Printable Version

Predicting Potential Failure

SwRI researchers team with industry to develop aircraft engine reliability software

By R. Craig McClung, Ph.D., and Michael P. Enright, Ph.D.

Commercial air travel is one of the safest modes of transportation available today — much safer per passenger mile than driving your car across town. However, the news headlines remind us from time to time that serious accidents can occur, and the resulting fatalities or injuries are certainly a tragedy to those affected. Furthermore, the significant increase in the total volume of commercial air travel projected for the coming decades means the number of accidents will inevitably go up unless incident rates are reduced.

Many different factors can contribute to aircraft accidents, of course, including pilot error, severe weather or poor aircraft maintenance. Engine failure is a relatively rare but still significant factor, and therefore an important target for improved design practices. The challenge is to improve engine reliability without compromising the performance or operating efficiency of these highly complex mechanical systems. Southwest Research Institute (SwRI) engineers, working closely with the aircraft engine industry, are making a major contribution to improvements in engine safety through their development of the DARWIN® computer program.

The crash landing of United Airlines flight 232 at Sioux City, Iowa, on July 19, 1989, captured the attention of the nation. The Stage 1 fan disk of the center engine in the DC-10 broke apart in flight and ruptured the plane’s hydraulic lines, severing the link between the pilots and all of the plane’s control surfaces. Remarkably, the experienced pilots were able to gain partial control of the aircraft by using the throttles of the two remaining engines, but 111 of the 296 persons aboard died during the landing. Subsequent investigations found that the titanium fan disk separated when a fatigue crack developed as a result of a metallurgical anomaly, known as “hard alpha,” that formed when the titanium ingot was produced.

As a result of the Sioux City incident, the Federal Aviation Administration (FAA) requested that the aircraft engine industry, through the Rotor Integrity Sub-Committee (RISC) of the Aerospace Industries Association (AIA), determine whether a “damage tolerance” approach could be implemented to reduce the rate of such uncontained rotor events. The industry working group concluded that this enhanced life management method could be a practical supplement to the traditional “safe-life” fatigue design method. The safe-life method considers nominal engine conditions (including minimum material properties), while the damage tolerance method specifically addresses anomalous conditions (such as rogue metallurgical features) that occur rarely but can have serious consequences. During the development of the damage tolerance approach, it became obvious to RISC that the capabilities and effectiveness of this emerging technology could be significantly enhanced by further research and development. In early 1995, the SwRI team, with guidance from RISC, proposed and was awarded a research project by the FAA to address the shortfalls in data and technology. The project team included four major aircraft engine companies, now known as GE Aviation, Pratt & Whitney, Honeywell and Rolls-Royce Corporation.


Dr. R. Craig McClung (right), a program director in the Mechanical Engineering Division, has more than 25 years of research and program management experience in fatigue crack growth, fracture mechanics, mechanical behavior of materials, structural life prediction and software development. Principal Engineer Dr. Michael P. Enright (left) specializes in reliability-based life prediction with an emphasis on probabilistic damage modeling, time-dependent system reliability and probabilistic fatigue and fracture.


DARWIN

The centerpiece of this “Turbine Rotor Material Design” (TRMD) research project was the development of the computer program that became known as DARWIN (Design Assessment of Reliability With INspection). DARWIN uses sophisticated risk assessment methods to determine the probability of fracture of a rotating engine component by integrating finite element stress analysis results, fracture mechanics models, material anomaly data, probability of crack detection and inspection schedules within a powerful graphical user interface (GUI). Early development of DARWIN focused exclusively on the threat of fracture caused by hard alpha anomalies in titanium rotating components. DARWIN and the enhanced life management process was one thrust of a multi-faceted FAA effort to reduce risk in titanium rotors, in conjunction with the development of improved titanium melting practices and better inspection techniques.

The scope of RISC, and ultimately of DARWIN, expanded following another incident on July 6, 1996, at Pensacola, Fla. An MD-88 experienced an engine failure just prior to takeoff. A fatigue crack — this time originating in a microstructural region that had been damaged during the drilling of a bolt hole in manufacture — caused separation of the fan disk. The resulting penetration of the cabin by engine debris caused two deaths and two serious injuries. Following the investigation, the FAA asked RISC and the TRMD project to address the threat of surface damage in all rotor materials, caused by anomalous manufacturing or maintenance practices, with an initial focus on bolt holes.

The success of the initial four-year TRMD project in addressing the hard alpha threat — and the new focus on surface damage — led to a six-year TRMD Phase II project that began in 1999 and concluded in 2005. The DARWIN software, which had matured to Version 3 by the end of TRMD-I, continued to expand capabilities through Versions 4, 5 and 6 during TRMD-II.

As DARWIN evolved, it captured the attention of the engine companies that had been guiding its development. Those companies realized that DARWIN was not just a research tool but also could potentially play a significant role in their own production design systems. They asked SwRI to consider licensing and supporting the software commercially so engine companies could use the tool with the confidence that training would be provided, questions answered, bugs fixed and new features added, thereby providing a stable path into the future. SwRI agreed to take on this role, and with some support from the FAA to establish the necessary infrastructure, it began licensing and supporting the code commercially in 2000. The DARWIN customer base has grown steadily as the code has continued to expand. Five aircraft engine companies, performing design in five different countries, are current DARWIN licensees, along with two foreign government laboratories. All licensing fees received by SwRI are dedicated to user support and training and to the further development of DARWIN capabilities for the benefit of licensees. All U.S. government agencies are eligible to receive a royalty-free license, and DARWIN has been distributed to three branches of the U.S. military.


DARWIN implements a general probabilistic fracture mechanics methodology that incorporates design inputs from component structural analysis and material properties as well as probabilistic characterizations of initial material or manufacturing anomalies, uncertain inspection times and inspection effectiveness. The probability of fracture is calculated statistically as a function of increasing flight cycles, taking into account the distribution of stress throughout the volume of the component.


How DARWIN Works

DARWIN is typically used by an engine manufacturer during the design or certification process. The analyst begins by importing the finite element structural model and corresponding stress analysis results into DARWIN. The GUI allows the analyst to display and manipulate the model and results, either in two-dimensional form to address inherent material anomalies such as hard alpha or in three-dimensional form to address surface damage. The GUI makes it easy to construct detailed fracture mechanics models for hypothetical cracks in different regions of the component. For each potential crack location, the GUI also facilitates the selection of appropriate statistical distributions describing the frequency and size of anomalies that may be present initially, the uncertainties associated with stress analysis results and fatigue crack growth life calculations, the probability of crack detection as a function of crack size, and the distribution of potential inspection times. A new feature in the most recent DARWIN version allows users to link their own algorithms to calculate the number of flight cycles required to form cracks at material anomalies. Some anomalies, such as the rarely occurring hard alpha, are so brittle that cracks form immediately. Advanced fracture mechanics algorithms are then used to simulate the growth of fatigue cracks in service, and advanced probabilistic algorithms are used to sum the risk in the component over the disk lifetime for all potential anomalies. These same algorithms identify which regions of the component contribute the most to total risk. The GUI is used to visualize all significant results and eventually to prepare formal written documentation of the analysis in standard formats. The GUI provides extensive online documentation of code functionality and theory.

DARWIN calculates the risk of fracture caused by the specific damage sources considered on each flight cycle and the total accumulated risk over the projected lifetime of the part. This total risk can be compared with a “design target reliability” (DTR) to determine if the design is acceptable, if the design must be modified, or if additional inspections must be added to reduce the projected risk. The FAA has published an Advisory Circular (AC 33.14-1, January 2001) that defines acceptable DTR levels based on historical experience and safety goals and describes an acceptable process for performing the reliability calculation to meet the appropriate Federal Aviation Regulations for high-energy rotating parts of aircraft gas turbine engines. Additional advisory material is currently under development by the FAA to address surface damage at bolt holes, and further advisory material for other anomaly types is anticipated. DARWIN is consistent with the processes documented in AC 33.14-1 and other draft advisory material.

DARWIN is particularly attractive to the engine industry because it combines state-of-the-art technology with a powerful user interface that enhances efficiency. DARWIN developers at SwRI have derived and implemented new families of highly accurate stress intensity factor solutions that calculate the driving force for the growth of fatigue cracks in complex three-dimensional stress fields. Sophisticated new probabilistic algorithms also have been developed that calculate the total component risk with high accuracy at speeds that can be orders of magnitude faster than conventional methods. Because risk calculations can take hours to complete for complex engine structures and detailed engine histories, these speed improvements can have a direct impact on design productivity. However, perhaps the most powerful feature of DARWIN is its GUI, which enables the analyst to visualize and manipulate all models and data with simple mouse maneuvers, reducing total analysis clock time while minimizing errors. One engine company reported that DARWIN enabled its engineers to complete a full certification analysis in 20 percent of the time normally required. In recognition of these and other advances, DARWIN earned an R&D 100 award from R&D Magazine as one of the 100 most technologically significant new products of 2000 (see Technology Today, Fall/Winter 2000).


The DARWIN GUI helps the user to visualize and manipulate a three-dimensional component model (top), identifying the location on the surface where a crack might initiate, slicing the model along the plane where this crack is most likely to grow, and determining the stresses on this plane that will govern the growth rate of the crack (middle). The DARWIN GUI also addresses inherent material anomalies that can occur anywhere in two-dimensional axisymmetric component models (bottom), helping the user to break down the total volume of the part into a series of zones with similar risk characteristics.


The DARWIN Team

SwRI engineers work closely with their technical colleagues at the four partner engine companies, who serve as a steering committee and provide detailed guidance and evaluation for new DARWIN versions. Other engine companies in the RISC industry group provide additional input and review. The interdisciplinary SwRI team comprises specialists in fracture mechanics, probabilistic analysis, software engineering, metallurgy and materials testing. Small business subcontractor Mustard Seed Software works closely with SwRI staff to develop the GUI, and a former member of the SwRI DARWIN team, now on the faculty of The University of Texas at San Antonio, contributes additional new technology. FAA staff members at the William J. Hughes Technical Center in Atlantic City, N.J., and the Engine and Propeller Directorate in Burlington, Mass., provide active review and guidance.
DARWIN development is continuing under a third FAA grant, “Probabilistic Design for Rotor Integrity” (PDRI), which began in 2005 and will continue through at least 2010. Under PDRI, DARWIN is being enhanced to address additional types of surface damage in rotor materials, as well as inherent material anomalies in nickel-based superalloy materials used for rotating components in the hottest regions of the engine. New algorithms also are being developed to improve user efficiency for all inherent anomaly problems, including hard alpha titanium.

Other Applications

DARWIN’s impact is not limited to commercial aviation. The U.S. Air Force Research Laboratory (AFRL) is funding research at SwRI to enhance DARWIN for the unique problems faced by military engines, and some licensees use DARWIN to support development of new engines for advanced military applications. The development of the general three-dimensional GUI has also opened up the potential application of DARWIN to structural components outside the engine world, such as the highly stressed and highly complex components in rotorcraft drivetrains.

DARWIN technology is also highly synergistic with the technology in two other SwRI software products, the NASGRO® computer program for fracture mechanics and fatigue crack growth analysis, and the NESSUS® software suite of probabilistic analysis tools. For example, NASGRO and DARWIN share many stress intensity factor solutions, while DARWIN and NESSUS share many features of the finite element model GUI. NASGRO and NESSUS, both of which are also R&D 100 winners, are each applied in many other industries, including aircraft, rotorcraft, spacecraft, automotive and offshore structures; gas turbine engines for power generation; pipelines; and others. 

More information about DARWIN is available at www.darwin.swri.org.

Comments about this article? Contact McClung at (210) 522-2422 or craig.mcclung@swri.org or Enright at (210) 522-2033 or michael.enright@swri.org.
 

Published in the Winter 2007 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Joe Fohn.

Winter 2007 Technology Today
SwRI Publications SwRI Home