Metal-Plated Connections in Sustainable Lightweight Construction: A Weak Link in Fire Conditions?

Abstract

Lightweight engineered trusses support sustainable construction with the benefits of mass production and fast construction at lower costs. However, the truss system has raised concerns due to premature failure in fire conditions. This study investigates the effect of a thin soot layer on the surface of the gusset plate and the teeth of the gusset plate on the temperature development within lightweight wood specimens in fire conditions. A 10 cm long, 8.9 cm wide, and 3.8 cm thick dimensional lumber (often called 2 by 4) partially covered by a gusset plate was exposed to a constant incident radiant heat flux. A total of 12 experiments were conducted with four different configurations, bare gusset plates with and without teeth and soot-coated gusset plates with and without teeth, at three different external radiative heat fluxes of 10, 15, and 20 kW/m². The exposure durations were set to be 60, 40, and 30 min, respectively, to allow the total applied amount of radiant energy for each specimen to be identical. Three thermocouples were installed at a depth of 13 mm from the exposed wooden surface: two beneath the gusset plate and one below the uncovered wooden surface, and an additional thermocouple was between the gusset plate and the wood surface. The obtained temperature data showed that soot-coated gusset plates absorb significantly more radiation and record higher temperatures within the specimens than the specimens with the bare gusset plates. It was also found that the bare gusset plate works as a protective layer for the wood at 20 kW/m², but not at 10 and 15 kW/m². The teeth certainly contributed to heat transfer increasing the temperatures within the wood higher than those without teeth, but the effect was only meaningful for the soot-covered specimens. Connection strength was also qualitatively analyzed and it was discovered that the bare specimen retained a strong connection between the gusset plate and wood. In contrast, the soot-coated specimen was easily removed by hand, even when exposed to the same heat flux. Applying these results to a realistic scenario, this loss in connection strength could result in truss failure and structural collapse, which may result in injury to or even death of the responding firefighters. Additional gusset plate protection measures may be necessary to prolong the connection strength and prevent structural collapse.

1. Introduction

During offensive firefighting operations, firefighters are at risk of severe injury or possibly death due to structural collapse of the floor or roof. These potentially fatal collapses are typically due to truss failure, often without any warning or indicators [1,2,3]. In general, the number of firefighter fatalities has been steadily declining, which also has resulted in a decreased number of fatalities due to structural collapse [4,5]. However, the percentage of fatalities due to structural collapse in residential homes has increased with most occurring during attack operations [4]. Modern residential homes often generate more adverse fire behavior than older legacy homes, which could result in more hazardous conditions for responding firefighters. Variables such as increased home size, open geometries, increased synthetic materials, and sustainable lightweight construction have allowed for quicker fire development and shorter times to collapse [6]. Looking at fire department response objectives and average fire department response times, structural collapse can occur before fire department arrival or shortly after arrival in the worst case [2,6,7,8,9]. Legacy homes had collapse times as long as 40 min after fire department arrival, allowing for a much longer time to perform offensive operations [6]. The changes that have increased fire propagation and shortened the time to collapse, such as lightweight construction, have possibly influenced the percentage of fatalities occurring in residential homes

Lightweight engineered trusses are often used to support both ceilings and floors. Online estimates place the percentage of new homes being made with them as high as 90% in the United States, indicating the popularity of the design for sustainable lightweight wooden residences [10]. They are common due to their various benefits such as easy mass production, safe handling, lower construction costs, thermal insulation, waterproofing, strength, and sustainability [1,10,11]. They generally consist of lightweight wooden members connected by metal plate connections, commonly referred to as gusset plates. Gusset plates were developed in 1952 by Carrol Sanford after a series of metal connection tests. The plate is commonly made from galvanized steel with teeth punched perpendicular to the plate with a length usually between 8 and 15 mm [10]. These trusses are typically not protected in any way and thus are prone to failure under fire conditions, which has resulted in the injury and death of firefighters [11].

Lightweight engineered trusses can fail in two locations: the gusset plate connection between wooden members or the wooden member itself. The gusset plate connection can fail in two ways: tooth withdrawal and failure of the metal itself. Tooth withdrawal is a common form of failure that can occur with a sufficient force pulling out the teeth or the failure of the wood beneath the plate [12,13]. Tooth withdrawal was found to be the mode of failure for a truss at elevated temperatures [13] possibly due to increased degradation of the wood beneath due to heat transfer through the plate’s teeth [1,3]. Charring of the wood beneath the plate or the thermal expansion of the plate relative to the wood can also result in a loose connection and potential tooth withdrawal [14]. Gusset plates are typically made of galvanized steel and thus are subject to the physical and thermal properties of steel. As temperatures increase, steel loses its structural strength, which may result in the physical breakage or bending of the plate exposed to fire conditions [15]. However, ductile failure due to tensile or shear conditions is not as common and has a longer failure process than tooth withdrawal [10]. To improve the strength of gusset plate connections concerning tooth withdrawal and ductile failure, the overall surface of the plate-wood connection should be increased [10] as well as increasing plate thickness, tooth length, and length of the plate, which is correlated with an increased number of teeth [13]. The failure of the wooden member itself may also occur during a fire. When the wooden member is exposed to elevated temperatures, natural charring may occur, which can provide temporary protection to the wood [16,17]. However, the structural strength of the wood is also decreased relative to the depth of charring [1,14,16]. This charring effect can also occur underneath the gusset plate connection, especially if the heat transfer beneath the plate is significant, resulting in tooth withdrawal from the charred, brittle wood beneath.

Trusses are typically located in concealed spaces within buildings, such as in attics or between floors. Trusses also contain large void spaces that can allow for fast, uncontrolled fire spread [3]. Because of the openness of the truss void, uncontrolled fires can affect a large surface area of wooden truss members on multiple trusses at once [16]. In many residential homes, trusses are not protected by sprinkler systems or adequate firestopping and draftstopping, further increasing the issue of uncontrolled fire spread [11,16]. This typically leaves gypsum board ceiling finish as the only line of defense against fire exposure to trusses in many homes. The lightweight wooden members have little inherent fire resistance and thus rely on the gypsum board to increase the fire resistance of the entire assembly [18]. Unfortunately, gypsum board fall-off can occur in fire conditions and is a significant variable in truss failure time [2,19]. Homes may also contain suspended, tiled, wooden, or other ceilings that offer little fire resistance. Once the ceiling is breached, fire can spread into the concealed space where it can then severely damage the trusses. Conditions in the attic, or other concealed spaces, can be just as severe as the rooms located beneath them. In a series of tests previously performed, attic spaces in simulated homes were breached by fire between 6 and 8 min and reached flashover conditions roughly 5 ½ min later with a peak temperature of around 500 °C before collapse [2].

One major report on gusset plate performance evaluated a gusset plate that was attached to a lightweight wooden member and exposed to a radiant panel [20]. The specimens were left entirely bare, and it was concluded that the gusset plates protected the wood beneath them when the gap between members was small or nonexistent, indicated by lower temperatures and reduced charring beneath the plate as opposed to the exposed wood. The authors indicated that sootiness may affect the radiation absorptivity of the plate. However, another study suggested that the gusset plate’s teeth transferred heat deeper into the wood beneath [13,21]. Other perspectives, including firefighters [20,22], heavy timber researchers [17,18], and other researchers [14,23,24] share the view that metal plate connections are a weak point that increases heat transfer into the wood exacerbating the structural integrity of the lightweight engineered trusses. Thus, the effect of the gusset plate connections on heat transfer into the wood beneath during fire conditions is not fully understood. Previous studies did not account for sooty conditions or high enough temperatures found in attics [2,13,20]. Firefighters’ concerns are based on field observations and investigations. Some previous research is focused only on metal plate connections between heavy timber joints [17,18], and other lightweight wood studies do not explicitly say why gusset plate connections are a weak point [23,24]. Based on these identified limitations, a certain gap exists in the understanding of how gusset plates perform in fire conditions and how performance is affected by certain variables such as heat flux applied to the surface, surface coatings, and the presence of gusset plate teeth.

The objective of this study is to investigate the effects of surface absorptivity and the presence of the teeth of the gusset plate on the heat transfer phenomena to the wood substrate under different external incident heat fluxes by measuring temperatures at various locations within the wood specimens. Independent variables that affect heat transfer phenomena such as heat flux applied, the presence of teeth (conduction into the wood), and the presence of a surface soot coating (radiation absorptivity) are examined experimentally. An increased understanding of these phenomena can lead to protection efforts of gusset plates that reduce firefighter fatalities and injuries.

2. Materials and Methods

2.1. Lightweight Wood–Gusset Plate Specimens Subjected to Radiant Heat Fluxes

Lightweight spruce pine lumber with a size of 3.8 cm in thickness and 8.9 cm in width (1.5 inches by 3.5 inches) was chosen in this study, as it is most commonly used in lightweight wooden residential buildings. The wood’s moisture content falls in the range of 9–14%. The lumber was cut into a section 10 cm long and a 7.6 cm long by 5.4 cm wide gusset plate was affixed to it. The gusset plate was purchased at a local hardware store and had 48 teeth that were 9 mm deep over a 54 mm by 76 mm surface area.

Four separate combinations of gusset plates were tested:

• Bare: no coating was applied to the surface of the plate or wood.
• Bare and toothless: no coating was applied to the surface and all teeth were removed except four corner teeth necessary to affix the plate to the wood.
• Soot-coated: candle flame soot was applied to cover the surface of the plate and wood.
• Soot-coated and toothless: candle flame soot was applied to the surface and all teeth were removed except four corner teeth necessary to affix the plate to the wood.

These four specimens tested the effect of teeth and surface characteristics on heat transfer into the wood. The soot-coating represents simulated fire conditions, where the gusset plates have been exposed to a sooty fire condition. Figure 1 shows the difference between the bare and soot-coated specimens; the exposed wood surface as well as the gusset plate were covered by soot in the soot-coated specimens.

A cone heater was used to generate external radiative heat flux as shown in Figure 2. A cone heater produces a constant heat flux after the user sets a temperature for the cone heater that is known to generate the desired heat flux. The temperature is correlated to heat flux through calibration with a heat flux gauge. During the testing, the temperature was maintained at the exact temperature required for the desired heat flux and did not fluctuate. All four specimen combinations were exposed to three separate incident radiant heat fluxes: 10, 15, and 20 kW/m² for 60, 40, and 30 min, respectively. These durations were selected to equal the total amount of incident radiant energy applied in all three cases.

The cone calorimeter was located within a temperature-controlled laboratory, where the ambient air temperature was maintained at 22 °C and no appreciable wind currents were present.

2.2. Instrumentation and Test Matrix

Three holes were created within the wood specimen using a drill press to ensure even depth and drilling angle. Four 24-gauge Type K thermocouples were inserted in the following locations.

▪

Between the plate and wood (TC1)

▪

Directly beneath a row of teeth of the gusset plate (TC2)

▪

Between the two rows of teeth of the gusset plate (TC3)

▪

Beneath the exposed wood (TC4)

TC2, 3, and 4 are located 13 mm beneath the surface, while TC1 is directly between the gusset plate and the wood. All thermocouple beads are located 45 mm deep in the specimen. Figure 2 depicts the thermocouple placement within the sample. A thermally conductive cement (Omega Bond 600 manufactured by Omega Engineering) was used to seal the thermocouples into the specimen.

A total of 12 experiments were conducted as shown in

Table 1. Test matrix.

Test Number	Test Name	Surface	Teeth Presence	External Heat Flux (kW/m2)
1	Bare-teeth-10	Bare	Teeth	10
2	Bare-teeth-15	Bare	Teeth	15
3	Bare-teeth-20	Bare	Teeth	20
4	Soot-teeth-10	Soot	Teeth	10
5	Soot-teeth-15	Soot	Teeth	15
6	Soot-teeth-20	Soot	Teeth	20
7	Bare-noteeth-10	Bare	No teeth	10
8	Bare-noteeth-15	Bare	No teeth	15
9	Bare-noteeth-20	Bare	No teeth	20
10	Soot-noteeth-10	Soot	No teeth	10
11	Soot-noteeth-15	Soot	No teeth	15
12	Soot-noteeth-20	Soot	No teeth	20

. To refer to the temperatures at different locations, the thermocouple name is used as a suffix to the test name. For example, Bare-teeth-10-TC1 indicates the temperature measured between the bare gusset plate with teeth and the wood surface exposed to 10 kW/m² radiant heat flux.

3. Results and Discussion

Multiple thermal reactions occur in the wood at high temperatures. Wood consists of three fundamental components: hemicellulose, cellulose, and lignin [25]. For spruce wood, the mass fraction of each component is 0.23, 0.44, and 0.31 with other minor components [26]. For each component, the temperature range of the reaction and their thermal effect on the temperature of the wood is different. Generic temperature ranges for wood pyrolysis are included below [27].

• Below 200 °C: endothermic dehydration.
• 200–500 °C: exothermic reaction of hemicellulose and lignin with two peaks around 275 and 365 °C, respectively.
• 300–400 °C: endothermic reaction of cellulose with a strong peak of around 360 °C.
• Above 500 °C: exothermic reaction of cellulose and endothermic reaction of hemicellulose and lignin.

These thermal reactions certainly influence the wood temperature in addition to the heat transferred from the hot external environment.

3.1. Effect of Surface Absorptivity (Soot-Coated vs. Bare)

The soot surface coatings had a significant effect on the temperatures measured both between the gusset plate and wood surface and within the wood specimen. The presence of a dark and rough texture of the soot-coated gusset plates resulted in a greater absorption of radiation compared to the relatively shiny bare plates. As shown in Figure 3, the temperature between the soot-coated gusset plate and the wood surface (TC1) was higher than that of the bare plates for all three external incident heat fluxes. For example, at 1800 s, Soot-teeth-20-TC1 is at 462 °C, which is approximately 130 °C higher than Bare-teeth-20-TC1 being at 335 °C. Likewise, soot-coated gusset plates consistently showed higher temperatures than bare gusset plates at the incident radiant heat fluxes of both 10 and 15 kW/m².

A new temperature, TC2.5, is calculated by averaging the values of TC2 and TC3 to develop a representative internal wood temperature beneath the gusset plate. It should be noted that while TC2 is located right beneath a row of teeth, its cross-directional location may be between two teeth. Similar to TC1, soot-coated specimens also recorded higher temperatures beneath the plate within the wood (TC2.5) as shown in Figure 4. The greater radiation absorbed by the soot-coated gusset plate was subsequently conducted through the wood beneath the plate leading to higher temperatures in the soot-coated specimens when compared to the bare specimens. For TC4, beneath the exposed wood, the result is slightly different for 20 kW/m² as shown in Figure 5; the bare specimen showed a higher temperature than the soot-coated specimen. Smoldering combustion with glowing embers on the surface was observed for both cases (bare and soot-coated) and is an exothermic oxidation reaction that contributed to the unexpectedly higher temperature for the bare specimen than the soot-coated specimen. This indicates that surface absorptivity is less important once active pyrolysis or smoldering ignition occurs in wood and that the corresponding temperature beneath the exposed wood is less predictable.

At the external heat flux value of 20 kW/m² for the bare surface cases, the gusset plate provided an apparent protection as the temperatures beneath the plate (TC2.5) are lower than the temperature beneath the exposed wood (TC4) as shown in Figure 6. It should be noted that the exposed wood surface showed significant smoldering combustion with embers at 20 kW/m² while the wood directly beneath the gusset plate less presented this phenomenon. This indicates that at 20 kW/m², the bare gusset plate can prevent smoldering combustion beneath the plate (TC2.5), thus providing some protection when compared to the exposed wood (TC4) where smoldering and higher temperatures did occur. This was not seen at the 10 kW/m² or 15 kW/m² heat flux level where less or no significant smoldering occurred on the exposed wood, due to the lower incident heat flux levels that did not support the same level of smoldering combustion and pyrolysis as the 20 kW/m² level. The same phenomena were observed for the bare toothless specimen at 20 kW/m² as shown in Figure 7.

Unlike the bare specimens, for the soot-coated specimens, the gusset plate did not provide any apparent protection for all heat fluxes. The temperatures beneath the gusset plate were higher than the temperature beneath the exposed wood at all three levels of heat flux as shown in Figure 8. Overall, these findings indicate that a dark, rough metal surface absorbs more radiation than a bare metal surface, which is then conducted into the wood and can thermally degrade the wood at a quicker rate.

3.2. Effect of Teeth

This experiment also investigated how the presence of teeth affects heat transfer into the wood. Again, TC2.5, an average of TC2 and TC3, represents the temperature beneath the gusset plate. For the bare specimens exposed to 20 kW/m², TC2.5 of the toothed specimen was higher than that of the toothless specimen, but the difference was within 15 °C as shown in Figure 9. The toothed specimen exposed to 10 kW/m² recorded slightly higher temperatures than the toothless specimen while the toothed specimen exposed to 15 kW/m² recorded a temperature slightly lower than the toothless specimen. This is surprising as the thermal diffusivity of galvanized steel is almost 10 times higher than that of spruce wood; the thermal diffusivities of galvanized steel and spruce wood are approximately 15 mm²/s [28] and 0.15 mm²/s [29], respectively. The potential reason for this small temperature difference needs to be further investigated. However, it may be conjectured that (1) the net energy absorbed on the bare gusset plate is not large enough to make a significant temperature difference, (2) lateral heat transfer through the wood between teeth precedes any meaningful downward heat transfer, and (3) a combination of these.

In the soot-covered specimens, the presence of teeth has a more pronounced effect on heat transfer into the wood, which supports the conjecture (1) above. At the 10 kW/m² level, the teeth had a negligible effect on heat transfer, similar to the bare cases, where the temperature differences are within 15 °C. However, at the 15 kW/m² level, the toothed specimen recorded a temperature around 50 °C higher than that of the toothless specimen, shown in Figure 10. The difference at TC2.5 for the 20 kW/m² specimens is not as large, about 25 °C by the end of the test. In general, the presence of teeth has a greater effect on heat transfer into the wood beneath in the soot-covered specimens, especially for 15 kW/m² and 20 kW/m².

3.3. Visual Damage Assessment

Physical damage of the toothed specimens was visually assessed after heat exposure testing. In Figure 11, the top view of the specimens is included. From left to right, the columns of specimens are 20 kW/m², 15 kW/m², and 10 kW/m². The top row of specimens (A, C, and E) indicates the bare, toothed specimens and the bottom row (B, D, and F) indicates the soot-covered, toothed specimens. More pyrolysis and thus surface damage occurred with higher incident heat flux and to the soot-covered specimens. Especially, in the 15 kW/m² exposure, the soot-covered specimen (D) apparently has more damage than the bare specimen (C), indicating that more radiation energy was absorbed by the soot-covered wood surface.

Figure 12 shows the bare, toothless specimen (left) and the soot-covered, toothless specimen (right) at a closer view and with the gusset plates removed. Both were exposed to 20 kW/m² heat flux. Limited pyrolysis was observed underneath the bare gusset plate specimen when compared to the soot-covered gusset plate specimen while similar pyrolysis was observed for the exposed wood surface. The bare gusset plate reflects radiant heat energy and limits the heat transfer to the wood surface more than the soot-covered specimens. As previously mentioned, the bare specimens exposed to 20 kW/m² were the only scenario where the gusset plate provided some thermal protection to the wood beneath. This indicates that when a gusset plate is covered in soot and exposed to higher incident heat fluxes, the higher temperatures beneath the plate damage the surface of the wood more quickly.

The toothed specimens exposed to 15 kw/m² were cut into to investigate the depth of damage beneath the plate when compared to beneath the exposed wood. Figure 13 shows the char depths of the soot-covered specimen (left) and the bare specimen (right). The char depths within each particular specimen are relatively consistent no matter if beneath bare wood or the gusset plate. However, Figure 13 does show that the soot-covered specimen has a thicker char depth than the bare specimen for both beneath the gusset plate and exposed wood areas.

3.4. Connection Strength Assessment

Before the gusset plates were cut into after heat exposure testing, the gusset plate needed to be removed first. Two toothed specimens with bare and soot-covered gusset plates were chosen for realistic analysis. For the bare specimen (Figure 13, right image), significant effort was needed to wedge the gusset plate teeth out from the wood beneath indicating that the teeth’s biting force was not compromised. For the soot-covered specimen (Figure 13, left image), however, the gusset plate was able to be removed by hand with very little effort. The wood beneath the gusset plate was very brittle and broke away with little force indicating that the biting force was significantly decreased. On the surface of the soot-covered specimen, significant damage between the teeth holes is present, where the brittle wood broke away even with gentle removal of the plate by hand. A similar result was observed at the 20 kW/m² level due to the higher levels of visual damage observed in the specimens. At the 10 kW/m² level, this phenomenon may not be as noticeable as there was only minor damage to the specimens. On the surface of the bare specimen, the holes created by the teeth are still separate and obvious, despite significant effort in wedging the gusset plate free. As gusset plates are exposed to a sooty environment and exposed to high heat fluxes during a residential fire, a loss in connection strength leading to truss failure and potential collapse is a serious concern for responding firefighters.

3.5. Limitations

This study examines the effect of surface absorptivity and the presence of teeth of gusset plates on temperature development within the wood specimen. To replicate and control the exact thermal exposure, specimens were placed under an electrically controlled radiative heating source. This thermal environment can be different in real fire conditions as convective heating and direct flame contact on both gusset plate and wood surfaces are possible. The incident heat flux on a lightweight truss system can be much higher than 20 kW/m² in real fire conditions. In addition, the specimens evaluated were not under any physical loads such that any variations induced by physical loads such as gaps between the gusset plate and wood surface were not considered. Potential endothermic reactions within the wood were not considered for the temperature comparison as a clear endothermic effect was not observed beyond 100 °C; the endothermic reaction by dehydration was clearly observed in the moderate slope decrease in the temperature near 100 °C.

4. Conclusions

Firefighters are at risk of death or severe injury due to structural collapse, especially in residential homes. Sustainable engineered trusses, composed of lightweight wooden members connected by metal plates, commonly referred to as gusset plates, are found in most modern residential homes. Previous literature and consensus on the performance of gusset plates during a fire is not consistent and some disagreement exists on whether the gusset plate protects the wood beneath or damages it quickly. This experiment attempts to build upon the current knowledge of gusset plate heat transfer phenomena during a fire by exposing small-scale specimens to an external radiative heat source. Four configurations were tested to evaluate the effects of a surface coating, in this case soot, on radiation absorption and the effects of gusset plate teeth on heat conduction into the wood. The specimens were exposed to three different heat fluxes and temperature measurements in various locations within the specimen were recorded. The most impactful results from the experiment are listed below.

• A bare gusset plate provides some protection for the wood beneath at a 20 kW/m² external radiant heat.
• Soot-coating on the bare gusset plate increases radiation absorption and does not protect the wood beneath at 20 kW/m² with significant charring occurring.
• The presence of teeth has a minor effect on the internal wood temperature for the bare gusset plates for an external radiant heat exposure up to 20 kW/m² but is more significant for soot-covered specimens.
• Connection strength is quickly lost in the soot-covered specimens when compared to the bare specimens.

Gusset-plated trusses, commonly located in attics, may quickly become coated in soot and exposed to high temperatures and high heat fluxes. The presence of soot increases the radiation absorbed by the gusset plate, leading to a quicker and higher increase in temperatures beneath the plate. The increased temperatures promote thermal degradation of the wood beneath the gusset plate, which then leads to loss of biting force by the plate’s teeth. A loss of biting force can result in tooth pullout, and ultimately failure of the truss and collapse of the structure. This experiment only tested small-scale specimens without any structural load. While further research is necessary with structural loading, it is expected that a load may increase the rate of failure due to the increased forces acting on the gusset plate–wood connection.

Future research should evaluate large-scale gusset-plated trusses that can account for limitations of this experiment such as convective heating, being under a structural load, and evaluating failure rates of the wooden members versus gusset plate tooth withdrawal. Future research can also further investigate protective coatings of gusset plates such as carbon [23] or graphite [24] phenolic sphere sheeting, gypsum board [13], intumescent paint [30], or other unique protection methods on heat transfer or failure times of lightweight, gusset-plated trusses. These protection methods, if found to be effective, may be incorporated into building codes by respective authorities, required by insurance companies, or introduced as a safer alternative by truss manufacturers in the future to help prevent or delay the collapse of lightweight engineered trusses during a fire. We hope our research will increase the body of knowledge on structural collapse in sustainable lightweight construction, leading to further testing, protection efforts, and hopefully a decrease in firefighter injuries and fatalities due to structural collapse while supporting sustainable lightweight construction.