Modellering och reglering av ureainsprutning på kraftvärmeverket FTG
The aim of this thesis is to improve the control of urea injection at the combined power and heating plant Framtidens Gärstad (FTG). Tekniska Verken i Linköping AB is the owner of the plant.The FTG plant consists of a boiler where garbage combustion is done. From the combustion nitrogen oxides are emitted. These nitrogen oxides are hazardous to the environment. To reduce the nitrogen oxides, injection of urea into the boiler is used. Urea is an organic compound of carbon, nitrogen, oxygen and hydrogen. When urea reacts with nitrogen oxides they transform into harmless nitrogen gas and water. In the plant the urea is injected by six lances.The control of the urea injection at FTG could be improved which would save some money for the company. This is the main reason for the aim of this thesis. It is the control of the total flow of urea to the lances that is studied in this work.Some literature about reduction of nitrogen oxides is studied, especially reduction using urea injection. There are a lot of factors that affect how good the reduction becomes. The most important factors are the amount of urea being used and the temperature of the flue gases where the reaction with urea takes place. A model with these two factors as inputs and the content of nitrogen oxides as output is derived. This is done with experiments in the boiler and system identification. The system is modelled as a linear system.The proposal about the improved control uses the temperature from a temperature measurement just below the urea injection in the boiler. This is a parameter that the existing control does not use. The temperature is divided into three intervals. For each interval different parameters for the function of the system and the nitrogen oxide controller are used. The nitrogen oxide controller in the proposed control is derived with help from a new method of controldesign called AMIGO.The identification models gave good results in two of the temperature intervals. The result for the third interval was not so good. This is probably due to lack of good data.The proposed control structure could for practical reasons not be tested online but preliminary tests using measurement data gave qualitatively reasonable results.In order to improve the results temperature dependence has to be treated more systematically.