Arlin Nelson, Sr. Systems Integration Engineer
MTS Systems Corporation
Arlin Nelson Shares Keys to Faster Fatigue TestingQ: How long has test cycle time been a major challenge for aerospace fatigue testing?
Nelson: Test cycle time concerns have been around since the 1950s, when the earliest analog controllers were introduced. External function generation was required with first-generation controllers, and it took a roomful of programmers to write the code. These controllers managed a small number of control channels, sometimes only one, and there was no such thing as null pacing. It could take a significant amount of time to restart tests after an interlock occurred.
Test setup and optimization was slow, tedious and error-prone with these systems, but they worked. Both the B-2 and F-22 aircraft were tested using analog controllers.
The next generation of analog controllers included features such as A/B compare and built-in single-channel function generators, but for more complex tests, external interlock chains and external function generators were still required. It was also difficult to collect data, and downright painful to make tests run fast.
The first digital controllers appeared the late 1980s, and brought huge advances in ease of use by automating many manual processes. They included internal function generators capable of handling large numbers of control channels, along with multichannel null pacing and bumpless starts. Due to computer processing and network speed limitations, however, there were still limits regarding load conditions and the amount of data that could be displayed at one time.
Q: That brings us to today. How do the latest digital controls perform?
Nelson: Due to a combination of advanced software technology and improvements in computing power, today’s digital controls can handle many inputs per channel as well as a staggering number of control channels and load conditions. The largest MTS controller to date manages 640 control loops in one system!
Furthermore, enormous amounts of data can also be collected, exported and distributed. You can even provide secure access to test data from a remote location, in near real time.
Q: With all of this technology, why is it so hard to make fatigue tests run faster?
Nelson: It’s mainly due to the fact that there are some things that test engineers can control, and others that they cannot. Making a test run at 80% of the fastest possible speed is extremely difficult and can take weeks or months to achieve. Getting up to 95% is just as hard as getting to the first 80%, and reaching 100% from there is at least as hard as getting to the first 95%.
In my opinion, most tests run at 80% of their fastest possible speed or less.
The ultimate testing speed you can achieve is most often determined during test design. The choices you make at the design stage will result in cycle times that you have to live with for the rest of the test. Some modifications are simple to handle in the testing stage, such as changing servovalve sites. Others are much harder to diagnose and remedy, such as having pin connections with too much play. There are other issues that will appear only after you decrease your cycle time.
Q: What steps can be taken to accelerate testing?
Nelson: Again, correct servovalve size is a great place to start. People tend to use too large of servovalves for their needs, but it’s important to remember that the larger the servovalve, the less stable the control channel. From my experience, I’d estimate that people err on the large side 60% of the time.
Optimizing proportional gain, or “P-gain,” is another early step to take. The bigger the P-gain, the faster you’ll go. As P-gain goes up, however, channel stability goes down. So the trick is to find the highest P-gain possible and remain stable. How you constrain the test article also impacts testing speed. Generally speaking, “floating” test articles run slower than fixed test articles.
Q: What advances have emerged on the technology side?
Nelson: Recent software innovations have enabled huge improvements in structural testing speed. First, there’s forward loop optimization, or FLO, which applies algorithms that automatically condition valve commands before they are sent to the servovalve. When used correctly, it can make a channel appear more stable.
Next, there’s PSO, which is short for Profile Segment Optimization. PSO automatically adjusts the transition time as a test runs to make the transition as fast as possible, while still staying within user-defined error limits.
And now we get to the game changer. MTS has developed a method that compensates for the cross-coupling of actuators through the test article. This feature, referred to as CCC and short for Cross-Coupling Compensation, fully compensates for the influences of all actuators present without compromising accuracy or introducing additional strains on test articles.
CCC eliminates the need to input cross-coupling information manually and allows the use of unit load cases to provide automated coefficient generation. Our customers have reported improving their test speeds by up to four times after employing CCC.
Q: What’s the takeaway for test labs looking to accelerate their testing?
Nelson: It’s that making a fatigue test run faster does not happen by accident. It all starts with smart test design, including a tight reaction fixture, rigid loading system, well-tuned actuator and ability maintain stable hydraulic pressures. All of this requires trial and error and experience.
Accelerated testing also requires taking full advantage of the latest software technology. Especially valuable are the automated capabilities available for optimizing the forward loop, specifying the proper profile transition times, and compensating for actuator cross-coupling.
>> DOWNLOAD MTS AEROPRO™ SOFTWARE BROCHURE
>> CONTACT MTS
>> SIGN UP TO RECEIVE OTHER VERSIONS OF FORCE & MOTION