Designing no-frost refrigerator air channel using response surface optimization and investigating its effects numerically

In recent years, the main goal of R&D studies that performed on refrigerators has been improving energy efficiency. In studies performed in order to reach this goal, air flow control and homogeneous temperature distribution are the subjects that being emphasized. It has been proved that, air flow and temperature distributions in refrigerators are being directly related to energy consumption, food freshness and shelf life. Because of that, optimization studies that will be performed on air flow will be based to R&D studies in following years. Up to the recent past, only experimental studies were being performed during R&D activities. Today, these experimental studies are being supported by CFD methods. As CFD methods are more economical and rapid than experimental methods, it is expected that usage of CFD will be increased in the future. Today, design studies that being performed on refrigerators are being supported and validated by CFD applications. However, these studies are mainly based on trial-error method by the experience that gained over years and require long efforts. Besides, as there is a tough competition in white goods industry, time for creating new refrigerator designs has shortened. Taking into account all of these, it is probably not easy to make major changes on refrigerator design in order to optimize air flow and temperature distributions. In this thesis it is mainly air channel, which is the major factor to arrange air flow in refrigerator, was focused on. Observing effects on temperature distribution while parametrically designing and CFD analysing of air channels and interpreting the results are main subjects of this study. These parametrical design activities were carried out by a CFD code and response surface optimization tool that found in a commercial CAE software. Response surface method, which is being used in numerical analysis field nowadays, was introduced in 1951. At the beginning, it was being used in agricultural applications especially in United States. Afterwards, the method was used in quality engineering applications like product process optimization studies. In 1995, the method was used in numerical analysis field for the first time. The first application was in aviation field, in which the method was used to optimize composite wings in high-capacity aircrafts. A year after that, In Japan, the method was used to optimize the automobile seat frames with a different concept. In response surface optimization, which uses statistical techniques, design parameters are defined as inputs; lower and upper limit values are defined for each input parameter; virtual experiment points are created and solved with respect to lower and upper limit values and this process is called the design of experiments; response surfaces are created according to the results of design of experiments; desired constraints and objective functions are set for each output parameter; and input parameter values are obtained as last step. By this way, it gets easier to obtain more accurate design parameters for the optimum design, compared to trial-error method. Within the scope of this thesis, No-Frost refrigerators were examined. A No-Frost refrigerator basically works according to vapor compression refrigeration cycle. Vapor compression refrigeration cycle consists of four processes, these processes are: compression in the compressor, heat rejection in the condenser, throttling in the capillary tube and heat absorption in the evaporator. The function of compressor is to make the refrigeration cycle continuous. It sucks and compresses low-pressure gas phase refrigerant, which comes from evaporator, to condensation pressure. After that, condenser turns the high pressure and high temperature gas refrigerant into low temperature liquid by rejecting the heat taken from the air inside refrigerator. Then high pressure and low temperature liquid refrigerant is sent to the expansion device, which is generally capillary tube. In capillary tube, refrigerant pressure drops down to evaporation pressure. Evaporator then takes heat from air to be cooled and turns low pressure liquid refrigerant into gas phase. With that, refrigeration cycle is completed. Beside compressor, condenser, capillary tube and evaporator, main element of a vapor compression refrigeration cycle is refrigerant. Recently, Isobutane (R600a) has been used as refrigerant in refrigeration cycles for environmental reasons. Unlike conventional refrigerators, No-Frost refrigerators have forced convection air circulation in their cabinets by means of a fan. The reason for these refrigerators are called "No-Frost" is air to be cooled is passed through the evaporator. While passing the evaporator, air contacts with cold plates of the evaporator and the humidity inside air turns into frost. Hence, dry air is sent to the refrigerator cabinet and frost formation is not observed on the walls of refrigerator. On the other hand, in conventional refrigerators humid air contacts with cold cabinet walls so frost formation takes place inside the cabinet. In this study, two different No-Frost refrigerator models, one has division plate and other has not, were examined. Division plate is a horizontal plate that divides refrigerator cabinet into two separate compartments. These two No-Frost devices have three air channel models and two refrigerator cabinet models that were analysed. The air channels mentioned, differ from with their number of air outlets. They have 5, 10 and 12 air outlets respectively. The air channel with 5 outlets has not a division plate on it, while air channels with 10 and 12 outlets have. Firstly, air channel flow geometries were created from refrigerator CAD geometries. After that, meshes at three different qualities were built up for the air channel which has 12 air outlets. The reason for choosing this air channel for mesh study was its complicated structure. Then, each mesh was solved with three different turbulence models and the results were compared with experimental results. The mesh parameters that approach the experimental results more accurately were determined and selected as reference mesh parameters for rest of the study. Within mesh studies, Standart k-ε, RNG k-ε and Realizable k-ε turbulence models, which are available in the CFD software, were used and compared with each other and also experimental results. It has been understood that Realizable k-ε method is the most accurate model for such kind of a study. Beside mesh generation studies, solution setup for numerical models was also made. Some assumptions and simplifications were made while modelling the flow in the refrigerator. For example, solution was prepared for steady state and the flow was assumed as incompressible. Secondly, air flow analyses for air channels and temperature distribution analyses for refrigerator cabinets were performed for both models with and without division plate. Air flow and mass flow rate calculation studies were performed for the air channels with 5, 10 and 12 outlets. On the other hand, temperature distribution studies were performed on the refrigerator cabinets with 5 and 12 outlets. The results obtained from these analyses were also compared with experimental results in order to check the sufficiency of CFD model and determine the imperfections. Points that need to be developed on air channels were also found and interpreted. After that, parametrical design studies were carried for imperfections found in previous step. These studies were carried out for each air channel by response surface optimization tool and new air channel geometries were created. After the parametrical design studies, outlet dimensions of the air channel with 5 outlets were changed in order to obtain desired mass flow rate at each outlet and desired homogeneous temperature distribution in refrigerator cabinet; modifications were made on division plate section of the air channel with 10 outlets; division plate of the air channel also was re-drawn for parametrical study in order to obtain desired mass flow rates at each outlet; location of the profile that directing air flow and located in division plate of the air channel with 12 outlets was changed parametrically in order to increase low mass flow rates at specific outlets and obtain homogeneous temperature distribution in bottom side of refrigerator cabinet. In final phase of the study, current and new designs of both air channels and refrigerator cabinets were compared and interpreted for their air flows and temperature distributions. With that, it became possible to compare new air channel and refrigerator cabinet designs, which were designed with response surface optimization, with current ones and experimental results from the points of mass flow rate and temperature distributions. Hence, the success of design optimization process with response surface method has been verified. In white goods industry, especially in refrigerator R&D area, there are many design optimization studies that uses CFD in the literature. In these studies, firstly initial situation or system is being analyzed using CFD. With that, existing temperature and air distribution is being obtained. After that, some design alternatives are being created with the light of existing system results. CFD analyses are also being performed for new desings in order to compare initial and new designs. Finally, some experiments are made for new system to verify numerical studies. However, this design process is totally a manual one. It is also limited with just a few number of designs as it is doubtful whether these designs are optimum ones or not. This thesis differs from previous refrigerator design optimization studies with its considerably automated design phase. In this thesis, design phase is not based on classical trial-error method as in the previous cases in literature. It is response surface optimization method, which made it possible to obtain most appropriate design(s) for this study. As a result, it has been seen that design process for refrigerator manufaturers can be easier, quicker and more accurate with response surface optimization method. There are a lot of components in refrigerator which are suitable to be designed with response surface optimization. In addition to air channel, other refrigerator components like door shelves, glass shelves, cabinet walls and even a full refrigerator can be parametrically designed by this method. So that, a refrigerator can reach its optimum states in respect to temperature distribution and cooling performance.