Performance testing?

isn't that the whole point though? by testing a knifes cutting performance you are taking into account edge geometry, sharpess, grain structure and a load of other stuff. isn't that what we mean by performance testing? if you made two knives with identical geometry, you'd only be testing steel and heat treat, not the knife.
Click to expand...

The performance of a knife definitely depends on steel and heat treat. Blade and edge geometry as well as handle design and material are very important, too.

Or should we define performance prior to discussing it:

Performance Type A:

Description of the knife's properties by means of physical, mechanical, mathematical and metallurgical data and expressions, taking into consideration the complete geometry, steel, heat treat, experiments, etc....

The problem here is IMHO to find a "neutral" test which complies with "real usage" and where all parameters mentioned above are considered.

Performance Type B:

Based on user feedback, testing, personal experience and knowledge of how to use a knife. This is basically very complicated as long as you do not have a way to document the properties experienced in usage.

There will always be people that insist on type A and others will stick to type B. You cannot convince either to accept the other type, probably.

Regards

Mark23
 
There will always be people that insist on type A and others will stick to type B. You cannot convince either to accept the other type, probably.
Click to expand...

certainly it seems that a few folks fall into these apparently mutually exclusive subgroups. however, i reckon that most guys have some common sense, and will tell you that both types of testing have their place, it would be foolish to rely on one without the other. you couldn't develop new high performance steels without the science part, any more than you an design a practical, efficient tool, without having an end use, and an end user.

i'd suggest a further criterion for testing, "within design intent" the "wolfen" guy over at osf berated a comment i made, i suggested that an opinel (french wooden handled penknife) would slice veggies better in the kitchen than a busse. his retort was something like "every one knows busse knives are only supposed to be compared to other big combat knives" which is perhaps, when taken out of context, not quite as daft or childish as it intially sounded. it all comes back as you said to a definition of performance.

but i have to get on with the "honey do" list.

btw mark, thanks for the cutting machine link but it won't work for me, all i get is titles at the top of the page. any chance you could post a pic here?

cheers, and
 
machine_02.jpg


The homepage is http://www.catra.org

Navigate using "Products" - "Testing" - "Knives, Blades and Cutting Edges"

Your browser has to support frames, IMHO, to show the page correctly.

Regards

Mark23
 
It still come down to how the individual will use the knife. Cliff Stamp tests knives under the harshest conditions and, it seems, with the intention of destroying them.

If you have a machine that will test cutting ability, wouldn't all the tested blades have to have exactly the same edge profile and thickness? Otherwise, it would be like comparing a straight razor with an ax.

I don't know how many of you remember the tests Blade Forums ran a few years ago. The knives I remember were a Busse Basic 9, a Cold Steel Trailmaster and a Mad Dog knife. (I don't remember the model)
I remember that Mad Dog didn't do very well and he howled like a wounded coyote. The Basic 9 outcut the Trailmaster and CS complained that the testing wasn't fair. Both the Basic 9 and the Trailmaster were broken. I don't remember Jerry complaining about the test.

By the way, the broken test blades were sold recently on e-Bay. I considered buying them just to see if Jerry would honor his warrenty.;) ;) :D
 
If you have a machine that will test cutting ability, wouldn't all the tested blades have to have exactly the same edge profile and thickness? Otherwise, it would be like comparing a straight razor with an ax.
Click to expand...

no they don't need to be the same, if you had two knives the same, they'd be the same. if you were developing a new heat treat method or something, you would want two identical geometries with the different heat treatment. you'd be comparing the heat treatments.

in this instance we are hypothetically testing different knives, they are different. we are simply testing the knifes ability to cut the given medium. every knife is a compromise of various functions or attributes, it's a basic premis that can't really be denied. if you had a paracord cutting test. then the chances are that your straight razor would look better than your thick axe head. but when you come down to a test like the wood splitting i mentioned the other day, that razor is going to get pretty dinged up at the first knot in the wood. these are two tools at different ends of the spectrum, if that makes any sense. one very robust, and one very fine. as far as knife design goes, this compromise can be summed up with a simple question. "how much cutting performance are you prepared to sacrifice, in favour of robustness?" the question applies as much to the knife designer as the knife buyer.

if we start comparing like to like a bit more, things get a bit more subtil. does a similar size knife with a hollow grind cut more or less easily than the sabre, flat or convex ground knife? does a clip point penetrate more easily than a tanto? going back to the fire starting again, how finely can you produce curls or fuzz sticks? it may seem that i stress the fire starting thing a bit much. but it's a real world application that is much more likely to occour than the cutting your way out of a downed helicopter for example, for pretty much everyone except the helicopter pilot.

as an aside. i can say that i have read a real account of a guy cutting his way out of a downed helo in vietnam, it was one of the testimonials that came with an Ek knife i bought in about 1989. but if that ek knife was enough to cut his way out of the helo, or my reeves knife is enough to support my 190lbs when hammered into the tree, why weigh myself down with more? there was a good artical in tactical knives a while ago. the guy was doing a review of a sog recondo. (i know cliff broke one of those easily enough) the reviewer he was an actual graduate of the recondo school, which i think makes him qualified to discuss the subject. he made some good points on the need for need for maximum performance with minimum weight and encombrance. the guys in his unit were issued with kbars, but they were considered too big and heavy for an already overloaded lurp. this might seem a digression, but i'm trying to illustrate the other compromises that should be taken into account.

cheers, and.
 
Well, since we've boiled this down to load concerns... you could just go with a knife that is robust and an outstanding cutter. Like... say... a Busse.:D

As a side note, there are plenty of combat and field use testimonials on the various Busse forums. Mine among them. But, since I was never in Viet Nam... they probably don't hold water.:rolleyes:
 
Zius,

The methodology you describe in your response to me is the opposite of what you suggested in the passage which I quoted. In the passage I quoted, you suggest that the purpose of the testing is to validate, i.e., prove something to be correct, but what you describe in your response is the process of trying to falsify, i.e., trying to reveal and winnow out the wrong ideas. Further, in your original post, you say that the testing is for the sake of the mathematics behind the design, whereas, in your response, you describe a methodology of refining the mathematics for the sake of actually making objects well.

In your response, you have come over to my point of view, not explained the errors in my point of view.

--Mike
 
In regards to removing the human element, this while sounding "scientific" is in fact the opposite, as you have now designed a test which the outcome doesn't allow useful predicitions.

To be specific, when people cut with knives there are all kinds of random loads. The blade twists turns, accelerates, stops, and spins; the vectors keep changing at a significant pace in both direction and magnitude.

If you replace the human hand and arm with a lever all of this goes out the window. The blade then sees a load that is very removed from actual use and thus the results can't be directly compared.

You are of course still testing something, but they would not allow a predicted performance in the hands of a user, because you have removed critical elements which influence the result.

-Cliff
 
Cliff,

What you say may be true to an extent. The mechanical arm would offer consistency to the testing, whereas the twisting and deflection of what you speak is normal when a person is doing the testing but there is no way to standardize or account for this.
 
It would offer a very consistent results which can not be used to predict performance in use because you are testing a different set of factors. You can of course account for skill, and power simply by using different users over a broad range of abilties.

When I first started recording performance back around five years ago, I realized quite quickly that the abilities changed dramatically when I loaned the knife out to friends. Someone at 5'5" and 145 lbs, could have a very differnt perspective than somone at 6' and 250+ lbs.

This also makes comparing review very difficult, because of people using relative terms in an absolute sense due to a lack of comparisons to other knives. You can easily get two people finding that a knife is difficult to do something with and another saying it is very easy, if one of them is a lot stronger than the other.

The solution was obvious, I imagined the reviews with a sliding scale which you would set based on personal skill and power. The results could easily be interpolated from a database of users at various skill and power levels. This is why I constantly send out knives I review to others. Most recently a half a dozen machetes were sent out.

Jerry's suggestion is also a valid approach, build an arm which has the ability to vary power and consistency of stroke. The performance can then be evaluated over a range of settings. The user can then pick out what it personally most relevant by doing a simple few tests.

For example on chopping, measure the depth of penetration, and compare the required number of perfect hits to the actual number taken to thus determine raw power and consistency of swing.

-Cliff
 
there are plenty of combat and field use testimonials on the various Busse forums. Mine among them. But, since I was never in Viet Nam... they probably don't hold water. :rolleyes:
Click to expand...

yeah well, me neither. and i already have a canteen to hold water, so i'd be intrested in reading 'em anyway. if you can point us in the appropriate direction...

In regards to removing the human element, this while sounding "scientific" is in fact the opposite, as you have now designed a test which the outcome doesn't allow useful predicitions.
Click to expand...

cliff i disagree completely. if this was indeed the case, then these same points would be equally applicable to the standard busse lateral stress test. it could be argued that these tests are invalid, as we are unable to predict the radius of the car door sill, that our hapless survivor is trying to force open.

i believe that jerry's requirements were for tests that were easily specifiable, and repeatable "live". surely in order to produce lots of scientific sounding numbers the ideal is to remove the human element and eliminate unquantifiable variation. exactly how a test method is carried out, is not as important as ensuring that the samples being examined are tested to the exact same protocols. it matters less that busse carry out lateral stress tess in a vice, and fallkniven do theirs about a radius. the comparison is only valid if both products are tested using identical methodology. you can, as zius has suggested, apply some clever maths to extrapolate the results of one test, and predict how the sample will perform in the other test. (to be honest i'm not that fussed, my maths ain't that good.)

there is a simple truism that applys to engineers and scientists alike. "if you can't repeat it, you haven't done it"

i still don't know exactly what the cutlery and allied trades machine does. the test i envisaged was a sort of automatic guillotine, the knife blade is secured to an arm, one end of which is secured at a fulcrum. as the arm moves up a line of rope is feed through under the knife blade. a pneumatic piston drives the arm back down and the force required to sever the rope is measured and plotted. the arm goes up the rope goes in and so on. there are finer points that could be haggled over, you could have a slicing rather than chopping motion, but with movment about the fulcrum you get a degree of slicing antway. do you have a chopping board as an end stop. in all cases the point's at which the pressure is measured and torque applied will be the same.

this test effectively compares the cutting ablity of the samples. this includes the efficiency of the geometry, and edge holding. edge holding is dependant on the steel and the effectiveness of the heat treat, as it applies to that steel. so this test covers those points, and the busse lateral stress test covers robustness.

jerry has commented about other manufacturers claims not having factual or statistical basis. well, a man from sog once told me that their knives were good "stabbers", unfortunately, according to the regs, i'm not permitted to stab my scouts :rolleyes: :D so it's rather a moot point for me. but on that note. other tests i'd like to see developed are:

ease of penetration into meat or a meat substitute, covers both combat and filleting fish requirements.

chopping ability, to include edge retention/deformation. this could be hard to do.

most acute angle of slicing ability, covers woodworking skills for survival, and affects chopping ablity. (try chopping with both a cs hawk and a gb axe, you'll see what i mean.)

so, how're we doing for developing a range of class "A" factual/numerical tests then? when do we start developing the class "B" user driven subjective testing program eh? :cool:

cheers, and.
 
hi again cliff, i just re-read your last post, you could adjust the pressure developed by the pneumatic arm and say that below such a level of force the knife failed to penetrate the rope/plastic/rubber. if you wanted to introduce more of a random movement you could mount the knife on a spring "wrist" or something. but that would be more difficult to quantify also, and make it more difficult for sog or cold steel to reproduce the test rig and produce a similar set of numbers that were comparable to everyone else's.

cheers, and.
 
I can't really add anything to Sargey's last post, as it is exactly what I'm thinking too.

Evolute, what I said in my first post was:
"The reason to do testing should be to validate the mathematics behind the design of the blade, not to approximate performance in the field."

In my second post, I tried to explain what I meant. As far as I'm concerned, my two posts on the subject don't contradict. The crucial part is: "C) Test the model, to see if the conceptual model approaches reality in laboratory conditions". This is, IMO, the basis on which all other testing should take place. Scientific testing without a very high consistency between calculated results of the conceptual model and the results of the laboratory testing is useless, since either the conceptual model or the testing is faulty. Usually it is the first.

But what I'm missing in this discussion, is that the conceptual models aren't even defined.

For pure bending, it is relatively easy. An amount of torque could also be introduced into the setup.

But making a conceptual model of chopping, what actually happens with the knife and the chopped material is much more difficult (being a dynamic instead of a static problem), and beyond my personal expertise. The same applies for cutting.

Maybe, since we are interested in the performance of knives that are already made, we may do things the other way around, and build an empirical (as opposed to mathematical) conceptual model based on testing of existing knives. To do this, we would have to make a couple of series, in which only one characteristic is changed. For instance, make a test series with blades that have completely triangular cross-section and a standard edge bevel, and vary the spine thickness. Then proceed to test it with a couple of standardised tests. Chopping and cutting tests using a mechanical arm, and finally a breaking test. That is the only way to gain some real scientific insight in what makes a knife work.

No doubt that it'll be very expensive, as I can think of a lot of variables. And all the tests must be done by a respected research institute.

Most variables would deal with blade geometry. To name a couple, just to illustrate what I'm thinking of:
- amount of belly: from a triangular blade (with a straight spine) throug the different amounts of roundness of the belly toward a rectangular blade.
- The acuteness of the edge bevel: from completely blunt to no edge bevel at all (pure scandi grind). On different blade thicknesses, to effectively measure both cutting and chopping abilities.
- Different grinds: from very convex, to very concave.
And so on.
 
I think that this discussion has yielded some pretty good info. It is also indicative of the problems we faced when designing our in-house blade testing protocol. As for our public blade testing protocol, we pretty much followed the protocol that had been generally accepted by the knife industry and had been used by the ABS.

We take a lot of flack for "rope cutting" but keep in mind that we aren't the ones who started using this test media and that in the 80's and 90's it was the undisputed and most widely accepted test media in our industry and remains so today.

Jerry
 
sargey :

...hese same points would be equally applicable to the standard busse lateral stress test.
Click to expand...

No that is the opposite. The Busse bend test is done in the worst way possible, thus the user can expect at least this performance and usually more.

However cutting done as described by a rigid arm sets an upper bound on performance which can never be reached by any user. In use the performance will always be less.

It is even worse because you are testing a *different* set of material properties in the rigid arm test than when cutting is done by hand.

... exactly how a test method is carried out, is not as important as ensuring that the samples being examined are tested to the exact same protocols.
Click to expand...

The primary concern is that the end result be meaningful. There is no use to a test which is very precise, but not at all accurate. Try getting a grant for research along those lines.

there is a simple truism that applys to engineers and scientists alike. "if you can't repeat it, you haven't done it"
Click to expand...

There seems to be some perception that "scientific" work implies there is no human element, and that human beings can not be repeatable. This flies in the fact of the facts that humans did scientific work before the digital age.

Try doing some blade testing yourself and seeing just how repeatable the results will be. If you want lots of numbers I can send you thousands which you can look at statistically and test for variance, skew and the like.

I would very much like to see an independent knife testing lab, sure. However this is taking a pretty big step considering the current state of hype. At first why not develop some standard tests which people could do easily, and get some solid information developing along those lines.

These tests could then be refined as time goes on, depending on the interest.

Yes, you could build a robot arm to be as uncontrolled as a human arm, this I noted clearly in the above. This however would not be very easy - check and see the current state of robotics. The obvious question is then what have you gained over a human arm.

-Cliff
 
i agree on some of your points re the difficulty of designing tests that are useful in predicting real world peformance, i disagree that the slicing test rig gives an unfair avantage over a human arm. the force required is quantifiable, the same portion of blade is used every time... how ever the real world is a very random place, attempting to design an unsubjective test to predict subjective results seems a bit pointless to me at the moment. but let's go back a bit, to the test categories as defined by mark23

"Performance Type A:

Description of the knife's properties by means of physical, mechanical, mathematical and metallurgical data and expressions, taking into consideration the complete geometry, steel, heat treat, experiments, etc....

The problem here is IMHO to find a "neutral" test which complies with "real usage" and where all parameters mentioned above are considered.

Performance Type B:

Based on user feedback, testing, personal experience and knowledge of how to use a knife. This is basically very complicated as long as you do not have a way to document the properties experienced in usage.

There will always be people that insist on type A and others will stick to type B. You cannot convince either to accept the other type, probably."

i replied:

"certainly it seems that a few folks fall into these apparently mutually exclusive subgroups. however, i reckon that most guys have some common sense, and will tell you that both types of testing have their place, it would be foolish to rely on one without the other. you couldn't develop new high performance steels without the science part, any more than you can design a practical, efficient tool, without having an end use, and an end user."

so i was for the moment just concentrating on a couple of ideas for catagory A tests. questions i had wondered about in the past, just looking to establish a benchmark. so that different brands and models of knives can be meaningfully compared by the consumer. i may never get 45 miles to a litre of petrol with the way i drive my wife's new car, but we knew when we bought that it was likely to be more frugal than one of the other choices. comparing several products to one another across several disciplines, is never going to predict exactly how each model will perform in the hands of every different user. but it will give the consumer meaningful info on which to base his choices. that can only be for the common good.

after that we can move onto subjective end use testing, by various panels or test teams. set out a number of different tasks, some of which may overlap the lab testing, chopping for instance is now going to be affected by balance, length of blade for draw cuts, as well as the edge retention, edge robustness and possible angle of attack. skinning mammals and birds, filleting fish, and preparing veggies. RTAKs are ok for slicing spuds, but try peeling one with it. field expedient trap manufacture. as many tasks as you like. with as big a sample as is viable.

after that it's up to the consumer to decide/prioritise which tasks are more meaningful for them.

cheers, and.
 
Holy Crap!!! Haven't been able to post lately but had a chance to watch the battles. Performance testing to me is, I learn about the product, handle it, buy it, keep it if it works and sell it if it doesn't. My testing has changed over the years from when I was a kid whittling and playing in the woods, to my years in the Corp as a radio operator, to my 18 years in the Army as a S.E.R.E. instructor and Grunt. Now my biggest blade challenge since retirement is cutting soft irrigation pipe, to easy camping trips. I find I use my BA3 more than my SHBM now. I have found that the Busse line now fills all my needs except for the folder ;) I've owned alot of knives but these are MY preference.
BTW, sargey, spent some time with your T.A.'s around Pembrook and in the Breckens a few years ago. They were top notch.
Bob
 
This thread is a PERFECT example of the problems of agreeing on a good set of tests for knives. Here you have a group of people getting along very well while still agreeing to disagree (that in itself is a RARE thing), think of how hard it would be for people who hate a particular brand of knives, for whatever reason, to agree. Or knife makers, who have a very valid financial interest in the tests.

Therein lies the problem.

Congratulations to everyone for a very fine thread!
 
sargey :

i disagree that the slicing test rig gives an unfair avantage over a human arm. the force required is quantifiable, the same portion of blade is used every time...
Click to expand...

Yes, and it will be very precise. Meaning the results will show little variance. This is however only half of the equation. The accuracy, or how close the measured results are to the quantity they are supposed to be estimating, depends on completely different factors.

If the goal is to decide which steel works best in machine cutters, that is an excellent test. However if it is to predict which steel works best in a human hand, it is a pretty lousy test because it ignores critical aspects of a human hand.

As a trivial example, lets assume you test breaks on cars, using very careful equipment which measures very precisely the stopping power, and repeatability of breaks when the break is applied in a very controlled manner (by machines). You then build the breaks around these test, optomizing them for performance in that machine.

The car is now brought out and used by people. People who don't, and even can't, apply force in the same manner, especially when stressed. The net result is that the breaks fail horribly. How did a test which was so precise and repeatable fail to be useful? Because it was not an *accurate* measurement of the intended goal.

First off look at what you intend to test. Determine the influences. Now make sure that your testing method includes the same influences. You can test them together or apart, but you must include all of them. If this isn't done, then gads of precison are meaningless. They just allow you to be really precise about how wrong you were.

As a very simple example. Lets assume that you want to measure the length of a blade, which is known to be 10.0" . You have two measurements, 11.0 +/- 0.5" and 8.7564 +/- 0.0001". One of these is more precise, one of them is more accurate. Which one is "better". The ideal goal of course is to have high accuracy and precison, however neither imply the other. Precision is also worthless without little accuracy, accuracy is however valuable with little precision.

To clarify, my main point isn't against controlled testing. I think it can be very valuable, and materials properties are the place to start. The next point would be getting makers and manufacturers to be more public about the blade specifics, balance, edge geometry and so on, and then get some rough standards for performance in place. Once these start to get accepted, then they can be refined.

-Cliff
 
You must log in or register to reply here.
Back
Top