Structural Health Monitoring of Advanced Aerospace Systems

David C. Zimmerman, Ph.D., Assistant Professor, UH
Mike Grygier, Ph.D., JSC
George H. James, Ph.D., Post-Doctoral Fellow, UH

THE DETERMINATION OF LOCATION and extent of structural damage is of significance in many engineering systems ranging from advanced aerospace structures such as the International Space Station Alpha, the Space Shuttle, commercial and military aircraft, future re-usable launch vehicles and spacecraft, as described in the NASA Millennium initiative, and to the nation's civil infrastructure such as bridges, dams, offshore platforms and large buildings. Inspection of these structures in the past has mainly been based on visual methods, with occasional employment of conventional non-destructive evaluation techniques by ultrasonics or acoustic emission. However, this methodology is time-consuming, costly, does not necessarily provide a quantitative measure as to the extent of structural damage, and, is not applicable to a wide range of advanced aerospace structures. For example, reusable spacecraft such as the Space Shuttle, is typically covered with thermal protection materials, making visual inspections difficult. Thus, it is necessary to develop a structural health monitoring system to detect and locate structural damage as it occurs. Our general concept for structural damage detection is depicted in Fig. 2.

Above. Timothy Cao (l.) consults with Dr. Mike Grygier (center) and Dr. George James, Post-Doctoral Fellow, on potential structural damage to portions of the tail of the Space Shuttle in flight.

Project Overview
The overall objective of the research program is to develop and validate procedures to allow engineers to enhance structural safety, maintainability, and performance, using global and local vibration measurements, structural modeling, and data processing algorithms. The key concept is in monitoring the "vibration signature" of the structure and utilizing observed changes to locate and estimate the extent of damage. In the literature, this approach is often termed System Identification (SI).

Figure 1. Direct inspection requires extensive extravehicular activity and does not always reveal structural anomalies.

The development of SI approaches to date has been focused on truss-type structures. Historically, many space structure concepts (SSF, ISSA, mission-to-planet-earth, . . .) utilized truss-type structures as the major structural component. Logically, laboratory verification of various methodologies has also focused on truss-type structures. Unfortunately, truss structures are the least complex structural element to model and test, so there exists a wide technology gap between successful methodology for truss structures and their follow-on performance for other types of structures. The specific focus of this study will be the development and experimental testing of system identification techniques for plate- and shell-like structures. Plates and shells are utilized in a wide range of engineering structures, including the laboratory modules of the ISSA, fuselage and wings of the Space Shuttle and aircraft, as well as most of the structures that we see in day-to-day life. In particular, the following topics will be investigated:

Project Accomplishments
Experimental Testing
The development of the Space Shuttle Orbiter vehicles in the late 1970s and early 1980s included detailed testing of several significant structural components. The Vertical Stabilizer Assembly (VSA) test article was one of these components. It was used as an acoustic test article from 1979 to 1981 and as a static fatigue test article in 1982 to verify that the static and fatigue response of the shuttle vertical tail met the design requirements. The VSA was recently returned to NASA-JSC for a series of experiments to support development and demonstration of model correlation and damage identification technologies. The assembled structure consists of the upper section of the fin for the vertical stabilizer, the two upper Rudder/Speed Brakes (RSB), two aluminum actuator mockups, and a steel transition flange (for attachment to ground). The Thermal Protection System (TPS) and all static load application pads had been removed prior to the testing discussed herein. The VSA is identical to the original OV102 (Columbia) vehicle design. Figure 3(a) shows the operational configuration of the orbiter vehicle with the upper rudder marked. The VSA test article was reassembled in the NASA-JSC Vibration and Acoustic Test Facility. Figure 3(b) shows the fully assembled VSA as configured for the tests. During the assembly process, modal testing was performed after the addition of each major subassembly. After full assembly, a series of repairable and nonrepairable damage cases were inflicted on the structure. Therefore an extensive data base was produced to study model correlation and damage identification procedures.

The fully assembled test article was instrumented with 56 Kistler 8632 and 8630 accelerometers and was excited with an MB500 shaker. The standard force input was 100 newtons RMS. Figure 4 provides the accelerometer locations and experimental geometry used in the testing. Frequency Response Functions (FRF's) were acquired from 4 to 300 Hz. This range included approximately 40 structural modes which could be extracted from the data.

Eight damage cases (each of which contained multiple levels of damage) were inflicted on the test article. The first four damage cases were repairable, meaning that the damage was easily reversed or repaired. Damage case #1 involved the addition of three 5 in x 6.5 in access panel cover plates to the rear spar. Figure 5(a) shows the lowest of these plates before installation (Damage Case #1(c)). The RSB's are not attached and the lower actuator mock-up is shown. Figure 5(b) shows the same plate after installation. Figure 6 shows the location of all three plates on the test article. Modal testing was performed before the installation of each plate to provide three levels of damage. The first baseline or undamaged data set was acquired after installation of all three plates.

Each rudder was connected to the actuator mockups using an upper and lower hinge attachment as well as an upper and lower RSB link. Damage Case #2 involved disconnecting each of these links one at a time from the actuators. This sequence provided four additional damage subcases as data were acquired after the removal of each link. Additional tests were performed after reinstallation of each link. Hence, four additional baseline data cases were produced which allowed some characterization of assembly changes. Figure 7 shows one of the RSB link actuator attach points on the left hand side of the lower actuator which connects to the right rudder.

Damage Case #3 involved removing two actuator-to-fin attach bolts one at a time. Each actuator was connected to the fin by six bolts (three on each side). The top left bolt on the upper actuator and the bottom right bolt on the lower actuator were removed in Damage Case #3. This provided two damage cases and two additional baseline cases with potential assembly differences. Figure 8 shows the bolt removed to produce Damage Case #3(a) (top left bolt of upper actuator).

Damage Case #4 involves removing the tip cap access panel, as shown in Fig. 9, in three stages. Figure 10 shows the access panel after complete removal (Damage Case #4(c)). This sequence provided three additional damage cases.

A linearity study was also performed in which three amplitude levels of input were used to assess nonlinearities in the structure. The different input levels were 100, 75, 50, and 25 newtons. This data set provided four additional baseline data sets which captured changes in the dynamic response attributed to nonlinear phenomena. This information is necessary to interpret the damage localization results provided in a later section.

Algorithmic Assessment and Development
Several algorithms developed in this program are discussed below.

Reduced Sensor Measurement Set
A problem shared with all developed approaches is that of the incomplete measurement problem. The incomplete measurement problem has two contributions: (1) experimental measurement of a lesser number of modes of vibration than that of the analytical model and (2) experimental measurement of a lesser number of degrees of freedom than that of the analytical model.

One approach to practically address problem (2) is to either reduce the analytical model to the test degrees of freedom or to expand the measured modal data to all degrees of freedom included in the analytical model.[1] Model reduction raises issues with damage location, as localized changes in the full model may become "smeared" throughout the reduced model. A problem observed with mode shape expansion is that errors introduced in the expansion process lead to false positive indications of damage. In this work, we extend the Minimum Rank Perturbation Theory (MRPT)[2] dynamic residuals and couple it with concepts of stiffness matrix disassembly[3] to arrive at an expanded dynamic residual. The development of an expanded dynamic residual using a Finite Element Model (FEM) of the structure and measured modal data is based on work initially proposed in Reference (4).

FRF Damage Detection Preprocessor
The Shuttle Modal Inspection System (SMIS) operational team faces the problem of extracting modal parameters from measured space shuttle frequency response functions. The objective in identifying modal parameters is to monitor trends in modal parameters changes of key components of the space shuttle. Unfortunately, the measured FRF's are complex, and the effort spent extracting modal parameters allows little time to monitor and assess measured changes. As part of this ISSO project, we have developed a rapid pre-processor for inspecting the measured FRF's directly for structural changes. This pre-processor can be used as a stand-alone damage detection algorithm. In addition, the algorithm can be used to select frequency bands which reveal damage and thus serve to focus the modal extraction efforts to these particular bands. Figure 11 shows the results of this algorithm applied to the shuttle vertical stabilizer assembly. The solid line in the upper figure is a measured frequency response function after the structure has been damaged. The dotted lines represent the 95 percent confidence bounds of the healthy measured FRF.

These bounds were obtained by testing the healthy structure under a variety of environmental and testing situations. The basic way to view these confidence bounds is to understand that the normal variations obtained during testing should all fall within these bounds. Therefore, any "excursion" from these bounds indicates that something (i.e. damage) has happened that cannot be explained by the environmental and/or testing variations. The lower figure shows the level of "excursions" between the healthy confidence bounds and the damaged frequency response function. The large number of excursions shows that some form of damage has occurred. A number of distinct frequency regions can be seen in which the excursions lie. These regions could be the focus of subsequent modal extraction.

Autonomous Dynamics Determination
Another problem faced by the SMIS team and anticipated for the ISS ground support team is the human effort required to extract modal parameters from measured time histories or frequency response functions (Fig. 12). In this work, performed in collaboration with researchers at NASA Langley Research Center, the concept of identifying modal parameters with no human intervention is being investigated. This technique has already been applied successfully to subsets of the VSA testing data,[5] and to on-orbit data obtained from the Russian MIR space station.

Other Accomplishments
Zero-Gravity Testing
Damage detection experiments have been completed on a truss structure in the NASA KC-135 "vomit-comet." This project involved eight undergraduate students (four from UH, four from the University of Kentucky), graduate students, the ISSO Post-Doctoral student, and two faculty members (D. Zimmerman, UH, S. Smith, UK). In addition, McDonnell-Douglas Aerospace in Houston served as the industrial review team. The purpose of the experiment was to identify differences in vibration response and damage detection capability between our 1-g ground validated methodology and 0-g. Figure 13 shows a NASA photograph of Eric Myers, a UH undergraduate student, getting ready to "impact" and catch the truss structure while "floating" in zero-g.

In addition to advancing technology, this experiment has been visible in the public eye. Numerous television segments (both regional and national) and newspaper articles (regional and campus) have reported on these activities. Sound and Vibration, a leading industry magazine, will feature this experiment as the cover story for its November, 1997 issue. Technical papers reporting on the experiment will be read at the International Modal Analysis Conference in February, 1998.

Advanced Sensor Technology
A number of advanced sensor technologies have been demonstrated at NASA-JSC by the ISSO fellow and former colleagues. The main challenge has been to identify near-term sensor technologies capable of measuring alpha joint outboard components of the International Space Station. The following sensors were evaluated for NASA:

References
[1]D. C. Zimmerman, S. W. Smith, H. -M. Kim, and T. Bartkowicz. "An Experimental Study of Structural Damage Detection Using Incomplete Measurements," ASME J. of Vibration and Acoustics 118.4 (1996) 543-50.
[2]D. C. Zimmerman and M. Kaouk. "Structural Damage Detection Using a Minimum Rank Update Theory," ASME J. of Vibration and Acoustics 116.2 (1994): 222-31.
[3]L. D. Peterson, S. W. Doebling, and K. F. Alvin. "Experimental Determination of Local Structural Stiffness by Disassembly of Measured Flexibility Matrices," AIAA Paper #95-1090, AIAA Structures, Structural Dynamics, and Materials Conf., 1995.
[4]G. H. James, D. C. Zimmerman, and T. Cao. "Development of a Coupled Approach for Structural Damage Detection," Proc., 35 Aerospace Sciences Meeting & Exhibit, Reno, NV, AIAA Paper No. 97-0362, 1997.
[5]R. S. Pappa, G. H. James, and D. C. Zimmerman. "Autonomous Modal Identification of the Space Shuttle Tail Rudder," ASME Paper DETC97/VIB-4250, 1997.

Contents
ISSO -- Institute for Space Systems Operations
1996-1997 Annual Report

Navigation Bar

foot-black.gif (4301 bytes)