Why joints loosen and split

How a joint fails depends on the type of force it receives and on its dimensions. This article sorts the failure modes reported in strength tests and describes which modes tend to appear in which joints. Loosening caused by the drying of wood is covered in a separate article, so only failure and deformation under load are treated here.

Failure modes

The failures reported in the tests fall into five types. The conditions of each test are given where the test first appears.

• Shear: the hooked part is sheared off in tension. It is a brittle failure in which the load drops suddenly, and it is the main failure of the koshikake-kama-tsugi (housed gooseneck scarf joint)^[1][2][3].
• Embedment: the faces of the ago (jaw) crush into each other along the grain. The load does not drop suddenly, and the joint shows ductility. It has been confirmed in tension of the koshikake-kama-tsugi^[3] and in bending of the okkake-tsugi (the sloped scarf splice without the pegs)^[4].
• Peg failure: in the kanawa-tsugi (rabbeted scarf splice with a square peg) in Toda's tension tests (sugi, repeated tension), lateral compression deformation of the peg and shear failure of the peg were seen^[2].
• Splitting: failure caused by lateral tensile force or shear force. An explanation on an-sd says the wood is torn along the grain^[8], while a paper in the Journal of the Japan Wood Research Society defines it as failure in which the wood is torn in the direction perpendicular to the grain^[10]. In the tension tests of Uchida et al. (spruce), compressive stress perpendicular to the grain arose in the koshikake-ari-tsugi (housed dovetail scarf joint), and the female piece split. In joints that use a shachi (wooden wedge) as well, the shachi is reported to have rotated and caused splitting of the female piece and cracks at the mechigai (a small locating tongue)^[1]. In the four-point bending tests of Kobayashi et al. (Douglas fir, okkake-tsugi without the daisen), splitting at both ends of the joint occurred first in every specimen, whatever the member depth^[4]. In tension of the okkake-daisen-tsugi (sloped scarf splice with square pegs) too, splitting from the mechigai caused by eccentricity was seen in some specimens^[2]. In the bending test modeled on a kanawa-tsugi in a sugi post (Agency for Cultural Affairs), the specimen without reinforcement also failed by splitting at the contact area at the top^[5].
• Bending fracture: the joint itself bends and breaks. Bending failure of the lower piece is reported in the tests of Kobayashi et al.^[4], and bending fracture at the root of the joint is reported in a specimen of the Agency for Cultural Affairs test that used wooden pegs and a steel band together as reinforcement^[5].

Failure that changes with dimensions

Even in the same joint, the mode shifts with dimensions. In Toda's tests, both the kanawa-tsugi and the okkake-daisen-tsugi failed mostly in shear when the member depth was small. As the depth increased, lateral compression failure and shear failure of the peg became more frequent in the kanawa-tsugi, and compression failure along the grain at the bearing faces became more frequent in the okkake-daisen-tsugi^[2]. The strength of the kanawa-tsugi did not rise when the joint length was increased at the same member depth, but rose when the member depth was increased^[2].

In the tension tests of the koshikake-kama-tsugi by Kijima et al. (Douglas fir, monotonic tension), the joint tended to fail by embedment when the ratio of shear area to bearing area was above about 9, and by shear when it was below. In the range where embedment governed, increasing the width of the gooseneck ago raised the maximum load, and in the range where shear governed, increasing the length of the gooseneck head did so; in both cases the gain levelled off beyond a certain dimension^[3].

In the tests of Kobayashi et al., specimens with a member depth of 120 mm failed in a brittle manner after splitting. The mean ductility factor (ultimate displacement divided by yield displacement) was 1.32 at 120 mm, 5.82 at 180 mm, 4.44 at 240 mm and 5.15 at 300 mm. The 180 mm and 300 mm specimens showed ductility through embedment of the ago, while the 240 mm specimens ended in a relatively brittle shear failure of the ago^[4].

Differences by joint shape

In the tests of Uchida et al. (spruce, 10.5 cm square, monotonic loading), the maximum tensile load decreased roughly in the order of the simplified gooseneck type (kanawa-tsugi and okkake-tsugi), the gooseneck type and the dovetail type. The okkake-tsugi reached 3,010 kg, the kanawa-tsugi modeled on an old example (furu-kanawa) 2,750 kg, the koshikake-kama-tsugi 1,087 kg and the koshikake-ari-tsugi 400 kg. However, the furu-kanawa that uses a shachi fell to 1,323 kg, about the same as the gooseneck type^[1]. The unit is kg as in the report (kilogram-force, kgf).

In the bending tests, all joints fell far below the 8,283 kg of the member without a joint: the kanawa-tsugi of today's standard dimensions reached 1,878 kg, the okkake-tsugi 1,443 kg, the koshikake-kama-tsugi 831 kg and the koshikake-ari-tsugi 220 kg^[1]. In the bending tests of the simplified gooseneck type, the joint rotating so that the shoulder faces shift, and cracks forming around the mechigai, are given as failures peculiar to this type. Cracks in the upper part of the side face appeared in the gooseneck type as well, after the maximum load^[1].

The work of the peg, and loosening

For loosening under load, the displacement of the joint measured in the tests is examined. In an example of the load-deformation relation in Toda's tests (Fig. 5), the displacement of the kanawa-tsugi at maximum load reached 10 mm, larger than that of the other joints in the same figure. The cause is given as the large lateral compression deformation of the peg^[2]. In the tension tests of Uchida et al., the load at the proportional limit of the koshikake-ari-tsugi was 198 kg, lower than the 2,525 kg of the okkake-tsugi, and its displacement at maximum load was 8.0 mm, larger than the 3.8 mm of the okkake-tsugi and the 2.8 mm of the koshikake-kama-tsugi^[1]. The tests differ, so the displacements of Toda and of Uchida et al. are not compared with each other.

In the Agency for Cultural Affairs test, adding wooden pegs as reinforcement encouraged splitting in the weak-axis direction^[5]. Takiyama et al. model the yield strength of the kanawa-tsugi in tension as determined by the smallest of three values: full lateral compression of the peg, shear failure beyond the ago, and splitting caused by the opening of the mechigai^[6]. These overlap in name with the failures seen in Toda's tests (lateral compression and shear of the peg in the kanawa-tsugi, and splitting from the mechigai in the okkake-daisen-tsugi), but no description of how the model corresponds to the tests could be found.

Shimoyama's commentary, in its explanations of the yatoi-sao-shachi-tsugi and of the joints with a through post, states that shachi pegs, komisen (square pegs) and split wedges make use of the elasticity and recovery of wood, so that loosening is unlikely to arise over the years. In another section it also writes that strength differs greatly depending on whether komisen and similar pegs have been driven tightly^[7].

Uchida et al. considered experiments that included the effect of drying at the planning stage but set them aside in order to focus on the relation between shape and strength^[1].

See them in 3D

• Kanawa-tsugi and okkake-daisen-tsugi: Kanawa-tsugi, Okkake-daisen-tsugi
• Gooseneck and dovetail joints: Koshikake-kama-tsugi, Koshikake-ari-tsugi
• All joints