Hi Don,

If this question related to aircraft or metalwork engineering I would agree with your comments about profiling to help reduce stress concentration but wood is not a homogenous material like metal and we know that it is particularly weak in horizontal sheer so I rather suspect that radius profiling would only offer a limited improvement. Has anyone actually done any research test work on this topic to determine and publish "evidence based" guidance (other than as referenced below) ?

Mo,

There is a discussion on this topic together with an illustration and test report findings on pages 283 - 285 in Cecil Hewett's English Historic Carpentry where he publishes the findings made by Dr. D. M. Brohn of Bristol (University) in 1979 on this topic. Dr Brohn tested and compared the performance of four different types of joist to beam joints and his published results are quite surprising in that well designed joints tend to result in the beam starting to fail rather than the joist so this is not such a straight forward topic to answer since it seems that joint configuration plays a significant part in the overall structural performance of the assembly.

Andrea,

What is the NDS Code basis for specifying this requirement ?

Is it "evidence based" ?

Regards

Ken Hume

Hi Ken,
Digging around a bit;
The NDS commentary available in the download above(section 3.2.3.1) says;
Notches are a special problem in bending members due to the stress concentrations occuring at the corners and the difficulty of calculating the effects of shear and perpendicular to grain stresses occuring at such locations. These stress concentrations can be reduced by using gradually tapered rather than square cornered notches.

That comment references the Forest Products Labs "Wood Handbook"
Downloading chapter 8;
http://www.fpl.fs.fed.us/documnts/fplgtr/fplgtr113/ch08.pdf
The reference seems to be referring to the section beginning at the bottom of 8-6 and continuing onto the next page.

References 91,92,132 in the commentary look promising also.

I remember a comment either in the NDS or the codebook I think stating that a gradually tapered notch reduces shear stress to nearly unnotched levels.
..found it;
"3.4.3(4) A gradual change in cross section compared with a square cornered notch decreases the actual shear stress parallel to grain nearly to that computed for an unnotched bending member with a depth of Dn (the remaining depth)

I also did a quick search on the FPL site and got 28 hits on the word "notch" It doesn't take much for those boys to go over this carpenters head. They do perform testing there, they drive the NDS and our codes and do respond to mail as well.

Also to expand on Andrea's comment. The 1/4 depth is for end notching, 1/6 depth on compression edge of heavy timber in outer thirds of span, none in center third. In glulam 1/10 of depth or 3" whichever is less for ends, none elsewhere.

I'm sure not an engineer though.

Ken discussed the center of the beam in the vertical face, the neutral axis. I will add that joists, or any members, only installed on one side of a beam cause the beam to be eccentricly loaded, and I have seen several historical examples of beams severly twisted from being eccentrically loaded. Try to make the joist notch extend to the center of the width of the beam to help prevent eccentric loading, if possible. Even a heavy partition wall which lands on the edge of a beam can cause twisting. A beam loaded on both sides shouldn't need to have the notches extend to center, assuming the loads are reasonably balanced.

I just saw some pictures of common purlins in an old frame which were laid across the rafters in oversized trenches (trenched purlins)and then wedged. This technique essentially replaces the wood removed for the trench in the rafter with the purlin and wedge, minmizing any reduction in strength to the rafter or beam.

Jim

Hi Jim,

I agree that a floor joist sitting in a simple open notch will tend to cause twisting of the carrying beam unless the beam is fully supported across its section e.g. as in the case of a sill. An open notch is not capable of acting as a moment connection and can withdraw. When floor joists are joined to one side only of an unsuppported carrying beam then twisting effects on the carrying beam can be substantially overcome by using properly designed joints e.g. as per fig 303 in Hewett (page 281 / 282) where a tenon with diminished haunch and peg sitting on the carrying beam centre line is employed. We have this arrangement in our granary where the floor joists run axially in one bay and transversely in the other i.e. connecting floor joists on one side of the centre and (one) end cross beam only. The cross beam is now about 400 years old and there are no obvious signs of twisting (yet!).

If a deliberate choice is made to employ debased joinery then a price will undoubtedly need to be paid at some point in the future.

Regards

Ken Hume