Mabel Sánchez

Cost-optimal design of reliable sensor networks

Several papers have been presented in the last years regarding the design of reliable sensor networks. In all these papers, the system reliability was maximized, constrained by a fixed number of sensors. In these models, the cost has played an indirect unclear role. A minimum cost model for the design of reliable sensor networks is presented in this paper. The connections with previous models are established, showing that they are a particular case of the model stated in this work.

Use of orthogonal transformations in data classification-reconciliation

Abstract In this paper, the use of orthogonal factorizations, more precisely the Q-R decomposition, to analyze, decompose and solve the linear and bilinear data reconciliation problem is further investigated. It is shown that the decomposition provides additional insight in identifying structural singularities in the system topology, allowing the problem to decompose into lower dimension subproblems. Energy balances are explicitly considered. Two examples of application are presented.

Simultaneous estimation of biases and leaks in process plants

In this paper, a recursive strategy is applied to identify gross errors (biases and leaks) and estimate their magnitude in linear steady state processes. A recursive search scheme is used first to isolate the candidate sources of gross errors and then simultaneous identification/estimation of gross errors is accomplished. A recently proposed equivalency theory is used to correctly assess results. Comparative studies of performance and accuracy of estimation are performed for some process networks when compared to existing techniques. Simulation results show that the proposed approach has higher performance to identify gross errors as well as to estimate their magnitudes.

Reallocation and upgrade of instrumentation in process plants

Abstract In several process plants the precision of parameter estimates is low because the installed set of instruments do not satisfy the new information requirements for different purposes, from simple data reconciliation to on-line optimization. This paper presents models to perform the upgrading of instrumentation at minimum cost to achieve maximum precision of selected parameters. Alternative equivalent models based on maximizing the precision of the parameters are capable on putting a bound on the capital cost. The intricacies of these apparently conflicting goals are explained and a unique procedure based on an MINLP model is presented.

Pladat: A package for process variable classification and plant data reconciliation

Abstract Process data are used in chemical plants for the purpose of control, performance evaluation, optimization, etc.. Measurements usually have some degree of error, therefore it is necessary to adjust them through a reconciliation procedure in order to obtain a reliable process knowledge. In this work a computer package (PLADAT) is described which allows the classification of the operational variables of a complete chemical plant. The results of this procedure are used to reduce the size of the reconciliation problem. An output set assignment of the process balance equations is used to classify the measured and unmeasured variables and provide a reduced set for reconciliation. This problem is solved using a nonlinear programming technique. Previously, statistical methods are applied for the detection and elimination of gross errors, which would invalidate the reconciled results. The application of PLADAT for the instrumentation revamping of an existing ethylene plant section with 150 streams and 45 units is discussed. The minimum amount of instruments required for the determinabi1ity of all unmeasured process variables is obtained in order to develop the complete mass and energy balances. Data reconciliation results for the demethanlzing sector are provided.

ModGen: a model generator for instrumentation analysis

Abstract A computer software tool for the automatic generation of steady-state process models to be used in instrumentation analysis was developed. We describe the program, called ModGen, discussing its main advantages and potential benefits. ModGen constitutes the front-end of a complete decision support system (DSS) for plant instrumentation design and revamp. This DSS is currently under development. The paper concludes with the description of ModGens application to the classification of unmeasured variables of an existing medium-size process plant by means of GS-FLCNs structural technique for observability analysis.

Sensor network design and upgrade for plant parameter estimation

This work concentrates on the comparison of two approaches for the design of sensor networks with parameter estimation purposes. After extending a Maximum Precision Model recently published to multiple parameter estimation and binary variables, its equivalence with the Minimum Cost Model is presented. Industrial heat exchanger units are used to illustrate the results.

Design of Sensor Networks for Chemical Plants Based on Meta-Heuristics

In this work the optimal design of sensor networks for chemical plants is addressed using stochastic optimization strategies. The problem consists in selecting the type, number and location of new sensors that provide the required quantity and quality of process information. Ad-hoc strategies based on Tabu Search, Scatter Search and Population Based Incremental Learning Algorithms are proposed. Regarding Tabu Search, the intensification and diversification capabilities of the technique are enhanced using Path Relinking. The strategies are applied for solving minimum cost design problems subject to quality constraints on variable estimates, and their performances are compared.

REMOVING SINGULARITIES AND ASSESSING UNCERTAINTIES IN TWO EFFICIENT GROSS ERROR COLLECTIVE COMPENSATION METHODS

Abstract Two collective estimation strategies, the Unbiased Estimation Technique (Rollins and Davis, 1992) and the recursive Generalized Likelihood Ratio (Keller et at.,1994), have been shown lo be very efficient in detecting and estimating multiple gross errors in linear process systems. However, these strategies run into singularities and uncertainties that prevent them from being used in automatic schemes. This paper uses a recently presented theory on gross error equivalency to explain when and how these singularities and uncertainties take place. The procedures presented by these two methods are modified to prevent the singularities from appearing and allowing their automatic implementation.

Gross measurements error detection/identification for an industrial ethylene reactor

Abstract This paper describes a strategy that allows the identification of gross errors for pyrolisis reactor measurements. The problem formulation is not specially suited for a particular case but has a wide range of application. A reactor model is formulated in terms of heat and mass balances and input-output mappings based on available measurements. The adjustment is done using historical data and a rigurous reactor simulation program. Neural networks trained with a Robust Back Propagation algorithm, relating variables in the convective zone, are essential to identify gross errors in crossover temperatures. The evaluation of the proposed scheme for gross error detection/identification shows a good performance.