Trying to estimate the force required to roll form a strip of aluminum

[Ryoko]

TL;DR Summary

I want to roll form a 0.040 x 1" aluminum strip into a U-channel using a series of roll stations. How much force/torque is required at each station (guestimate)?

I need to form a one inch wide strip of 0.040" 5052 aluminum (annealed) into u-channel. The pieces I'm wanting to make are too long to use a press brake. (They are 84" in length.) So I'm kicking around the idea of making a roll former which seems to be the preferred method in this situation. I have a design concept modeled using 7 roller stations with each station bending the strip in roughly 13 degree increments. The bad part is that the rollers are only 2.5" in diameter. This is a limitation imposed by my bench lathe. (This is a home/hobby project.)

I made some test rollers and I was able to crudely simulate doing a 15 degree roll by hand. So the small diameter rollers appear to be workable. What I don't know is how much force I need to apply to the rollers and how much force is going to be required to shove the strip thru all of the rollers. I need at least a ball park figure so I can select the bearing size and decide how much torque is going to be required to turn the rollers. I've looked at videos of commercial rollers and they all seem to use some really beefy chain drives and gear boxes. This has me a confused because I was able to do a single roller pass by just pulling a test strip thru with a pair of pliers.

My assumption is that the pressure between the rollers only needs to be high enough to insure enough friction to allow the strip to continue feeding and to close the gap between the strip and the rollers to insure the forming takes place. Some (very) crude napkin calculations suggest the linear force required to shove the strip from one station to the next due to forming is around 30lbs. But I made a lot of assumptions to get that number. That said, I'm guessing in the neighborhood of 50 in-lbs of torque to drive each station. Is this realistic?

 

[Baluncore]

How many hundred straps, (25 mm x 1 mm), do you need to roll ?

Measure the force needed to bend an inch of the strap, by holding a specimen in a vise, then bending to the required radius with an adjustable shifter and a tension gauge.

Al and steel stick together, that is good for friction drive, but the Al will bruise and scuff if it is not well lubricated, or coated in a protective paper layer.

Bend the Al just beyond a 'U', so it flexes back to the angle and form required. Do not pinch the Al strap tight between steel rollers, the surface will stretch, and then the 'U' will not be straight. The line through the first few rolls should not be straight and level, since then edges of the strap would then be stretched.

The area of the roller shaft bearings, need only be more than the sectional area of material being bent, unless you intend to coin the material between two rollers.

Annealed Al will work harden as it passes through the rollers. Each roller stage will take more force, unless the bends start at the middle of the 'U' and migrate out towards the edges, where the material is not yet work hardened. The inside radius of the 'U' might form itself.

 

[jrmichler]

TL;DR Summary: I want to roll form a 0.040 x 1" aluminum strip into a U-channel using a series of roll stations. How much force/torque is required at each station (guestimate)?

I was able to do a single roller pass by just pulling a test strip thru with a pair of pliers.

I spent three years in grad school studying sheet metal forming, so I can say authoritatively that pulling a test strip with a pair of pliers will get you the torque information as well as, or better than, spending a lot of time calculating. Use a C-clamp and a spring scale if you want actual numbers. The challenge to calculations is rubbing friction between the rollers and strip.

Ball bearings have static and dynamic load ratings. If you are making only a few or few hundred strips, you can load bearings up to the static load rating. That is a high enough load to start bending shafts, so if you make the roller shafts large enough, the bearings should be good enough.

The above is for a hobby setup. If this is for a factory production machine, the above approach will give you a pilot machine to work out the best dimensions of the rollers. Then put in the largest ball bearings that you can fit in. Machine design is a tradeoff between "get it done" and "paralysis by analysis". For a simple machine such as this, using bearings a size or two larger than necessary normally results in the lowest cost and shortest time to a reliable machine in a production environment.

 

[Ryoko]

Thanks for your reply. Fortunately, I don't need to roll hundreds of strips. More like a dozen or two. The volume is much too low to have them done with an outside firm. And where's the fun in that?

I did do a sample test with a jury-rigged press and a sliding weight on a level arm. It came out to be about 175lbs/in. But the corner radius was larger than desired. I attribute most of that to the crude nature of the test setup. I took a look at a couple of the online calculators for bending using a press brake and they are telling me that it would require 200lbs/in. That number would need to be doubled since the roller is doing two sides at once.

This has been a learning experience so far. I did make the mistake of using too much force between the rollers on the first attempts thinking it would make for a nice corner radius. It made the strips into pretzels. They were terribly crooked. Keeping the piece going straight is also still a challenge. I've found minor errors in alignment quickly get magnified with each roller.

It hadn't occurred to me that edges would need to stretch on the first few passes. That explains the curve I was seeing on my early tests. I'm guessing that's why I'm seeing the stations spaced out on some machines I've looked at online. I've arbitrarily decided on 5" spacing between rollers. This was done to relieve some of the pressure between stations to give room for the strip to assume the shape of the next station more than anything else.

It's interesting you mentioned the idea of forming the middle first and then working out. One of the machines I looked at online had a concave bow in the middle of the first 3 rollers. I assumed it was done to focus the pressure into the corners to get a better corner radius. But it makes sense they were trying to avoid the work hardening in the bend.

As mentioned earlier, it looks like it will require 200lbs/in to bend the material. The question remains how much torque that translates to for each roller. I determined that the strip makes contact with the roller about 0.4" off center from the roller axis. This is where the assumptions start. To make things simple, I assumed the bending force is directed purely vertical. I also assumed the force on the roller increases linearly from zero (where the strip first touches the roller) to its maximum directly over the center of the roller axis. I then solved for the resulting torque. This is where I came to a ball park figure of 50 in-lbs needed to turn each roller.

I don't want to mess this up by underestimating the required torques. The difference is having to use an expensive chain drive and reduction gearing versus being able to use inexpensive syncro belts and pulleys and a fractional horsepower motor to drive it.

 

[Baluncore]

Hand operated "ring roller" machines are low cost on eBay etc. The rollers could probably be modified or replaced, to roll the section you require.

Small rectangular channel stock is available from the aluminium industry as extrusions.

 

[Ryoko]

I spent three years in grad school studying sheet metal forming, so I can say authoritatively that pulling a test strip with a pair of pliers will get you the torque information as well as, or better than, spending a lot of time calculating. Use a C-clamp and a spring scale if you want actual numbers. The challenge to calculations is rubbing friction between the rollers and strip.

Ball bearings have static and dynamic load ratings. If you are making only a few or few hundred strips, you can load bearings up to the static load rating. That is a high enough load to start bending shafts, so if you make the roller shafts large enough, the bearings should be good enough.

The above is for a hobby setup. If this is for a factory production machine, the above approach will give you a pilot machine to work out the best dimensions of the rollers. Then put in the largest ball bearings that you can fit in. Machine design is a tradeoff between "get it done" and "paralysis by analysis". For a simple machine such as this, using bearings a size or two larger than necessary normally results in the lowest cost and shortest time to a reliable machine in a production environment.

I admit that I'm likely falling victim to paralysis thru analysis. I've made a couple attempts at doing this using different methods with mixed success. It's starting to get expensive. So I really want things to work this time around. The volume is pretty low, so abusing the bearings is not really a deal killer.

The largest bearing there's room for looks to be a 1/2"ID (FR8ZZ ) with a static load rating of 540lbs. I guess that's enough since there's two on each roller. I'd consider oilite-style bushings which would be easier to mount and leave more room. But ball bearings are less expensive and I'm being paranoid about the rolling resistance. The proposed shaft is 1/2" W1 drill rod. I need something with good dimensions and be soft enough to thread on the ends. I'm not going to heat treat them.

 

[Ryoko]

Hand operated "ring roller" machines are low cost on eBay etc. The rollers could probably be modified or replaced, to roll the section you require.

Small rectangular channel stock is available from the aluminium industry as extrusions.

Unfortunately, off-the-shelf channels don't have the dimensions or profile I need. That was the first place I looked.

 

[jrmichler]

This has been a learning experience so far. I did make the mistake of using too much force between the rollers on the first attempts thinking it would make for a nice corner radius. It made the strips into pretzels. They were terribly crooked. Keeping the piece going straight is also still a challenge. I've found minor errors in alignment quickly get magnified with each roller.

I have 20 years experience at machine design. This sort of thing is completely normal, and is known as "learning curve". A mechanical engineer with experience at roll forming might get it right the first time. An engineer with experience at machine design, but not roll forming, would expect at least one or two iterations to get it right.

I don't want to mess this up by underestimating the required torques. The difference is having to use an expensive chain drive and reduction gearing versus being able to use inexpensive syncro belts and pulleys and a fractional horsepower motor to drive it.

Get the roller part to work and run it with a hand crank. Then, if you really want to motorize it, measure the torque with spring scale to size the drive.

Semi relevant anecdote: I was working in a paper mill, and they wanted a machine to break apart run-together rolls of paper. So I sketched up a concept. Management saved money by asking for a hand hydraulic pump instead of a motorized pump. I warned them that it could take several hundred pump strokes for one cycle. Less than one hour after the machine was unloaded from the truck, a manager was in my office with orders to get the motor pump on order. That was one iteration, the second iteration was a change in wedge dimensions. That machine was then successful, and my first patent.

 

[Ryoko]

Technically, I'm an engineer (retired), but in electronics -- not mechanics. I know iterations are a fact of life. I like to plan ahead. All of the previous iterations have been hand-powered. Initially, this one will be too. But I want to have provisions in place for adding power later on without having to rebuild everything.

This is the 3rd concept to do this task. The first was a long press brake using a series of bolts to draw a blade down into a channel. It mostly worked. But it was a pain to use since it required tightening all the bolts evenly and the strip had a nasty tendency to shift producing an uneven part. I looked at using hydraulics. But the machine would need to be substantially beefed up and I really didn't want an 8ft, 250lb machine taking up space in my garage.

The 2nd concept used a single roller running along a track to force the the strip into a channel a little at a time. The roller would be extended a little deeper with each pass. This also mostly worked and was a lot faster and easier than dealing with a bunch of bolts. But trying to keep a 7 feet length of 1" wide aluminum centered was a challenge once again. This finally convinced me to not reinvent the wheel and go with the usual solution of using a roll former. I had originally avoided it because of the high parts count and all the machining that will be necessary.

 

[Baluncore]

But trying to keep a 7 feet length of 1" wide aluminum centered was a challenge once again.

Do you have any rollers or bearings pressing on the edge of the strap to maintain alignment with the rolls? Most deep groove bearings, have outer races that can be used as alignment rollers, without modification.

I find it better to push the sides in, rather than to press a strap into a fixed channel. A roller, that pushes the sides in, can also have a step or guide, that maintains strap alignment.

Rather than cascading rollers in a line, consider three separate machines, one that begins to close the section, another that pushes the sides together, followed by a final straightening pass.

Can you provide a sectional diagram, of the ideal profile, you wish to produce ?

 

[Ryoko]

The first machine acting as a long press brake had edge guides to start things straight. The problem was that as the strip disappeared into the channel, the edge guides no longer reached and the strip was free to shift. The second machine which incrementally rolled the strip into the forming channel had a similar problem.

There's nothing special about the profile. It's just a plain-old u-channel that's 3/8" wide and 3/16" tall. The inside radius is 0.040" -- same as the material thickness. That said, there can be a little slop in the profile. This project is to replace some trim on a project car and the factory pieces were not exactly perfect.

The 3rd proposed machine will either have slotted guides or rollers feeding it. I'm leaning towards guide rollers because they can be easily adjusted. I want the last stage to pinch the piece into the final dimension. I currently have a separate simple machine that does this. You just pull it thru by hand. I need to figure out a way to allow the last stage to move vertically and to rotate to deal with any twist. I don't have a mill, so the design has to work within tooling limitations. (Thank god for online laser cutting services.)

 

[Ryoko]

Finally got around to doing a rough pull test of a sample strip thru a roller. It was hard to get a consistent value because pulling evenly by hand proved to be a bit of a challenge. But It looks like the pull force was between 20 and 30lbs. So my guestimate was high, but not too far off. This works out to about 15 lb-ft of torque to feed the strip thru all of the rollers.

I decided to make my prototype consist of the first 5 rollers sets which translates to a bending angle of about 60 degrees. I'm now planning to do a second machine to do the last 30 degrees of bending by rolling against the sides of the strip horizontally while pressing against a center mandrel. The parts for the first machine are expected to arrive early next week.

 

[jrmichler]

This works out to about 15 lb-ft of torque to feed the strip thru all of the rollers.

The difference is having to use an expensive chain drive and reduction gearing versus being able to use inexpensive syncro belts and pulleys and a fractional horsepower motor to drive it.

My Milwaukee 1/2" variable speed electric drill has maximum speed of 850 RPM and maximum torque of 30 ft-lbs. Since this is a hobby machine, not a high production industrial machine, a variable speed drill might meet your drive needs.

 

[Ryoko]

My Milwaukee 1/2" variable speed electric drill has maximum speed of 850 RPM and maximum torque of 30 ft-lbs. Since this is a hobby machine, not a high production industrial machine, a variable speed drill might meet your drive needs.

I was thinking of using a drill to power it. They have the torque and I've got a couple lying here around I could use. I would only need a simple adapter to make it work. But it's hand cranking for now until I'm sure it will produce consistent results.