On The Pull - The Pros and Cons of Mechanically Testing Component Parts of Trees

As an ageing tutor, it becomes increasingly harder to understand the language that youngsters use, who come to be trained at College and, probably, vice versa. This effect makes me feel somewhat analogue in this digital age. However, some terms oddly persist, such as ‘hoovering’ (even when using a Dyson vacuum cleaner – no-one says “I’m dysoning”, do they?), ‘video’ (for the content of a digital streaming service) and ‘new record’ (for a pop single released digitally) – because there is always some continuity in terms, as well as some disparity. Recently, I asked a student where they found an odd ‘nugget’ of information for their work project, and they answered: “In my feed.” For a brief moment, confusion!

Terms are changing faster due to our greater reliance on the internet, which is growing ever greater (Google says) whereas the reading of books is on its way out (according to Bing). Will modern students gain an understanding of trees and tree terms through browsing the internet or through ‘book learning’? Besides ‘book learning’ is often seen as a pejorative term – employers often value a person with obvious practical knowledge and experience, not someone who has read and understood a book. However, for those starting to study the failure of trees, their understanding of the topic is best supported by some authenticated, peer-reviewed, well-worded and edited information or evidence, alongside gaining experience, as experience of tree failures is only gained slowly over many years. Good journal papers and scientific books can speed up this learning process considerably – whereas the internet is an information minefield, containing much unauthenticated nonsense, which many students struggle to navigate through without ‘falling down rabbit holes.’

Predicting a tree failure is a particularly difficult aspect of arboriculture: who hasn’t had the experience of having a tree condemned for its basal decay, only to find it is still standing 15 years later? Students and young people – that’s who haven’t! I look to take my students through key features that can help to predict a tree failure – such as actively growing cracks, abnormal doglegs in stems and branches (Fig. 1) or changes in a tree stem’s lean – being clear that there are always exceptions to any rule.

Mechanical Testing of Tree Components

One way of estimating the likelihood of a tree failure is to test the strength of component parts of a tree using accurate testing equipment. This process can be done outside or inside, with a wide range of equipment being available for this purpose. This article concerns the testing common to many scientific studies, which involves using universal testing machines (UTMs), typically used in an indoor setting. Instron? is a major manufacturer of UTMs – so, like a ‘hoover’, this category of machine is often referred to just as ‘an Instron’, even if the machine isn’t that trademark. When creating bespoken testing rigs for my allocated Instron? UTM, it was initially a surprise that I had to cater for very odd diameters of holding rings: 12.7 mm, 19.5 mm, 25.4 mm: then I realized that the Instron corporation was based in the US – and despite its sophistication, these machines are based around ‘the inch’ as a measurement, whereas I grew up in a (mostly) metric part of the world. Thank goodness an Instron? UTM doesn’t measure temperature too!

These UTMs are very versatile, allowing a wide range of test types, depending on the apparatus that are attached to them. They are also very robust and strong: you don’t want to get your shirt cuff or cardigan sleeve caught in the mechanism for, after about an hour, it will be ripping off your arm! Lol! Not really - most testing is done at an awfully slow rate and there is an emergency stop button as well, so don’t worry: it’s not like a tractor’s PTO… fortunately...?

Common Instruments of Torture

For component parts of trees, there are three common tests: tests for tensile strength, compressive strength and bending strength. A bending test puts areas of the opposing edges of the component into tension and compression, whilst the middle of the component may fail in shear (Ennos & Van Casteren, 2010). Torsional and shear strength of tree components can be estimated using a UTM, but that is less frequently reported in the academic record. Sometimes you will find all such tests described as ‘rupture tests’ – which I think is a bit vague (are we rupturing the structure in tension or compression, or both?).

The images here illustrate the bespoken test set-ups that I used to evaluate tree forks and the wood of trees for the completion of my PhD (Figs. 2 to 4).

The basic output of this type of testing is a force/displacement diagram, where the force used to move the UTM’s crosshead is plotted against the displacement of the crosshead. Plotting the force against displacement twenty times per second provides good graphical illustration of the performance of the material or component (Fig. 5)

There are a lot of advantages to this testing method, aside from potentially finding the relative strength of component parts of a tree, When you are a UK-based student completing the third year of a BSc(Hons) degree from September to March, you might wondering how you can possibly do a research project involving trees through the winter months (i.e., there are no leaves on deciduous trees, no growth rate in trees to speak of, no insect damage to monitor or treat, etc…). Using a UTM for mechanical testing of tree components is a ready fix for such a student’s dilemma. This research method also produces data very quickly: as soon as you press the ‘Run’ button, data is generated – a lot of it – in a very short period of time.?

Functional Anatomy

One advantage of this method that is not often mentioned in the associated literature is that, when testing tree components, the failure that the UTM induces can highlight specific tissues or cell types that are playing a key role, helping to explain the failure mode that occurs in that component. Figures 6a & 6b are a case in point: finding what looked like centrally placed interconnecting tissues within tree forks when splitting them apart was crucial evidence for the strengthening effect of axillary wood (the wood that forms under a branch junction’s bark ridge (JBR)). Split, cracked or extracted wood fibres tell an important part of the story of a tree failure, if observed carefully – and the slow failures induced by UTMs allow for such observations to be made reliably and repeatedly (Slater & Ennos, 2013).

?What Mechanical Testing Cannot Tell You

Unfortunately, most testing by UTMs does not mimic the loading that plants/trees receive in their natural setting – gardens, parks and woodland. Nature does not impose a force that results in a constant rate of displacement in tree tissues for prolonged periods of time, for instance – rather, branches and stems are bent and twisted erratically by wind gusts in sudden movements.

A first limitation, then, is that testing with a UTM often cannot provide the combination of speed and forces that act in Nature. For example, branches of trees often fail in a combination of bending and torsion (Avalos & Sánchez, 2014), and it would be difficult to develop a mechanical test that provides such a combination of forces in the anarchic way in which the wind applies these to the trees, at constantly changing proportions.

A second limitation to these tests is the speed of displacement needed to evaluate wood strength. The testing standards are set around evaluating dried timber samples – not greenwood – and test rates are typically slow (e.g., 10 mm/second of displacement) so that details of yielding and failure of the material can be recorded by the UTM. However, for greenwood, if the testing is conducted too slowly, then one can get ‘creep’ in the test samples, resulting in the specimens failing at a lower force than it would when tested at a faster rate. However, test the specimen at a fast rate of displacement and the reading may be inaccurate, due to inducing an artificial brittle-like failure and/or potentially shortening the dataset due to the rate of recording the UTM is capable of. There’s no winning, really, when setting a speed of displacement for testing greenwood with a UTM – it’s always a compromise.

A further limitation is being able to interpret the mechanical outputs of such tests (e.g., yield strength, bending strength, extent of plastic deformation) from such static testing, and then apply it meaningfully to a standing tree, which is a dynamic structure. Many conclusions of older mechanical studies concerning trees can now be disregarded as they fail to incorporate the key concept of mass dampening – that trees flex and bend their way out of failure, often performing a complex ‘dance’ with their twigs and limbs, to avoid damaging wind loading (James et al., 2006). Essentially, one must set ‘the particular’ in the context of ‘the whole’ correctly, to come to the right conclusion when analysing data from this form of testing.

My PhD research involved this kind of artificial mechanical testing, and to overcome these limitations, I chose to do some in-field assessments using accelerometers on trees as they moved in high winds (Slater & Ennos, 2016). This was not only an important counterpoint to my mechanical testing, but it was the first step towards finding that natural bracing was an important influence on the morphology and failure potential of tree forks. Natural bracing is not something that would have ever been discovered in the labs – although we did end up testing it there (Meadows & Slater, 2020).

There are many classic scientific errors (not just Type I and Type II errors in statistics). ?Errors worth looking out for, when studying arboriculture, are ‘hasty generalization’, such as claiming a factor of safety for ALL trees based on testing just a few specimens of only two species (Mattheck et al., 1993), omissions of important factors, misapplication of statistical testing (which happens more often than hot dinners, apparently ??) and systemic errors.

The worst draft paper I have reviewed thus far in my academic career was one that assessed the suitability of a range of trees for planting in coastal areas. They tested wood samples from many tree species – and pulled trees to assess the strength of their stems and anchorage. Their conclusion was that palm species were unsuitable to plant by the coast, as their wood was weak and they were easier to pull over by mechanical testing than other species (!) A clear error by omission – it is well known that palms are very successful as coastal trees because of the way that they shed fronds and bend their stems to align themselves with the wind direction, even in extreme gusts. Omit a key factor, and your findings will be misleading or meaningless.

In the case of pulling apart tree forks with a UTM, there has been an on-going systematic error for decades (Slater, 2021). This error relates to how stresses (in megapascals (MPa)) are calculated in such tests. Figure 7a identifies how previous studies have calculated leverage and subsequent stress levels when testing bifurcated specimens. However, the suggested pivot point for the stress calculation lies in-line with the apex (top) of the junction, which is illogical. To pull apart such a junction, one must apply leverage at a pivot point set lower down in the component (Fig 7b), to pivot and break open the junction (Slater, 2021). This is not a major error if the lever arm is long and the fork being tested is of a small size – but when the lever arm is set at just 52 mm and the forks tested are substantial in size, the bending strength of a fork may be underestimated by more than 40%.

By not attending to these potential errors, some authors on this topic espouse that they have proved that all forks in trees are defects – a fallacy that is being perpetuated, sadly, to the detriment of our tree care industry. Two of my former students have conducted studies, varying the length of the lever arm for testing juvenile tree forks. At 100 mm and 200 mm of lever arm, the forks nearly all break at their join, somewhere near their junction bark ridge. However, extend that lever arm to 400 mm and 800 mm, and there are many more failures of the branches, much fewer failures at the forks themselves. In essence, using longer branches results in greater flexure in those branches during testing, and as the branches curve the bending stresses concentrate at the point of greatest curvature, and thus failures occur more often in the branches and less often at the branch junction/fork (example shown in Fig. 6c). What these preliminary experiments show is that, for this component part (a branch junction) it really matters how it is tested, as to the conclusions one would draw: short lever arm = forks are weak; long lever arm = forks are a good match for the branches that they bear. Quite different conclusions from remarkably similar testing!

Previous claims that all tree forks are defects (e.g., Smiley, 2003; Gilman, 2003; Kane et al., 2008) need to be seen in the light of both of these systematic errors – and that the likelihood of fork failure, in the real world, varies significantly across woody plant species (i.e., to avoid a hasty generalisation from testing just a few species).

?

The Value of Mechanical Testing

Accepting that one is not mimicking Nature, mechanical testing can still provide us with useful technical information, so long as the comparisons being made are carefully thought through. For wood testing, any strength tables should be accompanied by the wood density of the associated samples, the location where the trees were grown and the trees’ age, as the context of such testing is so important. For component testing, one must consider the role that the tested component needs to play within the tree’s structure. For example, a horizontal branch has a different morphology and different bending strength than a vertically aligned branch, as they experience different forces as they develop, to which they acclimate. Trying to find a simple rule about the bending strength of all branches is, thus, a pointless pursuit, as such strength will vary widely, according to the alignment, structure and function of each branch.

.Although many of us would wish it to be the case, simple rules rarely apply to trees – especially with regards to tree failure. A failure is often the results of complex interactions of several factors, which the arborist/arboriculturist must judge in the light of tree-related science, but rarely directly supported by it.

Figure 8 shows the failure of a leaning beech tree; factors involved in this failure will have included the type of storm that was involved, the direction of the prevailing wind, the long-term poaching of the ground by cattle at the base of the tree and the subsequent root rot that was induced, alongside the progressive lean of the tree. These are all likely factors in the equation to predict this tree’s failure. Scientific research can consider events caused by multiple factors – but it needs a remarkably high number of replicates (into the hundreds) to determine the relevant significance of each one. This helps to explain where we currently are with our understanding of tree failures – some niche but accurate studies of single factors have been reported - some poor studies espousing simple global rules for ALL trees that confuse the overall picture – and, mostly, we are highly reliant on an arborist’s experience to assess risks from trees – as they can, in their wise old head, combine factors in the instance of a single tree that scientists have yet to combine in sufficient numbers to provide a reliable prognosis (e.g., what science may be able to offer is a statement such as: “of ten trees in state X, on average, 40% will fail within five years”).?

Conclusions

Attributed to several people is this insightful quote that helps to conclude this article: “In theory, there is no difference between theory and practice – in practice, there is!” For tree risk assessment, it is especially important that any theory or interpretation of an experiment closely matches with reality, or poor decisions will be made.?

Acknowledgements

For supervising my PhD, I must thank my long-suffering supervisor, Prof. Roland Ennos – who put me on the straight-and-narrow path every time that I wandered off from it. My thanks must also go to ex-students Chi Kin Lam and George Harding for their experiments to identify differences in mode of failure due to differing lever arm length for the mechanically testing of tree forks. For the illustrated diagrams used in this article, my thanks go to David Elwell.?

References

Avalos J and Sánchez A (2014) Evaluation of failure criteria in branch members under torsion and bending moment. Arboriculture and Urban Forestry 40 (1), 6-45.

Ennos A R?and Van Casteren A (2010) Transverse stresses and modes of failure in tree branches and other beams.?Proceedings of the Royal Society B: Biological Sciences,?277(1685), 1253-1258.

Gilman E F (2003) Branch to stem diameter affects strength of attachment. Journal of Arboriculture 29, 291-294.

James K R, Haritos N and Ades PK (2006) Mechanical stability of trees under dynamic loads.?American Journal of Botany?93(10), 1522-1530.

Kane B, Farrell R, Zedaker S M, Loferski J R and Smith D W (2008) Failure mode and prediction of the strength of branch attachments. Arboriculture & Urban Forestry 34, 308-316.

Mattheck C, Bethge K and Sh?fer J (1993) Safety factors in trees. Journal of Theoretical Biology 165, 185-189.

Meadows D and Slater D (2020) Assessment of the bending strength of bark-included junctions in Crataegus monogyna Jacq. in the presence and absence of natural braces. Arboriculture and Urban Forestry 46(3), 211-228.

Slater D and Ennos A R (2013) Determining the mechanical properties of hazel forks by testing their component parts. Trees: Structure and Function 27, 1515-1524.

Slater D and Ennos A R (2015) Interlocking wood grain patterns provide improved wood strength properties in forks of hazel (Corylus avellana L.). Arboricultural Journal 37, 21-32.

Slater D and Ennos A R (2016) An assessment of the movement behaviour of hazel bifurcations under dynamic wind loading using tri-axial accelerometers. Arboricultural Journal 38, 183-203.

Slater D (2021) The mechanical effects of bulges developed around bark-included branch junctions of hazel (Corylus avellana L.) and other trees. Trees: Structure & Function 35, 513-526.

Smiley E T (2003) Does included bark reduce the strength of co-dominant stems? Journal of Arboriculture 29, 104-106.

***This article first appeared in the Autumn edition of the Arb Magazine, published by the Arboricultural Association***