Technology Benefit Estimator Designed for Aerospace Propulsion Systems
Tailoring Code Enhanced To Predict Isothermal Fatigue Life of MMC's
Mechanical Loads Significantly Change Structural Damping and Natural Frequencies of CMC's
Optimality Criteria Method Provides Optimum Design for Select Structural Problems
Active and Sensory Responses Simulated for Smart Composite Structures
Efficiency of Elastoplastic Analysis Improved
New Higher-Order Theory Analyzes Functionally Graded Materials
TMF Damage Progression Characterized in Titanium-Matrix Composite
Isothermal Axial-Torsional Fatigue Data Bases Generated for Cobalt-Based Superalloy
APPLE Incorporates All Aeroelastic Analyses for Turbomachines and Propfans
Leakage Model Developed for Hypersonic Engine Seals
Active Control of Rotordynamic Vibrations Achieved
Reaction-Compensating Platform Preserves Microgravity Environment
Unstalled Flutter of Counterrotating Propfan Experimentally Investigated
Integrated Design Software Predicts Durability of Monolithic Ceramic Components
Postscan Interactive Data Display System Developed for Ultrasonic Scans
The T/BEST software is specifically designed for estimating the benefits and benefit sensitivities of introducing new technologies into existing propulsion systems. Key engine cycle, structural, fluid, cost, noise, and emissions analyses modules are used as a framework for interfacing with advanced technologies. The open-ended modular approach allows for modifying and adding both key and advanced-technology modules. T/BEST's hierarchical framework yields varying levels of benefit estimation accuracy, depending on the degree of input detail available. This hierarchical feature permits rapid estimation of technology benefits even when the technology is at the conceptual stage. As knowledge of the technological details increases, so does the accuracy of the benefit analysis obtained.
Included in T/BEST's framework are correlations developed from a statistical data base, which are relied on if insufficient information is given in a particular area (e.g., fuel capacity or aircraft landing weight). Statistical predictions are not required if these data have been specified in the mission requirements. The engine cycle, structural, fluid, cost, noise, and emissions analyses modules interact with data libraries to yield estimates of specific global benefits: range, speed, thrust, capacity, component life, noise, emissions, specific fuel consumption, component and engine weights, precertification test, engine cost, direct operating cost, life-cycle cost, manufacturing cost, development cost, risk, and development time.
At present T/BEST operates on stand-alone or networked workstations and uses a Unix shell or script to control the operation of interfaced Fortran-based analyses. T/BEST's interface structure works equally well with non-Fortran or mixed software analyses. It is designed to maintain the integrity of the analyses modules by interfacing with experts' existing input and output files. Parameter input and output data (e.g., number of blades and hub diameters) are passed through T/BEST's neutral file while copious data (e.g., finite element models and profiles) are passed through file pointers that point to the experts' output files. To make the communications between T/BEST's neutral file and the attached analyses modules simple, only two software commands, PUT and GET, are required. This simplicity permits easy access to all input and output variables contained in the neutral file. Both public domain and proprietary analyses modules may be attached with minimal effort while maintaining full data and analysis integrity and security.
Lewis contact: Dr. Edward R. Generazio, (216) 433-6018
Headquarters program office: OA
This direct introduction of life prediction capabilities into the synthesis cycle allows for the concurrent synthesis of both process and material and ensures explicit improvements in fatigue performance. This technique automatically identifies which residual stresses are important for the composite's critical failure modes. It also optimally controls their evolution irrespective of the applied loads and thermomechanical fatigue cycle.
The method was evaluated on a ceramic silicon carbide fiber (SCS-6)/titanium aluminide (Ti-24Al-11Nb) matrix composite. This intermetallic matrix composite system was selected for its significance as a candidate high-temperature material and the availability of experimental data on its fatigue performance. Excellent isothermal life predictions were obtained at both room and elevated temperatures (top graph). The robustness of the method under low-, intermediate- and high-cycle fatigue conditions was validated. In addition, optimal processing temperatures and consolidation pressures were predicted that maximize the isothermal life at room and elevated temperatures (bottom graph).
Bibliography
Lewis contacts: Dimitris A. Saravanos, (216) 433-8466;
Dr. Christos C. Chamis, (2l6) 433-3252
Headquarters program office: OA
The stress-strain curves showed nonlinearities in the loading portion, significant hysteresis, and a residual strain after loading, indicating that damage had occurred in the specimens within the first load cycle (top graph). There was a significant increase in damping after the first load cycle and further increases after 10 and 100 cycles (bottom graph)--with a corresponding decrease in the natural frequency. Interspecimen variation in damping and natural frequency was consistent; the sample that had the largest increase in damping showed the largest decrease in natural frequency. Results were very sensitive to support conditions but insensitive to atmospheric pressure. The results indicate that both damping and natural frequency are sensitive to damage and may be useful techniques for monitoring damage progression in CMC structures.
Lewis contact: Dr. J. Michael Pereira, (2l6) 433-6738
Headquarters program office: OA
Optimality criteria can be a useful tool to design, or modify an existing design of, a structure for displacement or frequency constraints.
Lewis contacts: Dale A. Hopkins, (216) 433-3332;
Dr. ;Surya N. Patnaik, (216) 433-8468;
Dr. Laszlo Berke, (216) 433-5648
Headquarters program office: OA
The static and dynamic responses of smart composite structures with embedded sensors and actuators were also obtained, and specialty finite elements were developed for this purpose. Evaluations on composite beams and plates have demonstrated that the mechanics can represent either sensory or active structures and can model the complicated stress-strain fields, including the interactions between passive and active layers (see graph), interfacial phenomena between sensors and composite plies, and critical damage modes in the material. The capability to predict dynamic characteristics under various electric circuit configurations was also demonstrated.
Lewis contacts: Dimitris A. Saravanos, (2l6) 433-8466;
Dale A. Hopkins, (216) 433-3332
Headquarters program office: OA
In addition to its many beneficial computational characteristics, MLS also provides a logical and efficient means by which to isolate and localize the numerical treatment of nonlinear behavior. Those areas where the nonlinearity is occurring can be confined within the appropriate substructures. Consequently, only those substructures affected by the nonlinearity need be recalculated after the first iteration of the solution process. Computational effort for each iteration is substantially reduced because the linear portions of the model remain unchanged and do not have to be "re-solved." For example, a three-level MLS model required only 110 wallclock seconds to perform each iteration of the Newton-Raphson solver, as opposed to 435 wallclock seconds for the model without substructuring on a single processor. Moreover, because only two substructures behaved nonlinearly, only two processors were needed to perform the nonlinear iterations in parallel--reducing the wallclock time to only 60 sec per iteration. The remaining processors on the network could be used for other tasks. Or, as with a clustered workstation network, a nested hierarchy of parallelism could be used to perform the calculation. That is, each nonlinear substructure would be solved concurrently with a parallel equation solver of its own.
It is anticipated that the MLS solution strategy will ultimately be used to solve nonlinear FE models comprising thousands of degrees of freedom. Furthermore, by using MLS's inherent ability to separate, isolate, and localize the various constituents or components of an analysis, fluid and solid formulations could be combined to achieve more realistic and relevant simulations. These types of parallel computer simulations will be able to provide an extremely high level of accuracy in a relatively short turnaround time.
Lewis contact: Dr. Shantaram S. Pai, (216) 433-3255
Headquarters program office: OA
The potential benefits of composites with tailored microstructures have led to increased activities in processing and materials sciences. Handicapping these activities, however, is the lack of appropriate computational strategies for the response of functionally graded materials that explicitly couple the material's heterogeneous microstructure with global analysis. Such coupling may be required in composites containing a relatively small number of large-diameter inclusions (e.g., fibers) along the dimension subjected to a thermal gradient (e.g., silicon carbide/titanium composites).
A new analytical approach has been developed at NASA Lewis to analyze the behavior of composites with tailored microstructures characterized by large-diameter reinforcement and thermal gradients. In this approach the microstructural and macrostructural details are explicitly coupled when solving the thermomechanical boundary-value problem. In contrast, standard micromechanical schemes based on classical homogenization procedures treat the local (micromechanics) and global (macromechanics) problems separately. Coupling the local and global analyses allows analysis of the response of functionally graded composites with continuously changing properties due to nonuniform fiber spacing or the presence of several phases. We have used this new "higher-order theory for functionally graded materials" (HOTFGM) to investigate how internal temperature and stress distributions affect an applied temperature differential across the thickness of a composite plate with different fiber spacings in the thickness direction and to investigate the force and moment resultants. Grading the composite microstructure appears to reduce the temperature distribution and thus yield more favorable stress distributions. This, in turn, reduces the in-plane force and moment resultants, maintaining the composite flat during a thermal gradient.
The manner in which the composite microstructure is graded must consider the sign of the thermal gradient. Future work will incorporate inelastic effects as well as temperature-dependent response of the constituent phases exhibited by advanced metal- matrix composites at elevated temperatures. The full potential of HOTFGM as a design tool will be realized when it is combined with an optimization approach.
Lewis contact: Dr. Steven M. Arnold, (2l6) 433-3334
Headquarters program office: OA
If TMC damage modeling is to be entirely successful under general TMF loading conditions, the fundamental mechanisms discerned from the various experimental studies must be accurately represented. Such studies often detail the damage mechanisms in qualitative fashion through fractographic and metallographic observations on specimens cycled to failure. Although this information is generally useful, it lack details about the actual progression of damage. To supplement this information, recent advances in TMF testing were introduced to quantify damage progression at the macroscopic level. Now, explicit measurements tracked isothermal stiffness (E) and CTE degradations as functions of accumulated TMF cycles. These quantitative data, combined with microstructural observations, assisted in further characterizing the damage progression. However, details of the progression of physical damage within the microstructure and its relative impact on the macroscopic mechanical properties remained unknown. The experimental investigation highlighted here was undertaken to provide insight to this matter.
The TMF damage progression in SCS-6/Timetal 21S [0]4 was thoroughly characterized at the macroscopic level by monitoring the E and CTE degradations to failure. A second set of specimens was cycled under identical conditions to 10, 50, and 90% of the cyclic life Nf and then subjected to destructive metallography to relate the extent of physical damage in the microstructure to the mechanical property degradation. Finally, a third set of specimens was cycled in identical fashion and tensile tested at room temperature to quantify residual strength at various stages of TMF damage progression, as shown in the graph. Damage began very early in cyclic life ( N < 0.1 Nf ) for both in-phase and out-of-phase TMF loadings. In-phase TMF damage was dominated by fiber breakage in the absence of matrix damage. The progression of microstructural damage was difficult to quantify because fiber failure is based on probability. Out-of-phase TMF loadings produced matrix cracks that began exclusively at the surface. Here, damage progression was clear, both the number of cracks and their inward spread toward the outer fiber rows with increased cycling. The specimen interrupted at 50% Nf revealed that major cracks were near, but had not yet reached, the outer fiber rows. The stage at which this event occurred (when localized fiber/matrix debonding and matrix crack bridging occurred) appeared to be reflected in the macroscopic property degradation curves and is estimated at approximately 60% Nf .
Lewis contact: Michael G. Castelli, (2l6) 433-3334
Headquarters program office: OA
The cobalt-based superalloy Haynes188 is used for the combustion liner in the T800 turboshaft engine for the RAH-66 Comanche helicopter and for the liquid oxygen posts in the main injector of the space shuttle main engine. Both components are subjected to multiaxial stresses resulting from thermal gradients and mechanical loading. NASA Lewis has generated isothermal, axial-torsional, fatigue crack initiation data bases on Haynes188 at 316 and 760 deg.C. The elevated temperature 760 deg.C (minimum ductility temperature for Haynes188) was selected because low-cycle fatigue life is governed by ductility and because a lower bound on fatigue life is usually required for design. The intermediate temperature 316 deg.C was selected because it is the subcreep regime for Haynes 188. Four types of computer-controlled fatigue tests (axial, torsional, and in- and out-of-phase axial-torsional) were conducted on uniform-gage-section, tubular specimens of Haynes188 at each temperature. For comparing fatigue data from all the tests, the von Mises equivalent strain range, which reduces a given set of multiaxial strain ranges to an equivalent uniaxial strain range, was computed for each torsional and axial-torsional fatigue test. Fatigue data are summarized in the graph for the tests conducted on Haynes 188 at 316 and 760 deg.C. Fatigue life relations, computed from the axial fatigue data at the two isothermal temperatures, are also shown to clearly indicate the differences and similarities in the axial, torsional, and axial-torsional fatigue data of Haynes 188.
The isothermal, axial-torsional fatigue data bases were used to verify the applicability of room-temperature multiaxial fatigue life prediction models to intermediate- and elevated-temperature axial-torsional fatigue data on Haynes 188. These isothermal data bases were also employed to devise a thermomechanical, axial-torsional fatigue testing program on Haynes 188 with 316 and 760 deg.C as the minimum and maximum temperatures.
Lewis contact: Dr. Sreeramesh Kalluri, (2l6) 433-6727
Headquarters program office: OA
Flutter and forced response are aeroelastic phenomena involving the interaction of unsteady fluid forces and a flexible structure. In turbomachinery aeroelastic analysis the equations representing the structural dynamic and unsteady aerodynamic behavior of a fluid and blade system are combined to obtain an aeroelastic model. The resulting equations can be solved by either a frequency-domain or a time-domain solution for the system stability and forced vibration. Using a frequency-domain aeroelastic analysis, one assumes sinusoidal motion for the structure and calculates the unsteady aerodynamic forces at a fixed interblade phase angle and operating condition. The governing equations in this case lead to an eigenvalue problem; the eigenvalues determine the stability of the system. In a time-domain aeroelastic analysis the equations of motion for both the fluid and the structure are integrated simultaneously in time to obtain the response to a perturbation from a steady operating condition. An unstable response indicates a flutter condition. The time-domain method does not involve the assumption of linearity required by the frequency-domain method. Thus, it allows the modeling of systems that contain structural and/or aerodynamic nonlinearities.
For several years NASA Lewis has been developing aeroelastic analyses for turbomachines and propfans. This work has resulted in individual codes with different aerodynamic and structural models. However, a single aeroelastic analysis system consolidating all computer codes did not exist. The availability of numerous computers--from desktop workstations to supercomputers, the development of graphical user interfaces, and the development of networking and concurrent engineering methods have paved the way to change this. All the existing aeroelastic analyses at Lewis are being incorporated on a single platform with a common input and output data base. Some user options of the system, named APPLE (Aeroelasticity Program for Propulsion at Lewis), are shown in the figure. Preliminary implementation and testing has begun on NeXT workstations.
Lewis contact: Oral Mehmed, (2l6) 433-6036
Headquarters program office: OA
Key elements of a leakage flow model for the braided rope seal have been determined. Flow models help designers to predict performance-robbing parasitic losses past the seals and to estimate purge coolant flow rates. The leakage model is based on Kozeny-Carmen relations for flow through porous media. The model treats leakage flow through and around the braided seal structure as a system of flow resistances analogous to a series of resistors in an electrical network. These elemental resistances are combined in accordance with their electrical analogs to form an overall effective seal resistance that characterizes the seal. The current work extends previous flow models to predict seal leakage flow as a function of preload and engine pressures. A specially developed seal rig was used to collect the required leakage flow rates for validating the seal leakage flow models as a function of simulated engine pressures and preloads. The seals tested were braided of commercially available alumina-silica fibers. Seals were constructed by overbraiding a dense uniaxial core with several two-dimensional braided sheath layers having a high braid angle.
The logarithmic form of the semiempiric resistance preload model characterizes the observed variation of seal leakage resistance with increasing preload and engine pressures by using a two-term correlation. As shown in the graph the correlation between the measured and predicted leakage rates was quite good. The predicted leakage correlated well for both preloads over the engine pressure range tested.
This rope seal technology was developed by NASA Lewis and Drexel University. NASA/military program changes have curtailed further development.
Lewis contact: Dr. Bruce M. Steinetz, (2l6) 433-3302
Headquarters program office: OA
A vibration control package containing sensors, actuators, controllers, and power amplifiers is adapted to the machine. Research continues to be conducted on each component of the vibration control package, as well as on applying the package to the machine. The following four research projects were successfully completed within the past year.
Computer simulation of the completed research package has been developed and will be used as a design tool. Within the computer simulation the actuators must be modeled. In the past actuators have been treated as ideal devices. This assumption neglects phase and amplitude changes at high frequencies and may lead to coupled control system/structural system instability that limits the amount of active stiff-ness or active damping which can be achieved. Research has been completed for simulating the coupled "electrical/mechanical" system to predict rotordynamic stability and unbalance response along with control system stability.
A hybrid controller has been successfully tested that uses piezoelectric actuators to control vibrations of a flexible rotor. The controller includes active analog components and a hybrid interface with a digital computer. The computer uses a grid search algorithm to select feedback gains that minimize a vibration norm at a specific operating speed. These gains are then downloaded as active stiffnesses and dampings with a linear fit through the operating speed range to obtain very effective vibration control.
A robust controller has also been designed for active vibration control. This controller is based on a novel frequency-domain technique rooted in quantitative feedback theory. The advantages are a plant-based design according to performance specifications and the ability to include structured uncertainties in the critical plant parameters, such as passive bearing stiffness or damping. Simulation studies have shown the effectiveness of this technique in attenuating vibration.
Research has also been completed in incorporating flexible casings into the computer simulation of active rotor vibration control. The trend toward lighter and more flexible designs in rotating machinery brings with it increasing demands for ways to dissipate the excess energy transferred to such structures by the action of dynamic forces. This work incorporates active feedback control mechanisms to suppress this unwanted vibration.
Lewis contact: Albert F. Kascak, (2l6) 433-6024
Headquarters program office: OA
At NASA Lewis we designed a three-degree-of-freedom reaction-compensating platform as a test bed for active attenuation of reaction forces caused by moving objects in a microgravity environment. Unique "linear motors" convert electrical current directly into rectilinear force. These reaction forces are used to counteract disturbance forces introduced to the platform.
The platform system consists of a passive spring-mass damper with added active components and sensors. The passive system attenuates forces at frequencies greater than the resonance and passes forces at frequencies below the resonance. Because the passive system provides at least -20-dB disturbance attenuation for frequencies above 88 rad/sec, the active system design should be most concerned with disturbance rejection below that frequency. The resonant frequency could be lowered by decreasing the spring constant, at the expense of larger platform excursion, or by increasing the system mass, which may not be desirable in spacecraft applications.
The displacers of the linear motors are constrained to vertical motions with respect to the platform and can thus react to vertical disturbance forces (along the z axis) and moments (about the x and y axes). The motors are each capable of 712-N maximum force. Below 4.8 Hz the force available is limited by the position constraint; above that frequency the position amplitude is limited by the maximum force constraint. Therefore, it is safe to attempt control at high frequencies, but commanding a large-amplitude control signal at low frequencies may be unsafe or ineffective.
Without accurate modeling of motor and composite system behavior, high-performance control is not possible. In particular, information on the force constant, mass, friction, maximum force and velocity, and bandwidth of each motor is needed before any active compensation using the motors can be attempted. Although the motors' electrical and mechanical characteristics are very like those of three-phase rotary motors, the mechanical stops prevent the use of rotary motor characterization techniques. Instead, techniques like those used in robotics were used to prevent motor damage.
The control consists of three discrete parts: the force feedforward controller, which directly responds to incoming forces read from the force sensors; the acceleration feedback controller, which responds to accelerations of the platform mass; and the motor position controller, which attracts the motors to equilibrium position, provides software damping for the motors, and also acts as a primary safety system.
The anticipated force disturbance rejection for the combined system is at least -20-dB attenuation for frequencies greater than 55 rad/sec, extending the lower bandwidth by 33 rad/sec below that of the passive system alone without increasing platform mass or decreasing spring stiffness.
Lewis contacts: Dr. Charles Lawrence, (2l6) 433-6048; Dr. Benjamin B.
Choi, (2l6) 433-6040
Headquarters program office: OA
A fluttering blade row (called F21) was tested alone and with a stable aft row (called A21). The blades were research models for the General Electric Unducted Fan engine. The main objectives were to determine how the second blade row affected the stability of the fluttering row and to investigate the flutter. Flutter boundaries over the full operating range of the rotor, including transonic tip Mach numbers, were mapped. A nonintrusive optical system, used to measure blade vibrations at flutter, provided complete blade-to-blade phase information and an indication of the blade mode shapes at flutter. Conventional blade-mounted strain gages were used to measure blade frequency.
We found that at most conditions the second (aft) row had a small stabilizing effect that increased when the power of the aft row was increased. This result can be seen in the graph, which shows the measured stability boundaries in terms of the basic operating parameters. The solid symbols above the dashed boundaries are conditions where the aft row was powered, with increasing power moving toward the top of the graph. It can be concluded from the experiment that flutter analyses developed for single-row propfans should give conservative results when used for counterrotating rows similar to the ones tested.
We also found two distinct flutter modes within the rotor's operating regime, both apparently single-degree-of-freedom instabilities associated respectively with the first and second natural blade modes. Another finding was that the flutter physics changed with relative tip Mach number. At lower relative tip Mach numbers the flutter depended on both angle of attack into the blades and relative tip Mach number, and the dependence differed with the amount of twist-bending coupling in the flutter mode. However, when a relative tip Mach number of 1.10 was reached, it became the limiting parameter.
Lewis contact: Oral Mehmed, (2l6) 433-6036
Headquarters program office: OA
Designing ceramic components requires specialized knowledge of statistics and fracture mechanics. Design accommodations for any given component shape and service environment require extensive numerical computational capabilities. Multidisciplinary research in fracture analysis, probabilistic modeling, model validation, and brittle structure design has been combined in CARES/LIFE and its accompanying documentation. Compiling the diverse elements of this technology into one package has made a comprehensive design tool readily available. CARES/LIFE enables the design engineer to assess the risk of fracture from various competing failure mechanisms and to determine or estimate overall failure probability for a component over its lifetime. Consequently, appropriate design changes can be made until an acceptable failure probability is predicted.
Through its use in an extraordinarily wide array of product designs, ranging from aerospace components to medical implants, the CARES/LIFE computer program has demonstrated a tangible public benefit. CARES/LIFE is the only public domain computer program of its kind available in the United States today. This technology applies to advanced ceramic materials, including silicon nitride and silicon carbide, and other brittle materials, such as glass and graphite. Many commercial products--turbocharger rotors, rocker arm pads and cam followers, poppet valves, radiant heater tubes, heat exchangers, and prototype ceramic turbines--are designed using this software. In addition, CARES/LIFE is used to design large infrared transmission windows, glass panels for skyscrapers, ceramic packaging for microprocessors, cathode ray tubes, and even ceramic tooth crowns and kneecaps.
Lewis contact: Noel N. Nemeth, (2l6) 433-32l5
Headquarters program office: OA
In ultrasonic contact scanning two front-surface and two back-surface ultrasonic pulses obtained with the pulse-echo configuration are digitized and stored at every scan location. These pulses are later Fourier transformed to the frequency domain and used in calculating ultrasonic reflection coefficient, attenuation coefficient, cross-correlation (pulse) velocity, and phase velocity. Images of these ultrasonic properties are then formed at preselected frequencies by using spectral analysis software.
This ultrasonic method is especially sensitive for quantifying global variations (such as pore fraction variations) in microstructure as well as for detecting isolated material defects in monolithic and composite materials. Microstructural variations are indicated by gray or color-scale variations in an ultrasonic image; at such locations there can be questions as to the validity of the indication.
PSIDD allows the operator to examine the digitized ultrasonic waveforms at scan locations and to verify whether the waveforms and the resulting ultrasonic properties are valid, and if valid, how the waveforms and properties differ from those at other scan locations. Additionally, by waveform analysis, PSIDD allows automatic detection and visualization of scan locations considered "good" that may in fact show distorted waveforms. Thus, it is a sensitive quality control device.
Digitized and calculated property data obtained from an ultrasonic contact scan and stored in files are retrieved through a direct-access data retrieval algorithm that allows display of data at any point. Waveforms are autoscaled in both the horizontal and vertical directions and can be individually enlarged to take up the entire video display for more detailed viewing. The user can choose to view the two back-surface waveforms with or without the ultrasonic system noise subtracted. The user has the choice of linear or natural cubic spline interpolation of spectra or property data. This form of analysis is likely to yield more accurate predictions of material behavior, and it can be the basis for artificial intelligence techniques that allow defect identification based on waveform shapes and property versus frequency behavior. This type of comprehensive analysis in such a convenient format is not commonly done in conventional ultrasonic testing.
Lewis contact: Dr. Don J. Roth, (2l6) 433-60l7
Headquarters program office: OA
Last updated 1995
Responsible NASA Official:
Gynelle.C.Steele@nasa.gov
216-433-8258
Point of contact for NASA Glenn's Research & Technology reports:
Cynthia.L.Dreibelbis@nasa.gov
216-433-2912
SGT, Inc.
Web page curator:
Nancy.L.Obryan@nasa.gov
216-433-5793
Wyle Information Systems, LLC