Evaluating Vibration Fatigue Correctly - Why Components Fatigue

Why Resonance Destroys Components

Manufacturing a component and performing physical tests are cost‑intensive and time‑consuming. In addition, real tests often only deliver the simple result “passes” or “fails”. Simulation can significantly reduce both development time and costs. It also reveals the root causes of potential failures so that optimization measures can be derived more efficiently. To illustrate a typical fatigue verification under vibration excitation, a water‑cooled electronic control unit is considered.

Repeated loads lead to cyclic stress and thus to material fatigue. Test standards (e.g., ISO 167503 and VW 80000) define load signals based on frequency content and amplitudes. With increasing frequency, component life decreases because the number of cycles per unit time increases. If the excitation frequency lies in the range of a natural frequency, resonance occurs. This increases both deflection and stress, reducing the lifespan further. VW 80000 also includes temperature cycles that influence fatigue life.

Stresses at Different Frequencies and the Corresponding Cycle Counts | © CADFEM / ID: YTOGTV / EZED1U 

Load inputs can be represented as time signals, e.g., acceleration over time. A Fourier transform decomposes this signal into sine components of different frequencies. The resulting power spectral density (PSD) shows which frequencies contain how much energy and which vibration components make up the original signal. As an example, we consider a water‑cooled ECU on a testbench subjected to a Random Vibration Test according to VW 80000 with a PSD profile from 20–2,000 Hz, typical levels, and temperatures from –25 °C to +100 °C – with the objective of meeting the release criteria for vibration fatigue. 

Analyses for a Vibration Fatigue Assessment

Once the fundamental fatigue mechanisms are understood, the next step is replicating the test specifications in simulation. Several analyses are required to evaluate fatigue strength using simulation: static structural analyses, modal analyses, and frequency‑domain analyses. For linear dynamics, the bolt preload contact condition is used as a restart point for the modal analysis. When calculating the modes, the contact condition is linearized, leading to a better representation of stiffness. 

Contact Condition LS2 and Restart Settings of the Modal Analysis | © CADFEM / ID: 9B3Y59

Because the temperature changes during the application of the PSD spectrum, stress distributions at minimum and maximum temperature are also calculated and used as static loads in the vibration fatigue evaluation. Ansys nCode DesignLife reads the results of these analyses and composes them into a duty cycle. The entire test procedure is split into events to include temperature load case stresses. The customized Ansys nCode DesignLife workflow can be added to the Ansys Workbench toolbox as a template.

Defining Linear Dynamics Efficiently

For the vibration fatigue analysis in Ansys nCode DesignLife, transfer functions of the stresses are required – i.e., the structural response to a harmonic excitation of 1g over the considered frequency range. Ansys uses the Mode Superposition Method (MSUP), which employs the previously computed mode shapes and enables very fast harmonic analysis. For precise resolution around natural frequencies, frequency clustering is applied.

Transfer Function With and Without Clustering | © CADFEM / ID: 87OOY9

However, every single frequency point generates its own result set, creating two challenges:

• Stress evaluation dominates computation time (approx. 90% of total runtime).
• Result files can easily reach hundreds of gigabytes.

This effect worsens for components with many cooling fins, due to dense natural frequency spacing requiring many frequency points.

The solution:
Disable the stress output of the harmonic analysis. Instead of writing extensive stress fields, Ansys writes only the modal contribution factors to the compact MCF file. nCode can reconstruct the complete transfer functions using the modal stresses and modal contribution factors. Result: drastically reduced computation time and file size with no loss of accuracy. Additional advantage when using HPC Platform Services (HPS): the modal analysis is stored on the cluster and does not need to be retransferred. If enough compute capacity is available, all three spatial directions can be computed in parallel.

Model with 1 million nodes and 23 modes in the evaluated frequency range. Comparison of File Size and Computation Times for a 3‑Axis Vibration Fatigue Analysis | © CADFEM Germany GmbH.

Calculating Damage with Ansys nCode DesignLife

After the transfer functions have been efficiently determined using the modal‑based harmonic analysis, they can be carried over directly into the fatigue life evaluation. For this purpose, Ansys nCode DesignLife uses an adapted nCode SN Vibration PSD Workflow, specifically designed for the modal‑based approach. The workflow imports both the modal RST file and the compact MCF files, and reconstructs the complete transfer functions from them. The associated PSD spectrum is defined directly within nCode, allowing the combined results from the modal and harmonic analyses to be used flexibly for different PSD profiles — provided that the frequency range of the PSD spectrum lies within the previously calculated transfer function range. The customized workflow is already integrated into the project’s toolbox and is therefore ready for immediate use.

For the fatigue life calculation, the following steps are carried out:

1. Importing the modal stresses: Ansys nCode imports the stresses from the RST file of the modal analysis, considering only the surface nodes.

2. Determining the transfer functions: For each excitation direction and for every node, the transfer functions are calculated by combining the modal stresses with the modal contribution factors.

3. Scaling with the PSD spectrum: The transfer functions are then scaled using the PSD spectrum defined in Ansys nCode, resulting in the PSD of Stress.

4. Determining the stress cycles: Based on the PSD of Stress, Ansys nCode generates a cycle histogram, typically using a distribution such as Lalanne or Steinberg.

5. Calculating the damage progression: Finally, the damage histogram is derived from the cycle histogram and the corresponding S‑N curve, quantifying the resulting fatigue damage.

Ansys nCode DesignLife Workflow | © CADFEM / ID: 76H5KO

Comparison of Results from Simulation and Shaker Test

Ansys nCode DesignLife provides, at the end of the workflow, both the damage per event and the allowable number of repetitions for the entire duty cycle. In the example, running all events through the workflow results in a total damage of D < 1 — meaning the component successfully passes the virtual system test. The evaluation also shows that the highest damage originates from excitation in the X‑direction. At the critical node, it becomes clearly visible which frequency contributes most significantly to the damage and thus plays a decisive role in determining the fatigue life. To validate these results, the control unit is subsequently subjected to a real shaker test. Comparing the frequency responses from the physical test and the simulation is essential to adjust the model to the actual boundary conditions.

Deviations often arise from modified mounting conditions, updated geometry versions, or differing damping values. Only through this correlation can the model and the real system be meaningfully aligned, allowing reliable conclusions to be drawn for future projects — a fundamental prerequisite for robust fatigue‑strength predictions 

Result (Damage / X at Minimum Temperature) from nCode DesignLife in Mechanical and Comparison of FEM Results with Measurement | © CADFEM / ID: 4LEL7G

Those who want to carry out similar analyses reliably themselves benefit from a systematic approach to fatigue assessment - regardless of whether the load originates from random vibration, sinus sweeps, shock, or static alternating loads. In the training Strength Assessment with Ansys nCode DesignLife, we provide exactly this methodological foundation: how to apply transfer functions correctly, build fatigue life models cleanly, interpret damage results with confidence, and correlate test bench measurements with simulation results. The workflow presented here is only a starting point - in the training, you will learn the full range of Ansys nCode methods and apply them directly to your own questions.