What's new
What's new

The lights out checklist

thunderskunk

Cast Iron
Joined
Nov 13, 2018
Location
Middle-of-nowhere
We did a ton of work in order to… well, prepare to run lights out. Said team hadn’t gotten there by the time I left, but other teams did. I’m on another team that’s much farther from lights out being possible, never mind automated parts loading. I threw together a “checklist” just to see what you guys thought and if there was anything you’d add or remove:

Machining I/O
  • Inputs
    • Power
    • Coolant
    • Way oil
    • Grease
    • Material
    • Cutting tools
  • Outputs
    • Machined Blanks
    • Chips
    • Worn/broken tools
    • Coolant mist
    • Scrap parts
  • Rotating
    • Tool holders
    • Fixtures


Lights out checklist
  • Stable process
    • Measurement system
      • GRRs used to ascertain the percentage of uncertainty of measurements. Tolerance dependent
      • Workmanship standard defined such as dents, surface color, etc
    • Process studies
      • Cp, Cpk, Pp, Ppk requirements defined
      • Characterization study: 100% of the parts for the unattended period are measured
      • Capability study: Performed once for the machine, once for each process and again for each process change thereafter. Sample size such as 4 parts per 100 across the unattended run
      • Tool life studies: the number of parts that can be made on each tool.
        • Create toolpaths which meet unattended goals, balancing life and cycle time
    • PFMEA
      • Identify high-risk steps in process and mitigate
  • Risk mitigation
    • Tool life
      • Usually fixed number of parts such as 80% of life study
      • Tool probing to check for breakage or wear
      • Tool life management: cycle through duplicate tools when one reaches life or probing indicates excessive wear
      • Integrated measurement system inspects part for tool wear and offsets machine automatically/cycles to next tool per tool life management
      • Robot could replace tools in smaller cribs
    • Emergency management
      • Fire suppression system
      • Network monitoring alerts
      • Spindle load monitoring
    • Part loading
      • Probing detects misload
      • Robot preventative maintenance
  • Feeding
    • Staging: hopper, autosaw, or barfeeder
    • Loading: robot or barfeeder
    • Flipping: robot or sub spindle
    • Offloading: robot or conveyor
 

J Gilles

Cast Iron
Joined
Feb 19, 2016
Location
PNW
For part loading, pressure sensors routed through ports on a fixture or chuck are very helpful to confirm if parts are seated correctly.

Spindle load monitoring can also be implemented for tool life monitoring if your loads aren't in the lower ~15% of your spindle power range.

Aside from the direct manufacturing aspects of lights out machining, having a robust and easy to use database of the parts you are making is critical. Ideally it is tied in to your in-process measurements as well as cmm reports and any other recorded measurements you take on the parts. Tying the code back to incoming material is also helpful to decide if it's a supplier issue. Laser marked barcodes on the parts are a great option for high volume production. Every station has a barcode scanner and tracks the movement of the parts through the system.

Once you get the system operating relatively smoothly, finding your new issues quickly is critical. When parts start coming out bad, you need to be able to go back and find the last good parts quickly and narrow down when it started so you can diagnose, learn, and work on eliminating the problem in the future. When all that takes is scanning a barcode to figure out when a part was loaded, machines, which machine it was on, etc you can spend more time improving the process.
 

Ox

Diamond
Joined
Aug 27, 2002
Location
West Unity, Ohio
We did a ton of work in order to… well, prepare to run lights out. Said team hadn’t gotten there by the time I left, but other teams did. I’m on another team that’s much farther from lights out being possible, never mind automated parts loading. I threw together a “checklist” just to see what you guys thought and if there was anything you’d add or remove:

Machining I/O
  • Inputs
    • Power
    • Coolant
    • Way oil
    • Grease
    • Material
    • Cutting tools
  • Outputs
    • Machined Blanks
    • Chips
    • Worn/broken tools
    • Coolant mist
    • Scrap parts
  • Rotating
    • Tool holders
    • Fixtures


Lights out checklist
  • Stable process
    • Measurement system
      • GRRs used to ascertain the percentage of uncertainty of measurements. Tolerance dependent
      • Workmanship standard defined such as dents, surface color, etc
    • Process studies
      • Cp, Cpk, Pp, Ppk requirements defined
      • Characterization study: 100% of the parts for the unattended period are measured
      • Capability study: Performed once for the machine, once for each process and again for each process change thereafter. Sample size such as 4 parts per 100 across the unattended run
      • Tool life studies: the number of parts that can be made on each tool.
        • Create toolpaths which meet unattended goals, balancing life and cycle time
    • PFMEA
      • Identify high-risk steps in process and mitigate
  • Risk mitigation
    • Tool life
      • Usually fixed number of parts such as 80% of life study
      • Tool probing to check for breakage or wear
      • Tool life management: cycle through duplicate tools when one reaches life or probing indicates excessive wear
      • Integrated measurement system inspects part for tool wear and offsets machine automatically/cycles to next tool per tool life management
      • Robot could replace tools in smaller cribs
    • Emergency management
      • Fire suppression system
      • Network monitoring alerts
      • Spindle load monitoring
    • Part loading
      • Probing detects misload
      • Robot preventative maintenance
  • Feeding
    • Staging: hopper, autosaw, or barfeeder
    • Loading: robot or barfeeder
    • Flipping: robot or sub spindle
    • Offloading: robot or conveyor


Well if I had to deal with all of that, I';d prolly just stay in bed. :ack2:

I just try to leave with all tools updated, the magazine full, the coolant topped up, the Bijur full, and the parts counter set for how long I want it to run - IF I want it to stop overnight.

You need to know how stable your process is, and I guess that's what your laundry list above is all about, but seems overkill.... But you also need to weigh the assets and liabilities. What is the worst case scenario if all goes to Schidt? Are you running 2" round 316SS? Or are you running 1/2" A36?

What are you able to gain? Is the machine backed up and you will be able to keep customers happy with sooner dates, and b/c of this your threshold for benefit over risk will move, compared to just getting this job out the door, or getting a second machine - depending on the size/duration of the project.


----------------

Think Snow Eh!
Ox
 

Orange Vise

Titanium
Joined
Feb 10, 2012
Location
California
Didn't read through the whole list but it sounds like you know what you're doing from a planning standpoint.

Now it's time to execute. Lights-out is a culture. It'll take time to figure out, so don't bite off more than you can chew.

For starters, what does the shop currently look like? Do you have operators standing in front of every machine during the day?
 

Strostkovy

Stainless
Joined
Oct 29, 2017
I am no expert on lights out, but from just trying to run certain machines without staring at them I think you'll want sensors for every step of the way. A quick go/no go prox sensor goes a long way to cutting down carnage.
 

thunderskunk

Cast Iron
Joined
Nov 13, 2018
Location
Middle-of-nowhere
So, assuming my team can see this post. I don’t think they will, but it’s always a safer bet. I’m not here to bash them as I genuinely believe they have a solid process; they’ve made it robust, easy to understand and train, no risk of crashing, etc. They don’t want to lose that and I don’t blame them. However… their machines are an early 2000s vintage with severely neglected PMs and are sitting on wood floors. Old controllers, too, so no fancy HSM toolpaths here. When the table moves from right to left, the level of the machine changes. It’s a one-shift operation with 5 mills, 5 lathes, and a whole bunch of manual processes. Tolerances are +/-.005” because they had to tolerance it with something. Mostly held for blending and fixturing the next op, not because of form, fit, or function. They get 6 hours of parts off of their tooling before it must be replaced. Material is tool steel: 1060 or A6. A single, automated 3-shift mill-turn center could replace their entire machining cell, I have no doubt. My goal as best I can is to convince them to go this route while respecting their people and processes.

I took the feedback thus far and updated the list. Ox is right: the statistical requirements are overkill for this scenario; they don’t typically measure parts, and if they do it’s with calipers. That’s left over from my past life, but I’ll keep it as I’m a sucker for some of that six sigma crap.


Machining I/O
  • Inputs
    • Power
    • Coolant
    • Way oil
    • Grease
    • Material
    • Cutting tools
    • NC Programs
  • Outputs
    • Machined Blanks
    • Chips
    • Worn/broken tools
    • Coolant mist
    • Scrap parts
  • Rotating
    • Tool holders
    • Fixtures


Lights out checklist
  • Stable process
    • Measurement system
      • GRRs used to ascertain the percentage of uncertainty of measurements. Tolerance dependent
      • Workmanship standard defined such as dents, surface color, etc
    • Process studies
      • Cp, Cpk, Pp, Ppk requirements defined
      • Characterization study: 100% of the parts for the unattended period are measured
      • Capability study: Performed once for the machine, once for each process and again for each process change thereafter. Sample size such as 4 parts per 100 across the unattended run
      • Tool life studies: the number of parts that can be made on each tool.
        • Create toolpaths which meet unattended goals, balancing life and cycle time
    • PFMEA
      • Identify high-risk steps in process and mitigate
    • Process creation
      • Fixtures designed for loading assurance and wear
      • NC programs created using CAM; stable process replicated no matter part shape or size.
      • Setup documentation
      • Robust print, program, inspection routine, and setup control
        • Kan ban tags for traceability
    • Consumables replenishment
      • Automatic coolant level monitoring and filling
      • Bijur oil with enough capacity to finish run
    • Chip management
      • Conveyor PMs
      • Robust integrated machine wash
      • Spindle-driven chip evacuation (the fan)
      • Drain for chip bucket
      • Chip bucket rotator
  • Risk mitigation
    • Defects/scrap
      • Kan ban tags traceable to back-track quality issues
      • Tie traceability to laser markings if possible
    • Tool life
      • Usually fixed number of parts such as 80% of life study
      • Tool probing and spindle load monitoring (for more than 15% load) to check for breakage or wear
      • Tool life management: cycle through duplicate tools when one reaches life or probing indicates excessive wear
      • Integrated measurement system inspects part for tool wear and offsets machine automatically/cycles to next tool per tool life management
      • Robot could replace tools in smaller cribs
    • Emergency management
      • Fire suppression system
      • Network monitoring alerts
      • Spindle load monitoring
      • Coolant overflow alarm
    • Part loading
      • Probing detects misload
      • Robot preventative maintenance
      • Proximity sensors on vise confirm parts are seated correctly
  • Feeding
    • Staging: hopper, autosaw, or barfeeder
    • Loading: robot or barfeeder
    • Flipping: robot or sub spindle
    • Offloading: robot or conveyor
 

triumph406

Titanium
Joined
Sep 14, 2008
Location
ca
I think your way overthinking this. Looks like you have created 'lights out by committee"

Practice running during day shift unattended. Then as things happen if any, change the process. Then you'll be ready for lights out

When I had a day job, and ran machines in the evening occasionally I'd run the machines overnight. The only issue I had was one machine had a small coolant tank, I'd fill that one to the brim to make sure it didn't run out. And I never ran programs that I hadn't run before.
 

triumph406

Titanium
Joined
Sep 14, 2008
Location
ca
One problem I've run into on a Fadal is running very large programs thru DNC overnight. Quite often I'll get there in the morning and find the machine froze at some point during the program. Haven't figured out whether it was the Laptop, or the machine. I suspect the machine.

The Makino with a Fanuc 0M is rock solid DNC'ing thru the night.
 

Emma99

Plastic
Joined
Jan 21, 2015
Location
Silicon Valley
We did a ton of work in order to… well, prepare to run lights out. Said team hadn’t gotten there by the time I left, but other teams did. I’m on another team that’s much farther from lights out being possible, never mind automated parts loading. I threw together a “checklist” just to see what you guys thought and if there was anything you’d add or remove:

Machining I/O
  • Inputs
    • Power
    • Coolant
    • Way oil
    • Grease
    • Material
    • Cutting tools
  • Outputs
    • Machined Blanks
    • Chips
    • Worn/broken tools
    • Coolant mist
    • Scrap parts
  • Rotating
    • Tool holders
    • Fixtures


Lights out checklist
  • Stable process
    • Measurement system
      • GRRs used to ascertain the percentage of uncertainty of measurements. Tolerance dependent
      • Workmanship standard defined such as dents, surface color, etc
    • Process studies
      • Cp, Cpk, Pp, Ppk requirements defined
      • Characterization study: 100% of the parts for the unattended period are measured
      • Capability study: Performed once for the machine, once for each process and again for each process change thereafter. Sample size such as 4 parts per 100 across the unattended run
      • Tool life studies: the number of parts that can be made on each tool.
        • Create toolpaths which meet unattended goals, balancing life and cycle time
    • PFMEA
      • Identify high-risk steps in process and mitigate
  • Risk mitigation
    • Tool life
      • Usually fixed number of parts such as 80% of life study
      • Tool probing to check for breakage or wear
      • Tool life management: cycle through duplicate tools when one reaches life or probing indicates excessive wear
      • Integrated measurement system inspects part for tool wear and offsets machine automatically/cycles to next tool per tool life management
      • Robot could replace tools in smaller cribs
    • Emergency management
      • Fire suppression system
      • Network monitoring alerts
      • Spindle load monitoring
    • Part loading
      • Probing detects misload
      • Robot preventative maintenance
  • Feeding
    • Staging: hopper, autosaw, or barfeeder
    • Loading: robot or barfeeder
    • Flipping: robot or sub spindle
    • Offloading: robot or conveyor
Thanks for documenting the process, I have a shorter list, but every shops situation and production is different, but understanding the full scope of lights-out manufacturing, is knowing if you are ready for primetime.
 

Springy

Plastic
Joined
Oct 22, 2021
So, assuming my team can see this post. I don’t think they will, but it’s always a safer bet. I’m not here to bash them as I genuinely believe they have a solid process; they’ve made it robust, easy to understand and train, no risk of crashing, etc. They don’t want to lose that and I don’t blame them. However… their machines are an early 2000s vintage with severely neglected PMs and are sitting on wood floors. Old controllers, too, so no fancy HSM toolpaths here. When the table moves from right to left, the level of the machine changes. It’s a one-shift operation with 5 mills, 5 lathes, and a whole bunch of manual processes. Tolerances are +/-.005” because they had to tolerance it with something. Mostly held for blending and fixturing the next op, not because of form, fit, or function. They get 6 hours of parts off of their tooling before it must be replaced. Material is tool steel: 1060 or A6. A single, automated 3-shift mill-turn center could replace their entire machining cell, I have no doubt. My goal as best I can is to convince them to go this route while respecting their people and processes.

I took the feedback thus far and updated the list. Ox is right: the statistical requirements are overkill for this scenario; they don’t typically measure parts, and if they do it’s with calipers. That’s left over from my past life, but I’ll keep it as I’m a sucker for some of that six sigma crap.


Machining I/O
  • Inputs
    • Power
    • Coolant
    • Way oil
    • Grease
    • Material
    • Cutting tools
    • NC Programs
  • Outputs
    • Machined Blanks
    • Chips
    • Worn/broken tools
    • Coolant mist
    • Scrap parts
  • Rotating
    • Tool holders
    • Fixtures


Lights out checklist
  • Stable process
    • Measurement system
      • GRRs used to ascertain the percentage of uncertainty of measurements. Tolerance dependent
      • Workmanship standard defined such as dents, surface color, etc
    • Process studies
      • Cp, Cpk, Pp, Ppk requirements defined
      • Characterization study: 100% of the parts for the unattended period are measured
      • Capability study: Performed once for the machine, once for each process and again for each process change thereafter. Sample size such as 4 parts per 100 across the unattended run
      • Tool life studies: the number of parts that can be made on each tool.
        • Create toolpaths which meet unattended goals, balancing life and cycle time
    • PFMEA
      • Identify high-risk steps in process and mitigate
    • Process creation
      • Fixtures designed for loading assurance and wear
      • NC programs created using CAM; stable process replicated no matter part shape or size.
      • Setup documentation
      • Robust print, program, inspection routine, and setup control
        • Kan ban tags for traceability
    • Consumables replenishment
      • Automatic coolant level monitoring and filling
      • Bijur oil with enough capacity to finish run
    • Chip management
      • Conveyor PMs
      • Robust integrated machine wash
      • Spindle-driven chip evacuation (the fan)
      • Drain for chip bucket
      • Chip bucket rotator
  • Risk mitigation
    • Defects/scrap
      • Kan ban tags traceable to back-track quality issues
      • Tie traceability to laser markings if possible
    • Tool life
      • Usually fixed number of parts such as 80% of life study
      • Tool probing and spindle load monitoring (for more than 15% load) to check for breakage or wear
      • Tool life management: cycle through duplicate tools when one reaches life or probing indicates excessive wear
      • Integrated measurement system inspects part for tool wear and offsets machine automatically/cycles to next tool per tool life management
      • Robot could replace tools in smaller cribs
    • Emergency management
      • Fire suppression system
      • Network monitoring alerts
      • Spindle load monitoring
      • Coolant overflow alarm
    • Part loading
      • Probing detects misload
      • Robot preventative maintenance
      • Proximity sensors on vise confirm parts are seated correctly
  • Feeding
    • Staging: hopper, autosaw, or barfeeder
    • Loading: robot or barfeeder
    • Flipping: robot or sub spindle
    • Offloading: robot or conveyor

Really gotta tie this in to your volumes here too. If these parts are running at high volume (which to me means >10k parts/year) in continuous production full time (ie not in batches) then this level of detail makes sense. If not - or if other systems can get the job done much faster and cheaper, to the extent that it can be entirely or mostly supervised, then a lot of this straight up isn't necessary. We run a lot of stuff overnight/unattended in batches that are usually only running a couple of days total; if that's the kind of runtime you're getting then you're going to lose far more money doing the statistical analysis than sorting scrapped parts. From what you're describing it sounds like even the baby steps of automation (eg put a bar feeder on a lathe) would improve things massively, no need to go from an antiquated process to an unmanned Fanuc factory in one go. Tolerances of 0.005 indicate that the product maybe shouldn't even be machined in the first place?
 
Last edited:
  • Like
Reactions: Ox

thunderskunk

Cast Iron
Joined
Nov 13, 2018
Location
Middle-of-nowhere
Really gotta tie this in to your volumes here too. If these parts are running at high volume (which to me means >10k parts/year) in continuous production full time (ie not in batches) then this level of detail makes sense. If not - or if other systems can get the job done much faster and cheaper, to the extent that it can be entirely or mostly supervised, then a lot of this straight up isn't necessary. We run a lot of stuff overnight/unattended in batches that are usually only running a couple of days total; if that's the kind of runtime you're getting then you're going to lose far more money doing the statistical analysis than sorting scrapped parts. From what you're describing it sounds like even the baby steps of automation (eg put a bar feeder on a lathe) would improve things massively, no need to go from an antiquated process to an unmanned Fanuc factory in one go. Tolerances of 0.005 indicate that the product maybe shouldn't even be machined in the first place?
The team I gathered this list from was about 6M parts a year; 3 shifts, 21 machines, 73 SKUs. All bar fed or blank fed Swiss turns.

New team has 600 SKUs, but they have blanks that reach volumes which fit lights out. And yes, baby steps is what we need current state, but on the reverse I’m trying to tell the story to the boss that there is a nuclear option: buy one new machine with new capabilities and automation that replaces 8 machining centers and call it a day. If we go that route, this is a list of stuff to consider that you may not be thinking about with our antiques.
 

Ox

Diamond
Joined
Aug 27, 2002
Location
West Unity, Ohio
"Blank Fed Swiss"

That's interesting!
Anything you can share on that?
Like maybe a pic of the part?


"buy one new machine with new capabilities and automation that replaces 8 machining centers and call it a day"

I'd be very interested in what "one machine" that you have identified that can replace 8 independent mills?
Possibly a Hydromat Icon 6xxx or similar?





-------------------

Think Snow Eh!
Ox
 

michiganbuck

Diamond
Joined
Jun 28, 2012
Location
Mt Clemens, Michigan 48035
I guess if the lights are out one may not see where to mark the checklist likely a key chain flashlight may be due.

A small boy is sent to bed by his father.​

A small boy is sent to bed by his father. Five minutes later:

"Da-ad..."

"What?"

"I'm thirsty. Can you bring me a drink of water?"

"No. You had your chance. Lights out."

"Five minutes later: "Da-aaaad..."

"WHAT?"

"I'm THIRSTY...Can I have a drink of water??"

"I told you NO! Lights out."If you say one more thing I'll have to spank you!!"

"Five minutes later...

"Daaaa-aaaad..."

"WHAT??!!"

"When you come in to spank me, can you bring me a drink of water?"
 
Last edited:

thunderskunk

Cast Iron
Joined
Nov 13, 2018
Location
Middle-of-nowhere
"Blank Fed Swiss"

That's interesting!
Anything you can share on that?
Like maybe a pic of the part?


"buy one new machine with new capabilities and automation that replaces 8 machining centers and call it a day"

I'd be very interested in what "one machine" that you have identified that can replace 8 independent mills?
Possibly a Hydromat Icon 6xxx or similar?





-------------------

Think Snow Eh!
Ox
No pictures; maybe in a few years when most of the stuff we were making is obsoleted by new tech. Copper 12’ barstock and copper blanks; all parts are .5” diameter and anywhere fro 2 to 4” long. All machines were Citizens of different flavors: L7, L220, and L320, even the blank fed. Swiss machine, but not actually Swiss turning going on with the blank fed machines. We’d even converted a blank fed to a bar fed machine as it was the same platform with little effort. Blank parts looked exactly like barfed parts with pre ops completed for this and that.

As far as the current team: we run one shift, and most of it is spent babysitting two-vise ops and saws. The machines are out of shape, are set at 50% rapids (for good reasons), cut a lot of air, don’t use modern toolpaths, wouldn’t have enough memory even if we did, aren’t level, spindles tapers are worn/dented, tool holders are out of shape, tools have high run out, a few have bent ball screws you can hear make noise once a rotation, the list goes on. I grouped turn center in with machining center, so 4 of those 8 are lathes. Same story, maybe they’re a little better shape. We don’t technically need to turn anything because of tolerance; it’s just for a profile.

My thought is a brother Speedo M300X3 with turn capability and organic robot. Automate it, run it 3 shifts, and we’ll replace 4 mills and 4 lathes easy. Could we fix problems on existing machines? Sure, but our average machine age is 20.
 
  • Like
Reactions: Ox

Orange Vise

Titanium
Joined
Feb 10, 2012
Location
California
They get 6 hours of parts off of their tooling before it must be replaced. Material is tool steel: 1060 or A6. A single, automated 3-shift mill-turn center could replace their entire machining cell, I have no doubt. My goal as best I can is to convince them to go this route while respecting their people and processes.

A DMG Mori NTX with barfed load and robot unload will not only replace the entire cell, but if spearheaded by a single, competent person, said person will displace the entire team.

This has deviated quite far from the original post. This goes way beyond figuring out a lights-out checklist.
 

Chips Everywhere

Aluminum
Joined
Apr 29, 2021
I don't have any input on running lights out, as I have enough problems running during the day. But I am curious, one of the things that requires a significant amount of operator input while running my parts is small chip management. Not through the auger, but the chips that come out of the coolant return, although my Haas auger likes to move the chips around inside the enclosure vs removing them from the enclosure. I fill up the return coolant filter tray every 30-45 minutes and risk overflooding if I don't empty it. How does one plan for that or address that issue if trying to run lights out?

Is the problem that it's a Haas? 🤔..... couldn't help myself. 😃
 

Orange Vise

Titanium
Joined
Feb 10, 2012
Location
California
I fill up the return coolant filter tray every 30-45 minutes and risk overflooding if I don't empty it. How does one plan for that or address that issue if trying to run lights out?

Is the problem that it's a Haas? 🤔..... couldn't help myself. 😃

Actually yes, this specific issue is a Haas problem. However, chip issues in general aren't by any stretch confined to Haas machines. There are plenty of higher end machines with chip removal problems, some of which are far more serious than the Haas chip basket issue.

For starters, the job has to be paired with the right machine. In your case, your machine could work for running smaller parts with less material removal, but probably not the parts you're currently running in a lights out setting. You'd probably want a different machine, maybe a horizontal that does a lot better with high volumes of chips.

Second, lights out can mean a lot of different things. It could mean 100% unattended production, or it can simply mean that the machines run without the supervision of highly skilled labor, but lower skilled maintenance personnel can check up on the machines every few hours to manually clear chips, swap out chip bins, top off coolant, etc.
 

Strostkovy

Stainless
Joined
Oct 29, 2017
One thing that is overlooked in my opinion is partial lights out operation. Instead of trying to get machines to run all night, just let something easy run for a few hours after everyone goes home and then stop.

An example of this is loading a sheet on a laser, hitting go, and leaving. An extra 2 hours on a 10 hour shift is a pretty solid increase in productivity, and you don't need nearly as much setup.
 








 
Top