Developments in reinforced glass beams

Abstract

Glass is gaining popularity as a structural material, this is mainly based on its transparent character. But due to its brittle failure behaviour the safety of these structures is an issue. Several concepts are found to guarantee the safety during the development of its structural application. One of these developments is to reinforce glass with other materials, this concept is known from concrete structures. After fracture of the glass (initial failure), the reinforcement will function as a tension element in order for the beam to maintain its integrity. This creates an additional load transfer mechanism, preventing total collapse of the structural system (ultimate failure) if designed properly.
The objective of this thesis is to further develop reinforced glass beams. In Chapter 1 this is started by defining an abstract concept called the ideal glass beam. This concept is defined as a glass beam which has increased initial failure resistance, ability to maintain its integrity, freedom in manufacturing and all without disturbing the clean transparency of the glass. For the reason of transparency potential all concepts are fibreglass based. Freedom in manufacturing is realised by the use of fibreglass fabric which can be cut as desired and laminated together with the glass panes, just as is done with fully integrated (GFRP) reinforcement. This results in the three reinforcement concepts; post-tensioned fibreglass reinforcement; fibreglass fabric reinforcement; and transparent fibreglass reinforcement. All these three concepts are not yet or only partially developed and thus require research. However only post-tension and fibreglass fabric are chosen to be elaborated in this thesis. The application of transparent fibreglass is briefly discussed in order to assess its plausibility.
Next to these design concepts, some issues exist regarding numerical calculations of glass systems. Most non-linear analysis run into convergence problems. Or computational intensive methods to account for the snap-back constitutive relation. This is due to the relative low fracture energy and high tensile strength with respect to concrete. A new kind of analysing method is in development which uses a series of linear analysis together with a damage model, reducing the material properties at each iteration. This method is called sequential linear analysis (SLA). The stability if this method creates the opportunity to analyse how alterations in design and numerical model influence the performance. These alterations come from differences in modelling approach in previous research, developments in SLA from research in concrete and design possibilities.
The design strategy and philosophy of a reinforced glass beams is different from the current design approach, where additional panes are used to prevent ultimate failure. What these differences are and how a reinforced glass beam should be designed is elaborated in Chapter 3. Here the initial and ultimate failure strength capacities are discussed, how their resistance in affected and how they should be designed in order to create a safe system. In Chapter 4 the origin, feasibility and validation of the concepts are elaborated. In order to validate the concepts and numerical method some validation goals are drafted partially consisting on the former mentioned design philosophy. These goals result in desired data requirements which is subsequently used to describe the experiments to acquire this data. Initially only two bending experiments are required, for validation of both post-tension and fibreglass reinforcement. Both are validated using a reference experiment, for post-tension this is a specimen without applied normal force and the fibreglass fabric reinforcement is validated using integrated GFRP strips. In addition to these experiments tensile data of the different reinforcements is obtained to be used as input during numerical modelling. Also pull-out experiments are done for the fibreglass reinforcements to have a more direct comparing approach and to spread the risk of unsuccessful execution of either experiment.
In Chapter 5 the experiments are designed based on the former mentioned requirements using analytical calculations. This is done to ensure theoretical feasibility of the experiments and to estimate the behaviour. The post-tensioned design consist of GFRP strips adhered on top and bottom after tensioning in order for them to function as reinforcement. No eccentricity is used te prevent crushing of the glass edges as the tensioning is introduced by head sections on either edge. These steel components function as actuators to introduce the normal force on the glass. The fibreglass fabric beams are designed to have equal ultimate failure resistance as the reference specimen.
In the experimental research in Chapter 6 the validation goals are repeated and each experiment is treated. Starting with the tensile experiments, ordinary force displacement experiments are executed on the Fibrolux GFRP strips. These experiments failed at lower values of tensile strength and Youngs’ modulus than expected based on the values given by the manufacturer. This is explained by stress concentrations due to the rectangular geometry and relative low uni-directional (UD) fabric within the composite, according to the manufacturer. The fibreglass composite experiment also failed at a lower resistance than expected. After failure loose fibres are visible implying full saturation is not achieved. Using the theoretical strength of the SentryGlass (SG) it is calculated that approximately 30% of the fibres contributed in the strength resistance. The pull-out experiments are designed as a double pull-out with one lower resistance side which should fail by de-bonding. The difference in resistance can be compared with a reference set which has equal shear resistance to analyse the bonding performance. But due to the same saturation issue as the tensile fibreglass specimen, these fibreglass specimen also failed earlier. Thus making it unable to compare the different reinforcements. The GFRO pull-out specimen did fail properly as is expected from the reference set. Furthermore, the bending experiment using fibreglass fabric reinforcement did perform accordingly. Despite the same internal slipping of the fibreglass, the specimen showed almost sufficient ultimate resistance. After the occurrence of the first crack, the resistance increases again but after some deflection a decrease is observed. This is the consequence of the same internal slip due partial saturation. The final experiment regarding post-tensioning had a fault during the post-tensioning process, resulting in an insufficient posttension force. The experiments did therefore shown no increase in initial failure. Although the post-tension experiment did not succeed, still a GFRP reinforced beam was experimented and a different failure was observed. This different failure mode occurred due to the low stiffness of the reinforcement resulting in high crack opening and earlier failure of the system.
The numerical analysis is done in Chapter 7. This chapter contains two parts, first the analysis part and secondly the validation part. During the first part a study is done on different parameter and design considerations. This study is based on an experimental data set from former research, this set is also used on a sequential linear analysis (SLA) and non-linear analysis (NLA) validation and is therefore the ideal case to advance the developments. A reference model is made based on starting points from previous research and each aspect is applied on this reference model to analysis the performance differences. This performance differences are analysed based on four aspects: Resistance, deformation, crack density and amount of cracks. After each analysis a conclusion is made using these performance aspects and is graded with respect to the reference case. Two major objectives in this study is to implement the lacking resistance of the SG interlayer and post-tensioning. For the interlayer, the characteristics of SLA are used to construct an equivalent post-cracked resistance using the reduction branches. The post-tension should be implemented as initial stress/strain conditions but as SLA is still in development this is not yet available. A two solutions are found based on the transformation of the material properties. Both result in the increased initial failure, but only one represents the post-cracked behaviour of a post-tension beam in the correct manner. The numerical models of the reference case and post-tension experiments are validated in the second part of this chapter. Initially the fibreglass fabric experiment were planned to be used as validation, but this is not done. Based on the complex behaviour of the reinforcement, numerical analysis of this experiment does not contributed to the objective. The validation of the numerical analysis is done based on the deviation which represents the difference in ultimate resistance with respect to the experiment. For the reference case the deviation in resistance and deflection to the experiments decreased when the findings are applied to the reference model. However the crack pattern is still an issue as T-shaped crack occurred where V-shaped cracks were expected. Also in the validation of the post-tension experiment this V-crack is not present, but due to the absence of SG the cracking pattern is lot more conform the observations in the experiment. This case not only has small deviations in resistance and deflection, but failed in equal manner as the specimen.
All results are summarised and discussed in Chapter 8, from this discussion the conclusions regarding the stated research objectives in the outline are drawn in Chapter 9. In this chapter is concluded that the suitability of GFRP post-tensioning could not be validated due to the fault in the process and the low Young’s modulus resulted in higher crack opening and earlier reaching of ultimate failure. The fibreglass fabric did achieve a fair ultimate failure resistance but as they did not have a higher resistance as the initial fracture they did not pass the validation criteria. This was due to insufficient saturation of fibreglass with SG, causing the fibres to slip on one each other, hence not able to reach their full capacity. The analysis of different aspects using SLA resulted in a development in analysing method. Using a grading system the influences on the performance of each aspect are indicated. Creating insight of the influences of different design and modelling considerations. To finalise some recommendations are in place, as this is only the beginning of a broad subject. Obviously, it is recommended to research the third concept as this could not be included in this thesis, to even further develop the ideal reinforced glass beam. But regarding the post-tensioning, more
effective systems are recommended as the initial failure increase in the post-tensioned design was quite low, for example, higher eccentricity and more stable structures. Solutions for higher saturation in the fibreglass fabric to decrease the slip and increase the resistance. Regarding the numerical analysis, implementation of SLA in 3D brick elements and a better more physical correct plasticity model are recommended. Next to these independent subjects reinforced glass in general requires the definition of different failure modes and the prevention of brittle failure modes. Also would it be important to perform force-controlled experiments, possibly in combination with an impact load. Since this is the major reason to apply reinforcements. More recommendations and conclusions are found in Chapter 9.
So in conclusion all the independent subjects converge to the contribution in the developments in reinforced glass beams. The freedom and benefits in application of integrated fabric reinforcement are known and more knowledge is obtained regarding design of reinforced glass structures and influences of (in)correct numerical modelling.