Excessive use of fossil fuels resources is adding several types of greenhouse gases which make the earth warmer. Emissions from ship's exhausts contribute to global climate change, too. The International Maritime Organization (IMO) has adopted regulations to reduce the emission of air pollutants from international shipping, such as major air pollutants, carbon dioxide (CO2), nitrogen oxides (NOx), and sulphur oxides (SOx) under Annex VI of the 1997 MARPOL protocol. Likewise, as regulations on the emission of major air pollutants have become internationally strict, the development of environmentally friendly vessels and engines is required. One of the globally accepted means of reducing emission gases is the use of more eco-friendly fuel, LNG (Liquefied Natural Gas). LNG as a marine fuel reduces air pollutants as referred compared to traditional heavy fuel oil (HFO). Recently, large engine manufacturers are developing LNG-fuelled marine engines. In order to use this cryogenic LNG as a fuel, it is necessary to change it back to a gaseous state. A heat exchanger is used to regasify LNG. The heat exchange takes place between LNG and glycol on the primary loop, and heat exchange occurs between glycol and steam on the secondary loop. These series of processes are called LNG regasification.
To control the temperature of the heat exchanger, it is necessary to model the heat exchanger. However, it is not easy to obtain an accurate mathematical model because the heat exchanger has non-linearity and time-varying characteristics. In addition, a fixed-gain controller is bound to have a limitation in its function if parameters of the heat exchanger are changed. Thus, various techniques have been studied to improve the adaptability and robustness of the controller. Recently, there has been using nonlinear PID (NPID) controller for the controlled system which have highly nonlinear and time-varying characteristics during operation.
Therefore, this thesis proposes two types of the nonlinear proportional, integral, derivative (NPID) controllers to control the glycol temperature of the regasification system for LNG-fuelled marine engines. The Fully-Nonlinear PID (F-NPID) controller has a structure that the error between the set-point (or reference input) and output (or the measured output) is scaled nonlinearly, and input into the controller to derive proportional, integral, and derivative controllers. The Partial-Nonlinear PID (P-NPID) controller uses the conventional linear PD controller and only I controller uses the method of F-NPID controller. In this case, the nonlinear functions are implemented by the Fuzzy model of Takagi-Sugeno (T-S) type. In addition, the error is continuously scaled so that outstanding control performance can be maintained even when the operating environment is changed, thereby improving the swiftness and the closeness of responses.
Also, the parameters of the two proposed controllers are optimally tuned in terms of minimizing the integral of the absolute error (IAE) objective function based on the genetic algorithm (GA). Meanwhile, it is necessary to examine the stability of overall feedback system that can be caused by introducing nonlinear functions during controller design. For this, the stability of the overall feedback system is analyzed by applying the circle stability theorems, which is often used for stability analysis of nonlinear problems.
The proposed controllers are verified their performances which are the set-point tracking, robustness against noise and parameter changes, disturbance rejection performances by comparing with two conventional PID controllers and a conventional NPID controller.