What's new
What's new

What limits automation in the machine shop?

pmtool

Cast Iron
Joined
Jul 10, 2011
Location
Portland, OR
I mean of course most of the machines have been automated for years, at least in their metal cutting processes, but I am still opening machine doors, loading, unloading, blowing off parts, measuring etc.

This is of course in small to medium job shops and even production shops.

Much of what operators do today is boring and repetitive.

I could see a robot at every cnc machine for anything but one off or very low quantity work. Especially now that there are robots that don't need cages and can work around people. Even a robot that would open the door, blow off the parts and unclamp and vise would make things faster and simpler for an operator.


So why do you think there is not more robotics in an average shop? Money, complexity of integration, fear that machines will take over jobs?

Will we see a major shift in this in say the next 20 or 30 years?
 
What? ROI!

I mean of course most of the machines have been automated for years, at least in their metal cutting processes, but I am still opening machine doors, loading, unloading, blowing off parts, measuring etc.

This is of course in small to medium job shops and even production shops.

Much of what operators do today is boring and repetitive.

I could see a robot at every cnc machine for anything but one off or very low quantity work. Especially now that there are robots that don't need cages and can work around people. Even a robot that would open the door, blow off the parts and unclamp and vise would make things faster and simpler for an operator.


So why do you think there is not more robotics in an average shop? Money, complexity of integration, fear that machines will take over jobs?

Will we see a major shift in this in say the next 20 or 30 years?

Though in the long run, every automation effort is rewarded, if the single job you are doing now does not justify the payback, it won't be done.

Under capitalization is the most common reason for business failure.
 
What limits automation in the machine shop?

I do.

Automation and cybernetics are the most childish achievements of the past 200 years!
 
So why do you think there is not more robotics in an average shop? Money, complexity of integration, fear that machines will take over jobs?

Will we see a major shift in this in say the next 20 or 30 years?

I think the main reasons are 1) Lack of understanding, 2) Lack of technical resources. 3) Lack of Technical Employees.

Automation and robotics, while reducing the number of labor jobs (button pushers), does generate requirements for "Higher Technically Skilled" employees. Someone has to set up, run and troubleshoot the robotics and automation systems. Those higher technically skilled employees are more difficult to find and require more training. They are also require a higher pay scale. Not much different that trying to find good machinists actually. The overall cost is much lower for the business though when implemented successfully. Getting from point A to point B is a difficult transition though and most smaller employers simply don't have the resources available to do it.
 
Well..
Its mostly fud, plus the fact that robot automation stuff is 98% sold to big corps.
Thus, the robot side is not used to dealing with the small shops, and has a high incentive to keep the extra costs piled up.

Automation is actually pretty easy .. potentially.
A lot of it is made harder than necessary, so that the current customers can keep making the big bucks on the integration projects.
 
I barley have time to program set up and keep 5 cnc's running never mind dicking around with a robot. I guess if you have enough help on the floor and free time to step up to this it would eventually become second hand and just another step in your set up process. I don't know how I would handle it and maybe I'm just being old school. I would be forever worried about parts being loaded accurately and fixtures being free of chips. How many times have you blown out you vise and fixture only to find chips still sitting on a parallel as you load the next part. I can just imagine walking away from my robot and coming back an hour later to a pile of junk. Now if I had a nice job with parts in the thousands that we would own for a few years I would jump at the chance to automate it properly. But for short run stuff my time and money is best spent on other things including extra man power.

Ron
 
I never make any real money on one time jobs. If a customer orders a high enough quantity of a part that I can achieve real efficiency...chances are it will be a repeat part. If it repeats automation is a no brainer.

What's stopping me is inertia. The first robot in the shop floor will be a huge time suck in terms of learning curve, I know it will pay off but I'm busy making parts right now. A job will come along where the obvious benefits make it irresistible, then hang on.

My business consists of operations that have already been largely adapted to robotic automation, it's only a matter of time.
 
Mostly in agreement with Overflow, with some additional points: The setup of automation for part-handling, and running a machining center in the way it is normally done, would likely be very time-consuming in the typical job shop environment, due to the hand-eye-brain coordination that is brought to the operator function by a human. While there has been a lot of improvement in the visual object recognition technology in recent years, the fact remains that for doing 5 lots of 100 pcs, and 2 lots of 250 pcs, all different in their physical characteristics, and all needing to be processed through the shop in the course of a week through several machines, the setup and programming requirements for the robots or automated part handling would be quite substantial. There is little in the way of intuitive, automatic orientation recognition that is available in the machine vision world as I perceive it currently. The issues noted by Overflow specifically about the part-clamping vs. vise cleanliness (as an example of detail) would be a major impediment with current technology, I imagine.

Long story short, a typical job shop that is going to run machining operations on job-lot qtys would be a major investment from day one, and would need to be set up that way from the beginning, I suspect. It might look very different from what we see in shops today, and would require some deep pockets to get it operating.
 
I think the main reasons are 1) Lack of understanding, 2) Lack of technical resources. 3) Lack of Technical Employees.

+1 To the above.

To lack of understanding I would add applications. I think there's an assumption that automation belongs on just one end of a spectrum. You don't have to be a high tech cutting edge machine shop to find ways to use automation. I'd love to hear about more examples of shops breathing new life into old iron. Although not necessarily closed loop control systems, I still see these as examples of automation for the old-iron shop: Installing DRO's on old iron, upgrading drive motors with VFD's, utilizing low cost PLC's (make a simple weld positioner with gun control), and use of bar-codes to track stock or jobs through a shop.

Automation doesn't have to mean a bunch of Roomba's running around sweeping up chips while Wall-E loads and unloads your CNC turning center. There's lots of room on the spectrum to apply this stuff. Would love to see some more examples of using it in shops with older tech.
 
First of all full-disclosure, I own a company that makes robotic machine tending cells (VersaBuilt) and a motorcycle clutch manufacturing business (Rekluse Motor Sports). There is robot technology that absolutely works in low volume production. We do it everyday at Rekluse. In my biased opinion I believe our machine tending robot is the best solution but there are other machine tending solutions available now that absolutely can work in short run production and improve your business.

For most machine shops that run even small production jobs (50+ pieces), the operator, the guy loading parts in and out of the CNC, is the most expensive resource in the shop. Operators also limit capacity to produce parts; even running two shifts a day, a robot can easily out produce two shifts at a fraction of the cost.

In my opinion, there are two types of robot systems that make sense for short production runs: the robotic pallet loaders (Erowa, Lang and others) and the VersaBuilt system. The key to all of these systems is the in-feed/out-feed and robot programming is entirely pre-engineered. The advantage of the robotic pallet loaders is that chip management and work holding is simple; about the same as typical operator tending. The disadvantage of these systems is work holding cost, setup time and the labor required to load and unload parts. If you want to run 50 parts, then 50 different pallets with work holding for the new part need to be setup. Each part must be loaded by hand for each operation and cleaned after each operation by an operator. Work holding on these systems can easily exceed the cost of the robot itself.

The VersaBuilt system uses soft jaws to load and machine parts. The advantage of the VersaBuilt system is inexpensive work holding and its ability to process a part through multiple CNC operations without operator intervention. The primary disadvantage of the VersaBuilt system is that soft jaw design is a bit more difficult. In addition to holding the part in the vise for CNC machining, consideration needs to be made in the soft jaw design for picking and placing the part on the shelf. Chip control is another consideration; the VersaBuilt system allows each operation to have a unique CNC wash program. One the basics are understood, work holding design with VersaBuilt soft jaws is not significantly more difficult.

My clutch manufacturing business has 5 robot cells running in our shop now and we will have 8 robots running by the end of the year. We installed our first robot about 18 months ago. We frequently get over 140 hours of CNC cycle time per week on our CNCs tended by robots. Our most productive non-robot CNC rarely gets above 40 hours of CNC cycle time. We have about 10 different part families we run on the robots with about 50 total part numbers robotically tended to date. Our average batch size on the robots is about 25 pieces and we typically load the robot with 2 or 3 different part numbers simultaneously. It takes about 1-2 hours of operator labor per day per robot to keep the CNC machine running 24 hours. Setup time for a job is typically less than 15 minutes.

My point is that robotics are totally achievable for the average shop. You don't need production jobs with orders for thousands of pieces to make robotics work. If you think you're too busy to make a robot work, I would suggest that getting a robot will give you way more time.
 
The Versabuilt list for 89k, not sure if there are dealer incentives or whatever that can/would apply. We are just starting to get into production machining here, I have mostly worked job shops so it s a big transition/learning curve for all involved. I think if I suggested we buy a 90k robot they would laugh me out of a job. 90k buys a new machine, or a crap load of tooling... If we spent the 90k (lets say) on a new machine they would have me focus on staggering ops or whatever to keep them both running. If we buy a robot we have to learn all the programming/setup, verify that we have parts that would run all night (no broken tools, no tolerances going out of spec, etc). I completely believe you (youngwerth) are seeing a ton more spindle time, but you also mention you have some families of parts. I guess depending on what you make would determine how valuable a robot load system would be...
 
There is robot technology that absolutely works in low volume production. We do it everyday at Rekluse. In my biased opinion I believe our machine tending robot is the best solution but there are other machine tending solutions available now that absolutely can work in short run production and improve your business.

I'm not trying to be a kiss-ass, but based on my passing familiarity with Versabuilt/Rekluse I think one of the impediments is lack of companies like that one actually making a serious, creative effort to bring new thinking and new solutions. Because I see that changing, I see significant strides in automation coming in the next 10-15 years.

It's a challenge to nail the right combo of business sense and engineering ingenuity to be successful in this industry, but I'm seeing HUGE gains in the number of start ups in machining and manufacturing. Not sure if it's post-recession desperation, a new wave of technical-minded workers, a "back to the land" style nostalgia for made objects, or just the normal ebb and flow of industry, but it's pretty exciting to watch it happening.
 
From my view point it is a lack of the knowledge to do it, combined with lack of will to pay someone (outsource) to do it, and a lack of patience to allow those already employed to learn to do it. Where I work there are different "tiers" ......not that things do not cross from one tier to the next.....but the people who would most likely get involved in these sorts of projects already have 6 or 8 other projects waiting to be done. These people (Engineers, Programmers etc...) are often considered overhead and/or non-value added.........depending on which manufacturing fad of the month is implemented at your shop. There is alot of resistance to pay someone to work on a project that may take several months to get working vs someone making a part 10 seconds quicker this week.
 
Much of what operators do today is boring and repetitive.

Much but not all.
It's that 5-10% that keeps a person on the job.
So many things that seem simple to a person's fingers that robots do not do well.
Reach around behind your back, pick up a randomly oriented part, spin it into the right orient and stick in the vise.
Simple and boring for a human, yet try to give this to a robot.
Pick parts from a bin and stuff them into a collet with a few thou clearance. A six year old can do it.

Both of the above can be done with a robot but not as easy as telling a low paid person to do it.
Robots get very upset if the part is upside down, people "get" this problem easily.
Knowing and solving these things with automation require higher paid and skilled people and added structure in the work environment.
Now the parts have to be sitting right side up and oriented.

How about changing a CNMG-432 insert in a holder. Want to train a robot to do this simple task?
Opening the box and grabbing the insert alone is going to produce some problems.
Yet you can train a human to do this in less than 10 minutes.

Automation can not think and my opinion is that we will all be long gone before it even starts down this path.
This skill will keep people in the loop for a long time.

Before too many jump up and down I should say that I am a big fan of robots and automation and have been beating my head against the AI wall for over 30 years.
At one time it just seemed to me that this would be easy and gosh, if I just had more CPU power and 2 meg of RAM ............
I do live where the highway signs still say "Welcome to Automation Alley".
Bob
 
The Versabuilt list for 89k, not sure if there are dealer incentives or whatever that can/would apply. We are just starting to get into production machining here, I have mostly worked job shops so it s a big transition/learning curve for all involved. I think if I suggested we buy a 90k robot they would laugh me out of a job. 90k buys a new machine, or a crap load of tooling... If we spent the 90k (lets say) on a new machine they would have me focus on staggering ops or whatever to keep them both running. If we buy a robot we have to learn all the programming/setup, verify that we have parts that would run all night (no broken tools, no tolerances going out of spec, etc). I completely believe you (youngwerth) are seeing a ton more spindle time, but you also mention you have some families of parts. I guess depending on what you make would determine how valuable a robot load system would be...

It's not that big of a learning curve. With systems like the VersaBuilt system or the Erowa-type of system, there is no programming. You can easily be making parts the same day a VersaBuilt system is installed. We will also do a turn-key design on your first part free of charge.

In any automation project, I highly recommend a CNC tool setter (checking for broken tools). I also suggest a tool probe (automatic tool wear offsets on critical features, detecting chip problems) but honestly we're not probing all of our parts. Both are very easy to program.

A system like ours will make sense if you are running the same job more than once in quantities greater than about 5 pieces. Yes, you invest a little more time designing and proving the work holding for a new part. The first part will take twice or even three times as long to prove out. After a couple of parts, you get comfortable with the process. Once your up to speed, it really takes very little extra time if any to prove out a new part. And each time you run that part after that first time you are making bank.

A $90K robot in front of a $90K CNC will out perfom 3 CNCs with 1/5 the labor cost. I just walked through our shop. All the operators left a couple of hours ago. I've got 5 machines running and all 5 will still be running in the morning.
 
We have a pretty fair bit of automation here. There are a few factors that make it difficult.

First, its just straight up too expensive and time consuming to set up for the single man shop. So the "one extremely clever owner" case, which could be well served running serious automation once it is in place, can't get there easily. Plus automated volumes cry out for a support staff, and then you're not a one man shop anymore.

For the medium sized shop is very difficult to significantly exceed the technological average for your area. This is where we struggle. Unfortunately, there aren't a ton of other companies around here doing things like we do. Each of my operators has more responsibility than average, because each of their skills is being more highly leveraged by the automation. Even paying well, there just aren't a ton of people out there. We basically only train new employees from scratch, and that's slow.

Large shops are already either union, in a profitable/comfortable niche or automated now.

All the automation in the world doesn't help without a man who knows how to run it. That limits how fast it can grow.
 
I think the key is shop size(small shops 3-25 employees total) aren't gonna benefit much if the people aren't properly trained. I saw this in shop with a production job that was given to a robot, and because there was no measurement built into the system, and no one checking parts, they went out of tolerance, and one lucky machinist(me) got to spend a full day fumblefuckin around with those parts to finish them thru their remaining ops to sell them.

Medium shops(25-50) might better benefit, but again training is key. It's also gonna take training(or perhaps thought re-education) of the employees to make them feel "good" about a robot that is taking away an excellent entry level job to the machinist field.

Personally I find it hard to believe that a repeat job that shows up 4 times a year with only 20-25 parts per time, is gonna make it worth while to automate with robots, or even a manual pallet changer.
 
A system like ours will make sense if you are running the same job more than once in quantities greater than about 5 pieces. Yes, you invest a little more time designing and proving the work holding for a new part. The first part will take twice or even three times as long to prove out. After a couple of parts, you get comfortable with the process. Once your up to speed, it really takes very little extra time if any to prove out a new part. And each time you run that part after that first time you are making bank.

A $90K robot in front of a $90K CNC will out perfom 3 CNCs with 1/5 the labor cost. I just walked through our shop. All the operators left a couple of hours ago. I've got 5 machines running and all 5 will still be running in the morning.

It takes that long to make five parts? Must be complicated....
 
...How about changing a CNMG-432 insert in a holder. Want to train a robot to do this simple task?
Opening the box and grabbing the insert alone is going to produce some problems. Yet you can train a human to do this in less than 10 minutes.

Automation can not think and my opinion is that we will all be long gone before it even starts down this path.
This skill will keep people in the loop for a long time...Bob

Just like any other component of lights-out machining (in-process gauging, breakage detection, alternate tools) the loading/unloading robot represents a large investment. Automation implies economy of scale. Once you're into that, you're not really a job shop.
 








 
Back
Top