The effect of biodeterioration on the structural connection performance of timber for conventional framing and mass timber has been investigated recently, but there is a need for additional data as well as for the development of analytical models to utilize these data. An empirical material model (seismic analysis of wood frame shear walls) was fitted to cyclic connection test data of four species of cross-laminated timber at different levels of biodeterioration by two brown-rot fungi. These model inputs were then analyzed to account for trends between wood species and fungal species. Weak trends were most prominent for initial stiffness, intercept load, and displacement at peak force. Correlations were poor with postyield and postpeak stiffness modifiers. These relationships were consistent both as a function of time and as a function of mass loss, but additional data are needed to more accurately predict the effects. The limited relationships likely reflect the variations in fungal decay across the test members.Abstract
One of the most used materials in the shift to mass timber construction has been cross-laminated timber (CLT). Substantial research has gone into structural characterization and design with CLT, ranging from floors, diaphragms (Popovski et al. 2016; Beairsto et al. 2022; Line et al. 2022a, 2022b), shear walls (Gavric et al. 2015; Popovski and Gavric 2016; Ho et al. 2017; Amini et al. 2018, 2021; van de Lindt et al. 2019; Moerman et al. 2023), and connections for the aforementioned elements. However, there are other concerns that can affect the structural integrity of the structure. Especially important are considerations relating to the risk of moisture accumulation that create conditions conducive to biodeterioration leading to losses in material properties of wood (Wang et al. 2018, Cappellazzi et al. 2020, Udele et al. 2021, Ayanleye et al. 2022). CLT is not immune from moisture intrusion and subsequent decay and could be subject to similar issues. Despite best intentions and efforts, some structures will be affected by moisture, and both the direct moisture effects, such as dimensional changes, and indirect effects, such as biodeterioration, need to be understood.
Understanding effects of biodeterioration on structural performance of CLT is critical, especially in load-bearing panels, but to date relatively few studies have examined the effects of decay on structural performance of building assemblies. To address this gap, Sinha et al. (2020) proposed a method to characterize biological degradation in building assemblies, especially connection systems. Udele et al. (2023) used this method to investigate wall-to-floor CLT connections after exposure to two brown-rot fungi for up to 40 weeks. Significant reductions in load-carrying and energy dissipation capacity were observed at the end of the exposure time, with a 74 to 90 percent reduction in energy dissipation in some assemblies. Yermán et al. (2022) showed that ultimate strength of nailed timber connections decreased by 40 to 60 percent after 34 weeks of exposure to brown- and white-rot fungi, but noted no significant change in stiffness. Kent et al. (2005) reported similar observations for nailed connections of oriented strand board to Douglas-fir exposed to brown-rot fungi for up to 30 weeks; a 0.1 change in specific gravity was associated with a 60 percent drop in energy dissipation of the connections but initial stiffness was not significantly affected.
Engineering models are useful for analyzing connections (Shen et al. 2013, Mahdavifar et al. 2019, Bhandari et al. 2023) and incorporating data from experimental investigations of connections into larger models. One common engineering model for wood lateral systems is the seismic analysis of wood frame shear walls (SAWS) model (Folz and Filiatrault 2004), which is a 10-parameter model, with five parameters modeling the envelope of the response and five parameters defining cyclic unloading, reloading, and degradation behaviors. The SAWS material model is implemented within OpenSees (McKenna et al. 2010) and was originally developed for conventional framed wood shear walls. As both conventional framed wood shear walls and most mass timber connections are heavily reliant on the behavior of small-diameter metal dowels, typically nails and screws, for their structural response, this model has been used for different mass timber connections (Shen et al. 2013, Mahdavifar et al. 2019, Mahr et al. 2020, Miyamoto et al. 2020, Bora et al. 2021). The envelope curve of the SAWS model has been used to model the effects of environmental changes on CLT connections both for elevated temperature (Mahr et al. 2020) and water intrusion (Bora et al. 2021). This suggests that the model could be useful for determining changes in hysteretic properties as a function of exposure to environmental effects.
Objectives
Understanding the effects of fungal biodeterioration on structural performance of mass timber connections is vital for ensuring the most effective application of these systems. Especially important is assisting designers in assessing and determining required repairs for existing structures affected by biodeterioration. The ability to effectively model the structural effects of biodeterioration can assist practitioners in both the design and analysis of mass timber structures. Therefore, the objectives of this study were to (1) develop model parameters for a common connection model (SAWS) from hysteretic results of a biodeterioration study (Udele et al. 2023); (2) investigate changes in the above parameters as a function of exposure time and mass loss; and (3) compare these trends across four CLT species exposed to two brown-rot fungi.
Materials and Methods
Experimental methods
Two CLT blocks were connected to form a T-shaped assembly using the A3 floor-to-wall connection developed by Amini et al. (2018) and van de Lindt et al. (2022), which was included in the 2021 special design provisions for wind and seismic design (AWC 2022). This connection used a bent 12-gauge (2.7 mm), A653 grade 33 galvanized steel plate (ASTM 2020) connected to the wall element using eight 8d common nails and to the floor element using two 16-mm bolts (Fig. 1). Each test specimen included two floor-to-wall connectors to avoid eccentricity.
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
A total of four CLT types was tested: V1 (Douglas-fir), V2 (spruce–pine–fir), V3 (southern pine), and CV3M1 (Norway spruce; APA 2018). All CLT pieces in this study were three ply, with 35-mm-thick plies for a total thickness of 105 mm. The two brown-rot species investigated were Rhodonia placenta and Gloeophyllum trabeum. Initial moisture contents were checked using a 50-mm pin-type probe attached to a Delmhorst RDM3 moisture meter to ensure blocks were approximately 12 ± 3 percent, per ANSI/APA PRG-320 (APA 2018). Since the aim of the study was to mainly characterize the performance of the connections, the pins were used to take moisture readings at three different locations around the A3 connectors to account for moisture variations. The CLT specimens were fully immersed in water for 4 weeks before inoculation to create conditions suitable for fungal growth and selected assemblies were periodically weighed to determine moisture uptake. Initial and final masses of the assemblies were used to determine moisture gain over the 4 weeks of soaking. Specimens were sterilized to avoid cross-contamination, and consequently inoculated with either of the brown-rot fungi. Sterilization was done in a Wellons kiln drier with temperatures set at 75°C for up to 14 hours. This method was sufficient to raise and hold the core of the CLT assemblies to a temperature of 70°C for up to 2 hours. The specimens were inoculated with fungal-colonized grain and placed in closed containers that were stored in a climate-controlled room at 27°C to accelerate fungal growth. Samples were harvested after 10, 20, 30, and 40 weeks. Exterior fungal growth was removed from harvested specimens and the specimens were air-dried before testing.
Dry controls and wet controls with no fungus were tested for comparison. Dry controls were specimens tested as made, whereas wet controls were soaked until moisture levels exceeded fiber saturation point, dried, and then tested. The purpose of these two treatment groups was to separate moisture-induced changes from fungal degradation. The wet controls were treated as exposure week 0 for both fungi. A total of 10 specimens was evaluated per test series (Table 1).
Testing was conducted using a 178-kN universal testing machine (Fig. 2). The specimens were tested quasistatically following the basic abbreviated CUREE protocol (Krawinkler et al. 2001) using ASTM E2126 (ASTM 2019) as a guide. The reference displacement for this, from previous testing by Bora et al. (2021), was 12.7 mm, and testing was conducted at a cyclic rate of 0.08 Hz. The displacement protocol included cycles with amplitudes through three times the reference displacement. Testing was concluded when one of the following occurred: all nails in the connection had fractured or fully withdrawn, complete delamination occurred between any of the plies in the wall piece, or the conclusion of the displacement protocol. Instrumentation included the actuator linear variable differential transformer to measure displacement and for feedback control and the actuator load cell for force measurement.
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
SAWS analysis
Load and displacement data were used to visualize hysteresis plots (Fig. 3a). The hysteretic plot was used to generate an envelope or a backbone curve by finding the peak force and corresponding displacement in each primary cycle of the CUREE displacement protocol. The hysteretic data can be used with different engineering models to analyze the response and observe trends. These engineering models can also be used in design by practitioners. Although the SAWS model involves 10 parameters, some of these parameters are esoteric and harder to derive and compare. As a result, only the parameters defining the envelope of the response were used in this study. These parameters were F0, the y-intercept of the postyield stiffness line, Du, the displacement at peak force, S0, the initial stiffness, R1, the ratio of the postyield stiffness to the initial stiffness, and R2, the ratio of the postpeak stiffness to the initial stiffness (Fig 3b).
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
These parameters were used to model the force at a given displacement before peak force using Equation 1 from Dolan and Madsen (1992) and Folz and Filiatrault (2001), whereas postpeak response is given by Equation 2 from Folz and Filiatrault (2001):
where FΔ is the force at a given displacement Δ and Fmax is the force when evaluating Equation 1 at Δ = Du.
Previous studies have used this methodology for examining mass timber (Mahdavifar et al. 2019, Mahr et al. 2020, Miyamoto et al. 2020, Bora et al. 2021). Similar to this study, both Mahr et al. (2020) and Bora et al. (2021) only looked at the characteristics from the envelope. The last five parameters (R3, R4, F1, alpha, and beta) apply only to the unloading and reloading branches of the hysteretic behavior until the response rejoins the backbone.
The modeling parameters were fitted using the simulated annealing optimization strategy from Matlab (Mathworks Inc. 2017). The primary objective of the optimization strategy is to minimize the percent difference in energy between the experimental and modeled results. Two parameters were taken directly from the data to simplify analysis, Du and Fmax. These two parameters were used to define a third parameter in terms of the others using Equation 3:
This simplification meant that only three parameters (S0, R1, and R2) needed to be optimized.
A maximum of 10,000 iterations was conducted on each specimen. The optimization was ended early if the objective function reached a value <2 percent; this occurred for most specimens.
Parameters extracted from each curve were used to calculate the mean, standard deviation, and covariance matrix for each parameter or set of parameters within a test series. These values were then regressed against fungal exposure time.
Results
Salient SAWS results
Detailed analyses of the hysteretic response including maximum force, stiffness, and ductility were presented in Udele et al. (2023). Regression analysis and pairwise comparisons of all these parameters were based on CLT species, inoculation time, and fungus species. Herein, only modeling parameters and their degradation with time or mass loss were investigated (Table 2) and representative experimental and optimized curves can be seen in Figure 4. It is important to note that the SAWS parameters do not allow for an offset at zero force. Some of the later tests exhibited a small slack behavior before picking up force. This was not due to the test fixture, but rather due to slack that developed within the connection itself as a result of deformation from wetting and fungal exposure. These results were not removed from the data, which reduced the fitness of the SAWS model for these samples, as evidenced by the lower values and higher variations for S0 and F0. Values tended to decrease as a function of time for the three primary parameters (F0, Du, and S0), but discrepancies were observed within these parameters where the property seemingly increased. These discrepancies were attributed to the high variance inherent in biological durability testing (Kent et al. 2005, Yermán et al. 2022). The fungi were introduced from the exterior surface of the assemblies and were likely to differentially colonize the wood, creating differing degrees of damage.
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
Linear regressions through both time and mass loss can be used to elucidate relationships between the SAWS parameters and the degree of degradation using degradation as a function of exposure time or as a function of percent mass loss. Although there was some variation between the wood species, in general they all experienced similar trends of direction, magnitude, and model fitness. As such, only one of the wood species, Douglas-fir, will be discussed in detail, and the regressions for all four species can be seen in Tables 3–6.
Both decay fungi show similar losses in the postyield intercept, F0, as a function of time, with the property reducing in value while increasing in variance as time progressed (Fig. 5). The linear fits through these data were weak, with R2 of 0.19 and 0.25 for G. trabeum and R. placenta, respectively. Fungal attack varies with moisture content, wood chemistry, wood anatomy, and wood grain orientation, among other factors. The large scale of the test assembly, compared with typical tests of fungal degradation, further increases the potential for differing degrees of decay in different areas of the same specimen. Larger differences were observed as a function of percent mass loss, with R. placenta exhibiting a slower degradation. Both of these fungi are widely used in decay tests because they are easily cultured and produced consistent mass losses; however, they are found in different niches. G. trabeum tends to be more prevalent in aboveground applications such as windows or door frames where moisture levels tend to be lower and more variable (Duncan and Lombard 1965). R. placenta is found in a variety of applications but is a common invader of Douglas-fir heartwood (Smith et al. 1987). R. placenta reduced the mass of the specimens at a faster rate with respect to time, but these losses had less impact on F0 than G. trabeum. Fitting changes in F0 as a function of mass loss exhibited weaker trends for both fungi when compared with exposure time.
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
One SAWS parameter that showed higher causality was ultimate displacement, Du (Fig 6). Displacement capacity decreased as a function of exposure time for both fungi, with the average Du dropping approximately 40 percent over 40 weeks or a linear rate of approximately 0.2 mm of displacement capacity a week. There were weak but suggestive coefficients of determination at 0.27 and 0.28 for G. trabeum and R. placenta, respectively. Similar trends were seen between F0 and Du with respect to percent mass loss, with R. placenta associated with slower degradation than G. trabeum.
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
The initial stiffness term, S0, showed little change with respect to either exposure time or mass loss (Fig. 7). Stiffness tended to be similar for the first 30 weeks of exposure; it decreased and became more variable at 40 weeks for both fungi. Assemblies exposed to R. placenta exhibited higher, more variable stiffness losses at 40 weeks. R2 values for both exposure time and mass loss were low for both fungi. Fungal degradation is rarely uniform across a member and this characteristic would tend to induce variable impacts on stiffness for individual elements. Eventually, this variability would decline as the decay process progressed.
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
As with the other parameters, R1, the ratio between postyield and initial stiffness, became more variable with time and mass loss, reflecting the inherent variability of biological activity; however, the general trend was positive for both decay fungi (Fig. 8). This positive trend did not suggest that stiffness increased as a function of exposure time or percent mass loss, since initial and postyield stiffness both declined. Postyield stiffness decreases more slowly than initial stiffness, thereby increasing the ratio as a function of time or increased mass loss. Despite some evidence of trends, the high variability once again resulted in low correlations for this parameter.
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
There was little correlation between the ratio of postpeak to initial stiffness (R2) and exposure time or percent mass loss for either fungus (Fig. 9). These poor correlations reflect the lack of consistent failure paths for each connection. Failures were generally dependent on local nail failures and these varied widely between assemblies as a result of differing degrees of decay. Additionally, some of the later tests had to be terminated because of wood failure away from the connection due to degradation. These failures sometimes occurred before a postpeak behavior was observed, further increasing variability.
Citation: Forest Products Journal 74, 2; 10.13073/FPJ-D-23-00064
Discussion
Although the variable nature of the decay process resulted in poor correlations between fungal attack and the SAWS parameters, there were some consistent trends. Three parameters (F0, Du, and S0) generally decreased as a function of time. The postyield stiffness modifier, R1, showed that postyield stiffness, R1S0, decreased as a function of exposure time, but at a slower rate than initial stiffness, S0, alone. Postpeak stiffness, R2, showed no observable trends as a function of exposure. Part of the lack of observable trends in R2 may be due to premature failure of pieces at higher exposure times, resulting in less available data for the regression. Property losses were consistently seen between the dry and wet controls, with the wet controls having higher ultimate displacement capacity, Du, but lower initial, S0, and postyield stiffness, R1S0. These results contrast with those of Bora et al. (2021), who found little difference between dry control and redried specimens. Differences in the drying protocol may help explain the results since Bora et al. (2021) air dried the specimens, whereas the specimens in this study were dried using a kiln at a slow drying rate.
Ultimately, none of the regressions through the properties as functions of either time or mass loss accurately determined biodeterioration-associated loss of connection properties. Decay is rarely uniform in a timber assembly and the scale of these test members made it highly likely that there would be considerable variation in the degree of decay. As a result, the degree of decay close to the connection could vary considerably between assemblies even if the average mass losses were similar. This likely contributed to the very high variance.
Another observation with the data was a flat or weak slightly positive trend for R1. This trend indicated that postyield stiffness losses occurred at the same or slower rate than those for S0. Thus, although there were postyield stiffness losses, they occurred more slowly than losses in initial stiffness. This contrasts with previous work on wetting effects that showed no obvious trend in R1 with prolonged wetting (Bora et al. 2021). This suggests that the effects of biodeterioration on postyield stiffness are unrelated to the moisture trend. Additionally, R2 showed no appreciable trend as a function of time or mass loss, suggesting that postpeak stiffness does not have a trend outside of the initial S0 losses. This finding on R2 lines up with the moisture trend in Bora et al. (2021), suggesting that, outside of damage causing more variation in the postpeak performance, no appreciable trend can be found in the postpeak stiffness.
One important factor in any study involving biodeterioration is the organisms involved. In this study, two brown-rot fungi were used. Brown-rot fungi generally preferentially use the hemicelluloses and cellulose, leaving a heavily modified residual lignin. However, brown-rot fungi have a range of decay capacities, which can vary depending on the wood species and environmental conditions. In this study, samples exposed to R. placenta exhibited higher property losses than those exposed to G. trabeum and these differences were also present with mass loss. This suggests that, although the correlation between assembly mass loss and connection properties was weak, mass loss did correlate with structural property losses for larger mass timber connections.
Although the data produced only suggestive conclusions, the application of the SAWS material appears useful for designers. There are limited data on the effect of biodeterioration on connection properties of mass timber and existing data focus primarily on strength and initial stiffness. Quantifying the effects of decay on postyield and postpeak behavior may make it possible to examine the effects of biodeterioration on structures subjected to an earthquake. Although more data are needed, fitting degradation data to engineering models appears to be an effective approach.
Conclusions
The SAWS modeling parameters were derived for CLT connection of four species after exposure to two different fungi for a total of 40 weeks. Statistically, the SAWS parameters were generally poorly correlated with fungal-associated mass losses of the CLT assemblies, but there were some trends with some parameters that suggest they could be useful with additional data. Overall, a few conclusions can be made:
-
SAWS parameters can be determined from hysteretic data and the parameters can be used to generally track trends due to biodeterioration.
-
Connection properties were affected by biodeterioration, showing average losses of 13 and 40 percent over 40 weeks of exposure for initial stiffness and ultimate displacement, respectively.
-
The variability of fungal attack and the large dimensions of the assemblies likely created nonuniform decay in the CLT that affected the results. The use of fewer variables (fewer fungi or wood species) and perhaps smaller assemblies would allow for additional replication with more uniform decay.
-
The results suggest that the postyield stiffness reduced at or more slowly than the rate of initial stiffness for all CLT and fungal species.
Future studies into the effects of biodeterioration on connection performance will help to solidify the findings of this study, especially on smaller-scale connections. Additional research using SAWS parameters and physical testing of full-scale walls would also assist practitioners in the confidence of modeling walls using small-scale deterioration connection data.
Specimen layout (all dimensions mm).
Physical testing apparatus.
(a) Backbone from test data and (b) seismic analysis of wood frame shear walls (SAWS) envelope parameters.
Representative seismic analysis of wood frame shear walls (SAWS) and hysteretic backbone curves for (a) a Douglas-fir sample exposed to Gloeophyllum trabeum for 10 weeks; (b) a spruce–pine–fir sample exposed to Rhodonia placenta for 40 weeks.
Seismic analysis of wood frame shear walls parameter distribution of F0 as a function of (a) time and (b) percent mass loss after exposure of Douglas-fir samples to Gloeophyllum trabeum or Rhodonia placenta.
Seismic analysis of wood frame shear walls parameter distribution of Du as a function of (a) time and (b) percent mass loss for Douglas-fir samples after exposure to Gloeophyllum trabeum or Rhodonia placenta.
Seismic analysis of wood frame shear walls parameter distribution of S0 as a function of (a) time and (b) percent mass loss for Douglas-fir samples after exposure to Gloeophyllum trabeum or Rhodonia placenta.
Seismic analysis of wood frame shear walls parameter distribution of R1 as a function of (a) time and (b) percent mass loss for Douglas-fir samples after exposure to Gloeophyllum trabeum or Rhodonia placenta.
Seismic analysis of wood frame shear walls parameter distribution of R2 as a function of (a) time and (b) percent mass loss for Douglas-fir samples after exposure to Gloeophyllum trabeum or Rhodonia placenta.
Contributor Notes
The authors are, respectively, Postdoctoral Scholar, Postdoctoral Scholar, Distinguished Professor Emeritus, and Professor and JELD-WEN Chair in Wood-Based Composites Sci., Wood Sci. and Engineering, Oregon State Univ., Corvallis (ian.morrell@oregonstate.edu [corresponding author], Kenneth.udele@oregonstate.edu, jeff.morrell@oregonstate.edu, arijit.sinha@oregonstate.edu). This paper was received for publication in December 2023. Article no. 23-00064.