Welcome To CEM

Materials Program

NASA Arbor

Composite Rotor Lifetime Testing

The University of Texas at Austin Center for Electromechanics (UT-CEM) has developed a 2 kW-hr flywheel battery for energy management on a hybrid electric urban bus. The battery will recover braking energy and store excess energy generated by the prime mover. The flywheel rotor, fabricated from high-strength composites, spins at 36,000 rpm at full charge (~825 m/s tip speed), and is housed in a vacuum enclosure to minimize windage drag. A cross-section of the flywheel system design is shown. Ensuring flywheel safety is a major issue that must be addressed in using flywheels for transportation applications. In support of this activity, the durability tests performed under Phase IV of the DARPA Flywheel Safety Program, focused on this flywheel design.

The transit bus preloaded flywheel design (pictured below) is assembled from multiple individual rings fabricated from high-strength carbon fiber impregnated with a toughened epoxy. All composite rings are assembled by interference fit which preloads the rings and maintains a radial compressive stress state at operating speed. A solid metallic hub or shaft serves as the foundation of the preloaded flywheel design. Testing under earlier phases of this program have shown that preloaded composite flywheels are robust, stable structures

CEM performed the single cycle failure test at Test Devices, Inc The purpose of this test, by performing an overspeed test, is to document the failure mode of an “as built” flywheel. This failure pattern serves as a baseline for comparison against the failure mode of similar flywheels after being subjected to cyclic fatigue loading, later in the program.

The single cycle, overspeed test, as predicted, revealed that the failure mode was not a hoop fiber burst, but a controllable loss of mass balance. The flywheel’s peak operational speed is about 36,000 rpm. The flywheel maintained excellent mass balance up to and past this speed. At about 42,000 rpm, the flywheel began to show slight balance changes, which was expected, since some of the flywheel’s internal rings had transitioned from radial compression to radial tension and physically separated. At about 47,500 rpm (1120 m/s), the flywheel showed a rapidly increasing mass balance shift due to further ring-to-ring and rim-to-metallic hub separation. Driving the test to higher speeds, at this point, would have resulted in increased vibrations leading to quill shaft failure and the flywheel dropping to the bottom of the spin test pit. CEM decided to stop the test so the flywheel could be safely spun down to rest, and available for post-test inspection. In conclusion, this test successfully characterized the “as built” overspeed performance of the first Phase IV (type I) flywheel.

CEM completed fatigue cycling of the first fatigue, Type I flywheel. Fatigue testing of this flywheel progressed smoothly, but with a slight anomaly, which occurred at about 6500 cycles. At 6500 cycles, a small amount of composite material was lost off the flywheel ring’s outer surface. The flywheel was sent back to CEM for inspection. The pictures show the section on the outer diameter that experienced the material loss. About 0.030” of material was lost over a length of about 5/8” axially. The damaged portion of the outer ring appeared to be a radial tensile failure. However, direct cause of this remains uncertain. Possible contributors are 1) towpreg material flaw, 2) contact damage, or 3) tensile peel aggravated by both elevated temperature effects (resin modulus reduction) and free edge stress states. However, a preliminary FEA was completed to evaluate free edge stresses with room temperature resin properties. Only a minor reduction in radial compression stress was noted (less than 15%).

The decision was made to send the flywheel back to Test Devices and resume testing at a slightly reduced maximum speed (39,200 rpm down to 36,000 rpm). Testing progressed normally to about 20,000 cycles with no reported problems. Since this flywheel experienced a problem, believed to be of an isolated, manufacturing nature (resin lean carbon fiber towpreg), the decision was to terminate this test and use the remaining test budget to perform additional testing on the second fatigue flywheel.

CEM completed fatigue cycling of the second fatigue (Type I) flywheel in April of 2002. This CEM flywheel was tested through duty cycles representative of the service application, which is for load leveling on a hybrid-electric Transit Bus. Over a period of six months, the flywheel had undergone speed excursions from 27,000 rpm to 36,000 rpm, with a peak tip speed of 825 meters/second. The test was performed at elevated temperatures of about 140o F. At about 93,500 cycles, a very slight mass center shift was detected (50 m/inch, equivalent to 28 gm-mm). The flywheel and spin pit drive assembly were inspected. No cause was found and testing resumed.

At about 101,000 cycles, small surface “tear” in one face of the flywheel was detected by the Test Devices’ health monitor system. The associated mass balance change was again minor (200 m/inch, 114 gm-mm equivalent). The test was interrupted and the flywheel rotor was inspected to explore the change. Shown in the picture is surface tear discovered at this point in the testing. Testing of the flywheel was again resumed, and proceeded smoothly through 112,000 cycles without any additional balance change.

On September 20, 2002, this flywheel was then subjected to an overspeed test. As hoped, this second fatigue flywheel performed nearly identical to the overspeed test on the “as built” flywheel, performed earlier in the program. Failure mode for the second fatigue flywheel was preload loss, leading to a mass balance shift. Maximum speed achieved, limited by vibration in the quill shaft to avoid quill failure, was about 1100 m/s. This overspeed test reveals that no structural degradation was identified between the second fatigue wheel (tested to 112,000 cycles) versus the “as built” wheel.

Successes from this test

• Significantly increased the number of fatigue test cycles achieved for a full-scale composite flywheel operating in realistic simulated service conditions. The test was conducted at actual operating speeds and at elevated temperature (average 140°F.)

• The Test Devices Rotor Health monitor demonstrated the importance of measuring small changes in mass distribution in real time. Although the small surface tear created only a slight change in rotor performance, the monitoring system readily detected the anomaly. This is a significant pathfinder to monitoring methods that may be employed in-service on future flywheel systems.

• For future space-based applications: This durability test exceeded the approximate 90,000 cycle requirement for space-based composite flywheels for energy load leveling, LEO missions for a 15 year service life.

The Center for Electromechanics has reached an important milestone in the development of high performance energy storage flywheel systems. A composite flywheel has just passed 100,000 test cycles, repeatedly accelerated between two duty cycle speeds. This is highlighted as a significant achievement for the newly forming flywheel industry, and this success, along with related achievements of others in the community, is helping to demonstrate the durability and reliability of composite flywheels operating for extended periods of time at elevated, cyclic stress levels. At the same time, these activities benchmark both facility and protocol requirements for performing flywheel durability tests.