Moisture-dependent mechanical properties of longitudinally compressed wood

Wood is a natural composite material, built up mainly of cellulose, hemicelluloses and lignin. These three components form the microstructure of cells (Da Silva and Kyriakides 2007). Cellulose chains have a slender shape, 4–5 μm long, and are arranged in bunches, called microfibrils (Gindl 2001; Tamer and Fauziah 2013). Hemicelluloses and lignin form the matrix of the cell walls, while cells are linked by the highly lignified middle lamella (Dinwoodie 1971; Gindl 2001; Inoue et al. 2008).

When the longitudinal compression of wood is combined with a thermo-hydro treatment, wood becomes more pliable. Prior to compression, the high-density and high quality hardwood (Buchter et al. 1993) has to be plasticized. The plasticization process means the softening of the bonds of hemicelluloses and lignin, so the binding force between cells and between microfibrils decreases. According to Lenth and Kamke (2001), increasing moisture content (MC) decreases the softening temperature of lignin and hemicelluloses. These are the reasons why softening of wood in hot and humid conditions is needed (Ansell 2012). During longitudinal compression, the specimen can be held for a predetermined time constantly at the applied compression level. This period has been called earlier relaxation, but a much more descriptive name is fixation, because the compressed length of the wood is fixated for a predetermined time. In this period, the stresses caused by the longitudinal compression of wood are relaxed and finally—depending on the duration of this period—the longitudinal deformation partially or completely remains after treatment. Fixation increases the effects of the longitudinal compression. Compression stress increases during the compression process and decreases continuously in the fixation phase with gradually decreasing intensity (Fig. 1). Since the rate of compression is constant in Fig. 1 (Báder and Németh 2018a), the shape of the strain–stress curve seems similar to the time-stress curve in the compression phase.

It is most important that the modulus of elasticity (MoE) further decreases along with increasing fixation time. The decrease in the MoE as a result of longitudinal compression is 59%. Its decrease due to compression followed by fixation for 1 min is 63%, and if fixation is used for more than 10 h after compression, the MoE decreases by 81%. Each value is with a uniform MC of 12% (Báder and Németh 2018a). In a previous study by Báder and Németh (2018a), a relationship was identified between compressive stress change during fixation, remaining shortening and some mechanical properties. Increasing fixation time increases the remaining shortening of the specimen, so its spring-back will be much weaker (Báder and Németh 2018a). The spring-back correlates well with the change of the compression stress during fixation, the MoE and the bendability. Fixation for a long time results in a more effective treatment, but it consumes much time. Thus, it may be used only if extreme wood properties are needed after this modification process. Considering the effective property changes, the productivity and the costs, the ideal fixation time was specified as 1 min by Báder and Németh (2018a), but it actually depends on the needs regarding the product properties.

The method of longitudinal compression and fixation was called “pleating” due to the buckling of the cell walls by the treatment. The physical and mechanical properties of pleated wood are most affected by the compression ratio and the time of fixation. As evidenced by Báder and Németh (2017a), the self-developed laboratory compressing machine produces specimens with equal compression ratio along the length. Unequal compression ratio along the length befalls very rarely and if such a mistake occurs, it is clearly visible (Báder and Németh 2018a), so the investigation of the equal compression ratio along the length was not necessary in this study. After treatment, the specimen is first wet, and it should be dried after adjusting its shape and before its final use. Different minimal moisture contents can be found in the literature as a limit of pliability, ranging from 15% (Vorreiter 1949; Ivánovics 2005) to 25% (Buchter et al. 1993; Szabó 2002; Kamke 2014). Pleated wood with over 15% MC has nearly plastic properties. Between 0 and 5% MC, the treated wood is more brittle than untreated wood, so it is no longer pliable (Vorreiter 1949). Unfortunately, after shaping and drying, pleated wood only keeps its shape until it stays dry. If it is re-moistened, its original shape will be partially recovered, similar to what Inoue et al. (2008) and Kúdela et al. (2018) described as it has a high shrinkage-swelling in the longitudinal direction. The easiest way to decrease the swelling of wood is the reduction of its moisture uptake. This might be done using different treatments, for example heat treatment in different atmospheres or impregnation (Inoue et al. 2008; Obataya and Higashihara 2017; Báder and Németh 2018c).

The described modification method results in a semi-finished wood product, which can be used for many purposes, where pliable wood is needed. Analysing the bending tests of pleated wood showed that the bending force and the modulus of elasticity decrease exceedingly, and ensure a high deformability of the wood, compared to untreated specimens (Báder and Németh 2018a). The technology was first patented by Hanemann (1917), but industrial production began only in the second half of the 1920s (Báder and Németh 2018b). The most modern equipment used today has been patented in 1989 (Thomassen et al. 1991). Longitudinally compressed wood is currently produced in Italy, Hungary and in the USA (Báder and Németh 2018b). This is an excellent raw material for the furniture industry, interior design, or areas where sharp corners are not allowed for safety reasons (ship and aircraft furniture) (Báder 2015).

The water content of wood can be divided into two physically different parts. Bound water is found in the cell walls and bonded to the hydroxyl groups of the main components of wood (Berthold et al. 1996; Thygesen et al. 2010; Ansell 2011). The change in bonded MC is called sorption. Generally, the decrease in MC is called desorption, while water uptake is resorption or adsorption (Siau 1984). Equilibrium moisture content (EMC) can be achieved at a constant temperature and relative humidity (RH) of the air over time. Fiber saturation point (FSP) is defined as MC above which the physical and mechanical wood properties do not change because of changes in its MC (Stamm 1964; Glass and Zelinka 2010).

Plasticization reduces EMC of wood at the beginning of the desorption by 3.7% (MC%). This difference later decreases, while at EMC below 10%, all moisture contents are the same (Fig. 2). Pleating results in a deviation between 0% and 1% (MC%) of adsorption compared to control specimens. FSP is the same for treated and control wood after adsorption. Thus, the sorption isotherm of pleated wood in comparison with untreated wood differs only slightly (Báder and Németh 2017b).

It has been well-known for a long time that MC affects not just the dimensions but also the physical and mechanical properties of wood (Huang et al. 2006; Hernández et al. 2014). Bauschinger published this fact in Munich in 1883, and many researchers dealt with this topic later, because of its practical and theoretical importance (Tiemann 1907; Vorreiter 1949; Kollmann 1951; Niemz and Sonderegger 2017). The change ratio is called coefficient of moisture dependence of the mechanical property, and is marked with α (Molnár 2004), B (Skaar 1988), E (Vorreiter 1949; Kollmann 1951) or K (Jiang et al. 2012). As a coefficient, α never has a unit of measurement. The moisture dependence of some mechanical properties can be seen in Table 1, and the moisture dependence of other mechanical properties in Online Resource 1.

The results of the various research studies in Table 1 do not match perfectly, but the trend is similar. The reasons for the differences are wood species, the number and selection of specimens, accuracy and sensitivity in MC and temperature or in data analysis. The mechanical properties of wood increase in some cases up to about 5–10% MC (Fig. 3a) (Kollmann 1951; Ansell 2011), compared to the oven-dry state (0% MC).

Further increasing the moisture content above 10% until the FSP, only decreases the mechanical properties (Tiemann 1907; Skaar 1988). Above FSP, the mechanical properties have a constant value (Fig. 3a). As described by Tiemann (1906), by increasing MC, the cell walls expand and as a result, fibrils move farther from each other. The structure loosens, and, as a consequence, the mechanical properties deteriorate. As described by Gerhards (1980), the conclusions drawn by different researchers differ in whether there is a linear (by Drow, Greenhill, James, Kitahara and Suematsu, Okohira, Sellevold, Tang and Hsu) or a more complicated (by Ishida, Leont’ev, Wilson) relationship between the MC and the mechanical properties, as shown in Fig. 3b.

Nowadays, prior to laboratory mechanical testing, the specimens are generally conditioned at a temperature of 20 °C and 65% RH, resulting in about 12% moisture content of wood (Gerhards 1980; Skaar 1988; EN 408:2010+A1 2012). In order to determine the sensitivity of the mechanical properties to MC change, the mechanical properties at 12% MC are selected as reference values (σ₁₂). Then, a well-correlated linear section should be added to the curve obtained from the measurement results, and the α coefficient can be computed by the slope of this linear section (Eq. 1):where \(\Delta \sigma /\Delta MC\) is the slope of the linear equation (Jiang et al. 2012). Based on the values of α, the mechanical properties of wood specimens can be calculated as described in Eq. 2, according to Kollmann and to different standards (Kollmann 1951; MSZ 6786 2004a, b, c; ISO 13061 2014a, c, etc.):where \(\sigma_{MC}\) is a particular mechanical property and MC is the moisture content of the specimen at the time of measurement. α coefficient for untreated wood is determined as 0.04 for modulus of rupture, 0.01 for modulus of elasticity, and 0.04 for compressive strength (MSZ 6786-15 2004a, b, c). Using Eq. 2, data comparable to the evaluation of research results are obtained. The equation can be interpreted within a range of 12 ± 3% by Hungarian standard MSZ 6786-15 (2004a), but researchers usually give a much wider range (Vorreiter 1949; Kollmann 1951; Lohmann et al. 1987; Niemz 1993; Molnár 2004).

The average fiber saturation point (FSP) for beech control specimens (BC) was 25.6 ± 0.6%. The FSP for control specimens of steamed beech (BSC) was 25.9 ± 0.4%, while the FSP for treated beech specimens (B1mA) was 25.4 ± 0.8%. The large difference of FSP from the published data (35.6%, Molnár 2004) could have been caused by drying the specimens to 0% moisture content (MC) at 103 °C temperature before saturating with water vapor, and the relative humidity (RH) slightly below 100% during the tests (Báder and Németh 2017b). The FSP of control and treated specimens is the same, about 25.6%. This result can also be proven by Fig. 4a, because there is a negligible change of mechanical properties when their MC is over 25% compared to their change below 25% MC. This confirms the correctness of the FSP test.

According to Báder and Németh (2018a), the mechanical properties of wood due to the applied steaming before pleating do not change compared to control specimens. In this research, there was no statistical difference between BC and BSC specimen groups, which is also evidenced by Fig. 4b, showing some typical stress–strain curves of untreated and steamed specimens at 12% MC, and typical stress–strain curves of pleated specimens at each examined MC level. The slopes of pleated wood at the beginning of each graph decrease with increasing MC, representing the change of the MoE. The initially peaky graphs gradually become rounded, thus the ductility of the treated specimens increases with increasing MC (Hayden et al. 1965). The properties of pleated wood shown in Fig. 4a, b were analysed between 0% MC and FSP, because above this value the change of mechanical properties can be considered constant (Stamm 1964; Glass and Zelinka 2010).

Various data can be found in the literature for the bending modulus of rupture (MoR). Wagenführ (2007) found 78–110–117 MPa, Kretschmann (2010) measured 105 MPa, Meier (2016) established 110 MPa, and Niemz and Sonderegger (2017) found 95 MPa for beech wood. In this research, the MoR for the control specimens averaged 111.7 MPa. The differences can be explained by the different dimensions and the different length and height ratios (L·h⁻¹) of the specimens, the different densities of wood specimens and the differences between 3-point and 4-point bending tests. Testing the MoR at 12% MC, it decreased by 19.2% due to longitudinal compression and fixation for a short time, compared to the control specimens (Fig. 4a). This is in agreement with the results of Báder and Németh (2018b) where the decrease in MoR was about 24%. As seen in the literature analysis, the MoR does not decrease at low moisture contents, until about 5% (Table 1 and Fig. 3a). The current tests produced about the same values for MoR at 2.4% and 9.2% MC, so it can be assumed that at around 5% MC, a maximum value can be found, in accordance with the published data (Kollmann 1951; Skaar 1988). Taking into account this finding, the change of MOR with the MC can be outlined (Fig. 5a), and the quadratic equation of the curve between 9% and 24% MC can be seen on the graph. A linear section was added to the curve obtained from the measurement results similarly to Jiang et al. (2012), taking into consideration that it functions only from 7% to 16% MC. The MC at room conditions rarely exceeds these values so this range is wide enough to correct the research results of MoR in daily wood testing, to get the standardized MoR at 12% MC. From the linear equation in Fig. 5a, it can be inferred that each 1% MC change results in an increase or decrease of 3.87 MPa for the MoR. Calculated in accordance with Eq. 1, the degree of change of the MoR for each percent MC change is α = 4.2.

Comparing Figs. 3a and 5a, the shapes of the MoR graphs of untreated and pleated wood are very similar as well as their α coefficients (Table 1).

The other α coefficients were calculated in the same way. The value of the bending stress at 4 mm crosshead displacement for control specimens was 95.1 MPa at 12% MC, and it is approximately halved by pleating, as seen in Fig. 4a. The graph in Fig. 5b has a slight curve over the whole range between the absolute dry state and FSP. The used linear approximation results in a high α value of 5.4, based on Eq. 1. Thus, the force needed for bending wood decreases very much due to the increased MC of pleated wood (Figs. 4b, 5b).

The results by Wagenführ (2007) are 9.2–13.0–13.5 GPa, Kretschmann (2010) measured 12.3 GPa, Meier (2016) calculated 14.3 GPa, and Niemz and Sonderegger (2017) found 13.0 GPa for the bending modulus of elasticity (MoE) of beech wood. In this research, the MoE for the control specimens averaged 9.89 GPa. The differences can be explained by the setup deviations between the 3 and the 4-point bending tests, besides the natural variation of the wood even within one log. MoE was 5.34 GPa after the longitudinal compression and fixation for a short time which is a 46.0% decrease compared to the control values, both at 12% MC (Fig. 4a). Based on the results, a curve with a similar shape to the MoR graph can be drawn (Fig. 5c). A simpler variant can be obtained through the linear approximation, which shows an α value of 5.0 in the range between 6% and 17% MC. Compared to the published value of 1–2% (Table 1), this result shows significant changes. MoE and pliability are closely related (Ashby and Jones 2003). Using bendability coefficient (k_bend) is a good possibility for pliability investigation. Comparing at 12% MC, k_bend increases by 136% as a result of pleating. The graph in Fig. 5d has a slight curve. The α coefficient of the approximating linear line is 4.2 and it is valid between 2% and 22% MC. Increasing the MC, the k_bend of pleated wood increased fourfold, while k_bend of the treated specimens at 12% MC were already 2.4 times higher compared to control specimens at 12% MC. This means a great increase in pliability. As earlier described, increasing MC decreases the softening temperature of lignin and hemicelluloses, thus these constituents are more likely in their plasticized state when the MC is high. For example, Olsson and Salmén (2003) and Engelund et al. (2013) affirmed that the softening of hemicelluloses occurs at room temperature around 70–75% RH. These facts explain the much better pliability of wood besides high MC not only if it is pleated, but even when it is untreated.

In the 4-point bending tests at 12% MC, the deflection at maximum load and the highest deflection before breaking of control specimens averaged 7.3 and 7.6 mm, respectively. These deflections increased by 143% and 149% for pleated specimens, respectively. Figure 5e shows the deflection of the treated specimens and also shows that the deflection at the breaking point diverges from the deflection at the highest bending force with increasing MC (Y axis on the right side of Fig. 5f), so it is even more difficult to break the specimens. The specimens approach the plastic state with increasing MC, as Vorreiter (1949) described. The α coefficient of the deflection ability of pleated wood is 7.6 and valid from 5% MC up to FSP. Above the FSP, the MoE slightly reduces (Fig. 4a), but surprisingly, the deflection of specimens drops at the FSP, and above the FSP it becomes constant (Fig. 5f). Free water above the FSP reduces the available free void volume and causes hydraulic pressure during bending, which generates stresses in the material (Stevens and Turner 1948; Peck 1957). A good explanation can be that the appearance of hydraulic forces induced by free-water in the mostly stressed area results in rupture.

The decrease in the MoE between 12% MC and the FSP was 25% (MoE%), similar to that of fixation for a long time compared to the specimens with fixation for 1 min, which was 19% (MoE%), both at 12% MC (Báder and Németh 2018a). These results predict that the compressed material and that fixated for a short time in its wet state and the fixated for a long-time material at 12% MC have similar bending ability. The increase in the highest possible deflection between 12% MC and the FSP was 120%, but in the case of the fixation for a long time compared to the specimens with fixation for 1 min it was 320% (Báder and Németh 2018a). The change in pliability is not as extensive as indicated by the decrease in MoE, but 120% is still a very high increase. The fracture during bending usually occurs on the side under tension because wood allows much less tensional deformation compared to the compressed deformation (Gaff et al. 2015). Thus, the neutral axis has to be moved as close to the tension side as possible to avoid its breaking. The appearance of the high-tension stresses during bending can be delayed or even removed by generating buckled cell walls. The cell walls of wood fixated for a long time are much more buckled than the cell walls that were compressed without fixation or only fixated for a short time (Báder and Németh 2018a). The reason for the smaller pliability is because the specimens fixated for a short time cannot stretch as much on the tension side during bending. A greater compression ratio with fixation for a longer time solves this problem but this is a very time-consuming procedure. This solution can be substituted to a certain extent by using the longitudinally compressed wood in a wetter state. For best pliability, the moisture content of the pleated beech wood must be close to its FSP, between 15 and 25% MC.

The published data for compressive strength parallel to the grain (σ_c) of beech wood are 48–65–70 MPa (Wagenführ 2007), 51.3 MPa (Kretschmann 2010), 57.0 MPa (Meier 2016), and 52 MPa (Niemz and Sonderegger 2017). The current research results for the control specimens averaged 53.1 MPa, consistent with published values. The comparison of the specimens with 12 × 12 × 18 mm³ and 20 × 20 × 30 mm³ (R × T × L) resulted in no statistical difference in σ_c. This is also true for control and treated specimen groups. The average σ_c result for treated specimens at 12% MC is 49.9 MPa, which can be considered the same as the result for the control specimens (Fig. 4a). In the range between 0% and 17% MC, the α coefficient is 3.2 (Fig. 5e), which is about the half of the data given in the literature for untreated wood (Table 1). Moreover, the proportional limit obtained during the compression test for pleated wood is about one-third lower compared to the proportional limit of untreated specimens, and in many cases the elastic range cannot be determined, neither the compressive stress at proportional limit. This is a consequence of the longitudinal compression, because the wood has been compressed beyond its elasticity limit during the pleating process. However, this type of longitudinal compression cannot be considered as a compressive strength test, because it was made under special circumstances when the material was in plastified state.

In this study, comparative bending- and compression tests were carried out with longitudinally compressed beech wood specimens with different moisture contents. The mechanical properties of wood change with the moisture content and the ratio of change differs from the untreated specimens.

This study indicates that compressive strength parallel to the grain exhibits the smallest sensitivity to changes in moisture content by an α value of 3.2%, followed by 4.2% for its modulus of rupture, 5.0% for its modulus of elasticity and 5.4% for stress at a 4 mm crosshead displacement. With increasing moisture content, the α value of both the bendability coefficient and the highest deflection during a 4-point bending test causes an increase of 4.2% and 7.6%, respectively. As a result of the research, it can be deduced that there is a negligible difference between the change of the modulus of rupture of the untreated and the treated wood due to the change in their moisture content. However, the changes of modulus of elasticity, bendability coefficient and compressive strength show a considerable alteration. For the best pliability during bending, the moisture content of pleated wood must be close to its fiber saturation point.