"Military Intelligence," Feature Article, August 2003

military intelligence

A smart system says a helicopter is good to go; a smarter one predicts for how long.

This article was prepared by staff writers in collaboration with outside contributors.

The U.S. armed forces rely on the H-60 to carry troops, cargo, even electronic-jamming equipment. Variations of the helicopter include Black Hawks, Pave Hawks, and Seahawks. The manufacturer, Sikorsky Aircraft Corp., says more than 2,500 H-60s are in service with the U.S. Army, Navy, Coast Guard, Air Force, and Marine Corps.

The Navy and Army have begun to equip the aircraft with sensors that will monitor key moving parts to make the H-60 more reliable, and to get the most out of each machine. Now, to make the system even smarter, the Navy has enlisted a couple of engineering firms in upstate New York to create intelligent software that one day may tell pilots in the air how much longer their craft will fly.

Accelerometers can monitor gearboxes and drivetrains so that technical teams on the ground can use the information to base maintenance decisions on need rather than timetables. This program is called the Health and Usage Monitoring System, or HUMS, an almost lighthearted acronym for military-speak when you consider that it involves sensing vibration.

HUMS tells technicians "so far, so good" or "needs fixing," as the case may be. But if a computer can reference incoming information against predictive models, it can make a prognosis of a part's remaining usefulness in terms of hours.

A predictive maintenance program for the H-60 helicopter includes a model of a transmission gear. Highlight colors show areas of maximum stress.

According to the lead engineer on the modeling project, Greg Kacprzynski of Impact Technologies LLC in Rochester, N.Y., "From day one, in the absence of strong vibration, the system predicts life expectancy. If vibration increases, the model calculates based on the new information."

According to Tony Ingraffea, a Cornell University professor who has been brought in as a consultant, in the best-case scenario, computers on the helicopters will give real-time feedback to pilots.

The engineers are modeling the spiral-bevel pinion gear in the helicopter's intermediate gearbox, one of three transmissions that step down the thousands-per-minute spin of the turbine to hundreds of rpm for the rotor.

William Hardman, diagnostic engineer at the Naval Air Systems Command in Patuxent River, Md., assembled data from two sample gears, which he had run to failure on test stands. In each test, a gear tooth was notched to hasten crack initiation.

One gearbox was subjected to various loads "to capture what you see in service," he said. A second started at 2,340 ft.-lbs. of torque and continued to 2,000 (3,173 newton-meters easing to 2,711) after damage became serious.

Gleason Corp. of Rochester, N.Y., provided gear data and a model of three consecutive teeth to cover the complete load cycle for any tooth. The model was translated into an Ansys file, according to Avinash Sarlashkar, director of technology at Impact. This software, from Ansys Inc. of Canonsburg, Pa., calculated physical values and generated the finite-element mesh for analysis.

According to Sarlashkar, who with Brad Lamirand did much of the Ansys work, simulations of crack propagation were done by Fracture Analysis Consultants Inc., Ingraffea's consulting firm in Ithaca, N.Y. Ingraffea, an ASME member, used a program called Franc-3D, developed by his research group at the university, the Cornell Fracture Group.

So far, the researchers have deemed their models accurate in calculating a crack length of about a half-inch, or 2 cm. Beyond that point, they say, load-sharing factors that were not modeled may have come into play. According to Sarlashkar, the project is in its second phase under a Small Business Innovative Research grant, and the model is being adjusted to match the test data more closely.

A crack in a gear tooth, whether it is accidental or intentional, will produce vibration, Sarlashkar said. The challenge is to get the complex events of gear failure to fit into a computer.