Periodic Reporting for period 1 - MFOptBF (Multi-fidelity design optimization of long-span bridges considering probabilistic wind-induced instabilities of flutter and buffeting and hydrodynamics)
Berichtszeitraum: 2018-08-20 bis 2020-08-19
Efficient transport structure is considered as fundamental for the economic growth and social cohesion of communities. Long-span suspension bridges have served to cross large physical barriers for decades as one of the most challenging structures in civil engineering. Several major suspension bridges are currently under planning; Coastal Highway E39 Route in Norway, Çanakkale Bridge in Turkey, Tokyo Bay Bridge in Japan to name a few.
The span length of suspension bridges has become longer and longer with the advances in construction technologies, yet the bridge structures become more flexible and more prone to wind-induced vibrations, which are key concepts for the bridge design. Among different structural elements for the design, the bridge deck shape is one of the most important factors to determine the structural response under wind.
For the construction of long-span bridges, huge material cost is required, and the reduction in material will be of great importance for the sustainable development of the world. In fact, steel production is one of the main industry sources other than power generation to release considerable amount of CO2.
The design optimization is a mathematical method of minimizing a cost function while satisfying structural requirements. For carrying out design optimization of bridge structures considering wind-induced vibrations require high computational cost. In general, Computational Fluid Dynamics (CFD) models are used for the estimation of wind forces acted on a particular bridge section. However, performing numerous precise CFD analyses is computational prohibitive.
In recent years, the use of surrogate modelling is becoming more popular to alleviate the high computational costs of numerical simulations. Surrogate models provide a structural response (output) from a limited number of data (input). Among such surrogate modelling, the multi-fidelity surrogate modelling uses two types of input data: few numbers of precise numerical simulations and many low-fidelity approximate models results. Since structural response based on few expensive data is complemented by many cheap data, it saves time for simulations. Therefore, the design optimization based on multi-fidelity surrogate modelling is computationally efficient while maintaining the precision of high-fidelity simulations.
The purpose of this project is to carry out the shape design optimization of suspension bridge deck using multi-fidelity methods considering wind instabilities. It is multidisciplinary involving in mathematical discipline of design optimization, wind, and structural engineering. This project will help to form efficient transport in the EU by providing method for the sustainable bridge designs.
The multi-fidelity optimization method was proven to be feasible and effective from this project. I hope that more and more industries will use optimization methods for their construction/production so that we could contribute to sustainable development.
The span length of suspension bridges has become longer and longer with the advances in construction technologies, yet the bridge structures become more flexible and more prone to wind-induced vibrations, which are key concepts for the bridge design. Among different structural elements for the design, the bridge deck shape is one of the most important factors to determine the structural response under wind.
For the construction of long-span bridges, huge material cost is required, and the reduction in material will be of great importance for the sustainable development of the world. In fact, steel production is one of the main industry sources other than power generation to release considerable amount of CO2.
The design optimization is a mathematical method of minimizing a cost function while satisfying structural requirements. For carrying out design optimization of bridge structures considering wind-induced vibrations require high computational cost. In general, Computational Fluid Dynamics (CFD) models are used for the estimation of wind forces acted on a particular bridge section. However, performing numerous precise CFD analyses is computational prohibitive.
In recent years, the use of surrogate modelling is becoming more popular to alleviate the high computational costs of numerical simulations. Surrogate models provide a structural response (output) from a limited number of data (input). Among such surrogate modelling, the multi-fidelity surrogate modelling uses two types of input data: few numbers of precise numerical simulations and many low-fidelity approximate models results. Since structural response based on few expensive data is complemented by many cheap data, it saves time for simulations. Therefore, the design optimization based on multi-fidelity surrogate modelling is computationally efficient while maintaining the precision of high-fidelity simulations.
The purpose of this project is to carry out the shape design optimization of suspension bridge deck using multi-fidelity methods considering wind instabilities. It is multidisciplinary involving in mathematical discipline of design optimization, wind, and structural engineering. This project will help to form efficient transport in the EU by providing method for the sustainable bridge designs.
The multi-fidelity optimization method was proven to be feasible and effective from this project. I hope that more and more industries will use optimization methods for their construction/production so that we could contribute to sustainable development.
The main objective of this project is to apply the multi-fidelity optimization technique to suspension bridges considering aerodynamic constraints.
First of all, in order to carry out a shape optimization of the bridge deck, two design variables were defined. The deck geometry was defined as the section fitted with guide rails and devices to reduce vortex shedding vibrations.
In order to understand the bridge performance under wind load, we use aerodynamic force coefficients called, lift, moment and drag coefficients. They are related to lift (L) moment (M) and drag (D) forces acted on the bridge deck by wind. Based on these data, we can study the suspension bridge under some aerodynamic instabilities of flutter and buffeting.
We create two types of Computational Fluid Dynamic simulation models to study these aerodynamic coefficients. One is called high-fidelity (HF) model, which is a precise model, but it takes a long time for computer simulation. The other is called low-fidelity (LF) model, which is not as precise as the HF model, but it runs quickly while giving enough insight of the force coefficient when the bridge shape is modified.
40 LF models and 17 HF models were simulated, and we recorded the results of the force coefficients. Then we created surrogate surfaces, which are prediction surfaces of the aerodynamic coefficients for all shape variations even where we have no data.
Once we have the surrogate models, we can get aerodynamic coefficient of any deck shape, and therefore we can carry out a design optimization. I created a structural model to obtain natural frequencies and mode shapes for modifying the shape. All these data are used to analyze the bridge performance under flutter and buffering.
Wind tunnel tests were carried out to validate CFD simulation models in the wind tunnel of the University of Coruña, Spain.
The resulting optimum shape considering flutter was more aerodynamic shape than the original design and the flutter performance has increased from 72.49 m/s to 94.5 m/s as a result of the optimization. Buffeting phenomenon will be included in the optimization routine in the incoming publication.
Once the optimum deck shape is determined, my next step is to optimize the thickness of the steel plates that form the bridge deck to optimize the steel weight needed to construct the bridge deck.
First of all, in order to carry out a shape optimization of the bridge deck, two design variables were defined. The deck geometry was defined as the section fitted with guide rails and devices to reduce vortex shedding vibrations.
In order to understand the bridge performance under wind load, we use aerodynamic force coefficients called, lift, moment and drag coefficients. They are related to lift (L) moment (M) and drag (D) forces acted on the bridge deck by wind. Based on these data, we can study the suspension bridge under some aerodynamic instabilities of flutter and buffeting.
We create two types of Computational Fluid Dynamic simulation models to study these aerodynamic coefficients. One is called high-fidelity (HF) model, which is a precise model, but it takes a long time for computer simulation. The other is called low-fidelity (LF) model, which is not as precise as the HF model, but it runs quickly while giving enough insight of the force coefficient when the bridge shape is modified.
40 LF models and 17 HF models were simulated, and we recorded the results of the force coefficients. Then we created surrogate surfaces, which are prediction surfaces of the aerodynamic coefficients for all shape variations even where we have no data.
Once we have the surrogate models, we can get aerodynamic coefficient of any deck shape, and therefore we can carry out a design optimization. I created a structural model to obtain natural frequencies and mode shapes for modifying the shape. All these data are used to analyze the bridge performance under flutter and buffering.
Wind tunnel tests were carried out to validate CFD simulation models in the wind tunnel of the University of Coruña, Spain.
The resulting optimum shape considering flutter was more aerodynamic shape than the original design and the flutter performance has increased from 72.49 m/s to 94.5 m/s as a result of the optimization. Buffeting phenomenon will be included in the optimization routine in the incoming publication.
Once the optimum deck shape is determined, my next step is to optimize the thickness of the steel plates that form the bridge deck to optimize the steel weight needed to construct the bridge deck.
As the optimization method is rarely used for civil structures, the introduction of this method to this field is very beneficial to the industries and our society for the realization of sustainable development. Since steel production is a big CO2 emitter behind power generation, the use of optimization method for the design of civil structures to avoid unnecessary steel production will impact the society. I presented my research in front of policy makers during the project. The application of the proposed method of multi-fidelity optimization is not limited to suspension bridges, but to any structures or industrial products. I will try to reach out industries, policy makers and general public for spreading this method and knowledge as much as I can.