Author: David Taylor MRIA, chartered engineer, is professor of materials engineering at Trinity College Dublin. He is interested in the failure of materials and structures, both in engineering components and in the natural world As materials go, bamboo is one of nature’s better ideas. Stronger than wood, lighter than metals and so fast-growing that it can be supplied in large quantities as a fully renewable source, it surely deserves a place among structural materials. In Asia, it is already being used in large quantities in buildings and scaffolding, and new composite materials are being developed in which bamboo fibres make up the reinforcing phase. [caption id="attachment_20969" align="alignright" width="263"]David Taylor Bamboo David Taylor MRIA, chartered engineer, is professor of materials engineering at Trinity College Dublin[/caption] So, a couple of years ago, I was surprised to discover that, when I searched the scientific literature, I could find no publications on the fatigue strength of bamboo. Having spent a long time studying fatigue in metals and other materials, and having seen for myself the tragic consequences of fatigue failure in many structures and components, it seemed unbelievable that people had used bamboo for so long without asking whether it might experience long-term failure as a result of cyclic loading. Working with my long-term collaborator in TCD’s Mechanical Engineering Department, Peter O’Reilly, and two final-year project students Patrick O’Hanlon and Lauren Keogh, we carried out a series of fatigue tests on samples of bamboo, the results of which were recently published in the International Journal of Fatigue*. These data provide a very basic and simple characterisation of fatigue in the material: there’s a lot more to do, but at least we have broken the ice.

Bamboo is extremely anisotropic

One interesting feature of bamboo is that it is extremely anisotropic. The mechanical properties measured by applying load along the length of the stem (known as a culm) are very much greater than those measured transversely. This anisotropy is greater than found in wood, and in any structural engineering material. The tensile strength of bamboo in the longitudinal direction is, strictly speaking, unknown, because if one applies force in that direction the sample will eventually fail by longitudinal splitting. It is almost impossible to get it to break across the culm in a controlled manner. Its strength is certainly more than 150MPa, making it about 10 times stronger in this direction than when loaded across the culm. Not surprisingly, then, we also found that its fatigue behaviour was very different depending on the testing direction. We cut short lengths of culm, which conveniently make samples which are almost perfect cylinders, and loaded them in one of two ways: parallel to the culm axis in longitudinal compression, or perpendicularly by applying compression across the diameter. In longitudinal compression we found that fatigue failure simply didn’t happen. Given enough force a sample would fail, by longitudinal splitting, but if it did not fail on the first application of a load then the same load could be applied a million times with no failure occurring. This kind of behaviour is unusual but not entirely unprecedented. It occurs in some ceramic materials and in engineering fibre composite materials in which the fibres are arranged uniaxially, all running in the same direction, if load is applied parallel to the fibre. Such engineering composites are rarely used in practice but, with bamboo, nature appears happy to put all its eggs in one basket, directionally speaking. We thought “maybe this is why there’s no information on fatigue in bamboo: perhaps it just doesn’t happen”. But it was a very different story when we loaded samples in the perpendicular direction. As well as being much weaker, they also displayed a large fatigue range. To minimise scatter and improve accuracy we tested samples in pairs. Cutting two samples from the same part of the plant, we tested one with a monotonically increasing load until it failed, establishing a strength value. We then carried out a fatigue test on the other sample from the pair, loading it with a sinusoidally varying force which had a maximum value set to some percentage of the strength of the first sample. In this way we established that fatigue failure can happen with applied loads as small as 40 per cent of the static strength and that, as in most other materials, the number of cycles to failure increases rapidly as the applied cyclic load decreases. [caption id="attachment_20975" align="alignright" width="300"]Bamboo Table Fatigue data for bamboo: number of cycles to failure as a function of applied load, expressed as a percentage of the static failure load, for samples loaded in compression across the diameter of the culm (click to enlarge)[/caption] In these tests failure invariably occurred by longitudinal splitting, which was expected: what was surprising was the location of the failures. Imagine the circular cross-section of the culm as the face of a clock, with compressive load applied along the 12 o’clock – 6 o’clock axis. In both the static and fatigue tests, failure occurred first by the formation of a crack at the 3 o’clock or 9 o’clock positions. This is surprising because, though theory would predict regions of high tensile stress in those locations, there are much higher tensile stresses at the 12 o’clock and 6 o’clock positions, on the inside of the tube.

Bamboo has a graded structure, harder on the outside and softer on the inside

The reason why failures didn’t occur there is still not completely clear but it is probably related to the fact that the material in the culm wall is not the same throughout: bamboo has a graded structure, being harder on the outside and softer on the inside. In this case cracks seem to form preferentially in the harder, more brittle regions. Another surprising finding was that this initial splitting, at 3 or 9 o’clock, did not lead to catastrophic failure. It was possible to continue loading the sample, with the same cyclic forces, until eventually a crack would form on the other side (9 or 3 o’clock) and then, finally, at 12 or 6 o’clock, before complete collapse of the sample occurred. This suggests a considerable capacity to tolerate defects, at least for this particular type of loading. In practice bamboo is not often going to be loaded in compression directly across the culm diameter as in our tests. A more common type of loading – both in structures made from bamboo and in the original plant – is bending applied to relatively long lengths of culm. A common mechanism of failure in relatively thin-walled tubes loaded in bending is so-called ovalisation, in which the initially circular section becomes elliptical, reducing its moment of inertia and creating high stress concentration points at the major and minor axis. Our diametral compression loading created a similar elliptical deformation and so may be a reasonable simulation of a common failure mode. Despite our precautions there was still a lot of scatter in the data, but this is quite common for brittle materials whose fatigue life is strongly affected by small pre-existing flaws and minor variations in material quality from place to place. We tested two different species of bamboo: one which is very commonly used as a construction material and another which we found growing in the large greenhouse at the National Botanic Gardens in Glasnevin, kindly donated by them. We also carried out some tests to investigate the effect of drilled holes and other defects introduced into the samples. This work, though very preliminary in nature, has established that fatigue is a likely mode of failure in bamboo culms, and the data we generated have allowed us to make some tentative recommendations regarding the use of bamboo in situations where it can be expected to experience repetitive loading. *Lauren Keogh, Patrick O’Hanlon, Peter O’Reilly and David Taylor (2015) Fatigue in Bamboo. International Journal of Fatigue 75: 51-56.