Job shop KPI Pegs

Longtime lurker first time caller.

Where do you peg your KPI’s?

I am attempting to put some context around a few touch points in our shop.

Example.

For 10 hour days we are averaging 3 hours and 12 min of feed time across 7 machines with 6 operators.

That leaves 6 hours and 48 min of time spent hunting for tools, set up, figuring out how to hold the thing, waiting on a program from the programmer, waiting for QC to return the given part to the machine and a host of other inefficiencies and time suck.

My question...for those that track this and other metrics...what is good?

It kind of sounds like you have 5 extra machines and operators.

kustomizer said:

It kind of sounds like you have 5 extra machines and operators.

Click to expand...

Ha BINGO!!

Since time immemorial:

"work expands so as to fill the time available for its completion"

All Parkinson did was write that and similar sets of historical fact down on paper!

Every Manager since Noah has had to deal with the issue.

USACE once had a formal test question that went into five pages of minutiae as to the what and how a Company was FUBAR in every way it could possibly BE FUBAR, and few that were not even possible.

What should you do as the newly appointed Company Commander?

The "School answer" was a brutally effective five Mike Foxtrot WORDS!

"Kick ass and take names!"


Go, thou, and do likewise!

CAVEAT: Whomever did the measuring is on that list, too!
Wouldn't be the first time that part of the tasking had gone wrong, would it?

Sheesh that is awful. Something is not right < understatement of the year!

We have 2 setup/run guys that keep 5 machines going pretty steady. Sure there is down time while trying to get 2 or 3 setups going, but for the most part, at least 3 of the 5 run all day....

edit: Also, IMO, the programmer should be detailing fixturing, or vise holding, size of parallels, etc. That is what I do here. Our setup guys rarely have to even figure on parallel sizes (for new programs from me). We do have some old legacy stuff that is not nearly as detailed that causes some trouble though.

Feed time. Is that coming from the control, like tool life monitor or something?
That doesn't monitor rapid time, tool change time, load/unload, etc.
What are your average cycle times? If you're poking lots of holes your feed time will be less than if your 3D-ing a mold cavity.

You're going to have to establish an acceptable baseline based on your unique situation and make adjustments from there.

To little information.

Are your parts all one-of or production? How complicated are the parts? How fussy are your tolerances? How large are your parts?

On the surface you aren't keeping your machines very busy. But machine spindle time is cheap. Are your operators busy doing tasks that pay, or are your operators wasting time chasing tools and materials and waiting for others to do their job?

It sounds to me like the shop is a poorly organized mess. Fixing that is above the pay grade of a machine operator or set up man.

This is one of a dozen or so things I’ve started tracking over the last 4 or 5 weeks.

The most abysmal looking one is the feed time #s. Cycle start time is probably 10% more than feed time across the board. I don’t have it in front of me at the moment.

Our feed times on 40 hours a week have gone.

32% last week
44%
43%
34%
41% first week tracking

The spread on that first week was 5 hrs 12 min on the worst performing machine and 31 hours on the best.

It tightened dramatically last week.

We run around 100 jobs a year. Real low quantity stuff for the services or large primes.

I’m tracking these things because I realized that the outcomes for direct labor on jobs were completely all over the map.

Which is partly because the guy bidding labor is different from the guy programming. Programming guy has been at times has high as 7x the programming set up and run times of the guy bidding labor (his boss).

And it goes on and on.

This is what happens when you move from gut sense to tracking and logging everything I guess!

All super revealing threads I’m pulling on.

You need to make sure your KPI is congruent with the goals if the business. Too often people look for off the shelf KPIs and try to hold the business accountable to those and it just doesn't work.

For example, we are doing a lot of super complicated prototype work. But we also have a 5th axis with a pallet changer. A feed time metric for the shop would get me nowhere. Instead, the prototype machines are held to set up time reduction metric (we measure set up time weird), and the pallet is held to 140 hr/wk cycle time metric.

Make sure the KPI fits the business goals.

As mentioned earlier, you are providing too little information. Those numbers could be great or abysmal depending on circumstance. To me you sound like a manager trying to apply the latest management system that might not even have a chance of producing positive results in your situation and is a 100% waste of time.

I just looked up KPI on Wikipedia, it appears to me unless you are comparing apples to apples in a manufacturing environment the numbers are going to be meaningless. Unless you are an OEM, or a job shop with frequent repeating jobs of the same quantities please tell me how KPI is anything more than fun with numbers.

I will give you a good example of meaningless numbers even in the same department at an OEM. Decades ago I worked at a large aerospace connector manufacturer. They seemed to love trying out the latest fad in management systems, but they always tracked employee efficiency. In my department, contact primary those numbers were based on the numbers of parts produced compared to what 100% up time would produce. Contact primary was divided among 2 machine groups, Swiss cam automatics, and CNC Swiss.

On the Swiss cam side of the aisle there was a group of 4 machines that had auto multi bar feeders and ran the same family of parts out of brass and leaded nickel copper for the whole 7 years I was there. A set-up usually involved adjusting a kicker cam. Most parts were either double ended pins, or a crimp pot pin, very simple parts that would often run 3 straight shifts without even a tool sharpening or even a diameter adjustment.
Since those were the easiest group to run they were usually assigned to the weakest machinist on each shift.
On first shift a guy named Emilio ran them and on second shift it was Frank the whole time I was there.
The numbers on those always came back in the 90's week in and out.

On the flip side in difficulty of the 40 or so cam machines were the MS-7 Tornos. They had all the bells and whistles, full attachments on the front and back working. Even a top guy took longer than a shift to set one up even if that is all he did. Due to the complexity of the parts it was rare for one of them to run a trouble free shift. The average job on those ran out in about two weeks. Efficiency averaged in the 60's to low 70's in that area. Of the cam machine guys only the top men worked on the MS-7s.

Then there was the CNC Swiss area where it was rare to have a job stay on a machine for more than a couple days, many took longer to set-up than to run. Also often that area got hit with so many hot short run jobs there were more machines on set-up than set-up men. A machine without an operator earned a big fat ZERO for efficiency and that hit was divided among all the guys assigned to that area. CNC Lathe efficiency was usually in the 20's to 30's. Since in that era like myself CNC guys were ex cam auto guys, the CNC Lathe guys were the most valuable employees in the department as they could be used in a pinch in the cam auto side of the aisle.

The department was pretty much run by two lead men, myself over the CNC side and a guy over the cam machines.
The guy on the cam machine side could not get the hang of computers so I dealt with most of the data keeping for the whole department. I was always over joyed when I received last month's efficiency report and told to look it over for a meeting all department heads must attend in the auditorium tomorrow.

I always was hit with the same old crap, it was like upper management had no common sense and no memory retention. They did seem to get it through their thick skulls why the CNC Lathe area would put up low numbers as they knew we did a lot of prototypes and specials. The cam machine area was a different story, it seemed they shuffled managers around quite a bit so in those meetings we were answering to the clown of the month.
I don't know how many times I had to explain the Frank and Emilio were the lowest skilled in spite of their "numbers" and years with the company, while the low man Joe, was one of the best in that area, his low numbers were a direct result of being given all the tough jobs. I told them if them if Frank was assigned the jobs Joe had his efficiency might not crack single digits.

Justaguy11 said:

This is one of a dozen or so things I’ve started tracking over the last 4 or 5 weeks.

snip

Which is partly because the guy bidding labor is different from the guy programming. Programming guy has been at times has high as 7x the programming set up and run times of the guy bidding labor (his boss).

snip

All super revealing threads I’m pulling on.

Click to expand...

Just curious, who is closer, the bidder or programmer?

Couple examples, one shop I worked at quoted all our jobs at solid carbide drilling speeds and feeds..... of which we had ZERO.

Programmer might be too conservative with speeds feeds depth of cut, OR he/she might realize, oh this thing is going to need 2 fixtures and soft jaws designed and programmed.

Also worked at a shop where almost everything was underbid. The owner (and quoter) just assumed everything was easy, or we were a bunch of lazy asses milking the jobs. Same guy quoted a job (I will never forget) for something like 40 hours. We had a discussion and we basically concluded (we were told they had it done somewhere else and the shop couldn't do it right in the end) we could charge a premium more or less because work like that was kind of a specialty for us, and mostly went to the same customer, of which was the 40 hour job. First day guy is setting it up, it was something like 8x8x20. He gets like 2 hours in, and the boss was like "how's that job coming? What's taking so long?"

Mike1974 said:

Just curious, who is closer, the bidder or programmer?

Couple examples, one shop I worked at quoted all our jobs at solid carbide drilling speeds and feeds..... of which we had ZERO.

Programmer might be too conservative with speeds feeds depth of cut, OR he/she might realize, oh this thing is going to need 2 fixtures and soft jaws designed and programmed.

Also worked at a shop where almost everything was underbid. The owner (and quoter) just assumed everything was easy, or we were a bunch of lazy asses milking the jobs. Same guy quoted a job (I will never forget) for something like 40 hours. We had a discussion and we basically concluded (we were told they had it done somewhere else and the shop couldn't do it right in the end) we could charge a premium more or less because work like that was kind of a specialty for us, and mostly went to the same customer, of which was the 40 hour job. First day guy is setting it up, it was something like 8x8x20. He gets like 2 hours in, and the boss was like "how's that job coming? What's taking so long?"

Click to expand...

Pardon me as 25 years self employment in a small shop, the last ten years with data capped internet, the last trade show I attended was Westec after the Rodney King riots, so maybe I should change my name to Bob stuck in 1995. All that being said I could not imagine the person bidding the work not consulting the programmer and or the lead machinist on a cycle and set-up time estimation. I am a cut and paste programmer, so I have no idea how much time would be involved using CAD software to estimate a cycle time. The last place I worked for the man was a job shop, the mill programmer used GIBBS, for the lathes I was Mr Cut & Paste. Both of us kept a part library. Mine fit in a couple drawers as nothing was bigger than 20mm diameter. The owner/job bidder always consulted us on cycle times, it was pretty easy to find something similar to use to give him a quick answer. I think some times old school quick and dirty works good enough.