The Effects of Heat-Treatment on Tensile Elastic Modulus and Creep in Hickory Wood: Applications to Bow-Making Nicholas Wenner 23 May 2014 ME348 Independent Research Project 2 Table of Contents: Motivation and Objectives ………………………………. 3 Background ………………………………. 4 Equipment and Experimental Procedure ………………………………. 5 Results ………………………………. 9 Discussion ………………………………. 11 Conclusions ………………………………. 14 Appendix I: Stress Calculations …………………………………. 16 Appendix II: St. Louis’ Experiments ………………………………. 17 Appendix III: Selections from Esteves and Pereira (2009) ………………………………. 19 3 Motivation and Objectives: This experiment was conceived and carried out with an eye towards designing an archery bow from natural materials. Over several meetings with a local bowyer, an interest in the effects of heat-treatment on bow wood developed based on anecdotal evidence suggesting that certain forms of heat-treatment can improve the elastic qualities of wooden bows, giving both increased draw strength in a given bow and decreased levels of string follow (the tendency of a bow to deform and lose strength over time and use). In The Traditional Bowyer’s Bible, Vol. 4 (2008), bowyer Marc St. Louis discusses his informal experiments with the heat-treatment of bows, finding potential for both increasing the draw weight of a given bow and decreasing its string follow (see Appendix II for details). St. Louis discusses the following benefits of heat-treatment: “A bowyer can make heat-treated bows of a much narrower profile than an untreated bow.” “Dry heat can also be used to temper static recurves on a bow after they have been steam bent, and this can reduce the mass of the limb tips.” “Heat-treating and reducing the mass of the tips is advantageous for those wishing to use lighter arrows than normal, as light tips reduce handshock. This can also be beneficial to those wanting to make bows to compete in flight archery.” “In effect, bows made from the better second-string wood species such as HBB or elm can exceed the performance of the higher density wood species.” “The process can be used to make very high performance bows, improve the performance of simple flatbows, or even to revitalize older bows. The tempering process reduces mass, thereby increasing performance. The reshaping of the wood and adding reflex increases string tension, which also improves performance.” It also “seems to increase a bow’s resistance to moisture incursion, which is always a concern with whitewood bows.” St. Louis also mentions that heat-treatment can be taken too far, at which point any benefits are overcome by an increase in the brittleness of the wood. St. Louis gives some guidance to the appropriate level of heat-treatment: “A good rule of thumb when tempering: brown is good, black is bad.” The general aim of this experiment is to build on the evidence St. Louis provides by more rigorously quantifying the effects of heat-treatment on wood and providing more specific guidelines for the proper time and temperature needed. Wood is a highly diverse and complicated material. The effects of a given heat-treatment will depend strongly on the type of wood, temperature, time, humidity, and many other factors. Due to limitations of resources and the relatively small scope of the assigned project, this experiment aims only to provide a small-sample exploration of the effects of one particular heat-treatment on one particular wood species. The heat-treatment was chosen based on reports given by St. Louis 4 (2008) and informed by a review of the scientific literature on wood heat-treatment provided by Esteves and Pereira (2009). Hickory wood (Carya illinoensis) was chosen based on availability and on reports from both a local bowyer and from St. Louis on its “excellent” response to heattreatment. Background: Esteves and Pereira (2009) offer an excellent review of the effects of heat-treatment on wood, showing that wood in heat-treatment undergoes complex chemical and anatomical changes with effects on factors including mass, equilibrium moisture content, dimensional stability, rot resistance, mechanical properties, color, wettability, thermal conductivity, and factors involved in the finishing and gluing processes. Of the effects Esteves and Pereira discuss, the most applicable to bow making are: Mass loss Decreased equilibrium moisture content, resulting (among other things) in increased dimensional stability Mechanical properties o Increased brittleness o Time and temperature-dependent effects on modulus of elasticity “The modulus of elasticity seems to increase for softer treatments and decrease for more severe treatments.” See Appendix III for more information on each of these effects. 5 Equipment and Experimental Procedure: Equipment: Two hickory wood specimens of clear grain and approx. dimensions of 1.5” x .5” x 6.25” Table saw, band saw, disc sander, and wood planer for sample preparation Thermolyne small benchtop muffle furnace FB1415M (120V) Dial calipers (resolution .001”) Digital scale (resolution .1 gram) Two single element, pre-wired strain gages and installation materials P 3500 strain indicator Gear motor driven, deflection controlled loading system Four-point bending loading fixture Voltmeter for monitoring applied load Sample Preparation: Because only two strain gauges were available for the purposes of this experiment, special effort was taken to maximize homogeneity between the experimental and control samples. Both samples were cut from the same piece of hickory wood in adjacent growth rings. The samples were cut to dimensions as equal as possible. Dimensions were chosen to approximate those of a bow limb and to accommodate the size of the loading fixture. Original Dimensions: Experimental (1.455” x .456” x 6.274”), Control (1.460” x .463” x 6.277”) Care was taken to maintain as similar grain orientation and growth ring size as possible: 6 The samples were prepared with the grain oriented in the direction that would be found in a bow: Mass and dimensional data was taken in both specimens immediately prior to heat-treatment and immediately prior to loading. Moisture content was not measured due to lack of appropriate equipment. Heat-treatment: The experimental specimen was heated in a benchtop muffle furnace at 200 degrees C for 91 minutes. This relatively mild treatment was chosen based on a conclusion by Esteves and Pereira (2009) that the modulus of elasticity “seems to increase for softer treatments and decrease for more severe treatments.” (See Appendix III for more details.) 7 The specimens were left side-by-side for about 1 week to allow the heat-treated specimen to come back to equilibrium moisture content. Loading and Data Collection: Strain gauges were attached with cyanoacrylate adhesive to the underside of each specimen as close to the centers as possible and oriented as closely as possible with the longitudinal axis. Gauges were attached to the sides of the specimens that would experience tension under bending. The samples were placed in the four-point bending fixture in the loading system. 8 Data was taken in two rounds, the first for determination of elastic modulus and the second for testing the effects of creep. In the first round, each sample was loaded in increments of about 30 pounds, pausing at each increment to record loading and strain values with a photograph. The intent was to load each specimen to 1% strain, an approximation of the strain experienced by a bow at full draw. The loading machine was unable to apply sufficient load, however, reaching a maximum in the control of about .47% strain and about .37% in the experimental specimen. Each specimen was unloaded immediately and residual strain measured. In the second round, each specimen was loaded continuously and at the same rate up to a maximum strain of .35% and left there for 60 minutes, at which point the specimen was unloaded and residual strain measured. A specimen under approximately .35% strain. 9 Results: Modulus of Elasticity: Modulus of elasticity increased by 26% with heat-treatment. The stress-strain curve demonstrated highly linear behavior: Creep: No significant creep was observed in either specimen. Both specimens behaved elastically both when unloaded immediately after reaching a maximum load (Phase 1) and after being held for 1 hour at 3500 microstrain (Phase 2). Creep (με) 10 Mass Loss: Due to equipment malfunction, mass data taken prior to heat-treatment was erroneous. Comparing the densities of the control and the heat-treated specimens, however, provided an indirect measure of mass loss, yielding: Percent mass loss: 1.7% Dimensional Shrinkage: Dimensions of both specimens were taken immediately prior to treatment of the experimental specimen and immediately prior to collecting stress-strain data on the loading machine. To separate dimensional changes due to heat-treatment from those due to changing environment, shrinkages observed in the control were subtracted from those observed in the experimental specimen, yielding a “corrected” percent shrinkage. Corrected percent shrinkage due to heat-treatment 11 Discussion: Modulus of Elasticity In “The Wood Handbook” (2010), Kretschmann gives the following values: Assuming 12% moisture content in the wood used in this experiment, the elastic modulus of the control specimen was found to be 58% greater that reported by Kretschmann. While this could be due to natural variation between wood specimens, Kretschmann also reports a coefficient of variation for elastic modulus of only 22%, leaving a 36% increase unexplained. According to The Wood Database (http://www.wood-database.com), the strength characteristics of Carya illinoensis are influenced by the spacing of the growth rings, with wood from fastergrowing trees (wider spacing) tending to be harder, heavier, and stronger than that from slowergrowing trees (closer spacing). The relatively high elastic moduli of the specimens in this experiment could be explained, then, by the relatively high density of the specimens, with the specific gravity of the control being .88 (33% greater than the value reported by Kretschmann). While the magnitude of the elastic moduli reported is perhaps unexpected, there is regardless a clear increase of elastic modulus (26%) between the control and the experimental specimens, verifying and quantifying the effects reported by St. Louis (2008). The surface stresses reached in the experiment (12721 psi in the control specimen and 12795 psi in the experimental) are approaching the modulus of rupture reported by Kretschmann (13700 psi). Creep: That no significant creep was observed in either specimen could be due to relatively low strains and relatively short loading times. String follow in bows results from repeated shooting from full draw (at approximately 1% strain). Strains in this experiment were limited to .35% due to 12 limitations of the available equipment. Creep in bows may also be a result of dynamic bending effects, which were not explored in this experiment. Mass Loss: The observed mass loss is on the order of that expected by data given in Esteves and Pereira (2009). (See Appendix III, Figure 1.) Studying the effects of the observed mass loss (1.7%) on bow performance would be an interesting avenue for future study. Dimensional Shrinkage: The Wood Database (http://www.wood-database.com) gives the following values for dimensional shrinkage with moisture content (green to oven-dry): Radial: 4.9%, Tangential: 8.9%, Volumetric: 13.6%, T/R Ratio: 1.8. The longitudinal shrinkage is negligible. That only negligible amounts of radial shrinkage occurred in the heat-treated specimen and that the T/R ratio was not maintained at approximately 1.8 suggest that a mechanism alternative to moisture loss contributed to the relatively large tangential shrinkage observed in the specimen (~1.9%). The corrected percent shrinkage values reported in this experiment are likely to be underestimates of the real values given that heat-treated wood has been shown to have higher dimensional stability than untreated wood (Esteves and Pereira, 2009). That significant tangential shrinkage occurred with heat treatment fits with St. Louis’ observations that heat-treating the belly of a bow causes it to become slightly concave as well as smaller in width. Sources of Error General While great effort was taken to ensure homogeneity between the two samples, inevitable differences between them must account for some of the reported changes. The relation provided by the instructor to convert measured voltage to applied load was of unknown accuracy. Strain gauge application was not perfectly aligned with the longitudinal axis of the material. The contribution to error due to this alignment, however, is small for small angles: If the angle were off by 1 degree (and it was surely less than that), the error incurred would be only about 0.02%. The strain gauge was not perfectly placed in the center of the beam. This is mitigated by the fact that the internal moment is expected to be constant throughout the middle section of a beam in four-point bending. 13 Temperature compensation for changes in resistivity and for differences in the coefficients of thermal expansion between the gage and the substrate were not taken into account. Temperatures could have varied on the order of 1-4 degrees C. Transverse sensitivity of the strain gage was not taken into account. It was not possible for the top and bottom loading fixtures to be perfectly aligned, resulting in four-point bending that was asymmetrical to some extent. Modulus of Elasticity Deformation of the sample at the loading points clearly occurred, potentially affecting measured strains. In calculating stress, elastic strain under pure bending with uniaxial bending stress was assumed. Creep Loading profile over time between the two samples could have varied stochastically between the two samples as a function of deformation at the loading points. Mass Loss Equipment malfunction rendered pre-heat-treatment mass values unusable and mass loss had to be calculated between the experimental and the control specimens via density. Dimensional Stability While the average of several measurements were taken for each dimension and each dimension was measured in roughly the same location pre-heat-treatment and preloading, inevitable variation in the specimen’s shapes could have accounted for some of measured dimensional changes. 14 Conclusions: The general aims of this experiment were to build on the evidence provided by St. Louis (2008) about the effects of heat-treatment on bow wood by 1) more rigorously quantifying those effects and 2) providing more specific guidelines for the proper time and temperature needed. For a particular heat-treatment in a particular wood species, the first goal was achieved: For a heat-treatment of 200 C for 1.5 hours, an increase in elastic modulus of 26%, no significant change in creep, mass loss of 1.7%, and significant tangential shrinkage (1.9%) were observed. The second goal was approached, showing that the particular heat-treatment explored had positive effects on the elastic modulus of hickory wood, but not exploring the effects on modulus of rupture and the effects of different heat-treatments. While necessarily lacking in scope, this study served well as a next step in the attempt to quantify the effects observed by craftsmen in their trade and – perhaps – to provide those craftsmen with useful theoretical frameworks and predictions. Special thanks to bowyers Richard Baugh and Jim Langell for guidance, inspiration, and sharing of resources, time, and materials. Recommendations for future research: Further inquiry o Test more wood types at a greater range of heat-treat temperatures and times. Identify the times at a given temperature at which elastic modulus begins to decrease. o Test creep over a greater range of maximum strains and loading times. o Test in dynamic loading environments to more accurately reflect the loading and unloading seen by a typical bow. o Provide frameworks for estimating temperature based on metrics available to the bowyer (e.g. temperature of thermometer held at surface, color of wood reached). o Take data on how modulus of rupture is affected by heat-treatment. (Requires thinner samples or loading machine with higher range to reach sufficient strains.) o Study the effects of the observed mass loss (1.7%) on bow performance. o Study the effects of the observed increase in modulus of elasticity (26%) on bow performance. Improvements on methodology o Multiple samples and multiple tests with each sample. 15 o Compare samples from the same growth rings but different (but adjacent) longitudinal locations (as opposed to the same longitudinal locations but different (but adjacent) growth rings) to maximize homogeneity between the experimental and the control specimens. o Take mass data before as well as after heat-treatment using a scale with precision of at least .1 gram. o Take data on moisture content. o Use thinner samples (or a loading machine with a higher range) to achieve strains typical of a bow at full draw (on the order of 1% strain). 16 Appendix I: Stress Calculations The modulus of elasticity, E was calculated from the slope of the stress-strain curve, assuming elastic strain under pure bending with uniaxial bending stress. σ = 3My / (2bh3) Where the applied moment, M is given as: M = F/2*l Where l is the distance between the inner and outer loading points. In this experiment, the lower loading fixture spanned 5.5 inches and the upper loading fixture spanned 3.5 inches, such that l was 1 inch. 17 Appendix II: St. Louis’ Experiments St. Louis (2008) explores the use of heat-treatment to introduce recurve into bows by drawing them backward and applying heat to the tensioned belly. Recurve bows have tips that curve away from the archer when the bow is strung, storing more energy than a simple straight limbed bow and permitting a shorter bow for a given performance. Using a heat gun, St. Louis heats the wood until it turns “brown.” Measuring the heat at the surface of the wood with a thermometer in one occasion, he measured about 400 degrees F (204 C). Elm static recurve bow About fifteen minutes at four inches from a hot plate with the bow reflexed. Belly turned a “dark brown.” Gave off a pleasant aroma. Back was hot enough that he could only put his hand on it for a second or two and water brushed on the back started to steam immediately. Belly shrank from the heat, leaving the back wider than the belly. In the end the bow was reflexed several inches and the color turned from pale white to dark brown. Left for several days to come back to equilibrium moisture content. Noticed the belly had become slightly concave as well as shrunk in width. From the side, slight discoloration could be seen up to about halfway through the limb. Shavings from the belly seemed “crisper.” Seemed to be a definite increase in draw weight. Drawing well past brace height had “hardly any difference on reflex.” During tillering, “the limbs just did not want to take any appreciable amount of set.” Tillered to 50# at 28” draw, where it kept “at least 50 percent of the reflex” he had heated into it. It was “very fast”: nearly 180 fps with a 10 grain per pound arrow. It was “stable and shot like a bow at least 10# heavier in draw-weight.” It “gave nothing away in performance” compared to the high-performance sinew-backed reflexed recurves he had made in the past and it was much easier to make and much more stable with humidity. Years later, after having shot “many arrows” with the bow, it still holds a “substantial reflexed profile” when unbraced. Eastern Hop Hornbeam static recurve bow Same procedure as previous. Again noticed shrinking of the belly. Image shows about 1/8” belly over a limb about 2” wide. Heat-treated again until the bow started to turn black. 18 Tillered same as before. Also retained “more than 50%” of the several inches of reflex he introduced. Performance was “spectacular”, but the bow began to fail not long after finishing it. Large fractures formed on the belly (compression) side. Eastern Hop Hornbeam pyramid flat bow “Though being relatively short for the draw length, it still retained nearly 2” of reflex.” Eastern Hop Hornbeam flat bow with shallow static recurves and a bit a set with about 1” string follow The limbs returned to nearly their original straight profile and again became somewhat concave. The draw weight had “increased substantially.” The limbs did not take “any set at all” after being pulled back “a few times.” At 28” draw, the draw weight increased from the original 52 lbs. to 60 lbs. After tillering back to the original draw weight, the bow shot “an additional 10 yards, or an additional 6-8 fps” for a 500 grain arrow. The bow was averaging 156 fps with highs of 160 fps. White ash native style D bow Held by hand about 12” above hardwood coals for about 30 minutes per limb. It “lost a bit of reflex during the process” but still had about 2” before tillering was started. After repeated pulls to full draw, the bow held 1” of the initial reflex. 19 Appendix III: Selections from Esteves and Pereira (2009) The following selections are taken from: Esteves and Pereira (2009). “Heat treatment of wood,” BioResources 4(1), 370-404. Mass Loss: “Mass loss of wood is one of the most important features in heat treatment and is commonly referred to as an indication of quality. Several authors studied mass loss with heat treatment and concluded that it depends on wood species, heating medium, temperature, and treatment time (Fig. 1).” Equilibrium Moisture Content: “The main effect of the heat treatment is the decrease in equilibrium moisture content. The reduction was already reported in 1920 by Tiemann, who showed that the drying at high temperatures decreased the equilibrium moisture of wood and consequently its swelling and shrinking. This is the basis for all heat treatment processes. As in mass loss, improvement of equilibrium moisture content depends on wood species, temperature, time, and type of treatment.” “The minimum temperature necessary to perform a heat treatment is 100oC according to 20 some authors. Kollmann and Shneider (1963) carried out tests with beech wood, oak, and pine at temperatures 70oC- 200oC and 6 -24 hours, and concluded that the absorption of water decreased at temperatures higher than 100oC, decreasing with the increase in treatment time. The same was confirmed by Nikolov and Encev (1967) and D ́Jakonov and Konepleva (1967). However, other authors disagree and believe it depends on the wood species. For example, Kollmann and Fengel (1965) reported that wood degradation begins at 100oC for pine but only at 130-150oC for oak.” “The reasons for the decrease of the equilibrium moisture are as follows: There is less water absorbed by the cell walls as a result of chemical change with a decrease of hydroxyl groups; there is enhanced inaccessibility of cellulose hydroxyl groups to water molecules due to the increase of cellulose crystallinity; and cross-linking occurs in lignin.” “The lower equilibrium moisture content might affect positively the strength properties of heat-treated wood, but this effect is superseded by the degradation of the chemical compounds.” Mechanical Properties “The downside of the treatment is the degradation of mechanical properties. The effect on MOE is small, whereas static and dynamic bending strength and tensile strength decrease. Brittleness of wood increases with the deterioration of fracture properties due to the loss of amorphous polysaccharides. The degradation of hemicelluloses has been identified as the major factor for the loss of mechanical strength, but also the crystallization of amorphous cellulose might play an important role. Polycondensation reactions of lignin, resulting in cross-linking, have been mentioned as having a positive impact mainly on longitudinal direction.” Modulus of Elasticity “The modulus of elasticity seems to increase for softer treatments and decrease for more severe treatments. Results reported by Esteves et al. (2007b) with steam heat-treated pine wood (Pinus pinaster) showed a small increase until about 4% mass loss, followed by a decrease for higher mass losses. With the same treatment conditions, heating time, and temperature, the reduction of MOE was higher for the treatment in air, and relationship prevailed also when comparing at constant mass loss.” “Results reported by Esteves et al. (2007b) with steam heat-treated pine wood (Pinus pinaster) showed a small increase until about 4% mass loss, followed by a decrease for higher mass losses. With the same treatment conditions, heating time, and temperature, the reduction of MOE was higher for the treatment in air, and relationship prevailed also when comparing at constant mass loss.” “In static bending tests (Fig 8) with Pinus radiata wood treated at 120oC, 150oC, and 180oC during 6 to 96 hours, Kim et al. (1998) showed that there was a close relationship between the decrease of bending properties (MOR, MOE and WML) and the process conditions (time and temperature). The work for the maximum load (WML) suffered an accentuated decrease, while the modulus of elasticity was affected less.” “Shi et al (2007) studied the mechanical behaviour of Quebec wood species heat-treated using the Thermowood process and concluded that the modulus of rupture decreased 21 between 0% and 49% for heat-treated spruce, pine, fir, and aspen, while for birch the modulus increased slightly (6%) after the heat treatment. Heat- treated spruce and pine modulus of elasticity decreased between 4% and 28%; however for fir, aspen, and birch the modulus generally increased. Mburu et al. (2008) with heat- treated Grevillea robusta wood found reductions on MOR and MOE reaching about 65% and 28%, respectively.” “Rusche (1973a, b) made heat treatments with and without oxygen using pine and beech and concluded that the modulus of elasticity decreased significantly for mass losses from 8 to 10%. Similar results were reported by Vital et al. (1983), with Eucalyptus saligna treated at 105-155oC for 10-160 hours. Mitchell (1988) studied heat-treated Pinus taeda at 150oC and 1, 2, 4, 8, and 16 h with equilibrium moisture of 0%, 12%, and green, in oxygen, nitrogen and air, and found that the MOE decreased irregularly with the time of treatment, decreasing more for green wood. For the heat treatment in air, MOE decrease was 14 times higher in green than in dry wood. In nitrogen there was no decrease of MOE, while with air the decrease was smaller than with oxygen. Santos (2000) reported for heat-treated Eucalyptus globulus a steep increase of the modulus of elasticity from 15,974 MPa to 27,646 MPa, although the time and the temperature of the treatment were not mentioned. Different results were reported by Esteves et al. (2007b) for the same wood treated at temperatures between 180 and 210oC. They found a slight increase at the beginning of the treatment, followed by a decrease. Sailer et al. (2000) did not find differences in the modulus of elasticity for oil and air heat-treated wood at 180oC, 200oC, and 220oC, and only for impact bending that decreased 51% in the case of oil and 37% in air. Kamdem et al. (2002) used spruce and beech treated by the French method (Rectified wood) between 200oC and 260oC, and obtained a decrease of 11% and 20% for MOE, 8% and 40% for MOR, respectively for spruce and beech. Goroyias and Hale (2002) studied the heat treatment of Pinus sylvestris chips for OSB production between 200oC and 260oC for 20 minutes. MOR did not decrease significantly for 200oC and 210oC, decreased slightly for 220oC, 230oC, and 240oC, and decreased significantly for 250oC and 260oC. The variation in MOE was similar with a decrease for temperatures higher than 240oC.” “Kubojima et al. (1998) made some vibrational studies with Picea sitchensis, and observed that the Young modulus in longitudinal and radial directions increased in the first two hours of treatment and remained constant afterwards for wood treated at 120oC and 160oC. At 200oC, the Young modulus increased in an initial phase, decreasing after that. The shear modulus in longitudinal and radial directions increased in an initial phase, becoming constant for 120 and 160oC, while for 200oC increased initially and decreased afterwards. The Young modulus increased with the increase of cellulose crystallinity and with the decrease of wood moisture. The effect of crystallinity prevails in the beginning of the treatment but with the continuation of the treatment the heat degradation is dominant, leading to the decrease of the Young modulus.” “Kubojima et al. (2000a) reported that the effects of heat treatment were similar in green and dry wood. The same authors (Kubojima et al. 2000b) also reported that the Young modulus and the bending strength increased in the beginning of the treatment, and decreased afterwards, more for the treatments in air than in nitrogen.” “Korkut et al. (2008a) studied heat-treated Scots pine wood and concluded that compression strength parallel to grain, bending strength, modulus of elasticity in bending, janka-hardness, impact bending strength, and tension strength perpendicular to the grain 22 decreased. Similar results were presented by Korkut et al. (2008b) for Red-bud maple (Acer trautvetteri).” “According to Boonstra et al. (2007b) heat-treated wood can be used in construction if the stresses that occur in construction are taken into account. These authors treated Norway spruce construction wood and obtained a 6% reduction in bending strength and a 17% increase on MOE. They also mentioned that only a combination of several defects, such as large knots, enclosed pith, and an abnormal slope of grain, decreases the bending strength and MOE of treated posts … “