Here's the conclusion up front: how to test ASRS throughput isn't about how fast a stacker crane can move. It tests three things: single-crane dual-cycle capacity, sustained aisle-level operation, and mixed-flow performance under a realistic slot distribution. It's one of the questions I get asked most often, and the order never changes: validate the method first, then the data, then the equipment. A wrong test method makes every number meaningless — including the flattering ones.

The Pitfall I've Seen Three Times: Wrong Method, Fake Numbers

Fall of 2023, an e-commerce peak-season DC in East China. The contract said 90 pallets per hour per aisle. Acceptance day produced 61. The vendor swore the cranes were fine and blamed our WMS for slow task dispatch. We didn't buy it. Two days of standoff.

Then I spent a morning watching the slot map, and there it was. The first test cycled only the nearest third of the rack, while the contract's 90 was rated on randomized putaway and retrieval across the full aisle. Shorter travel should have been faster — except the crane spent the whole time accelerating and braking in that short zone, never reaching cruising speed. Throughput died in the deceleration zones.

We redistributed test inventory across the full aisle per the contract's random-slot rule. Second run: 87 pallets per hour. The last 3 pallets? Traced to our WMS dispatching tasks one at a time — the crane was literally waiting for work. We switched to batched dispatch with task prefetch, and it passed. That project taught me the rule I still use: when the acceptance number fails, don't start by blaming the equipment. Eight times out of ten, it's the test method or the scheduling.

Here's the counterintuitive part most buyers miss. The vendor owns "the crane can move." The WMS/WCS owns "the crane stays busy." A throughput test that fails is usually a scheduling problem wearing an equipment costume. I say this as the WMS person in the room — it's often my own side's fault, and I'd rather find that out during acceptance than during peak season.

How to Test ASRS Throughput: Align the Method Before Arguing About Numbers

The night before test day, I run the same checklist on every project. Call it my night-before-go-live checklist:

  1. Pull the contract appendix and confirm three things: test duration (four to eight hours of continuous operation is typical industry practice), slot distribution rules (random vs. zoned), and allowed exceptions (cold storage, overweight loads — is there a derating clause?). Anything vague gets an addendum signed before test day, never during the argument.
  2. Pre-stage inventory across the rack per the contract ratio. Nothing stacked at the near end. The empty-to-full slot ratio should mirror real operations, not the vendor's wishful thinking.
  3. Prepare task data ahead of time: inbound/outbound mix per the agreed ratio (say, 40% inbound, 60% outbound), imported into WMS before the test. Nobody hand-keys task orders on test day.
  4. Agree on the timing rule: the clock starts when the first pallet leaves the inbound conveyor and stops when the last pallet reaches outbound. Both sides keep a log and co-sign on the spot.
  5. Test the single crane before the aisle. Dual-cycle baseline first; if that fails, the aisle test is pointless.

There are four test methods. Which one applies depends on what the contract says:

| Test method | What it measures | When to use it | | Dry-run, no load | Mechanical motion is sound | Right after installation; warm-up only, not acceptance | | Single-crane dual cycle | One crane's put-away-plus-retrieval ceiling | First acceptance step; the equipment baseline | | Continuous in/out operation | Sustained aisle-level throughput | The core of acceptance; the contract value usually means this | | Mixed-flow endurance | Long, stable run at realistic ratios | Final gate before go-live; four to eight hours |

Whichever method the contract names, run it exactly as written — improvising how to test ASRS throughput on the day is how vendors end up grading their own homework.

Engineers observing continuous pallet infeed during ASRS acceptance testing

Missed the Contract Value: A Four-Step Playbook

Early 2024, a pharmaceutical cold-storage DC in North China. Acceptance stalled at 58 pallets per hour against a contract value of 80. The vendor blamed our task mix. We blamed their WCS for never pairing tasks — inbound and outbound dispatches went out separately, so the crane ran single cycles all day, quietly surrendering nearly 30% of its capacity. One WCS update with task-pairing logic took throughput from 58 to 76. The remaining gap? Motor derating kicking in at minus 18 degrees Celsius, already covered by the contract's cold-climate clause. We signed an addendum: acceptance at 95% of contract value under sub-zero conditions. 76 cleared the bar. Signed off.

When the number misses, work these four steps:

Step one: retest. Change slot distribution, task mix, and the person holding the stopwatch — one variable at a time. Three runs, take the average. A single data point proves nothing.

Step two: isolate. Split the gap into three buckets: raw crane capacity (the dual-cycle score), WCS task scheduling (is it pairing cycles? is the task stream breaking?), and slot strategy (near-end congestion or far-end starvation). Most projects land their gap in the last two buckets.

Step three: negotiate the qualifiers. Cold temperatures, overweight or odd-shaped loads — if the contract has derating clauses, walk through them line by line. If it doesn't, sign an addendum now. This is the concession vendors accept most readily.

Step four: negotiate the money. Withhold acceptance payment in proportion to the gap, or go phased: go live at measured throughput, give the vendor a deadline to optimize, release the retention when it passes. Remember the trade-off nobody says out loud: refusing to sign burns your labor-substitution cost and your go-live delay, every single day. The vendor can afford to wait. You can't. Within 5% of target, push for a qualifying addendum and release. Beyond 10%, withhold payment or demand a remediation deadline. No middle option.

Throughput Test Failure Points I See Every Year

First, staged slots. Everything parked at the near end makes the number look good and collapses in the first week of production. Second, testing outbound only, or day shift only — real operations are mixed, and the test should be too. Third, the WMS-to-WCS interface gets tested in a bypassed "direct mode" and then production traffic goes back through the normal chain, adding two or three seconds of task latency per cycle. Throughput drops off a cliff. Acceptance must run through the production path. A test that bypasses it is no test at all.

One line to close: throughput acceptance is won on test method, task scheduling, and negotiation order — not on the crane's spec sheet. Get the method right and the numbers defend themselves. Get it wrong and your signature is just a down payment on future pain.

By Lin Xiao, WMS Implementation Consultant