the work schedules of onsite investigators, the many information types sent for one incident, the arrival rate of that information as large files take time to transmit, per-type processing times, the schedules of offsite analysts by skill level, the consolidation of results, and the time to transmit the finished report.

The result

The proof, and the payoff

With the workflow modeled, the team ran experiments to see how the new technology would actually change turnaround time. The finding was clear: the vendor's solution would produce no appreciable improvement once it sat inside the real system of schedules and handoffs.

The proposal went from fast track to shelved in a matter of weeks. Avoiding a purchase that would not have helped saved $40 million on a single project, for the modest cost of building the model that revealed it.