Lars Malcolm Pedersen

A nonlinear observer design for fuel cell hydrogen estimation

We present an observer design to estimate the partial pressure of hydrogen in the anode channel of a fuel cell. A precise knowledge of this pressure is of importance to ensure reliable and efficient operation of the fuel cell power system. Our design makes use of a monotonic nonlinear growth property of the voltage output on hydrogen partial pressures at the inlet and at the exit of the channel. By treating the slowly varying inlet partial pressure as an unknown parameter, an adaptive observer is developed that employs a nonlinear voltage injection term. We then study the robustness of this observer against variations in the inlet partial pressure, and analyze its sensitivity to other modeling errors. We also prove a robustness property of the observer against the parameter estimation error, which means that it can be implemented with alternative parameter estimator designs.

Control of natural gas catalytic partial oxidation for hydrogen generation in fuel cell applications

A fuel processor that reforms natural gas to hydrogen-rich mixture to feed the anode field of fuel cell stack is considered. The first reactor that generates the majority of the hydrogen in the fuel processor is based on catalytic partial oxidation of the methane in the natural gas. We present a model-based control analysis and design for a fuel processing system (FPS) that manages natural gas flow and humidified atmospheric air flow in order to regulate 1) the amount of hydrogen in the fuel cell anode and 2) the temperature of the catalytic partial oxidation reactor during transient power demands from the fuel cell. Linear feedback analysis and design is used to identify the limitation of a decentralized controller and the benefit of a multivariable controller. Further analysis unveils the critical controller cross coupling term that contributes to the superior performance of the multivariable controller.

System level dynamic modeling of fuel cell power plants

In a joint effort between United Technologies Research Center and the business unit UTC Fuel Cells system level dynamic models are developed and deployed for the design and analysis of fuel cell based power systems. The article describes the scope and challenges within this effort as well as some of the methods and tools used. System level modeling of fuel cell systems is a challenge that few modeling tools can handle due to the heterogeneous nature of the system. The joint work has resulted in a set of proprietary model libraries for unique component designs, building upon publicly and commercially available model libraries and tools. The model libraries have been used to build system models in both stationary and automotive applications. Examples of models that show the model complexity and some simulation results are given in the paper. Since the system models contain a large number of components, the number of the variables for a full system model usually exceeds 500. Simulation of large, complex systems pose special problems not seen when working with smaller models and really stretch the capabilities of the simulation tools. The paper also discuss some lessons learned in this respect.

Multivariable controller design for a hot rolling mill

This paper describes a controller design for a hot rolling mill. The main purpose of the algorithm is to improve the control of the cross-width thickness profile of the plates. This is obtained by designing a controller which makes independent thickness control possible at the two sides of the rolling mill. This is achieved by first linearizing the positioning systems using feedback linearization and then using linear quadratic eigenspace design on the linearized multivariable system. Integral control is included to ensure zero stationary thickness error. The design is done using derived dynamical multivariable models. To ensure that the design is stable, the stability of the system is investigated using the small gain theorem. The performance of the controller is evaluated using models estimated from data obtained from the hot plate mill at The Danish Steel Works Ltd.

An adaptive observer design for fuel cell hydrogen estimation

We present an observer design to estimate the partial pressure of hydrogen in the anode channel of a fuel cell. Our design makes use of a monotonic nonlinear growth property of the voltage output on hydrogen partial pressures at the inlet and at the exit of the channel. By treating the slowly-varying inlet partial pressure as an unknown parameter, an adaptive observer is developed that employs a nonlinear voltage injection term. We then prove robustness of this observer against variations in the inlet partial pressure, and analyze its sensitivity to measurement errors in the voltage.

On the reheat furnace control problem

This paper develops a model and control algorithm for the temperature control of steel blocks (slabs) passing through a furnace burner zone. The slab temperature is controlled by varying the zone temperature. A model for a single slab passing through the furnace is developed and the parameters of this model are estimated. Data from the reheat furnace no. 2 at The Danish Steel Work Ltd. are used for the identification. The single slab model is used for building a model for all slabs in the burner zone. A slab temperature reference function (heating curve) is found from the slab temperature model. A nonlinear controller is designed to minimize the deviation from this heating curve. The stability and controllability of the system is also analyzed.

Thickness control for a plate mill

The purpose of this paper is to describe the design of a multivariable control law for a hot plate mill. The design is performed on models adapted to the plate mill at The Danish Steel Work Ltd. The model for the rolling mill is divided into nonlinear models for the hydraulic systems performing the thickness control and a multivariable linear model for the rolling stand. The parameters of the models are found using measurements from the rolling mill at The Danish Steel Work Ltd. Most of the measurements are collected during rolling, but plate thickness measurements are performed after rolling using special equipment. In general, good agreement is obtained between measurements and model outputs. The multivariable nonlinaer model is used for designing a multivarialbe thickness control law which can be seen as an unified approach to control mean thickness and plate wedge. The result is a second order nonlinear controller with rolling forces and roll positions as inputs and control signal to the servo valves as outputs. Since it is not possible to implement the multivariable control law for the time being, the multivariable control law is simplified with the purpose of using it as a basis for an optimization of the existing thickness control equipment at The Danish Steel Works Ltd. This optimization can be seen as the first step in the practical verification of the multivariable control law. (Less)

Control of a Double Side Shear

Abstract The purpose of this paper is to design a differential position controller for the pinch rolls for a double side shear at The Danish Steel Works Ltd. The purpose of the differential position controller is to ensure that plate edges remain straight despite various disturbances. First a model is derived for the double side shear. The model is a first order differential equation with two unknown parameters. These parameters are found in the system identification which is based on data from 7 plates. The identification shows that one of the parameters varies considerably from plate to plate. The differential position controller is designed to handle the parameter variations using a robust stability criterion. Simulations show that the performance of the system is satisfactory, despite the parameter variations. Comparisons of results from the existing control system with simulations using the new control law indicate a considerable improvement of performance. The control law will be implemented in the near future. The results from the implementation will reveal if the theoretical results are correct.

Multivariable Controller Design for Hot Rolling Mill

Abstract This paper concerns the controller design for the hot rolling mill at the Danish Steel Works Ltd. The design is done using derived dynamical multivariate models. The main objective of the design is to separate the two sides of the rolling mill. This is obtained by first linearizing the positioning systems using feedback linearization and then using eigenspace design on the linearized multivariable system. To ensure that the design is stable, the stability of the system is investigated using the small gain theorem. The performance of the controller is evaluated using estimated models based on data from the plate mill.

Modeling and Identification of Hydraulic System on Rolling Mill

Abstract The identification of a hydraulic system on a hot rolling mill is described. The rolling mill is in operation at The Danish Steelworks Ltd. in Frederiksvaerk, Denmark. The hydraulic system is used for thickness control of the rolled plates. A non linear model for the system is derived and transformed to discrete time. The data are collected during normal operation, which implies that the system is operating in closed loop. It is in this connection shown that unique parameters can be obtained for the system despite of this. The identified model is able to predict the output well.