نوع مقاله : مقاله پژوهشی
1 دانشجوی دکتری / گروه مهندسی هوافضا، پژوهشگاه هوافضا و دانشگاه صنعتی امیرکبیر، تهران، ایران
2 عضو هیات علمی / پژوهشگاه هوافضا، وزارت علوم، تحقیقات و فناوری، تهران، ایران
3 عضو هیات علمی / دانشکده مهندسی هوافضا ، دانشگاه صنعتی امیرکبیر، تهران، ایران
عنوان مقاله [English]
When designing complex products such as space thrusters, accurate simulation models are needed to evaluate and improve the design during development. The implementation of these accurate simulation models is often expensive and time-consuming. Surrogate models or metamodels are simplified models of accurate and expensive simulations that can be used to reduce some computational costs during studies or design optimization. The closer the surrogate model is to the real model, the more accurate the solution and the lower the percentage of error. These models with high accuracy are called metamodels. The purpose of this article is to design the metamodel of the liquid single-propellant thruster system using the kriging method, which can predict the behavior of the model to some extent. The purpose of this article is metamodeling of liquid monopropellant propulsion system by kriging method, which can predict the behavior of the model to some extent. The disciplines related to the liquid monopropellant propulsion system are divided into five parts: high-pressure gas tank, liquid fuel tank, injector, catalyst bed and nozzle. First, according to the input and output variables of each discipline, the design of the experiment has been done using the Latin hypercube sampling method. Then, using the kriging method, metamodel and distribution diagram of design points related to each of the subjects are extracted. In addition to the mass metamodel of each of the discipline, for the injector, the metamodel related to the mass flow rate of the fuel, for the catalytic bed, the characteristic speed, and for the nozzle, the specific impulse of the engine was also produced. Also, four Gaussian, Exponential, Linear and Spherical functions with degree two were compared for each of the metamodels in the kriging method. In this comparison, it was observed that due to the same coefficient of determination, the Gaussian function has less error than other functions and, as a result, better accuracy.