Robotic Cycle Times: Actual vs Capaticy

Date 02-19-2010 15:35

Sounds to me like you need a time study or at least an investment in an inventory of Depends to eliminate the bathroom breaks.
Seriously though, I 'm not sure that without actually seeing your widgets and how your system operates that thats a question this forum can help you with other than with some general suggestions (like a time study).

Maybe our resident web link magician/guru can guide you to some internet assistance.

By eekpod

Date 02-19-2010 18:19

I can't say Iv'e seen many of us comment on robotic welding production here, but maybe you'll get lucky, good luck.
Chris

By ssbn727 (*****)

Date 02-19-2010 20:06 Edited 02-19-2010 20:10

On paper, I would say: "Most definitely!!!" However, things aren't always the way we see it on paper!!! ;) Let's see... 400 parts per hr. per 8 hr. day = 3200 parts and the difference between projected and actual is on average somewhere between 600-800 parts short of actual productivity from projected which means that you're losing on average of anywhere between one and a half to two hours worth of production per day and that needs to be accounted for as it adds up when one starts to consider weekly, then monthy productivity... You definitely need to find out why!!! ;) :) ;)

The easly way you can figure that one out is as Jeff mentioned, to do a revised time study... This time instead calculating what your robot is capable per day, calculate it down to the hour instead, and this way one can clearly distinguish the discrepancy on an hourly basis with respect to the shortfall of the daily output of what you currently produce which 600-800 less than projected... I would also check again what time it takes per part unit cycle to complete within the cell just to make sure everything is indeed "copastetic!"

You also need to consider any gaps in delivery of parts to the cell as well as any maintenance down time within each shift including any changes in wire spools or downtime from refilling the shielding gas to the cell, etc.

And if you already know how much time it takes to complete one complete cycle in the cell, then it shouldn't be too complicated to see where they're clearly slacking, or there is indeed a logistical logjam within the flow of your materials/components to the cell causing delays also... In other words every aspect of production has to be considered including parts delivery to the cell along with every other factor you already previously mentioned as well. ;)

Respectfully,
Henry

By Metarinka (****)

Date 02-19-2010 21:32

ever gone down the floor and watch the operator (or if possible run the machine yourself for a few hours)? I've found the biggest thing is to run the machine and time yourself and see where the increase in time is coming from.
on all our semi auto machines, We have a theoretical time which is: weld cycle time + load/offload cycle time. this will give me the cycle time in minutes. Then I do any periodic maintenance/ downtime (wire change, tip dressing, consumables etc) These are usually calculated on a per part basis or arc-on time. (wire change every 100 parts, tip change every 30 minutes of arc-on time etc). That will give me a practical part count for a day.

Finally I add in human factor, which is breaks, dropping the screw driver, something gets jammed, etc etc. I've found realisitically part count is usually about 5-25% less of my practical estimation based on the human factor. Line starving, or inefficiency with load/unload are usually the biggest culprits.

As an example we had a practical cycle time (including consumables changing) of about 4:15 seconds for one operation. realistically it would be about 5:00 minutes. The culprit was line starving, and offloading. If the operator didn't stay on top of things and get a good buffer of parts, you would have to stop production to load pallets, plus any rejects took additional time to log and off load.

Just my experience. Also if time estimation is done just on one part and not averaged out, it's easy for people to rush and get it done as fast as possible to make cycle times lower. Very few workers can or will maintain blazing fast part loading, tool switching etc for 8 hours. Would you?

By Duke

Date 02-24-2010 01:57

Are we talking about an 8 hour 'day', or a 24 hour, 3 shift day? Do shift changes (TWO operators BS'ing for ??? minutes) figure into calcs?

By ssbn727 (*****)

Date 02-24-2010 02:05

Good point, as it does changes things significantly especially if they're BSing and there's nobody present during the shift change which would mean that the robot is also down during that time period also. ;)

Respectfully,
Henry

By Lawrence (*****)

Date 02-24-2010 02:28

The only thing I can add to benefit the conversation depends totally on your work group.

This would be operator involvement in the timestudy. If they are go-getters and have a production based attitude they will be the key to accuracy and improved productivity. They know what slows them down (for good or evil) and will probably be the ones to identify the solutions when it comes down to brass tacks.

If they feel invited to help rather than being opressed you will win.

In many places workers feel like engineering/mgmt don't listen to their ideas.. We all know what happens in the long run with this senerio.

Of the relationship is oppositional now, it may be an opportunity to make huge gains if you can get them onboard with a winsome presentation of your plan.

By 803056

Date 02-24-2010 20:14

I read the same post in another forum and I simply have to ask the question. "Why does your robot need a bathroom break?"

It would seem like the robot is taking advantage of the free oil changes more frequently than is necessary. Robbie might be less inclined to abuse the privilege if there was at least a token charge for the service.

I think most of us are on the same page with this one. A visit to the production floor to see the actual operation should reveal the bottlenecks.

Best regards - Al

By DaveBoyer (*****)

Date 02-25-2010 02:21

If You want to see how fast Your operators can work, put them on piece work, but be sure they understand that they only get paid for the acceptable parts.

By Ke1thk

Date 03-01-2010 16:55

Great feedback.

I've done time studies and I know what the robot's program capacity is. I know how much time each non-production deduction is and how many of each is expected. My problem is convincing my boss that my robot is not being maximized. I thought if there were industry "standards" or "benchmarks" that accounted for the time differences (like 10% or 20%) I could better state my case.

A problem is that I work at a paternal organization (family owned). I think my operators are lazy. One is a nephew. I would like to argue for a new set if I had standards or benchmarks from an outside source, like our forum.

Thanks,

Keith

By Metarinka (****)

Date 03-01-2010 17:08

who integrated the system? PreTech, wolf robotics? I would give your integrator a call and see if they have any info about cycle time estimation. Also going down and doing an audit might help or running the cell your self. If you can go watch the worker for 4 hours and all of a sudden productivity jumps up 15% well then someone is taking the easy train.

In my experience after everything is said and done we would take 5-15% off and readjust that as our baseline. I don't think there's an industry standard number, but most people don't actually work for 8 hours a day, linger in a few minutes late from lunch etc. If the cell needs human interaction every cycle for load/offload then these things will eat into it. We had that issue, once we fixed our line starving issue we developed a batch loading/unloading robot which only needed human interaction ever 32 parts. That way if you got distracted for a few minutes the robot would do it's thing with each batch taking 30 minutes to run.