Inductively cured glued-in rods in timber using Curie particles

Author’s Accepted Manuscript Inductively cured glued-in rods in timber using Curie particles Till Vallée, Michael Adam

www.elsevier.com/locate/ijadhadh

PII: DOI: Reference:

S0143-7496(16)30104-X http://dx.doi.org/10.1016/j.ijadhadh.2016.05.005 JAAD1848

To appear in: International Journal of Adhesion and Adhesives Received date: 8 February 2016 Accepted date: 11 May 2016 Cite this article as: Till Vallée and Michael Adam, Inductively cured glued-in rods in timber using Curie particles, International Journal of Adhesion and Adhesives, http://dx.doi.org/10.1016/j.ijadhadh.2016.05.005 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Inductively cured glued-in rods in timber using Curie particles Till Vallée1, Michael Adam2 Fraunhofer Institute for Manufacturing Technology and Advanced Materials IFAM, Wiener Straße 12, DE 28359 Bremen, Germany Phone + 49 421 2246-474 | Fax - 430 Keywords — A: adhesives for wood, epoxides; B: plastics, wood and wood composites; C: dynamic mechanical analysis; D: cure / hardening, glass transition temperature.

Abstract Commonly used adhesives for glued-in rods in timber engineering, cold curing two-component (2K) epoxies or polyurethanes, only harden relatively slowly (usually in hours to days), which is magnitudes of times longer than mechanical fastening. Additional constraints associated with aforementioned 2K adhesives arise from the fact that they usually necessitate some minimum temperature below which polymerisation can take place, respectively that pot life is strongly affected by (higher) ambient temperature; both limitations restrict the possibilities to bond onsite, depending on the location to a limited number of months of a year. Induction heating was investigated in the light of a potential timber engineering application, herein fast curing of glued-in glass fibre reinforced rods in timber. As none of the materials involved is susceptible to electromagnetic induction, metallic particles were mixed to a 1C epoxy. Induction of adhesives with particles has long been plagued by issues related to non-uniformity of heat distribution in the bondline, unless extensive temperature monitoring and control is exerted. To overcome these issues, metallic particles consisting of Mn-ZnFerrite, a Curie material that is not susceptible to induction heating beyond its Curie temperature, were used. The experiments showed that, by matching Curie temperature with the temperatures at which the adhesive cures, induction heating can be performed without any external control, thus freeing the process from monitoring equipment. Inductively cured glued-in rods reached shear strengths at the upper end of what is usually reported in literature. Compared to current practice of cold curing 2K polyurethanes or epoxies, the proposed method leads to bonded joints achieving full strength within minutes, instead of hours, or days, which represents a substantial gain in processing time, almost unaffected from outside temperature conditions.

1

[email protected]

2

[email protected]

1.

State-of-the-art

Adhesively bonding is among the oldest joining techniques known to humans [1], including for structural [2], and architectural [3] purposes. For a large part of the 20th century, adhesive bonding, with the prominent exception of prefabricated glued laminated timber, has fallen in relative disdain for civil engineering applications. This is also true in timber engineering, where most load bearing connections are still realized using mechanical fasteners. However, in recent decades, bonding as a joining technique experiences a revival [4], in particular with the introduction of synthetic adhesives, as epoxies and polyurethanes. Bonding particularly suits the anisotropic and brittle nature of timber, which, in contrast to pinned connectors, does not interrupt fibres and allows for a much smother load transfer between loaded members [5]. Adhesives permit the structural connection of timber with a wide variety of other materials, as for example concrete [6][7] to form timber-concrete structures, steel [8] as a very convenient way to build up complex timber structures, or glass [9] to develop architecturally appealing structural elements. Within timber engineering, glued-in rods represent a particular class of bonded joints in which load is transmitted from timber elements by means of rods through a layer of adhesive; this specific type of joints is widely been investigated [10], in particular for retrofitting existing structures [11]. Standard practice is to employ metallic rods, usually textured, or threaded rods, to improve mechanical interlocking. However, for a series of reasons, including improved resistance to corrosion in humid or acid environments, lower weight, easier and faster handling and installation, and lower heat conduction into the joint in case of fire [12], fibre reinforced polymers (FRP) can be considered to act as glued-in rods. Most available studies focus on Glass-FRP (G-FRP) rods, and Carbon-FRP (C-FRP) bars with even higher strength are generally discarded because timber strength is the limiting factor [13]. Commonly used adhesives for glued-in rods in timber engineering, cold curing two-components (2K) epoxies or polyurethanes, only harden relatively slowly (usually in hours to days), which is magnitudes of times longer than mechanical fastening. Additional constraints associated with aforementioned 2K adhesives arise from the fact that they usually necessitate some minimum temperature below which polymerisation can take place; respectively that pot life is negatively

affected by (higher) ambient temperature. Depending on the location, bonding onsite can only occur in a limited number of months of a year. Adhesive curing can be accelerated by a series of methods, including UV light [14], radiation [15], microwaves [16]; however, the most widely used method remains increasing curing temperature. Temperature increase acts directly on the curing kinetics, and its effect is most commonly described using Arrhenius’ Law [17]—which practitioners often apply in form of the rule of thumb that temperature increase of 10 °C halves the curing time. For smaller parts, or standardized parts manufactured in large numbers, temperature increase is very often achieved by oven curing; for the highly individualized and larger parts typically encountered in civil engineering, including timber engineering, oven curing is not an option, especially when manufacturing on-site. In the context of fast curing adhesively bonded joins, discarding ovens and direct radiation (e.g. infrared lamps, lasers or heating sleeves), the most widely used techniques to generate heat is electromagnetic induction. If metallic adherends are considered, induction heating acts on them, and the adhesive is heated via thermal conductivity. If inductive heating is to be used on non-conductive adherends, it is necessary to ensure that the adhesive reacts to electro-magnetic fields; this is mostly achieved by adding appropriate susceptors; typically metallic particles, or electrically conductive meshes. Two classes of susceptors can be adjunct to the adhesive [18] : firstly, large(r) sized components, as for example plates, or more commonly meshes, embedded within parts to be bonded; secondly, significantly smaller sized components, usually particles, directly mixed with the adhesive. Compared to meshes, particles embedded in adhesives adapt easily to a large variety of geometric situations; accordingly, handling adhesives filled with particles does not require any specialized skills or equipment. Typical particle susceptors encountered are Magnetite (Fe 3O4,[19]) and Maghemite (Fe2O3, [20]); additionally, a series of specifically designed susceptor particles have been developed, one example being MagSilica® [21]. Further information related to induction heating can be found in [25], which addresses the influence of particle size, coil geometry, frequency of the alternating current etc.

Although adjunction of susceptors allows remotely located induction coils to induce thermal energy, associated methods have often been “plagued by non-uniformity of heating of the bondline” [18]. It was nevertheless used in a series of practical applications, as for example to accelerate the curing of adhesively bonded structural composite tubes [22] with diameters up to 92 mm (35/8”), with NickelZinc-ferrites susceptor particles. MAHDI et al. [23] have compared oven-cured and induction-cured adhesively bonded single lap joints, and shown that there is no significant difference in strength and fracture toughness. The authors used a 2K epoxy on G-FRP adherends, electromagnetic susceptibility was achieved by means of a stainless-steel mesh. SUWANWATANA et al. [24] demonstrated that inductive-heating (using Nickel particles in poly-sulfones) can generate bond strengths of polymer– matrix composites comparable to those achieved in autoclave process; however, with the benefit of cycle times reduced with an order of magnitude. Recently, VALLÉE et al. [25][26] reported on the inductive heating of glued-in metallic using a 1C-epoxiy adhesive; induction heating proved functional, ultimately leading within 5 minutes to sufficient strength in the bondline to yield the steel bars in pull-out tests. However, the curing process had to be monitored, and induction power constantly adjusted (via thermocouples fixed on the rods) to prevent overheating of the adhesive. Glued-in G-FRP bars in beech, in combination with a Pre-Applicable Structural Adhesive (PASA®, a 2K epoxy) inductively cured were investigated by ADAM et al [27]. Although ultimately induction curing proved as efficient as oven-curing, leading to similar shear strength of the joint, it took the authors specific adaptation of the coil shape to keep temperature differences inside the adhesive layer within tens of degrees centigrade. Additionally, it proved necessary to control the induction power using embedded thermocouples within the adhesive layer. All metals and alloys that exhibit magnetic properties lose the ability to be susceptible to external electromagnetic fields beyond a material specific temperature [28]: this temperature is labelled Curie temperature (TC) in case of ferromagnetic materials; for antiferromagnetic materials, the analogous temperature is labelled Neel temperature (TN). In the context of this paper, Curie and Neel temperature are used indiscriminately as the point beyond which a material loses its magnetic properties, and thus no longer generates heating if subjected to electromagnetic induction— this feature is schematically

illustrated in Figure 1. The loss of electromagnetic susceptibly is dependent on particle size [29][30][31], pressure [32][33] or applied strains [34]—however, corresponding influences do not occur under conditions relevant for practitioners. For practical considerations, the Curie temperature is a property of the particles. For a large variety of alloys, Curie/Neel temperatures depend upon the relative contents of the components. To illustrate this on one example, Figure 2 depicts for Ni-Fe-alloys the influence of the alloy’s composition on the Curie temperature: at 30% Ni-content, TC is roughly 100 °C, TC increases up to 600 °C for a Ni-content of 70% [35]. Accordingly, it is possible to select specific Ni-Fe compositions to target any given Curie temperature (at least below 600 °C). The composition of an alloy does not only influence its Curie temperature TC, but also (amongst others) its magnetic permeability μ. For illustrative purposes again, Figure 2 shows that TC and μ do not follow similar trends; for practical applications in which a specific performance is required, several parameters will have to be balanced for selecting the right susceptor alloy. It is tempting to take advantage of Curie temperatures in the context of inductive heating, since ideally the specific phenomena associated would act as a switch to the induction process, with shutting down the heating beyond TC. Using Curie particles as susceptors would automatically lead to a substantial simplification of the process by reducing the sophistication of the technical equipment to monitor temperatures, respectively the skill of the induction operator. Mixing particles to an adhesive may adversely affect the mechanical properties of the adhesives: it is thus primordial to verify that the filled adhesive still achieves the required mechanical performance. Adding particles might also carries the possibility to raise issues of durability—e.g. corrosion of metallic particles; this matter has to be considered—and experimentally verified prior to any practical application. Besides these fundamental aspects, specific prerequisites to consider for Curie-particles acting as automatic temperature regulator in induction heating processes are: 

Firstly, curing dynamics of adhesives is governed by a combination of temperature and curing times, with limits imposed on the maximum curing temperature by chemistry. It is thus

fundamental to adapt the Curie temperature of the particles to the optimal curing temperature of the adhesive, which in turn means selecting particles according to the adhesive envisaged— or vice-versa, the adhesive in function of the particles. 

Secondly, the relationship between electromagnetic susceptibility (expressed as magnetic permeability) and temperature in Curie metals (and alloys) is not as simple as following the definition stated above; in general magnetic permeability, if plotted against temperature, drops more or less sharply around TC, which in turn means that corresponding heating effects might not completely cease to occur beyond the Curie temperature.



Thirdly, as for any filled adhesive, care must be taken to ensure that the particles added are equally distributed in the bulk and that they remain in place until hardening. Effects as segregation or percolation must be avoided, and in any case be verified.

The objective of the research presented herein is the inductive heating of glued-in rods using a combination of an epoxy and susceptor particles (Magnetite and a Mn-Zn-Ferrite-alloy) for the accelerated curing of G-FRP rods glued into timber. The paper guides through the necessary experimental determination of appropriate curing temperatures depending on the adhesive selected, and the corresponding selection of Curie particles. Based thereof, it is investigated how stringent the temperature control in the bonded joint is accomplished, and how uniform temperatures are distributed therein. Lastly, tensile tests of the joints will assess the mechanical performance of the process.

2.

Experimental investigations

2.1.

Glued-in-rods

For the subsequent tests, specimens were manufactured from blocks of beech (Fagus sylvatica) cut from boards stored prior to bonding in constant climate (25 °C) and conditioned to approximate moisture contents of 12%. Shear strength of the beech specimens from the same origin was determined in previous studies and amounted to 14.3±1.32 MPa. The glued-in rod samples tested consist of GFRP rods (=4mm) with an embedment length of 35 mm inserted in 30 x 30 mm² sections of timber.

Prior to inserting the rods, an oversized hole (2 mm wider in diameter than the rods, 6mm in total) was drilled centrally up to a depth of 35 mm; beyond that depth, a smaller hole with a diameter of exactly 4 mm was drilled to a depth of 5 mm to allow for a controlled centring. The rods consisted of pultruded G-FRP smooth bars made of glass fibres embedded in a vinyl-ester resin with a nominal diameter of 4 mm; tensile strength was determined in tests to amount for 813±23 MPa. The rods were inserted into the holes, a process in which excessive adhesive was spilled out. Four different experimental series were defined, corresponding to adhesives mixtures defined in Table 1; except for mixture M.0, for which only one sample was manufactured, all series consisted of five individual samples. Within the frame of this study, G-FRP rods and timber, both exhibiting very low magnetic permeability, are assumed not to interfere with the magnetic field generated during the induction. 2.2.

Adhesive

The adhesive used in this research was DELO-MONOPOX AD066, a 1C unfilled epoxy resin. According to its data sheet, “fast and high-strength” curing proceeds at temperatures between 120 °C and 150 °C, with curing times being dependent on temperature (e.g. 20 min at 130 °C). Lap shear strength on aluminium is stated to amount for 35 MPa, while tensile strength reaches 50 MPa; glass transition temperature is indicated as being 132 °C, without further specification according to which curing regime. Based on the data sheet, a curing temperature of 150 °C was targeted. In order to verify, and supplement, the information given by the data sheet, a Differential Scanning Calorimetry (DSC) was performed. For the DSC, temperature was first raised to 150° following a ramp of 100K/min, the kept isothermal at 150 °C for 4 min, then again raised following a ramp of 10K/min up to 250 °C. The amount of heat required to increase the temperature of the adhesive sample, compared to an inert reference, is continuously measured as a function of temperature; divided by the mass of the sample it is labelled heat flow.

2.3.

Susceptor particles

Two types of susceptor particles were considered in this study. Firstly, magnetite, Fe3O4, for which [36] reports Curie temperatures in the range of 773 to 850 K (roughly 500 to 576 °C), and initial magnetic permeability of around 610-3 H/m. The magnetite used herein was provided in form of particles with a size <44 μm. The second type of susceptor used in this study consisted of Mn-Zn-Ferrite particles with a size of 100-200 μm on average, as shown in Figure 3. According to the datasheet, initial magnetic permeability of the selected particles peaks at a temperature of around 125 °C to drop sharply at 155 °C, as depicted in Figure 3, which corresponds to its Curie temperature. In contrast to the magnetite, the drop in permeability is well marked, which matches the targeted curing temperature. According to the product’s data sheet, initial magnetic permeability below TC is in the range of 0.8-1.010-3 H/m, to sharply drop to zero after TC. The specific Curie temperature of the particles was experimentally determined using a modified Thermo-Gravimetric Analysis (TGA) in which a magnet’s force on a sample was monitored for temperatures up to 600 °C. Concisely, susceptor samples were placed in a TGA device, and then subjected to increasing temperatures, from 25 °C to 600 °C for the magnetite, and from 25 °C to 300 °C for the Mn-Zn-Ferrite particles, respectively. A magnet placed below the samples exerted a force that added to the dead weight thereof; the latter combination of forces was measured through the test. The idea being that when the temperature reached, or surpasses the material’s TC, a significant reduction in the magnetic force would result; the outcome of these experimental investigations are presented in Figure 4. Regarding durability, more specifically corrosion of the metallic particles, both particle types proved chemically stable with regard to corrosion under atmospheric conditions (cf. Fig. 4): up to 600 °C for Magnetite and up to 300 °C for Mn-Zn-Fe, respectively. Since durability of filled adhesives was not in the focus of this research, no specific tests were performed with regard to this important topic. Within this research, different configurations of susceptor particles were mixed by hand with the adhesive four mixtures are summarized as in Table 1. All mixtures exhibited sufficient stability with regard to sedimentation.

2.4.

Inductive heating of the glued-in-rods

Inductive heating was achieved by inserting all specimens inside a coil; the complete setup is depicted in Figure 5. The coil consisted of a hollow copper tube ( = 5.8 mm, t = 0.9 mm, inner/outer diameter = 40/55 mm, 11 windings with a length of 120mm) which was water-cooled so to prevent any heating of the coil. The coil was the electrically connected to the oscillating circuit that generated an alternating electromagnetic field with a frequency of 373 kHz; under full power, 6 kW were generated. No provisions were taken to determine the strength of the magnetic field within the coil, and no physical measurements allowed tracking back the shape of the latter. However, since the coil length (120 mm) was more than three times longer than the zone to be inductively heated, it was reasonably posited that the generated magnetic would be homogeneous along the 35mm long adhesive layer. Each specimen was submitted to the following induction regime: raising the induction power to full 100%, subsequently maintaining it without any control for 7 min, and finally shutting down the apparatus. In practice this can be considered equivalent to a non-monitored heating procedure, because embedded thermocouples were not used to regulate the induction regime. To allow for temperature readings during the induction process, thermocouples (provided with wave filters to reduce the electromagnetic interferences) were fixed directly on the G-FRP-rods. These gauges measured the temperature at the adhesive-rod interface along their length. For each series, one specimen was instrumented with three thermocouples: counting from the bottom of the rods the first was placed at 2 mm, the second in the middle at 17.5 mm, and the third one at 33 mm. For the remaining specimens of each series, only one thermocouple was fixed in mid-bar. 2.5.

Mechanical testing of the glued-in rods

Mechanical testing was performed on a universal testing machine (Zwick 1476, with a load capacity of 100 kN, fitted with a load cell of 20 kN) in displacement control (3 mm/min) at room temperature (+25 °C). Before testing, all specimens were left to cool for more than 24 hours. Besides loaddisplacement curves, determined from the crosshead displacements, ultimate loads were gathered.

3.

Results and discussion

3.1.

Adhesive characterization

The curing characteristics of the adhesive, as experimentally determined by means of a DSC are reported in Figure 6. The data logged during the test shows that after around 3.7 min heat flow peaks at a value of 8.67 W/g. Beyond that, heat flow asymptotically drops to almost zero (0.13 W/g reached at after less than 7 min). Another way to represent this data is to express the curing progress by means of the conversion degree, herein defined as the fraction of cumulated heat flow at each time step, W(t), relatively to the total heat flow, W (estimated as being the total heat provided). At each time step, cumulated heat flow is obtained through integration of the heat flow in the isothermal part of the curing regime. Plotting the conversion degree  = W(t)/ W, as done in Figure 7, shows that almost fully completed (to 99%) after less than 7 min, which corresponds to 4 minutes of curing at 150 °C. Accordingly, at a temperature of 150 °C, curing is almost complete after 4 min. 3.2.

Susceptor characterization

The estimation of the Curie temperatures of the susceptors using the modified Thermo-Gravimetric Analysis (TGA) yielded in the result displayed in Figure 4, where dead weight is augmented by the traction forces exerted by the magnet (refer to section 2.3). The data shows the magnet exerts a force corresponding to 2–3% of its dead weight up to the temperature range of 550-600 °C; beyond that range, the magnetic forces markedly drop to reach zero at approx. 600%. The temperature zone at which the Curie effects take place is relatively broad, since significant reductions start to be noticeable around 500 °C. Most important for this publication is the fact that for magnetite no Curie effects are likely to occur during the curing of the adhesive, which is targeted to be around 150 °C. Compared to the magnetite, results for the Mn-Zn-Ferrite particles are significantly different, as indicated by Figure 4. The drop in magnetic properties is well marked, with a very narrow temperature window around the specific value 150-155 °C. The tests, in essence, confirm that the Curie temperature Tc is adequately approximated by 155 °C.

3.3.

Inductive heating

The first series of temperature measurements aimed at determining the temperature distribution along the rods during the induction process. One specimen of each series was instrumented with three thermocouples (bottom, middle and top). The first observation is that the induction process using susceptor mixture M.0, cf. Figure 8.a, which only consisted of Mn-Zn-Ferrite particles, results in a relatively slow temperature increase of a mere 25 °C relatively to the initial room temperature, even after almost 7 minutes. Additionally, although temperatures at the top and in mid-rod are almost equal, temperatures at the bottom are noticeably lower; the temperature distribution within the sample thus cannot be considered homogeneous. Because of the poor heating performance of the specimen with susceptor mixture M.0, no further tests were carried out with adhesives containing solely Mn-ZnFerrite particles. For all other cases, and almost equally for all susceptor mixtures, as depicted in Figure 8.b—d, temperatures quickly, within around 90 sec, rose from room temperature to 150160 °C; then, very sharply, the increase of temperatures stopped and stabilized to almost constant 155 °C, on average. In the experiments carried out with the mixtures M8 and M12, and solely at the top of the rod, temperatures exhibit an overswing (cf. Fig. 8-c/d), to then fall back to the trend followed along the rod. The magnitude of this effect seems more important for the adhesive with the higher Magnetite content. Since the overswing was observed neither at the bottom, nor at the mid-position, it cannot be explained by the exothermy of the curing reaction itself. Discarding the overswing, differences in temperature at different positions along the rod, although perceptible, remain in a narrow range of ±5 °C. The close correlation of temperature along the rods led to the decision to limit temperature readings to the sole mid-position of the remaining samples. The scattering of temperature readings increases with increasing magnetite content, as show the diagrams depicted in Figure 8.b—d. In a second set of experiments, after rejecting adhesive mixture M.0, all remaining specimens with mixtures M.4, M.8, and M.12 were inductively heated following the very same regime described above; the difference being that only the temperature in mid-rod was monitored. All corresponding results are presented in Figure 9, where the observations made previously on the first set of specimens are confirmed; temperature quickly rose to 150-160 °C, and then stabilized. While the stabilization of

temperatures for series M.4 yields into an almost constant temperature over time, a slight increase is noticeable for M.8 (increase of approx. 1.4 K/min), an increase even more pronounced for M.12 (increase of approx. 2.3 K/min). Additionally, temperatures for all specimens remain in a relatively narrow temperature window (± 8 °C), indicating a high reproducibility of the induction heating process, even with no temperature control. The significant differences in inductive heating performance between the Curie and magnetite particles are evident. Mn-Zn-Ferrite particles alone fail at generating sufficient heat, and substituting as few as 4% Mn-Zn-Ferrite particles by magnetite dramatically enhances the thermal response. It is tempting to track back this effect to the difference in their respective magnetic permeability: for the Mn-Zn-Ferrite particles μ ≈ 0.8-1.010-3 H/m, for magnetite μ ≈ 610-3 H/m. However, the difference of magnetic permeability alone, although significant, does not offer a simple explanation for the fact that the dramatically faster heating of M.4 does not extend beyond TC, where readings suggest almost constant temperatures. Moreover, it does not elucidate why subsequent doubling, or tripling, of the magnetite content does not proportionally change the thermal behaviour of the particle mix; as temperature readings indicate that the heating rate beyond TC is only marginally changed from almost constant for mixture M.4 to a mere 1.4 K/min for M.8, respectively 2.3 K/min for M.12. The metallic particles were randomly distributed within the adhesive, it is thus reasonable to posit a homogeneous heat generation within the bulk – assuming a homogeneous magnetic field. Conversely, effects related to adhesive layer thickness, respectively thermal conductivity within the adhesive bulk are unlikely to explain the observed relationship between heating behaviour and the different mixtures of the particles. One possible explanation is hinted to by the relatively rapid cooling of the specimens, once the induction process is shut down: Figure 8 and Figure 9 both show that within 9 min, temperatures inside the joints drop from 150-160 °C to 25 °C. Obviously, the timber is poor at insulating the heated mix of adhesive and particles. The induction process had to balance thermal influx in the system, which occurred in the particle-filled adhesive, with thermal losses radiated through the outer surfaces of the specimen—via thermal conduction within the timber. Conversely, below TC, both particle types

together achieve a positive energy balance, resulting in temperature increases. Beyond TC, however, with the Mn-Zn-Ferrite particles no longer generating significant heat, magnetite particles alone merely achieve to maintain the energy balance (M.4, with almost no temperature increase), respectively develop a much smaller excess in thermal energy (M.8 and M.12, with very small temperature increases). 3.4.

Mechanical testing

Mechanical testing occurred 24h after induction; considering that temperatures after the induction process dropped within minutes to room temperature, as consistently indicated by Figure 8, this is magnitudes longer than needed to cool down. Loads and displacement, as displayed in Figure 10, were directly taken from the UTM, with no external transducer used. Load displacement curves exhibited a slight tendency to flatten around half the ultimate load reached. Failure, in most cases, occurred by adhesive failure along the smooth FRP rod, and only in limited occurrences was mixedadhesive-cohesive failure observed. In all cases, failure could be classified as brittle. Averaging all test results. failure loads amounted to 5.19±0.34 kN. Very similar averaged loads were obtained in each individual series (cf. Table 2); this suggests that failure loads and magnetite content are not correlated. Scattering of loads at failure was relatively low, with a variance of a mere 7%, again on average over all series, but also in each individual series. If related to the net cross section of the G-FRP rods, failure loads correspond to tensile stresses of 413±27 MPa, safely below the corresponding strength of 813±23 MPa. If smearing loads at failure along the bonded rod-adhesive surface, at which adhesive failure was mostly observed, shear stresses at failure of 11.80±0.78 MPa result. Achieved shear strengths thus compare to values reported in literature: e.g. 11.8 MPa with epoxy bonded G-FRP rods in C35 grade timber [37], 8.1–13.8 MPa for G-FRP rods in pine and oak [38], 3.6 MPa with G-FRP rods glued with epoxies in softwoods [39]. For displacement at failure, 1.54±0.41 mm if averaging regardless of susceptors mixture, there is a slight tendency for larger deformations with increasing magnetite content. If compared to the overall

average, deformations are 24% lower for series M.4, 9% lower for M.8, and 14% higher for M.12, respectively. Since no mechanical characterization of the adhesive-particle-mix was performed, no attempts were made to explain this difference.

4.

Conclusions

This paper investigated the possibilities offered by using Curie susceptors, which switch-off their magnetic properties beyond their Curie temperature, to inductively heating bonded joints, so to lift the constraints associated with temperature control and induction power control. The selected particles’ Curie point, 155 °C, matched well the temperature range in which the associated adhesive is able to cure rapidly. Differential Scanning Calorimetry showed that at that temperature, curing of the adhesive completes in less than 4 min. However, matching Curie and curing temperatures alone proved not a sufficient condition for fast curing, as experimentally shown in the heating test with only Mn-Zn-Ferrite particles mixed to the adhesive. The susceptibility of the adhesive-susceptor mixture was increased by substituting small proportions (4%, 8%, and 12%) of Mn-Zn-Ferrite particle with magnetite. In doing so, the best of both materials was used for the induction process: fast initial heating of the probes thanks to the magnetite, and temperature control owing to the Mn-Zn-Ferrite particles. The subsequent induction heating tests clearly showed that temperature was effectively controlled, and relatively equally distributed along the bonded rods; this in turn allowed to free the induction process of any form of temperature control. However, the experiments also indicated that the magnetite content could not be chosen randomly: beyond a critical proportion, herein estimated to be around 12 %, temperatures did not stabilize to an almost constant value, and scattering of the resulting temperatures did increase. It is posited that the investigated specimens did balance thermal energy influx with thermal losses at their outer surfaces, which depends on their geometry, but also from the environmental temperature. The magnetite content is thus a pivotal parameter for practical applications, in particular when considering inductive heating on-site, under low temperatures: chosen too low potentially slows the curing process; selected too high results in a less controlled process.

The inductively cured G-FRP rods bonded in beech manufactured resulted in glued-in rods of reproducible good quality: variance of the experimentally determined pull-out strength did not exceed 7%; and nominal averaged shear stresses at failure of 11.80 MPa were attained—which compares well with experimental results reported on glued-in rods. The research clearly indicates that, under laboratory conditions, induction leads to glued-in rods similar in strength to those manufactured considering cold curing 2K-adhesives practitioners usually use. Compared to current practices, full strength is achieved within minutes, instead of hours, or days, representing a substantial gain in processing time. Additionally, the heating process is largely independent from outside conditions, potentially allowing to extend the work on site all over the year. With the suggested use of Curie particles matching the curing temperature of the corresponding adhesive, the induction process is completely freed from any temperature monitoring and control, which lessens the burden for practitioner’s qualification. However, before amending the state of the art, a series of complementary questions have to be answered, among them quantifying the effect of relatively high mass percentage of metallic particles on the mechanical properties and the workability, respectively the durability of the particles embedded within the adhesive.

Figures

Figure 1: Curie material; magnetic permeability drops significantly beyond a its Curie temperature, labelled TC, which in turn significantly reduces its ability to generate heat by means of induction; note that depending upon the material considered, the drop from pre- to post-Curie behaviour occurs within a temperature range that differs in width

Figure 2: Influence of the composition of Ni-Fe-alloys on the Curie temperature and initial magnetic permeability (redrawn from [35])

Figure 3: Mn-Zn-Ferrite particles (left) Microscopy showing particle size (100-200μm), (right) magnetic permeability (redrawn with data from the manufacturer’s datasheet)

Figure 4: Results of the modified TGA analysis for both susceptor particles

Figure 5: Experimental setup for the induction process; a) general geometry of the specimens, b) the G-FRP-rods with thermocouples in mid-rod, c) the timber placed in the middle of the coil, d) general view of the specimens ready to be inductively heated

Figure 6: Thermo-analytical characterization of the adhesive by means of Differential Scanning Calorimetry (DSC)

Figure 7: Thermo-analytical characterization of the adhesive by means of Differential Scanning Calorimetry (DSC), evolution of conversion degree during the isothermal phase

(a) Mixture M.0

(b) Mixture M.4

(c) Mixture M.8

(d) Mixture M.12

Figure 8: Temperature readings along the rods for a representative of each of the four series

(a) Mixture M.4

(b) Mixture M.8

(c) Mixture M.12

Figure 9: Temperature readings in mid-rods for all specimens involving mixture M.4, M.8, and M.12

Figure 10: Load-displacement curves of the tested glued-in rods

Tables Table 1: Adhesive-susceptor mixtures, all percentages in mass

Mixture M.0 M.4 M.8 M.12

Adhesive 60% 60% 60% 60%

Magnetite 0% 4% 8% 12%

Mn-Zn-Fe 40% 36% 32% 28%

Table 2: Experimental results on the mechanical testing

Series

Load at failure [kN] Average

Std.-Dev.

M.4

5.02

0.23

M.8

5.33

M.12 Overall

Variance

Displacement at failure [mm]

Shear stress at failure at FRP-rod [MPa] Average Std.-Dev.

Average

Std.-Dev.

5%

1.17

0.08

11.42

0.52

0.40

8%

1.39

0.17

12.11

0.91

5.19

0.36

7%

1.75

0.61

11.80

0.82

5.19

0.34

7%

1.54

0.41

11.80

0.78

References [1]

Zipkin AM, Wagner M, McGrath K, Brooks AS, Lucas PW. An experimental study of hafting adhesives and the implications for compound tool technology. PLoS One 2014;9:e112560. doi:10.1371/journal.pone.0112560.

[2]

Adams R.D., Comyn J. WW. Structural adhesive joints in engineering. 2nd ed. London, Weinheim, New York, Tokyo, Melbourne, Madras: Chapman & Hall; 1997.

[3]

Keller T, Haas C, Vallée T. Structural Concept, Design, and Experimental Verification of a Glass Fiber-Reinforced Polymer Sandwich Roof Structure. J Compos Constr 2008;12:454–68. doi:10.1061/(ASCE)1090-0268(2008)12:4(454).

[4]

Vallée T, Tannert T, Fecht S. Adhesively Bonded Connections in the Context of Timber Engineering - A Review. J Adhes 2015;8464:150901092817008. doi:10.1080/00218464.2015.1071255.

[5]

Tannert T, Vallée T, Hehl S. Experimental and numerical investigations on adhesively bonded timber joints. Wood Sci Technol 2012;46:579–90. doi:10.1007/s00226-011-0423-1.

[6]

Cachim PB, Leitão de Oliveira CA, Negrão JHJ de O, Maia de Oliveira FM. Glued composite timber-concrete beams. I: Interlayer connection specimen tests. J Struct Eng 2010;136:1236– 45. doi:10.1061/(ASCE)ST.1943-541X.0000228.

[7]

Negr o JH, Oliveira F, Oliveira C, Cachim PB. Glued Composite Timber-Concrete Beams. II: Analysis and Tests of Beam Specimens. J Struct Eng 2010;136:1246. doi:10.1061/(ASCE)ST.1943-541X.0000251.

[8]

Vallée T, Tannert T, Hehl S. Experimental and numerical investigations on full-scale adhesively bonded timber trusses. Mater Struct 2011;44:1745–58. doi:10.1617/s11527-0119735-8.

[9]

Vallée T, Grunwald C, Milchert L, Fecht S. Design and dimensioning of a complex timberglass hybrid structure: the IFAM pedestrian bridge. Glas Struct Eng 2016. doi:10.1007/s40940016-0007-4.

[10]

Tlustochowicz G, Serrano E, Steiger R. State-of-the-art review on timber connections with glued-in steel rods. Mater Struct 2011;44:997–1020. doi:DOI 10.1617/s11527-010-9682-9.

[11]

Steiger R, Serrano E, Stepinac M, Raj??i?? V, O’Neill C, McPolin D, et al. Strengthening of timber structures with glued-in rods. Constr Build Mater 2015;97:90–105. doi:10.1016/j.conbuildmat.2015.03.097.

[12]

Bainbridge R, Mettem C, Harvey K, Ansell M. Bonded-in rod connections for timber structures - Development of design methods and test observations. Int J Adhes Adhes 2002;22:47–59. doi:10.1016/S0143-7496(01)00036-7.

[13]

Lorenzis L De, Scialpi V, La Tegola A. Analytical and experimental study on bonded-in CFRP bars in glulam timber. Compos Part B Eng 2005;36:279–89. doi:10.1016/j.compositesb.2004.11.005.

[14]

Goss B. Bonding glass and other substrates with UV curing adhesives. Int J Adhes Adhes 2002;22:405–8. doi:10.1016/S0143-7496(02)00022-2.

[15]

Thalacker VP. Radiation processing of coatings and adhesives - an overview. Int J Radiat Appl Instrumentation Part 1990;35:18–29. doi:10.1016/1359-0197(90)90051-I.

[16]

Zhou S, Hawley MC. A study of microwave reaction rate enhancement effect in adhesive bonding of polymers and composites. Compos Struct 2003;61:303–9. doi:10.1016/S02638223(03)00061-8.

[17]

Laidler KJ. The development of the Arrhenius equation. J Chem Educ 1984;61:494–8. doi:10.1021/ed061p494.

[18]

Tay TE, Fink BK, McKnight SH, Yarlagadda S, Gillespie JW. Accelerated Curing of Adhesives in Bonded Joints by Induction Heating. J Compos Mater 1999;33:1643–64. doi:10.1177/002199839903301704.

[19]

Zhao D-L, Zhang H-L, Zeng X-W, Xia Q-S, Tang J-T. Inductive heat property of Fe3O4/polymer composite nanoparticles in an ac magnetic field for localized hyperthermia. Biomed Mater 2006;1:198–201. doi:10.1088/1748-6041/1/4/004.

[20]

Verde EL, Landi GT, Carri o MS, Drummond AL, Gomes A, Vieira ED, et al. ield dependent transition to the non-linear regime in magnetic hyperthermia experiments: Comparison between maghemite, copper, zinc, nickel and cobalt ferrite nanoparticles of similar sizes. AIP Adv 2012;2:032120. doi:10.1063/1.4739533.

[21]

Maurer A, Lammel C. Rapid bonding of non-metallic materials. Adhes Adhes 2014;11:26–9. doi:10.1365/s35784-014-0261-2.

[22]

Wetzel ED, Spurgeon WA, Yungwirth CJ. Induction Bonding for Structural Composite Tubes. Army Res Lab Aberdeen Proving Gr Md Weapons Mater Res Dir 2002.

[23]

Mahdi S, Kim H-J, Gama BA, Yarlagadda S, Gillespie JW. A Comparison of Oven-cured and Induction-cured Adhesively Bonded Composite Joints. J Compos Mater 2003;37:519–42.

doi:10.1177/0021998303037006776. [24]

Suwanwatana W, Yarlagadda S, Gillespie JW. Hysteresis heating based induction bonding of thermoplastic composites. Compos Sci Technol 2006;66:1713–23. doi:10.1016/j.compscitech.2005.11.009.

[25]

Vallée T, Kohl D, Vorholt F, Böhm S. Die Fünf-Minuten-Gewindestange. adhäsion Kleb Dicht 2015;59:36–9. doi:10.1007/s35145-015-0534-1.

[26]

Vallée T, Kohl D, Vorholt F, Böhm S. The 5-minute-rod. Adhes Adhes 2015;12:28–31. doi:10.1007/s35784-015-0540-6.

[27]

Adam M, Lühring A, Popp M, Fecht S, Vallée T. Pre-Applicable Structural Adhesives for Timber Engineering: Glued-in G-FRP Rods. Int J Adhes Adhes 2015;67:1–19. doi:10.1016/j.ijadhadh.2015.12.034.

[28]

Sgobba S. Physics and measurements of magnetic materials. CERN- 2011;2010-004,:pp.39– 63.

[29]

Bertoldi DS, Bringa EM, Miranda EN. Analytical solution of the mean field Ising model for finite systems. J Phys Condens Matter 2012;24:226004. doi:10.1088/0953-8984/24/22/226004.

[30]

Parry LG. Magnetic properties of dispersed magnetite powders. Philos Mag 1965;11:303–12. doi:10.1080/14786436508221858.

[31]

Lopez-Dominguez V, Hernàndez JM, Tejada J, Ziolo RF. Colossal Reduction in Curie Temperature Due to Finite-Size Effects in CoFe 2 O 4 Nanoparticles. Chem Mater 2013;25:6– 11. doi:10.1021/cm301927z.

[32]

Bose SK, Kudrnovský J, Drchal V, Turek I. Pressure dependence of Curie temperature and resistivity in complex Heusler alloys. Phys Rev B - Condens Matter Mater Phys 2011;84. doi:10.1103/PhysRevB.84.174422.

[33]

Leger JM, Loriers-Susse C, Vodar B. Pressure Effect on the Curie Temperatures of Transition Metals and Alloys. Phys Rev B 1972;6:4250–61. doi:10.1103/PhysRevB.6.4250.

[34]

Sadoc A, Mercey B, Simon C, Grebille D, Prellier W, Lepetit MB. Large increase of the Curie temperature by orbital ordering control. Phys Rev Lett 2010;104. doi:10.1103/PhysRevLett.104.046804.

[35]

Davis JR. Nickel, Cobalt, and Their Alloys. 2000. doi:10.1361/ncta2000p013.

[36]

Levy D, Giustetto R, Hoser A. Structure of magnetite (Fe3O4) above the Curie temperature: a

cation ordering study. Phys Chem Miner 2011;39:169–76. doi:10.1007/s00269-011-0472-x. [37]

Serrano E. Glued-in rods for timber structures—An experimental study of softening behaviour. Mater Struct 2001;34:228–34. doi:10.1007/BF02480593.

[38]

Arriaga F, Íñiguez-Gonzáles G, Esteban M. Bonding shear strength in timber and G-FRP glued with epoxy adhesives. Wood Res 2011;56.

[39]

Gonzales E, Tannert T, Vallee T. The impact of defects on the capacity of timber joints with glued-in rods. Int J Adhes Adhes 2016;65:33–40. doi:10.1016/j.ijadhadh.2015.11.002.