How we work / Line study
Find the bottleneck
We measure your line, build a model of it and test fixes on the computer first. Then we build the one that pays.
- 42 s/unitCut59% busy
- 44 s/unitWeld62% busy
- 71 s/unitPaint100% busy
- 45 s/unitPack63% busy
Two versions
Both end with a ranked list of fixes and what each one earns.
- Line you run today
Line study
Output is below what the machines should do, and nobody agrees why. We measure the line, then test fixes on a model of it.
- Line you are about to build
Expansion study
There is nothing to measure yet. We start from your current line and the machine makers' data, and size the new one: stations, buffers, robots.
How a study works
- 1 day1
Walk the line
Scope and fixed price. We map stations, buffers and changeovers with your supervisors, and agree what to improve.
You get: written scope, fixed price
- 2–3 weeks2
Measure
Real cycle times. We read run state from your machines, or log it with clamps and counters, across a full run with changeovers.
You get: cycle times, stops and reasons
- 1–2 weeks3
Model
Checked against your week. We build a computer model of the line and compare it to your measured output before we trust it.
You get: a model that matches your numbers
- 1 week4
Test the fixes
On the model first. A buffer, an operator, a camera, a robot, a new layout. We try each and rank them by payback.
You get: ranked fixes, with payback
When it pays
And when it doesn't. We tell you on the plant walk.
- New line or expansion
Size it before you buy it
How many stations, how much buffer, how many operators. Cheaper to find out on the model than on the floor.
- Many products, one line
Changeovers eat the week
We test run order and batch sizes before you change the schedule.
- Mobile robots
How many robots, not how fast
Fleet size depends on routes, charging and peak demand. We test your layout before you buy vehicles.
- Layout change
Move a cell on paper first
See what a new layout does to flow and wait times before anyone moves a machine.
- Output stuck
Stops nobody writes down
Measurement shows which stoppage costs the most output. It is often not the loudest one.
- When it doesn't pay
The answer is already clear
One machine is plainly the limit, or there is no way to measure. We say so and go straight to the fix.
A typical study
- Problem
- Output is below the rated speed, and nobody agrees which station sets it.
- We do
- Measure every station for two to three weeks, build and check the model, then test the fixes your team suggests and ours.
- Timeline
- 4–6 weeks
- Typical range
- $10k–15k
- How you'll know it worked
- The model matches your measured output within a tolerance we agree up front. Each fix has output gained and payback in months.
- If you go ahead
- The study fee is credited toward the fix. If you stop at the report, that is all you pay.
What you get
Measured data
Cycle times, stops and stop reasons, in a spreadsheet you keep.
The model, in your name
Handed over with its assumptions, so your team can rerun it.
Ranked fixes
Each fix tested on the model, with output gained and payback.
A written report
Findings, assumptions and what we would do first. Short enough to read in one sitting.
An honest "not worth it"
If the model says a fix won't pay, the report says so.
A fixed-price fix
If you want us to build it, the study fee comes off the price. Or take the report to anyone.
Questions plants ask
Why not just buy simulation software?
Packages exist. They need clean data and someone who can build and check a model. We supply both, and we then build the fix.
What does it cost?
Most single-line studies are $10k to $15k at a fixed price, agreed after the plant walk. An expansion study is scoped after the walk. If you buy the fix from us, the fee comes off.
We don't have data. Can you still do it?
Yes, that is the first step. We read run state from your PLCs, read-only, or log it with current clamps and counters. If we can't measure a station, we say so rather than guess.
What if the bottleneck is obvious?
Then skip the study. A day on the floor and a few hours of downtime data answer a lot. We will say so on the plant walk.
How accurate is a model?
As accurate as the data behind it. We check it against your measured output and show you where it does not match.
Is this the same as connectivity?
No. Connectivity is a live downtime board that stays on your floor. A study is a one-off answer. The same links can stay on afterwards and become the board.
Will you push us to buy your fix?
No. The report lists fixes anyone can build, including doing nothing. You can take it to another integrator.
Who builds the model?
Andrew Nash, who created Aurrigo's airport simulator, millions of times faster than real time. Frank Walsh handles the machine signals and the fixes that need hardware.
Can funding help?
Maybe. Programs differ on whether a study is an eligible cost, and most won't cover costs from before you apply. Ask the program first. See tariff relief
Tell us what's slowing your line down.
Book a plant walkPick a time on our calendar, or send us a message.