[How do I?] 3D Modeling the scraping process

[Rex Walters]

In my never ending quest to understand more about hand scraping, I decided to create a somewhat crude but instructive model of the process. Not sure if I've learned anything new from the model, but I thought others might find the pretty pictures interesting. (laugh)

The scraper blade has a radius, typically 60-90mm. It also has a negative rake angle of about 5-10 degrees:



This isn't a precise model, it's just for illustration and reasoning about the process (in particular, I'm trying to think about how to increase the bearing percentage vs. the number of points per inch). The radius, rake-angle, etc. aren't modeled exactly. In practice, you'd normally grind the rake bevel on both sides of the blade for convenience (just flip the blade when it gets dull).

The blade and scraper assembly has some flex/give. Fine scraping is easiest with the longer biax-style blades because they have more flex. The shorter blades are best for flaking where you want to create deep grooves.

Anyway, when the blade meets the surface of the part, the blade is pushed forward and it creates a furrow or groove in the surface of the part:



Once you've ground down the very highest areas and got the part relatively flat (with about every square inch showing at least 5 points of bearing or so) you lay down a grid of scrape marks on the surface:



Unlike what this simple model shows, you'd normally have the scrapes longer than wide. As shown, you want to strive for equal spacing both side-to-side, as well as between rows (fore and aft). Unlike what's shown, you want the space between the scrapes (in both directions) to be about the same as the scrape marks themselves.

At the start of the cut, as the blade is pushed forward it cuts into the surface of the part and starts removing metal. If you maintain a constant downward force on the scraper, the pressure between the flexed blade and the hardness of the part soon equalizes so the depth of the furrow stays fairly constant.

At the end of the stroke (the top of the image) material piles up and forms a burr that you'll need to stone off. I've not modeled the burr, but here is what they look like in the real world on cast iron:



If you run your fingers lightly over the surface, you'll feel these burrs quite easily, and your fingernails will catch on them. It's very, very important that you stone off all of them every single time you complete a pass. Not only will they throw off your results and create confusing markups, they will also scratch up a granite surface plate surprisingly easily.

The other thing to note about the arrays of scrape marks is that you want to stagger them. This isn't something you need to pay too much conscious attention to as human variability inevitably causes each row to start in different locations, but it's worth knowing what you don't want:


Don't do this! If you ever notice "rivers" forming like this, you're doing it wrong. Try to stagger each row (in practice it's harder to create the bad example, but I goofed when I cloned the scrape marks in my model and thought I'd point it out).

The scrape marks are normally angled at roughly 45 degrees from one side of the part. Then next pass of scraping rows then proceeds angled from the other side (90 degrees from the first pass):



[To be continued]

 

[Rex Walters]

After completing one pass across the surface in each direction, you'll have made a fairly complex topology of scrape marks:



At this point, though, some of the original surface is still around. If you were to blue up this plate it would look like this (with the shaded portion indicating the blue):



After a couple more "blind" passes (just scraping even rows and ignoring the blue) it would still show some of the original surface:


After two or three crossed passes like this, you should target the individual blue areas. Something like this to scrape off all the blue marks:



After pinpointing just these last remaining spots of the original surface, you are left with high spots like this:




At this point, all of the marked high spots were "low spots" scraped in a previous pass.

In order to increase the number of blue spots (shaded spots) per inch, you would decrease your stroke length, narrow the stroke a bit (with a tighter radius blade or by raising the back of the scraper slightly), and simply repeat the process. Each time you just scrape away the areas of blue, breaking the larger areas into smaller and smaller points.

I've not (yet?) modeled continuing to achieve higher PPI. Again, not sure if I learned anything from this so far, but it was interesting to me at least!

 

[TakeDeadAim]

If your getting burrs on cast iron your scraper has the wrong angle. The cut should not be blunt into the material at its end. As you get better you learn to back off on the pressure toward the end of the stroke which allows the cutter to move back out of the cut and nearly eliminates burrs from the work. I think its difficult to computerize a process that is very much a by hand operation.

 

[AndySomogyi]

Very neat, @Rex Walters! How did you create your models software wise?

I'm actually a physicist at my day job, and I develop mathematical models and simulation software for mechanics and materials. I'm learning scraping now (teaching myself in order to fix my broken mill), and I'm interested in the physics of the cutting edge and material removal. I'm going to look into adapting my code to do these kinds of simulations where you have finite-element models (the scraper blade) in contact with ductile fracture work surface.

I made my own scraper, and it didn't work too well, so I'm trying to understand how different stiffness of the scraper and cutting edge rake angle effect the cut. What I think would be really cool is write a model that simulates stiff vs flexible scrapers, and then actually test these out experimentally.

 

[Janderso]

Well done Rex. Holy cow!
I'm still working on a carbide lapping machine. Can't scrape without a blade.
Could I borrow your brain for a few days?
When are you not going to need it?
A physicist? We have a physicist in our forum?
Makes us mortals feel kind of average.

 

[Bob Korves]

Very neat, @Rex Walters! How did you create your models software wise?

I'm actually a physicist at my day job, and I develop mathematical models and simulation software for mechanics and materials. I'm learning scraping now (teaching myself in order to fix my broken mill), and I'm interested in the physics of the cutting edge and material removal. I'm going to look into adapting my code to do these kinds of simulations where you have finite-element models (the scraper blade) in contact with ductile fracture work surface.

I made my own scraper, and it didn't work too well, so I'm trying to understand how different stiffness of the scraper and cutting edge rake angle effect the cut. What I think would be really cool is write a model that simulates stiff vs flexible scrapers, and then actually test these out experimentally.

Andy, when I first started being interested in scraping, I, like you, tried to figure it all out for myself. I am a good reader, with trained learning skills, and understanding and retention of many subjects I have become interested in. I usually do quite well with reading some well written "how to" books by others experienced in the field I am interested in, and then jumping in with both feet.

I tried the same method with scraping, and essentially got nowhere. Looking on Hobby Machinist, I found a nearly local scraper enthusiast who had attended Richard King's classes and also studied on his own, Mike Walton, "Ulma Doctor" on this site. Mike kindly invited me to his shop and showed me what scraping is about, and let me use some of his tools while teaching me the basics in a half day hands on "class." Well, I learned some of the basics that day, all good stuff from a really good teacher, but it did not get me over the hump of being able to move forward by myself. Several years later I bit the bullet and attended one of Richard King's week long classes in my area. Those of us in the class were fully immersed in learning from Richard as well as from others of the 10 or so students attending, along with a few of Richard's skilled former students who dropped by. I still was moving ahead very slowly until about the middle of the week, when everything finally started to click into place. Being immersed in it (and having multiple skilled scraper hands to watch and to get help from) made all the difference in the world. I am still not the world's best scraper hand, but I understand the basics of it in a visceral way and can make good progress on my own projects. I think the real issue is that scraping metal to achieve high precision results is simply not something that the rest of our lives prepares us for, and reading about making surfaces flat, really flat over a full area, with quite simple tools and techniques, is simply a step too high to pull ourselves up. Watching various scraping videos on sites like YouTube will get you a bit of good knowledge along with a great amount of very poor, useless stuff, but no muscle memory backed by mastery, and the difference is quite difficult for a novice to be able to sort them out. A professional class, with a highly experienced grinder hand, who can explain things well, or a local mentor who can help you regularly to keep you on track, is an important requirement -- in my opinion.

Another bonus of the class is learning about the basics of rebuilding metal working machines, an extension and related partner to learning scraping for completing real life projects. It is not all about learning to scrape something flat, it is about how we tie those skills to successfully completing real life projects.