The study of wood fatigue spans over 70 years, with growing interest driven by environmental concerns. Wood's high stiffness-to-weight ratio and capacity to store atmospheric carbon give it a significant advantage over materials like steel and concrete.
However, as a hygroscopic, heterogeneous, and orthotropic material, its response to mechanical loads is difficult to predict, leading to over-design and higher material consumption.
With the increasing frequency of seismic events and storms, understanding wood's behaviour under cyclic loads is essential for designing safe and durable structures.
Historically, wood was believed to be immune to fatigue, a notion supported by its use in World War II aircraft. Fatigue was often mitigated by considerations for creep.
Current design standards, such as Eurocode 5, lack comprehensive fatigue-based design rules due to the absence of reliable analytical models. The next generation of standards aims to address this gap.
Despite progress, wood fatigue remains a specialised field. Since fatigue accounts for up to 90 percent of material failures, a better understanding is crucial for optimising the use of wood as a renewable resource.
This review synthesises existing literature on the cyclic fatigue of wood, covering influencing factors, materials, measurement methods, predictive models, and practical applications.
Influence Of Physical Properties
Firstly, density is a key physical indicator correlated with many mechanical properties. It is closely linked to moisture content and varies with growth conditions and within a single tree (e.g., due to knots, earlywood/latewood).
Research generally indicates that higher density improves fatigue resistance, as it is associated with better mechanical properties under constant load tests. While some studies show conflicting results, the consensus is that higher density wood exhibits superior fatigue performance.
Besides, moisture content significantly influences mechanical properties. Up to the Fibre Saturation Point (FSP, typically 25-30%), an increase in moisture reduces the modulus of elasticity, strength, and fatigue life. For instance, LVL samples at 35 percent moisture content showed significantly shorter fatigue life compared to those at five percent.
Moisture interacts with temperature, and rapid moisture changes can induce internal stresses, leading to damage accumulation. High-frequency fatigue testing can cause localised temperature increases, potentially leading to drying and further altering the material's response.
In addition, temperature directly affects mechanical properties; the modulus of elasticity and strength decrease as temperature rises. Indirectly, temperature influences relative humidity, which in turn affects moisture content.
Wood performs better than steel in fire scenarios because it chars on the surface while retaining a load-bearing core. Research shows that elevated temperatures accelerate damage formation under cyclic loading, reducing fatigue life.
Thermal cycling causes expansion and contraction, potentially leading to microcracks.
For wood's orthotropic nature, it means its fatigue response is highly dependent on the load direction relative to the grain.
In wood materials, different loading conditions result in distinct mechanical behaviours. Longitudinal tension exhibits the highest fatigue strength, where failure is characterised by longitudinal crack growth and fibre pull - out.
Perpendicular tension, on the other hand, shows the lowest static and fatigue strength because the cellular structure is weak in this direction. Longitudinal compression has a lower fatigue life than tension; damage begins with crushed cell walls (kinks), eventually leading to a characteristic 45-degree shear failure.
Flexural load is the most common test for wood, with failure involving tensile cracks on one side and crushing on the other. When stress is normalised to static strength, fatigue life is largely independent of species.
Shear load has a shear strength lower than tensile or compressive strength, yet its normalised fatigue strength is comparable to other loading directions.
In torsion, fatigue behaviour differs between hardwoods and softwoods, with hardwoods showing progressive damage while softwoods fail suddenly.
The size of a structural element influences its load-bearing capacity due to the increased probability of defects in larger volumes (the ‘weakest link’ theory).
While some studies on engineered wood like LVL have not found a significant size effect in fatigue, others warn that results from small specimens cannot be directly transferred to full-scale structures without consideration of this effect. The size effect in wood fatigue remains an area requiring further research.
Influence Of Load Type & Shape
The load ratio R (min stress/max stress) and mean stress are critical parameters. A positive mean stress (tensile) is more detrimental than a negative one (compressive). The relationship between mean stress and stress amplitude (σa) for a given fatigue life is best described for wood by a polynomial Gerber line, while the linear Goodman diagram provides a conservative approximation.
The waveform of the load cycle (sinusoidal, triangular, square) affects fatigue life. Square waveforms are the most destructive, as they combine creep and fatigue into a single damage mechanism. Sinusoidal loading is more critical than triangular, and both are more representative of real-world conditions than square waves.
Frequency has a substantial impact. Lower frequencies (e.g., below 0.5 Hz) allow creep to dominate, leading to shorter fatigue lives. The influence of frequency is more pronounced at higher load levels and higher moisture contents. At very high frequencies (e.g., 1 kHz), the influence becomes negligible, but such frequencies can cause internal heating and drying, artificially improving fatigue life.
Wood & Wood-Based Materials
The natural variability of solid wood makes standardising its fatigue response challenging. Fatigue life appears to be largely independent of wood species when stress is normalised to static strength.
Knots act as stress concentrators, typically reducing fatigue life, though their higher density can sometimes have a compensating effect.
Products like Glued Laminated Timber (GLT), Laminated Veneer Lumber (LVL), and Cross-Laminated Timber (CLT) offer greater homogeneity and dimensional stability.
There is no clear consensus on whether engineered wood has a longer fatigue life than solid wood.
The type of adhesive and manufacturing process play a significant role. Some studies indicate that engineered wood suffers less damage accumulation during fatigue, as evidenced by a smaller reduction in the modulus of elasticity.
Wood-based composites are classified into two groups: homogeneous and inhomogeneous. In homogeneous composites, wood is present as small particles or even nanofibres, together with another component of the composite material. Examples of homogeneous wood composites include wood polymers and wood cements.
Inhomogeneous composite materials combine the positive properties of two or more materials in their primary form of use. Examples include composites made of timber and concrete, wood reinforced with glass fibres, and composites made of wood and aluminium.
A comprehensive review of their fatigue properties would require a separate study. Nevertheless, we mention some examples of composite materials and their response to fatigue.
Yeoh et al. investigated the behaviour of a composite of timber and concrete. After two million load cycles, no obvious loss of stiffness was observed in the rectangular notch-connected floor beams.
Davids et al. examined the response of used beams reinforced with glass fibres. Similarly, after two million cycles, they did not observe a significant reduction in the static strength of the Glulam timber beams.
Yang et al. studied the response of a wood–polymer composite to fatigue. As expected, wood–polymer composites exhibit greater predictability than solid wood.
Few researchers are familiar with wood fatigue, as it remains a highly specialised research area. The problem is that allowable design stresses are based on static strength, even though fatigue accounts for 90 percent of material failures. As a result, structural elements are oversized to compensate for the unknown factors related to fatigue.
To preserve wood as a renewable resource amid increasing demand for green and sustainable products, it is essential to optimise wood-based structures and products. Therefore, wood fatigue is an important research area that should be addressed by more researchers.
Measurement Methods & Models
Current timber design standards, such as EN 1995-2, incorporate fatigue through simplified methods based on constant amplitude loading and reduced allowable stresses.
This approach is conservative and not ideal for structural optimisation. To better predict fatigue life, various experimental methods and predictive models have been developed.
Monitoring wood during fatigue tests involves tracking force, displacement, and other parameters like crack growth and energy dissipation. Key non-destructive methods include:
Acoustic Emission (AE) is a prominent technique for detecting crack formation and growth. AE sensors record high-frequency sound waves released by micro-damage within the material.
Studies show that the rate of acoustic events increases significantly under cyclic loading compared to static loading, providing an early warning of impending failure. It has potential for locating crack tips and estimating remaining fatigue life.
Digital Image Correlation (DIC) is a non-contact, optical method that measures full-field deformations by tracking patterns on a sample's surface between consecutive images. It is highly effective for mapping strain distributions and identifying localised damage.
While real-time processing at high fatigue frequencies remains challenging, DIC has been successfully used to monitor phenomena like humidity-induced swelling during fatigue tests.
Strain Gauges are widely used for measuring local deformations. Their main limitation in wood is the material's heterogeneity; measurements from a small gauge may not represent the global response. Despite this, they remain a reliable and commonly used tool for both static and fatigue testing.
Other methods like Ultrasonic testing (for measuring dynamic modulus of elasticity) and Linear Variable Differential Transformers (LVDTs for displacement) are also used but have specific limitations regarding sample size or real-time monitoring capability.
Predictive models for wood fatigue generally fall into two categories: those based on the number of cycles to failure (S-N curve approach) and those based on the duration of load. However, pure forms of these models are often inadequate for wood, leading to the development of more sophisticated alternatives:
(1)Damage Accumulation Models: These models use a damage variable (α) that evolves from 0 (undamaged) to 1 (failure). While simple, they typically represent only time-dependent damage and do not fully capture the effects of load oscillations.
(2)Energy Models: These models propose that failure occurs when the accumulated energy from loading cycles reaches a critical value. They account for wood's viscoelasticity by considering elastic, viscous, and viscoelastic strain components. However, they can be complex due to challenges in separating the energy that actually causes damage from that which is recoverable.
(3)Number of Cycles Models: This approach uses linear fracture mechanics. For constant amplitude loading, Paris' law relates the crack growth rate per cycle to the stress intensity factor. For variable amplitudes, the Palmgren-Miner rule sums the damage fraction from each stress level. These models are simple but do not account for frequency or creep effects.
(4)Damaged Viscoelastic Material (DVM) Model: This is one of the most comprehensive models for wood fatigue. It assumes an initial crack and defines damage as the energy released during its propagation. The DVM model uniquely combines time-dependent (viscoelastic) behaviour with cyclic loading effects, making it particularly suitable for wood. Its main limitation is the requirement of a pre-existing crack.
Among these, the DVM model is considered the most applicable as it best adapts to the viscoelastic nature of wood under cyclic loads.
Finite Element Method (FEM) simulations are powerful tools for predicting the fatigue life of complex wooden structures. Successful simulation requires selecting appropriate damage criteria and material models.
(1)Damage Criteria: Anisotropic failure criteria like Hashin and Tsai-Wu, originally developed for fibre-reinforced composites, are often applied to wood due to its similar microstructure. The Hashin criterion, for example, can distinguish between different failure modes like fibre fracture and matrix crushing.
(2)Advanced Constitutive Models: Sophisticated 3D material models have been developed that incorporate orthotropic elasticity, plasticity, and isotropic damage. These models can simulate the stress-strain response under variable-amplitude cyclic loading and are calibrated using experimental data from monotonic and cyclic tests.
(3)Cohesive Zone Models (CZM): These are effective for simulating crack growth in quasi-brittle materials like wood. CZMs describe the nonlinear process of damage formation ahead of a crack tip and have been successfully used with a modified Paris' law to simulate fatigue damage.
Commercial software packages like ABAQUS and ANSYS are typically used for these complex simulations, with dedicated modules for composite materials becoming increasingly available for wood-based products. These tools allow for the optimisation of designs to be both lighter and safer against fatigue failure.
Findings
We reviewed the current work, research, findings, and results in the field of wood fatigue. Density and moisture content are the most important influencing factors in static strength testing, and they play an equally important role in fatigue.
Other factors influencing fatigue include fibre orientation, mean stress, R-ratio, load cycle waveform, frequency, and temperature. Many of these influencing factors are interdependent, and their effects are shown in Table 1 below using a colour scale.
Table 1: Wood fatigue influential factors.
The factors are divided into physical properties and loads. Some properties and loadings have two states, for example, low density (LD) and high density (HD), or low mean stress (LMS) and high mean stress (HMS).
Combinations of different states of physical properties and stresses are indicated by abbreviations, such as the combination of low density and low mean stress (LD-LMS), where the influence of the combination is indicated by a colour, as shown in the table legend. The evaluation criterion ranges from very unfavourable to very favourable.
The descriptive states correspond to quantitative values from 0 to 1, from which we have obtained an evaluation of the influence of the individual factors. The quantitative evaluation of the influence is determined from the difference between the average values of the individual states of the physical properties or loads.
For example, the overall rating of the influence of density is the difference between the average value of the rating at low density and at high density. Similarly, the overall influence rating is also determined for mean stress, which has the highest value, confirming the appropriateness of the influence rating method.
Throughout the review, we discussed research by many authors who used various wood species, engineered wood, and composites in fatigue tests to determine fatigue strength, investigate influential factors, and develop prediction models.
We examined the equipment and methods used for collecting experimental data, as well as the models used to predict and simulate fatigue life. Finally, we identified areas where fatigue is the main cause of failure and noted that wood has good resistance to fatigue.
The most important findings on the response of wood to cyclic loads under various influencing factors are shown in Table 1 above. Additional findings that determine approaches for addressing the response of wood to cyclic loads are as follows:
•The α−N curve is monotonically decreasing up to 109 cycles.
•The researchers set the fatigue strength at 107 cycles, for softwood (spruce, fir and Japanese cypress) it is 15–35 percent of the static strength.
•The energy loss per load cycle is initially higher, then decreases in a few cycles and remains approximately constant until shortly before failure, where it increases exponentially in just a few cycles.
•Due to the influence of frequency, the waveform of the load cycle and the time under load must not be neglected.
•The square waveform of cyclic loading is the most destructive.
•The Goodman line describing the relationship between αm and αa, is conservative for wood, the relationship is better described by a polynomial Gerber line.
•At sufficiently low load frequencies (<0.5 Hz), the influence of creep begins to dominate over fatigue.
•Damage from creep occurs earlier than damage from fatigue if the criterion is time under load, while the accumulation of damage from fatigue occurs faster.
•Knowledge about fatigue cannot be transferred directly from tested samples to structural elements due to the size effect.
•The deviation of the load angle from the fibre direction should not exceed 10 degrees for elements subjected to fatigue.
•The process of quality control and the classification of wood into strength classes can damage the wood and reduce its resistance to fatigue.
•The dependence of fatigue on the frequency of cyclic loading is greatest in compression and least in bending.
•At lower frequencies, up to 100 Hz, the influence of frequency on fatigue is greater, while at higher frequencies, such as one kHz, it is negligible.
•It is not yet clear if engineered wood is more resistant to fatigue because it has the positive properties of solid wood and greater homogeneity.
The findings presented in Table 2 are a direct result of the literature review and additionally summarise the effects of frequency, temperature, and load direction in relation to moisture content and heat generation on fatigue strength.
Table 2: Effect of the combination of influential factors on wood fatigue.
Influencing FactorInfluence on Fatigue
Frequencystronger influence with higher humidity and lower frequency.
High temperaturestronger influence on fatigue strength reduction at higher humidity.
Load outside the
anatomical directionstrong influence on the reduction of fatigue strength.
Compressive loadhigher fatigue strength.
High frequency, low loadlow heat generation.
High load, low frequencyhigh heat generation.
Bending stress in
tangential directionbetter fatigue strength.
Future Prospect
Certainly, unexplored areas themselves suggest directions for future work. Although our overview is comprehensive, it does not cover all aspects. A separate study on the fatigue of different types of glulam and wood composites compared to solid wood is needed.
Further investigation into the fatigue of joints would also be valuable. There are many potential applications in the field of measurement methods.
In particular, it would be beneficial to test the integration of sensors. For example, ultrasound could be used to measure the dynamic modulus of elasticity under cyclic loading, while deformation or stress could be measured using the DIC method.
Most importantly, wood fatigue testing methodology should be standardised, including predetermined specimen size ratios, waveforms, frequencies, load ratios, load directions (including specimen fixtures), moisture content, and density classes.
Overall, expanding research on wood fatigue will contribute to optimised structures and products, reducing the impact on the renewable resource, wood.
