A simple optimisation approach adapted to design a megawatt scale YASA generator
In the preliminary design process of a megawatt scale Yokeless and Segmented Armature (YASA) generator for wind turbine application, sizing equations are used to generate the possible generator designs with the same rated power and rated rotational speed.
Optimisation process is needed to choose the best YASA generator design based on the requirements and constraints of the application. The optimisation of the YASA generator involves many conflicting objective functions and constraints, such as outer diameter, total power losses, cost of active materials and structural mass. Taken into account of these objectives functions, a design reference map can be generated as shown in Figure 1. This design reference map visualises how the outer diameter, power losses, cost of active materials and structural weight changes with the air-gap flux density Bg and outer-to-inner diameter ratio kdunder a constant current loading.
To solve these conflicting objective functions in the optimisation of a YASA generator, a simple optimisation approach is adapted which allows the users to prioritise one or more objectives based on their requirement. First, the weight (wm) of each objective function is computed with the pseudo-weight vector approach. Once the weights of each objective function of all the possible YASA generators are obtained, they are compared with a standard weight (swm) for each objective function defined by a user. The total deviation of the weights of the objective functions from their corresponding standard weights (swm) is calculated with the equation,
where M is the total number of the objective functions. It is important to note that the sum of the standard weights for all objective functions should equal to 1. Using this method, designers can set the standard weight for each objective function according to their requirement.
For example, to design a 1MW YASA generator, the constraints applied are outer diameter below 5 m and cost of active materials below € 100,000. It is common that some objective functions can be justified as being more important than the others. Of all four objective functions considered here, power losses should be prioritised as it not only determines the generator’s efficiency but also will affect the design of the cooling system. Structural weight follows considering these generators may be used in offshore wind farm, which heavy weight may cause problem during transportation and installation. As the outer diameter and cost active materials already set as constraints, all the values within that range are acceptable but lower value will be great in this case. The standard weights for each objective function are: power losses - 0.6, structural mass - 0.2, outer diameter - 0.1 and cost of active materials - 0.1. The calculated weight deviations and optimised YASA model is shown in the design reference map (Figure 1). The total weight deviations are the blue lines plotted while the YASA generator model with the lowest total deviation to the standard weight is the red dot on the map.where M is the total number of the objective functions. It is important to note that the sum of the standard weights for all objective functions should equal to 1. Using this method, designers can set the standard weight for each objective function according to their requirement.
Figure 1: Design reference map of 1MW YASA at 5×104 A/m with weight ratio and the optimised YASA machine (Bg=0.87T and kd=0.83) plotted. (The aimed weights on the objective functions are: Aimed weight of total power losses is 0.6, structural mass is 0.2, cost of active materials is 0.1 and outer diameter is 0.1. Constraints are outer diameter ≤ 5 and cost of active materials ≤ €100,000.)