A Generator Rotor-stator Cooling Direct Optimization by ANSYS MOGA CFD simulation is a very advanced and efficient way to design better electrical generators. The cooling of the spinning rotor and stationary stator is critical for performance and life. A Rotor-stator Cooling Direct Optimization study uses a smart algorithm called MOGA (Multi-Objective Genetic Algorithm). This Optimization by ANSYS MOGA fluent tool automatically tests hundreds of different cooling designs inside the computer.

This report details an ANSYS Optimization CFD analysis where the MOGA algorithm is connected to ANSYS Fluent. The goal is to find the best possible size for an air injection slot to maximize cooling. This is done by using a Periodic Boundary condition fluent allows, which makes the simulation fast and accurate. This Direct Optimization ANSYS process helps engineers find the best design much faster than traditional methods, leading to generators that run cooler, last longer, and are more powerful. For more advanced turbomachinery CFD simulations and optimization tutorials, visit CFDland Turbomachinery Simulations.

Figure 1: A conceptual diagram of the air injection cooling system for the Generator Rotor-Stator, the subject of this MOGA CFD optimization.

Simulation Process: Fluent-CFD & MOGA Setup, A Periodic Model for Direct Optimization

The simulation process for this Generator Rotor-stator Cooling study began with a simplified 3D geometry. To make the simulation very efficient, only a single, repeating section of the full 360-degree generator was modeled. This is a standard and highly effective technique in turbomachinery, and the Periodic Boundary Condition in ANSYS Fluent was used to make this small section behave as if it were part of the complete machine. The rotation of the rotor at 2000 rpm was modeled using the Multiple Reference Frame (MRF) method, which calculates the rotational effects without needing to physically move the mesh, saving a great deal of computer time. A high-quality, fully-structured mesh was created to ensure the flow physics were captured accurately. Fully developed velocity profiles are applied at both rotor inlet and stator inlet using the profile method in Fluent. The air injection inlet has a constant temperature of 301 K and acts as the cooling source in this optimization study. The rotor wall has a constant heat flux of 5000 W/m² applied as a thermal boundary condition. This heat flux boundary condition forces heat into the cooling air and creates the temperature gradients needed to calculate the Nusselt number.

Figure 2: The simplified 3D geometry model of a single periodic section of the Rotor-stator with the air-injection slot.

Figure 3: The high-quality, fully-structured hexahedral grid used in the Fluent CFD simulation to ensure accurate results in the narrow air gaps.

Direct Optimization Setup with ANSYS MOGA

The direct optimization process uses the Multi-Objective Genetic Algorithm (MOGA) built into ANSYS Workbench to automatically find the best design parameters for air injection geometry. Two input parameters control the air injection inlet dimensions: the width and length of the injection slot. These design variables are constrained within specific ranges shown in Table 1, which ensures the optimizer only tests physically realistic geometries. The MOGA algorithm works by creating multiple design candidates, running Fluent CFD simulations for each one, and selecting the best performers to create the next generation of designs. This evolutionary optimization approach efficiently explores the design space and finds optimal solutions even when the relationship between geometry and performance is complex and nonlinear.

Table 1: Input Parameters and Constraints for Direct Optimization in ANSYS

The objective function for this CFD optimization is to maximize the Nusselt number on the rotor wall, which measures the heat transfer efficiency of the cooling system. Nusselt number is a dimensionless parameter that shows how much better the convective heat transfer is compared to pure conduction. Higher Nusselt numbers mean better cooling performance, so the MOGA optimizer tries to find air injection dimensions that create the highest possible Nusselt number. During the optimization loop, ANSYS Workbench automatically updates the geometry, regenerates the mesh, runs the Fluent CFD simulation, calculates the Nusselt number from temperature and heat flux data, and feeds this result back to the MOGA algorithm. This process repeats for many generations until the optimizer converges to the best design configuration that maximizes rotor cooling efficiency.

Post-processing: CFD Analysis of the Automated Optimization Results

The direct optimization process using ANSYS MOGA started from an initial rotor-stator cooling design with an air injection slot measuring 6 mm width and 3 mm length. This baseline CFD simulation in ANSYS Fluent produced a Nusselt number of 40.129 on the rotor wall, which served as the reference point for all optimization improvements. The MOGA algorithm then automatically generated and tested 99 design points across the entire design space defined by the parameter constraints in Table 1. The History Chart of P1 (width parameter) clearly shows how the optimization algorithm explored different air injection widths throughout the iterative process. In the early stages, between design points 1 and 40, the width values fluctuated dramatically between 0.002 m and 0.01 m as the MOGA performed broad exploration of the design space. After approximately 40 design points, the width parameter stabilized around 0.0095 m to 0.0098 m, indicating the optimizer found a promising region in the design space where better cooling performance occurs. This convergence behavior is typical in genetic algorithm optimization and confirms the MOGA successfully identified optimal air injection dimensions that maximize heat transfer efficiency in the generator rotor-stator system.

Figure 4: The history chart from the ANSYS MOGA optimization, showing how the algorithm learned and converged on the optimal air injection width over 99 design iterations.

The tradeoff chart in Figure 5 confirms this finding and adds another piece to the puzzle. This chart plots every successful design the MOGA tested. We can clearly see a large cluster of green dots (the best designs) in the top-right corner. This is not a random pattern; it is a clear design rule discovered by the algorithm: to get the best cooling, both the width and the length of the air injection slot should be made as large as possible.

Figure 5: The tradeoff chart from the ANSYS MOGA results, visually showing that the best-performing designs (green dots) are those with both large width and large length.

Final Candidate Solutions and Performance Improvement

The final results in Table 2 show the success of this process. The MOGA presented five final “best” candidates. The top performer, Candidate Point 1, had a width of 9.8 mm and a length of 58.2 mm. This optimized design achieved a Nusselt number of 46.374.

Table 2: Final Candidate Solutions from MOGA Optimization in ANSYS Fluent

The most important achievement of this simulation is the 15.6% improvement in the Nusselt number, found completely automatically. For a generator designer or manufacturer, this is invaluable.

1. Increased Power or Efficiency: A 15.6% better heat transfer means the rotor will run significantly cooler at the same power level. This increases the lifespan and reliability of the generator’s insulation. Alternatively, it means the manufacturer can increase the power output of the generator by 15.6% and the new, optimized cooling system will be able to handle the extra heat.
2. Drastic Reduction in R&D Time and Cost: Finding this optimal design manually would require an engineer to guess, model, and test dozens or hundreds of designs, taking weeks or months. The MOGA algorithm did this automatically in a single run. This dramatically shortens the product development cycle and saves a huge amount of money on engineering time and computing resources.
3. Provides a Clear Design Guideline: The simulation provides a simple, powerful conclusion: for this generator design, maximizing the area of the air injection port is the best strategy for improving cooling. This is a clear, data-driven directive for all future designs.