Simao P.D.,INESC Coimbra |
Simao P.D.,University of Coimbra
European Journal of Mechanics, A/Solids | Year: 2017
The paper presents a procedure for the stability analysis of columns that are sensitive to shear deformations, the so-called “weak-in-shear” columns that can be found, in the engineering practice, in build-up or composite columns, or in elastomeric bearings. Two distinct formulas are commonly used to compute the critical load for shear sensitive columns: the Engesser and the Haringx formulae, the latter enabling significantly higher loads. They differ on the choice of the cross section's shear stresses resultant, and this duality has been object of much passionate discussion during the last decades. This problem is analysed here under the perspective of the Generalized Beam Theory, and a specific mode for shear deformations was developed using two distinct strategies: i) a linear shear formulation, corresponding to the Timoshenko beam theory for which cross sections remain plane after deformation, and ii) a nonlinear shear formulation, for which shear warping is allowed in order to accomplish with the condition of null shear distortions at the section's contour. A total potential energy is defined, assuming a linear elastic behaviour, and the correspondent functional is rendered discrete by means of the Rayleigh-Ritz method. Finally, the traditional stability procedures are applied and the critical loads are computed. The Engesser critical load is derived by applying the stability procedures to the total potential energy associated with the linear shear formulation. A parametric study on the critical behaviour of a shear deformable column and a set of conclusions end the paper. © 2016 Elsevier Masson SAS
Simao P.D.,INESC Coimbra |
Simao P.D.,University of Coimbra
Computers and Structures | Year: 2017
The post-buckling analysis of shear-deformable prismatic columns under uniform compression is studied using the Generalized Beam Theory. The member's deformed shape is described by a linear combination of three modes of deformation: axial elongation, bending and the shear deformation. The total potential energy is computed, based on the relevant relations of the theory of elasticity, and is rendered discrete through the Rayleigh-Ritz method. The buckling behaviours of the pinned-pinned columns are validated by comparison with available critical load formulae. The post-buckling behaviours, for three representative lengths, are computed by searching nontrivial equilibrium points in the neighbourhood of the critical state. The results, described by the load-displacement plots of the equilibrium paths, are symmetric and stable. Furthermore, the stability of the post-buckling paths, measured by the magnitude of their concavity in the neighbourhood of the critical state, increases with increasing values of cross section shear stiffness and of the member's length. © 2017 Elsevier Ltd
Soares S.,University of Coimbra |
Antunes C.H.,INESC Coimbra |
Araujo R.,University of Coimbra
Neurocomputing | Year: 2013
In the last decades ensemble learning has established itself as a valuable strategy within the computational intelligence modeling and machine learning community. Ensemble learning is a paradigm where multiple models combine in some way their decisions, or their learning algorithms, or different data to improve the prediction performance. Ensemble learning aims at improving the generalization ability and the reliability of the system. Key factors of ensemble systems are diversity, training and combining ensemble members to improve the ensemble system performance. Since there is no unified procedure to address all these issues, this work proposes and compares Genetic Algorithm and Simulated Annealing based approaches for the automatic development of Neural Network Ensembles for regression problems. The main contribution of this work is the development of optimization techniques that selects the best subset of models to be aggregated taking into account all the key factors of ensemble systems (e.g., diversity, training ensemble members and combination strategy). Experiments on two well-known data sets are reported to evaluate the effectiveness of the proposed methodologies. Results show that these outperform other approaches including Simple Bagging, Negative Correlation Learning (NCL), AdaBoost and GASEN in terms of generalization ability. © 2013 Elsevier B.V.
Branco F.G.,INESC Coimbra |
Godinho L.,University of Coimbra
Construction and Building Materials | Year: 2013
Most recent European acoustic design codes and regulations establish a maximum value for impact sound insulation on pavement slabs. These requirements demand the implementation of technical solutions such as floating floors, with the introduction of a resilient layer under the finishing pavement layer. Technical solutions such as floating concrete slabs (placed over synthetic foam or natural fibers layers), or floating pavements (like wooden floors built over synthetic foam layers) became quite common on recently built constructions. A possible alternative solution to floating pavements is the use of lightweight soft layers, applied over the structural concrete slab. These lightweight materials may present high quality results on the reduction of impact sound transmission. In the present work, lightweight mortar slabs were tested, and the impact sound insulation for different materials was quantified. Different types of cement mortar containing expanded polystyrene, expanded cork and expanded clay granulates were compared. The acoustical performance of these solutions was evaluated through laboratory tests, using an acoustic chamber with small dimensions, which allows comparing several solutions, on similar test conditions, in an expedite way. The influence of the type floor covering used over the lightweight mortar layer was also analysed. Different types of materials were tested. © 2013 Elsevier Ltd. All rights reserved.
Dias L.C.,INESC Coimbra |
Dias L.C.,University of Coimbra |
Domingues A.R.,INESC Coimbra
Applied Energy | Year: 2014
Multi-criteria evaluation methods are often used for sustainability assessment. One such method is the Multi-criteria Spider-gram Cumulative Surface Area (MCSA score) recently used for this purpose (Nzila et al., Multi-criteria sustainability assessment of biogas production in Kenya. Applied Energy 93:496-506, 2012). This paper presents results illustrating that the MCSA method results and rankings might be biased by the arbitrary order of the criteria. This paper also addresses the way the comparison of two alternatives can be biased by the presence of a third (possibly irrelevant) alternative. Such dependence bias is due to the use of an internal normalization operation, a problem shared by some multi-criteria analysis methods. The paper concludes with a few suggestions to avoid such biases. © 2013 Elsevier Ltd.
Lourenco R.P.,INESC Coimbra
Information Polity | Year: 2013
The current financial crisis has reinforced the need for transparency and accountability in the context of open government. As such, governments have promoted transparency initiatives by developing portals where a huge number of datasets is made available. But simply to disclose datasets in a centralized web portal might not address the variety of citizens' and other stakeholders' information seeking scenarios. To characterize such scenarios we considered the role of scientists and researchers as 'information brokers' and analysed transparency assessment literature and the way information was being searched for in those exercises. Based on this analysis, we propose an overall data disclosure strategy which contemplates several different types of information sources (other than just one centralized web portal) and lay out a set of desired characteristics for those sources. We then use the proposed strategy as a framework to assess the Portuguese panorama with respect to information disclosure, and also to practically illustrate the strategy and its components. The assessment exercise showed that the proposed strategy could indeed be used not only as an open government policy guideline, but also as an open government (transparency) assessment framework. © 2013 - IOS Press and the authors. All rights reserved.
Alves M.J.,INESC Coimbra
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012
In this paper we propose a multiobjective particle swarm optimization (MOPSO) algorithm to solve bilevel linear programming problems with multiple objective functions at the upper level. A strategy based on an achievement scalarizing function is proposed for the global best selection and its performance is compared with other selection techniques. The outcomes of the algorithm on some bi-objective instances are compared with those obtained by an exact procedure that we developed before. The results indicate that the algorithm seems to be effective in solving this type of problems. In particular, the proposed selection technique provides a good convergence towards the Pareto front. © 2012 Springer-Verlag.
Soares A.,INESC Coimbra |
Gomes A.,INESC Coimbra |
Antunes C.H.,INESC Coimbra
Renewable and Sustainable Energy Reviews | Year: 2014
In a smart(er) grid context, the existence of dynamic tariffs and bidirectional communications will simultaneously allow and require an active role from the end-user concerning electricity management. However, the residential end-user will not be always available to manage energy resources and decide, based on price signals and preferences/needs, the best response actions to implement or the best usage of the electricity produced locally. Therefore, energy management systems are required to monitor consumption/generation/storage and to make the best decisions according to input signals and the user's needs and preferences. The design of adequate algorithms to be implemented in those systems require the prior characterization of domestic electricity demand and categorization of loads, according to availability, typical usage patterns, working cycles and technical constraints. Automated demand response actions must be tailored and chosen according to this previous analysis of load characteristics. In this paper, a characterization of household electricity consumption is presented and an operational categorization of end-use loads is proposed. The existing potential for demand response to a diversified set of management actions is described and a tool to assess the impact of implementing several actions with different rates of penetration of energy management systems is presented. The results obtained show the potential savings for the end-user and expected changes in the load diagram with a decrease of the aggregated peak electricity demand and a smoothed valley. © 2013 Elsevier Ltd. All rights reserved.
Sarabando P.,INESC Coimbra |
Dias L.C.,INESC Coimbra
Computers and Operations Research | Year: 2010
The additive model of multiattribute value theory is widely used in multicriteria choice problems. But often it is not easy to obtain precise values for the scaling weights or the alternatives' value in each function. Several decision rules which require weaker information, such as ordinal information, have been proposed to select an alternative under these circumstances. We propose new decision rules and test them using Monte-Carlo simulation, considering that there is ordinal information both on the scaling weights and on the alternatives' values. Results show the new rules constitute a good approximation. We provide guidelines about how to use these rules in a context of selecting a subset of the most promising alternatives, considering the contradictory objectives of keeping a low number of alternatives yet not excluding the best one. © 2010 Elsevier Ltd. All rights reserved.
Girao Coelho A.M.,Technical University of Delft |
Girao Coelho A.M.,INESC Coimbra
Computers and Structures | Year: 2013
This paper presents a finite element analysis of partial strength steel joints. The joint configurations are drawn from a previous experimental study, with good prediction shown by the finite element model. Failure of the joint is confined to the end plate and is predicted by means of micromechanical models based on void growth mechanisms and coalescence, which are practical to apply to finite element analyses without additional modelling effort. Parametric studies are carried out to investigate the structural behaviour with variations in beam depth and thickness of the end plate. Quantitative assessments of resistance and rotation capacity are undertaken. © 2012 Elsevier Ltd. All rights reserved.