(2) A fail-safe strength evaluation. See § 23.571(b) for pressurized cabin and § 23.572(a)(2) for wing, empennage, and associated structure.
(3) A damage tolerance evaluation. See § 23.573(b) and § 23.571(c) for pressurized cabin and § 23.572(a)(3) for wing, empennage, and associated structure.
b. For wings, empennage, and associated structure, you may be able to show compliance to the fatigue requirements by comparing your design to an existing design. This method of substantiation includes showing that the structure, operating stress level, materials, stress concentrations, and expected uses are equivalent, from a fatigue standpoint, to a similar design that has had extensive satisfactory service experience. See § 23.572(a).
c. For metallic commuter category airplanes, you must use a damage tolerance evaluation. If you can show that damage tolerance is impractical for a particular structure, you may use a fatigue strength, or safe-life evaluation. See § 23.574.
d. For all categories of small airplanes constructed with composite materials, you must use a damage tolerance evaluation. AC 20-107A provides guidance for composite structures.
e. The following flowchart may help clarify which method of fatigue evaluation is required for your application.
FIGURE 1. WHICH FATIGUE REGULATIONS APPLY TO MY NEW TYPE DESIGN?
(Amendment 23-48, effective February 9, 1996.)
Is your structure constructed of composite materials?
NO Required: Damage Tolerance Evaluation YES 14 CFR §§ 23.573(a), 23.575 See AC 20-107A Required: 1.) Damage Tolerance Evaluation
14 CFR §§ 23.574(a), 23.575
Are you seeking certification YES in the commuter category?
or,
2.) If applicant can substantiate that the damage tolerance approach is impractical for the application, they may use a safe-life approach. 14 CFR §§ 23.574(b), 23.575 Required: For metallic structure in normal, utility, and acrobatic categories, an applicant may choose any of the following:
NO
1.) Safe-Life Evaluation For pressurized cabin, 14 CFR §§ 23.571(a), 23.575 For wing, empennage, and associated structure, 14 CFR §§ 23.572(a)(1), 23.575
2.) Fail-Safe Evaluation For pressurized cabin, 14 CFR §§ 23.571(b), 23.575 For wing, empennage, and associated structure, 14 CFR §§ 23.572(a)(2), 23.575
3.) Damage Tolerance Evaluation For pressurized cabin, 14 CFR §§ 23.571(c),
23.573(b), and 23.575 For wing, empennage, and associated structure, 14 CFR §§ 23.572(a)(3), 23.573(b), and 23.575 5
f. The most recent amendment to part 23 that revised the fatigue requirements was Amendment 23-48, effective February 9, 1996. Always check the latest amendment to part 23 to verify that the regulations referenced in this AC have not been superseded or that additional regulations have not been added.
1-4. Which metallic components should I include in the fatigue evaluation?
Although the fatigue regulations discuss “those parts of the airframe structure whose failure would be catastrophic,” a 1998 policy letter clarifies that you are required to include only certain sections of metallic airplanes in your fatigue evaluation. For airplanes certificated in any small airplane category, these sections include the pressurized cabin, wing (includes tandem wings, canards, winglets and tip fins), empennage, and associated structures. You do not need to include other portions of the airplane, including the landing gear and engine mounts.
1-5. Are the small airplane fatigue re gulations different in other countries?
a. The FAA fatigue regulations for small airplanes are the same as those of many foreign countries. The FAA and the CAA (Civil Aviation Authority) of these countries have worked together to harmonize the part 23 regulatory wording.
b. Even though the regulations may be identical, some differences in policy, guidance, and acceptable means of compliance may exist between the FAA and the CAA of foreign countries. These differences include:
(1) The CAA of some countries may require fatigue evaluation of the engine mounts and landing gear.
(2) The CAA of some countries may require different scatter factors in a safe-life evaluation.
1-6. How do I show compliance to the fatigue requirements for alterations, modifications, or changes to the design?
a. The fatigue performance of the structure should be evaluated for any alteration, modification, or change to the type design. P hysical changes to the design that may affect fatigue performance include changes that increase the external or internal loads, alter the distribution of internal loads, change the stress concentrations, or change the materials or method of construction. You should also ev aluate the effect of any change to the airplane’s operational characteristics that invalidate the assumptions used in the original fatigue substantiation.
Changes to the operational characteristics that may be important for fatigue include higher design airspeeds or higher average speed. They also include changes to the maximum allowable weight and center of gravity envelope, changes to the average weight and center of gravity location, and engine or propeller changes.
b. Modifiers and applicants for supplemental type certificates (STCs) often do not have access to the Type Certificate holder’s original fatigue substantiation data. In these cases, applicants often complete a “comparative fatigue analysis” in which the fatigue performance of the modified and unmodified airplane are compared. Applicants use the results of the comparative analysis to determine the appropriate changes to the safe-life or inspection program of the modified airplane.
CHAPTER 2. SAFE-LIFE FATIGUE EVALUATION
2-1. What is a safe-life fatigue evaluation?
A safe-life fatigue evaluation is the assessment of a structure to ensure that the structure is able to withstand, without catastrophic failure, the repeated loads of variable magnitude expected in service throughout its operational life. The “safe-life” is a point in the airplane’s operational life, expressed in hours or flight cycles, at which the structure must be taken out of service or replaced. The structure is reti red or replaced at the safe-life to prevent the structure from developing fatigue cracks. An applicant may determine the safe-life of a structure using the following methods:
a. full-scale testing,
b. component testing, or
c. analysis supported by test evidence.
2-2. What is the definition of a fati gue failure in a safe-life structure?
For a safe-life structure, fatigue failure is the development of a detectable crack. A detectable crack is one that can be detected by common inspection methods, or the inspection methods required in the maintenance instructions. Since a safe-life evaluation usually does not include demonstration of crack growth rates or residual strength capability, we assume that the development of a detectable crack may result in catastrophic failure of the structure.
2-3. What are the steps in evaluati ng the safe-life of a structure?
We can summarize the essential elements of any safe-life evaluation as follows:
a. Estimate or measure the expected loading spectra for the structure. The loading spectra should include all loads you expect the structure to experience in normal operation. This includes gust loads, maneuver loads, and pressurization loads. Although you are not required to include the landing gear and engine mounts in your fatigue evaluation, you need to include appropriate taxi loads, landing impact loads, engine loads, vibration, buffet, and propeller slipstream effects in your loading spectra.
b. Conduct a structural analysis, which includes consideration of the effects of stress concentrations. Your structural analysis should include determination of gross stress levels and determination of the stress concentration factors, Kt, for all details in principal structural elements. Your analysis should consider, in detail, structural joints and fittings, paying particular attention to eccentrically loaded joints and the bearing and bypass stresses in the joint. The structural analysis should identify locations where fretting may occur. Identify areas where corrosion may develop or extreme thermal environments may affect fatigue performance.
c. Conduct a fatigue test of any structure that cannot be related to a test background to establish the structure’s response to the typical loading spectrum expected in service. Full-scale fatigue testing is the most reliable method of determining the response of the structure to expected operational loadings. Fatigue substantiation by analysis is also acceptable. However, the stress-life (S-N) or strain-life (ε–N) data used in the analysis must be based on fatigue tests of similar structures or adjusted to account for differences between the actual structure and the tested structure.
d. Determine reliable replacement times by interpreting the loading history, variable load analysis, fatigue test data, service experience, and fatigue analysis. The reliability of the 7
replacement times is usually achieved through the use of fatigue scatter factors or other statistical adjustments.
e. Provide data for inspection and maintenance instructions and guidance information to the operators. This data should include inspectio n methods, inspection thresholds and intervals.
2-4. What loading spectra should I use?
a. Appendix 1 presents gust and maneuver flight load spectra in both graphic and tabulated form for the following types of airplanes and usage:
(1) Single-Engine Executive Usage (non-pressurized, engine size greater than 185 hp)
(2) Single-Engine Personal Usage (non-pressurized, engine size less than or equal to 185 hp)
(3) Single-Engine Instructional Usage (non-pressurized)
(4) Single-Engine Acrobatic Usage (non-pressurized)
(5) Twin-Engine General Usage (non-pressurized)
(6) Twin-Engine Instructional Usage (non-pressurized)
(7) Pressurized Usage
(8) Special Usage (including survey and aerial application)
b. For other types of loads:
(1) Appendix 1 also provides ground load spectra, including landing impact and taxi loads.
(2) The loading spectrum for pressurized cabins should include the repeated application of the normal operating differential pressure, and the superimposed effects of the gust and maneuver flight loads and external aerodynamic pressure.
(3) If the wing center section skin panels are affected by cabin pressurization, the wing loading spectrum should include the effects of pressurization.
(4) Reference 1 provides additional guidance on load spectra development and fatigue evaluation for the empennage, and for configurations with canards (or forward wings) and winglets (or tip fins).
c. Appendix 1 also discusses the gust load formula you should use to develop the gust load spectra for your airplane.
d. Your loading spectra should include the ground-air-ground cycle (GAG). The GAG cycle represents the range of the maximum and minimum loads expected to occur on a per flight basis. Typically, the minimum load results fr om landing or taxi conditions and the maximum load from the gust or maneuver spectra. For typical airplanes certificated in the normal category, two-thirds to three-fourths of the total fatigue damage on the wing may be caused by the GAG cycle.
e. While positive and negative load cycles are considered to occur randomly in service, the high positive and negative loads of a given type of repeated loading tend to occur at the same time. For testing and analysis purposes, the high positive load cycles are typically combined with the high negative load cycles of the same frequency.
2-5. How were the load spectra in Appendix 1 developed?
a. The FAA developed the flight load spectra in Appendix 1 based on a statistical analysis of the data presented in Reference 2. We used the ground load spectra directly from Reference 3 for the Appendix 1 ground load spectra.
b. The Appendix 1 flight load spectra include an increment (1.5 standard deviations) added to the average measured load frequency. The increment accounts for the variability in loading spectra experienced from individual airplane to airplane. The magnitude of the increment was selected to maintain the probability that a component will reach its safe-life without a detectable fatigue crack established by the scatter factor (see paragraph 2-15).
Reference 4 provides additional details on the development of the Appendix 1 flight load spectra.
2-6. Can I use other loading spectra in place of the Appendix 1 spectra?
The Appendix 1 spectra are the FAA’s approved load spectra for use in small airplane fatigue evaluations. Applicants may propos e alternate spectra in certain situations. If an applicant proposes alternate spectra, the Small Airplane Directorate Standards Office must approve these spectra. Any spectrum proposed by an appli cant should maintain the probability that a component will reach its safe-life without a detectable fatigue crack established by the scatter factor (see paragraph 2-15 and Reference 4). Si tuations in which applicants might consider alternate spectra include the following:
a. The applicant may develop gust spectra for pressurized usage based on power spectral density (PSD) methods. It is acceptable in this case to use PSD mission analysis techniques to account for typical operating mission profiles.
b. The applicant has access to a large database of measured spectra data on a similar configuration or on an airplane with similar usage. This type of data is particularly useful in situations involving severe or unique usage, such as acrobatic and agricultural usage.
2-7. What should I consider in developing a mission profile?
In developing the mission profile for your airplane, you should use the following assumptions:
a. Flight time
(1) Single-Engine and Twin-Engine (non-pressurized) 0.65 hours
(2) Pressurized 1.1 hours
(3) Single-Engine special usage (low level survey) 2.0 hours
(4) Twin-Engine special usage 3.0 hours
b. Airplane Speed
(1) The speed for determining miles flown should not be less than 0.9 VNO or 0.9 VMO.
(2) For special usage, the speed for determining miles flown may be 100 knots or
0.9 VA, whichever is less.
c. Gross weight and load distribution
(1) You should base estimates of the gross weight and distribution of disposable load on conservative estimates of typical operating conditions.
(2) One acceptable means of compliance, used by some small airplane applicants, is to assume the weight condition that gives the highest 1-g stress.
d. The mission profiles of certain airplanes should consider the unique aspects of the usage. Some examples follow:
(1) Instructional usage . Takeoff and landing traini ng, or “touch-and-go” training, is a significant portion of the student pilot training curriculum. Incorporate “touch-and-go” training into the mission profiles of any airplane used for instructional purposes. Each “touchand-go” should be treated as a short duration flight, 6 to 10 minutes in length.
(2) Mixed usage . Some airplanes are certificated in more than one category. A common example is an airplane certificated in both the acrobatic and utility/normal categories.
In general, the safe-life should be based on the usage that results in the shorter life. Some applicants have estimated a mix of missions, the percent of time spent operating in either category, as a way of accounting for the different types of usage. In using this approach, it is not acceptable to use a mix of missions within a single flight. For purposes of fatigue evaluation, a single flight is operated within a single category and within a single type of usage.
(3) Severe usage. Some airplanes may be used more severely than you had assumed in the fatigue evaluation. Common examples of this include airplanes considered to be in the single-engine personal or executive usage category used for pipeline and utility patrol, for instruction, or for short duration commuter and air taxi flights. Some applicants address this issue by placing a statement in the limitations section of the Airplane Flight Manual (AFM) that certain types of usage require a re-evaluation of the structure by the Type Certificate (TC) holder. The AFM lists these t ypes of usage and requests the owner to contact the TC holder regarding the usage.
2-8. Does the FAA require a flight test to validate the loads I use in my fatigue evaluation?
In most cases, the FAA will not require flight test validation of external loads or the internal stresses used in a fatigue evaluation. Typically, it is acceptable for an applicant to substantiate that the predicted external loads conservatively represent actual flight load conditions. However, it is in the applicant’s best interest to accurately know the internal loads, or stresses, at critical fatigue details. A one percent increase in stre ss level can shorten the expected fatigue life by more than five percent. Due to the sensitivity of fatigue life to stress level, an applicant should consider flight-testing to measure strains at or near fatigue critical structural details of the principal structural elements.
2-9. Does the FAA require validation of the stresses I use in my fatigue evaluation?
In determining whether validation or substantiation of the stresses used in a fatigue evaluation is necessary, the applicant and the FAA should consider the following:
a. For full-scale fatigue testing, an applicant should substantiate that the loads applied in the test either accurately or conservatively represent actual flight load conditions. Substantiation may be shown with comparisons of applied and predicted external loads. Substantiation should include demonstration that the applied external loads produce conservative stresses in the fatigue critical components. Strain gages should be used on the full-scale fatigue test and compared and correlated against the static test article or a validated analysis of the structure. Caution should be taken when using strain gages in areas of high stress gradients.
b. Component testing is different from full-scale testing in that the component test omits substantial portions of adjacent structure that may impart local stresses on the tested component.
Because of the omitted structure and other simplifications inherent in component testing, an applicant should validate that the stresses in the component test accurately or conservatively represent the stresses in the actual structure. One acceptable method of validating the internal loads is with strain gages on the airframe static test article. Strain gage measurements should be compared and correlated against the static test article or a validated analysis of the structure.
Caution should be taken when using strain gages in areas of high stress gradients. Consider applying unique conditions on the static test article to better represent fatigue loading conditions.
Loads applied in the static test should conservatively represent actual flight load conditions.
c. For fatigue evaluation by analysis, an applicant should validate the internal loads used in the analysis. In validating the internal loads, the applicant should validate the component’s total stress field, including magnitude and orientation. One acceptable method of validating the internal loads is with strain gages on the airframe static test article. Strain gage measurements should be compared and correlated against the static test article or a validated analysis of the structure. Caution should be taken when using st rain gages in areas of high stress gradients.
Consider applying unique conditions on the static test article to better represent fatigue loading conditions. Loads applied in the st atic test should conservatively represent actual flight load conditions.
d. Most modifiers and applicants for supplemental type certificates (STCs) evaluate the effect of their modification using a comparative fatigue analysis. In this case, it is usually sufficient for an applicant to substantiate that the loads and stresses used in their analysis conservatively account for the effects of their modification. The substantiation should incorporate the validation of stress analysis methods used for static strength substantiation.
2-10. How do I substantiate a fatigue safe -life with full-scale fatigue tests?
Full-scale fatigue tests are the most reliable method for establishing the fatigue performance of a structure. A full-scale test may use a complete, or nearly complete, airframe with flight, ground, and pressurization loads applied to the entire structure. A full-scale test may also be a series of separate tests of the individual major subassemblies of the airframe, such as the complete wing assembly, the empennage assembly, and the pressurized cabin. In testing major subassemblies, you should simulate the loads applied from any omitted structure and any fretting that would have been present with the omitted structure. For example, in testing the pressurized cabin, accurately simulate the loads and deflections, including assembly preload, imparted on the cabin structure by the wing or empennage.
2-11. What should I include in the test plan for my full-scale fatigue tests?
Your test plan should include at least the following items:
a. Description of the structure you will test. Describe whether you will test a complete airframe or separate major assemblies. Your test plan should include an analysis of the loading effects of any omitted structure and how you will simulate the loads imparted on the test article from these omitted structures. You may omit secondary structures from the full-scale test article provided the omitted structure does not contribute to the fatigue performance of the primary structure. The effects of secondary structure should be substantiated by analysis or previous test experience. Pay particular attention to the effects of fretting of any secondary structure you plan to omit.
b. Description of the test apparatus. Describe how the test ar ticle will be constrained and what effects the attachment to the test fixture will have on distributing internal loads in the test article. Describe how the test loads will be applied. The test set-up should be capable of accurately controlling the application of test loads. The set-up should record the time history of applied test loads.
c. Validation of internal loads. See paragraphs 2-8 and 2-9 of this AC.
d. Comparison of predicted internal loads and applied fatigue test loads. Make sure that the applied fatigue test loads produce internal loads that conservatively represent the predicted internal loads. You could show this with calibrated strain gages on the fatigue test article.
e. Description of loading sequences. The most realistic method of applying fatigue test loads is a random application of loads on a flight-by-flight basis. However, you may apply test loads in ordered loading blocks. The block length should be no greater than the number of flight hours (unfactored) that can be repeated ten times during the expected life. The sequence of loads should be from low to high to low within each block and use at least six load levels to describe each block. You should apply the ground-air-ground cycles either individually or in frequent blocks.
f. Description of truncated and clipped loads.
(1) The highest load level applied in the test should not exceed limit load or the load level that will be equaled or exceeded ten times in the life of the part. The magnitude of these high loads should be reduced, or clipped, down to the load level that will not be equaled or exceeded ten times in the expected life of the part. The intent of this precaution is to prevent the development of localized yielding.
(2) Your test loads should extend to the lowest level that causes significant fatigue damage unless you otherwise account for such loads. To expedite testing, you may omit, or truncate, lower level loads and simulate the fatigue damage resulting from these loads by applying a limited number of loads at higher load levels. However, enough cycles must be applied to account for the effects of fretting.
(3) Validate the load truncation level. You may do this with either coupon tests or analysis using data obtained from fatigue tests of similar structures.
g. How, when, and where you will inspect the structure.
(1) You should periodically inspect the test article to check for the development of detectable cracks. Use an anal ytical fatigue analysis of your structure to help determine the locations where cracks are likely to develop. Star t inspections well before the predicted life of the structure. Later inspections should be fre quent enough to allow an accurate estimate of the time of crack development. Your test plan should define the type of inspection and equipment required to perform the inspection. This will insure familiarity and proper calibration of the inspection equipment before the start of the test.
(2) Consider a teardown inspection of the structure after completion of the test. A teardown inspection often reveals unpredicted fatigue problems and can be valuable in addressing continuing airworthiness issues in the future.
h. Describe your definition of a completed test. The detection of a fatigue crack is not necessarily the end of a fatigue test.
(1) When testing a complete wing assembly, if a crack develops on one side, you may want to continue the test to determine when a crack develops on the other side of the wing.
You would then compute the safe-life based on two test results. Continuation of the test in this circumstance is acceptable provided the failed structure does not affect performing the test on the other wing.
(2) Some applicants choose to repair the cracked structure and continue the test.
Consider the effects of any repair on the distribution of internal loads and the associated effects on fatigue performance in the repaired area. We use the initial crack detection to determine the safe-life. However, applicants use the informa tion gained from continuing the test to provide a greater understanding of the fatigue performance of the structure and as a database in addressing continuing airworthiness issues. You should obtain concurrence from the responsible Aircraft Certification Office (ACO) before installing any repair on the test article.
(3) Decide before you start the fatigue test when you will stop if no detectable crack is found. Some applicants set a “design life goal.” Once the fatigue test article reaches this goal (“design life goal” multiplied by the appropriate scatter factor) without a detectable crack, they choose to stop the test at that point. Other applicants choose to let the test continue and set the safe-life as high as the test results will allow.
i. Approval of test plan and conformity inspection. The responsible ACO should review and approve your test plan. The FAA must also perform a conformity inspection of the test article and test setup.
2-12. Do I need to complete the test before the FAA will issue the type certificate (TC)?
a. The FAA recognizes the significant time often required to complete a full-scale fatigue test; therefore, the FAA may issue a type certificate before the applicant completes the full-scale fatigue test. In this instance, the safe-life specified in the limitations section of the Airplane Flight Manual (AFM) should be based on one of the following:
(1) The safe-life shown by an on-going test at the time of TC issuance. The safelife should be computed as shown in the following paragraphs.
(2) The safe-life substantiated by a fatigue analysis. Your analysis should follow the guidance on fatigue analysis included in paragraphs 2-24 through 2-26 of this AC. Your analysis may consider information gained from the on-going fatigue test but should not be a prediction or extrapolation based solely on the on-going test.
b. Your certification plan, compliance checklist, and methods of compliance document should discuss how to determine the safe-life for TC issuance. In developing your certification plan, consider that a revision to the limitations section of the AFM may be all that is required to document a safe-life greater than that specified at the time of TC issuance. However, for a shorter safe-life, an airworthiness directive (AD) may be necessary.
2-13. How do I calculate the fatigue safe-lif e using full-scale fatigue test results?
a. You can calculate the safe-life from the results of a full-scale fatigue test using the following equation:
The Safe-Life = The mean test life ÷ The scatter factor Equation 1
b. The following two paragraphs discuss the calculation of the mean test life and the purpose and derivation of the scatter factor.
2-14. How do I calculate the mean test life?
a. The calculation of the mean test life might seem a simple issue. However, the statistical distribution of fatigue life follows a log-normal statistical distribution; therefore, we should not use the arithmetic mean. We should instead calculate the mean based on the logarithms of the test lives. You may use the following two equations to calculate the mean test life based on the logarithms of the individual specimen test lives.
n log10 (meantest life)= 1 ∑log10 (test lifei ) Equation 2n i =1 10log10 (meantest life)meantest life = Equation 3
b. The following example calculation shows the use of Equations 2 and 3.
Given: Specimen 1 test life: 100,000 hours Specimen 2 test life: 200,000 hours Using Equation 2:
log10 (meantest life)= 1 [log10 (100,000) + log10 (200,000)]2 = 1 [5.0 + 5.30103]= 5.150515 2
Using Equation 3:
105.150515meantest life = = 141,421 hours
c. The mean test life using the logarithmic mean (141,421 hours) is less than you would calculate using the arithmetic mean (150,000 hours).
2-15. What are the scatter factors for full-scale fatigue tests?
a. The purpose of the scatter factor is to account for the statistical variability inherent in the fatigue performance of built-up structures. The equation for the minimum scatter factor, given below, assumes a log-normal statistical distribution of fatigue life.
ns +1Z σ s Scatter FactorFST = 10 p n Equation 4 where: FST denotes full-scale test.
Zp is the normal distribution variate for the specified probability of a detectable crack-free safe-life.
σ is the standard deviation of the population of fatigue test lives. For use in this equation, σ , is a percentage of the log of the mean test life for the built-up or monolithic structure.
ns is the number of fatigue specimens tested.
b. Reference 3 specified a value for Zp = 3.511. This value for the normal distribution variate equates to a probability of 99.9777 percent. For any individual structural component in a
small airplane, the FAA’s specified probability that the component will reach its safe-life without a detectable fatigue crack is 99.9777 percent.
c. The value for the standard deviation of the fatigue test life is usually taken from historic data. This is due to the high cost of testing sufficient numbers of full-scale test articles to develop an accurate value for the standard deviation. For built-up aluminum structures, Reference 3 specified a value for σ = 0.14. The following table summarizes the commonly accepted values for the standard deviation for various structural metals.
Table 1. Standard Deviation of Log Fatigue Life for Built-up Structures Constructed of Common Structural Metals (Reference 5) Base Material of Built-Up or Monolithic Structure σ, Standard Deviation (as percent of log life) Aluminum alloys 14% Steels (ftu = 100 – 200 ksi) 17% Steels (ftu = 200 – 300 ksi) 25% Titanium alloys 20%
d. These values for the standard deviation are for complete, built-up structures. Values for the standard deviation given in material handbooks, such as Reference 6, are from simple coupon tests and should not be used to determine full-scale fatigue test scatter factors. An applicant may propose alternate values for the standard deviation provided they have adequate substantiating data. Any alternate values for the standard deviation must be coordinated with the Small Airplane Directorate Standards Office.
e. The table below summarizes the required full-scale test scatter factors for aluminum structures. You may recognize th e commonly known scatter factor of 4.0 from Reference 3. If your structure is made of a metal other than aluminum, you need to use Equation 4 to calculate the correct scatter factor.
Table 2. Full-Scale Fatigue Test Scatter Factors for Aluminum Structures For Use with Safe-Life Structures Number of Test Specimens Required Scatter Factor Probability of Detectable Crack-Free Safe-Life
99.9777% (Zp = 3.511) Standard Deviation of Log Fatigue Life, σ
0.14 (14%) for Aluminum Structures 1 4.96 2 4.00 3 3.70 4 3.54 15
2-16. What about statistical confidence inter vals on full-scale fatigue test results?
a. Certification authorities, military specifications, and industry standards often require the use of confidence intervals in the analysis of test results. A common example is the requirement for material strength allowables to be given as an “A-basis,” defined as 99 percent probability of survival with 95 percent confidence. A common requirement for fatigue is 99.9 percent probability of survival with 95 percent confidence. The confidence, or confidence interval, is a statistical method of estimating the possible range for the true mean of a population based on the test results of only a few specimens. The confidence interval gives a lower and upper boundary for the likely value of the population mean. In establishing design allowables, or design lives, we are interested in the lower boundary for the mean.
b. In the statistical model we have used to determine the scatter factor, each individual fatigue critical component has a probability of reaching its safe-life without a detectable crack of
99.9777 percent (with no confidence interval on the mean). This probability is equivalent to the following:
(1) With two test results (scatter factor = 4.0): 99.9 percent probability with 91.3 percent confidence (or 99.822 percent probability with 95 percent confidence).
(2) With one test result (scatter factor = 4.96): 99.9 percent probability with 93.95 percent confidence (or 99.868 percent probability with 95 percent confidence).
c. The scatter factor model we are using provides probabilities nearly equivalent to alternate fatigue requirements that use confidence intervals. However, be aware that the CAAs of other countries often require the use of a somewhat larger scatter factor that provides 99.9 percent probability with 95 percent confidence.
2-17. What counts as one test specimen, and what counts as two test specimens?
a. Structural components or details that occur more than once in a test article may count as two or more test specimens provided the details have the all of following:
(1) Same structural function.
(2) Same geometry.
(3) Same materials.
(4) Same stress level, both flight and ground.
(5) Same stress concentration factor.
(6) Same flight and ground load spectrum.
b. Testing of a complete wing, both the left and right side, generally counts as two test specimens. Similarly, certain frame, stringer, cutout, and other details that occur symmetrically in a complete pressurized cabin may count as two or more test specimens.
c. Examples of testing that might count as only one test specimen include details in the vertical tail, the keel beam in a pressurized cabin, and non-symmetric details in the pressurized cabin, such as entrance and emergency exit doors.
d. A 1996 policy letter further clarifies when a test article counts as only one test specimen. An applicant asked, “Does the fact that each fuselage fitting assembly consists of two parts, back to back, mean that the time to initial failure of the one assembly counts as 2 (two) results?” Our answer was, “If each part can independently sustain ultimate load for the assembly, then our answer is yes. If each part cannot sustain ultimate load, and safe-life