Air flow study in a two branches manifold

This project consists on a cooling system for a wind turbine generator. The main concept used to cool the generator is based on a manifold, which consists in an arrangement of pipes used to redistribute the flow of a fluid or gas, typically from a single inlet to a number of outlets, or vice versa. The first idea was to design a manifold with an inlet and four outlets using an equation solver program called EES, SolidWorks Flow Simulation as an application of computational fluid dynamics (CFD) analysis and a wooden model built and tested, and try to get the total control of the system. A first step has been to carry out all the calculations for a two branch model. This has given us a general idea of how the flow works and how it is distributed between the branches, and this two branches model (instead of four) simplifies the calculations of the model at first. This model, which its main branch is 30 mm wide and the outlet branches widths are 19.5 and 21.5 mm, has been introduced to an EES program, designed in SolidWorks and built in wood to obtain the pressures, velocities and flows for the circulating fluid along the conduits. Once the results for the three methods have been obtained, these has been analyzed and compared between them to draw some conclusions about the similarities or differences of the data from the three methods. After the first analysis we have decided to build again a two branches model but with narrower conduits, of 5 mm both the main branch and the outlet branches. This second model has been designed to see if the flow behaves the same way when it circulates through narrow conduits. Here we have had to design a new method to measure the parameters inside the thin conduits. This method, which we called “Sphere method”, consists of a small lead ball which moves more or less depending on the pressure applied on its surface. Once compared to the other two computational methods, it has been tested in the wide branches model and compared to the results obtained there measuring with a Pitot tube, and has resulted to be more accurate than expected, so we have decided to enhance it by working with more lengths to calculate the velocity profile curve and improve the calibration of the threads so the results were even more precise. In other matters we have studied the pressure loss in a tubular pipe which is narrower in the middle. This study has been performed using three computer programs; the already known EES and SolidWorks Flow Simulation Studio, and CFDesign. This last comparison has been expected to show the differences between three methods based on theoretical calculations