FIGURE S1-1: Typical Example Conventional Life Management Process
Methods and Materials Data Life Estimation
Development and Validation Tests Service Life and Product Assurance
S1-3
g. The requirements for life estimation are represented by the elements in FIGURE S1
1. Life limits are to be specified for any engine rotor component where a failure of that component would result in a hazard to the aircraft. Within the cyclic life span, it is assumed that continued safe operation of the specific rotor design, up to the stated life limit, is not contingent upon interim or periodic inspections. The key elements of life estimation include:
(1) Flight Cycle Definition: A flight cycle should be defined that considers the most severe and typical operation of the engine. The flight cycle provides the operational definition so that the stress and thermal analyses can capture the transient and cyclical loading.
The part lives are quoted in terms of this flight cycle. The flight cycle should include the various flight segment such as start, idle, takeoff, climb, cruise, approach, landing, reverse and shutdown. The hold times at the various flight segments should correspond to the limiting installation variables (aircraft weight, climb rates, etc.). The corresponding rotor speeds, internal pressures, and temperatures during each flight segment should be adjusted to account for production tolerances and installation trim procedures, as well as engine deterioration that can be expected between heavy maintenance intervals. The range of ambient temperature and takeoff altitude conditions encountered during the engines’ service life should also be considered.
(2) Thermal Analysis: The flight cycle along with the corresponding performance data, act as boundary conditions for the thermal analysis. Temperature levels and thermal gradients should be determined throughout the flight cycle and should be calibrated and verified by engine rotor temperature data.
(3) Stress Analysis: Analytical and empirical engineering processes are applied to determine the stress distribution for each structural part. The analyses evaluate the effects of engine speed, pressure, metal temperature and thermal gradients at many discrete flight cycle conditions. From this, the component’s cyclic stress history is constructed. Stress concentration effects due to geometric discontinuities such as dovetails, boltholes and contour changes should be included.
S1-4
(4) Life Prediction System: The life prediction system is based upon test data obtained from cyclic testing of specimens and components fabricated from representative production grade material to account for the manufacturing processes that affect LCF capability.
Sufficient testing should be performed to evaluate the effects of elevated temperatures and hold times, as well as, interaction with other material failure mechanisms such as high cycle fatigue (HCF) and creep. Test data should be reduced statistically in order to express the results in terms of minimum LCF capability (1/1000 or alternately -3 sigma) and the safe-life should be quoted to initiation of a fatigue crack. Appropriate prior service experience gained through a successful program of parts retirement or precautionary sampling inspections, or both, may be included to expand the database.
S1-5
SECTION 2
ENHANCED LIFE MANAGEMENT PROCESS
1. History has shown that the safe-life philosophy has served the turbine engine industry and flying public well, and provides a solid base for further enhancement. Modifications to the conventional life management procedures should augment, not supplant, current approaches, which are based on the safe-life philosophy.
a. By adding a new element, damage tolerance, to the existing conventional life management process, the enhanced life management process has been defined to further minimize the occurrence of uncontained rotor failures and thus improve flight safety (see FIGURE S2-1).
b. Use of the enhanced life management process depicted in FIGURE S2-1, will result in damage tolerance assessments being conducted on critical titanium alloy rotor designs. These will be fracture-mechanics based probabilistic risk assessments, the results of which will be compared to the agreed upon design target risk (DTR) values. Designs that satisfy these DTR values will be considered to comply with the requirements of § 33.14. The engine manufacturer will have a variety of options available to achieve the DTR values. They include, but are not limited to, component redesign, material changes, material process improvements, manufacturing inspection improvements, in-service inspections, and life limit reductions.
c. A general methodology for conducting the fracture-mechanics based probabilistic analyses mentioned above is presented in Section 3 (Damage Tolerance Implementation).
S2-1
FIGURE S2-1: Enhanced Life Management Process Safe Life Prediction Methods & Material Tolerance Damage New Element Added to Data Assessment Existing Safe Life Process Life Management Process Development Service Life & Validation & Product Tests Assurance S2-2
SECTION 3
DAMAGE TOLERANCE IMPLEMENTATION
1. Approach
a. Fracture-mechanics based probabilistic (FMBP) assessments are a key feature of the enhanced life management process. The results of these assessments will provide the basis for evaluating the relative damage tolerance capabilities of candidate engine designs, and will also allow the engine manufacturer to balance the designs for both enhanced reliability and customer impact. The results will be compared against agreed upon DTR values (see paragraph 5 of this Section).
b. The necessity for further risk reducing actions on the part of the engine manufacturer will be based upon whether or not the design under consideration satisfies the desired DTR values at both the individual component level and the overall engine level. If the targets are met, then the design is considered to comply with § 33.14. The manufacturer may conduct quantitative parametric studies to determine the influence of key variables such as inspection methods, inspection frequency, hardware geometry, hardware processing, material selection and life limit reduction. The manufacturer may then make changes to the design or field management of the part, or both, in order to achieve the agreed upon DTR values, as shown in FIGURE S3-1. This approach provides the individual engine manufacturer with the flexibility to develop an optimal engine design solution consistent with customer requirements and company policies, procedures, and available resources. An example assessment utilizing this methodology is described in paragraph 2 of this Section, and provided in Appendix 1 of this AC.
c. FMBP assessments will usually be performed during the engine component detail design phase. Paragraph 2 of this Section defines an assessment methodology applicable to material melt-related induced anomalies. It contains a standardized list of inputs for conducting FMBP assessments, and a process to refine a design to achieve the desired DTR values.
2. Methodology
a. Probabilistic risk assessments may be conducted using a variety of methods, such as Monte Carlo simulation or numerical integration techniques; the methodology S3-1
and assumptions used to determine the risk should be approved by the FAA. FIGURE S3-1 conceptually depicts a melt-related anomaly probabilistic assessment. When performing an FMBP assessment, use of the standardized inputs and default data presented in this AC will achieve consistent FMBP assessment results across the industry.
b. A list of standardized inputs is provided below. Default input data is described in paragraphs 3 and 4 of this Section. Use of this default data in probabilistic assessments requires no further demonstration of applicability or accuracy. However, where this data is used, the guidelines accompanying this data should be followed by the engine manufacturer to ensure applicability. Use of input other than these default values may require additional validation to verify applicability, adequacy, and accuracy.
c. Probabilistic risk assessments should incorporate the following inputs as part of a basic analysis: component stress and volume, material properties, crack growth lives, anomaly distribution, design service life, inspection PODs, and maintenance exposure rate.
(1) Input. The following criteria are defined as basic input data.
(a) Anomaly Distribution. For melt-related (hard-alpha) assessments, use the default anomaly distributions outlined in paragraph 3 of this Section, or FAA approved company-specific data. Company-specific data should be developed using the same process employed to derive the default anomaly distributions. This process is described in a technical paper titled “Development of Anomaly Distributions for Aircraft Engine Titanium Disk Alloys” (see Section 5). Anomalies will be treated as sharp propagating cracks from the first stress cycle.
(b) POD. Default PODs and instructions on use of individual company values are contained in paragraph 4 of this Section. Subsurface assessments should consider the effects of ultrasonic inspection only. Surface assessments should likewise consider the effects of fluorescent penetrant inspection (FPI) and eddy-current inspection (ECI).
(c) Maintenance Exposure Interval. For the purpose of assessing inspection benefits, exposure interval curves for the engine, module, or component in question should be modeled in the analysis.
(d) Stress. Variable influencing crack propagation life. Input should encompass the most limiting operational principal stresses. Subsurface S3-2
assessments should incorporate the appropriate subsurface and near surface stress distribution.
Surface assessments should incorporate the appropriate surface stress distributions, including the effects of stress concentration. Flight cycle(s) for design LCF certification, or actual usage (if known) should be used to establish the stress profile. The influence of major and minor cycles needs to be considered since the cyclic damage accumulation can be dramatically different for crack propagation as opposed to initiation. Note that the method described in this AC has been calibrated against industry experience without consideration for the effect of surface enhancements such as shot peening on the predicted crack propagation lives. As a result, it is inappropriate to include the beneficial effects of such enhancements on any calculations at this time. This issue may be addressed in future revisions of this document.
(e) Volume. Affects the probability of having a defect in the material.
Represents the volume of material at a specific level of stress. For subsurface assessments, represents the volume of material at various levels of subsurface stress. Surface assessments should incorporate volume constituting a thin layer ("onion skin") around the surface of the part.
Where a non-axisymmetric feature, such as a series of holes in a disk web has a localized stress concentration, the decision on whether it makes a significant contribution to the probability of burst must be based on the combination of mass of material at high stress and the size of the anomaly which would cause the part not to reach its safe declared life. While the method described in this AC assumes axisymmetric features, a non-axisymmetric feature can be included by reducing the cross-sectional area to ensure that the total volume when integrated around the whole circumference is equal to the volume at high stress.
(f) Materials Data. Average cyclic crack growth rate properties of the base material generated in an air environment, should be used as the default condition to calculate anomaly propagation life.
(g) Propagation Life. Number of cycles for a given size anomaly to grow to a critical size. Based upon knowledge of part stress, temperature, geometry, stress gradient, anomaly orientation, and materials properties. Linear elastic fracture-mechanics should be used for the calculation of propagation life. Default conditions should assume anomalies to be in the worst orientation to the stress field.
(2) Calibration
(a) Each engine manufacturer should calibrate its analytical prediction tools by conducting the industry test case detailed in Appendix 1. The test S3-3
case consists of a probabilistic analysis of a simulated titanium ring disk, using specified inputs and scenarios. Test case results in the ranges from 1.27E-09 to 1.93E-09 (for the “no inspection” case) and from 8.36E-10 to 1.53E-09 (for the “with in-service inspection” case) are considered acceptable. Test case results outside of these ranges may indicate problems with either the probabilistic assessment technique or the assumptions made.
(3) Output
(a) Component Level Assessments. The probabilistic assessments and prediction of event potential should be calculated over the entire anticipated service life of the part. This result should be expressed as the number of predicted events for each cycle, and be designated as the predicted “component event rate”. For multiple stage components such as spools, the assessment should be conducted for each stage. The predicted component event rate should then be compared to the component level DTR value to assess design acceptability.
(b) Engine Level Assessments. Once all critical titanium rotors in a given engine configuration have satisfied the component level DTR value, the cumulative event rate for these components should be calculated and compared to the engine level DTR value for acceptability.
(4) General Comments. At this time, standardized inputs and default data for FMBP assessments are available only for titanium material melt-related (hard-alpha) anomalies.
Future updates to this AC may include material anomaly information for other rotor grade materials such as Nickel as well as inputs and default data for manufacturing and maintenance induced anomalies. The corresponding default data for DTR values for titanium materials are specified in paragraph 5 of this Section.
(a) The FAA strongly encourages engine manufacturers to incorporate fracture resistant design concepts where possible. Credit may be given for fracture resistant or burst resistant engine design features that clearly demonstrate, through both analysis and test, a reduction in the potential for rotor failure due to the presence of unanticipated material melt induced anomalies.
S3-4
(b) Because the design of an aircraft turbine engine rotor is a lengthy process involving numerous iterations, each of which can substantially alter the initial calculated DTR values, it is important that the DTR values be satisfied at the time of engine certification.
Some FMBP assessments may also occur several years after the engine enters service because of such influences as design changes associated with in-service problems or changes in the analytical results due to evolving predictive capability. It is important to state here that the DTR values should be satisfied at both the individual component and overall engine levels throughout the life of the hardware.
S3-5
Pea k Stres s Condition Probabilistic Assessment Part Design Life FIGURE S3-1: Typical Elements of a Titanium Melt Related Anomaly Risk Assessment S3-6
3. Anomaly Distributions. A key input distribution associated with FMBP assessments is the size and rate of occurrence of the anomalies. This type of information is statistical in nature and can be presented in a form that plots number of inclusions that exceed a particular size in a specified amount of material.
a. Titanium Melt-Related (Hard-Alpha) Distributions.
(1) Hard-alpha anomaly distributions that apply to fully machined components have been developed to characterize a type of inclusion found in titanium (See Appendix 3).
The distributions are applicable to titanium rotor components manufactured after 1995; using the triple vacuum arc remelt (VAR) process or, the cold hearth melt (CHM) plus VAR processes.
For example, FIGURE A3-7 represents the anomaly distribution for a component which was inspected at both the billet and the sonic shape to at least a #3 FBH (flat bottom hole).
(2) The anomaly distributions contained in Appendix 3 may be used to determine compliance with paragraph 5 of this Section, DTR requirements. The background of the development of these distributions are contained in a technical paper titled “Development of Anomaly Distributions for Aircraft Engine Titanium Disk Alloys.” Briefly stated, the distributions were developed by modeling a complex series of interrelated steps that simulated the entire component manufacturing process from billet conversion to final machining. The final distributions were validated based on field experience. This process resulted in a number of distributions that could be used to determine a relative risk reduction, but not an absolute level of risk.
(3) Individual engine manufacturers who desire to utilize an alternate anomaly distribution or an improved inspection should use the methodology contained in reference No. 5 of Section 5 to create the alternate distributions. The alternate distribution must be substantiated by the appropriate background data. An alternate distribution should include the following: (a) three dimensional inclusion data, (b) inspection POD data, (c) should account for potential undetected, uncracked and unvoided inclusions, and (d) should be based upon substantial field experience.
b. Manufacturing/Maintenance Induced Anomaly Distributions. At the present time, there are no default data distributions for use in FMBP assessments for manufacturing or maintenance based anomalies.
S3-7