L. Gimeno

Reactive scheduling approach for multipurpose chemical batch plants

Abstract In this paper a reactive scheduling technique for multipurpose plants is presented. A Mixed Integer Linear Problem (MILP) formulation is used based on the State Task Network (STN) representation of Kondili et. al. (1993). In order to obtain short response times a rolling horizon is utilized, which is reasonable as far as it is considered a reactive context where perturbations in the future are expected. To bring into consideration time constraints outside the rolling horizon a lookahead characteristic is introduced.

A New Spline Algorithm for Non-Linear Filtering of Discrete Time Systems

Abstract For dynamic systems of order greater than or equal two the computational burden involved in the reconstruction of a posteriori densities needs efficient numerical tools. This work presents spline-based filters, where the proposed method obtains the same accuracy as in the classical ten sor product approach considering a reduced set of points. The feasibility of the non-linear filter is shown by some simulation examples for systems of first and second order.

A Comparison of Fault Detection Methods and Adaptive Identification Algorithms

Abstract This paper is concerned with the problem of identification of timevarying linear stochastic systems. Slowly time-varying systems are first considered. Recursive identification algorithms are described in U-D factorized forms. In presence of abrupt parameter changes, the tracking ability of identification algorithms is improved by use of a fault detection method resulting in adaptive identification algorithms. A comparison of performance of various adaptive identification algorithms is carried out through a study in simulation and experimental results are given to show the influence of the tuning parameters of such algorithms

Scheduling heuristics based on tasks time windows for APS systems

Abstract Advanced Planning and Scheduling systems (APS) are intensively utilized in the industry, namely because they allow generating and analyzing planning scenarios very quickly. They are not well suited for make-to-order situations since due dates are not considered hard constraints to be fulfilled. Processing time windows allow to take into account due dates in the development of APS enhanced heuristics.

Constraint Propagation Tools in Multipurpose Batch Plants Short Term Planning

Abstract Constraint propagation tools, used in scheduling problems, are utilized in a short term planning problem in a batch plant to obtain a planning frame in the form of batches processing time windows and equipment units loading. An extended propagation mechanism is proposed to deal with storage constraints.

Mould Level Control in Continuous Casting Machines. Identification Aspects

Abstract This paper deals with the level control of molten steel in the mould of a continuous casting machine. Existing control techniques are described and an approximate linear model is developed. The resulting model is in the form of a MIMO matrix polynomial stochastic difference equation. The control problem is stated as a minimum variance control problem, wearing effects introducing the need for an adaptative controller. Two recursive methods for parameter estimation in MIMO systems are studied, Extended Matrix (EM) and a new extension of the Approximate Maximum Likelihood (AML) method. In order to lower the computational time each row of model parameters is estimated separatedly; EM leads directly to this simplification, AML needs the utilization of a particular canonical form. Simulation results are presented to compare the two algorithms.

A holistic framework for supply chain management

Abstract A new approach to Supply Chain Management (SCM) is proposed that considers different structures conceived as module components of a holistic framework. In this way, decision support is supplied at different levels and for different functionality. Thus, the system may be also used for multi-objective assistance to decision making, while systematically considering the time and other intrinsic constraints associated to the structure of the supply chain. This approach is applicable to the different supply chain layers, from the individual single sites to the global Supply Chain optimisation, including multi-site (single company) competitive strategic decisions, or the independent management of a similar objective (i.e. benefits, client satisfaction, etc.) for different companies.

A time-windows approach for enhancing the capabilities of batch scheduling systems: An application to simulated annealing search

Simulated Annealing (SA) is utilized for a short term scheduling problem where batches have to be scheduled inside processing time windows obtained in a planning phase. It is well known that SA, when used for solving a short-term scheduling problem, can generate a large quantity of infeasible candidates in heavily constrained situations, since it is difficult to filter out infeasible candidates based only in their structure, basically an ordering relation between the tasks. It is shown that tools from Constrained Based Search scheduling techniques can be utilized to implement a filtering procedure for the problem, determining which interchanges lead to feasible situations.

Time windows for short term planning in multipurpose batch plants

Abstract A planning approach is presented for short term mode of operation in multipurpose batch plants. Planning output is given in the form of processing time windows for intermediate products. Time windows are established from the time profile demand of final products and raw materials availability plan input by the user. Equipment units capacity analysis is used to assure a feasible situation and an acceptable plant loading.

Constraints propagation techniques in batch plants planning and scheduling

Abstract The problem considered is make to order situations where products due dates and raw materials delivery times are constraints that the user must try to satisfy or negotiate. Those constraints, coupled with recipe and resource constraints, can be conveniently handled through constraints propagation techniques both in planning and scheduling. an interactive planning system is discussed where the user gets, after each decision, a feedback in terms of batches processing time windows, equipment units load and estimated resources utilization. APS-like scheduling heuristics exploit these measures to implement internal procedures to select batches in the sequential scheduling procedure.