In the world of power electronics and electrical machine design, managing heat is paramount. The electrical performance of components like busbars, motors, and transformers is intrinsically linked to their operating temperature. Ansys provides a powerful solution for this challenge by allowing engineers to couple Maxwell for electromagnetic simulation with Icepak for thermal analysis. This process can be done in two ways: through one-way coupling or two-way coupling.
A one-way couple is a straightforward handoff: Maxwell calculates the electromagnetic losses (like ohmic loss), and those losses are passed to Icepak as a fixed heat source to calculate the temperature. A two-way couple creates a more realistic feedback loop: Icepak calculates the temperature, which is then fed back to Maxwell to update material properties that change with temperature, leading to recalculated losses, which are then sent back to Icepak. This iterative process continues until the solution converges, capturing the interdependent nature of electrical and thermal performance.
Let's explore how to set up both types of simulations using a practical busbar example.
One-Way Coupled Simulation
We'll begin with a built-in Ansys example for the DC conduction of a busbar, which can be found in the examples menu under Maxwell/General/DC Conduction
This model features a copper busbar with six current inputs and a single sink at the opposite end.
After solving the initial DC conduction analysis in Maxwell, we can visualize the results for surface current density, voltage, and ohmic loss. The ohmic loss represents the heat generated due to the electrical resistance of the copper.
To perform the thermal analysis, we can send these losses to Icepak. Right-click on the Maxwell design in the project tree and select Create Target Design.... We'll set the Link Type to EM Loss, the Target Design Type to Icepak, and for this example, we will simulate Natural Convection.
This action automatically generates a new Icepak design where the busbar geometry and the calculated EM losses are already imported and configured. The analysis is ready to run. This is a one-way coupling.
Solving this Icepak simulation gives us the steady-state temperature distribution. The result shows a maximum temperature of 52.9°C
Setting Up Two-Way Coupling
The one-way coupling is useful, but it operates on the assumption that the material properties in the Maxwell simulation are constant. In reality, the electrical conductivity of copper decreases as its temperature increases. To capture this effect, we need to establish a two-way coupling.
First, right-click on the Icepak Setup and select Add 2-Way Coupling.... In the dialog, you can specify the number of coupling iterations; for this analysis, we'll use five.
Next, we must define how copper's conductivity changes with temperature on the Maxwell side.
-
Add a Thermal Modifier to the Material: In Maxwell, edit the properties for copper. In the View/Edit Material window, check the box for Thermal Modifier to make its properties visible. Then, for Bulk Conductivity, select Edit.
-
Input Thermal Coefficients: The thermal modifier can be defined using a quadratic equation. We'll input the coefficients that describe how copper's bulk conductivity changes with temperature.
Deriving Accurate Thermal Coefficients
While default values are available, using coefficients derived from experimental data provides the highest accuracy. The coefficients used here were found by fitting a curve to published experimental data. For a detailed guide on this process, please see our video:
Here is a summary of the steps covered in the video:
-
Source Reliable Data: The process starts with sourcing trustworthy experimental data, such as from the "Electrical Resistivity of Copper, Gold, Palladium, and Silver" by R. A. Matula, published by Purdue University.
-
Pre-process the Data: The raw data, typically in Kelvin and resistivity (), is converted to Celsius and conductivity (), since . The data is also filtered to the relevant operating temperature range of the device.
-
Fit the Quadratic Model: A least-squares regression is used to fit the processed data to Maxwell's thermal modifier equation,
which solves for the required coefficients, C1 and C2.
With these settings in place, running the simulation will now execute the two-way coupled analysis.
The Payoff: Comparing the Results
After the two-way coupled simulation converges, we can compare the temperature results directly. The one-way simulation predicted a maximum temperature of 52.9°C, while the two-way simulation shows a maximum temperature of 56.4°C
Why is the Two-Way Coupled Temperature Higher?
The temperature rises more in the two-way coupled case because it captures a critical real-world feedback loop.
As copper heats up, its electrical resistivity increases, which means its conductivity decreases. According to the formula for ohmic losses, a lower conductivity leads to higher losses for a given current distribution.
This creates the following cycle:
-
Maxwell calculates initial ohmic losses based on a reference temperature (e.g., 22°C).
-
Icepak solves for the temperature distribution based on these initial losses.
-
The new, higher temperature is fed back to Maxwell.
-
Maxwell updates the copper's conductivity to a lower value based on this higher temperature.
-
These lower conductivity values result in higher calculated ohmic losses.
-
These increased losses are sent back to Icepak, which calculates an even higher temperature.
This loop repeats with each coupling iteration until the system reaches a stable thermal and electrical equilibrium. The one-way simulation misses this entirely; it assumes the conductivity is fixed at its initial value, thereby underestimating the total heat generated.
In short, in the one-way simulation, Maxwell assumes the copper's properties never change. In reality, as copper gets hotter, it resists current more, which makes it generate even more heat. The two-way coupling captures this feedback, allowing the simulation to converge on a more physically accurate—and in this case, higher—final temperature.
Conclusion
For designs where operating temperatures can significantly influence material properties, a two-way coupled simulation between Maxwell and Icepak is essential for accurate performance prediction. While a one-way coupling provides a good baseline, the iterative feedback of a two-way analysis captures the critical interplay between electromagnetic and thermal physics, leading to a more robust and reliable design.
For a more detailed look into this process, watch our accompanying video here: