Thermo-mechanical properties of RTM-made carbon fibre/polyimide composite attaching collar under transient heating

Shengd JIANG,Chuyng LUO,*,Peng ZHANG,Jinwen BAO,Peipei CAI,Xufeng XIA

a Shanghai High Performance Fibers and Composites Center (Province-Ministry Joint), Center for Civil Aviation Composites, Donghua University, Shanghai 201620, China

b AVIC Composite Technology Center, Science and Technology on Advanced Composites Laboratory, Beijing 101300, China

c Luoyang Optoelectro Technology Development Center, Luoyang 471009, China

KEYWORDS Polyimide composites;Progressive damage model;Resin transfer moulding;Thermo-mechanical properties;Transient heating

Abstract This study focuses on the thermo-mechanical properties of Carbon Fibre/Polyimide Composite (CFPC) attaching collars under transient heating.The CFPC attaching collars were fabricated by a high-temperature resin transfer moulding process, and their thermo-mechanical properties under the conditions of simultaneous transient heating and bending load were investigated.The results show that the attaching collar tends to fail at 118%of the limit load.The failure mode includes the fracture of the connecting screws,local extrusion damage of the hole edges,and slight ablation damage at the outer plies.And there is no observable residual deformation in the composite attaching collar.Furthermore,considering that the material properties vary with temperature,a progressive damage model based on the sequential thermo-mechanical coupling method was established to study the failure mechanism of the attaching collar.Finally,the damage factor of the CFPC was calculated to assess the safety status of the attaching collar.The results show that the primary damage modes of the composite attaching collar are intralaminar failure,which mainly occurs at the heat insulation layer and the hole edges,and these slightly affect the structural bearing capacity.A good correlation between the experiment and FEA is obtained.The test methods and analysis models proposed contribute to the safety assessment of composite structures under transient heating.?2022 Chinese Society of Aeronautics and Astronautics.Production and hosting by Elsevier Ltd.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1.Introduction

Advanced composites have been extensively used in aerospace structures owing to their excellent stiffness and strength properties.1,2With the increasing speed of aircraft, thermomechanical coupling problems caused by aerodynamic heating are becoming increasingly prominent.High temperatures can change the thermo-mechanical properties of materials, which reduces the strength and modulus of materials and causes catastrophic failure of structures.3Therefore, studying the mechanical behaviours and failure mechanisms of composites and structures at high temperatures is of great significance to ensure the safety of composite structures in service.Many experimental studies have focused on mechanical properties,4,5impact properties,6–8aging properties,9and fatigue properties.10,11Most studies concur that the resin matrix softens and undergoes plastic deformation at high temperatures,which results in interfacial debonding between the fibre and matrix.Furthermore, the different thermal expansion coefficients of the fibre and matrix cause thermal stress at the interface, which induces microcracks and causes deterioration in the mechanical properties of the composites.12

Experiments can intuitively observe the failure process and characterise the properties.However, there are several aspects of failure that cannot be measured because most of the damage occurs inside the structure.In contrast,numerical methods can conveniently study the strain and stress distributions and damage evolution.The Progressive Damage Model (PDM) based on the Finite Element Method (FEM) has been proven to be an effective method for revealing the damage process and failure pattern of composite structures.13By using PDM, the stress and strain distribution characteristics of composites in high-temperature environments can be quantitatively analysed.Furthermore, the distribution of microcracks caused by thermal stress has been determined, and the progressive damage process of composites under high temperatures has been studied.14,15Over the past decades, researchers have proposed a wide range of computational models for predicting damage evolution in composites at different scales.Using FEM,16,17Discrete Damage Models (DDMs),18,19and multiscale models,20,21PDMs have been successfully applied to investigate the damage evolution and failure process of various composite structures,such as perforated laminates,22–24T-joints,25–27and other similar simple structures.However, for complex structures such as cabins28and connecting rods,29it is necessary to simplify the models and failure criteria using a userdefined material constitutive model to improve the poor calculation efficiency and convergence while ensuring good accuracy.In particular, when modelling thermo-mechanical coupling under transient heating, the numerical solution usually diverges or fluctuates because the temperature and material properties both vary sharply.Undoubtedly, it is more difficult to predict the damage process of complex composite structures at high transient temperatures by PDM owing to the strong coupling between thermal and mechanical loading.Consequently, analysis models considering the coupling of transient high temperatures and mechanics have rarely been reported in the literature until now.

High-speed aircraft (e.g., missiles, rockets, and hypersonic vehicles) usually operate in rapid heating environments.30For example, during launch, supersonic missiles undergo an extraordinarily rapid heating process.Consequently, investigating the damage process and failure mode under transient heating is important to ensure the safety of composite structures in such service environments.However, most of the above studies have focused on steady high-temperature performance rather than transient heating.In fact, the thermomechanical properties of materials under steady high temperature are not exactly the same as those under transient heating.A number of studies have shown that the strength limit of materials under transient heating is significantly higher than that under long-term constant temperature.31–35Moreover,the temperature gradient along the thickness direction under transient heating is far larger than that of steady high temperature.When hypersonic vehicles are flying at high speed, the aerodynamic force and aerodynamic heat change continuously with time,as well as the material properties.These factors lead to a strong thermo-mechanical coupling effect, which makes both experiments and simulations extremely complicated.Accordingly,the failure mechanism and behaviour of composite laminates under transient heating are quite different from those under steady high temperature.It is noteworthy that the thermo-mechanical coupling test and thermal strength analysis are prerequisites for ensuring structural safety in service.Nevertheless, there have been few investigations on these subjects by experimental or numerical methods hitherto.Therefore, we analysed a composite attaching collar, which is an important and typical component in radar-guided missile body connections, and investigated its transient thermomechanical properties.First, a Carbon Fibre/Polyimide Composite (CFPC) attaching collar was manufactured by a hightemperature Resin Transfer Moulding (RTM) process, and a thermo-mechanical coupling test was conducted by combining a high-temperature transient heating test system and a hydraulic servo loading system.Second, a Finite Element Analysis(FEA) model was established based on the sequential thermo-mechanical coupling method for progressive damage analysis.Finally,the FEA results were analysed and compared with the experimental results.

2.Experimental and modelling methodology

2.1.Material and layup of attaching collar

The composite attaching collar was prepared using a hightemperature RTM process.Meanwhile, the titanium alloy inserts were embedded in the screw holes to reduce the stress concentration between the screws and composite.The reinforcement was CF3031 carbon fibre plain weave fabric with a ply thickness of 0.225 mm made by WeiHai-TuoZhan Fiber Limited Company in China,and HT-350RTM polyimide resin provided by AVIC Composite Corporation ltd.was used as the matrix.The material properties are presented in Table 1.As shown in Table 1, the resin viscosity was only 0.81 Pa?s(<1 Pa?s) after being kept at 280 °C for 2 h, and the process continued for 3–4 h.The glass transition temperature of HT-350RTM is 392 °C, and its long-term working temperature is above 350°C,which is similar to the PETI-330 polyimide resin developed by NASA.36The geometry and dimensions of the composite attaching collar are shown in Fig.1.The composite attaching collar was composed of an inner bearing layer and an outer heat insulation layer, whose layups were[(±45)/(0,90)/(±45)/(0,90)/(±45)/(0,90)/(±45)/(0,90)/(±45)/(0,90)/(±45)/(0,90)3/(±45)/(0,90)/(±45)/(0,90)/(±45)/(0,90)/(±45)/(0,90)/(±45)/(0,90)/(±45)]and[(±45)]21s,respectively.As observed in Fig.1, the outside diameters of the attaching collar were 176.5 mm and 204 mm in the front and rear ends,respectively.In addition, the thicknesses of the front end and rear end were 1.25 mm and 5.50 mm,respectively.The lengthsof the heat insulation layer and attaching collar were 63.4 mm and 101.4 mm, respectively.

Fig.1 Configuration and layup of composite attaching collar.

2.2.Manufacturing process of attaching collar

The RTM forming die of the composite attaching collar,which was composed of an upper mould, a lower mould, a mandrel,and outer combinative mandrels, was designed and manufactured according to the structural characteristics of the attaching collar(see Fig.2).The inlets were set at the rear end(thickwall end)of the attaching collar,while the outlets were located at the front end (thin-wall end).The upper and lower moulds were connected using bolts, and the pins were used for positioning.

The RTM process for the attaching collars mainly included carbon fibre fabric cutting and draping, mould closing, resin injection, curing, and demoulding.First, the mould was cleaned with acetone and evenly coated with a hightemperature release agent.Then, the carbon fibre fabric containing a special polyimide tackifier was cut and draped on the inner mandrel according to the designed layup (as shown in Fig.3(a)).Second, the attaching collar preform was placed on the lower mould,and the outer combinative mandrels were assembled in sequence.Finally, the upper mould was closed,and the forming die was placed on a hot press for heating.Subsequently,consolidation was conducted.During consolidation,the polyimide powder was first placed in the injection tank and heated to 280 °C at a rate of 5 °C/min and held for 1 h.The mould and injection tank were depressurised by vacuum at the same time.Subsequently, the resin was injected into the mould with 0.1–0.5 MPa pressure until the resin completely impregnated the preform.Then, the mould was heated to 370 °C at a rate of 5 °C/min and held for 2 h again.Next,the mould was cooled to room temperature,and consolidation was completed.According to the ultrasonic A-scan results,the quality of the composite attaching collar was quite good without delamination.Finally, a net-shaped composite attaching collar was obtained by machining and assembling with titanium alloy inserts (as shown in Fig.3(b)).

Fig.3 Composite attaching collar.

2.3.Thermo-mechanical coupling tests

Considering the service environment of the composite attaching collar, a thermo-mechanical coupling test under transient heating was conducted to study the high-temperature mechanical behaviour of the attaching collar.Fig.4 shows the test fixture and temperature control points of the composite attaching collar.As shown in Fig.4, the test system was mainly composed of a loading fixture, fixed clamp, quartz lamps, lamp bracket, reflector, thermocouples, and composite attaching collar.The loading fixture and fixed clamp were coated with a TR-37 heat insulation coating(Aerospace Research Institute of Materials and Processing Technology, China) of 2.5 mm thickness to prevent heat conduction from the fixture to the attaching collar.Before the test,the front end of the attaching collar was bonded with the loading fixture by a twocomponent room-temperature curing epoxy adhesive, and the rear end of the attaching collar was assembled with a fixed clamp using eight countersunk screws of diameter 4 mm.The fixed clamp was then installed on the steel beam.Type K thermocouple wires with a diameter of 0.511 mm, accuracy of ± 0.75% temperature, and maximum temperature of 704°C were used for temperature measurement.Three thermocouples were directly attached at the corresponding measuring points on the internal and external surfaces of the specimen(see Fig.4)using a silicone resin adhesive provided by Harbin Institute of Technology.In order to obtain good temperature response, the thickness of the adhesive layer was controlled within 0.15 mm.Finally, the static loading device, lamp bracket, quartz lamps, and reflector were installed sequentially.The temperature measuring points were in the middle section of the attaching collar, and four thermocouples were arranged at the attaching collar.Measuring points 1 and 2 were the inner wall temperature measuring points, Measuring point 3 was the central air temperature measuring point, and Measuring point 4 was arranged on the outer wall of the attaching collar for temperature control.A bending load was applied to the front end of the loading fixture using a hydraulic servo loading system, and the temperature was provided by a high-temperature transient heating test system,which was controlled overall by adjusting the direct current voltage value at both ends of the quartz lamps to change their working power.To obtain a uniform temperature field, each of the 25 quartz lamps of length 300 mm were arranged uniformly along the circumference.The surface temperature of the specimen rapidly reached the set value by the heat radiation of the quartz lamps.A PT100 temperature acquisition system was used to record the temperature.The load and temperature were controlled by an MTS complex coordinated control system.The loading mode was load control.The displacement survey point was placed at the top of the loading fixture,and the displacement was automatically recorded by the loading system.Before the formal test,the transient heating system and hydraulic servo loading system were verified by a pretest with a 30% load at 200 °C.Fig.5 shows the control curves of the temperature and load.As shown in Fig.5, the surface temperature of the attaching collar reached 520 °C at 13.5 s,and the load reached 6.0 kN synchronously.Subsequently,if the specimen was not destroyed,the load gradually increased to 25.0 kN at 36.5 s while holding the temperature at 520 °C.

2.4.Progressive damage analysis modelling methodology

Fig.4 Test fixture and temperature control points of the composite attaching collar.

Fig.5 Practical control and feedback curves of temperature and load.

Fig.6 FE model of composite attaching collar assembly.

To reveal the failure mechanism of the composite attaching collar under transient thermo-mechanical coupling,a 3D finite element model of the attaching collar assembly, as illustrated in Fig.6, was established using ABAQUS37based on the sequential thermo-mechanical coupling method.The damage process of the attaching collar was simulated using PDM.The fibres in the 0° and 90° directions were consistent with the axial and circumferential directions of the attaching collar,respectively.Considering the requirements of sequential thermo-mechanical coupling progressive damage analysis, the element thickness along the radial direction was the same as the actual ply thickness of the fabric, which indicated that the mesh was subdivided ply-by-ply along the radial direction with a total of 700148 elements.The loading fixture was coated with a 2.5 mm thick TR-37 heat insulation coating, and the fixed clamp was set as a rigid body.Considering the influence of temperature on the material properties and the fact that the temperature distribution in the attaching collar was not uniform,the material properties as a function of temperature were used for the thermo-mechanical coupling simulation to accurately obtain the structural response trend.Although it is well known that the direct thermo-mechanical coupling method is an effective method for analysing structures under steady heating,its calculation efficiency is very low.Additionally,the convergence is poor when the model involves transient heating and damage evolution of complex laminated composite structures.In fact, in the direct thermo-mechanical coupling analysis,it is necessary to solve the heat conduction problem and the thermal stress problem simultaneously to determine the temperature, displacement, and stress inside the structure, which increases the difficulty of solving and leads to oscillation in the decoupling process.In contrast, in the sequential thermomechanical coupling method, the temperature fields of all elements are first calculated through heat conduction analysis.Then, the node temperatures calculated by FEA are imported into the structural analysis model.After setting the thermomechanical properties of the materials and boundary conditions, the failure indexes can be calculated by a User-Defined Field(USDFLD).Then,thermo-mechanical coupling progressive damage analysis can be realised.This method can improve the calculation efficiency while ensuring good prediction accuracy.Because the total deflection of the attaching collar was within 5 mm and quasistatic loading occurred, the additional strain term in the heat conduction equation can be ignored.Therefore,the sequential thermo-mechanical coupling method,which only considers the effect of temperature on elastic deformation,is more suitable for solving practical engineering problems.Moreover,the strength and modulus parameters varying with temperature can be considered in the simulation model,which is more essential for the thermo-mechanical coupling calculation.Consequently,a flowchart for progressive damage analysis based on the sequential thermo-mechanical coupling method was adopted, as shown in Fig.7.

According to Fig.7, a thermo-mechanical coupling finite element model was established.Then, the element type was set to an 8-node hexahedral element (DC3D8), and heat conduction analysis was carried out first.The temperature boundary condition was applied to the outermost part of the model,as shown in Fig.5.Each increment size was set as 0.5 s.According to the actual test time, the total time was set as 22.0 s.After heat conduction analysis,the functions of temperature and time at each node were imported to the next analysis step,and the element type was adjusted to an 8-node reducedintegrated hexahedral element(C3D8R)for stress analysis.As shown in Fig.6, a reference point was created at the centre of the upper end of the loading fixture to couple with the nodes of the upper end.Then, a concentrated force was applied to the reference point in the x direction and gradually increased to 6.0 kN at 13.5 s.Afterwards, the load reached 25.0 kN at 36.5 s with an increment of 0.5 s, which corresponded to the increment in heat conduction.The six degrees of freedom of the rigid base in the three directions were constrained according to the actual test constraints.The interaction between the composite attaching collar and the rigid base was set as hard contact.The failure criteria and stiffness degradation rules of the materials play a critical role in PDM.The failure criterion was applied to estimate the damage in elements, while the degradation rule was used to reduce the stiffness of the damaged elements.In comparison with the maximum stress or strain criteria, the Hashin criterion38has been proven to be more effective for predicting the failure of laminates because it considers the interaction between fibre and matrix,especially when the composite structures are under complex stress conditions, such as screw connections.Therefore, the Hashin criterion was used to evaluate the damage to the CFPCs.Furthermore, considering that the fracture of metal screws is usually plastic failure, the von Mises criterion39was adopted to calculate the damage characteristics of metal screws based on the USDFLD.The mechanical and thermal properties of the materials are listed in Table 240and Table 3, respectively,where T is temperature.The change in the material parameters with temperature was realised by linear interpolation.The strength and modulus of each element changed with temperature during the analysis.Hence, the strength and modulus parameters were reset at each increment according to the real-time temperature (see Fig.7).Stiffness reduction is an effective method to simulate the progressive damage process of laminated composite materials.13,15Here, the reduction in stiffness was dependent on the rules listed in Table 4.

Fig.7 Progressive damage analysis flowchart based on sequential thermo-mechanical coupling method.

3.Results and discussion

3.1.Experiment

Fig.5 shows a comparison between the controlled temperature and the feedback temperature in the actual test.As shown in Fig.5, they are in good agreement in the first 13.5 s.From 13.5–15.0 s, the feedback temperature has a small peak (i.e.,approximately 550 °C), which is 30 °C higher than the controlled temperature, and then the feedback temperature quickly drops and returns to the controlled temperature.This might be due to the large thermal inertia of the attaching collar, and the control accuracy of the test system being insufficient, which leads to a small fluctuation in temperature at the final heat-up stage.Furthermore, the specimen fractures at 20.5 s with a failure load of 11.794 kN, which is 118% of the limit load (10.0 kN).Fig.8 shows the temperature measured at the inner part of the attaching collar, which indicates that the maximum temperature of the internal measuring point of the specimen is 30.0°C and the maximum air temperature of the central point is 28.0 °C.This indicates that the composite has excellent thermal insulation properties.Meanwhile, there is rapid heating at Measuring point 3 at the end of the test caused by the rapid entry of hot air into the specimen after the failure of the attaching collar.

It is evident from Fig.9 that the experimental load–displacement curve includes two rigidity mutation regions at loads of 4.0 kN and 7.1 kN at the corresponding temperatures of 318 °C and 558 °C, respectively, where t is time.The firstrigidity mutation is possibly due to the assembly clearance and deterioration in material properties at high temperatures,while the second rigidity mutation is primarily caused by the damage or fracture of the screws.The ultimate failure load of the specimen is 11.794 kN,and the maximum displacement at the loading position is 5.52 mm.After the test,it was observed that the screws at the rear end of the attaching collar were fractured by shearing (see Fig.10).Furthermore, no residual deformation or damage is observed in the attaching collar, which indicates that the composite attaching collar meets the design requirements of thermal strength.As the thermal conductivity of metals is significantly higher than that of composites, the temperature in the screws increases more rapidly during the test than that in the composite attaching collar, which causes high thermal stress under a large temperature gradient,as well as deterioration in the material properties of the screws.Consequently, the screws are more vulnerable than the composite attaching collars at high temperatures.Fig.10 also shows that an obvious gap occurred at the hole edges, which is probably due to extrusion of the titanium alloy inserts to the hole edges leading to local damage at the hole edges.Furthermore, ablation traces are observed on the surface of the attaching collar,as well as some pores,which indicates that the polyimide resin decomposed at the transient high temperature and resulted in damage at the outer plies.It is noteworthy that the maximum temperature (i.e., 520 °C) of the attaching collar exceeded the glass transition temperature of polyimide resin (i.e., 392 °C)and was near its decomposition temperature (i.e., 537 °C).However, the CFPC attaching collar maintained good integrity without obvious damage.The above findings indicate that the failure behaviour of composites under transient heating is different from that under steady high temperature.The CFPC attaching collar still maintained good strength and stiffness under transient high temperature.These results are in agreement with the conclusion asserted by Jiang et al.33that polyimide composites still have a certain bearing capacity at a transient high temperature of 500 °C.

Table 2 Mechanical properties of structural materials used for composite attaching collar.40

Table 3 Thermal properties of structural materials used for the composite attaching collar.

Table 4 Mechanical property degradation rules of composite.

Fig.8 Temperature at measuring points at inner part of composite attaching collar.

Fig.9 Load–displacement curves of composite attaching collar.

Fig.10 Failure mode of composite attaching collar.

Fig.11 shows the schematic failure process of the composite attaching collar.In fact,there is a small assembly clearance between the attaching collar and the fixed clamp after screw assembly (Fig.11(a)).Therefore, the attaching collar first rotates around the axis of the Nos.4 & 5 screw connections under a bending load, resulting in an increased clearance in the upper end at the root of the attaching collar and a decrease in clearance in the lower end until it contacts with the end surface of the fixed clamp (Fig.11(b)).Subsequently, as the load increases, the attaching collar rotates around the contact line until the end of the test (Fig.11(c)).Consequently, before the lower end of the attaching collar contacts the end surface of the fixed clamp,the force of the eight screws can be regarded as symmetrical along the axis of the Nos.4&5 screw connections.After contacting the end surface of the fixed clamp, the eight screws are subjected to linear loading instead of the original symmetric loading with the No.8 screw as the fulcrum,and the load decreases from the No.1, Nos.2 & 3, Nos.4 &5,and Nos.6&7 screws to the No.8 screw in sequence,which results in the failure order of the screws in the test process being Nos.1–8.On the other hand, the titanium alloy inserts extrude the hole edges under bending loading, which results in local damage at the hole edges.Obviously, the load and damage of the first broken screw is the largest.Hence,according to the fracture sequence of the screws, the hole edge of screw No.1 was damaged the most.As shown in Fig.11(c),the insert squeezes the hole edge downwards, resulting in extrusion damage at the lower part of the hole, and the insert moves along the damage direction.Consequently, there is a gap between the insert and the upper part of the hole.Since the maximum temperature on the surface of the attaching collar is close to the decomposition temperature of the polyimide resin, the resin on the surface was slightly decomposed under transient heating, which leads to local pores appearing at the intersection of warp yarns and weft yarns of fabric.

Fig.11 Failure process of composite attaching collar.

3.2.Thermo-mechanical coupling analysis

Fig.12 shows the calculated temperature distribution of the composite attaching collar at t = 20.5 s.It is observed that the temperature at the outer surface of the attaching collar reaches 520 °C at 20.5 s, while the inner surface remains at a low temperature except for the hole edges and the front end.This is because the thermal conductivities of titanium alloys and steel are greater than that of carbon fibre reinforced polyimide composites.Therefore, the heat on the surface is preferentially conducted from the titanium alloy inserts and steel screws to the hole edges of the composite attaching collar.As the front end is thinner and the heat conducts to the inner part of the metal actuator more quickly,the temperature at the front end is relatively higher than that at the inner surface of the rear end.Moreover, owing to the hysteresis of heat conduction,although the outer surface temperature of the attaching collar reaches a maximum, the inner surface is still at low temperature.The calculated temperature of the middle section of the attaching collar is 29.4 °C, which is consistent with the maximum temperature of 30.0 °C in the test.

The load–displacement(P–δ)curve and displacement distribution of the composite attaching collar calculated by FEA are shown in Figs.9 and 10,respectively.Fig.9 demonstrates that the predicted P–δ curve from FEA is close to the experimental curve,except for the two stiffness mutations in the experiment.The numerical results are slightly steeper than the experimental results.This is because the finite element model applies perfect boundary conditions,and assembly clearance is not considered in the model.However, slipping caused by assembly clearance and material property deterioration induced by high temperature would significantly change the experimental structural stiffness.Moreover, the loading rod also slightly affected the displacement of the experiment.Therefore,the predicted curve is steeper than the experimental curve.As shown in Fig.10,the attached collar rotates along the contact line between the lower end of the attaching collar and the fixed clamp.Furthermore,Figs.11(b) and (c) clearly reveal the magnitude and location of the contact stress of the attaching collar.As shown in Figs.11(b) and (c), the contact stress increases gradually as the load increases.At the end of the test,the maximum contact stress is 375 MPa, accompanied by slight damage at the contact point.The maximum failure load predicted by FEA is 11.677 kN, which is consistent with the experimental result(i.e., 11.794 kN).Thus, it can be considered that progressive damage analysis based on the sequential thermo-mechanical coupling method proposed in the present study can precisely calculate the ultimate failure load of the attaching collar under coupling transient heating and bending loads.

Fig.12 Temperature distribution of composite attaching collar at t = 20.5 s.

Fig.13 shows the progressive damage process of the connected screws.The initial failure occurs at the top screw(No.1) caused by shearing at t = 16.0 s.As the temperature and load increase, the damage propagates to the adjacent screws,i.e.,Nos.2&3,Nos.4&5,and Nos.6&7 in sequence(shown in Fig.13(b)), and the No.1 screw totally ruptures at t = 19.5 s.Almost at the same time, the Nos.2 & 3 screws rupture, resulting in the collapse of the attaching collar at t=20.5 s(shown in Fig.13(c)).The Nos.4&5 screws are also seriously damaged.This is consistent with the experimental phenomena shown in Fig.10.The damage locations of the composite attaching collar at t = 20.5 s estimated by the 3D Hashin criterion are shown in Fig.14.It is shown that the damage is not severe.Furthermore, intralaminar damage mainly (i.e., warp-wise fibre failure, weft-wise fibre failure,and in-plane shear failure)occurs at the hole edges(see Figs.14(a),(b),and(d)),while the interlaminar damage(i.e.,interlaminar tensile and shear failure)primarily appears at the junction(shown in Figs.14(c), (e), and (f)).Since the hole edges are mainly squeezed by the inserts,the damage modes are in-plane failure.A sharp ply drop causes relatively high interlaminar stress at the junction.These results are in accordance with the experimental findings.

Fig.13 Progressive damage process of metallic screws.

Fig.14 Failure indexes of composite attaching collar at t = 20.5 s.

Figs.15 and 16 show the evolution of warp-wise fibre damage and weft-wise fibre damage, respectively.The damage in the warp-wise direction initiates at the outer plies of the hole edge at t = 16.0 s (see Fig.15(a)).Subsequently, the damage propagates to the inner plies of the hole edges as the load and temperature increase (see Figs.15(b) and (c)).The weftwise fibre damage initiates at the outer plies at t = 13.0 s,and then the damage propagates to the hole edges at t = 16.0 s (see Figs.16(a) and (b)).Similarly, the weft-wise fibre damage propagates to the inner plies of the hole edges as the load and temperature increase(see Fig.16(c)).Nevertheless, the damage at the hole edges only appears at the outer plies of the heat insulation layer (see Figs.15 and 16).This damage is mainly due to the temperature being the highest at the outer plies and hole edges, which results in higher thermal stress at the heat insulation layer and hole edges,as well as significant deterioration in the material properties in this region.Although there is some damage to the composite attaching collar, the damage locations shown in Fig.14 indicate that the damaged plies mainly belong to the heat insulation layer and have an insignificant effect on the bearing capacity of the composite attaching collar.Therefore, the CFPC attaching collar still has a certain residual strength.

Fig.15 Warp-wise fibre damage process (X1) of composite attaching collar.

Fig.16 Weft-wise fibre damage process (X2) of composite attaching collar.

The PDM based on the sequential thermo-mechanical coupling method can effectively simulate the failure process and intuitively reveal the damage mechanism of the composite attaching collar under coupling transient heating and bending load.Moreover, the FEA results show that the damage that occurs at the composite attaching collar is earlier than that at the screws.This phenomenon cannot be achieved by experimental observation, and it reveals the failure mechanism of the composite attaching collar.However, it must be pointed out that the high-temperature material properties used in the present models are based on long-term high-temperature test results.Meanwhile, the variation in the material parameters with temperature is realised by linear interpolation.In fact,these results are not exactly consistent with the experiments and affect the precision of the FEA results to some extent.Therefore, in the future, a more accurate model should be developed by introducing true material properties.Nevertheless, the present model can provide a reference for evaluating and predicting the thermal strength of complex composite structures under transient heating.

4.Conclusions

This study focuses on the thermo-mechanical behaviours of a CFPC attaching collar by a high-temperature RTM process under coupling transient heating and bending load.The thermo-mechanical coupling test results reveal that the attaching collar fractures at the 118% limit load, and the failure mode involves the rupture of screws at the root,local extrusion damage of the hole edges, and slight ablation damage at the outer plies.Meanwhile, there is no evident residual deformation in the composite attaching collar,which meets the requirements of thermal strength.A sequential thermo-mechanical coupling method based on USDFLD is proposed to analyse the performance of the complex structure under multiple physical fields.Considering that the material parameters vary with temperature,the PDM is developed to simulate the failure process of the composite attaching collar.The results show that the PDM can provide insight into the asymmetric progressive damage process of the composite attaching collar and predict the failure process of screws precisely.The damage factor of the composite was calculated to assess the safety status of the attaching collar.The results show that the damage to the composite caused by thermal stress and bearing stress is mainly distributed at the heat insulation layer and the hole edges,which slightly affects the structural bearing capacity.Among the six failure modes, the intralaminar failures (i.e., warpwise fibre failure, weft-wise fibre failure, and in-plane shear failure)are the predominant damage patterns in the composite attaching collar.The FEA results are in good agreement with the experimental results.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This project was supported by the Young Elite Scientists Sponsorship Program by the China Association for Science and Technology (No.2016QNRC001) and the Science and Technology Commission of Shanghai Municipality, China (No.19DZ1100300).