i AN EXPERIMENTAL INVESTIGATION INTO THE BEHAVIOR OF GLASSFIBER REINFORCED POLYMER ELEMENTS AT ELEVATED TEMPERATURES By KENNY ZONGXI QIAN B.S., Northeast Forestry University, 2009 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Civil Engineering 2014
ii 2014 KENNY ZONGXI QIAN ALL RIGHTS RESERVED
iii This thesis for the Master of Science degree by Zongxi Qian Has been approved for the Civil Engineering program By Yail jimmy Kim, C hair Frederick Rutz Nien Yin Chang Nov/20 /2014
iv Zongxi Qian. (M.S., Civil Engineering) An Experimental Investigation into the Behavior of Glass Fiber Reinforced Polymer Elements at Elevated Temperatures Thesis directed by Associate Professor Yail Jimmy Kim ABSTRACT This thesis presents a literature review and results of an experimental study about the effects of high temperatures and cyclic loading on the physical and mechanical properties of pultruded glass fiber reinforced polymer (GFRP) square tubes used in civil engineering structural applications. Most laboratory researches have focused mainly on the effect of elevated temperature on the compressive strength of the GFRP square tubes. Limited research has focused on the tensile strength of GFRP coupons under eleva ted temperatures. Dynamic Mechanical A nalyses (DMA) was performed to assess the viscoelastic behavior including the glass transition temperature of GFRP. Sixteen GFRP coupons were tested under elevated temperatures to investigate the tensile strength and t he effect of elevated temperatures to the tensile strength of GFRP. The results of an experimental program performed on fifty GFRP square tubes with different designs in 1.83m at normal temperatures were discussed to investigate compression performance. A nother experimental program was performed on 20 GFRP square tubes with different designs in 1.22m under elevated temperatures. The experiments results were discussed and showed that the compressive strength of GFRP material was influenced by several factor s including the glass transition
v temperature and the connection bolts. Failure modes under 25 C and 75 C were crushing and the failure modes with the temperatures above 75 C were not typical crushing due to the glass transition of GFRP. Sixteen GFRP square tubes with length of 0.61m were tested with the same experimental program under elevated temperatures as the control group. Twelve GFRP square tubes with the same size were subjected to cyclic loading under elevated temperatures to investigate the effect of the cyclic loading to the compression properties of GFRP material. According to the experimental results and the discussion, the stiffness was reduced by the cyclic loading. On the contrary, the influence of the cyclic loading was not obvious compared t o the GFRP specimens subjected to normal displacement control loading. The higher temperature made the stiffness of GFRP more sensitive to the cyclic loading. The form and content of this abstract are approved. I recommend its publication. Approved: Yail Jimmy Kim
vi ACKOWLEDGEMENT This thesis would not have been possible without the support of so many people in many forms. It was the product of a large measure of serendipity with people who have changes the course of my life. When I was a young boy, I was dreaming of getting a master degree from a good university in the future. This dream came from my parents, my grandma and everyone appeared in my life. Thanks to all those of whom I have come across and have led to this path that I chose. First, I must take to opportunity to acknowledge and thank my parents for their endless support and understanding. I am the only child in the family and I am very thankful for their help at any time that I met problems. I would like to thank my wife for th e love and company when I need her and also taking care of the family when I was busy working on my thesis. With this believe in mind, it is with immense gratitude that I acknowledge the deepest appreciation to my advisor, Associate Professor Yail Jimmy Kim, for his patience and guidance. Without his encouragement, I wou ld have given up working on my thesis. I t would be very hard for me to finish this. Also, I would like to thank the faculties in department of mechanical. I must thank Tom Thuis, Jac Corle ss, Yakachi Chris for their patient support with the experiments part. I could finish this thesis without them.
vii This thesis work was supported by University of Colorado.
viii TABLE OF CONTENTS Chapter 1. Introduction ................................ ................................ ................................ ................ 1 2. Literature Review ................................ ................................ ................................ ....... 2 2.1 GFRP ................................ ................................ ................................ .................... 2 2.1.1 Processing Technologies ................................ ................................ ............... 2 2.1.2 GFRP Application ................................ ................................ ......................... 4 2.1.3 GFRP Techniques ................................ ................................ ......................... 5 2.1.4 Ba ckground of GFRP and GFRP Rebar ................................ ....................... 6 2.2 High Temperature Effect ................................ ................................ ..................... 7 2.3 Glass Transition ................................ ................................ ................................ ... 9 2.4 Decomposition ................................ ................................ ................................ ... 11 2.5 Effect of Elevated Temperatures on GFRP Materials ................................ ....... 11 2.6 Behavior of Concrete Members Strengthened with GFRP at High Temperature ................................ ................................ ................................ ................................ .. 13 2.7 Performance Based Fire Safety Design ................................ ............................. 14 2.8 History of Performance Based Codes ................................ ................................ 14 2.9 The Advantages and Disadvantages of Performance Based Design .............. 16
ix 2.10 Fatigue Damage of GFRP ................................ ................................ ................ 17 2.11 Fatigue of GFRP Reinforcement ................................ ................................ ..... 18 2.12 Cyclic/Fatigue Loading of Structural Members ................................ .............. 19 2.13 Acoustic Emission and Fatigue Behavior of GFRP ................................ ......... 20 3. Experimental Program ................................ ................................ ............................. 28 3.1 GFRP Coupon Tensioning Tests with Elevated Temperatures ......................... 28 3.1.1 Coupons Design ................................ ................................ .......................... 28 3.1.2 Pr eparation ................................ ................................ ................................ .. 29 3.1.3 Test Setup and Test Process ................................ ................................ ........ 30 3.2 Dynamic Mechanical Analysis ................................ ................................ .......... 30 3.3 Compression Tests of GFRP Columns with Different Designs ......................... 31 3.3.1 Tests Specimens ................................ ................................ .......................... 31 3.3.2 Setup and Instruments ................................ ................................ ................. 32 3.3.3 Tests Procedure ................................ ................................ ........................... 33 3.4 Compression Tests of GFRP Columns with Different Designs in 1.22m under Elevated Temperatures ................................ ................................ ............................. 33 3.4.1 Design of Specimens ................................ ................................ ................... 34 3.4.2 Setup and Instruments ................................ ................................ ................. 35 3.4.3 Tests Procedure ................................ ................................ ........................... 36
x 3.5 Compression Test of Plain GFRP Columns in 0.61m under Elevated Temperatures ................................ ................................ ................................ ............ 37 3.6 Cyclic Loading Compression Test of Plain GFRP Columns in 0.61m under Elevated Temperatures ................................ ................................ ............................. 37 3.6.1 Specimens and Setup ................................ ................................ .................. 37 3.6.2 Experiment Procedure ................................ ................................ ................. 38 4. Results Analysis and Discussion ................................ ................................ ............. 44 4.1 DMA Analysis of GFRP ................................ ................................ .................... 44 4.1.1 Storage Modulus ................................ ................................ ......................... 44 4.1.2 Loss Modulus ................................ ................................ .............................. 45 4.1.3 Damping Factor (Tan Delta) ................................ ................................ ........... 46 4.1.4 Comparison of Glass Transition Temperatures and Summary ................... 46 4.2 GFRP Coupons Tensioning Tests ................................ ................................ ...... 47 4.2.1 Load vs. Axial Displacement Behavior ................................ ...................... 47 4.2.2. Stress vs. Strain Behavior ................................ ................................ .......... 49 4.2.3. Failure Modes of Tensile Tests ................................ ................................ .. 51 4.2.4. Conclusions ................................ ................................ ................................ 53 4.3 Compression Strength of GFRP Square Tubes ................................ .................. 54 4.3.1. Compressive Strength and Axial Deformation ................................ .......... 54
xi 4.3.2 Stress and Strain ................................ ................................ .......................... 56 4.3.3. The Failure Mode of the Compressive Tests ................................ ............. 57 4.3.4 Summary ................................ ................................ ................................ ..... 59 4.4 Compressive Tests under Elevated Temperatures ................................ ............. 60 4.4.1 Strength and Axial Displacements ................................ .............................. 60 4.4.2 Stress and Strain ................................ ................................ .......................... 64 4.4.3. Failure Modes ................................ ................................ ............................ 65 4.4.4. Summary ................................ ................................ ................................ .... 66 4.5 Compressio n Tests of GFRP Columns in 0.61m under Elevated Temperatures ................................ ................................ ................................ ................................ .. 67 4.5.1 Strength Axial Displacement ................................ ................................ ...... 68 4.5.2 Failure Modes ................................ ................................ ............................. 70 4.6 Cyclic Loading Compression Test of GFRP Columns in 0.61m under Elevated Temperatures ................................ ................................ ................................ ............ 71 4.6.1 Effect of Cyclic Loading to the Compressive Strength .............................. 71 4.6.2 Th e Effect of the Cyclic Loading on the Stiffness of GFRP Columns ....... 72 5. Numerical Modeling ................................ ................................ ................................ 96 5.1 Euler Buckling Formula ................................ ................................ ..................... 96 5. 2 Finite Element Analysis Model ................................ ................................ ......... 97
xii 5.3 The Comparison of Theoretical Critical Load and Critical Load from ANSYS 98 5.4 The Buckling Load of GFRP Columns with Different Wall Thickness .......... 100 5.5 Buckling Load of GFRP Columns with Different Modulus of Elasticity ....... 101 5.6 Buckling Load of Columns with Access holes ................................ ................ 102 5.7 Buckling Analysis of Columns u nder Elevated Temperatures ........................ 103 6. Summ a ry and Recommendations ................................ ................................ .......... 118 Reference ................................ ................................ ................................ ................... 121 Appendix ................................ ................................ ................................ .................... 124
1 1 Introduction GFRP which is known as glass fiber reinforced polyme r is first developed in the mid 1930s. However, GFRP was used merely at the early days. Till 1967, the architectural advantages of GFRP were discovered accidentally with the attempted destruction of ristic house was constructed using only GFRP. When the attractions were no longer necessary, people decided to destroy it. Amazingly, the wrecking ball merely bounced off the structure. From that point, civil engineers started to pay attentions to this new material full of potentials. By 1994, nearly 600 million pounds of composite materials had been used by building industry. GFRP provides one more choice for designers when they are looking for a good construction material.
2 2. Literature Review 2.1 GFRP 2.1.1 Processing T echnologies There is a variety of processing technologies which can be used to manufacture FRP Material. Only a few are introduced in this section. As shown in Fig 2.1, h and layer up is probably the most traditional and easiest method to produce FRP composite material. The process first is to place the resin on the surface of the mold, and then place a layer of fibers. The basic process will be repeated several times till there are enough layers of fi bers and resin. However, air must be removed between the fibers within the matrix, using a roller to squeeze it out. A layer of Gel coat is usually applied between the mo ld and Resin to reduce the cohesiveness. As shown in Fig 2.3, s pray lay up process is very different from hand lay up process. The differece comes from the way how the resin and fiberglass is applied to the mould. Spray up utilizes a low cost open mould, normal temperature curing resin. Compared to hand lay up process, it has a quite obvio us advantage when produce units in large size. Chopped fiber reinforcement and catalyzed resin are sprayed on to mould with a spray gun. As with lay up process, entrapped air is manually removed by roller.
3 As shown in Fig 2.5, v acuum assisted resin transf er molding is another method forming and shaping FRP Composite material. Fibers are layered on a solid mold base or core material which is designed to be reinforced. A vaccum bag is applied on top of the fibers. By generating a vaccum, the vaccum bag is co mpacted due to the atmosphere pressure, thus, entrapped air is squeezed out. Fabric absorb excess resin and the resin supply lines keeps supplying resins to the system. This process technology can meet the requirement for automation and ensure a good quali ty of the product as well. Filament winding Fig 2.7 is an automatic process to produce FRP structural components. It consists of creel with winding continuous fiber tow to form a tubular unit. After it come through the separator combs, the fiber is impre gnated with polymer resin. With the guide, the fiber is synchronized to move during the rotation of the rotating mandrel. Thus, the fiber with resin on it are laid down in a designed pattern. The main advantage of filament winding process are its high prod uction rate and relatively low manufacturing cost. Fig 2.8. Based on the reasearch, the most commonly used production process is pultrusion process.It is developed around 60 years ago by a person considered by many to be continuous cross section profile. It starts with creels holding rolls of continuous fiber roving. The fibers is pulled out from the rack and guided to the resin bath which is the impregnated system. After come out of the resin, all the fiber filaments
4 The fiber is organized and shaped into desired shape by this custom tooling. The surfacing veil is usually a dded to increase structure and surface finish. The excess resin is also removed in this step. Once the resin pregnated fiber is processed, the composite will go through a heated steel mould. The mould is usually heated to a constant temperature with severa l temperature zone through out its length. The temperature is the core factor of curing the composite. The composite is pinched and pulled out after being cured by the pulling system. At the end of this pultrusion process is cutting off. 2.1.2 GFRP A pplication There are a few buildings and bridges that were constructed with GFRP. For example, the first all GFRP pedestrian bridge was built in Taiwan. Yeou fong Li did the relevant research on it. The bridge was built in Taijing National Park of Taiwan which is located in a chloride concentrated environment. Structures made of conventional materials close to the seashore find themselves defe nseless to corrosion. This GFRP bridge has been in used over a decade now. The all GFRP pedestrian bridge was made of four continuous GFRP I girder and GFRP decks in its superstructure. The for Design of FRP Pedestrian Bridge for a deflection not more than L/500. This case fully proved that GFRP can totally meet the requirement as a structure material.
5 The most common way GFRP are used nowadays is used as the reinforcement in concrete bridge. Based on the researches people did before, GFRP reinforced concrete bridge deck or slab has lower m aintenance cost compared to traditional steel bar reinforced concrete elements. Considering the corrosion problem, a waterproofing membrane and a layer of asphalt are typically placed over the steel reinforced concrete to protect the steel from corrosion. However, the cost of these can be totally saved if the GFRP reinforced concrete is adopted. Due to the light weight property, the GFRP bars can lighten the whole structural system, which can save some material cost as well. High tensile strength, light wei ght, and resistance to corrosion make GFRP an attractive alternative to steel. Typically, GFRP materials have higher tensile strength relatively to steel yield strength of 414MPa [60 ksi] and u ltimate tensile strength of 552 MPa [80ksi], however, steel has larger Mod ulus of elasticity which is 200GPa [29,000 ksi]. As a result, the GFRP failure rapture is considered to be brittle failure, the strain is around 2%. However, steel ruptures at approximately 0.2%. 2.1.3 GFRP Techniques The steel has been using for a long time s however, there is some disadvantages that s, light weight, high strength and resistance to corrosion, low maintenance and durability. The most two most important factors are the high strength to weight ratio and corrosion resistance which make it a
6 sound alternative for steel in some design project. LRFD Design specification for FRP Pultruded Structures is one of the national specification s for GFRP. In order to keep the consistence, most of t he standard use GFRP rebar is to replace steel rebar. And the similar method like the steel was adopted in the GFRP rebar design calculations. As the least expensive type of fiber, glass fiber is now the most commonly used fiber. The glass fiber can be div ided into three classes E glass, S glass and C glass. The E glass is designated for electrical use and S glass is for high strength. The C glass is for high corrosion resistance which is uncommon for Civil Engineering application. Most of the glass fibers are applied in Civil Engineering are E glass. The diameters may range from 2 to 13 10 6 m. The glass fiber strength and modulus change tremendously due to the change of temperature. The temperature is a big concern when this material is applied. 2.1.4 Background of GFRP and GFRP R ebar It is believe that the first project manufactured with GFRP was a boat hull completed during 1930s. The mold was made from foam. After that, GFRP were widely used in Air force, the Navy and oil industry. However, people started to pay attention at this new material was because of another building built totally with GFRP in 1950s, the the wrecking ball hit the futuristic GFRP house, it simply bounced the structure. It
7 was this event that fully highlighted the high strength property of GFRP. Nowadays, GFRP is more used as GFRP rebar in GFRP reinforced concrete. In order to keep the consistence with steel rebar, GFRP rebar has the similar surface, (Figure 2.9) Ribbed (a), Sand co ated (b), Wrapped (c), Deformed and Helical. Due to its light weight property, it is applied a lot in bridge deck slab to decrease the self weight. GFRP AASHTO LRFP Bridge Guide Specifications for GFRP reinforced concrete bridge deck and traffic railings i s the code. Besides, some researches were done in order to know GFRP better. Mathieu Robert1 and Brahim Benmokrane did the research of the performance of GFRP under high temperature. David Trejo, Paolo Gardoni, Jeong Joo Kim, and Jason Zidek did the resea rch of long term performance GFRP reinforcement. However, this area still need more research to help have a better applying in engineering. There is no research date found in long term high with high temperature. 2.2 High Temperature Effect All the fibe r reinforced composites are composed of axial particulates embedded in a matrix material which is also called resin. Because when GFRP material is loaded, the load is loaded on matrix directly, then transmitted from the matrix to the fibers, thus, the stre ngth and modulus of GFRP can be very high. Hence, interfacial bonding between fibers and resin is important for the transmission of applied load. GFRP materials consist of strong fibers surrounded by a relatively weaker resin material. Although the fibe r itself can retain their strength at elevated and high
8 temperature, most of the polymer matrix materials are vulnerable at elevated temperatures. As shown in the strength of carbon fiber seldomly change when the temperature increase. The strength of g lass fiber decreases approximately 10% when the temperature increase from 0 C to 200 C. However, the strength of glass fiber reinforced polymer decreases tremendously from 100 C to 150 C. Obviously, the drop of the strength of GFRP is not due to the st rength change of the glass fiber because the matrix materials like epoxy and polyurethane are so vulnerable at high temperature. The strength of resin materials start decrease significantly as the temperature start increase. From the figure 2.10 it is ind icated that after around 70 C, the resin materials totally lose their strength. Polymers can be categorized into thermoplastics and thermosets. Figure 2.11. A thermoplastic material is high molecular weight polymer without cross link, that is, in linear or branched structure as shown in (a) and (b). A thermosets has one more thing compared to thermoplastics, the covalent bonds that cross link molecular chains as shown in (c). Due to the cross links, thermosets materials can only be processed once while thermoplastic materials can be processed again by heating to the proper temperature. Nowaday, thermoset polymers are more commonly used for GFRP as the resin in engineering industry. During engineering production of thermosetting resin, the thermoset materials are usually liquid or malleable prior to curing and designed to be moulded into their final shape.
9 2.3 Glass Transition The glass liquid transition (or glass transition for short) is the reversible transition in amorphous materials from a hard and relatively brittle state into a molten or rubber like state. As temperature increases up to a certain level, the cross link of m olecular chains start to break. Therefore, the molecular chains lose the constraints from the cross link and start to slip to each other. The molecular chains slipping means the deformation of the material which will link to an obvious decrease of the mate rial E modulus, starting from the onset temperature of glass transition. A variety of methods of measurement have been devised, based mainly on measurements of the coefficient of expansion as a function of temperature. 1. Differential Scanning Calorimetr y Figure 2.12&13 (DSC) this is probably the most traditional and common technique for polymers. DSC applies a heat flow technique and compares the amount of heat supplied to the test sample and a similarly point. T g is typically calculated by using a half height technique in the transition region. Like the name of this method, difference in the amount of heat required for increasing the temperature of a sample and reference is measured as a function of temp erature. When the state of the material changes, the heat content, which means the energy required for increasing unit temperature alters. In this case, the temperature where the change of the heat content happen s is the glass transition point.
10 2 Thermal Mechanical Analysis (TMA ) TMA is used to measure coefficient of thermal expansion of polymers. Basically, TMA is the measurement of a change of a dimension or a mechanical property of the sample while it is subjected to a temperature. In Thermal mechani cal analysis, a constant usually small load is applied on the sample. The measurement expansion of the specimen can be used to determine the coefficient of linear thermal expansion. As the temperature increase, the expansion coefficient will increase gradu ally until the temperature reach the point of glass transition. At the glass transition point, the expansion coefficient changes significantly. 3. Dynamic Mechanical Analysis (DMA) DMA is probably the most sensitive and accurate method for T g determina tion. It measures the response of the specimen to an applied cyclic strain as the temperature increases gradually. How the material response varies with the temperature is recorded and plotted in graph using a variety of factors. Several typical approaches for reporting T g includes: Onset of the storage modulus curve; Peak of the loss modulus curve; Peak of the Tan Delta curve. E Peak: Occurs at the middle temperature and is more closely related to the physical property changes attributed to the glass transition in plastics. It reflects molecular processes and agrees with the idea of T g as the temperature at the onset of segment al motion. Tan Delta Peak: Occurs at the highest temperature and is used historically in literature.
11 states of a polymer. The height and shape of the tan delta change syste matically with amorphous content. 2.4 Decomposition When the temperature exceed glass transition point and goes even higher, the primary bonds of a thermoset polymer are broken and the material decomposes. This process is called decomposition process d uring which the mass of the polymer changes significantly. Thermogravimetric analysis (TGA) is commonly used to determine changes in weight in relation to changes in temperature, and has been seen to be a very accurate method to characterize a decompositio n process. During Thermogravimetric analysis, the TGA instrument continuously measures the weight of a samples as it is heated to temperatures of up to 2000 C. As the temperature increases, various components of sample are decomposed. Thus the weight of the composite specimen decreases. 2.5 Effect of Elevated Temperatures on GFRP Materials Nowadays, GFRP rebars are the most common way to use GFRP materials. However, GFRP composites demonstrate different thermomechanical properties compared to conventi onal structural materials such as steel reinforcement. Several studies were conducted to evaluate the static and fatigue behavior of the GFRP reinforced concrete
12 structural elements, but the fire resistance studies are not sufficient. Thus, the coefficient and property data which are used in GFRP structural design are scarce. When GFRP reinforcement is applied in concrete, the difference of the thermal expansion of GFRP rebar and concrete may be a concern. Because the difference may cause the distress of th e GFRP reinforced concrete. ( Gentry and Hussain 1999). As mentioned in previous pages, glass transition temperature is one of the most important factors for the constituents of a GFRP composite. The glass transition temperature is defined as a temperature beyond which morphological changes in the polymetric resin of GFRP take place. The glass fiber usually begin to soften around temperatures of 650 C to 970 C and melt above 1225 C. ( Rostasy 1992 ), while most of the commercially available resins have a glass transition temperature of below 100 C. Once the morphological change of the resin happens, the load carrying capacity of resin decreases significantly. The bonding between resin and glass fiber is broken which causes t he failure of GFRP composite. As the temperature increases, mass variation of GFRP is worth noting. It is measure by TGA, the major weight loss occurred between 300 and 450 C. The drop due to thermal degradation of the polymer can be up to 18%. (Mathieu Robert & Brahim Benmokrane).
13 2.6 Behavior of Concrete Members S trengthened with GFRP at High Temperature Starting from early nineties the possibility to replace steel reinforcement bars with GFRP rebars has been increasingly investigated. Compared to steel reinforcement rebar, the main concern of this composite material is its high temperature performance. There are probably not a commercially available structure material that can insulate fire and stay strong in fire forever, but the time available in a fire before the structure collapses matters. Therefore, what really matter in this case and need to be investigated is not the reaction to fire of GFRP rebar itself but its capability to maintain the load together with concrete when the temperature increases. Large amount of GFRP are used in bridge deck as reinforcement. Rebar embedded in the bridge deck in the direction of the beam are usually lapped in the same longitudinal position. So when the fire hap pened, the deck can usually sustain the load. The real concern is the bond behavior of GFRP rebars and concrete.( Valter carvelli 2013). The Canadian code provides a design procedure in fire situation. Marco Andrea Pisani and Carlo Poggi developed a set of GFRP reinforced concrete bridge deck slabs with a variety of temperatures and lapping scheme of the rebars. The main conclusion they got is that the localized heating temperature generates damage in concrete and partial debonding of GFRP rebars without ca using the collapse of the element. The reinforcement geometry in the overlapping areas is the most important factor that influence the load carrying capacity.
14 The tensile strength of steel is around 800 MPa to 900MPa. However, based on test, the tensile strength of E glass fiber can be up to 3445 MPa. So when the GFRP rebar is used in the tensioning area of the bridge deck slab, the tensile strength of it does not need to be concerned. Figure 2.14. 2.7 Performance Based Fire Safety Design The safety of buildings in fire depends on three main elements: fire prevention, suppression and extinction; evacuation of occupants; structural integrity. Even though the first two factors are non structural, they play an important role to ensure the robu stness and functionality of structure to allow for adequate rescue time. In performance based methods design procedure can be relatively simple because the strength of the member is calculated based on the expected temperatures in concrete and reinforcing steel at the time of the required fire endurance. However, for the structures strengthened with GFRP, it is recommended that ignoring the GFRP completely in the fire. (ACI 440) which could be conservative. Figure 2.15. 2.8 History of Performance Based Cod es Performance Based approaches have been permitted by nearly every U.S. building code in the past 100 years. When human first tried to build a building, there is no code or rule, they just wanted to build something can be used with decent functionality.
15 This endeavor has been guided by the Aristotelian notion that design is a process that seeks a convergence of form and function: a physical means that can support certain human needs or activities, subject to certain conditions and constraints. The concept s of performance based regulations and performance based design code started the fast evolution and attracted the designers in 1970s. In 1970s, people started shifting the t building as an integral instead of a certain amount of structural elements. This huge change did a tremendous progress on the evolution of fire safety design. In 1971, N elson (1971) selected systematic approach to the fire safety analysis during a Federal Building in Seattle. Later that year, Benjamin (1971) of National Bureau of Standards (NBS) presented an event logic diagram of fire safety elements in fire safety analy sis. These two documents from Benjamin and Nelson formed the early frame of systematic analysis approach of building fire design. In 1974, National Fire Protection Association Technical Committee on Systems Concepts for Fire Protection published and event Vaughan Beck (1979) developed a risk assessment model to identify cost effective building fire safety design solutions which can also satisfy the requirement of occupant fire safety. The two key f actors in Beck model: The Expected Risk to Life and the Fire Cost Expectation.
16 In 1980s, Performance Based Fire Safety Design (1980) developed much faster. SFPE symposium on Systems Methodologies and some Application. Ministry of Construction undertakes project on the Development of the Fire Safety Design Method (Japan). The Building Regulations are published for the first time as a performance based document (U.K.). Beck collaborates with the National Research Council Canada (NRCC) on fire risk assessmen t modeling (Australia and Canada). The Building Code for U.K. was modified in 1991. The new thing for this 1991 version was the reference to the use of approved documents. At that time, many of the designers were still reluctant to seek alternative designs because of the lack of guidance. In 1996, The British Standard Institute (BSI) drafted The Application of Fire Safety Engineering Principles to Fire Safety in Buildings. Nowadays, within ASCE 7 design. With Performance based procedure, both the structural and non structural components need to be reliable and not less than the expected load carrying capacity under strength procedure. Figure 2.16. 2.9 The Advantages and D isadvantages of Performanc e Based Design Fire safety design codes can be prescriptive or performance based. Currently, more and more fire safety design codes adopt the performance based design method. Compared to the traditional method, performance based codes are usually less conservative and mo re cost effective which is more environmental and currently
17 designed. Because the performance based design considers the whole building as an integral, it has greater flexibility and more possibility for new technology and materials. For prescriptive based based unusual building is being designed, perform ance based codes are better choice (CAIRD RAMSAY). However, there are some disadvantages of performance based design that worth noting. It requires more design experience by engineers, meanwhile, it increases design costs and time. Because different engi neers may have different opinions towards one problem or a decision that is went through during design, it could bring less certainty of approval and greater time for approval. The design part is more complicated and flexible, thus, it needs more new tools methodologies and a fire engineered approach to maximize the benefits. Figure 2.17&18. 2.10 Fatigue D amage of GFRP The damage behavior of composites is quite complex, especially if compared with the case of homogenous materials. Because of their natur al form, damage in composites have various failure modes( matrix cracking, interfacial debonding, delamination, fiber rupture). Fatigue damage in composites is one of the damage which need to be
18 concerned. It can cause the change of strength which is defin ed as residual strength, the decrease in stiffness, and debonding of fibers from matrix ( Ellyin and EL.Kadi 1990). C. Colombo, F. Libonati and L. Vergani did the research of Fatigue damage of GFRP. The results turned out that 2.11 Fatigue of GFRP R einf orcement As composite materials are increasingly used in infrastructure projects like bridge and building, their durability need to be quantified. For instance, a bridge is usually designed to be used for 50 years or more. Thus, if the GFRP can be durable and maintain a decent load carrying capacity is rather important. As the reinforcement in concrete bridge slab, the GFRP rebars have to withstand vibrant, cyclic, highly dynamic or quasi static interval loads, fatigue has been investigated since 1960s. SN behavior, fatigue damage, the degradation of stiffness and strength are all tested. Using cyclic loading to imitate the long term loading is the m ost common method that is adopted in fatigue research. The number of the cycle ranges, thus there are static failure, low cycle fatigue and high cycle fatigue. However, with respect to fatigue life, several applications face even higher cycle numbers in th e so called very high cycle fatigue range beginning at 10 8 cycles (T. J. Adam). As high cycle fatigue test, a glass fiber reinforced polymer used as a blade material for wind turbines was tested by J. Andersons and J. Korsgaard (1999). The fraction percent age of fibers was 41% based
19 on the test. The longitudinal Elastic modulus E and tensile strength were around 30 2GPa and 672 21MPa. 2.12 Cyclic/Fatigue Loading of Structural M embers Quite a while ago, engineers discovered that if a certain load was repeatedly applied and removed to and from a structure, the structure will break after a certain number of cycles. Like human being, a structure has a life time, too. The process of accumul ating damage and finally to failure due to cyclic loading is called fatigue, an insidious cause of loss of strength. Based on the researches done by other researchers, even when the maximum cyclic stress level applied was much lower than the ultimate stren gth of the structural element, the failure still happens after a certain number of cycles. However, as they reduces the magnitude of cyclic stress, the part would was originall Typically, there are three different kinds of fatigue loading: (1) Zero to max to zero: Where a part which is carrying no load is then subjected to a load, and then, the load is removed. The chain used to haul lugs be hind a vehicle. (2) Varying loads superimposed on a constant load: The suspension wires on a suspension bridge are an example for this case. The load from the dead load of the bridge is the constant load, plus, the live load from the transportation acts a s the varying loads.
20 (3) Fully reversing load: A tensile stress of some value is applied to an unloaded structural element and released, then a compressive stress of the same value is applied to this structural element and released. A rotating shaft with a bending load applied to it is a good example of this type of load. Annie Christine Retika from University of Minnesota Center of Transportation studies have done the research about mechanical fatigue effects on GFRP rebar concrete bond. The experimenta l results indicated that mechanical cycles lowered the stiffness of the specimen in a certain load range. Above the certain ranger, the stiffness degradation is not significant. Some steel elements were also tested with cyclic loading to compare to those G FPR specimens. The results showed that mechanical cycling only affect the stiffness of the specimen within the load range of the mechanical cycles as observed in GFRP specimens. 2.13 Acoustic Emission and Fatigue B ehavior of GFRP Acoustic emission (AE) is the sounds waves produced when a material undergoes stress (internal change), as a result of an external force. For composite material like GFRP, some little failure in some area may happen before the whole structural element fails, which makes acoustic emission a rather good way to identify the modes of failure. Fatigue failure in composites is the net results of the accumulation and interaction of different types of distributed damage. There are three fatigue failure
21 modes: matrix crackling, interface debonding and fiber failure. These three types of failure may happen at the same time on the same structure as a very important part. One kind of failure could initiate a second one as well. The acoustic emission technique has the inherent characteristics of monitoring active processes on line.
22 Figure 2.1 Hand lay up method Figure 2.2 Hand lay up method Figure 2.3 Spray up chopping process Figure 2.4 Spray up chopping process
23 Figure 2.5 & 2.6 Vaccum assisted resin transfer molding Figure 2.7 Filament wind process
24 Figure 2.8 Pultrusion process Figure 2.9 GFRP rebars Figure 2.10 Thermal Mechanical properties of Materials
25 Figure 2.12 & 2.13 Differential Scanning calorimetry(DSC) Figure 2.14 Thermal mechanical performances at elevated temperatures Figure 2.11 Molecular structure of composite materials
26 Figure 2.15 Thermal properties of glass fibers Figure 2.16 Performance based code flow chart Figure 2.17 Performance based fire safety design flow chart
27 Figure 2.18 Performance based fire safety design Figure 2.19 The influence of the heating rate on measured glass transition temperature
28 3. Experimental P rogram 3.1 GFRP Coupon Tensioning Tests with Elevated T emperatures 3.1.1 Coupons D esign Sixteen coupons of GFRP were made into dumbbell shape with each dimen sioned as following Fig 3.1. Based on the information from Strongwell which is the producer of the GFRP materials, the longitudinal tensile stress of GFRP coupon should be approximately 207 MPa. However, the loading capacity of the available MTS grips is 10kN. In order to make the section of the coupon small enough so MTS machine can break it before the load run out of the limitation of MTS, f igure 3.1 the section of coupons were made into 6.2mm*3.2mm. Besides, the failure point of the coupons during tensile tests need to be controlled at the mid point so that the laser extensometer can measure the deformation of the coupons. And if the coupons fail at the end sides of it, there is going to be a problem because the grips may influence the tests result. That is why the coupons were designed into a dumbbell shape. And the length of the wider part of the dumbbell were designed to be 70mm so that th e attached area between the grips and the coupons are big enough to provide friction that larger than the tensioning force.
29 3.1.2 Preparation The GFRP coupons were tested in four categories under four elevated temperatures: 25 C, 75 C, 125 C and 175 C. For the coupons tests in 25 C category, no temperature by furnace was applied because 25 C is very close to the normal temperature in the lab. For the coupons tests in 75 C, 125 C, 175 C, a furnace was used cooperated with MTS tester. All the cou pons in these three categories were exposed to the certain temperatures for 30 minutes to make sure the GFRP coupons are heated thoroughly. Because the coupons would expand due to the increase of the temperature, the expansion will influence the deformatio n measurement result from the laser extensometer. Thus, the pre heat procedure lasted 30 minutes until the attached on the coupon for laser extensometer. The MTS LX laser extensometer was used to measure the deformation of the coupons during the tensile tests. The MTS family of LX laser Extensometers are highly accurate, non contact extensometer. The LX measures extension or strain by scanning the specimen and detecting the location of reflective tape markings. As the marks move during the test, the laser tracks and records their exact position. Wintest 7 software was applied to record the load, displacement, stress and strain data during the test.
30 3.1.3 Test Setup and Test P rocess The specimens were positioned to the MTS machine for mechanical tens ile tests. As shown in Fig 3.2 The GFRP coupons attached with reflective tapes in the mid were gripped by MTS we dge action grips The laser from laser extensometer was set t o be pointing at the reflective tape on the coupon. After being exposed to the certain temperatures applied by furnace, the strain reading on the laser extensometer should stay constant which indicates that the GFRP coupons were thoroughly heated. After th e setup and pre test procedure, the tensile tests started. The monotonic tension was applied at a loading rate of 0.01mm/sec until the specimen fail. The cross section property, temperature, strain, stress, load and deformation were recorded by Wintest 7 s oftware. 3.2 Dynamic Mechanical Analysis Dynamic Mechanical Analysis (DMA) Fig 4.1 was used to locate the glass transition temperature of the GFRP. The viscoelastic property of GFRP was studied by dynamic mechanical analysis where a sinusoidal force is applied to, and the response was measured and recorded. The sinusoidal force was applied through the rod placed in the mid point of the GFRP strip. The sensor on the bottom of the support detected and recorded the response of GFRP strip. The data recorded includes stiffness, loss
31 stiffness, storage stiffness, modulus, loss modulus, storage modulus, damping and phase angle. After the specimen and instrument were set up, the furnace started heating. From 30 C to 230 C, in every other 5 C, DMA test was pe rformed. The change of all the parameters were recorded by Wintest 7. It took the furnace about 215 seconds to heat the specimen up every each 5 C. The DMA tests was done using the ElectroForce 3200 Series III test instrument. 3.3 Compression Tests of GFRP Columns with Different D esigns 3.3.1 Tests S pecimens Fifty GFRP square tubes were made from GFRP with 10 different designs: (1) T3 PC as shown in Fig 3.4 (e) was designed to be plain square tubes in 6ft with the section of 3in*3in which is 76.2mm* 76.2mm. The thickness of the four sides is 0.25in which is 6.35mm. (2) T3 2BSB as shown in Fig 3.4 (a) consists of one 4ft square tubes and one 2ft square tubes. The cross section of them are also 76.2mm*76.2mm. One column with section of 63.5mm*63.5mm wa s in the mid of these two tubes for connection. Two ASTM A325 Portland bolts were used for connection at the splice. (3) T3 4BSB as shown in Fig 3.4 (c) is very similar to T3 2BSB. The only difference is that the two parts of T3 4BSB were connected using four A325 bolts.
32 (4) T3 AH1S as shown in Fig 3.4 (g) is 6ft square tubes with section of 76.2mm*76.2mm. The thickness is 6.35mm. There are two access holes with the diameters of 44.5mm on one side of the GFRP square tube. (5)T3 A H2S as shown in Fig 3.4 ( i) is 6ft square tubes with section of 76.2mm*76.2mm. The thickness is 6.35mm. There are two access holes with diameters of 44.5mm on two opposite sides of the GFRP square tube. (6)T35 PC as shown in Fig 3.4 (f) has the same design as T3 PC with a differe nt size of 88.9mm*88.9mm. (7)T35 2BSB as shown in Fig 3.4 (b) has the same design as T3 2BSB with a different size of 88.9mm*88.9. (8)T35 4BSB as shown in Fig 3.4 (d) has the same design as T3 4BSB with a different size of 88.9mm*88.9mm. (9)T35 AH1S as s hown in Fig 3.4 (h) has the same design as T3 AH1S with a different size of 88.9mm*88.9mm. (10)T35 AH2S as shown in Fig 3.4 (j) has the same design as T3 AH2S with a different size of 88.9mm*88.9mm 3.3.2 Setup and I nstruments The specimens were all in 6ft and have two holes for bolting near both ends of the columns. ASTM 325 bolts were used to fix the columns on fixtures. The ASTM A325
33 diame used in structural connection particularly therefore have short thread lengths than regular hex bolts. Because of the way the columns were fixed, the boundary condition should be considered as fixed fixed. MTS 270kN was used to perform the compression tests of the GFRP columns. 3.3.3 Tests P rocedure Fig. A compression test program was developed first for the experiment using the terminal controller of MTS test instrument. The loading procedure is displacement controlled as 2mm per minute. At the end of the test when the failure happened, the load dropped due to the crush of GFRP columns. Based on the experiment program in terminal, the failure was defined to be the point wh en the load dropped to 20% of the peak load. Thus, the MTS stopped loading once the failure was detected. All the axial displacement and loading data were recorded by the sensor in MTS mechanical tester. All the columns in different categories were tested in the same way to keep the consistence of the whole tests.
34 3.4 Compression Tests of GFRP Columns with Different Designs in 1.22m under Elevated T emperatures 3.4.1 Design of S pecimens Five types of GFRP columns were tested with exposure to elevate d temperatures. (1) T3 1.22 PC was designed to be plain square tubes in 4ft which is 0.61 m with the section of 3in*3in which is 76.2mm*76.2mm. The thickness of the four sides is 0.25in which is 6.35mm. (2) T3 1.22 2BSB consists of two same square tubes in 2ft which is 0.61m. The cross section of them are 76.2mm*76.2mm. One column with section of 63.5mm*63.5mm was in the mid of these two tubes for connection. Two ASTM A325 Portland bolts were used for connection at the sp lice. (3) T3 1.22 4BSB consists of two same square tubes in 2ft which is 0.61m. The cross section of them are 76.2mm*76.2mm. One column with section of 63.5mm*63.5mm was in the mid of these two tubes for connection. Four ASTM A325 Porland bolts were used for connection at the splice. (4) T3 1.22 AH1S is plain square tubes in 4ft which is 1.22m with two access holes with the diameter of 44.5mm on one side of the column near the end of it. The cross section of the columns are 76.2mm*76.2.
35 (5) T3 1.22 AH2S is plain square tubes in 4ft which is 1.22m with two access holes with the diameter of 44.5mm on two opposite sides of the columns near the end of it. The cross section of the columns are 76.2mm*76.2mm. Only one specimen was tested under one temperature p oi nt for each kind of GFRP column. 3.4.2 Setup and I nstrument s This phase of the experiments used the same MTS mechanical experimental instrument which was used in the normal temperature compression tests mentioned in previous section. The columns in this section were also fixed on the fixtures with two A325 bolts on each side of it. The boundary condition of it was fixed fixed. Flexible silicon rubber fiberglass insulated heating pads were used to provide the columns elevated temperatures of 75 C, 125 C and 175 C. The heating pads which were used are two silicon rubber heating pads in the same size of 25mm*122mm. It has the advantages like lightweight, thin, and flexible. The power of this heating pad is 10W/in 2 so columns can be exposed to high temper atures rapidly and be warmed up to the certain temperatures soon. Operation temperature of the heating pad can be up to 232 C. Because the temperature in this test can be up to 175 C, fire resist wires were used to wrap the heating pad round the GFRP colum ns after it was installed and fixed on the MTS. Thermal coupon was used to measure the actual temperature of the columns.
36 All the temperature data was recorded by the thermal coupon terminal. At the end of the test, the temperature time graph was plotted. 3.4.3 Tests P rocedure The same test program which was used in the previous section test with the loading rate of 2mm per minute. However, it take some time for GFRP to be thoroughly heated up to a certain temperature. Thus, the specimens were heated for 30 minutes under the certain temperatures before the MTS start loading. Manual control was the way to keep the heating pad maintaining a stable tem perature. When the temperature went higher than the proposed temperature, the heating pad was turned off until the temperature read from the thermal coupon terminal is 3 C lower than the target temperature. Therefore, in the temperature time graph, the tem perature is kept to be 75 C C, 125 C C, 150 C C. Although the error exists, it is relatively acceptable. The temperature of the columns were manually controlled to be like this during the whole tests. The compressive loading and axial displacement was recorded by MTS terminal.
37 3.5 Compression Test of Plain GFRP Columns in 0.61m under Elevated T emperatures These specimens were tested in the same way as previous compression test. However, the length of the GFRP specimens were 0.61m. The columns were wrapped in a spiral way using heating pad in the size of 25mm*122 mm Totally 16 same GFRP columns were tested in this compression test. 3.6 Cyclic Loading C om pression Test of Plain GFRP Columns in 0.61m under Elevated T emperatur es 3.6.1 Specimens and S etup To figure out the influence of cyclic loading with elevated temperatures on the load capacity and other mechanical properties of GFRP columns, twelve GFRP square tubes were tested subjected to cyclic loading with elevated te mperatures. Twelve specimens were all GFRP square tubes with cross section of 88.9mm*88.9mm*6.35mm fixed on MTS using A325 bolts. The length of the specimens were 0.61m. The elevated temperatures were applied in the same way using silicon heating pad.
38 3. 6.2 Experiment P rocedure Based on the experimental results from the compression test of GFRP columns with the same length of 0.61m, average compressive strengths under different temperatures were determined. A new testing program in MTS was created to apply a cyclic loading to the se GFRP specimens. Before the tests start, the columns were exposed to a certain temperature using silicon heating pad for 30 minutes to make sure that the columns were thoroughly warmed up. Then the specimens were started subjected to a compressive load w ith the rate speed as 2mm per minute. The fixture on the bottom kept moving 2mm per minute upward, meanwhile, the compressive load to the column increased. At the point when the compressive load reached 10% of the average compressive strength of the 0.61m columns from previous tests, the loading stopped and released back to zero. The first cycle ended. Then the loading cycle started again until the compressive load reached 20% of the average compressive strength of the 0.61m columns from previous tests, the loading released. The cyclic load was applied repeatedly until the failure happened. During the whole tests procedure, the temperature of the columns were measured and recorded with a thermal couple and a thermal couple terminal.
39 Figure 3.1. The coupon specimen with cross section as 6mm*3.2mm (a) (b) (c) (d) Figure 3.2 Failure modes of coupon tensioning tests under elevated temperatures: (a) at 25 C; (b) at 75 C; (c) at 125 C; (d) at 175 C
40 (a) (b) (c) (d) Fig ure 3.3 Failure modes after manually separated Table 3.4 Mechanical properties of GFRP columns Property Value Width 76.2 mm 88.9mm Thickness 6.35mm Cross section area 1774.19mm 2 2096.77mm 2 Tensile strength ( longitudinal) 207 MPa Tensile modulus 17.9 GPa Ultimate strain 0.012 Coefficient of thermal expansion (longitudinal) 1.210 5 mm/mm/ C
41 (a) (b) (c) (d)
42 (e) (f) (g) (h)
43 (i) (j) Figure 3.4. GFRP column designs in Phase one compression tests: (a) T3 2BSB; (b) T35 2BSB; (c) T3 4BSB; (d) T35 4BSB; (e) T3 PC; (f) T35 PC; (g) T3 AH1S; (h) T35 AH1S; (i) T3 AH2S; (j) T35 AH2S
4 4 4. Results Analysis and D iscussion This chapter presents the results and evaluation of experiments described in Chapter3. 4.1 DMA A nalysis of GFRP 4.1.1 Storage M odulus The storage m odulus ( into the stiffness of a materi al. As mentioned in the previous chapter, Storage Modulus was detected in the dynamic mechanical analysis using Bose Electro Force instrument. The curve of Storage Modulus temperature was plotted as Fig. This curve is useful in accessing the molecular b asis of the mechanical properties of polymer composites since it is very sensitive to structural changes such as molecular weight of polymer matrix, cross linking density and fiber/matrix interfacial bo nding as well. As shown in Fig 4.2 respect to temperature. However, at the temperature around 115 C, a general declining trend up to 135 C was observed, and then specimen went above 135 C, i.e. rd ASTM E 1630, the glass transition
45 rapid reduction occurring in the region between 115 and 135 C corresponding to glass transition temperature of GFRP. In general, the T g is dependent upon chain flexibility and the free volume associated with the chemical structure. 4.1.2 Loss M odulus of sinusoidal deformation. Loss modulus is a visco elastic response of the material which is extremely sensitive to molecular motions. Loss Modulus: ical analysis was shown in Fig 4.2. As shown in Fig 4.2 atures but rises to a peak at around 130 C, which corresponding to the maximum heat dissipation per unit deformation. Onset of loss modulus indicates viscoelastic behavior of resin, on the other hand, peak of loss modulus characterizes crosslink density, degradation and stability. temperature. The standard ASTM D 4065 also suggested that glass transition temperature should be taken as the peak of the loss modulus. Although there is a different between the glass transition temperature got from storage modulus curve and
46 4.1.3 Damping Factor (T an D elta) important factor that characterize the damping property of materials. As shown in Fig 4.2 the damping factor was fairly stable at low temperature but increased to the peak at around 140 C. The onset of Tan delta is quoted as glass transition temperature. 4.1.4 Comparison of Glass Transition Temperatures and S ummary Glass transition temperatures of the GFRP specimen as evaluated from the storage modulus, loss modulus and Tan delta curves are respectively 115 C, 130 C and 140 C. As it can be seen, due to the differences in methodology, the glass transition te mperature of GFRP materials by dynamic mechanical analysis may vary significantly. From the experimental results, it was revealed that the most conservative method is to take the glass transition temperature as onset point of loss modulus, while the least conservative method is the Tan delta peak. In the structure design, the most conservative result should be adopted. However, for research purpose, the glass transition temperatures from different methods should all be considered.
47 4.2 GFRP Coupons T ensionin g T ests 4.2.1 Load vs. Axial D isplacement B ehavior Fig 4.4 presents the load vs. axial deformation curves of all GFRP coupons from each one of the series. The load a xial displacement curves of those coupons under temperature of 25 C show that the whole tensioning procedure can be considered to be linear elastic since the line is almost straight. As it can be seen, the tangent of the line at the beginning were relatively larger. That is because the coupons were fixed on the MTS grips and the displacements were measured through the laser flection from the laser sticker which were put at the mid of the coupons. At the beginning of the tensioning procedure, the surface at the gripping part of the coupons deformed under the compression o f the grips. The way the grips are design is that when there is a tensioning force between these two grips, there will be a compression force appear between the grips the specimen so that the friction be big enough to hold the specimen against the huge te n sion force. Thus, as deformations happened at the gripping part on the ends of the coupons, the deformation at mid where the deformation was measured axial displacement cu rves. As shown in Fig 4.4 (a), the ultimate tensioning streng th of GFRP coupons was around 6 kN under 25 C with the average deformation approximately 2.6mm. There were some difference between these three curves in this series under 25 C. However, as a
48 kind of composite material, the difference between the ultimate tensioning strengths of different specimens were total ly acceptable. As shown in Fig 4.4 (b), the ultimate tensile strength of GFRP coupons under 75 C was around 5kN which decreased 16%from the on e under 25 C. The axial displacement was around 2.5mm. What is the modulus of elasticity was stable when the temperature increased from 25 C to 75 C, the axial displacem ent should have decreased in proportion to the reduction of tension strength. It is indicated from above, the modulus of elasticity changed when the temperature increased from 25 C to 75 C. As mentioned in the discussion of the dynamic mechanical analysis, increased obviously before the temperature reached the glass transition temperature. This can explain why the change of tension strength and axial displacement of GFRP coupons were not in proportion to eac h other even when temperature only changed in a range lower than the tested glass transition temperature. As shown in Fig 4.4 (c), the t ensioning strength was around 4 kN under 125 C. Based on the dynamic mechanical analysis did in the previous section, the glass transition temperature of GFRP is approximately 125 C. The tension strength of the GFRP coupons decreased from 5kN to 4kN which is 20%, meanwhile, the axial displacements increased to somewhere around 4mm. Since when the temperature is at glass t ransition point, the bonding between polymer resin and glass fiber is usually broken, the modulus of elasticity decreases tremendously. Thus, compared to the specimens tested under 25 C and 75 C, the specimens tested under 125 C has a very
49 obvious differen t change on the modulus of elasticity. As it can be seen, the ultimate tensile loads at different temperatures from 25 C to 125 C had a variation that was obvious (33%). At higher temperature up to 175 C, the ultimate load decreased much faster compared to the specimens tested in lower temperatures. The displacements of the specimens in this series were much larger than the low temperature series as well which ranged from 5mm to 10mm. 4.2.2. Stress vs. S train B ehavior As shown in the Fig 4.3 the stress strain curves of the specimen tested under 25 C, the tension procedure is almost perfectly linear elastic. As it can be seen, the average ultimate tensile stress of GFRP coupons under 25 C was around 250MPa. Compared g w ell GFRP specification, Fig 4.3 the results was a little lower than that. However, considering the individual variation between composites specimens, this difference should be acceptable. Based on the ultimate tensile stress and strain, the E, Modulus of elasticity was calculated to be 16.7GPa. Based on the manufacture specification, the Modulus of elasticity is 17.9GPa. The difference between the experimental results and manufacture properties was close. As shown in the Fig 4.3 (b), as the temperature inc reased from 25 C to 75 C, the stress strain curve is not as linear as the specimens under 25 C. The ultimate stress decreased to around 200 MPa from 250MPa (20% decreasing). Based on the stress and strain, the modulus of elasticity of the GFRP coupons unde r 75 C is
50 approximately 13.3GPa. Compare to the modulus of elasticity curve from d ynamic mechanical analysis Fig 4.3 which was 12.2 GPa, the results from coupons tests were relatively larger (9%). As it can be seen that, the variation among the coupon speci mens in the same group that under the same temperature exists. That is because the cross section of the GFRP coupon tested were rather small as 19mm 2 which means for the glass fiber reinforced polymer composite material, the uneven distribution of the gla ss fiber happened during the production of the GFRP makes a big influence on the strength of the coupon specimens with the tiny cross section area. warming up process of the GF RP coupons because all these coupons were placed in the furnace under the specific temperatures for 20 minutes before the tensile tests started. For GFRP coupons that with a tiny size, this 20 minutes warming up time is totally enough in this case. Thus, t he main reason for the stress error should come from the manufacture process. The stress strain curves of the coupons under 125 C were presented in Fig4.3 (c). The average ultimate tensile stress of GFRP coupons under 125 C were measure to be 100 MPa. Base d on the ultimate tensile stress and strain, the modulus of elasticity under 125 C was calculated, 6.7GPa. Compared to the modulus of elasticity from dyna mic mechanical analysis in Fig 4.3 (9GPa), the difference between these two was relatively large as 25 %. One thing should be noticed in the stress strain curve in Fig 4.3 (c) is that, at the beginning of the curve, the tangent was larger, as the strain increased, the speed of the stress increasing was getting lower and lower. On the
51 figure, it was shown as the decreasing of tangent of the curve. In this case, because the temperature was around the glass transition temperature point, materials properties had changed, the parts on the coupon where the grips at were weaker than it was at normal temperature. As it was mentioned, the deformation happened at the gripping part first, however, the deformation was measured at the laser stickers in the mid point, so the tangent at the beginning for the curves were relatively larger. The shapes of the stress strain curves changed most obviously at 175 C as shown in Fig4.3 (d). In this case, the temperature of 175 C is way above the glass transition temperature. Therefore, the tangent at the beginning of the stress strain curves was extremely larg e due to the deformation at the gripping parts caused by the high strain of GFRP coupons under 175 C was reduced to 0.1. Based on the Stress and strain, the modulus of elasti city of GFRP coupons was 10GPa. Compared to the modulus of elasticity from dynamic mechanical analysis, this result was 33% larger. 4.2.3. Failure Modes of Tensile T ests As shown in Fig 4.4 there were two kinds of failure modes that happened to the GFRP coupons. For the coupons tested under 25 C, 75 C and 125 C, the ultimate load were varied, however, the failure modes were the almost the same. At shown in Fig 4.4 the glass fiber is still covered by the resin material, or, the resin fiber bonding wa s still there after the failure of the GFRP coupons. On the contrary, for the coupon
52 tested under 175 C, the fiber could be seen which means the resin fiber bonding was broken. After took these coupons off from the MTS, they were observed and the coupons i n first three groups could be easily separated into two part. Only for the series 175 and post failure observations, the first three series which were tested under 25 C, 75 C and 125 C failed because of the failure of the whole composite material, the coupons tested under 175 C failed because of the resin. Based on the research in the literature review part, the fact is known that the strength of GFRP composite materials comes from th ree components as the strength of glass fiber, the strength of the resin material and the resin fiber bonding. Research on glass transition temperatures of polymer without reinforcement was done by Wei Sun using dynamic mechanical analysis method. Accordin g to his experimental results, the glass transition temperature of plain polymer material varies from 65 C to 75 C respect to t he change of heating rate. On the other hand, the glass fiber has a very good thermal resistance property whose softened tempera ture is above 800 C. Thus, in the first series with 25 C, the coupons failed at the ultimate tensile strength close to the properties in manufacture specification, the series with 75 C, 125 C failed at a relatively lower ultimate strength because the resin fiber bonding was weakened by the higher temperatures. Since the glass fiber only gets soft when the temperature is above 800 C, the reason the coupons were weak at higher temperatures was due to the failure of bonding. As the Fig 4.4 shown, in the series under 175 C, the glass fiber
53 temperature which is 175 C is too high for the resin materials that it nearly melted. The bonding force between the resin and glass fiber almost lost thoroughly. Therefore, enough to break the glass fiber. 4.2.4. Conclusions The thermal tensile b ehavior of GFRP coupons measured by laser extensometer and MTS was investigated in this work. The experimental results were analyzed and the failure modes were observed. The following conclusions were drawn: The modulus of elasticity of GFRP coupons decre ased as the temperatures increased. The ultimate tensile stress decreased as the temperatures increased. And reduction of the GFRP tensile stress changed gradually when the temperature was below the glass transition temperature, the tensile stress of GFRP coupons decreased tremendously at the temperature of 175 C. Based on the knowledge of the glass transition temperatures of polymer resin material and glass fiber, the main reason of the decreasing of the tensile stress for those coupons under 75 C and 125 C was the weakening of the resin fiber bonding. However, according to the failure modes of those coupons under 175 C, the main reason of those coupons failures was the totally platicization of the polymer resin materials.
54 All the data from the experiment s were close to the results from dynamic mechanical analysis with reasonable error. Those errors may due to the variation of the heating rate because different heating furnaces were used in dynamic mechanical analysis and coupon tensile tests. 4.3 Compres sion Strength of GFRP Square T ubes 4.3.1. Compressive Strength and Axial D eformation As introduced in the previous chapter, GFRP square tubes (columns) in 1.83m and the cross section as shown in Fig were with different design were tested with MTS. The ultimate compressi ve strength was shown in Table 4.2 For those specimens with cross section of 76.2mm*76.2mm*6.35mm, the ultimate compressive strength of the PC ( plain columns) group was the largest as 252.63kN. The columns with 4 bolts at the connection parts and those with 2 bolts have the compressive strength which were pretty close: 229.24kN and 233.78kN. The reason that why columns with 4 bolts have a larger compressive strength than the columns wi th 2 bolts may be more bolts made the splice work better and more like a plain columns. For the columns have access holes, the columns with access holes on one side were obviously larger than those with access holes on two opposite sides, two on each side. The T3 1.83 AH1S columns have the larger average compressive strength as 223.92kN, the T3 1.83 AH2S columns have the smaller as 200.96kN. In the tests
55 when the failure happened, almost all the failure modes turned to be crushing, so the key perimeters of the columns should the cross section area. Due to the two more access holes on T3 1.83 AH2S, the cross section area was small, thus, the reduction of the compressive strength could be explained. For those specimens with cross section of 88.9mm*88.9mm*6.35 mm, similar to those specimens with small cross section, the PC (plain columns) have the largest compressive strength As shown in fig 4.15 the load axial displacement plots of the GFRP columns in the compression tests, the tangent of the plots were very close to each other in one tests group. The perimeters that varied in the same plot is the compressive strength of columns with different design. The compressive strength were influenced mainly by the cross section area of the columns. For some of the colu mns tests groups, the compressive load went to zero after failures happened. That is because the program of MTS which controls the compression test has been changed. For the plot whose load ed to detect the failure as the load dropped. However, for the plot whose load axial displacement curve go to zero, the failure is detect once the compressive load decrease 50% of the maximum. As it can be seen in Table 4.2 297.40kN, the average compress ive strength of T35 1.83 PC is the largest among all the specimens. The average compressive strength of T35 1.83 2BSB is 284.14kN and the average compressive strength of T35 1.83 4BSB is 265.24kN (93% of 284.14). So the assumption in the previous section w as wrong
56 since the specimens with more bolts in splice have the smaller compressive strength. The difference between 4BSB and 2BSB specimens should depend on the manufacture error instead of the number of the bolts. One of the governing factors that is con sidered as manufacture errors is the irregular cross section of the splice which will reduce the compressive strength of GFRP columns due to the stress concentration. The average compressive strength of T35 1.83 AH1S and T35 1.83 AH2S are 247.60kN and 258 .60kN. As this was discussed, the failure modes in our tests were almost perfect crushing, thus the tests results should be influenced by cross section area of the specimens. Based on the assumption, the compressive strength of T35 1.83 AH2H should be slig htly smaller T35 1.83 AH1S. However, the experimental results turned to be opposite of the inference. The compressive strength of T35 1.83 AH2S is 258.74kN averagely. The compressive strength of T35 1.83 AH1S is 247.60kN. The reason of this will be discuss ed later in the chapter. 4.3.2 Stress and S train As shown in Fig 4.16 the ultimate strain of the GFRP specimens are approximately 0.06. The compressive stress of the specimens are around 140MPa averagely. According to the property information from man ufacture, the average compressive stress is 207MPa which is 48% more than the experimental results. The average
57 compressive stress of T35 1.83 PC is 141.67kN. The T35 1.83 AH2S turned to be the specimens with the largest compressive stress as 168.86kN. A s shown in Fig, all the plots with compressive strain curves of five different specimens, they look quite similar. T35 AH1S and T35 AH2S are always with the larges compressive strength, plus, the other three as T35 PC, T35 2BSB and T35 4BSB are rather clos e to each other since their stress are close. So, what is the reason that the experimental compressive stress results are smaller than that mechanical properties provided by manufacture? 4.3.3. The Failure Mode of the Compressive T ests As shown in Fig 4.11 all the typical failure modes were recorded. For all the GFRP specimens, there were several kinds of failure modes. In the T35 PC tests, all the failure happened at where the bolts were at. This can totally explain why the compressive stress of GFRP specimen were smaller than the data provided by manufacture. Because of the existing screw holes, the cross section over this point is smaller than the other points with no screw holes. Thus, for the cases that failure happened at bolts, the cross section used in the calculation was too big without the consideration of the reduction due to the screw holes. For T35 1.83 2BSB and T35 4BSB, failure happened at spl ice and bolts. As shown in Fig4.11 (a) and (b), the failure happened at splice. However, the compr essive strength of the specimen which was shown in Fig 4.11 (a) as much larger than that shown in
58 Fig(b). The reason for this was mention in the previous discussion: due to the production error as irregular cross section or the unevenness of the splice, the COV(coefficient of variation) is large. The influence from the stress concentration to very well. For T35 AH1S and T35 AH2S, as it is mentioned, the specimens with access holes on two opposite sides have the larger compressive strength than that with access holes only on one side. As shown in Fig 3.11, the failure modes of all the T35 AH1S were the same. Since the cross section of the specimens have an asymmetrical cross se ction whose cross section area is smaller on one side, the specimens failed into that direction. Because of the untypical cross section, the failure mode is untypical as well. Thus, the experimental results of the T35 AH1S is smaller than that of T35 AH2S. However, when the same assumption goes to T3 1.83 AH1S and T3 1.83 AH2S, the average compressive strength of T3 1.83 AH1S is 223.92k, T3 1.83 AH2S is 200.96kN. The average compressive strength of T3 1.83 AH1S is 150.09MPa, and that of T3 1.83 AH2S is 166 ultimate compression strain and failure modes showed that the asymmetrical failure still influences the compression stress in this case. The influence might not be that tremendous as it was for T35 AH1S and T35 AH2S, but it still exists and reduced with a cross section with smaller side length.
59 4.3.4 Summary This section presents a study GFRP square tubes with two different cross sections. Specimens with five different designs were tested using MTS co mpression program. The numerical findings are summarized as follows: (1) Square tubes made of glass reinforced polymer, GFRP, these columns had characteristics of light weight, great corrosion resistance, high axial stiffness to weight and strength to weig ht ratios. (2) The GFRP assembled with two splices using ASTM325 bolts have relatively lower strength compared to the GFRP plain columns. However, if the splice cross section is even, and the influence from stress concentration is not obvious, the strengt h is acceptable. (3) The experimental results show that the strength and stress were influenced by the type of the failure mode. The columns with access holes on one side has a potential risk of the strength reduction due to the asymmetrical cross section. It is better to have access holes on two opposite sides. (4) The agreements between the experimental results and the mechanical properties information provided by manufacture are decent. Because the failure location varies, the failure cross section area changed as well. Over all, very good agreements between manufacture data and experimental results were found. (5) The main failure modes of all the GFRP specimens in this phase of the experiments were crushing failure. Although some partial bucking happen ed due to
60 the asymmetry of the cross section, the governing failure mode was still crushing. Therefore, the most important factor of the compressive strength is the cross section area. 4.4 Compressive Tests under Elevated T emperatures 4.4.1 Strength and A xial D isplacements As shown in Table 4.3 the length of the specimens were 1.22m with the same design as the specimens in the previous section. Compared to the specimens in 1.83m, the specimens with length of 1.22m have very similar results in the com pression tests under normal temperatures. That should be acceptable because all the failure modes were crushing for which the cross section area is the main governing factor to the compressive strength. As it can be seen in the Table 4.3 the specimen T3 1 .22 PC 1 was not tested because of the deficiency of the GFRP materials. However, the results of T3 1.22 2BSB 1, T3 1.22 4BSB 1, T3 1.22 AH1S and T3 1.22 AH2S 1 were very close to the results from the previous tests. The temperature record of these specime ns during the experiment al procedure were shown in Fig 4.18 The temperature of the GFRP specimens were maintained quite close to the target temperature. As shown in Fig 4.18 the compressive strength of the GFRP specimens decreased as the temperatures inc reased. For the plain GFRP specimen, the specimen under 25 C
61 was not tested. However, the compressive strength of specimen which was tested under 75 C was as large as 213.60kN. Based on the glass transition temperature from previous dynamic mechanical anal ysis, the compressive strength should decrease very fast when the temperature is above around 125 C. Therefore, the strength of the specimen tested under 75 compressive will decrease tremendously because of the glass transition. But when the temperature went to 125 C, the compressive strength decreased to 51kN. According to the fact that the glass transition temperature of GFRP is around 125 C, the results here make sense. Because 125 C is just at the glass transition temperature, the glass transition should be just start. But in experiments, the results is much lower than the expected results based on the compressive strength provided by the manufacture and the glass transition tempera ture. This error may be caused by the error when the temperature was measured by the thermal couples. When the temperature of the specimens were measured during the experiments procedure, two temperature sensors were used, placed in two different place on the specimens. The larger number among these two reading from both temperature sensors will be adopted as the recorded time. Thus, the error from this temperature measuring method maybe big. Therefore, in the plain GFRP specimens tests, when the temperatur e of the thermal coupons showed 125 C, the real temperature of the GFRP columns could be somewhere higher than that. At least, the decreasing trend of the compressive strength make sense.
62 Based on the experimental results, for T3 1.22 2BSB, the compressiv e strength was 216.70kN. As the temperature increased to 75 C, the compressive strength decreased to be 64.40kN. This is so much lower than the expected compressive strength since the temperature is way below the glass transition point. However, the temper ature was measured using thermal coupons. In the experiments, the reading on the thermal coupons increased very fast to the target temperature which is 75 C. But in order to let he whole GFRP specimen be warm up thoroughly, the heating pad will keep workin g on the specimen for 15 20 minutes. There is one thing that needs to be noticed is the bolts used to connect two splices of the column. Since the temperature is measured using thermal coupons attached on the surface of the columns. Plus, the columns were wrapped with the silicon heating pad. The temperature of the bolts which were pretty far to the attaching thermal coupons may be much higher than the target temperature which was measured using thermal coupons. The testing method made the error happened. H owever, this warmed that when the GFRP materials were used cooperated with steel bolts under relatively high temperatures. Even though the GFRP can stands for the high temperature for some time, the thermal conductivity property of steel will make a big di fference to the compressive strength of GFPR materials under high temperature. As shown in Table 4.3 the compressive strength of the specimens T3 1.22 AH1S was 244.40kN. As the temperature increased to 75 C, the compressive strength of the specimens decr eased tremendously to 136.10kN. However, based on the dynamic mechanical analysis, the glass transition temperature is around 125 C which is way
63 larger than 75 C. The reason for the big decreasing of the compression strength may be related to the shape of the cross section. Plus, the glass transition temperature of the polymer resin material in GFRP is around 70 C, and 75 C is a little bit higher than the glass transition temperature of the resin material. This cross section may make the specimen more sensi tive to the change of the temperature. Therefore, the specimen T3 1.22 AH1S decreased around 45% as the temperature increased from 25 C to 75 C below the temperature. When the temperature increased to 125 C, the compression strength of T3 1.22 AH1S was 34. 30kN, the compression strength of T3 1.22 AH1S was 28.20kN that only decreased 20% from the strength under 125 C. Since the glass transition temperature of GFRP was tested to be approximately 125 C. It would be more reasonable to say that, in this case, th e governing factor of the compression strength is not the glass transition temperature GFRP based on the experimental results. On the contrary, the glass transition temperature of the polymer resin material should be the key factor. The same reason can al so be used to explain the change of T3 1.22 AH2S with elevated temperatures. As it can be seen, the compression strength under normal temperature was 201.20kN which is slightly smaller than that of T3 1.22 AH1S. Since under normal temperature, the failure modes were just crushing, the compressive strength were only effected by the cross section area. For the specimens tested under 75 C, the compressive strength decreased to 87.60kN. After the temperature reach the glass transition temperature of GFRP, the c ompression strength was very small as 19.40kN. As the temperature increased to 175 C, the compressive
64 strength was only 16.8kN. Just like the cased of T3 1.22 AH1S, the governing factor of the compression strength is not the glass transition temperature of GFRP. As shown in Fig 4.18 the axial displacements of the specimens had a regular pattern as well. For the Fig 4.18 (a)(c)(e) which were the plots of the load axial displacements of T3 1.22 AH1S, T3 1.22 AH2S and T3 1.22 PC, the load axial curves under 25 C and 75 C are almost straight lines. Thus, the compression process could be considered to be linear. For the specimens under 125 C and 175 C, the axial displacements were very small with displacements around 3mm. For T3 1.22 2BSB and T3 1.22 4BSB, the plots pattern was slightly different to the other plots in Fig 4.18 The load and displacements of the specimens under 75 C were smaller due to the connecting bolts. 4.4.2 Stress and S train As shown in Table 4.3 the compressive strains of T3 1.22 2BS B and T3 1.22 4BSB were very close as 122.14 and 128.00 under normal temperature. Although the T3 1.22 4BSB was not tested under 75 C, the fact was still obvious that the number of compressive strength of the GFRP specimens. The compressive strength of T3 1.22 4BSB under 125 C was slightly larger than that of T3 1.22 2BSB. On the contrary, the compressive stress of T3 1.22 2BSB was slightly larger than that of T3 1.22 4BSB.
65 These exp erimental results all proved the number of the bolts used for connection Therefore, for the colum ns with the same cross section area, the larger the strength is, the larger stress will be. For the columns with different cross section area, things will not be the same since strength equals the stress multiplied by the cross section area. For the column s with access holes only on one side, T3 1.22 AH1S, the cross section area is 1491.93mm 2 for the columns with access holes on two opposite sides, T3 1.22 AH2S, the cross section area is 1209.68 2 As it was mentioned in the previous section, the compressio n strength of T3 1.22 AH1S were larger than that of T3 1.22 AH2S since the cross section was the governing factor. However, the ultimate compressive stress of T3 1.22 AH1S under 25 C was slightly smaller than that of T3 1.22 AH2S, and it is larger than tha t under 75 C, 125 C and 175 C. Considering the existing of the error during the experiments process, the ultimate compressive stress of T3 1.22 AH1S and T3 1.22 AH2S should be the close which means the design of ny significant to the failure mode or the ultimate compressive stress of the GFRP specimens. 4.4.3. Failure M odes T he failure modes of all the specimens were crushing. Which is similar to the specimens in length of 1.83m, T3 1.22 AH1S were crushed into the side where the
66 access holes located at. For the specimens tested in this phase with elevated temperatures, the failure happened at the mid point of the columns, the end of the columns and at bolts as well. For most of the columns tested under 75 C, th e compression strength of the material was not large. Thus, when the failure happened, the sound of the failure was not like that sound in the tests under 25 C. At the failure, the polymer resin material seemed to be melted due to the high temperatures. Ba sed on the program setting in the MTS terminal, once the failure was detected, the load and displacement record would stop working. As it can be seen, the deformation of the GFRP specimens was larger in the higher temperature which can also found in the e xperimental results. After the failure of the specimens under higher temperatures, there was no glass fiber came out. On the contrary, under normal temperature, the glass fiber came out after the failure. This indicated that, for the crushing failure under normal temperature, the specimens failed because the glass fiber failed. However, when the temperature is as high as 75 C or maybe higher, the polymer resin material failed before the compressive load reach the compressive strength of the glass fiber. 4 .4.4. Summary In this section, the experimental results including load axial, stress strain and failure modes were discussed. All these discussion are summarized as follows:
67 1. The compression strength of glass fiber reinforced polymer is influenced by t emperatures very much. Its compression strength could decrease to less around 10% of the compression strength under normal temperature. 2. Due to the difference of thermal conductivity between steel and polymer, bolts which are used for connection in GFRP may have a tremendous effect on the compressive strength of GFRP under relatively high temperature. 3. Based on the experimental results and the failure modes, when the temperature is below the glass transition temperature of GFRP which is around 125 C, the reason for the failure of GFRP column is the failure for polymer resin material instead of the glass fiber. If there is another choice of the resin material that with a higher glass transition temperature, the high temperature performance of GFRP would be better than it is now. 4. For the GFRP material manufactured by Strongwell, the property working temperature should be around 75 C. However, if the GFRP material is used cooperated with steel bolts, what need to be noticed is that the temperature of t he steel bolts must be below 75 C.
68 4.5 Compression Tests of GFRP Columns in 0.61m under Elevated T emperature s 4.5.1 Strength Axial D isplacement The length and cross section were shown in Table (). Since all the specimens in the tests were all the same, the average compressive strength and average ultimate compressive stress were calculated. As it can be seen, the compressive strength of specimens under 25 C was averagely 276kN. As the temperature increased to 75 C, the average compressive strength of the specimens decreased to 212kN which is very close to the results in the experiments done in previous section. Since the temperature 75 decent compressive strength at th at moment. However, when the temperature increased to 125 C, which is around the glass transition point of GFRP, the average compressive strength of specimens was 79kN (28.6% of the compressive strength under normal temperature). When the temperature incre ased to 175 C, the compressive strength was averagely 33kN (12% of the compressive strength under normal temperature). As it was discussed, all the data looks very reasonable and accurate. Compared to the experimental results in the previous section, the drop of the compressive strength of specimens was more significant, which make sense. Plus, for T3 1.22 PC tested under 75 C, the compressive strength was around 210 as well. However, for the other
69 specimens with bolts and access holes, the compressive str engths were a lot lower. Based on the specimens got here, the fact that the bolts and access holes made the specimens more sensitive to the change of the temperature was proved. In this case, the compressive strength of specimens had an obvious drop when the temperature increased from 125 C to 175 C. This is also very easy to explain: the glass transition temperature is approximately 125 C, so the compressive strength should change rapidly. As shown in Fig 4.20 the whole compressing process were linear for specimens under 25 C and 75 C. The axial displacement of T35 0.61 tested under 25 C was around 4.5mm. The axial displacement of T35 0.61 tested under 75 C was around 4mm. Since the responses of the specimens under 25 C and 75 C were both linear the displacement should be in proportion to the load. As the temperature increased to 125 C and 175 C, the responses were not linear anymore. As shown in Fig 4.22 and Fig 4.23 the strengths of the specimens under same temperature were close to each oth er. However, the displacement varies obviously. As it can be seen from the Fig 4.23 the T35 0.61 175 2 and T35 0.61 175 3 had very similar strengths and very different axial displacement. Plus, the ultimate axial displacements of the GFRP columns under 17 5C were extremely large compared to the axial displacements under 125C. The average axial displacements of the specimens under 175C were approximately twice of the axial displacements under 125C.
70 4.5.2 Failure M odes As shown in Fig 4.20 (b), for th e specimens tested under 25C, the failures happened at the bolts area. Based on the experimental results in the previous chapters, the failure modes of GFRP columns under normal temperature are all crushing. Therefore, the cross section area is the govern ing factor for the compression strength. Thus, due to the bolts holes on four sides of the GFRP column, the cross section where the bolts holes were at is the most vulnerable cross section where the failures happened. When the temperature went to 75 C, the point of the failures changed to the area at around 1.5 inch far from the bolts. The failure mode was still crushing according to the picture taken after the tests. Plus, one of the differences between crushing and the failure due to the glass transition of GFRP was the s ound. When the crushing failure happened to the GFRP columns, there was always a very loud sound. On the contrary, when the failure happened due to the glass transition of GFRP or high temperature, the sound was small. In this case for the specimens tested under temperature of 75C, the sounds were still loud. So the failure modes under 75C should be seen as the same as the failure modes under normal temperature considering the compressive strength and how it looked like after failure. As the temperature went to 125C, the strength decreased largely from 212kN to 79kN. According to the results from dynamic mechanical analysis, 125C is around the glass transition point of GFRP. So the compressive strength changed tremendously. Besides, as s hown in Fig 4.22 there was no glass fiber appeared outside of the resin material which means glass transition
71 happened in resin mate rial of GFRP. For the specimens tested under temperatures of 175C, the failure modes were very similar as it was for the s pecimens under 1 25C. However, as shown in Fig 4.23 the specimens failed at four sides of the columns instead of only two sides. As it can be seen from the Fig 4.23 the displacement of the specimens under 175C is much larger than it was under 125C. 4.6 Cyclic Loading Compression Test of GFRP Columns in 0.61m under Elevated T emperatures 4.6.1 Effect of Cyclic Loading to the Compressive S trength As shown in Table 4.4 the compressive strength of specimens under elevated temperatures were not obvious ly influenced by the cyclic loading subjected to the specimens. The average compressive strength of specimens T35 Cyclic 25 was 288.7kN, which was very close to the compressive strength of specimens T35 0.61 25 276kN. As the temperature went to 75 C, the a verage compressive strength of specimens T35 Cyclic 75 was 204.4kN. The results were closed to those of T35 0.61 75. The average compressive strength of specimens T35 cyclic 125 was 51.6kN. The difference between the specimens that was subjected to cyclic loading and normal loading were relatively large at the temperature of 125 C. The average compressive strength of spec imens T35 Cyclic 175 was 54.7kN. Compared to the average compressive strength of specimens T35 0.61 175, the compression strength of the
72 cyclic loaded GFRP columns was larger. However, for composite materials like GFRP, the variation of the compressive strength could be large. The reason of the large difference between the compression strength of T35 0.61 125 and T35 cyclic 125 could be the different heating time. As it can be noticed, the heating time for the cyclic loading was much longer than that of T35 0.61 125. Since 125 C was just around the glass transition temperature of the GFRP, the length of the heating time may have a tremendous effect on the glass transition of GFRP. 4.6.2 The Effect of the Cyclic Loading on the Stiffness of GFRP C olumns As it can be seen in th e Fig 4.25 the tangent s of the Load axial displacement curves were not constant which means the stiffness of the GF RP specimens varied after being subjected to cyclic loading. As the more cycles the specimens have been through, the stiffness became smaller. As shown in Fig 4.25 the tangent of the load and axial displacement curve became smaller as the compression test was in process until the GFRP specimens failed. As shown in Fig 4.25 the shaded area stands for the variation of t he stiffness of GFRP specimens. The wider the shaded area is, the larger the variation of the stiffness is. As it can be seen, as the temperature increased from 25 C to 175 C, the width of the shaded area became wider and wider which means under higher temperature, the effects of the cyclic loading to the stiffness of GFRP is more significant than it is under normal temperature.
73 (a) ( b) Figure 4.1 (a) Dynamic mechanical analysis testing machine; (b) The setup of dynamic mechanical analysis, the specimen is 1.72mm*10.46mm*25.4mm.
74 (a) (b) (c) (d) Figure 4.2 Three way to determine the glass transition point; (a) Tan delta; (b) loss modulus
75 Table 4.1 Compression test of Columns in size of 3.5in*3.5in*6ft
76 Table 4.2 Compression test of Columns in size of 3in*3in*6ft
77 (a) (b) (c) (d) Figure 4.3 Stress strain data of coupons at elevated temperatures: (a) at 25C; (b) at 75C; (c) at 125C (d) at 175C (a) (b) 0 50 100 150 200 250 300 350 0.000 0.005 0.010 0.015 Stress(MPa) Strain coupon 25 1 coupon 25 2 coupon 25 3 0 50 100 150 200 250 300 350 0.000 0.005 0.010 0.015 Stress(MPa) Strain coupon 75 1 coupon 75 2 coupon 75 3 coupon 75 4 0 50 100 150 200 250 300 350 0.000 0.005 0.010 0.015 Stress(MPa) Strain coupon 125 1 coupon 125 2 coupon 125 3 coupon 125 4 0 50 100 150 200 250 300 350 0.000 0.050 0.100 0.150 Stress(MPa) Strain coupon 175 1 coupon 175 2 coupon 175 3 coupon 175 4 0 2 4 6 8 10 0 5 10 Load(kN) Axial displacement(mm) coupon 25 1 coupon 25 2 coupon 25 3 0 2 4 6 8 10 0 5 10 Load(kN) Axial displacement(mm) coupon 75 1 coupon 75 2 coupon 75 3 coupon 75 4
78 (c) (d) Figure 4.4 Load displacement data of coupons at elevated temperatures: (a) at 25C; (b) at 75C (c) at 125 C (d) at 175C (a) (b) (c) Figure 4.5 Failure modes of T3 2BSB: (a) fail at bolts 229.65kN; (b) fail at bolts 219.54kN; (c) fail at the end of the column 238.61kN 0 2 4 6 8 10 0 2 4 6 8 10 Load(kN) Axial displacement(mm) coupon 125 1 coupon 125 2 coupon 125 3 coupon 125 4 0 2 4 6 8 10 0 5 10 Load(kN) Axial displacement(mm) coupon 175-1 coupon 175-2 coupon 175-3 coupon 175-4
79 (a) (b) (c) Figure 4.6 Failure modes of T35 2BSB: (a) fail at splice 310.07kN; (b) fail at bolts 243.31kN; (c) fail at bolts 283.15kN (a) (b) (c) Figure 4.7 Failure modes of T3 4BSB: (a) partial bucking at one end of column 231.29kN; (b) fail at splice 251.10kN; (c) fail at one end of column 209.52kN
80 (a) (b) (c) Figure 4.8 Failure modes of T35 4BSB: (a) partial buckling at bolts 276.37kN; (b) fail at splice 259.16kN; (c) fail at bolts 244.81kN (a) (b) (c) Figure 4.9 Failure modes of T3 PC: (a) partial buckling at bolts 235.76kN; (b) fail at bolts 258.65kN; (c) fail at bolts 238.08kN
81 (a) (b) (c) Figure 4.10 Failure modes of T35 PC: (a) fail at bolts 313.72kN; (b)fail at bolts 279.40kN; (c)fail at bolts 290.01kN (a) (b) (c) Figure 4.11 Failure modes of T3 AH2S: (a) at access holes 211.12kN; (b) at access holes 184.83kN; (c) at access holes 199.40kN
82 (a) (b) (c) Figure 4.12 Failure modes of T35 AH2S: (a) at access holes 242.77kN; (b)at access holes 254.96kN; (c) at access holes 260.38kN (a) (b) (c) Figure 4.13 Failure modes of T3 AH1S: (a) at access holes 245.11kN; (b) at access holes 232.16kN; (c) at access holes 234.94kN
83 (a) (b) (c) Figure 4.14 fail ure modes of T35 AH1S: (a) at access holes 221.51kN; (b) at access holes 255.65kN ; (c) at access holes 246.43kN (a) (b) (c) (d) 0 50 100 150 200 250 300 0 5 10 15 Load (KN) Axial displacement (mm) T35-4BST1 T35-AH2S1 T35-2BSB1 T35-AH1S1 T35-PC1 0 50 100 150 200 250 300 0 5 10 15 Load (kN) Axial displacement (mm) T3-4BST1 T3-AH2S1 T3-2BSB1 T3-PC1 T3-AH1S1 0 50 100 150 200 250 300 0 5 10 15 Load (kN) Axial displacement (mm) T3-4BSB2 T3-2BSB2 T3-AH1S2 T3-PC2 T3-AH2S2 0 50 100 150 200 250 300 0 5 10 15 Load (kN) Axial displacement (mm) T352BSB3 T354BSB3 T35AH1S3 T35PC3 T35AH2S3
84 (e) (f) (g) (h) (i) (j) Figure 4.15 Load displacement data from MTS compression tests. 0 50 100 150 200 250 300 0 5 10 15 Load (kN) Axial displacement (mm) T3-4BSB3 T3-2BSB3 T3-AH2S3 T3-AH1S3 T3-PC3 0 50 100 150 200 250 300 0 5 10 15 Load (kN) Axial displacement (mm) T35-AH2S4 T35-4BSB4 T35-2BSB4 T35-AH1S4 T35-PC4 0 50 100 150 200 250 300 0 5 10 15 Load(kN) Axial displacement (mm) T3-PC4 T3AH1S4 T32BSB4 T34BSB4 T3AH2S4 0 50 100 150 200 250 300 0 5 10 15 Load(kN) Axial displacement(mm) T352BSB5 T354BSB5 T35AH2S5 T35AH1S5 0 50 100 150 200 250 300 0 5 10 15 load(kN) Axial displacement(mm) T3-2BSB5 T3-AH2S5 T3-AH1S5 T3-PC5 0 50 100 150 200 250 300 0 5 10 15 Load (kN) Axial displacement (mm) T35 AH2S2 T35 2BSB2 T35 4BSB2 T35 PC2 T35 AH1S2
85 (a) (b) (c) (d) (e) (f) 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T35 4BST1 T35 2BST1 T35 AH1S1 T35 AH2S1 T35 PC1 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T3-4BST1 T3-2BST1 T3-AH1S1 T3-AH2S1 T3-PC1 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T3-4BST2 T3-2BST2 T3-AH1S2 T3-AH2S2 T3-PC2 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T35 4BST2 T35 2BST2 T35 AH1S2 T35 AH2S2 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T3 4BST3 T3 2BST3 T3 AH1S3 T3 AH2S3 T3 PC3 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T35 4BST3 T35 2BST3 T35 AH1S3 T35 AH2S3 T35 PC3
86 (g) (h) (i) (j) Figure 4.16 stress strain data of the specimens 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T3 4BST4 T3 2BST4 T3 AH1S4 T3 AH2S4 T3 PC4 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T35 4BST4 T35 2BST4 T35 AH1S4 T35 AH2S4 T35 PC4 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T3 2BST5 T3 AH2S5 T3 AH1s T3 PC5 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T35 4BST5 T35 2BST5 T35 AH1S5 T35 AH2S5 T35 PC5
87 Table 4.3. Test specimens in 1.22m76.2mm76.2mm at elevated temperatures Figure 4.17 Test specimens in 1.22m76.2mm76.2mm at elevated temperatures 0 50 100 150 200 250 300 0 50 100 150 200 Load (kN) Temperature (C) T3-PC T3-2BST T3-4BST T3-AH1S T3-AH2S
88 (a) (b) (c) (d) (e) Figure 4.18 Load displacement of specimen (1.22m76.2mm76.2mm) at elevated temperatures. 0 50 100 150 200 250 0 2 4 6 8 10 Load (kn) Axial displacement (mm) AH1S-25 AH1S-75 AH1S-125 AH1S-175 0 50 100 150 200 250 0 2 4 6 8 10 Load (kN) Axial displacement (mm) 2BSB-25 2BSB-75 2BSB-125 2BSB-175 0 50 100 150 200 250 0 2 4 6 8 10 Load (kN) Axial displacement (mm) AH2S-25 AH2S-75 AH2S-125 AH2S-175 0 50 100 150 200 250 0 2 4 6 8 10 Load (kN) Axial displacement (mm) 4BSB-25 4BSB-125 4BSB-175 0 50 100 150 200 250 0 2 4 6 8 10 Load (kn) Axial displacement (mm) PC-75 PC-175 PC-125
89 (a) (c) (d) (e) Figure 4.19 stress strain of specimen (1.22m76.2mm76.2mm) elevated temperatures. 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain AH1S25 AH1S75 AH1S125 AH1S175 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain 2B2 5 2B7 5 2B1 25 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain AH2S25 AH2S75 AH2S125 AH2S175 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain 4B25 4B125 4B175 0 50 100 150 200 0.000 0.002 0.004 0.006 0.008 Stress(MPa) strain T3 PC 75 T3 PC 125 T3 PC 175
90 (a) (b) Figure 4.20 (a) Load displacement data of 2ft GFRP columns compression tests under 25 C; (b) Failure modes of columns (a) (b) Figure 4.21 (a) Load displacement data of 2ft GFRP columns compression tests under 75 C; (b) Failure modes of columns 0 50 100 150 200 250 300 0 2 4 6 Load KN Axial displacement(mm) T35-0.61-25-1 T35-0.61-25-2 T35-0.61-25-3 0 50 100 150 200 250 300 0 2 4 6 Load (kN) Axial displacement (mm) T35-0.61-75-1 T35-0.61-75-2 T35-0.61-75-3
91 (a) (b) Figure 4.22 (a) Load displacement data of 0.61m GFRP columns compression tests under 125 C; (b) Failure modes of columns (a) (b) Figure 4.23 (a) Load displacement data of 2ft GFRP columns compression tests under 175 C; (b) Failure modes of columns 0 50 100 150 200 250 300 0 2 4 6 Load (kN) Axial displacement (mm) T35-0.61-125-1 T35-0.61-125-2 T35-0.61-125-3 T35-0.61-125-4 0 50 100 150 200 250 300 0 1 2 3 Load (kN) Axial displacement (mm) T35-0.61175-1 T35-0.61175-2 T35-0.61175-3 T35-0.61175-4
92 Table 4.4. Test specimens in 0.61m 88.9mm88.9mm subjected to cyclic compressi on loading at elevated temperatures. ID Section area (mm 2 ) T (C ) P(kN) uc (MPa) T35 Cy clic 25 1 2096.77 25 243.00 115.89 T35 Cy clic 25 2 25 289.47 138.06 T35 Cy clic 25 3 25 333.62 159.11 T35 Cy clic 75 1 75 211.68 101. 00 T35 Cy clic 75 2 75 212.28 1 01.24 T35 Cy clic 75 3 75 189.10 9 0.19 T35 Cy clic 125 1 125 63.40 3 0.22 T35 Cy clic 125 2 125 39.66 1 8.9 T35 Cy clic 125 3 125 51.73 2 4.67 T35 Cy clic 175 1 175 49.56 2 3.64 T35 Cy clic 175 2 175 58.28 2 7.80 T35 Cy clic 175 3 175 56.47 2 6.93 (a) (b) 0 50 100 150 200 250 300 350 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 25 1 0 50 100 150 200 250 300 350 0 2 4 6 Load (kN) Axial displacement (mm) T35 cyclic 25 2
93 (c) Figure 4.24 Load displacement of GFRP columns subjected to cyclic compression loading at 25 C (a) (b) (c) Figure 4.25 Load displacement of GFRP columns subjected to cyclic compression loading at 75 C 0 50 100 150 200 250 300 350 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 25 3 0 50 100 150 200 250 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 75 1 0 50 100 150 200 250 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 75 2 0 50 100 150 200 250 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 75 3
94 (a) (b) (c) Figure 4.26 Load displacement of GFRP columns subjected to cyclic compression loading at 125 C (a) (b) 0 10 20 30 40 50 60 70 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 125 1 0 10 20 30 40 50 60 70 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 125 2 0 10 20 30 40 50 60 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 125 3 0 10 20 30 40 50 60 70 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 175 1 0 10 20 30 40 50 60 70 0 2 4 6 Load (kN) Axial displacement (mm) T35 Cyclic 175 2
95 (c) Figure 4.27 Load displacement of GFRP columns subjected to cyclic compression loading at 125 C 0 10 20 30 40 50 60 70 0 1 2 3 4 5 6 Load (kN) Axial displacement (mm) T35 Cyclic 175 3
96 5. Numerical M odeling In the experiments which were described in previous chapters, all the failures were crushing failures. To expand the experimental findings, the finite element program Ansys was used to create models of GFRP columns with different length, modulus of elastic ity and thickness as well. Because the columns in the experiments crushed before the buckling happen, the numerical modeling was used to analyze the buckling load and buckling mode of GFRP columns. Depending on the L e /r (the slenderness ratio), columns ca n be short columns or slender columns. For short column, Strength can be computed by considering only the column section properties. However, for long column, axial load capacity is significantly affected by length, loading conditions of column. 5.1 Eule r Buckling F ormula Euler postulated the phenomenon of elastic buckling as: 2 E I / L e 2 P cr = Critical axial load E = Modulus of Elasticity of the columns material r = radius of gyration defined by I = Ar 2 L e = effective length (boundary condition is considered)
97 A plot of the buckling load vs. the slenderness ratio, a so c alled column curve which shows the reduction in critical load with increasing slenderness. The quotient L e /r, known as the slenderness ratio, is an important parameter in the classification of compression members. For a member of sufficiently large slenderness ratio, a long column, buckling occurs at a stress lower than the proportional limit. The Euler formula is appropriate to this case. For some very short c olumns, failure occurs by compression, without buckling, at stresses exceeding the proportional limit. Between these two extreme cases, it is intermediate column. The range of L e /r depends on the material under consideration. A commonly acceptable range is : long columns are those for which L e /r larger than 100; short columns are those with L e/ r smaller than 30; 5.2 Finite Element Analysis M odel The general purpose finite element analysis program ANSYS Mechanical APDL 14.5 was applied to predict the behavior of GFRP columns subjected to axial loads. The Eigen buckling analysis was performed to analyze the buckling load and buckling failure mode. In general, a finite element analysis in ANSYS consists of three steps: 1. Preprocessing: defining the problem. It includes: Define points/lines/areas/volumes; Define element type and material properties; geometry of structure elements. 2. Solution: assi gning loads, constrains and solving;
98 3. Post processing: further processing and viewing of the results; The results includes lists of nodal displacements, element forces and moments, deflection plots and stress contour diagrams. 5.3 The Comparison of Th eoretical Critical Load and Critical L oad from ANSYS To cover all types of columns including short column, intermediate column and long column, columns with L e /r varies from 27 to 150 were modeled using ANSYS. The element type was defined to be 3D 4node 1 81 shell in ANSYS with modulus of Fig(). The thickness of the shell was made the same as the GFRP columns tested in previous chapters which was 0.00635mm. All the models wer e meshed into elements with length of 25.4mm. Because both ends of the columns were fixed in the experiments, to keep the consistence, the boundary condition was defined in a way that one of the end was fixed and another end was constrained on transverse d irection and rotations. The load was applied to uniformly distribute on the bottom cross section of the columns. Models of columns with L e /r as 27, 30, 50, 75, 100, 125 and 150 were created. Based on manufacture brochure, the longitudinal compressive str ess of the GFRP materials
99 is 207MPa, the cross section area of the columns modeled was 2096.77mm 2 The compressive strength when the columns crush can be easily to get: 2096.77mm 207N/mm 2 = 434kN Which means if the buckling load of a GFRP column is larg er than 434kN, the column will crush before it buckle. Euler formula was adopted to determine the theoretical buckling load of these columns as well. From the Fig 5.1 the variation trend of GFRP columns buckling load from ANSYS and Euler formula were sho wn. There was a rather significant difference between the buckling loads from ANSYS and Euler. As the table() shown, At the point that slenderness equals 27, the buckling load determined by ANSYS was 50.6% of the buckling load determined by Euler buckling formula. At point that slenderness ratio equals 50, the buckling load determined by ANSYS was 54.1% of the buckling load determined by Euler buckling formula. At point that slenderness ratio equals 75, the ratio between these two buckling load from differe nt methods was even bigger which means the results were closer. Recall the derivation of Euler Formula: EI ( d 2 y / dx 2 ) + P y = .0 This is the equation the analysis began with. The term d 2 y / dx 2 actually is an approximation to the curvature. As the L e /r increases, the error from d 2 y / dx 2 decreases.
100 Based on the buckling data shown in Fig 5.1 the L e /r limitation is around 30. As shown is Fig5.1 the experimental results from the previous chapter turned to be very close to the ANSYS modeling results. 5 .4 The Buckling Load of GFRP Columns with Different Wall T hickness GFRP columns with wall thickness of 4mm, 7mm, 10mm, 13mm and 16mm were modeled using ANSYS. Due to the change of the wall thickness, the cross section area changed, also the moment of inertia. Based on the calculation, the radius of significantly. In order to do the comparison between short columns and long columns, the length of the columns were defined to be 1.81m and 9.1m which means the effective length of the columns models were 0.91m and 4.55m. The boundary condition and materia l properties were define to be the same as previous modeling. As shown in Fig 5.2 the buckling load of columns were proportional to the wall thickness of the columns. There were a significant difference between the theoretical results from Euler and res ults from ANSYS. As shown in Table 5.1 as the thickness of the wall increases, the difference between the results from Euler formula and ANSYS modeling decreases. As shown in failure mode figure converted from ANSYS, the displacement in the mid caused by buckling increased as the thickness increases. The displacement of the column with thickness of 4mm is 0.276m; the displacement of column with thickness of 7mm is 0.326m; the displacement of
101 column with thickness of 10mm is 0.380m; the displacement of colu mn with thickness of 13mm is 0.418m; the displacement of column with thickness of 1 6mm is 0.437m. As shown in Fig 5.3 the displacement due to the column buckling is not in proportion to the wall thickness of the column. However, the displacement due to bu ckling increases when the wall thickness increases. 5.5 Buckling Load of GFRP Columns with Different Modulus of E lasticity To expand the experimental results, some columns with different modulus of elasticity were modeled. Based on the Euler buckling f ormula, P cr = 2 E I / L e 2 the critical load should be in proportion to E. Thus, columns with the same cross section but different modulus of elasticity were modeled to verify this inference. The L e /r range was set to be from 25 to 150, with the cross section of 82.55mm*82.55mm*6.35mm. As the results shown in Fig 5.3 the buckling load from both ANSYS and Euler formula were in proportion to their material modulus of elasticity. The buckling load sle nderness ratio plots Fig()of the columns with different modulus of elasticity starts from 20GPa to 60GPa all have the same shape as the one with modulus of elasticity of 19.3GPa from manufacture.
102 5.6 Buckling Load of Columns with A ccess holes Two sets of columns were modeled using ANSYS. All the columns in this section have the same cross section of 0.08255m*0.08255*0.00625. One set of the columns have two access holes on one side, another set of the columns have four access holes two opposi te sides, t wo on each side (Figure 5.7) Columns with slenderness ratio from 25, 50, 75, 100, 125, 150 were modeled for eigen buckling analysis. As the results from ANSYS shown in Table(), when two columns have the same length, the column with access holes on two sid es of it has the smaller buckling load then the columns with access holes on only one side. However, the difference of the buckling load is not sign ificant. As it is shown in Appendix the failure happened closed to the mid point of the column. Although th e account of the access hole on the columns different to the buckling load. For the columns with the slenderness ratio of 25, the buckling load difference due to the di fferent account of access holes is relatively larger. Because for short columns, the access holes are relatively closer to the mid point where the bucking failure happened since the distance from the access holes to the bottom end of the columns are certai n. On the contrary, for the columns with larger slenderness ratio, the buckling load difference due to the different account of access holes is ignorable. For the columns with slenderness ratio of 150, the length of the columns is 10.15m which is more than 30 times as the distance from the access holes to the end of the column. Therefore, the access holes has no influence to the
103 buckling load in this case. For the columns whose strength is less than that computed based on section properties, axial load and moment capacities are significantly affected by length, loading conditions of column instead the cross section area. In the ANSYS eigenbuckling analysis, the failure mode were all set to be buckling which is not realistic, thus, for the short columns, the buckling critical load of it is larger than the strength. 5.7 Buckling Analysis of Columns under Elevated T emperatures In order to expand the experimental results and also do the comparison between the experimental results and numerical results, buckling analysis of GFRP columns at elevated temperatures. The size the columns modeled was the same as the columns tested in th e previous chapters. Based on the experimental results of the modulus of elasticity from dynamic mechanical analysis, the modulus of elasticity under elevated temperatures were set to be 19.3GPa, 12.2GPa, 9.0GPa, 7.3GPa respectively at 25, 75, 125 and 175 C. According to the buckling analysis done with ANSYS, the buckling load versus temperat ures figure was plotted in Fig 5.5 As it can be seen, the buckling load of the column at 25C was 276.6kN, the buckling load of the column at 75C dropped to 176kN, th e buckling load of column at 125C was 128kN and the buckling load at 175 C.
104 Figure. 5.1 Critical load determined by Euler formula and ANSYS with the change of L e /r Table 5.1 Critical load determined by Euler formula and ANSYS with the change of L e /r L e /r L e (m) P cr (kN) real length(m) Buckling load from Ansys 27 0.91 546.20 1.83 276.60 50 1.69 159.60 3.38 86.3 75 2.54 70.93 5.07 40 100 3.38 39.90 6.77 23.3 125 4.23 25.54 8.46 15.4 150 5.07 17.73 10.15 11 Figure 5.2 Critical load determined by Euler formula and ANSYS with the change of the wall thickness 0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 400.00 450.00 500.00 0 50 100 150 200 Critical load(kN) Le/r theoritical critical load ANSYS buckling load Experiment result 0.0 200.0 400.0 600.0 800.0 1000.0 1200.0 1400.0 1600.0 0 5 10 15 20 Pcr(kN) Thickness of column wall(mm) Euler Pcr ANSYS Pcr
10 5 Table 5.2 Critical load determined by Euler formula and ANSYS with the change of the wall thickness Thickness of wall(mm) A(mm^2) I(mm^4) r Le/r Euler(kN) ANSYS(kN) 4 1321 1503621 33.7 27.0 345.5 169.7 7 2311 2644050 33.8 26.9 607.6 307.3 10 3302 3784479 33.9 26.9 869.6 455.7 13 4293 4996230 34.1 26.7 1148.1 615.7 16 5283 6225813 34.3 26.5 1430.6 787.2 Fig 5.3 D isplacement wall thickness Table 5.3 Critical load determined by Euler formula and ANSYS with modulus of Elasticity of 19.3GPa E=19.3 GPa Le/r Le(m) Critical load Pcr(kN) Real length L(m) Buckling load ANSYS(kN) 25 0.85 638.47 1.69 320.70 50 1.69 159.62 3.38 86.18 75 2.54 70.94 5.07 39.98 100 3.38 39.90 6.77 23.29 125 4.23 25.54 8.46 15.37 150 5.07 17.74 10.15 10.97 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 5 10 15 20 Pcr(kN) Thickness of column wall(mm) Euler Pcr Euler Pcr
106 Table 5.4 Critical load determined by Euler formula and ANSYS with modulus of Elasticity of 20GPa E=20 GPa Le/r Le(m) Critical load Pcr(kN) Real length L(m) Buckling load ANSYS(kN) 25 0.85 661.63 1.69 332.34 50 1.69 165.41 3.38 89.47 75 2.54 73.51 5.07 41.43 100 3.38 41.35 6.77 24.14 125 4.23 26.47 8.46 15.93 150 5.07 18.38 10.15 11.37 Table 5.5 Critical load determined by Euler formula and ANSYS with modulus of Elasticity of 40GPa E=40 GPa Le/r Le(m) Critical load Pcr(kN) Real length L(m) Buckling load ANSYS(kN) 25 0.85 1323.26 1.69 664.66 50 1.69 330.82 3.38 178.62 75 2.54 147.03 5.07 82.87 100 3.38 82.70 6.77 48.28 125 4.23 52.93 8.46 31.86 150 5.07 36.76 10.15 22.74 Table 5.6 Critical load determined by Euler formula and ANSYS with modulus of Elasticity of 60GPa E=60 GPa Le/r Le(m) Critical load Pcr(kN) Real length L(m) Buckling load ANSYS(kN) 25 0.85 1984.90 1.69 996.99 50 1.69 496.22 3.38 267.94 75 2.54 220.54 5.07 124.30 100 3.38 124.06 6.77 72.42 125 4.23 79.40 8.46 47.80 150 5.07 55.14 10.15 34.12
107 (a) (b) 0.00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 0 50 100 150 200 Axis Title Axis Title E=20GPa Euler buckling load ANSYS buckling load 0.00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 0 50 100 150 200 Axis Title Axis Title E=40GPa Euler buckling load ANSYS buckling load
108 (c) (d) Figure 5.4 Critical load determined by Euler formula and ANSYS with different Modulus of Elasticity 0.00 200.00 400.00 600.00 800.00 1000.00 1200.00 0 50 100 150 200 Axis Title Axis Title E=60GPa Euler buckling load ANSYS buckling load 0.00 200.00 400.00 600.00 800.00 1000.00 1200.00 0 50 100 150 200 19.3GPa 20GPa 40GPa 60GPa
109 Table 5.7 Critical load of columns with access holes on one side determined by Euler formula and ANSYS One side AH Le/r Le(m) Critical load Pcr(kN) Real length L(m) Buckling load ANSYS(kN) 25 0.85 638.47 1.69 316.42 50 1.69 159.62 3.38 86.07 75 2.54 70.94 5.07 39.93 100 3.38 39.90 6.77 23.27 125 4.23 25.54 8.46 15.36 150 5.07 17.74 10.15 10.97 Table 5.8 Critical load of columns with access holes on one side determined by Euler formula and ANSY S two side AH Le/r Le(m) Critical load Pcr(kN) Real length L(m) Buckling load ANSYS(kN) 25 0.85 638.47 1.69 311.86 50 1.69 159.62 3.38 85.94 75 2.54 70.94 5.07 40.04 100 3.38 39.90 6.77 23.27 125 4.23 25.54 8.46 15.32 150 5.07 17.74 10.15 10.97
110 Fig 5.5 Buckling load Elevated Temperatures Fig 5.6 Critical load of columns with access holes on one side (1AH), two sides(2AH) and normal columns without access holes(No AH) determined by Euler formula and ANSYS 0 50 100 150 200 250 300 350 0 50 100 150 200 Buckling load (kN) Temperature ( C) Le/r=25 Le/r=50 Le/r=75 Le/r=100 Le/r=125 Le/r=150 0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 25 50 75 100 125 150 P cr (kN) L e /r 1AH 2AH No AH
111 (a) (b) (c) (d) (e) (f)
112 (g) (h) Fig 5.7 GFRP column model in ANSYS (a)column with Le/r=25; (b) column with Le/r=50;(c) column with Le/r=75; (d)column with Le/r=100; (e)column with Le/r=125; (f)column with Le/r=150 (g ) column with access holes on one side; (h) column with access holes on two sides.
113 (a) (b) (c) (d) (e) (f) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 -0.05 0.00 0.05 10 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0.00 0.20 0.40 25 0 0.5 1 1.5 2 2.5 3 3.5 4 0.00 0.50 1.00 50 0 1 2 3 4 5 6 0.00 0.50 1.00 75 0 1 2 3 4 5 6 7 8 0.00 0.50 1.00 100 0 1 2 3 4 5 6 7 8 9 0.00 1.00 2.00 125
114 (g) Fig 5.8 Deformation of GFRP columns with different slenderness rat io after buckling (a) (b) (c) 0 2 4 6 8 10 12 0.00 0.50 1.00 150 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.00 0.50 TH=4 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.00 0.50 TH=7 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.00 0.50 TH=10
115 (d) (e) Fig 5.9 Deformation of GFRP columns with wall thickness after buckling 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.00 0.50 TH=13 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.00 0.50 Th=16 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0.00 0.50 1.00 1AH25 0 0.5 1 1.5 2 2.5 3 3.5 4 0.00 0.50 1.00 1AH50 0 1 2 3 4 5 6 0.00 0.50 1.00 1AH75
116 (a) (b) (c) (d) (e) ( f) Fig 5.10 Deformation of GFRP columns with different slenderness and access holes on one side after buckling 0 1 2 3 4 5 6 7 8 0.00 0.50 1.00 1.50 1AH100 0 1 2 3 4 5 6 7 8 9 0.00 0.50 1.00 1.50 1AH125 0 2 4 6 8 10 12 0.00 1.00 2.00 1AH150
117 (a) (b ) (c) (d) (e) (f) Fig 5.11 Deformation of GFRP columns with different slenderness and access holes on two sides after buckling 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0.00 0.50 1.00 2AH25 0 0.5 1 1.5 2 2.5 3 3.5 4 0.00 0.50 1.00 2AH50 0 1 2 3 4 5 6 0.00 0.50 1.00 2AH75 0 1 2 3 4 5 6 7 8 0.00 1.00 2.00 2AH100 0 1 2 3 4 5 6 7 8 9 0.00 1.00 2.00 2AH125 0 2 4 6 8 10 12 0.00 0.50 1.00 2AH150
118 6. Summ a ry and Recommendations The overriding purpose of this study was to determine the performance of glass fiber reinforced polymer at higher temperatures. To accomplish that goal it became necessary to reach some prerequisite goals. Determining the mechanical properties of GFPR mate rials at normal temperatures and determining the glass transition temperatures were both done as the basis of the GFRP columns tests under elevated temperatures. Related to these effort, it became necessary to reach an understanding about the GFRP backgrou nd including the manufacture process, application of GFRP material in industry, and also and glass transition temperature of GFRP. Based on the literature review and experimental results, the conclusion was as follow: Nowadays, there are not enough resear ch that have been done about glass reinforced polymer especially on its performance when subjected to higher temperatures. Most of time, glass fiber reinforced polymer is used in bridge deck as GFRP rebar. Due to lack of the knowledge in GFRP on its therma l properties, development of GFRP was limited tremendously. According to the dynamic mechanical analysis of GFRP, the glass transition temperature is approximately 130 C which means the compressive strength of GFRP will decrease significantly once the tem perature reach that temperature point. Over all, the compressive strength of glass fiber reinforced polymer decreases as the temperature increase. Considering the glass transition temperature is around 130 C, the test temperature was set to be 25, 75, 125 175 C. The Strength of glass fiber
119 reinforced temperature decreased when the temperature increased from 25 C to 75 C. However, the drop of the compressive strength was not significant. When the temperature increased from 75 C to 125 C, the drop of the co mpressive strength was relatively larger up to 70% as the temperature almost reach the glass transition temperature. As the temperature increased from 125 C to 175 C, the compressive strength decreased to be only 12% of the compressive strength under normal temperature. Different configurations and different length of the columns were applied in the tests. The results show as follow: The steel bolts used for connect ion in columns makes the columns more vulnerable when the columns are exposed to higher temperatures. The access holes make the columns more sensitive to the change of the temperature. The compressive strength of the columns is not influenced significantl y by the length of the columns. To simulate the fatigue failure, the columns were subjected to cyclic loading. The the compressive strength of the glass fiber reinforced p olymer. On the contrary, the stiffness decreased after the GFRP columns were subjected to the cyclic loading. Based on the experimental results and discussion about GFRP, the GFRP can be used compressive strength totally. However, its fire resistance ability is not good as good as concrete. If there are
120 steel bolts used for connection in the GFRP columns, the steel bolts part should be isolated from the high temperature.
121 R efere nces Turi, Edith A., Thermal Characterization of Polymeric Material, Second Edition, Volume I, Academic Press, Brooklyn, New York, 1997, P. 980. AASHTO, LRFD Bridge Design Guide Specifications for GFRP Reinforced Concrete Bridge Decks and Traffic Railings, American Association of State High way & Transportation Officials, Washington, DC, 2009 Chris P. Pantelides, Ruifen Liu, GFRP Reinforced lightweight precast bridge deck, University of Utah, 2011. Vinod Kumar Vankanti, Venkateswarlu Ganta, Optimization of process parameters in drilling of GFRP composite using Taguchi method, Warangal, AP, India, 2013. David Trejo, Long term performance of GFRP reinforcement: technical report, Texas Transportation Institute, Texas, 2009 M.R. Ehsani, H. Saadatmanesh, S. Tao, Design recommendations for bond of GFRP rebars to concrete, American Society of Civil Engineers, 1996. C. Eugene Buth, Crash Tests Evaluation Performance of GFRP Reinforced Bridge Rail, Project Summary Report 0 4 138 S, Texas transportation institute, the Texas A&M University system, 2003. Haomin Helen Wang, Test of Glass fiber Reinforced Polymer (GFRP) Anchors, the Graduate School of the University of Texas at Austin, 2013. Mathieu Robert and Brahim Benmokrane, Behavior of GFRP Reinforcing Bars Subjected to Extreme Temperatures, American society of civil engineers, Journal of composites for construction, 2010. S.K. Foster and L.A. Bisby, High Temperature Residual Properties of Externally Bonded FRP Systems, Qu SP 230 70, Canada, 2009. ACI (2004) Guide Test Methods for Fiber Reinforced Polymers (FRPs) for Reinforcing of Strengthing Concrete Structures, ACI 440.3R 04, American Concrete Institute. 44pp. Francesco Micelli, Antonio Nanni, Durab ility of FRP rods for concrete structures, Construction and Building Materials 18 (2004), Lecce, Italy, 2004.
122 J.L.V. Coelho and J.M.L. Reis, Effects of strain rate and temperature on the mechanical properties of GFRP composites, CEP 24.210 240, Rua Passo de Patria, 156. Jose Miguel da Costa Pires, Mechanical behavior at elevated temperatures of GFRP pultruded composite profiles, M.Sc. Disseration in Civil Engineering, Tecnico Lisboa, 2012. Benjamin Tang, Fiber Reinforced Polymer Composites Applications in USA, DOT Federal Highway Administration HIBT 10, Korea/USA Road Workshop Proceedings, Washington, 1997. Chowdhury, E.U., Green. M. F., Bisby, L. A., Thermal and mechanical characterizati on of fiber reinforced polymers, concrete, steel, and insulation materials for use in numerical fire endurance modeling, Structures under Extreme loading, Proceedings of Protect 2007, Whistler, B.C., 2007, pp. 1 10 Valter Carvelli, Marco Andrea Pisani, C arlo Poggi, high temperature effects on concrete members reinforced with GFRP rebars, Department of Architecture, Built Environment and Construction Engineering, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133, Milano, Italy, 2013. ACI (2004) G uide for the Design and Construction of Structural Concrete Reinforced with FRP Bars, ACI 440.3R 04, American Concrete Institute. Green, M. F., Benichou, N., Kodur, V.K.R., Bisby, L.A, Design guidelines for fire resistance of FRP strengthened concrete st ructures, Eighth International Conference on FRP in Reinforced Concrete Structures(FRPRCS 8), Patras, Greece, 2007, pp. 1 10. Brian J. Meacham, The evolution of performance based codes and fire safety design methods, Building and Fire research Laboratory of the National Institute of standards and Technology, Bethesda, MD, 1998. G Caird Ramsay, Fire Safety Engineering: Role in performance based codes. Scientific Services Laboratory, Australian Government Analytical Laboratories, Victoria, Australia, 1999 TP Sathishkumar, S Satheeshakumar and J Naveen, Glass fiber reinforced polymer composites a review, Journal of reinforced plastics and composites 2014 33. Kishore V. Pochiraju, Gyaneshwar P. Tandon, Gregory A. Schoeppner Editors, Long term Durabilit y of Polymeric Matrix Composites. ISBN 978 1 4419 9307 6. 2012.
123 Garret KW, Bailey JE. Multiple transverse fracture in 90_ cross ply laminates of a glass fibre reinforced polyester. J Mater Sci 1977;12:157 68. Kim HC, Ebert LJ. Fatigue life limiting parameters in fibreglass composites. J Mater Sci 1979;14:2616 24. Lorsch P, Adam TJ, Sinapius M, Horst P. Challenges for the experimental investigation of VHCF of polymer composites. In: Proceedings of the 5th Int. conference on very high cycle fatigue VHCF 5, Berlin; 2011. Hosoi A, Arao Y, Kawada H. Transverse crack growth behaviour considering free edge effect in quasi isotropic CFRP laminates under high cycle fatigue loading. Compos Sci Technol 2009;69:1388 93. Hosoi A, Sato N, Kusumoto Y, Fujiwara K, Kawada H. High cycle fatigue characteristics of quasi isotropic CFRP laminates over 108 cycles (initiation and propagation of delamination considering interaction with transverse cracks). Int J Fatigue 2010;32:29 36. Adam TJ, Horst P. Investigation of the very high cycle fatigue behavior of GFRP cross ply bending specimens. In: Proceedings of the 16th European conference on composite materials ECCM16, Seville; 2014.
124 A ppendix (a) (b) 0.00 20.00 40.00 60.00 80.00 100.00 120.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 75 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 0.00 500.00 1000.00 1500.00 2000.00 75
125 (c) (d) 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 3000.00 75 0.00 20.00 40.00 60.00 80.00 100.00 120.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 75
126 (e) (f) 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 0.00 500.00 1000.00 1500.00 2000.00 125 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 125
127 (g) (h) 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 75 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00 200.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 3000.00 175
128 (i) (j) 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 75 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00 0.00 500.00 1000.00 1500.00 2000.00 125
129 (k ) (l) 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 0.00 500.00 1000.00 1500.00 2000.00 125 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 125
130 (m) (n) 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 0 500 1000 1500 2000 75 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00 200.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 175
131 (o) (p) 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00 200.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 175 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00 200.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 175
132 (q) (r) 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 0 500 1000 1500 2000 2500 75 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 0.00 500.00 1000.00 1500.00 2000.00 125
133 (s) Figure A 1 Temperature Time recorded by thermal coupon termina l 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 0.00 500.00 1000.00 1500.00 2000.00 2500.00 75
134 (1) ( 2 ) (3) (4)
135 ( 5 ) (6) ( 7 ) ( 8 )
136 ( 9 ) ( 10) ( 11 ) ( 12 )
137 ( 13 ) ( 14 ) ( 15 ) ( 16 )
138 (17) (18) (19) (20)
139 (21) (22) (23) (24)
140 (25) (26) (27) (28) Figure A 2 Failure modes of GFRP columns
141 Figure A 3 F urnace controller Figure A 4 Laser extensometer and MTS cooperated with furnace
142 Figure A 5 Furnace and MTS