2. Theoretical framework
2.5. Calculation algorithms
The current chapter presents the algorithms that were used for the proposed design. Each of the following subchapters defines an individual calculation method, explains its principle and states its source.
2.5.1. Transformed area method
This method is used when the structural member that needs to be analysed is not homogenous (i.e. it is comprised of more than one material). According to Philpot (2011), using this method, the original cross section (comprised of two materials) can transformed into an equivalent cross section consisting of only one material. The method takes into account the difference between elasticity moduli and converts the transformed material into the original one with a different width.
The advantage provided by this algorithm was that the cross section used in the current project could be analytically analysed. Specifically, by converting the orthotropic section to a homogenous one, the bending stresses could be determined with the Euler-Bernoulli beam equation, due to the fact that it now fulfils all the simple bending theory conditions. Furthermore, shear stresses could be determined with the beam shear or Zhuravskii shear stress formula. The detailed description of this algorithm and its application to the current project is presented in Appendix 6.
2.5.2. Corrosion of steel embedded in GFRP
Steel is a material well known for its degradation over time under the influence of oxygen and moisture. This process is known as corrosion and it affects the lifespan of a structure when it manifests.
It is important for FiberCore Europe to know whether incorporating steel in a GFRP cross bridge deck will affect the lifespan of 100 years that can currently be achieved. For this purpose, Wilken (2015) performed corrosion tests on steel embedded in GFRP in order to ascertain the behaviour. The experiment concluded that by fully encapsulating the steel inside the GFRP, and covering the latter with a layer of gelcoat and one of topcoat the steel will be prevented from corroding, as long as the protective outer surface remains intact. Therefore, for the current project, due to the fact that only fully encapsulated steel is considered in the design, the assumption that it will not corrode is valid.
Table 2 - Unit consistency
12 2.5.3. Adhesive bond strength between steel and GFRP
An important aspect to consider when analysing hybrid materials is the strength of the connection between the individual components. In order to determine the stresses that occur in the bonding layer between a steel plate and GFRP skin, Wilken (2015) tested the adhesive bond strength between SR235JR structural steel and glass fibre reinforced Synolite 1967 polyester with respect to two different pre-treatments. The experiment concluded that the shear strength of the adhesive bond was 5,9 MPa.
He then states that this is considered to be a conservative value due to the nature of the test which consisted of small sized samples. Therefore, the 5,9 MPa was considered the design strength value of the adhesive bond between steel and GFRP.
2.5.4. Thermal stresses
Most materials exhibit changes in dimension in reaction to changes in temperature, specifically expand when warmed up and contract when cooled down. The degree to which a material changes its dimensional properties as a result of changes in temperature is indicated by its coefficient of thermal expansion (CTE), often indicated by the Greek letter α. When materials with different CTE’s are interconnected and movement is constrained, thermal stresses occur as a result of changes in temperature. Through Classical Laminate Theory the CTE of laminates can be determined, since the individual components vary depending on the fibre volume content and the fibre orientation. During the current project, the CTE of the laminate was used, not the ones of the component materials. The laminate’s CTEs in the longitudinal (i.e. x) and transverse (i.e. y) direction as used in typical bridge designs by FiberCore Europe, (2016) are: 𝛼𝛼𝑥𝑥 = 8.22 ∗ 10−6 𝑚𝑚𝑚𝑚∗𝐾𝐾; 𝑚𝑚𝑚𝑚𝑎𝑎 𝛼𝛼𝑦𝑦= 3.71 ∗ 10−5 𝑚𝑚𝑚𝑚∗𝐾𝐾 For composite bridges, there are no regulatory guidelines for thermal loads. In practice, the temperatures used for calculations of thermal expansions and stresses were derived from the NEN-EN-1991-1-5 standard with national annexes NEN 1 and NEN 2. This standard prescribes temperature ranges for bridges made from concrete and steel. The reaction of composite bridges to temperature is more similar to concrete bridges than to steel bridges, mainly due to its relatively low conductivity and therefore the temperatures prescribed for the former structures are used. The temperatures used in the current design are defined as follows:
• Maximum temperature range contraction: ΔTN.con = T0 - Te.min = 27 °C
• Maximum temperature range expansion: ΔTN.exp = Te.max – T0 = 22 °C
Wilken (2015) performed an experiment on thermal stresses whose conclusion was that the compressive and tensile stresses in the individual materials are considerably low compared to allowable stresses in the materials. This result was considered as an initial assumption in the current report’s calculations.
Appendix 2 provides a more detailed description of the algorithm used to calculate the thermal, expansion and corresponding stresses generated in the two materials, as derived from the information provided above.
2.5.5. Dynamic behaviour
According to Feldmann & Heinemeyer (2008), lightweight footbridges have small mass, which reduces the mass inertia and which lowers natural frequencies, resulting in a greater risk of resonance.
Feldmann & Heinemeyer (2008) then explain that resonance occurs if the frequency of the bridge coincides with the frequency of the excitation. Pedestrian induced excitation is an important source of vibration of footbridges and the loading caused by it is unsteady, transient and oscillating in a small
13 range of excitation frequency. It is therefore clear that dynamic responses play a fundamental role in the design of vibration susceptible structures. Vibrations of footbridges can lead to serviceability problems, as effects on the comfort of pedestrians might occur.
Since the bridge designed in the current project is modelled as a simply supported beam, the first natural frequency can be calculated with the following formula:
𝑓𝑓𝐾𝐾,𝑑𝑑𝑑𝑑𝑑𝑑 =𝐾𝐾𝑛𝑛
2𝜋𝜋 ∗� 𝐸𝐸𝐼𝐼
𝛾𝛾𝑚𝑚,𝑆𝑆𝑆𝑆𝑆𝑆∗ 𝛾𝛾𝑐𝑐𝑐𝑐,𝑣𝑣∗ 𝑔𝑔 (𝑞𝑞𝑀𝑀+ 𝑎𝑎𝑑𝑑𝑑𝑑∗ 𝑏𝑏𝑒𝑒𝑓𝑓𝑓𝑓) ∗ 𝐿𝐿4
For a uniform beam with the aforementioned support and loading conditions, Kn has a value of 9.87.
This value is based on perfectly hinged supports without any restraining moment. The actual frequency is higher. For this reason, FiberCore Europe studied the influence of abutment supports and the added stiffness of the railings on 20 bridges after installation and determined that the natural frequency can be increased by 18%, incorporating a so called Panos-factor yield for the harmonic constant. Therefore,
𝐾𝐾𝑛𝑛 = 9.87 ∗ 1.18 = 11.65
It can be observed that the flexural rigidity, weight and the dimensions of the deck have considerable influence on the natural frequency, with the length having the largest impact. Additionally, it can be noted that with increasing the dimensions and weight, the natural frequency decreases.
A critical range is determined by the dominant contribution of the first harmonic which characterises pedestrian effects. For longitudinal vibrations, this range is calculated as: 1,25 Hz ≤ 𝑓𝑓𝑖𝑖 ≤ 2,3 𝐻𝐻𝐻𝐻.
(Feldmann & Heinemeyer, 2008)
The Eurocodes do not specify a limit for the frequency, only for maximum vertical accelerations.
However, a relation between deflection and natural frequency is provided in EN 1991-2:2003 6.4.4 [Note 8] in the shape of:
𝑚𝑚0=17,75
�𝛿𝛿0
Where n0 represents the natural frequency and δ0 the deflection at mid-span due to permanent loads.
2.5.6. Creep behaviour of GFRP members
GFRP members creep over time. The creep is translated into design parameters as long term creep. In order to realistically determine the deflection of a bridge deck, not only deflection due to self-weight and live load are required, but also the deflection at the end of the lifespan, defined to be 100 years.
In order to determine the additional deflection caused by creep, the CUR-aanbevelingen 96.
(Civieltechnisch Centrum Uitvoering Research en Regelgeving, 2003) provides an algorithm whose purpose is to determine an adjusted lower value of the laminate’s elasticity modulus. The explicit algorithm that was used for determining this value is shown in Appendix 3.
2.5.7. Buckling of composite plates
In case of large magnitude concentrated loads, such as wheel loads of maintenance or accidental vehicles being applied on a bridge deck, the webs can buckle. In order to check the stability of the webs, an algorithm also provided in the CUR was used.
14 The algorithm only assumes uniformly distributed loads being applied on edges of composite plates, therefore it was only used to determine a critical buckling factor under the ULS uniformly distributed load which was afterwards used to check the FEM results. Having established that the FE model was accurate, the buckling factor under wheel load was determined.
The critical buckling factor represents the bearing capacity of the plate before it buckles. In order to determine the critical buckling load (i.e. the load under which the plate buckles), the critical buckling factor has to be multiplied with the applied load. Therefore, in order to prevent a plate from buckling, a critical buckling factor larger than 1 is required.
The formula for calculating the abovementioned factor together with the required parameters are presented in Appendix 4.
2.5.8. Shear stresses between two different materials
Another important check that had to be made was shear stresses in the adhesive bond between the steel plates and the GFRP skins. In order to determine these stresses analytically, the Zhuravskii shear stress formula can be used, with a modification that accounts for the difference in elastic moduli between the two materials and the location of the adhesive bond with regards to the neutral line of the cross section. The detailed algorithm can be seen in Appendix 5.
15