Safety factors in sheet pile wall design: considerations by using traditional methods and FEM analysis

The purpose of the work is: define and calculate a factor of collapse related to traditional method to design sheet pile walls. Furthermore, we tried to find the parameters that most influence a finite element model representative of this problem. The text is structured in this way: from chapter 1 to 5, we analyzed a series of arguments which are usefull to understanding the problem, while the considerations mainly related to the purpose of the text are reported in the chapters from 6 to 10. In the first part of the document the following arguments are shown: what is a sheet pile wall, what are the codes to be followed for the design of these structures and what they say, how can be formulated a mathematical model of the soil, some fundamentals of finite element analysis, and finally, what are the traditional methods that support the design of sheet pile walls. In the chapter 6 we performed a parametric analysis, giving an answer to the second part of the purpose of the work. Comparing the results from a laboratory test for a cantilever sheet pile wall in a sandy soil, with those provided by a finite element model of the same problem, we concluded that:in modelling a sandy soil we should pay attention to the value of cohesion that we insert in the model (some programs, like Abaqus, don’t accept a null value for this parameter), friction angle and elastic modulus of the soil, they influence significantly the behavior of the system (structure-soil), others parameters, like the dilatancy angle or the Poisson’s ratio, they don’t seem influence it. The logical path that we followed in the second part of the text is reported here. We analyzed two different structures, the first is able to support an excavation of 4 m, while the second an excavation of 7 m. Both structures are first designed by using the traditional method, then these structures are implemented in a finite element program (Abaqus), and they are pushed to collapse by decreasing the friction angle of the soil. The factor of collapse is the ratio between tangents of the initial friction angle and of the friction angle at collapse. At the end, we performed a more detailed analysis of the first structure, observing that, the value of the factor of collapse is influenced by a wide range of parameters including: the value of the coefficients assumed in the traditional method and by the relative stiffness of the structure-soil system. In the majority of cases, we found that the value of the factor of collapse is between and 1.25 and 2. With some considerations, reported in the text, we can compare the values so far found, with the value of the safety factor proposed by the code (linked to the friction angle of the soil).

Meta-learning for informed model searches in Automated Machine Learning

Day by day, machine learning is changing our lives in ways we could not have imagined just 5 years ago. ML expertise is more and more requested and needed, though just a limited number of ML engineers are available on the job market, and their knowledge is always limited by an inherent characteristic of theirs: they are humans. This thesis explores the possibilities offered by meta-learning, a new field in ML that takes learning a level higher: models are trained on other models' training data, starting from features of the dataset they were trained on, inference times, obtained performances, to try to understand the relationship between a good model and the way it was obtained. The so-called metamodel was trained on data collected by OpenML, the largest ML metadata platform that's publicly available today. Datasets were analyzed to obtain meta-features that describe them, which were then tied to model performances in a regression task. The obtained metamodel predicts the expected performances of a given model type (e.g., a random forest) on a given ML task (e.g., classification on the UCI census dataset). This research was then integrated into a custom-made AutoML framework, to show how meta-learning is not an end in itself, but it can be used to further progress our ML research. Encoding ML engineering expertise in a model allows better, faster, and more impactful ML applications across the whole world, while reducing the cost that is inevitably tied to human engineers.

Machine Learning in Automatic Tabulation for Constraint Model Reformulation

Combinatorial decision and optimization problems belong to numerous applications, such as logistics and scheduling, and can be solved with various approaches. Boolean Satisfiability and Constraint Programming solvers are some of the most used ones and their performance is significantly influenced by the model chosen to represent a given problem. This has led to the study of model reformulation methods, one of which is tabulation, that consists in rewriting the expression of a constraint in terms of a table constraint. To apply it, one should identify which constraints can help and which can hinder the solving process. So far this has been performed by hand, for example in MiniZinc, or automatically with manually designed heuristics, in Savile Row. Though, it has been shown that the performances of these heuristics differ across problems and solvers, in some cases helping and in others hindering the solving procedure. However, recent works in the field of combinatorial optimization have shown that Machine Learning (ML) can be increasingly useful in the model reformulation steps. This thesis aims to design a ML approach to identify the instances for which Savile Row’s heuristics should be activated. Additionally, it is possible that the heuristics miss some good tabulation opportunities, so we perform an exploratory analysis for the creation of a ML classifier able to predict whether or not a constraint should be tabulated. The results reached towards the first goal show that a random forest classifier leads to an increase in the performances of 4 different solvers. The experimental results in the second task show that a ML approach could improve the performance of a solver for some problem classes.