Edge Retention Testing and Analysis

[Broos]

Pretty neat - you can see the lines showing elastic deformation of the edge roughly 20 microns from the edge!

 

[tnelson]

I don't have a direct response to your thinking out loud, Broos. I too will engage in some of that:

From Test for measuring cut resistance of yarns, Hyung-Seop Shin, D. C. Erlick, D. A. Shockey:


The blade (Stanley utility blade, 2 um mean radius, 30 deg included) made one perpendicular cut through a yarn (made of 330 12-μm-diameter polybenzobisoxazole (PBO) fibers) of Zylon (a cut resistant material they were testing) material. The cut energy was 0.105 joule. Before the cut is completed, the edge deforms and expands to a radius greater than 5 um (evidenced by the fact that a harder ceramic blade of 5 um radius was outperforming it).

I think you can imagine what work even on this scale (1/10 of a J) directed at the edge not in line with the axis of the blade can do.

kel_aa,

Nice pictures!!!

Broos,

Good thoughts above. I think Wayne and Phil are among the best at testing blades on rope by hand, but we recognized we were not. We decided to eliminate the "human" factor and went with the CATRA test.

My personal feeling is that machine testing will be representative of good human testing. There will probably be a scalar factor, but I think the trends will be the same when comparing different steels (all other factors being the same).

I think it would be interesting to do a round robin assessment of hand testing versus machine testing. My hypothesis is that the larger the number of people testing by hand, the larger the deviation will become. This isn't saying it is bad testing or data. I think I would imply that without a "standard" results will vary and discriminating between results is very difficult.

Those are my thoughts.

TN

 

[hardheart]

My personal feeling is that machine testing will be representative of good human testing. There will probably be a scalar factor, but I think the trends will be the same when comparing different steels (all other factors being the same).

I think it would be interesting to do a round robin assessment of hand testing versus machine testing. My hypothesis is that the larger the number of people testing by hand, the larger the deviation will become. This isn't saying it is bad testing or data. I think I would imply that without a "standard" results will vary and discriminating between results is very difficult.

I agree, but think that it really hurts the value of the testing when discussed openly. Number of cuts are repeated well enough, but size of rope, type of cutting, what criteria were used to determine a stopping point, etc. would tend to get lost. There's a thread in the main board about a 10V blade doing 999 rope cuts, second place was Infi at around 300. That's quite different when compared to the number of cuts in this test, Wayne Goddard's, Phil Wilson's, or Busse's 2K+ demonstration. Still, a lot of people will just see lots of rope being cut and take the highest number as being best, or least trustworthy.

Each test has it's value, they just don't compare well.

 

[Broos]

Maybe a requirement for good human testing is having the same person do all the tests! I question whether I could avoid a lot of lateral forces after cutting as much as those guys do in a test - my hand might feel like super putty...

If we specified the rope, backing, knife, sharpening procedure, and stopping point (but different scales used), do you guys think we could get decently similar results from tester to tester? Do we need to specify the slice or draw amount or would the sharpening procedure make that unnecessary?

I've kind of thought we could get decent results from tester to tester, but you guys got me thinking about it and doubting it now. I'm having trouble with the scale *(is everyone going to use the same mfg/model scale?) and sharpening differences from tester to tester.

Even assuming the numbers are all over the place, do you think with a good test procedure using different human testers, we could at least get the same order of results (from knife to knife or steel to steel)? I hope we could.

 

[Wayne G.]

We test one knife against another and these knives are in the thin ground hunting knife class. We keep all factors the same as is humanly possible. The test knife results show as a percentage of total edge holding ability compared to a knife of known value. That's why I say hundreds or thousands of cuts mean nothing if there is no comparison made. We do same day, same rope, same sharpening comparisons whenever possible.

As an example I could say I had a Rockwell test machine and that tested a blade at 69Rc. That could be a true statement of my test results but it means nothing because the machine is out of calibration. The 69Rc on the dial is a number that doesn’t mean anything because a legitimate comparison wasn’t made. Once the test machine is adjusted with an accurate standard, the 69Rc would show up as a false reading.

Wayne G

 

[tnelson]

Having watched Wayne sharpen a knife, he has developed a technique and standard of his own that I believe is repeatable. Wayne has a smooth motion that is really consistent. Now, if we could just download the program from Wayne's arm, then distribute to all that want to test knives, we might get some repeatability between different people.

Broos,

I think you have hit many of the important criteria that would be required to "standardize" a test procedure. The one criteria that I think is impossible to control is the rate, and length, of the stroke. In addition, the load applied during the stroke. These will govern the load on the edge of the blade, and the amount of material that wears across the edge. I believe (no actual data) that these will have a significant effect on the results.

TN

 

[Wayne G.]

I don't have a direct response to your thinking out loud, Broos. I too will engage in some of that:

From Test for measuring cut resistance of yarns, Hyung-Seop Shin, D. C. Erlick, D. A. Shockey:



The blade (Stanley utility blade, 2 um mean radius, 30 deg included) made one perpendicular cut through a yarn (made of 330 12-μm-diameter polybenzobisoxazole (PBO) fibers) of Zylon (a cut resistant material they were testing) material. The cut energy was 0.105 joule. Before the cut is completed, the edge deforms and expands to a radius greater than 5 um (evidenced by the fact that a harder ceramic blade of 5 um radius was outperforming it).

I think you can imagine what work even on this scale (1/10 of a J) directed at the edge not in line with the axis of the blade can do.

It seems to me that the blade was (perhaps) not true sharp. In other words, it had a false sharp, wire edge. Most blades of that type come with a perfectly lined up wire. Cuts soft stuff good, fails on tough things. Probably 80% of knives being sold, handmade and factory do not have a true sharp edge when they are sold.

I've got a theory that some of the issues with the CATRA and REST tests have to do with that pesky wire edge not being properly removed.

Wayne G

 

[Broos]

TN, You are probably right - & those are variables we do not have to worry about with the CATRA machine. It would still be interesting to see the results with numerous testers all doing the same test with the same blades, and then try to see what we could figure out from the results. Regardless of how the results come out we could learn something, even if it was that the test procedure we attempt will not provide consistent results from tester to tester. Even that result would be enlightening.

Thinking a little more on the lateral force issue, further assuming that human induced (unsteady & transient) lateral forces are critical to edge retention results, I wonder if this may be able to be replicated with a machine - couldn't just a little play in linkages, in the proper planes, create similar lateral forces (some shakes and shimmy's) with a machine slice? Kind of against logic to introduce this in a test, but I wonder if it may be pretty easy to induce similar forces...

Wayne,
I'd like to sit through your sharpening demonstration - hopefully I can someday. Reading back through your posts I read that you make a stroke or two at an increased angle to remove the wire edge. Could you explain how you tell that the wire edge is removed? Also thanks for your sense of humor - a couple of your early posts had me chuckling this morning.

 

[Wayne G.]

Wayne,
I'd like to sit through your sharpening demonstration - hopefully I can someday. Reading back through your posts I read that you make a stroke or two at an increased angle to remove the wire edge. Could you explain how you tell that the wire edge is removed? Also thanks for your sense of humor - a couple of your early posts had me chuckling this morning.

THE WIRE EDGE
It is important to understand the wire edge and how to get it off of the freshly sharpened blade. The wire edge is formed as the two sharpening bevels meet at the edge. What I call a wire is a thin piece of blade material that bends back and forth from the action on the stone. The action on the stone to bring the edge up to sharp does not remove the wire, it creates it. When the wire edge is not lined up the knife will not appear to be as sharp as it could be, yet sufficient material may have already been removed. Since the edge does not feel sharp it is often worked some more on the stone and many knives are worn out prematurely from this overworking. When the wire edge is lined up the blade will appear to be sharp. I refer to this phenomenon as "false sharp". The blade may shave hair and slice paper, but when the edge contacts any type of hard substance the wire edge bends over and the knife will quit cutting. When the wire edge is pulled out in use it leaves micro flats where the steel pulls out. The wire edge must be carefully removed with very light action on the stone.

REMOVING THE WIRE EDGE
You will feel the hook of the wire with a fingernail and it will be on the opposite side from the last stroke on the stone. My way to remove the wire edge is to stroke the edge very lightly on the finish stone at an angle of around 30º. The strokes are alternated from one side to the other and are very light. As the blade is stroked with the light cuts the wire is abraded off, leaving the true sharp edge. It might seem that the steep angle used to get the wire edge off would slow down the cutting ability of the knife. When properly done, not enough material is removed to blunt the edge. With the wire edge removed you will have a true sharp edge and one that will hold an edge up to the full potential of the glade.
It is very easy to lose the sharp cutting angle at the edge when trying to get the wire edge off by buffing or stropping. If you prefer a buffed or stropped edge, take the wire off on the stone, then buff or strop. The angle on the strop or buff is critical, if too steep, the edge gets rounded off and the edge loses its true sharpness.
This applies to hunting and working knives.
Be aware that most cooks and meat cutters work with the wire edge on their knives. The constant steeling keeps the wire lined up and the burnishing effect prolongs use without returning to a stone.
My finish stone is the Norton Fine India (aluminum oxide).

 

[Kohai999]

My finish stone is the Norton Fine India (aluminum oxide).

What size stone, Wayne, and how do you feel about the combination stones(medium/fine)?

With water or oil as the lubricant?

If oil, what is your take on the Norton Tri stone systems?

Best Regards,

STeven Garsson

 

[bladeymae]

I attended Wayne's sharpening seminar at the Blade show in Atlanta. Wayne used a Norton combination stone that was crystalon on one side and fine India on the other. If i recall correctly it was 320 grit on the India side. He mentioned "TruGrit" as a source for the stones and i think they are available in 8" and 11" sizes. As a lubricant he recommended using "Simple Green" and he sprayed the stone with it.Wayne alternated strokes from one side of the blade to the other until he achieved a burr on the crystalon side of the stone and then repeated the process on the India side of the stone. Using a stick with a piece of leather attached he stropped the blade about 5 times on each side. The leather was loaded with green chromium oxide buffing compound. Wayne then sliced pieces from a newspaper and made a loop in one piece and sliced the ink off of the paper! I enjoyed the seminar and benefited from the instruction Wayne presented.

Johnny

 

[Wayne G.]

Johnny,

You forgot to apply the bald head shiny spot filter before taking the picture. (grin)

That's hard working editor Shackleford winding his camera in the background.

The class/demo is always fun for me and I'm glad to spread the word on getting kinves truly sharp.

Wayne

 

[2knife]

Entertaining debate. A good hoot! Meeting Wayne was the highlight of the Blade Show for me.

STeven,
NIce show of balls, funny commentary. Cliff, being a burr. You can't help but stick your neck out to say it.

I didn't know, or understand half or even a quarter of this thread. But, I could follow the drama. And, here we are with a bald spot..
David

 

[TBDAugs]

Miss seeing you guys there in Happy Valley! Keep up the good work. Loved the forum as it recalled the 5 years of arguing the merits on whether this technology was worth the investment to pursue. Charles sent me a FF'd prototype 'personal defense' knife I could have used in past lives. Truely a thing of beauty and much, much bigger than either Carls' or Tracy's.

Bravo Zulu

Cdr. "Augs" Flak, USN-retired


Searching PGPUB Production Database...


Results of Search in PGPUB Production Database for:
IN/"Flak, Richard": 8 applications.
Hits 1 through 8 out of 8





PUB. APP. NO. Title
1 20060289608 Friction stirring of high softening temperature materials using new surface features on a tool
2 20060255094 METHOD FOR CONSTRUCTION OF PRESSURE VESSELS WITH A LINER USING FRICTION STIRRING PROCESSES
3 20060157531 Single body friction stir welding tool for high melting temperature materials
4 20060049234 Friction stirring and its application to drill bits, oil field and mining tools, and components in other industrial applications
5 20060032891 Solid state processing of materials through friction stir processing and friction stir mixing
6 20060032333 Solid state processing of industrial blades, edges and cutting elements
7 20050252341 Solid state processing of hand-held knife blades to improve blade performance
8 20050156010 Applications of friction stir welding using a superabrasive tool

 

[tnelson]

Miss seeing you guys there in Happy Valley! Keep up the good work. Loved the forum as it recalled the 5 years of arguing the merits on whether this technology was worth the investment to pursue. Charles sent me a FF'd prototype 'personal defense' knife I could have used in my operational days. Truely a thing of beauty and much, much bigger than either Carls' or Tracy's.

Bravo Zulu

Cdr. "Augs" Flak, USN-retired

Augs,

You bugger! How'd you get a prototype of the "personal Defense"? I'll have to get Charles to send me a FF Bush Master because.....size does matter.

Stop and see us next time your in Happy Valley.

TN

 

[Cliff Stamp]

C is uniquely identified, but a and b can be traded off in a correlated way to get an acceptable fit. Since neither a nor b is a physical parameter of interest, but just a fit coefficient, it doesn't give me heartburn that there is a range of a and b values.

Unless the correlation coefficient is absolute zero, which is very rare, the other parameters can always compensate to some degree, that of course is the definition of that coefficient. As well, there is no such thing as unique parameters at all in nonlinear analysis, as opposed to linear fitting which is convergent to a unique parameter set, nonliner fitting is always sensitive to the initial seeds.

I suspect, but don't know, that treating a and b as independent may be where Cliff had some trouble with the fit parameters.

The method I used nor the computer I ran it on was not bothered about the results, neither had an emotional attachment, nor paused for an instant during the calcuations. They simple produced a result which showed theparameters were undefined as the data variance was too large due to the calculations performed on the raw data which magnified the data. As noted clearly many times, the actual raw data was perfectly fit by the model, as is every set of edge retention data I have seen to date.

I have finally started the edge retention article, as of right now it just defines the basic process of which there are two general types, cutting a specific amount of material, or cutting until there is a specific amount of blunting. As the responce is highly nonlinear these will give very different results when you compare knives. Right now the evaluation group is comparing a P/M D2 blade vs wrought D2 and I will use their results to expand the article. If you have any questions about the methods or there are issues you wish to see discussed in detail then you are welcome to participate in the discussion. The first results have already been posted and have been tabled for review already.

http://www.cutleryscience.com/forum/read.php?7,23,23#msg-23

As Mike's result were in fact identical for both blades (the error ranges actually overlapped) there was no need for significant detailed analysis. But you are still free to comment on what was done and presented in the PDF file. I am hoping that the edge stability comparisons that Paul is likely to do will show a difference and I will perform the comparisons in both ways to highlight the methods. I will also clean up the code I use (not in awk) and make it public for anyone who wants to run such comparisons on their own data. I will probably write it in excel/v-basic as most everyone has that now and it is fairly straightforward to use for basic numerical analysis and graphing. I would like to run it all in gnuplot as that is free but some of the calculations are not straightforward, it is not meant for that.

It would be interesting to see a FFD2 at 67 HRC vs CPM D2 at 64/65 HRC which will probably be done at a later date and ideally none of the testors will know which blade is which. Lots of other points of interest one as simple as just 1095 at 66 HRC vs FFD2 at 67 HRC to remove any hardness advantage and then check edge stability and long term slicing aggression (or just 1095 vs M2 at the same hardness). But that is months away, right now there will be lots of data of various types coming in with the Military and Sorg from various people doing various types of work. We will also get to see directly how the quantitative work done by Mike and Paul compares to the guys just using the knives in their work and hobbies.

-Cliff

 

[cds4byu]

Unless the correlation coefficient is absolute zero, which is very rare, the other parameters can always compensate to some degree, that of course is the definition of that coefficient.

I agree. But in this case, when I used octave to do the fit, the correlation coefficients were not even close to zero. According to what I understand from your answers, you didn't even look at the value of the correlation coefficients because you used gnuplot to do the nonlinear fit, and gnuplot doesn't report the correlation matrix.

The method I used nor the computer I ran it on was not bothered about the results, neither had an emotional attachment, nor paused for an instant during the calcuations. They simple produced a result which showed the parameters were undefined as the data variance was too large due to the calculations performed on the raw data which magnified the data.

You keep saying the parameters were undefined. I don't believe the parameters were undefined. I believe the a parameter had a value, and the uncertainty range of the parameter included both positive and negative numbers. This does not make the a parameter "undefined". Undefined has a specific mathematical meaning, and a is not "undefined" in the specific mathematical sense. If you want to claim that the parameters are undefined, you should be specific about the non-standard definition of "undefined" you are using.

As noted clearly many times, the actual raw data was perfectly fit by the model, as is every set of edge retention data I have seen to date.

I believe that the "transformed" data is also perfectly fit by the model. It fits just fine. I have yet to be able to understand what you mean when you say the transformed data is not fit well. I'd welcome your clarification.

It would be interesting to see a FFD2 at 67 HRC vs CPM D2 at 64/65 HRC which will probably be done at a later date and ideally none of the testors will know which blade is which. Lots of other points of interest one as simple as just 1095 at 66 HRC vs FFD2 at 67 HRC to remove any hardness advantage and then check edge stability and long term slicing aggression (or just 1095 vs M2 at the same hardness). But that is months away, right now there will be lots of data of various types coming in with the Military and Sorg from various people doing various types of work. We will also get to see directly how the quantitative work done by Mike and Paul compares to the guys just using the knives in their work and hobbies.

This would be interesting data to see. I'm glad you're pursuing it.

I think there is another factor to edge life, which is resistance to chipping. I believe that FFD2 has a significant advantage over 1095 at equivalent hardness in chip resistance, due to its extremely fine grain size. However, I don't yet have what I consider to be a geometry-independent test of chip resistance.


Carl

 

[Kohai999]

I have to admit that I did not understand what Cliff was saying above, I'm glad that you did!

Thank you from the bottom of my non-PhD heart for sticking with this...it has been really helpful and informative for all of us.

If I like the design, I plan on buying the "tactical" when it comes out.

Best Regards,

STeven Garsson

 

[db]

Unless you or someone else sends Cliff a free FFD2 blade I’ll be very surprised to see him test one.

 

[Broos]

In these FFD2 threads, Cliff objected to Wayne Goddard's method of looking at the reading on a scale while making a slice in the manilla rope. Cliff was also very quick to make the claims of bias in regards to these tests.

Now cutleryscience has presented testing that induces wear by slicing manilla rope, and measures the slicing ability by cutting twine at a much slower slice speed (vs. the manilla cut), with the force of the cut at a 45 degree angle to the edge.

To me this seems to be a very difficult measurement to make accurately (though GunMike obtained very consistent results), and that the sharpness being measured is not the same as the sharpness that is causing the wear (this was another of Cliff's objections to the FFD2 CATRA tests).

To my mind spotting the point on the edge where the twine breaks, then measuring the length of the cut is a more difficult measurement than looking down at a scale and reading it (a good constant slice is the hard part of Goddard's test).

Goddard's test has the additional advantage of measuring the slicing ability directly from the slicing that is causing the edge wear (like CATRA).

Which measurement has less noise? Is there bias shown by cliff's numerous objections to the FFD2 tests and not objecting to any aspect of this test that he "sponsored"?