Sizing an AMR fleet before you buy the robots
WareMax lets operations engineers sweep AMR fleet size against throughput and cycle-time tails on a seeded scenario, producing multi-seed confidence intervals so fleet-sizing decisions rest on tails, not means.
You are about to commit capital to a pod-to-person AMR fleet and you need to know how many robots. Too few and stations starve; too many and robots congest each other. The answer is a tail question, and WareMax is built to answer tail questions.
The workflow
- Encode your warehouse as a scenario YAML — topology, stations, order arrivals, and traffic capacities.
- Sweep
robots.countacross a range with multiple seeds per point. - Read throughput and the cycle-time tail at each fleet size, with confidence intervals across seeds.
waremax sweep scenario.yaml --param robots.count --range 6..20 --seeds 10Why discrete-event, not continuous
A fluid-flow model tells you the mean robots-per-station and mean travel time. It says nothing about the moment a station queue builds up or two robots contend for the same node. Those moments decide whether an order misses its due time. WareMax models them explicitly.
Honest caveat
WareMax models RMFS dispatching only — not AS/RS, conveyors, or human pickers. If your sizing question depends on those, WareMax sizes the AMR-dispatching part and leaves the rest to you.
Related: comparing dispatching policies · how it works. Built by Skelf Research.
Frequently Asked Questions
Why not just use a spreadsheet or a fluid-flow model?
Fleet-sizing decisions are tail decisions — queue build-ups and congestion spikes — that mean-based models hide. WareMax is a discrete-event simulator, so it reports the cycle-time tail honestly.
How confident can I be in a sweep result?
WareMax runs multi-seed sweeps and reports confidence intervals. Because each run is deterministic under its seed, the intervals reflect scenario variation, not simulator noise.