Laptop Controlled Setworks

[Old Greenhorn]

And I just went and bought me some new spandex today. Loading the video now.

Buying it is one thing, showing it off is another and ain't nobody ready for that.
 Thanks for putting yet another indelible picture in my head.

[Andries]

Jake, now you're really just tryin to break the interweb are'ntcha? 

[barbender]

Customsawyer...the man, the hat, the...spandex?🤔😂

[cxsmith]

Nothing wrong with running a through and through sawing pattern: In small logs it's one of the more effective ways to achieve decent recovery and so long as you understand the limitations of it and your logs and the desired product are okay it's a useful technique to know.

Provided you use it right. Most people do not know how to use it right. Or maybe it's the only one saw only one step attitude. Sawyers with a headsaw and resaws see logs differently because the headsaw can be used to prepare material for the resaws to do the work



 

That there is also a through and through pattern and I use a modified version of it quite regularly ( modified= remove first 2" board, roll and remove second 2" board, split straight up the pith and then saw out the wings for inch to edger). I find it yields high grade, stable, quarter sawn feedstock out of little solid logs and while I'd rather not have to saw those logs they're part of what we get and being able to make money from them is important.

EDIT: Yes every one of the 4 x 1's will turn into a banana... but thats why we overcut them as 1" x the limit of the log and then feed them through an edger. 4 x 1 is the target size post edger, not the size the mill cuts to

Can I pick up one of those hawt scantily clad assistants now???

I couldn't find "banana" in woodopedia but I'm assuming that's the end result of cupping? If so then do sawyers add extra height to compensate so they can get the right real dimensions after edging? Do the big mils do that or would they be unwilling because that would screw up their stickering so they'd rather just take the loss?

Adding a no-go region for the pith wouldn't be too hard, but too bad I probably couldn't reuse the code from check avoidance. Getting computer vision to recognize how far it extends might be the more challenging part.

This thread is what makes FF great.   The breadth of knowledge and willingness to share helps a new guy begin sorting out the prior knowledge in the industry.

We've gotten off into one of the areas that appears to be challenging to the OP - that's an understanding of the process and techniques of sawing logs into lumber.

With that said, I believe the OP has considerable homework to do for his software to be successful.   He needs to determine if the software/setworks/optimization that's out there can be improved with his innovations. Without a comparison and evaluation, the new software could be behind what's already there. On the other hand, it could be major advancement. There is room for improvement.

Another piece of the homework is learning the sawing process, the characteristics of logs and what makes optimization challenging. It's impossible to do process improvement without a strong understanding of the process.

The third piece is developing relationships in the industry.  This may be the most difficult in a highly competitive industry that is wary of newcomers of any sort.

Yeah, the fact that there's room for improvement is a big motivation. Ideally it would be in the softwoods since AFAIK that's the biggest market, but I bet that also has the most work already done for optimization.

...Once you see the inside of the log you have a better idea of what you can produce. But you can't X-ray the log before you start.

Actually they're working on that: https://microtec.us/products/ct-log/

[Magicman]

"Banana" can be severe crook or bow and also a combination of both.  Simply think about how a banana peel looks, or even a propeller.

Because of the different stresses within a log, removing flitches and boards is very often not a simple matter.  The log is the boss, not a computer program nor setworks.

[beenthere]

cxsmith

Quote from: Ianab on Yesterday at 04:44:57 AM

...Once you see the inside of the log you have a better idea of what you can produce. But you can't X-ray the log before you start.

Actually they're working on that:

"working on that" is the crux of the comment.  Some 55 years ago ('68 ) and since,  recall similar graphic images and predictions for computerized sawing of logs and also breaking down softwood and hardwood lumber into production of clear parts and cuttings. All efforts, one might say, to get support for research funding aimed at the gathering of digital information to plug into the software to make the decisions that then might control the breakdown (sawing, laser, chipping, etc.) of the wood material.

The graphics are getting better, and little by little the machines to handle logs being processed at high production rates and innovative turning of logs as a result of improved digital shape information are being justified. 

But the actual value of the product and what even 10% improvement in recovery will, at the moment, not justify the expected expense of CT. But there will always be dreams of change. 

These high production mills now process softwood at a very high throughput. Based on the shape of the log, the opening face and iterations of sawcuts made and on through the system to spit out lumber in many forms. Once through the system, just grade the output and assign the best use, be it structural dimension material such as 2X . Some of that ends up in machine-graded engineered lumber products, while some of the remaining dregs at the bottom is sold at HD-like box stores. 

But good luck to cxsmith to meet the challenge of writing software for the best information available. 
Enjoying the comments, but hope cxsmith will break out only the significant parts of the quotes rather than including the lump sum.

[cxsmith]

But the actual value of the product and what even 10% improvement in recovery will, at the moment, not justify the expected expense of CT. But there will always be dreams of change.

These high production mills now process softwood at a very high throughput. Based on the shape of the log, the opening face and iterations of sawcuts made and on through the system to spit out lumber in many forms. Once through the system, just grade the output and assign the best use, be it structural dimension material such as 2X . Some of that ends up in machine-graded engineered lumber products, while some of the remaining dregs at the bottom is sold at HD-like box stores...

A 10% increase in yield means that you buy 9% less logs. Being generous and assuming that logs cost only a quarter of your revenue that's a reduction in cost equal to almost 2.5% of revenue. If your mill produces 100 MMBf/year at $500/mbf that's $1.25MM saved per year, so tens of millions of dollars of value. I think that an x-ray machine could easily be made for less than that.

"Banana" can be severe crook or bow and also a combination of both.  Simply think about how a banana peel looks, or even a propeller.

Because of the different stresses within a log, removing flitches and boards is very often not a simple matter.  The log is the boss, not a computer program nor setworks.

Looking at the output of already existing software like this it seems like the optimizer has a plan. So if the log is the boss then why does this software exist in the first place?

[Magicman]

No argument from me sir, but you were the one asking questions about bananas.  ??

[SawyerTed]

Here is a link to an article that discusses 7 identified lumber recovery factors.  A mill has to focus everyday on these seven, among them is sawing decision making.  

https://www.fpl.fs.usda.gov/documnts/fplgtr/fplgtr39.pdf 

There are six other factors of which optimization is tangentially connected to only a couple for example log quality and target product.  

One big item is the recovery rate downstream which @cxdmith 's linked article  gives some time to.  

The pallet mill in the article replaced a 40 year old manual line with an USNR optimized line including an optimized end dogging overhead carriage head rig, gang and edger with optimization.  

Finding 10% more recovery in an optimized head rig will be hard to do.  Finding 25% recovery in the manual pallet mill doesn't surprise me.  To me that points to sawyer skill level and need for production speed to keep everyone downstream working. 

[Ianab]

Here is a link to an article that discusses 7 identified lumber recovery factors.  A mill has to focus everyday on these seven, among them is sawing decision making.

Also coming into the equation is the balance of production speed (some patterns just take longer to saw), vs Volume recovered vs Value recovered. 

Quarter sawing is a good example of that. It's slower production, and lower recovery, but it can produce more value, IF you have the market for that wood (and the log is suitable quality). 

The decisions also have to factor the mills current stock and orders into it's optimisation. It's all very well deciding that 2x12s are worth more then 3 x 2x4s, on paper. But if the current orders are for 2x4s and you have a big stack of 2x12s on hand, that changes the decision making as well. 

What a computer program is good at is very quickly running a series of simulations and working out what is probably the best solution for that log, based on a variety of factors. How you weight them will vary. But a large commercial mill might be operating on a 10% profit margin? So at that level, improving returns by even 1 or 2 % is significant. 

[cxsmith]

Here is a link to an article that discusses 7 identified lumber recovery factors.  A mill has to focus everyday on these seven, among them is sawing decision making.

...What a computer program is good at is very quickly running a series of simulations and working out what is probably the best solution for that log, based on a variety of factors. How you weight them will vary. But a large commercial mill might be operating on a 10% profit margin? So at that level, improving returns by even 1 or 2 % is significant.

Right, like I did in my back-of-the-envelope calculations on tomography, in a decently-sized mill 1% could be worth hundreds of thousands per year under very conservative assumptions.

But a scenario I'm curious about is: suppose you've got a sawmill and you've sawed all your high quality hardwood logs for the day. You could mill some softwood logs, but it wouldn't be worth the effort. But if the primary breakdown were totally automated starting from putting the logs into a log deck, and finishing with the mill using its head to push the flitches out the back, with you only having to run them through the resaw to make them into boards, would that change the balance? Your margin would still be no greater than the big mills, but you already bought your sawmill so you'd just be getting more utilization out of it. Would the effort saved by automating that part of the process be enough to tip it into being worthwhile?

[Ianab]

Problem is that making an uneconomic job 5 or 10% more efficient probably doesn't make it economic. You can saw low value softwood construction lumber with any sawmill, but is it worth it? Probably is if it's your logs, and you need the lumber because you have cut out multiple middlemen.  But if you are buying logs, and trying to sell the lumber wholesale? Deduct your operating costs and maybe you are making $5 an hour, or less. Better to go and flip burgers, and save the overhead expenses and wear on the equipment. 

[longtime lurker]

I like food analogies, because everyone needs to eat so we can all relate to it.

So I'm running a cafe or diner with a regular clientele and a reputation for good food and I decide that after hours seeing as I already own the grill that I might try competing with McDonalds.

Look I'm a believer in automation wherever possible and not afraid to invest in technology that will make my business more profitable. But part of being profitable is knowing where not to compete with the guy down the road. If I'm going to run another shift, it won't be to compete with a mega mill I cannot compete with.

I believe there is a lot of scope for improvement in the current generation of technology. And I also believe that as the cutting edge - pun intended - of sawing technology advances that there will be a degree of trickle down with that. But if you asked me how to make the currently available equipment better my answer would be to teach an experienced sawyer to code, because it takes 10 or 15 years of full time sawmill operation to get from proficient to experienced compared with 6 months to be able to code reasonably well.

I would suggest that if you're serious about designing even the most basic of optimising software programs for sawmill applications that the best thing you could do is spend some time sawing logs to get a grasp of the how what and why. I'm a bit far away to help with that but this website has a widespread member base and I'm sure someone would be willing to give you some experience, or point you in the direction of someone who might.

[customsawyer]

Most of the big mills around me run 24 hours a day. Only a few of them are switching from hardwood to softwood. Most are either all hardwood or all softwood. The ones that are switching it isn't no big deal because it is about the same as switching from grade logs to pallet logs. They basically have multiple head saws and just switch one at a time and keep on running. Some are also running circle mills and band mills for head saws.
You may have answered this and I didn't see it. Are you trying to set this up for smaller operations or for the big mills?  

[SawyerTed]

Wondering out loud sort of like an investment banker might...

Are the percentages of possible increased recovery over existing recovery rates in optimized mills worth chasing?

Is the law of diminishing returns in play here?

Will it cost more in optimization to get a few percentage points more recovery than that recovered material is worth?  

Every manufacturing step takes time, add up the time of all the steps, can new optimization systems take less time and be more efficient too?  

Are we getting more high grade material or are we just getting more material that requires more processing ($) to be usable?

Aren't optimization companies and partner mills already hyper-focused on greatest recovery already?

Is there enough meat left on the bone to support investment?  

Where's the niche for the new optimization?  The big mills which may already have optimization or manual mills seeking optimization?  Both will be hard to get established for different reasons.  

Lots to think about.  

[longtime lurker]

Mechanisation is one future, exporting logs to places where labour costs are lower is another. But I don't see much long term joy for anyone dependant on hired labour at low production volumes in the developed world.
 As always when the crunch comes the big guys will get bigger, the little guys will tuck their belts in and survive, and the guys in the middle will get squeezed out.

Technology will always trickle down to the bottom. And for the big guys there will always be upgrades to come 

I'm mindful of that all the time. I'm somewhat niche, but not so niche I live in a vacuum and am not affected by market forces. And I'm big enough that I cannot run without hired help. So my mid term goals are to increase efficiency to improve output per man day while at the same time building in the financial resilience to be able to run hard regardless of short term issues. But those are contradictory requirements so it's going to take some juggling.

(An alternate future has me sitting on the beach with a long cold glass in one hand and my hot babe in the other watching a fishing line, but the sawdust affliction is still too strong for that.) 

[Kenb68]

I've been in the Automation business for 33 years. I've own my own integration business for 22 years.

1 - what 3d vision system are you looking at?
I've worked with some major brands keyence, cognex and they are not cheap 2 - 5 k or more. I always put a disclaimer in my quotes that have vision. I won't garrentee it will work. To many factor can screw them up, ambient light, environmental etc. Vision was always my last resort. 50% of our vision system end up not working.

2 - how are you going to scan a log? How long is the log? Let say you want to scan a 15' log. Do you move the camera on a 15' slide?

3 - your going to have to rotate the log for the vision system. You ll need an encode to know the 360 degrees of the log compare to what the vision system is seeing.  You ll have to send measurements from the vision system to plc memory. High spots low spots. (Unless you have a way to pass the log thru a 3d scanner and some software on the pc can do your calculations then send a motion profile to a plc or cnc controller, big bucks)

Then you have a huge problem with rotating the log. Ok we have and Encoder. But what if the log slips on the hydraulic rotating and it doesn't match the the Encoder postion to what the vision is seeing.

4 - rotating the log, the mill has hydraulics, you can do positioning with hydraulics but you ll need a specail controller. That's separate programming from the plc .

I've seen steppers motors mention, they are to small to move anything that needs alot of force. You'd have to go to servos. Servos are huge dollars and then you need axis controller cards for the plc. Those aren't cheap either.

Still have the issue of the log slipping and losing your postion. Bark breaking off, wet log, weird shape.

Just a quick mind dumb as I read this thread. I'm retired,  and own a woodmizer LT15 wide. Been milling for a year , sawing lumber for a house me and the Mrs are building.

You'd have 2 to 3 times the cost of a good hydraulic mill just for controls. I really don't think it would work.

Sorry if I'm a buzz kill.

[moodnacreek]

Mechanisation is one future, exporting logs to places where labour costs are lower is another. But I don't see much long term joy for anyone dependant on hired labour at low production volumes in the developed world.
As always when the crunch comes the big guys will get bigger, the little guys will tuck their belts in and survive, and the guys in the middle will get squeezed out.

Technology will always trickle down to the bottom. And for the big guys there will always be upgrades to come

I'm mindful of that all the time. I'm somewhat niche, but not so niche I live in a vacuum and am not affected by market forces. And I'm big enough that I cannot run without hired help. So my mid term goals are to increase efficiency to improve output per man day while at the same time building in the financial resilience to be able to run hard regardless of short term issues. But those are contradictory requirements so it's going to take some juggling.

(An alternate future has me sitting on the beach with a long cold glass in one hand and my hot babe in the other watching a fishing line, but the sawdust affliction is still too strong for that.)

You are so confused.

[beenthere]

I've been in the Automation business for 33 years. I've own my own integration business for 22 years.

1 - what 3d vision system are you looking at?
I've worked with some major brands keyence, cognex and they are not cheap 2 - 5 k or more. I always put a disclaimer in my quotes that have vision. I won't garrentee it will work. To many factor can screw them up, ambient light, environmental etc. Vision was always my last resort. 50% of our vision system end up not working.

2 - how are you going to scan a log? How long is the log? Let say you want to scan a 15' log. Do you move the camera on a 15' slide?

3 - your going to have to rotate the log for the vision system. You ll need an encode to know the 360 degrees of the log compare to what the vision system is seeing.  You ll have to send measurements from the vision system to plc memory. High spots low spots. (Unless you have a way to pass the log thru a 3d scanner and some software on the pc can do your calculations then send a motion profile to a plc or cnc controller, big bucks)

Then you have a huge problem with rotating the log. Ok we have and Encoder. But what if the log slips on the hydraulic rotating and it doesn't match the the Encoder postion to what the vision is seeing.

4 - rotating the log, the mill has hydraulics, you can do positioning with hydraulics but you ll need a specail controller. That's separate programming from the plc .

I've seen steppers motors mention, they are to small to move anything that needs alot of force. You'd have to go to servos. Servos are huge dollars and then you need axis controller cards for the plc. Those aren't cheap either.

Still have the issue of the log slipping and losing your postion. Bark breaking off, wet log, weird shape.

Just a quick mind dumb as I read this thread. I'm retired,  and own a woodmizer LT15 wide. Been milling for a year , sawing lumber for a house me and the Mrs are building.

You'd have 2 to 3 times the cost of a good hydraulic mill just for controls. I really don't think it would work.

Sorry if I'm a buzz kill.

Very well summarized Kenb68. Obvious that you have "been there".  ;D
Wrestled with these limitations for a good many years. 
One that I recall back in the day, was a system developed to put boat-patches in veneer sheets destined to be face veneer for plywood. Where operators would cookie-cut out defects such as knots to replace with boat-shaped wood plugs to qualify for clear face grade plywood. The system used optics to locate defects at one step on the assembly line, with the second step being the punch to stamp out the defect, and the third station to press the glued-up boat patch in the cut-out.  The system was to be a high-light of a wood manufacturers machinery show to illustrate the advance in automation in the wood industry. Worked to perfection in the building where this system was "invented" and fabricated. But move it to the show in Louisville and under different lighting conditions, defects were located, cut-out punched out, and boat patches pressed in..  however the defects were in one place, the cut-outs in another, and the boat patches pressed into the veneer in another spot. Very embarrassing for the developers and the show featuring this breakthrough. Blamed on the lighting conditions in the show pavilion. 
Just one illustration of what I think Kenb68 is suggesting.

Another might be when brake company out of Michigan years back, purchased a large sawmill complex in California with the thought that they could use their advanced automation techniques producing brake parts to automate the production of wood products. Based on the successful sensing of flaws in brake parts using optics, they launched a several million dollar plan to automate the chop saw process of removing defects from millwork (only the chop saw to begin with, but the rip saws were to be the second stage in the plan). Long story short, the system would locate most of the defects to be chopped out, but some were border-line and would be marked with a light-reflective ink pen which was detected by reflected light. Turned out the only person that could reliably decide what the optical system would miss, was the research tech who developed the "automated" system. He would mark the troublesome defects ahead of the detectors and was referred to as the uhi (upline human intervention). The brake company said they had about 6 million invested because they just knew if it worked for brake cylinders and the like, it surely would work for the easier wood cut-up manufacturing business. 

[longtime lurker]

I think that the fact that someone once failed doing something doesn't make it impossible, just difficult. As technology changes and improves in other fields that technology - some of it, and with modification - will find a home in the wood processing sector. That doesn't mean it will be cheap or trouble free, or even workable. But every failure comes with a series of lessons that can be applied to the next attempt: over and over until you get it right. (and then everyone will say how easy it was and the chinese will infringe your patents but yanno - that's life)

25 years ago the concept of a three axis CNC router being able to cut out a chair complete with inlaid carvings was bound to fail.
50 years ago any band under 6" wide was a toy, and anything under 3" wide was a hobby tool and had no place in the professional wood products sector.
And I'm sure when the first guy tried to hook up an outsize handsaw to a waterwheel that that was never going to work as well. "Just use your broadaxe man and stop wasting time with that foolishness"
It just takes time and a lot of investment for technology to make it from the space shuttle to the front of your car in a workable format.
And there's also some of that story about how the space shuttle design was determined by the width of a horses rump to factor into it as well.

And my feel is that the latter - the width of the horse determining roman roads thus railroad gauges, and railroad guages affecting the size of a part that can fit through a tunnel - is where a lot of the optimisation technology is not transitioning well. Yet. Wood is dynamic and subject to piece by piece factors, and the program doesn't yet know how to register those factors on a consistent basis. But thats just a matter of pattern recognition and picking the appropriate response, absolutely no different to taking in visual information with an eyeball and having experience tell you its going to pull to the left so we need to oversaw that way to compensate.

[Southside]

Along those same lines imagine how different the world would be today if the first guy who looked at a cow and said "I think I am going to drink what comes out of those dangely things" had been looking at a bull.... or the sheer luck of the guy who spotted a chicken and said "the next thing that comes out of that is going to be breakfast", 50/50 odds there.... :D

[barbender]

Well all of those early experimenters had failures, Southside. We just hear the success story. The guy with the first dairy didn't repeat the story of when he milked the bull first, or had the chicken poop omelet. That's why everyone thinks success comes easy😂

[Ianab]

Along those same lines imagine how different the world would be today if the first guy who looked at a cow and said "I think I am going to drink what comes out of those dangely things" had been looking at a bull.... or the sheer luck of the guy who spotted a chicken and said "the next thing that comes out of that is going to be breakfast", 50/50 odds there.... :D

To be fair, early humans weren't dumb.  Lot's of critters with less smarts recognise eggs as food, not poop. And "Hey that cows calf is getting a feed there, maybe I can too" isn't a huge leap of logic. Especially when human biology works on the same principle.  
Thought experiment about the log scanning. I get what Ken is saying about the practicality. It's currently used in large scale mills where they are basically production lining the sawing. Logs come in a stream and got though a scanning station. Then the computer decides how to set the various saws and edgers on the chain. It takes maybe 60 seconds for multiple saw heads to break down the log, and there are several in the chain at any one time. Multi-million dollar setup. 
But from a smaller mill, what say you loaded the log, then ran a scanner(laser?) down the log, with maybe 3 scanners, on a separate carriage, that's normally parked at the far end of the mill. 3 scan heads should be able to cover 90% of the log as it sits on the bunks, Now you have a good model of the log, size, taper, sweep,and most of the serious defects etc. From there compute your best sawing pattern and carry on. 
Like a lot of things the prototype might cost $1 mil to develop and work out the bugs. But once you do, the actual hardware and computing power is off the shelf, and not out of the question as an add-on for a small sawmill. 

[Kenb68]

I think this is a great discussion. Everything is doable but is it economical to add to an very expensive control system to a simple band saw? Lip stick on a pig?

So a high end milling system is 6 - 10 mill. Maybe a new mill could be designed that has the Automation that a sawyer could afford that's In the 200- 500k range?

Yes and in time as technology gets better and cheaper this will be doable at a cost effective price.

Does anyone with a lot more sawing experience have ideas to get around the issues i have listed? Let's keep this discussion going.


This is a hydraulic servo postion controller I've used in the past.

Products

[customsawyer]

Ken I think that your post is very informative. The one thing that I would say is that any scanning needs to be done before it gets to the mill. The purpose of this is to speed up the operation. Once the log lands on the mill, the mill needs to be sawing. Not turning the log so it can be scanned.