Random and Field Data Replication (FDR) are two test options available in the VibrationVIEW software. Both may appear as suitable test modes in many circumstances. So, how do you decide between a random or an FDR test? In this article, we list several factors to consider.
Fundamental Differences: Random and FDR
Figure 1. Acceleration waveform for a random vibration test.
Many test standards require random vibration tests to produce realistic conditions in the lab. Random test profiles are defined as a breakpoint table, which is outlined by the standard or test lab.
A random vibration test simultaneously outputs random vibration over the test’s frequency range for a defined period. The acceleration waveform for a random test displays a “ragged” plot because the test excites the device under test (DUT) with random amplitude vibrations across the defined frequency range (Figure 1).
Figure 2. Acceleration waveform for an FDR vibration test.
The FDR software builds a test from field data. FDR plays back an imported recorded time-history file as a control reference.
By replicating a waveform from the field, FDR includes the higher accelerations that a random vibration test may average out. If the field vibration is not Gaussian random, then the FDR test profile is not. The result is a 1:1 replication (Figure 2).
Comparison of Random and FDR Tests
The goals of random vibration testing and field data replication both involve reproducing accurate conditions in the lab. Whether to implement one or the other depends on the DUT’s operational environment, test requirements, and lab capabilities.
Random Vibration Testing
Random vibration tests are largely more reflective of real-world conditions than sine tests. Rarely does a DUT experience one single-tone vibration at a time, but rather a combination of excitations over a range of frequencies. As previously explained, random vibration tests simultaneously run multiple excitations over the test frequency range.
However, random tests are not an exact replication of the real-world environment, only an approximation. A random test profile can over- or under-test a product if the team does not compare it to field conditions.
Random vibration test control assumes a Gaussian distribution of data, which is not always the case in the real world. The averaging methods software programs use for random testing can “average out” extreme peaks that could possibly be damaging. The engineer would need to reincorporate the high acceleration peaks or confirm that the peaks pose an acceptable risk.
That said, random vibration testing is a widely accepted method for validation and qualification. It provides statistical averages often required in a testing environment.
Random testing also has the benefit of test acceleration. In VibrationVIEW, engineers can apply Kurtosion control to safely accelerate test profiles, bringing the DUT to failure more quickly. They can also access other advanced random test capabilities, like fatigue damage spectrum (FDS), TruClip, and virtual channels, to adjust their tests for safety and accuracy.
Random Potential Applications
- Testing to specifications with predefined/standardized PSDs
- Component durability across varied environments
- Early validation of a new design
- Accelerated life testing to failure
- Benchmark design revisions under identical conditions
- Multi-environment product qualification
Field Data Replication
Most significantly, engineers can use FDR for a 1:1 replication of individual operating environments. If a product operates in a single environment, its vibration data can be recorded and replicated in the lab.
For example, imagine a qualifying road test for a vehicle is the same number of laps on a defined test track. In this instance, the engineering team can use FDR to replicate a recording of the vehicle’s vibrations on the test track in a lab. Instead of taking every iteration of the vehicle to the road test, they can bring the road test to the lab.
However, an FDR test represents one specific environment, and most DUTs experience various conditions throughout their operating life. Some programs require engineers to test isolated conditions, and FDR caters to these types of applications.
Additionally, accelerating an FDR test is not easy. Consequently, if an engineer intends to use an FDR test to bring a DUT to failure, they must run numerous cycles of the recording.
With FDR, there is no need to approximate the field environment, and it is useful in many situations. However, this test method does not replace random vibration testing. As the replicated waveform is identical to the field data, it doesn’t capture potential variability. FDR provides a representation of one instance in the environment, while a random test provides a statistical average.
FDR Potential Applications
- Replicating a defined environment without iterative field testing
- Root-cause investigation of field failures
- Testing a single dominant operating condition
- Comparing lab results to real-world behavior