Abstract
Inspired by a case study, this paper reports a successful application of VNS to the production planning and scheduling problem that arises in the glass container industry. This is a multi-facility production system, where each facility has a set of furnaces where the glass paste is produced in order to meet the demand, being afterwards distributed to a set of parallel molding machines. Since the neighborhoods used are not nested, they are not ordered by increasing sizes, but by means of a new empirical measure to assess the distance between any two solutions. Neighborhood sizes decrease significantly through-out the search thus suggesting the use of a scheme in which efficiency is placed. over effectiveness in a first step, and the opposite in a second step. We test this variant as well as other two with a real-world problem instance from our case study.

Abstract
Cutting and packing problems involving irregular shapes is an important problem variant with a wide variety of industrial applications. Despite its relevance to industry, research publications are relatively low when compared to other cutting and packing problems. One explanation offered is the perceived difficulty and substantial time investment of developing a geometric tool box to assess computer generated solutions. In this paper we set out to provide a tutorial covering the core geometric methodologies currently employed by researchers in cutting and packing of irregular shapes. The paper is not designed to be an exhaustive survey of the literature but instead will draw on the literature to illustrate the theory and implementation of the approaches. We aim to provide a sufficiently instructive description to equip new and current researchers in the area to select the most appropriate methodology for their needs.

Abstract
Nesting problems are particularly hard combinatorial problems. They involve the positioning of a set of small arbitrarily-shaped pieces on a large stretch of material, without overlapping them. The problem constraints are bidimensional in nature and have to be imposed on each pair of pieces. This all-to-all pattern results in a quadratic number of constraints. Constraint programming has been proven applicable to this category of problems, particularly in what concerns exploring them to optimality. But it is not easy to get effective propagation of the bidimensional constraints represented via finite-domain variables. It is also not easy to achieve incrementality in the search for an improved solution: an available bound on the solution is not effective until very late in the positioning process. In the sequel of work on positioning non-convex polygonal pieces using a CLP model, this work is aimed at improving the expressiveness of constraints for this kind of problems and the effectiveness of their resolution using global constraints. A global constraint "outside" for the non-overlapping constraints at the core of nesting problems has been developed using the constraint programming interface provided by Sicstus Prolog. The global constraint has been applied together with a specialized backtracking mechanism to the resolution of instances of the problem where optimization by Integer Programming techniques is not considered viable. The use of a global constraint for nesting problems is also regarded as a first step in the direction of integrating Integer Programming techniques within a Constraint Programming model.

Abstract
Purpose - The purpose of this paper is to solve the problem of committing electric power generators (unit commitment, UC), considering network constraints. Design/methodology/approach - The UC is first solved with a local search based meta-heuristic, following the assumption that all generators and loads are connected to a single network node. For evaluation purposes, the economical production levels of the units committed are computed by running a pre-dispatch algorithm where network constraints are not included. If a good quality solution is reached, an economic dispatch (ED) with network constraints is performed, where the geographic location of generators and loads are considered. Therefore, the production level of each committed generator is performed that leads to the global lowest solution cost, regarding both the generators' costs and constraints and the power system network constraints. Findings - The algorithm proposed is computationally efficient, given the time available for decision making. In addition, the solution for this algorithm, in terms of minimization of total costs, is generally better than the solution of the two phases approach. Some contractual and legal aspects related with the injection in network connections can also be included in the model. Practical implications - UC with network constraints has a large potential of use, especially for small and medium size power systems. It reflects reality in a closer way and provides a more complete and realistic knowledge about the system in operation. Originality/value - The paper presents an approach where theED with network constraints is integrated with the UC procedure. The model described is currently implemented in an EMS package offered in the market - making it a case of successful transfer from science to industry.

Abstract
Due to its combinatorial nature, the Unit Commitment problem has for long been an important research challenge, with several optimization techniques, from exact to heuristic methods, having been proposed to deal with it. In line with one current trend of research, metaheuristic approaches have been studied and some interesting results have already been achieved and published. However, a successful utilization of these methodologies in practice, when embedded in Energy Management Systems, is still constrained by the reluctance of industrial partners in using techniques whose performance highly depends on a correct parameter tuning. Therefore, the application of metaheuristics to the Unit Commitment problem does still justify further research. In this paper we propose a new search strategy, for Local Search based metaheuristics, that tries to overcome this issue. The approach has been tested in a set of instances, leading to very good results in terms of solution cost, when compared either to the classical Lagrangian Relaxation or to other metaheuristics. It also drastically reduced the computation times. Furthermore, the approach proved to be robust, always leading to good results independently of the metaheuristic parameters used.

Abstract
This paper presents and analyzes an incremental system for clustering streaming time series. The Online Divisive-Agglomerative Clustering (ODAC) system continuously maintains a tree-like hierarchy of clusters that evolves with data, using a top-down strategy. The splitting criterion is a correlation-based dissimilarity measure among time series, splitting each node by the farthest pair of streams. The system also uses a merge operator that reaggregates a previously split node in order to react to changes in the correlation structure between time series. The split and merge operators are triggered in response to changes in the diameters of existing clusters, assuming that in stationary environments, expanding the structure leads to a decrease in the diameters of the clusters. The system is designed to process thousands of data streams that flow at a high rate. The main features of the system include update time and memory consumption that do not depend on the number of examples in the stream. Moreover, the time and memory required to process an example decreases whenever the cluster structure expands. Experimental results on artificial and real data assess the processing qualities of the system, suggesting a competitive performance on clustering streaming time series, exploring also its ability to deal with concept drift.