Timber testing 2

[glass99]

I have noticed is that there is a large delta between average timber strength test values and those prescribed by NDS, presumably for statistical variability reasons. Red oak for example has a tested average bending strength of 15,000 psi vs a specified allowable of 1,150 psi, even though a factor of safety of only ~1.84 is required by the NDS.

If you were to say to build a high performance truss from relatively small timber components and tested a sample from every component, could much higher strengths be used?

 

[phamENG]

That would probably be a question for the building official. Will they accept independent testing in lieu of codified requirements? For new construction, I think most would probably say no.

What are the conditions of the tests to obtain the average of 15,000psi? Are those clear specimens with 0 degree grain orientation, no checking, no knots, etc.? The average is adjusted to somewhere in the neighborhood of 5% (95% of the distribution above the value) and that is used as the reference design stress. If it's visually graded, you'll have the reference stress(about 10% of the average from clear specimens) x reduction factors for grain angle, imperfects, knots, and all of that. So for your red oak example, 15,000psi average from clear specimens x .10 = 1500psi. 1150/1500=0.7667. So you have a total reduction of 23.33% for whatever imperfections are allowed in that grade of lumber. That sounds about right to me.

In historical projects, where the lumber can't be judged based on modern stress values (because those stress values are updated periodically based on real world industry tests at lumber mills and are being effected by climate, bio-engineering and breeding of trees, etc.), it is not uncommon to test a representative sample and then visually grade the lumber in situ. For critical elements like exposed scissor trusses in a church spaced at 20' o/c, you might do every one individually, whereas the floor joists under the sanctuary might get every 20th joist. The reference design stress value for each case is then adjusted based on a rational variability (near 0 for the truss where everything was checked, and much higher for the joists where 1/20th were checked).

 

[SlideRuleEra]

Red oak for example has a tested average bending strength of 15,000 psi vs a specified allowable of 1,150 psi...

For wood, the "bending strength" is called the Modulus of Rupture (15,000 psi for red oak). MOR is the loading required to break the wood member by bending.

Allowable bending strength (1,150 psi for red oak) is much less than MOR because wood's Modulus of Elasticity (MOE) is correct for a limited range.

In other words, if a wood member is loaded in bending it will deflect proportional to stress (per it's MOE) up to the allowable bending strength. Allowable bending stress is the practical upper limit of wood's MOE range.
As loading continues to increase, the wood member's deflection increases rapidly... but does not break until stress reaches the MOR.

If a wood structure is heavily loaded, as you mentioned, it would be expected to distort dramatically (and likely not be usable for it's intended purpose), but not fail until some member was stressed to it's MOR.

Note that for steel the upper limit of the MOE range is considered to be the "yield strength". Wood does not have a "yield strength", it just becomes more or less "structurally unusable" without breaking (or, more likely, distortion of wood cause a connection to fail... bringing down the structure before any wood member reaches it's MOR).

http://www.SlideRuleEra.net

 

[glass99]

What are the conditions of the tests to obtain the average of 15,000psi

This was for a 1.5" diameter timber handrail which was basically "pristine" in respect to grain orientation, knots etc.

95% confidence interval implies two standard deviations. If you have test data which has an average value of 15000psi and a standard deviation of 4000psi, you are down to 7000psi even before any safety or reduction factors have been applied. We in the structural glass world tend to deal with statistical issue through redundancy (primarily lamination) rather than really high factors of safety. A factor of safety of 13 is really high. What if you really do the work to make sure that every element has no knots and the grain direction is 0 degrees +/-1 deg?

As loading continues to increase, the wood member's deflection increases rapidly

Good point, but this could be taken into account in the calculation. I am imagining a feature element type application, and not a floor joist, so 10x engineering effort per square inch is ok.

 

[phamENG]

I think your best bet, if you plan to go down this road, is to review ASTM D143, Test Methods for Small Clear Specimens of Timber, D2555 Practice for Establishing Clear Wood Strength Values, and D245 Standard Practice for Establishing Structural Grades and Related Allowable Properties for Visually Graded Lumber.

What you're describing is, essentially, creating your own "grade" of lumber. To that end, and I apologize for hijacking your thread, but does anyone out there know of any classes or certifications for engineers in this regard? I've done it some as needed, but with very conservative assumptions. It's a useful skill in my region, and I'd like to hone it a bit more.

 

[SlideRuleEra]

phamENG - "Southern Pine Inspection Bureau Grading Course", Fall 2020

glass99 - Before grading to push bending stress to max, you may want to check horizontal shear for the increased loading that higher bending stress permits. Note that no matter how "good" the lumber grade, allowable horizontal shear does not improve. Ultimate horizontal shear will be much higher than the allowable values published, but still should be checked.

Same thing for Compression Perpendicular to Grain (the supports and possibly the loading location).

http://www.SlideRuleEra.net

 

[dhengr]

Glass99:
It seems to me that you are trying way too hard to make a wooden product, act/perform like glass steel, aluminum, etc. which are all homogeneous and generally isotropic materials, while wood is neither of these. For something like you are talking about, I wouldn’t have a problem with studying the grading rules which apply to various grades of run-of-the-mill construction lumber of a given species, and understanding how gross items (defects, deficiencies) like knots, checks, poorly oriented grain, etc. affect their strength, and then allowing some adjustment in their allowable stress values. I doubt that I would be comfortable with a change from 1.15ksi to 15ksi, just because a clear/perfect sample showed the latter as the Modulus of Rupture (MOR) in a lab bending test. Wood also shows a surprising ability to tolerate impact and short term loading, which might be your salvation on the system you are looking at. Otherwise, the virgin material is just to variable, foot by foot in length, and cubic inch by cubic inch in volume to allow anything near the MOR to be used for design. You just have to build a few samples of your structural system and test them, working together, to destruction to start to pin down an increase in element strengths.

 

[glass99]

Does the NDS or other recognized literature give you procedures for doing the testing and statistics? Like all codes, the NDS allows for testing right up front, but doesn't seem to have a lot of details on how. For example, if I build a whole truss and proof load it, can I use the building code FOS of 2.5 for testing? Assume that its part of a facade system and is primarily wind load governed. If I do my testing and QA, I will have designed a truss in Red Oak at a stress of 15000/2.5 = 6000psi.

dhengr:
I think of timber like natures carbon fiber. Crappy QA, but much better carbon footprint. I'm looking to align the forces with the grain.

SlideRuleEra: good point about the shear strength parallel to the grain

 

[SlideRuleEra]

Does the NDS or other recognized literature give you procedures for doing the testing and statistics?

The research that forms the basis of NDS is performed by the U.S Department of Agriculture Forest Products Laboratory (FPL). Being a government agency, virtually everything they do is in the public domain and readily available. Suggest you start with their publication "Wood Handbook, Wood As An Engineering Material". You may gets some leads to pubs spelling out testing procedures.

IMHO, for short term loading you probably can significantly beat the published allowable design values, but 6000 psi is still too high... 2500 psi, good chance, 3000 psi, maybe.

Keep in mind that the 10 minute NDS Load Duration Factor (for wind loading) is 1.6. Allowable stress on "Select Structural" Red Oak is already 1.6 x 1150 psi = 1840 psi.

Also...

See "Determining Allowable Design Stresses for Timber using ASTM Standards D2555 and D245"

http://www.SlideRuleEra.net

 

[dhengr]

Glass99:
Your thinking of carbon fibers and wood is somewhat suspect, and part of your error is trying to use 15ksi/2.5. Your jump from a 1.5” dia. handrail to a truss is quite a leap. The problem is that the stresses are aligned by the structure shape/geometry, the support system, parts interactions, and the loading, and you have some control over the first three, and only some ideas of the forth, and fairly limited control over the grain orientation/alignment and imperfections or other natural defects/deficiencies of the wood. I’d talk with the AHJ (it kinda needs their blessing to be approved, or the ICC-ES.org people who do this kind of testing for the code world) and lay out a testing plan/protocol. I’d test a few to failure, study what failed first, improve the design, test again to a new/improved failure strength, maybe a different kind of failure, and finally divide that failure load/strength by 2 or 2.5, or some such, for a max. allowable value. And, if you test 100 times, you’ll have a better average than if you only test twice. The idea that any code would tell you (lay out the steps) for testing is crazy. There are only four trillion things and ideas which need testing and the next guy would want to do the test another way, maybe upside down or something. We, as engineers, should be able to develop an appropriate, and unbiased, testing program to prove our design to a reasonable degree of satisfaction of most practical and knowledgeable observers.

 

[phamENG]

So here's a tangentially related question. We're talking a lot about building a couple trusses and breaking them. I get the MiTek/USP guys doing it, but does this actually happen in the real world for individual projects? Who on earth has the money to do that, and how can I start working for them?

 

[glass99]

dhengr: carbon fiber is also highly anisotropic. Its quite disconcerting to know how poor carbon is if you get the fiber orientation a little bit wrong. The first thing you want to do is use unidirectional fiber, but then you realize you can't...

@slideruleera: thank you, I think that's exactly what I'm looking for

@phamENG: yes you are right that getting people to pay is the hard part. Time and uncertainty are actually the bigger hurdles. My usual argument is that my $1MM feature wall makes your $1BB skyscraper worth $1.1BB (especially if it makes it look green). Budgets for lobby renovations for example are frequently in the range of $30MM, so peeling off a few million for a gorgeous timber wall is not out of the question. They easily spend a couple of million on millwork as it is. Building 10 real trusses and burning one for a test is usually doable. High end residential is another good market for this kind of thing.