Structural Optimization and Self-tuning of Marine Boilers
Authors
Agesen, Mads Kronborg ; Uggerhøj, Hans Jørgen
Term
10. term
Publication year
2009
Pages
143
Abstract
Dette projekt præsenterer et proof-of-concept self-tuning system til Mission OS (TM) kedelsystemet. Derudover udvikles en metode, der sikrer ren forbrænding, selv når aktuatorerne i brændstofsystemet er mættede (aktuator-saturering). Kedlen er modelleret som tre ulineære delsystemer: kedlen, fødevandssystemet og brændstofsystemet. Til design af SISO-regulatorer (enkelt-indgang/enkelt-udgang) simplificeres de ulineære modeller til integratormodeller, dvs. simple modeller, hvor output akkumulerer input over tid. Begge modeltyper er verificeret mod måledata fra det faktiske kedelsystem. Regulatorerne er designet til brug i det self-tuning system, som under opstart automatisk identificerer centrale modelparametre: servomotorernes karakteristik, fødevandsventilens opførsel, opbygning af damptryk og et kurvebånd, der beskriver olie/luft-forholdet til iltregulering. Ud fra disse parametre beregnes regulatorindstillinger, inden systemet går i normal drift. Brændstofsystemet styres med pulsbreddemoduleret (PWM) positionsregulering af olieventil og luftspjæld. Opgaven med at følge kurvebåndet, selv når aktuatorerne er ved deres grænser, formuleres som et minimeringsproblem og løses med et modelprædiktivt referencekorrektionssystem, der forudsiger systemets respons og justerer referencen, så iltniveauet holdes konstant. Endelig er der udformet en Kalman-estimator, der estimerer middelvandstanden i kedlen og det ikke-målbare dampforbrug for at vurdere muligheder for bedre vandstandsregulering. De udviklede metoder er testet i simulering og viser tilfredsstillende resultater.
This project presents a proof-of-concept self-tuning system for the Mission OS (TM) boiler system. It also develops a method that keeps combustion clean even when fuel-system actuators are saturated (at their limits). The boiler is modeled as three nonlinear subsystems: the boiler, the feedwater system, and the fuel system. For designing SISO controllers (single-input/single-output), the nonlinear models are simplified to integrator models—simple models where the output accumulates the input over time. Both the detailed nonlinear models and the simplified integrator models are validated against measurement data from the actual boiler. The controllers are intended for use in the self-tuning system, which during startup automatically identifies key model parameters: servo motor characteristics, feedwater valve behavior, steam pressure build-up, and a curve band that describes the oil/air ratio used for oxygen control. Based on these parameters, controller settings are computed before normal operation. The fuel system is controlled with pulse-width-modulated (PWM) position control of the oil valve and air damper. The task of following the curve band even under actuator saturation is posed as a minimization problem and solved with a model-predictive reference correction system that predicts the system response and adjusts the reference to keep the oxygen level constant. Finally, a Kalman estimator is designed to estimate the average water level in the boiler and the unmeasurable steam consumption, to assess potential improvements in water-level control. The developed methods are verified through simulation and show satisfactory results.
[This abstract was generated with the help of AI]
Keywords
Boiler ; Constrained control ; Kalman ; Marine ; SISO ; Self-tuning
Documents