Par 6 Page 13
4. Maximum once-per-revolution (1P) vibratory load;
5. Climb;
6. Cruise;
7. Descent;
8. Landing;
9. Reverse thrust;
10. Taxi back; and
11. Shut-down.
(2) Data Reliability. The data should be developed as in paragraph 5.k. of this AC.
c. Damage Tolerance Data.
(1) General. Developing the fatigue strength characteristics of composite propeller components for a damage tolerance fatigue evaluation should be done using full-scale test specimens. This establishes the damage growth, strength, failure location, and failure mode and mechanism of the components. The damage tolerance discussion focuses on propeller blades, but may be applied to any composite propeller component. The damage tolerance approach is based on the following factors/assumptions:
(a) Damage is inherent in the structure or inflicted in service and may grow with the repeated application of loads;
(b) The propeller or propeller components will be inspected at intervals to assess the extent of damage; and
(c) When damage reaches the maximum permissible flaw size, the propeller or propeller component will be retired.
(2) Tests to Determine Failure Mechanism. A test should be conducted to determine the failure mechanism, using knowledge of the propeller blade loading spectrum and blade design.
This is the first step in the fatigue verification cycle, to verify the failure mechanism determined in the blade design phase. This also confirms the location of critical defects that will be used in subsequent fatigue tests.
Page 14 Par 6
(a) Test Specimen. The specimen should represent actual type design. This test verifies the critical stress locations and failure mechanism determined during the blade design process.
No artificial defects or repairs should be incorporated. The predicted high stress (failure initiation) point should be far from any naturally occurring flaws.
(b) Loading. The blade should be loaded to accurately simulate the predicted critical loading environment. This may not be the highest steady load condition. For example, a low rotational speed condition with a high vibratory bending load may be more damaging due to increased transverse stress and increased compressive stress. Most composite designs are more sensitive to the LCF part of the spectrum. Therefore, particular attention should be paid to conditions that produce high fatigue loads but not high cycles. Loads induced during these and other conditions can be substantially higher than normally occurring design load conditions such as take-off rotation. To accelerate testing, alternating bending loads may be increased above the actually occurring loads to initiate and propagate damage. Some examples of these conditions are:
1. High power;
2. High yaw;
3. High g loading due to maneuver or gust loading;
4. Ground air ground;
5. Resonant vibration loads occurring during initial run-up or run-down in a crosswind environment; and
6. Thermally induced stresses.
(c) Monitoring Failure Mechanism. The initiation and growth of damage should be monitored using non-destructive inspection (NDI) techniques. The NDI techniques to be used on the type certificated product and defined in the Instructions for Continued Airworthiness should be evaluated during this test. Since the test determines failure mechanism, the test should be run to structural failure. The definition of this failure should be the point at which the stiffness of the specimen has begun to rapidly degrade and, if possible, the component has been brought to total separation.
(2) Test to Determine Flaw Growth. This portion of the testing process determines the rate of flaw growth as a function of loading. The purpose is to determine the load that will propagate a flaw and the rate of flaw growth vs. loading.
Par 6 Page 15
(a) Test Specimens. Test specimens should be manufactured with the allowable manufacturing anomalies and/or permissible repairs, as appropriate, located in the most critical location. The most critical location should be determined in the design analysis and confirmed in the failure mechanism test in paragraph 6.c.(2) of this AC. If more than one critical defect has been identified, then all should be evaluated. If all defects result in the propagation of the same type of flaw, then only the propagating flaw should be analyzed. For example, if voids, resin pockets, and delaminations are possible manufacturing anomalies, and all defects result in the initiation and growth of delamination, then only delamination growth should be analyzed.
(b) Loading. As indicated in paragraph 6.c.(2)(b) of this AC, steady loads should represent normally occurring loads, and alternating loads should be applied to produce flaw growth. If flaw growth cannot be produced within the projected loading conditions, elevated loads should be used to produce flaw growth. Constant amplitude testing should be conducted at several different alternating load conditions.
(c) Monitor Failure Mechanism. Flaw growth should be monitored using NDI techniques. During the test the NDI techniques to be used on the type certificated product and defined in the Instructions for Continued Airworthiness should be evaluated. Since the purpose of this test is to determine flaw growth data, the test should be run until sufficient data has been collected to accurately model flaw growth. During flaw growth testing, the failure mode should be the same as in the failure mechanism test. If there is an inconsistency, the test and the design should be evaluated. If there is a second failure mode present, it should either be addressed through design or be independently tested.
(d) End of Life Condition. The end of life condition is established in conjunction with the service life. The component at the end of life condition meets all airworthiness loading requirements. Therefore, the component in its end of life condition is still safe. The loading requirements should account for resonant conditions when applicable. Resonant conditions may be applicable when the stiffness of the component changes sufficiently to change the natural frequency.
(e) Growth Rate. Figure 4 shows the progression of a composite blade flaw growth.
The flaw in this case is the delamination area as a percentage of the total area defined by the end of life condition for the blade. The flaw growth test involves the application of a steady and alternating load while measuring the flaw growth at an interval sufficient to establish the flaw growth rate. The measured flaw growth rate is generally non-linear, as the internal stresses in the component redistribute themselves as the flaw grows. For the case shown, the growth rate lessens with increased flaw size. The conservative approach to a non-linear flaw growth rate in which the growth rate lessens with increasing flaw size is to assign a linear value to the growth data, as long as the linear value predicts flaw growth to the end of life condition before the actual failure. All components should be treated in a similar manner when assigning flaw rates. For designs that do not show a decreasing growth rate with increasing flaw size, a much more conservative approach should be used when predicting blade life and inspection intervals. The test began with a known defect size and ended prior to the end of life condition defined as 100% of the flaw size. This practice is acceptable as long as the failure mode and its progression are
Page 16 Par 6 known. When the failure mode is unknown or unpredictable this method should not be used.
For this practice to be acceptable, the initial failure test should show ample margin between the point when the assigned flaw rate predicts the end of life condition and when the actual end of life condition occurs. Multiple flaw growth tests are needed to establish the flaw growth rate as a function of load.
(f) Flaw Growth Curve. The flaw growth rate results from multiple flaw growth tests are used to develop the flaw growth curve shown in Figure 5. The flaw growth rate curve is established for data collected at uniform steady load test conditions. The flaw growth rate curve is then used to compute the flaw growth for each propeller load cycle using a flaw growth accumulation algorithm. The scales shown in Figure 5 are above 100% because flaw growth testing was conducted at loads above the limit load.
(g) Spectrum Load Testing. Spectrum load testing verifies the flaw growth accumulation algorithm used for subsequent flaw growth evaluations. A load spectrum includes operating conditions representative of the intended operation for the propeller. The load spectrum may include a typical flight consisting of no load, increase to take-off thrust and maximum 1P vibratory load, reduce to a climb conditions, reduce to cruise conditions, increase to maximum reverse conditions and return to no load. Within this spectrum a high amplitude low cycle limit maneuver load should be periodically applied.
(h) Inspection Intervals. Paragraph 9.b.(3) of this AC discusses the development of inspection intervals.
7. METALLIC COMPONENT FATIGUE LIMITS.
a. Discussion. The fatigue limits for metallic components are determined by developing fatigue data through coupon and full-scale testing, previous testing, past experience and acceptable published data. This data should be used to develop S-N diagrams and Goodman diagrams suitable for a safe-life propeller fatigue evaluation and da/dn diagrams for damage tolerance evaluation of the applicable propeller components.
b. Safe-Life Fatigue Limits. Developing fatigue strength characteristics for metallic components follows the two phase approach presented in paragraph 6.b. of this AC. The first phase is coupon testing for both static properties and fatigue properties to supplement the fatigue (S-N) curve shape, standard deviation, and statistical distribution. The second phase is full-scale specimen testing to establish the strength level, statistical distribution, failure location, and failure mode and mechanism for the material system and geometry of the propeller. The development of the fatigue properties may be supported by previous testing, past experience and acceptable published data, when available.
Par 7 Page 17
(1) Loading. Steady loads should represent normally occurring loads; alternating loads should be applied to produce the desired failure results. Testing may be conducted in steps with constant amplitude alternating loads. If the alternating load level is below the endurance limit, failure will not occur within the number of cycles considered acceptable to verify the endurance limit. This is considered a run-out condition. The alternating load should be increased and the test should be resumed at ever increasing load levels until failure occurs. LCF should also be addressed. Due to the shape of S-N diagrams for metallics, the initial emphasis of full-scale testing of metallic components is usually HCF. Coupon tests have limited value in areas with complex geometry.
(2) Data Reliability. The data should be developed as in paragraph 5.k. of this AC.
c. Damage Tolerance Data. A damage tolerance evaluation of metallic components usually uses classical fracture mechanics techniques. Alternate methodologies include a safe-life approach coupled with experimental assessment of the effect of damage, using coupons or fullscale components. Equating damage to a crack may be extremely conservative for some materials; an alternate method for quantifying the effect of the damage on life should be used in those cases. This AC assumes that when damage tolerance methods are applied the component has been designed using damage tolerance and fail-safe principles.
(1) Tests to Determine Failure Mechanism. The full-scale tests should be used to define the fracture location for an undamaged component. When including the effects of damage, it should be assumed that the damage occurs at the fracture site.
(2) Tests to Determine Crack Growth. Crack growth data for the material of interest may be obtained from acceptable published data. If no data is available, the data should be obtained by experiment, using appropriate coupon configurations and taking into account the effects of stress ratio (R).
(a) Crack Growth Curve. The experimental data should be used to produce a crack growth curve in the form of da/dn (amount of crack extension per cycle) versus ∆K (stress intensity range) as a function of stress ratio (R). Figure 6 illustrates such a curve. The curve shape is sigmoidal when presented on a log-log plot. In the mid-region the growth is linear with a slope of “m.” In the initial region, called “near threshold,” the plot exhibits curvature and becomes asymptotic to the threshold stress intensity range (∆Kth). In the final region, the plot exhibits opposite curvature and becomes asymptotic to critical stress intensity range (Kic).
(b) Spectrum Load Considerations. Commercial and government computer programs, such as NASA/FLAGROW, NASGRO, and FASTRAN, contain analytical crack growth algorithms that sufficiently account for growth under load spectra anticipated in service for metals. These techniques may be used for performing crack growth calculations from the initial detectable crack to the critical crack length. The loading should include all of the segments of the spectrum unless an analytical damage assessment shows that specific segments produce
Page 18 Par 7 negligible damage. Consideration should be given to material statistical scatter. Generally this is accomplished using typical growth data and applying scatter factors on inspection intervals.
(c) Inspection Intervals. Paragraph 9.b.(3) of this AC discusses the development of inspection intervals.
8. METALLIC HUB AND RETENTION FATIGUE TESTS.
a. General. The complex nature of the hub and retention structure with bolted joints and bearings makes it difficult, if not impossible, to measure critical stress locations. Therefore, the fatigue test should be conducted on full-scale components, in a test rig that simulates the orientation of operating loads, so that applied test loads result in the stresses that match or exceed operating stresses in critical areas. Hub and retention testing develops S-N curve fatigue data or shows that the hub has a sufficient life through run-out testing at an appropriate load or spectrum loading. Run-out testing should be supported by previous testing, past experience, and acceptable published data, when available.
b. Loading. The centrifugal load, thrust, torque, cyclic aerodynamic loads and engine cyclic loads for reciprocating engines should be applied simultaneously during fatigue testing. The test rig in Figure 7 is one type has been shown to be suitable for testing with simultaneously applied loads for high power installations (1500 shp or greater) with high 1P loading. This type of test rig assesses the interface between the hub and propeller shaft flange. The hub is mounted to the actual propeller flange and shaft unit, which has been adapted to the test rig. The blade loads are input by means of actuators that are aligned in the direction of the applicable steady and vibratory blade loads. Centrifugal loads are also applied to load the blade retention. Since the hub loading is complex, sound judgement is required to approximate the complete loading pattern for the hub. The test rig in Figure 7 approximates that the blade vibratory and steady loads are aligned in the same direction. The suitability of any test rig should be determined before conducting the fatigue testing.
9. FATIGUE EVALUATION. Section 35.37 requires a fatigue evaluation. The fatigue evaluation should be conducted on a typical or the intended airplane. The intended airplane may be the airplane used during airplane certification to conduct the vibration tests and evaluation used to show compliance with §§23.907 or 25.907. The typical airplane may be the airplane used to develop design criteria for the propeller or another appropriate airplane.
a. Airworthiness Limitations. The propeller fatigue evaluation establishes the mandatory replacement times (life limits) for the propeller and, in some cases, mandatory inspections.
Compliance results for a typical airplane are preliminary, as propeller vibratory stress on the intended airplane has not been measured. The evaluation is a design evaluation on a typical airplane. The evaluation provides an acceptable level of assurance, before installation on the intended airplane, that the propeller will be structurally acceptable for use on that airplane.
When the fatigue evaluation is conducted on the intended airplane, the evaluation may be directly applicable to §§23.907 or 25.907. In either case the airworthiness limitations must be identified and appropriately documented in the Airworthiness Limitations Section of the
Par 9 Page 19 propeller Instructions for Continued Airworthiness. The Instructions for Continued Airworthiness should note that the applicability of the Airworthiness Limitations should be reevaluated for each new airplane installation.
b. Approaches. There are a number of different approaches to fatigue evaluation. This AC presents fatigue limits that support safe-life and damage tolerance evaluations. AC 20-66A provides guidance on conducting the fatigue evaluation. The following concepts should be considered when conducting the fatigue evaluation:
(1) Unlimited Life. When it is shown that all stresses are below the endurance limits established for the component, the component is said to have unlimited life. Components with unlimited life will be removed from service for reasons other than fatigue. In addition, when the safe-life of a component is shown to be greater than 70,000 hours, and it is shown that the component will be safely retired from service for reasons other than fatigue before its safe-life, the component may be said to have unlimited life.
(2) Safe-Life. Safe-life is the component fatigue life reduced by an appropriate scatter factor that accounts for the variability of the fatigue evaluation process. The fatigue life is determined by combining the airplane loading spectrum with the fatigue data using a damage summation algorithm (safe-life evaluation). Mandatory replacement times are established for parts with safe-lives. Unless substantially justified the following scatter factors should be applied:
(a) Metallic scatter. A scatter factor of three or greater should be used for metallic components.
(b) Composite scatter. A scatter factor of 10 or greater should be used for composite components.
(3) Damage Tolerance Inspection Interval. For damage tolerance methods the inspection interval is determined by the relationship between the time the damage reaches maximum permissible flaw size as defined during certification (detectable damage) and the end of life condition. The maximum permissible flaw size is established during certification by considering the inspection method, the inspection interval, and the end of life condition. The inspection interval permits multiple opportunities, usually three, to find the damage before the component reaches the end of life condition. The inspection method should also be evaluated to determine the probability of detection (POD). Inspection methods should have a POD of 90% probability with 90% confidence. When the POD is less than 90% probability with 90% confidence, the inspection frequency should be increased. The component should be removed from service when damage is detected at the maximum permissible flaw size. The Airworthiness Limitations Section of the Instructions for Continued Airworthiness establishes these inspections as mandatory.
Page 20 Par 9
(4) Limit Load Fatigue Test. The propeller blade should be able to withstand limit loads without detrimental permanent deformation or deformation that would result in a hazardous propeller effect. This should be demonstrated by a fatigue test, since the propeller is a rotating device. Appropriate life factors should be applied during the test. The test should be performed on a blade in end of life condition.
Original signed by JJP on 9/17/01 Jay J. Pardee Manager, Engine and Propeller Directorate, Aircraft Certification Service
FIGURE 1
LOAD AND LIFE FACTORS CYCLES (Log Scale) LOAD (Linear Scale) Life Factor Load Factor Typical S-N Diagram FIGURE 2 II Cycles I III Damage Residual Strength Load Capacity Applied Load
1.0
COMPOSITE BLADE DAMAGE DEVELOPMENT
FIGURE 3
• Peak str ess n ear the blade root
• R ratio adjusted for flight condition simulated
• Forced response test Hub Fixture Steady Bending Load Vibratory Load Hydrauli c Shaker Base
MEAN STRESS TEST Steady Ce ntr if ugal Load Bl ade Hub Fixture Electromechanical Shaker
ROOT ZERO MEAN STRESS TEST Vibratory tip motion
• Peak stress n ear the blade root
• R ratio equal zero, does not match flight loads for blade root
• Reson ant test First flatwise mode Bl ade Hub Fixture Electromechanical Shaker
TIP ZERO MEAN STRESS TEST Vibratory tip motion Second flatwise mode
• Peak stress n ear t he blade tip
• R ratio equal zero
• Reson ant test
EXAMPLES OF BLADE TEST RIGS
FIGURE 4
FLAW GROWTH DATA 0 10 20 30 40 50 60 70 80 90 100 0 5,000,000 10,000,000 15,000,000 20,000,000 Cycles Flaw Area (% of maximum) Test performed at constant amplitude Assigned growth rate FIGURE 5
COMPOSITE FLAW GROWTH RATE 1 10 100 1000 1 10 100 1000 Alternating Moment - % of limit load Growth Rate - % of growth rate at limit load
FIGURE 6 Typical Crack Growth Curve Shape
1.00E-09
1.00E-08
1.00E-07
1.00E-06
1.00E-05
1.00E-04
1.00E-03
1.00E-02 1 10 100 1000 ∆∆∆∆K da/dN Kc or final failure Threshold ∆∆∆∆Kth Linear portion with slope "m"