Optimization approaches for robust and durable reinforced concrete and fibre concrete structures under consideration of scale bridging polymorphic uncertainty modelling
Abstract of the planned project content, taking into account the results from the first project phase
The overarching objective of this project is the development of computational reliability and optimization methods for the lifetime-oriented design of reinforced and steel fibre concrete structures, which allow to follow the influence of polymorphic uncertainties throughout the life of the structure. Interacting loading and environmentally induced deterioration processes and the propagation of uncertainties through multiple spatial and temporal scales are accounted for by scale bridging models in conjunction with physically sound models for reinforced and fibre reinforced concrete and multifield structural analysis.
For this purpose, the sub-modeling strategy developed in the 1st funding phase will be extended in the 2nd phase to a hydro-mechanical coupled model, in order to record the influence of deterioration processes by multiscale models in conjunction with physically sound durability models. Furthermore, the aim is to quantify the uncertainties resulting from the fibre orientation by means of a number of simulations of the casting process and subsequently to generate statistical data for the fibre concrete structure, which is based on efficient substitute models.
a) Input parameter of a fuzzy random field (exemplified Young's modulus quantified as stochastic distribution with anisotropic fuzzy correlation parameters); b) Realization of the fuzzy random field; c) Combination of a convolutional neural network (CNN) and a recurrent neural network (RNN) to predict time variant fuzzy stochastic structural processes
Optimization algorithms for time variant polymorphic uncertain a priori and design parameters will be developed using particle swarm optimization. Time variant design parameters lead to high dimensional design spaces. In order to reduce the dimension of the optimization problem, approaches will be investigated, where the optimization problem is split into several interacting subproblems, which are solved sequentially or in parallel. This means that the objective function and the constraints of the subproblems are changing during the optimization. An important issue will be the investigation of the sensitivity of different surrogate measures of the polymorphic uncertain objective function w.r.t. the optimal design. Beside time variant parameters, spatial correlated stochastic distributed input material parameters, which can be quantified as random fields, will be considered. Because of the high input space resulting from the random fields, convolutional neural networks will be used to map the input random fields onto the quantity of interest.
To combine time variant polymorphic uncertain a priori and design parameter and spatial correlated stochastic distributed material parameters, convolutional neural networks and recurrent neural networks are combined.
a) Two stages of the computational simulation of the casting process; b) Illustration of the spatial distributon of the fibre orientation tensor as ellipsoids and histograms at two selected locations from multiple realizations
The developed numerical models for reinforced concrete and fibre reinforced concrete and the methods for uncertainty quantification and optimization are applied to optimize the life time of a prestressed concrete bridge strengthened by a fibre reinforced concrete layer.
Essential project goals and objectives
- How do uncertainties at the material scale (fibre distribution, crack roughness, steel-concrete bond properties) and human induced imprecisions (e.g. position of reinforcement) propagate through spatial and temporal scales and affect the structural reliability?
- Can simulations of the casting process help reducing uncertainties in the design of fibre reinforced concrete structures?
- How does the concrete and reinforcement design (fibres vs. conventional reinforcement and hybrid designs) and the associated levels of uncertainty affect the durability and lifetime of a structure?
- Is it possible to determine the concrete cover size for an optimized durability oriented design, considering uncertain interactions between reinforcement design, cracking, and the ingress of corrosive substances?
- How does a (retrofitting) design of a prestressed concrete structure, using high performance fibre reinforced concrete materials and considering polymorphic uncertainty, affect the fatigue performance and, consequently, the expected lifetime of the structure?
