Effects of Postcure and Degradation in Wet Layup Carbon/Epoxy Composites Using Shear-Based Metrics

Abstract

Non-autoclave-cured wet layup composites are used extensively in applications ranging from civil and marine infrastructure to offshore components and in transmission power systems. In many of these applications the composites can be exposed to elevated temperatures for extended periods of time. While residual tensile characteristics have been used traditionally to assess the integrity of the composite after a thermal event/exposure, it is emphasized that fiber-dominated characteristics such as longitudinal tensile strength are not affected as much as those associated with shear. This paper reports on the investigation of shear related characteristics through off-axis and short-beam shear testing after exposure to temperatures between 66 °C and 260 °C for periods of time up to 72 h. It is shown that the use of shear test results in conjunction with tensile tests enables better assessment of the competing effects of postcure, which results in an increase in performance, and thermal degradation, which causes drops in performance. Off-axis-to-tensile strength and short-beam shear strength-to-tensile strength ratios are used to determine zones of influence and mechanisms. It is shown that temperatures up to 149 °C can lead to advantageous postcure related increases in performance whereas temperatures above 232 °C can lead to significant deterioration at time periods as low as 4 h. The use of shear tests is shown to provide data critical to performance integrity showing trends otherwise obscured by just the use of longitudinal tensile tests. A phenomenological model developed based on effects of the competing mechanisms and grouping based on phenomenon dominance and temperature regimes is shown to model data well providing a useful context for deign thresholds and determination of remaining structural integrity.

1. Introduction

Fiber-reinforced polymer (FRP) composite elements and components processed using non-autoclave manufacturing processes are finding increasing application in civil, marine, offshore, electrical transmission, and transportation infrastructure. In addition, these materials are also extremely effective in strengthening and retrofitting aging/deteriorating infrastructure elements thereby increasing overall service life. Due to the higher stiffness, and relative inertness to most environmental conditions, carbon fibers are often the preferred reinforcement. While the fibers themselves do not degrade in the presence of heat till about 500 °C when oxidation occurs [1,2], the resin and the fiber matrix bond are affected at significantly lower temperatures initiating at levels close to, and above, the glass transition temperature [3]. The overall response of polymer composites to elevated temperature can be divided into 4 primary stages. The first stage extends till just before the glass transition temperature T_g. Exposure to elevated temperature in this phase can result in an increase in both T_g and mechanical properties due to postcure reactions [4], wherein the exposure activates molecular mobility in the resin resulting in an increase in glass transition temperature [5]. Since a number of composite systems under consideration are cured under ambient- or moderate-temperature cure regimes there is potential for significant increase in T_g through this initial short-term exposure. The second stage is in the range of 150 to 300 °C, within which response is dominated by changes in the matrix [6] and at the fiber-matrix interface level. The effect in the initial phase of this stage results in a competition between posture and degradation [7,8]. Past this stage the temperature causes degradation in the polymer resulting in matrix cracking and weakening of the fiber-matrix interface [4,5] and further deterioration because of the mismatch in coefficients of thermal expansion of the fiber and resin which initiate cracks providing paths for oxygen to penetrate into the composite [9]. This results in oxidation of the resin and further growth of microcracks [6] which causes a decrease in characteristics of mechanical performance. At the upper end of this stage the temperature causes bonds in the polymer network to break resulting in thermal decomposition [10] and significant decreases in material characteristics after which the resin itself can char and ignite also resulting in mass loss [11,12].

From the perspective of operational logistics there is a need for a comprehensive understanding of materials response and residual properties at temperatures below the decomposition temperature. For example, engine components and other elements close to the exhaust can face normal operating temperatures as high as 150–300 °C [13]. High voltage overhead conductors and supports used in electrical power transmission have operational temperatures as high as 180 °C due to heat generated during transmission [14,15]. Offshore structures, and components, even those not engulfed in fire, but adjacent to it, can be subject to temperatures in this range for extended periods of time [16,17]. Components and equipment such as rods [18], risers [19] and tools used in downhole operations can have operating temperature requirements between 150 °C and 232 °C [20]. Temperatures around 200 °C are also expected to be faced by FRP strengthening systems used in the rehabilitation of concrete infrastructure protected by external insulation [21]. Similar temperatures are expected to be faced by composite structural liners used to rehabilitate deteriorating cooling water piping systems in nuclear power stations and other facilities.

The effects of fire, and elevated temperature exposure, on mechanical characteristics of composites have been reviewed by Firmo et al. [22], Singh and Sethi [23], Bazli and Abolfazli [24], Moritz and Mathys [25], and Bai and Keller [26], among others. Models have been reviewed by Moritz et al. [27] and further developed recently by Correia et al. [28], Li et al. [29], and Zhang [30], and applied to prediction of service life after exposure to elevated temperature by Li and Xian [31]. Most of the experiments and models reported to date have focused on the evaluation of tensile strength and modulus as an effect of temperature. It should be noted that in unidirectional composites the mechanical characteristics in the fiber direction are unlikely to show significant deterioration till there is microcracking in the matrix and fiber-matrix debonding, i.e., at temperature levels well above T_g. Hawileh et al. [32] reported that carbon fiber-reinforced polymer (CFRP) composites showed brittle fracture with drops in strength and stiffness of 30% and 15%, respectively, between 100 and 150 °C. Changes in failure mode were noted between 200 and 250 °C from fiber fracture to splitting due to loss of matrix adhesion resulting in drops of 28% and 42% in strength and stiffness, respectively, with levels increasing to 71% and 46%, respectively, at 300 °C. At this level there was a loss of resin due to burning and decomposition exposing fibers. A similar response was seen with glass fiber-reinforced polymer (GFRP) composites but the use of carbon/glass hybrids resulted in a minimal drop in modulus of 9% at 300 °C. Cao et al. [33] reported a 40% drop in tensile strength when CFRP sheets and hybrids were exposed to a temperature of 200 °C, while Wang et al. reported a 50% drop at 300 °C with very little further change till 400 °C [34], all of which were well above the T_g. Foster and Bisby [35] had reported that CFRP and GFRP showed good residual performance after exposure to elevated temperatures even exceeding T_g with large reductions only being noted at temperatures close to 4T_g.

In a FRP composite exposed to elevated temperature the weak links are the resin and the fiber–resin interface. As the glass transition temperature of a resin is approached, the resin begins to soften and initiates the transition from glassy to rubbery resulting in decreasing ability to transfer load and interfacial shear, not just between the fibers and resin, but also between layers of reinforcement. Thus, while the focus to date in the literature has largely been on tensile characteristics, primarily in the fiber direction, the critical area of concern from a performance integrity perspective should be shear and flexural response wherein interlaminar and intralaminar integrity and characteristics serve as the metrics of performance, and are affected to a great extent by matrix level deterioration. There have been limited studies in this direction. Liu et al. [36] reported that failure load in short term shear beam shear (SBS) tests decreased linearly with elevated temperature exposures of up to 130 °C with the rate of decrease being as high as 0.415 MPa/°C and 0.303 MPa/°C for 2 carbon epoxy systems tested. Earlier work by Lundahl and Kreiner [37] had also shown a drop of 30% in SBS strength exposed to 105 °C conditions. Nema et al. [38] reported a linear drop in peak flexural strength from 90 to 200 °C with an 80% decrease at 200 °C as compared to levels attained under ambient condition levels, although the decrease in modulus only occurred after 170 °C. Zavatta et al. [13] reported an increase in SBS strength followed by a period of relatively little change until 198 °C and then a steep decrease at 260 °C, which represents 145% of T_g, with drops of 71% and 89% after 24 and 72 h exposures, respectively. Kodur et al. [39], Bazli et al. [40], and Karbhari and Hong [41] all emphasized that shear related characteristics at elevated temperature should be an area of increased concern.

Acknowledging that there are few investigations of effects of thermal aging on interlaminar properties [38] this paper focuses on the elucidation of shear response of wet layup carbon/epoxy composites under conditions of elevated temperature exposure. The focus is on developing a better understanding of the stages of performance evolution due to initial progression of cure and the later deterioration of properties as well as the transition period of competition between these two phenomena, using test data and simple models to provide insight and predict response based on the dual effects of temperature, and time period, of exposure. This work is part of a larger study into developing a comprehensive understanding of the effects of extended periods of thermal aging on ambient temperature cured composite systems.

2. Materials and Test Conditions

2.1. Material System Details

A unidirectional carbon fabric of aerial weight of 644 g/m² using T700 carbon fibers in untwisted 12k tow was impregnated using the wet layup process with a difunctional Bisphenol A/epichlorohydrin derived liquid epoxy (Epon 828 based) using a polyetheramine-modified polyoxypropylenediamine curing agent to form composites of two-layer thickness impregnating each layer sequentially using manual roller-based pressure, without use of a vacuum bag, to achieve impregnation in the 0₂ and the 0/90 configurations. Panels were cured at room temperature, mimicking field conditions for seven days and 30% relative humidity after which they were postcured for 72 h at 60 °C to provide a uniform baseline of cure progression. Fiber mass fraction was determined using acid digestion procedures following ASTM D3171 [42] using specimens of nominal size of 25.4 mm × 25.4 mm × 2.3 mm to be 60% (i.e., a fiber volume fraction of 49.8%) with a standard deviation of 2 to 2.5% over all specimens. Void content was found to vary between 2 and 3%. Glass transition temperature, determined using differential scanning calorimetry, was noted to be 72.4 °C with a standard deviation of 1.89 °C.

2.2. Specimens and Test Methods

Given the focus on interfacial properties, two different test configurations were used for this investigation in addition to the standard unidirectional tensile test using ASTM D3039 [43]. In order to assess shear response the ±45 configuration recommended by Pindara and Herakovich [44] was used following ASTM D3518 [45] with specimens cut from the 0/90 panels at the appropriate orientation to acquire the required ± 45 orientation with a length of 254 mm, a width of 12.7 mm. and a gauge length of 140 mm, with a test displacement rate of 1.27 mm/minute. When a ± 45 laminate is loaded in uniaxial tension a biaxial state of stress is induced within each lamina with the maximum off-axis shear strength determined as:

$${\tau }_{12}^{max}= {\displaystyle \frac{P_{max}}{2A}}$$

(1)

where P_max is the maximum load at failure and A is a cross-sectional area. Since any weakening effect of temperature on resin softening will lead to layer separation and sliding these specimens are expected to provide greater emphasis on the deteriorative phenomena. Overall shear response was also characterized using the short-beam shear (SBS) test following ASTM D2344 [46] with a specimen width of 6 mm, a length to thickness ratio of 6:1, and a crosshead speed of 1 mm/minute with short-beam shear strength determined as:

$${\sigma }_{SBS}= {\displaystyle \frac{0.75 P}{bh}}$$

(2)

where P is the failure load and b and h are the specimen width and thickness, respectively. At each exposure condition tests were conducted on three off-axis specimens, and five specimens each for tension and short-beam-shear. In addition to tests conducted under ambient conditions (nominally 23 °C and 30% relative humidity), specimens were tested after exposure to temperatures between 150 °F and 500 °F at intervals of 50 °F, i.e., at 66 °C, 93 °C, 121 °C, 149 °C, 177 °C, 204 °C, 232 °C, and 260 °C for periods of time ranging from one hour to 72 h. Specimens were placed in furnaces at the specified temperature and were tested after exposure for the specified period of time and cooling back to 23 °C. Further details on procedures and individual residual characteristics as well as microscopy results were previously reported in [47].

Prior to discussion of test results, it is important to contrast the two shear related test methods used and clarify the areas of overlap and differences. The off-axis shear test uses a ±45° fiber orientation resulting in the dominant shear response being matrix dominated with the ability for fiber rotation and capturing effects of Poisson constraint due to layup. It is designed to induce as close to a pure in-plane shear state in the composite test specimen as possible. In comparison, the SBS test is conducted in the flexural (3-point bending) mode with fibers bridging the span. The use of a small span to thickness ratio induces high interlaminar shear stresses near the mid plane induced through out-of-plane loading resulting in failure in a mode II interlaminar shear mechanism. Due to the matrix dominance of this mode on the response to in-plane shear the off-axis test is a good indicator of the effects of postcure since the thermal aging-based progression of polymerization affects the crosslink density and hence the stiffness of the matrix indicated through changes in in-plane shear modulus. Thermal aging also leads to microcracking, fiber matrix debonding, and interlaminar delamination, all of which are better seen through the SBS test, which is also sensitive to the onset of fiber-matrix debonding and matrix embrittlement. Thus, the off-axis shear test is better for in-plane property sensitivity and detection of the effects and extent of postcure, while the SBS test is a better indicator of interlaminar strength and aging-related degradation. A qualitative comparison of the two tests based on six key criteria is shown schematically in Figure 1. On this basis, the assessment of results from both tests is expected to yield a deeper level of insight into the effects of thermal aging on response with special focus on postcure and deteriorative effects of thermal aging.

3. Results and Discussion

3.1. Effects of Thermal Aging

Overall results for changes in off-axis shear strength, SBS strength, and longitudinal tensile strength, as a function of temperature, and time, of exposure are shown in Figure 2a–c, respectively.

As can be seen from Figure 2a, except for specimens under ambient conditions where the effect of residual postcure is expected, the off-axis shear test shows weakening due to the softening of resin which causes separation of layers and sliding of fibers. There is a drop in the first 2–4 h of exposure to elevated temperature followed by a much slower asymptote to the final value. At the two highest temperatures of 232 °C and 260 °C, however, the rapid, albeit slightly slower, deterioration continues with levels at 260 °C after 16 h being such as to invalidate the tests themselves due to heat induced twisting and failure by separation. In comparison, the SBS strength (Figure 2b), which is an indicator of inter-/intra-laminar response, clearly shows effects of posture with all specimens increasing in performance over the first 2–4 h followed by an asymptotic response with slow decreases, which, however, remain above the initial unexposed levels. The exceptions to this are at the two highest temperatures wherein a significant decrease in trend is noted after 24 h at 232 °C, and after 4 h at 260 °C. The longitudinal tension results (Figure 2c) also show the initial postcure results in an increase over the first 2–4 h followed at lower temperatures by a slight asymptote and then a gradual decrease. Exposure at the highest temperature, 260 °C, results in a rapid decrease after 4 h of exposure.

The overall trends and interactions between the mechanisms of post cure, transition, and deterioration can be viewed clearly using 3D plots as in Figure 3a–c for off-axis shear strength, SBS strength, and tensile strength, respectively.

To enable ease of comparison across all three test sets, the results are normalized by the respective unaged/unexposed values (57.51 MPa for off-axis strength, 39.45 MPa for SBS strength, and 502.26 MPa for longitudinal tensile strength). As can be seen from Figure 3a the overall trend in off-axis strength tracks with increases in temperature and time of exposure with a sharp decrease observed at temperatures higher than 149 °C with greater effect of both time and temperature. The degradation surface is steep at this point although it does exhibit a transition zone that is associated with matrix softening and weakening of the fiber-matrix bond. In contrast the SBS strength as shown in Figure 3b indicates an increase initially, with the increasing trend persisting even as long as 60 h of exposure at temperatures lower than 149 °C. Deterioration below initial unexposed levels is only noticed at, and above, 204 °C, and at longer periods of exposure. The uneven surface between the initial increase and the later steep decrease shows that the SBS strength is extremely sensitive to changes in both matrix stiffness and interlaminar adhesion which effectively occurs in the relatively larger resin zones between layers of fabric. Both these show the effects of competition between the increase in cross-linking due to postcure mechanisms and mechanisms of matrix microcracking and fiber-matrix debonding. The steep drop at the higher temperatures and longer periods of exposure is due to mechanisms of thermal oxidation and resin embrittlement. As can be expected with the longitudinal fiber-dominated characteristic of a unidirectional composites the degradation surface for tensile strength as seen in Figure 3c shows local variation but is fairly asymptotic and flat as compared to the other two characteristics suggesting dominance of the fibers with drops being due to transverse matrix cracks and fiber-matrix debonding. Based on the overall response, 3 zones can be generically identified as related to time-temperature interaction. Zone I is posture dominant. Zone II is transitionary and is represented by a plateau with minor local variation wherein strength characteristics largely stabilize Zone III represents the dominance of deterioration mechanisms such as oxidation, extensive microcracking and debonding, chain scission, and layer separation. The extent of the three zones is described in

Table 1. Overall Summary of Strength Response by Zones.

Characteristic	Zone I (Postcure Dominant)	Zone II (Transition)	Zone III (Deterioration Dominant)
Off-axis Strength	≤93 °C, ≤18 h	93–149 °C, 6–36 h	≥149 °C, ≥12 h
SBS Strength	≤149 °C, ≤60 h	149–204 °C, 24–48 h	≥204 °C, ≥24 h
Tensile Strength	≤121 °C, ≤36 h	121–177 °C, 12–48 h	≥177 °C, ≥24 h

.

In general, the transition threshold occurs at temperatures of 121 °C, 177–204 °C, and 149–177 °C, for off axis shear strength, SBS strength, and longitudinal tensile strength, respectively, with the first being highly sensitive to matrix deterioration and degradation of interfacial shear, and the second being sensitive to matrix embrittlement and stiffening only at higher temperatures and longer periods of exposure.

These comparisons help establish trends and thresholds and can be summarized schematically as in Figure 4, wherein the transitionary zone is seen to range over a large set of exposure temperatures from 93 °C to 204 °C and from 6 h to 48 h. While the schematic provides areas of dominance and a transition there is still a need to add further clarity to the overlap between the mechanisms of thermally activated postcure and deterioration.

This can be more clearly seen through the comparison of the percentage change in the three characteristics from their baseline values for the case of exposure for 72 h at all temperatures and the case of exposure at 149 °C (which is within the transition region) in Figure 5a,b, respectively. As seen from Figure 5a, initially, the trends shown by the change in tensile strength and SBS strength (increasing with temperature) are opposite to those of off-axis strength. This, however, changes at 149 °C after which all three show decreasing trends. Figure 5b, in contrast shows a decrease in the levels of both off-axis strength and tensile strength as a function of time of exposure at 149 °C with the SBS strength showing an initial slow decrease with a jump at the 8 h level, followed by minor variation between 24 and 72 h. Both these emphasize the competition between different mechanisms at play and the ability (or inability) to comprehensively assess changes through a single test, necessitating further study both for the purposes of better understanding of materials level changes, and in developing design thresholds for performance.

Most civil infrastructure design standards/guidelines dealing with the aspects related to thermal aging and exposure of FRP composites to in-service related temperatures focus primarily on retention of tensile strength and modulus with restrictions on use only being as related to glass transition temperature. This stems from their focus on fiber-dominated response modes for thresholds of mechanical performance and operational limits as related to temperature. The latter are generically set by selection of levels conservatively distant from the glass transition temperature to avoid effects of thermal change. For example, the guideline for externally bonded systems developed by the American Concrete Institute—ACI440.2R [48] only specifies factors for tensile properties, whereas the Eurocode [49,50] focuses on residual tensile strength and bond to the concrete surface but not the shear properties of the FRP. In contrast, both Federation internationale du beton, fib [51] and the Japan Society of Civil Engineers (JSCE) [52] emphasize tensile properties but also acknowledge the necessity of considering degradation of matrix dominated properties. From results discussed in this investigation and as reported previously [38,47,53,54,55] it is clear that even in cases where there is good retention in tensile characteristics the deterioration at the interfacial and inter/intra-laminar levels could be much higher [56] potentially resulting in misleading [57] and non-conservative estimates of overall response and service life, reiterating the issues related to shear characteristics raised by Bazli et al. [40]. Hong et al. [56] emphasize that the sole use of the longitudinal tensile test is “unsuitable” for adequate evaluation of degradation of CFRP composites.

3.2. Assessment Through Strength Ratios

While the assessment of mechanical properties can provide a direct metric of performance under specific conditions of exposure, they often fail to capture interactions between different failure modes, and the relative susceptibility of constituents (fiber, matrix, interface) and mechanisms, to the extent of exposure. In this context, the consideration of strength ratios, specifically off-axis shear-to-tensile strength, and SBS shear-to-tensile strength, offer a powerful approach to provide a more nuanced view of interactions and competing mechanisms. It should be noted that tensile strength in unidirectional composites is predominantly governed by fiber properties and fiber alignment. By expressing shear strength (SBS or off-axis) as a ratio to tensile strength, the influence of fiber degradation is intrinsically normalized enabling the ratio to isolate degradation phenomena that more specifically impact matrix behavior, fiber-matrix interfacial adhesion, and intralaminar strength, which are often more sensitive to thermal aging. Since carbon fibers are not affected within the temperature range considered, tensile degradation proceeds slowly due to the stability of fibers whereas matrix dominated shear properties exhibit nonlinear or staged responses due to postcuring effects, plasticization or thermal softening as the exposure temperatures approach T_g, and chemical degradation (oxidation, chain scission), and embrittlement due to elevated temperature exposure. These complex transitions can be hidden when results related only to shear strength are analyzed, but become evident when these results are interpreted relative to tensile strength.

Off-axis-to-tensile strength ratios (expressed as a percentage) are shown in Figure 6. Exposure at all temperature levels indicate an initial drop at shorter periods of exposure indicating that the off-axis shear strength deteriorates more rapidly than the tensile strength. It is noted that the resin in the ±45 layup carries a significant portion of the shear load, and this results in a reduction in strength due to matrix softening. This weakens the fiber-matrix interface and can also result in microcracking and initial relaxation of internal stresses that may exist due to cure shrinkage thereby further disrupting the off-axis response.

An increase in time of exposure is noted to cause an increase in the ratio at low temperatures (66 °C, 93 °C, 121 °C and 149 °C) and asymptotic response with minor increases, or variation, at the higher levels of 177 °C and 204 °C, with continued drops to the two highest temperatures. The increase after the first few hours is related to the effects of postcure which can also be noted by the transition in response between 121 °C and 149 °C. At 121 °C there is a rapid increase in ratio after 8 h reflecting a more optimal posture regime with the subsequent decline in overall values of the ratio between 149 °C and 204 °C indicating irreversible degradation at the interlaminar and fiber-matrix interface levels. At the 66 °C and 93 °C exposure temperatures there is insufficient thermal energy to result in postcure effects dominating over those of matrix softening with the requisite levels only being reached at 121 °C after 8 h of exposure. At this point there is sufficient energy to promote significant cross linking and interface densification to negate effects of matrix softening. The drop in ratios between 121 °C and 149 °C signifies that continued thermal aging again surpasses the benefits of postcure although the level is not great since asymptotic response is seen in the ratio suggesting that the region between 121 °C and 149 °C, as identified earlier, is the transition regime between postcure related maximization of crosslinking and matrix cohesion and interfacial strength, and the degradation of the resin along with the loss of interfacial and inter-/intra-laminar shear transfer. Thus, the behavioral regimes can be identified as detailed in

Table 2. Identification of Behavioral Regimes Based on Off-Axis Shear-to-Tensile Strength Ratios.

Regime	Temperature Range	Trend	Description
1	Low (66–121 °C)	Ratio decreases initially, then increases and stabilizes	Competing mechanisms of initial softening followed by postcure and interfacial strengthening
2	Mid (149–204 °C)	Ratio decreases then stabilizes	Initiation of deterioration dominance with later stabilization in a semi-equilibrium state
3	High (232–260 °C)	Ratio decreases continuously	Dominance of irreversible degradation through resin decomposition, fiber-matrix debonding, and matrix microcracking, with char formation at longer periods of exposure

and as shown schematically in Figure 7.

There is clearly a nonlinear interaction between time and temperature, with time being the dominant factor controlling degradation initially at low temperatures, and degradation being temperature activated and time-accelerated at elevated temperatures suggesting behavior that is consistent with thermally activated kinetics of the form:

$$D\left(T,t\right)=D_0 e^{{\displaystyle \frac{-E_a}{RT}}}·t^n$$

(3)

where D is a metric of degradation, in this case the off-axis-to-tensile strength ratio, D₀ is a temperature independent constant, T is the temperature in degrees Kelvin, t is time and E_a is the activation energy. However, the existence of three zones as shown in Figure 7 would suggest that it is unrealistic to expect a single activation energy across the competing time-temperature space and a model would necessitate a two- or three-stage mechanistic formulation.

While the off-axis test reflects in-plane shear performance, which is particularly sensitive to changes in matrix shear stiffness and intra-layer damage the SBS test is related to inter- and intra-laminar shear performance and is strongly influenced by resin toughness and through thickness effects. It is also a comparatively easy and fast test to perform and is used as a means of rapid quality assessment. Thus, the addition of the SBS-to-tensile strength ratio, in combination with the previously described off axis-to-tensile strength ratio, enables further insight into shear response differentiating between through-thickness and within ply-plane (off-axis) response and mechanisms. Together, these could amplify effects of mechanisms such as postcure, interface and intralaminar weakening as well as matrix level changes from initial softening and microcracking, to delayed deterioration, thus providing a more comprehensive assessment of durability. The SBS strength-to-tensile strength ratios (expressed as a percentage) as a function of temperature, and time, of exposure are shown in Figure 8.

Except for the ambient exposure which results in an initial increase due to greater posture effects on SBS strength, all others show an initial drop followed by regions of stabilization and partial recovery. At lower temperatures between 66 °C and 121 °C the drop is for the first 2–4 h indicating the competition between matrix softening and postcure with moderate/negligible micro-cracking and irreversible matrix level damage. At temperatures between 149 °C and 204 °C the initial drop is followed almost immediately by a slow increase in ratio indicating the predominance of postcure at the inter-/intra-laminar level at rates faster than that which could be noted from a longitudinal tensile test since the effect of postcure in a fiber-dominated characteristic is lower. At 232 °C the ratio remains high initially with a slow decrease in the first 8 h followed by a slight increase between 8 and 24 h after which the values drop due to damage accumulation and the increasing effects of surface micro-cracking and fiber-matrix debonding. At the highest level of 260 °C the drop initiates right from the beginning with the rate increasing rapidly after 8 h and continuing till about the 24 h period after which the rate decreases. This is clear indication of matrix embrittlement and interface failure which is far more pronounced through SBS strength. It is interesting to note that just as with the off-axis-to- tensile strength ratios there is a significant change in response between exposures at 121 °C and 149 °C. In this case the values at 121 °C do not drop as much as those at lower temperatures in the first four hours, but the recovery is far steeper and results in ratios higher than that of the initial state, indicating substantial progression of polymerization and stiffening of the matrix due to postcure. The ratios reach levels of about 9% at the 48 h levels from an unexposed level of 7.85%. In comparison at 66 °C the level after 48 h is 7.67%, i.e., a net decrease of 0.18%. Thus, the effect of postcure is seen to be the maximum in this temperature range. At the higher exposure temperature of 149 °C, the level of increase is significantly less reaching only 7.47%, still below the initial value. The overall interplay between mechanisms can be shown schematically in Figure 9, indicating a more complex level of competition than that described earlier by the off axis-to-tensile strength ratio in Figure 7, suggesting that it could again be modeled using a two- or multi-stage model of the following type:

$${\sigma }_t={\sigma }_0+ a\left(1-e^{-k_{1}t}\right)-b\left(1-e^{-k_{2}t}\right)$$

(4)

where σ₀ is the initial SBS strength, a & k₁, are parameters representing posture with higher temperatures yielding higher values of a & k₁ signifying faster rates and levels of postcure. Similarly, b & k₂ are parameters representing thermal degradation.

It is important to recognize the differences in the two test methods. Off-axis tests are based on mechanisms related to in-plane shear in the resin and the interface resulting in a relatively uniform distribution of shear while the SBS test is shear dominated at the interlaminar level due to the short span with high shear gradients at supports and under the load nose. It is useful to note that the former has moderate sensitivity to changes at the resin level and moderate-high sensitivity at the interface level whereas the latter shows high sensitivity to both. The combination thus provides a more comprehensive assessment of mechanisms and their progression as a function of temperature. Based on the combination of trends from the two ratios the overall exposure regime can be divided into 3 regions as listed in

Table 3. Summary of Temperature Regimes and Mechanisms.

Region	Temperature Regime	Off-Axis-to-Tensile Strength Ratio	SBS-to-Tensile Strength Ratio	Key Aspects
1	66–121 °C	Sharp dip followed by a strong recovery	Milder dip, followed by stabilization, and recovery	Postcure improves performance at the matrix and interface levels enhancing off-axis and tensile performance to a greater extent than SBS strength, i.e., in-plane properties are improved more than interlaminar properties. SBS performance is more sensitive to resin softening and local deterioration
2	149–204 °C	Moderate decrease with the level of deterioration increasing with temperature followed by an asymptotic region	Milder dip with level of drop decreasing with temperature followed by a slow recovery	Interface and matrix level deterioration dominates degradation in in-plane directions. SBS retention is due to multi-axial stress redistribution
3	232–260 °C	Continuous decline with failure prior to the full period of exposure at the highest temperature	Delayed decline with a level of stability, or slower decline, after 24–48 h following an initial more rapid period	Dominance of irreversible mechanisms of resin degradation and delamination/layer separation with off-axis mechanisms showing greater sensitivity to temperature, and time, of exposure

.

3.3. Modeling of SBS Response

Based on the previous discussion the assessment of more than one characteristic is essential for the comprehensive understanding of materials response to periods of thermal aging. While the off-axis test provides important information it lacks the level of sensitivity to thermal aging and interlaminar shear that are necessary. Since the SBS test is also easier to conduct and is widely used it is the more effective test to use and hence we focus on the evolution of SBS strength for the development of a model. As a first approximation the two-staged phenomenological model suggested in Equation (4) can be used with parameters a, b, k₁ and k₂ being determined through a best fit regression with values as listed in

Table 4. Kinetic Parameters for the 2-Stage Model.

Temperature of Exposure (°C)	a	k1	b	k2
23	4.8298	1.7164	2.3093	0.9999
66	6.6072	0.8078	0.5405	0.0200
93	8.2506	0.4674	1.6558	0.0947
121	6.7081	0.8290	0.0000	0.0000
149	7.1895	3.0200	0.0359	0.0000
177	6.3215	9.9999	99.9999	0.0002
204	6.2393	3.7743	99.9999	0.0003
232	9.4483	0.6735	53.3874	0.0030
260	65.3800	0.3174	99.9999	0.1214

, where a and k₁ represent postcure amplitude and rate, and b and k₂ represent the degradation amplitude and rate. Near zero values for b and k₂ indicate postcure dominance.

As shown in Table 4 the amplitude of postcure, a, increases with temperature till 93 °C, after which there is a level of variation between 121 °C and 204 °C, and then a rise to a maximum at 260 °C which matches the attainment of the highest value of SBS strength at that temperature. The variation, however, also suggests a level of competition in the intermediate range of temperatures. The rate parameter, k₁,increases significantly between 121 °C and 177 °C, signifying postcure dominance in terms of rates. It is of interest to note that within this range the value of k₂ is either zero or close to zero emphasizing the dominance of postcure in this band over deterioration. The high values of b after 177 °C indicate the initiation of deterioration with the highest rate of k₂ at the highest temperature indicating deterioration dominance in the later stages after attainment of a postcure induced peak. At levels higher than 149 °C mechanisms of degradation start to overcome effects of post cure in the later stages. A comparison of predicted trends with experimental values is shown in Figure 10. Since the overall response across temperature levels as shown in Figure 2b is clustered with values within a tight band only a few temperature regimes are shown to enable a clear assessment of the level of correspondence. As can be seen while the model provides reasonable fits it does not, as such, provide detailed insight at a level higher than interpretive and hence the development of a more involved model that differentiates between phases and mechanisms is desirable.

Figure 2b indicates that overall response of time, and temperature, of exposure on SBS strength can be divided into two phases, the first of which shows an increase from the unexposed level to a peak, and the second following this shows either a plateau or a decrease from the peak level at rates depending on the temperature of exposure. The initial increase is related to the direct effects of postcure as reported earlier by Zavatta et al. with an increase in the rate of SBS gain decreasing as time and temperature of exposure temperature increase [13]. As noted, earlier levels of SBS strength are affected by the competition between the two mechanisms of postcure and thermally induced deterioration with the time to attain the peak level of SBS strength decreasing with temperature of exposure from 24 h at 66 °C to 16 h at 93 °C. This is followed by a constant level of eight hours between 121 °C and 204 °C, after which it decreases further to 4 h at the two highest temperatures with the highest value of SBS strength of 50.13 MPa being attained after four hours of exposure at 260 °C. This level corresponds to a 27.1% increase due to postcure. It is thus of interest to not just compare the effect of temperature on the maximum SBS strength attained but also the rate of increase over the 72 h period as well as the rate of decrease after attainment of the peak level as shown in Figure 11.

A significant correlation can be seen between the two rates of change in SBS over the full period of exposure and till the time of attainment of the peak SBS strength related to the effect of postcure. This emphasizes the effect of the competition between the two primary mechanisms and thus the designation of two zones of responses as pre- and post-peak, wherein postcure mechanisms dominate in the first zone and deterioration related mechanisms dominate in the second. This was discussed earlier by Rothenhausler and Ruckdaeschel for epoxy resins [58]. It can be surmised that absent the effect of deteriorative mechanisms the SBS strength would attain a peak level related to full polymerization and then maintain that level through additional periods of exposure. Ideally, taking deterioration into account, the actual level of SBS strength determined at any point of time, t, due to exposure temperature, T, is the difference between the peak level and the amount of deterioration. The competition between mechanisms of postcure and deterioration would result in the idealized peak level not being attained with response deviating from that of the postcure evolution as shown schematically in Figure 12.

In order to predict overall response a number of aspects need to be considered including: (1) determination of evolution of post cure to a hypothetical maximum performance level based on temperature of exposure, (2) determination of a point in time when effects of deterioration cause deviation of response from the idealized postcure curve, and (3) determination of post peak response where the SBS performance is lower than the maximum possible postcure induced performance and is determined by the level of deterioration at every point in time at the given temperature of exposure. Using the maximum SBS strength attained of 50.13 MPa as the idealized peak level of SBS strength in conjunction with the initial data for SBS evolution with temperature and time of exposure in the pre-peak region, the time taken to reach the level of 50.13 MPa, t_peak, assuming no deterioration, can be estimated using a hyperbolic function of the following type:

$$t_{peak}= {\displaystyle \frac{c}{d+T}}$$

(5)

where c (= 1320.20) and d (= −48.80) are constants for the specific set of materials used herein and T is the temperature of exposure in degrees Celsius. A comparison of predictions using Equation (5) with direct extrapolation of the rising response SBS strength from experiments is shown in Figure 13 indicating good correlation.

The increase in SBS strength through time at a given temperature was determined through a power law of the following form:

$${\sigma }_t= {\sigma }_0+g{t}^h$$

(6)

where σ_t and σ₀ are the SBS strengths at time = 0 (unexposed) and time = t, respectively, g is a scaling constant and h is the exponent that relates the kinetics of post-cure. An iterative least squares analysis was used to determine values of g and h, as reported in

Table 5. Kinetic Parameters for Equation (6).

Temperature of Exposure (°C)	g	h
66	3.25	0.27
93	3.62	0.32
121	3.62	0.38
149	4.05	0.38
177	4.06	0.38
204	4.34	0.38
232	4.4	0.49
260	4.4	0.74

, by setting values found at 66 °C and 260 °C as the lower and upper bounds and then resetting bounds at each increment of temperature.

It can be seen from Table 5 that the kinetics of postcure, as represented by h, increase with temperature between 66 °C and 121 °C and then attain a level of constancy between 121 °C and 204 °C, followed by a sharp increase. This follows the trends described earlier in Figure 11, suggesting a nonlinear set of reaction mechanisms as described in the earlier section on strength ratios wherein the materials response is a result of complex interactions, and competition, between thermally activated posture and deterioration kinetics. Based on the commonality of trends the overall response can be divided into 3 phases with the first being from 66 °C to 121 °C showing steady increment in posture effects, and the second being between 121 °C and 204 °C showing minor variations and almost balanced kinetics between posture and deterioration. The end of this phase indicates the onset of deteriorative increases, leading to the third phase from 204 °C to 260 °C which represents dominance of deterioration. Equation (6) can then be modified using Table 4 to develop a best fit equation for each of the three phases as a function of time and temperature as follows:

$$\begin{gathered} \sigma =39.45+1.246 ·T^{0.231}·t^{0.002T+0.137}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\mathrm{f}\mathrm{o}\mathrm{r} 66 °\mathrm{C} \le \mathrm{T} < 121 °\mathrm{C} \\ \sigma =39.45+1.246·T^{0.231}·t^{0.387}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em} \hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\mathrm{f}\mathrm{o}\mathrm{r} 121 °\mathrm{C} \le \mathrm{T} < 204 °\mathrm{C} \\ \sigma =39.45+1.246 ·T^{0.231}·t^{0.0064T-0.955}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\mathrm{f}\mathrm{o}\mathrm{r} 204 °\mathrm{C} \le \mathrm{T} \le 260 °\mathrm{C} \end{gathered}$$

(7)

It should be emphasized that Equation (7) describes response that is initially postcure dominated assuming that the peak value of 50.13 MPa is attained through exposure at each temperature level. As depicted by Figure 12 there is a deviation from this idealized response due to the increasing effects of thermally induced deterioration. While both sets of mechanisms exist on the response curve the point of deviation represents the point where deteriorative mechanisms overcome the effects of ongoing cure progression. The point of deviation, i.e., the point in time where the actual SBS strength response is noted to deviate from the idealized curve predicted by Equation (7) can be expressed as:

$$t_{deviation}=m{T}^n$$

(8)

where m and n are 716.17 and −0.916 for this specific material. A comparison of predicted values with those determined experimentally shows an R² of 0.94, indicating good correlation. It should be noted that the format used in these equations follows that traditionally used to assess polymer relaxation. The use of Equation (8) in conjunction with Equation (7) then provides an estimate of the time at which SBS strength response deviates from the idealized response based on cure progression, and the actual value of the SBS strength at this point. A comparison of experimental values (using interpolation between points as appropriate) and those provided through Equation (8) for time and Equation (7) for SBS strength are given in Figure 14 showing good correspondence. For purposes of comparison since experimental data was interpolated between times of exposure the standard deviations of points closet to the point of deviation are shown for the experimental data. It is noted that results from Equation (7) are well within the standard deviation with the maximum percentage difference between the data being 2.49% with a majority being below 0.5%.

As can be seen from Figure 2b post peak response follows one of three different modes depending on temperature of exposure with lower temperatures resulting in an increase in SBS strength, and the highest temperatures resulting in a decrease with the intermediate range showing an increase followed by a gradual decrease, in line with trends also reported by others earlier [47,54,59]. Given this, and continuing with the empirical model format from before, the post-peak response can be expressed as a two-staged equation of the following form:

$$\sigma = f_1\left(T\right)t+ f_2\left(T\right)$$

(9)

where f₁ and f₂ are functions of temperature of exposure and t is the time of exposure. Consideration of the commonality of response trends for sets of temperature regimes as identified earlier leads to the functions as shown in

Table 6. Two-Stage Functions for Post-Peak Response.

Group	${\mathit{f}}_1\left(\mathit{T}\right)$	${\mathit{f}}_2\left(\mathit{T}\right)$
1: 66–177 °C	$\left(-1.39\ast {10}^{-5}\ast T^2\right)+\left(3.07\ast {10}^{-3}\ast T\right)-0.161$	$\left(5.83\ast {10}^{-4}\ast T^2\right)-\left(1.4\ast {10}^{-1}\ast T\right)+7.05$
2: 177–232 °C	$\left(-3.84\ast {10}^{-5}\ast T^2\right)+\left(1.42\ast {10}^{-2}\ast T\right)-1.36$	$\left(4.29\ast {10}^{-4}\ast T^2\right)-\left(1.69\ast {10}^{-1}\ast T\right)+16.9$
3: 232–260 °C	$\left(-1.51\ast {10}^{-2}\ast T\right)+3.36$	$\left(-4.55\ast {10}^{-1}\ast T\right)+106$

.

The inclusion of the quadratic term for temperature highlights the complex dependencies of SBS strength wherein long-term exposure results in deterioration with rates depending on both time and temperature [13,38,47,60,61]. A comparison of results through the use of Equations (7) and (9), in conjunction with Equation (8) to determine the point of deviation, with experimental results shown in Figure 15.

It can be seen that the simplistic empirical model provides reasonable estimates for overall response over the sets of time and temperature investigated with the exception of the highest temperature where the model does not capture the steep decrease followed by the slower asymptote. It should be emphasized that this model is not a detailed mechanistic description but rather is intended to be a first order approximation to enable predictive design capabilities within the ranges already investigated and to highlight the modes, and overlap of the mechanisms, of posture and deterioration due to thermal aging. These could also be used to assess design values and structural integrity limits based on predetermined performance level thresholds.

Pricop et al. [62] cite a factor of safety of 3.5 to 4 for pressure vessels and offshore structures. Using the upper value of 4 provides a design threshold of 9.86 MPa (i.e., 0.25 * 39.45 MPa). Thus, at temperatures of 165 °C and 220 °C the model indicates exposure times of 1140 h and 430 h, respectively. These provide engineers and means of estimating the time for which exposure can be reasonably expected to be safely taken by the component/structure. Copello et al. [63] discuss the use of reserve strength ratio (RSR) values for platforms, and Stacey and Sharp [64] cite a required RSR level of 1.8 to 2.5. Assuming a value of 1.9, as an example, use of the model predicts that the threshold would be attained after 665, 565, 380, and 240 h of exposure to temperatures of 170 °C, 190 °C, 210 °C and 230 °C, respectively, again providing useful information for engineers, inspectors, and designers. While the factors used above are not specific to composites, their use helps illustrate the value of the model as a first approximation under specific exposure conditions

4. Summary and Conclusions

The current study provides additional insight into the response of wet layup composites to extended periods of elevated temperature exposure such as would be expected under operating conditions or through the proximity to fire while still being outside the zone where immediate charring could occur. The traditionally used mechanical characteristic of longitudinal tensile strength is a fiber-dominated property and hence remains unaffected even at temperatures exceeding the glass transition temperature leading to inadvertent conclusions regarding the post-exposure integrity of the composite. Both the off-axis and SBS test results reveal a more complex interplay between postcure and degradation mechanisms. The off-axis shear test, which is governed by in-plane matrix shear, exhibits rapid declines in strength at elevated temperatures, emphasizing its sensitivity to resin softening and interfacial debonding. The SBS strength test highlights the initial effect of postcure through increases in performance transitioning to a stable regime followed by deterioration through microcracking and inter-/intralaminar weakening. The results discussed in this paper provide a deeper understanding of the effects of competing effects of mechanisms of postcure, which result in further progression of polymerization resulting in enhancement of both tensile and shear characteristics, and the effects of thermally induced deterioration at the level of the resin and fiber-matrix interface, and on inter- and intra-laminar response.

It is shown that strength ratios provide a normalized and mechanistically relevant method for isolating matrix and interface effects. They also assist in highlighting critical transitions across three regimes, i.e., lower temperatures where postcure dominates, improving performance, intermediate temperature regimes which show the competition between thermally induced postcure and deterioration and higher temperatures due to which irreversible matrix and interface level degradation dominate.

The development of simple models reveals that while two-stage exponential models can effectively fit SBS trends, a more robust three-phase predictive model incorporating postcure kinetics, deviation onset, and degradation allows for more precise estimation of overall performance. Importantly, these models can be used in engineering practice to define conservative operational limits and residual strength thresholds under realistic exposure conditions. This investigation emphasizes a pressing need to revise design standards and durability predictions for composite systems to incorporate interlaminar and matrix-dominated properties. Current code practices that focus solely on fiber-dominated tensile behavior may significantly overestimate remaining structural integrity after periods of thermal exposure. It is shown that the inclusion of matrix-sensitive metrics and degradation-informed modeling provides a pathway to safer and more resilient applications of FRPs in elevated temperature environments.