33.70-2 8/28/09 1 Crack propagation under out-of-phase stress vs. temperature transient loadings is a complex phenomenon. Compared to isothermal tests, non-isothermal experimental results suggest that the crack propagation rate can be higher than predicted at the temperature of the maximum stress when the maximum temperature during the flight is greater than the temperature at the maximum stress (see figure 2-1).
2 While some OEMs have developed models to predict this phenomenon, no industry standard is yet available.
3 The following design guidelines provide a default approach for the probabilistic analysis described in this AC.
4 Design Guidelines (Default Option). These are not intended to become a general industry standard. For the major and minor stress cycles over the full flight, the temperature to consider is the most life limiting between:
• Temperature at maximum stress during the cycle, and
• Maximum temperature during the flight 5 Design Guidelines (Alternative Option). A crack propagation model accounting for out of phase temperature vs. stress history for the full flight can be used if it is validated with the OEM laboratory and field experience on the considered material.
Figure 2-1. Stress and temperature transient loadings over a flight mission major cycle max stress Stress Temperature flight max temperature temperature at major cycle max stress time Stress Temperature major cycle min stress 6
33.70-2 8/28/09
3. Anomaly Distributions .
a. A key input distribution associated with PDTRA 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 specifies the number of anomalies that exceed a particular size in a given amount of surface area.
b. Manufacturing-Induced Anomaly Distributions.
(1) Manufacturing-induced anomaly distributions that apply to circular hole features in disks have been developed to characterize the size (depth) distribution and frequency of said anomalies (see Appendix 2). The distributions apply to circular holes machined in titanium, steel, nickel, or powder nickel rotor components.
(2) The anomaly distributions contained in Appendix 2 may be used to determine compliance with paragraph 6 (“Design Target Risk”) of this chapter. The background pertaining to the development of these distributions is contained in the technical paper “Development of Anomaly Distributions for Machined Holes in Aircraft Engine Rotors.” The distributions were developed by modeling crack sizes detected during component/engine testing or through inservice experience using an EIFS (Effective Initial Flaw Size) approach. The final distributions were validated based on OEM field experience. This process resulted in a probability of exceedance curve (see figure A2-1) that can be used to determine a relative risk reduction but not an absolute level of risk.
(3) Individual engine manufacturers who want to use an alternate anomaly distribution or an improved inspection capability, or extend the PDTRA process to features other than circular holes should use a methodology similar to that contained in the paper “Development of Anomaly Distributions for Machined Holes in Aircraft Engine Rotors.” This paper contains instructions on how to develop an alternate distribution, which must be substantiated by the appropriate background data. Alternate manufacturing-induced anomaly distributions should:
(a) include fracture and crack find data,
(b) include inspection POD data, and
(c) be based upon substantial field experience.
4. Manufacturing Credits Applicable to Manufacturing-Induced Anomaly Distributions.
a. To encourage the use of best practices when manufacturing circular holes in rotor lifelimited parts, a process has been established to take credit for available manufacturing process validation and controls that reduce the risk of hole damage during manufacture. This process includes:
(1) Process validation to a modern standard as described in FAA Report number DOT/FAA/AR-06/3.
33.70-2 8/28/09
(2) Single point boring as a method of finishing a hole. This is restricted to titanium alloys.
(3) Honing all alloys as a method of finishing a hole.
(4) The provision of a coolant monitor to ensure the flow of coolant towards the cutting edge.
(5) Power monitoring to a standard detailed in the FAA Report number DOT/FAA/AR-06/3.
(6) Other process monitoring, including feed force, vibration, etc., to a standard similar to that for power monitoring above.
(7) Inspection for detection of non-geometric anomalies.
b. The degree of allowable manufacturing credit is a function of which of the above available actions are actually employed during hole making. Calculation of manufacturing credit is outlined in Appendix 6. The manufacturing-induced probabilistic damage tolerance approach has been developed excluding the benefits of enhanced manufacturing validation and process controls. Hence, the credit is applied as a reduction factor to the results of the manufacturinginduced probabilistic assessments prior to comparing with the hole feature DTR.
5. Default Input - Nondestructive Evaluation (NDE) POD.
a. Detection of non-geometric anomalies, such as highly distorted grain structure, white or amorphous layer, and bent grain, which have dominated the experience of cracking in rotors, is quite distinct from detecting cracks and geometric anomalies such as nicks or dents. Cracks that may develop from non-geometric anomalies can be detected by conventional methods, such as eddy current, penetrant or ultrasonic inspection, when the part is inspected at shop visit. These methods, however, may be of very limited benefit at new manufacture. If credit for inspection at new manufacture is claimed, applicants must demonstrate that the capability of the technique on the target anomaly types. The default POD data supplied in this AC in Appendix 4 only refers to the detection of cracks.
b. The capability of individual NDE processes, such as eddy current, penetrant, or ultrasonic inspection, for the detection of local material anomalies (discontinuities or potential anomalies), is a function of numerous parameters, including the size, shape, orientation, location, and chemical or metallurgical character of the anomaly. In addition, the following four parameters should be considered when assessing the capabilities of an NDE process:
(1) The material being inspected (for example, its composition, grain size, conductivity, and surface texture).
(2) The inspection materials or instrumentation (for example, the specific penetrant and developer, the inspection frequency, instrument bandwidth and linearity).
(3) The inspection parameters such as scan index.
33.70-2 8/28/09
(4) The inspector (for example, his or her visual acuity, attention span, and training).
c. The “default” POD data supplied with this AC are characteristic of inspection capability that has been measured under typical, well-controlled conditions. These default POD values are provided primarily to facilitate selection of nondestructive inspection techniques that are best suited to support attainment of damage tolerant inspections. Manufacturers should recognize, however, that although properly applied inspections should result in capability similar to these default values, they are strictly applicable only under the conditions under which they were acquired (see Appendix 3). POD curves are described in Appendix 4, figures A4-1 and A4-2, as listed below:
(1) Appendix 4, Figure A4-1. Mean POD for Fluorescen t Penetrant Inspection of Finish-Machined Surfaces.
(2) Appendix 4, Figure A4-2. Mean POD for Eddy Current Inspection of Finish Machined Surfaces.
NOTE: Refer to Appendix 1 for an exam ple of the use of this data and to Appendix 3 for the NDE Applicability of these POD curves.
6. Design Target Risk.
a. The DTR is a benchmark relative risk level selected to enhance the overall safety of high-energy rotating components. Since no machine or device is 100-percent reliable, it is inappropriate to require a level that is technologically unachievable. Nevertheless, the goal is to achieve a significant and distinct improvement over and above current rotor designs.
b. For manufacturing-induced anomaly assessments, designs must meet the specified feature DTR to be considered acceptable. (Note: Only circular hole feature assessments and DTR’s are currently available.)
c. Design Target Risk values are specified in Table 2-1 below:
Table 2-1. Allowable Risk for Circular Holes Component Feature Design Target Risk (feature events per component published service lifetime) Circular Holes 2.0 E-05 9
33.70-2 8/28/09 CHAPTER 3. “SOFT TIME INSPECTION” ROTOR LIFE MANAGEMENT
1. Approach . The overall life management process encompasses a wide spectrum of design, manufacturing, and product support issues. This section addresses only one facet of that process, the assurance of structural integrity using inspection techniques and intervals derived from a damage tolerance (fracture mechanics-based) assessment. The inspection philosophy is intended solely to protect against anomalous conditions. It is not intended to allow operation beyond the safe-life limit specified in the Airworthiness Limitations Section of the Instructions for Continued Airworthiness.
a. When probabilistic assessment indicates risk levels greater than the desired target, many strategies can be used to reduce the predicted risk to the appropriate level. However, only the inservice inspection option is addressed here.
b. The industry data on uncontained fracture experience summarized in SAE reports AIR 1537 (1959 through 1975), AIR 4003 (1976 through 1983), and SP1270 (1984 through 1989) was used to guide the development of the inspection philosophy. These reports indicate that maintenance-induced uncontained failure rates were comparable to the failure rates for anomalous conditions (material and manufacture). This data suggests that additional inspection requirements, if not properly integrated into the normal maintenance scheduled for the engine, would have no net benefit to reduce uncontained failure rates.
c. The inspection philosophy presented here evolved from the desire to easily integrate inspections into the operation of the engine yet still achieve measurable reduction in the uncontained failure rates. The inspection philosophy advocates the use of opportunity inspections rather than forced inspections at ‘not to exceed’ intervals. These opportunity inspections occur due to the ‘on-condition’ maintenance practices used by operators today. Although opportunity inspections occur at random intervals, they can be treated statistically and used effectively to lower the calculated risk of an uncontained event.
d. Opportunity inspection refers to those instances when the hardware in question is available in a form such that the specified inspection can be performed. This condition is generally viewed as being reduced to the piece part; however; opportunity inspections can be performed on assembled modules. For example, an ECI of a disk bore may be specified on an assembled module whenever the module is available. This inspection is an opportunity inspection based upon module availability rather than piece part availability.
e. Whenever possible, the designs should use opportunity inspections to meet the DTR levels. However, in some instances, the probabilistic analysis may indicate an unacceptable risk level when using just opportunity inspections and some additional action may be required to meet the DTR. One of the options to mitigate this risk is to force inspection opportunities by specifying disassembly of modules or engines when a cyclic life interval has been exceeded.
There are many options on how to implement forced disassembly. Options range from mandatory engine removal and subsequent teardown at not to exceed cyclic limits (“hard-time” limits) to mandatory module teardown when the naturally occurring module availability exceeds the specified cyclic life inspection interval of one of the parts contained within that module (“soft-time” limits). This AC advocates the use of the piece part soft-time inspection option and 10
33.70-2 8/28/09 only suggests that forced disassembly of modules be considered when it is required to meet the DTR levels.
f. The soft-time inspection philosophy retains the “on-condition” maintenance practice and minimizes the impact of additional module disassembly. The inspection requirement exists only after the engine has been removed from the aircraft for a reason other than the inspection itself and is in a sufficient state of disassembly to allow access to the module containing the component in question. An available module containing a part with Cycles Since Last Inspection (CSLI) in excess of the soft-time interval must be disassembled to a condition that allows inspection by the procedure specified by the engine manufacturer. The engine manufacturer must evaluate the risk associated with parts that become available for inspection before the soft-time interval to determine if the CSLI can be reset.
g. The maintenance impact of the soft-time intervals should be considered during the design phase. Use the probabilistic analysis, summarized in Chapter 2, along with the anticipated engine removal rate and the module and piece part availability to develop designs that not only achieve the design target but also result in acceptable soft-time intervals and procedures, if such action is required.
h. When invoked, the soft-time inspection approach establishes interval limits beyond which rotor components must be inspected when the rotors are available in modular form. The soft-time inspection requirement is not intended to affect the current practice of forced inspection programs to address safety of flight concerns that arise in the course of engine operation and maturation. These safety of flight concerns should continue to be addressed through aggressive inspection programs, which are communicated using service bulletins.
i. Applicants should recognize that the inspection assumptions made in the probabilistic risk assessment must be communicated and implemented accurately to the field by using the ALS of the ICA. These assumptions must also be validated by the review of engine removal rates and module and piece part availability data. For example, the ALS must call out an eddy current inspection if that was an assumption in setting the original soft-time interval. Similarly, the amount of inspected material should correspond to the analysis assumptions. Likewise, if field experience suggests that the opportunity inspection intervals are in excess of the assumed rates in the probabilistic risk assessment, then appropriate corrective action, such as a modified inspection plan, is required.
j. The soft-time inspection interval and reference to the corresponding inspection procedures must be specified in the ALS of the ICA. This information must be provided for all rotor parts with specified retirement life limits that require any inspection plans beyond opportunity inspections to meet the DTR levels. The required inspection information should also be included in the individual ALS of the ICA with the other rotor inspection requirements. The manufacturer must also provide necessary information to focus the prescribed inspections to those areas of highest relative risk.
2. Inspection Scenarios . The following scenarios clarify th e action that should be taken at a maintenance inspection opportunity. Note that the inspection plans may vary for each part, depending on the outcome of the probabilistic assessments.
33.70-2 8/28/09
a. Maintenance Opportunity - Hardware Available For Opportunity Inspection. Hardware available in the condition to perform the specified opportunity inspection must be inspected by the procedures specified in the ALS of the ICA. This is a mandatory inspection.
b. Maintenance Opportunity - Module Below Soft Time Interval. Hardware accessible in the assembled or partially disassembled module may be nondestructively inspected by the procedures specified in the ALS of the ICA. The CSLI may be reset to zero, provided the engine manufacturer has assessed the risk impact associated with this action. This is a discretionary inspection.
c. Maintenance Opportunity - Module Above Soft Time Interval. Hardware listed in ALS of the ICA must be made available for nondestructive inspection, using the specified procedures.
This inspection must be performed whenever the module is available and the CSLI for any contained hardware that exceeds the inspection cycle limit. This is a mandatory inspection.
3. Soft Time Inspection Interval Example. An example of Soft Time Inspection Interval development is provided in Appendix 2 of AC 33.14-1.
Appendix 1 APPENDIX 1. MANUFACTURING-INDUCED ANOMALY CALIBRATION TEST CASE
1. This appendix provides detailed instructions for the calibration of a probabilistic risk assessment methodology for manufacturing-induced anomalies in circular holes. The package includes all required input data for the test case; analysis guidelines; and an analysis section that permits manufacturers to estimate the level of acceptability of their risk calculations and gain insights on intermediate results.
2. Test Case Input Data.
a. Problem Description. The test case geometry (figure A1-1) consists of a titanium ring disk under simple cyclic loading for 20,000 cycles. The maximum speed is 5,700 RPM and an external pressure load of 33 MPa (4.786 ksi) is applied on the outer diameter to simulate blade loading. The disk contains 40 circular holes of ½ inch (1.27 cm) diameter at a radial location of
14.27 inches (36.2458 cm).
Figure A1-1
GEOMETRY ROOM TEMPERATURE TEST CYCLE
P SPEED 5,700 rpm L R1 R2 Rbh RING R1 = 0.425m (16.732 in) L = 0.0254 m (1.00 in) o t
R2 = 0.3 m (11.811 in) P = 33.0 Mpa (4.786 Ksi)
Rbh = 0.3625 m (14.27 in)
b. Anomaly distribution curve. The anomaly distribution is provided in Appendix 2.
Finished part manufacturing inspections are fully accounted for in this curve. No additional modifications are necessary. The anomaly distribution is defined in terms of depth and is dependent on the Length to Diameter (L/D) ratio for the hole. The anomaly distributions should be linearly extrapolated when anomaly sizes are required outside the range of data provided.
c. POD. The POD curves used to determin e the effect of an in-service inspection are contained in Appendix 4. The default curves to be used are the Mean (50 percent Confidence) POD for eddy current inspection of finish-machined surfaces with the reject level set at one-half the calibration notch threshold response. For this test case, we assumed that this curve applies to the whole surface, including the entire hole and the front and aft face of the component.
d. Material Data. Two sets of material data are provided:
(1) Physical properties. Data required:
Appendix 1 Density: 4,450 kg/m 3 or 0.161 lb./in3
Young’s modulus: 120,000 MPa or 17.4E3 ksi Poisson's ratio: 0.361
(2) Crack Growth. Assume the following da ta represents air crack propagation. Crack propagation rate:
K threshold = 0.0 MPa √m or 0 ksi√in Fracture toughness= 64.5 MPa √m 58.7 ksi√in For K ratio (Min K/Max K) = 0.0 da/dN = 9.25 E-13 (∆K)3.87 (da/dN in m/cycle and ∆K in MPa√m)
- or - da/dN = 5.248 E-11 (∆K)3.87 (da/dN in in/cycle and ∆K in ksi√in ) For K ratio (Min K/Max K) = -1.0 da/dN = 1.281 E-14 (∆K)3.87 (da/dN in m/cycle and ∆K in MPa√m)
- or - da/dN = 7.2684 E-12 (∆K)3.87 (da/dN in in/cycle and ∆K in ksi√in )
(a) The above data apply at the test case component temperature.
(b) Crack propagation data are stress ratio dependent.
(c) Crack propagation data for a stress ratio of 0.0 was taken from MCIC-HB-01R, “Damage Tolerant Design Handbook, A Compilation of Fracture and Crack Growth Data for High Stress Alloys,” vol. 1, dated December 1983 (page 411.257, figure 4.113.104). It represents generic Ti 6-4 Paris fit data. These data are provided for example purposes only and do not constitute a recommendation for analyzing actual components. The R-ratio effect is idealized to illustrate variations caused by FM analytical techniques and may not represent actual material behavior.
(3) Tensile Properties:
Yield = 834 MPa or 121.0 ksi UTS = 910 MPa or 132.0 ksi
3. Test Case Analysis Guidelines. Analytical guidelines fo r probabilistic assessments are provided with the intent to minimize the variations of the manufacturer’s results due to analytical assumptions. The case presented is based on a typical surface anomaly probabilistic fracturemechanics approach. The hole is subdivided into four locations as identified in figure A1-2; the 14
Appendix 1 relative risk or probability of fracture (POF) is calculated for each location. Results for the highest risk location are multiplied by the number of holes and by the surface area per hole to arrive at the total hole set POF or relative risk. This analytical approach can be broken down into five basic steps (described further in paragraphs “a” through “e” below):
a. Stress analysis,
b. Hole set definition including surface area and number of holes,
c. Crack growth model definition,
d. Crack growth calculations, and
e. Relative risk calculation.
Figure A1-2 1 4 3 2
a. Stress analysis. The level of mesh refine ment of the part model is left up to the individual manufacturer’s discretion. However, the manufacturer should take steps to ensure that the final answer does not change by a significant amount (5-percent on relative risk or POF) if a finer mesh is chosen. The limiting operational principal stress is the hoop stress. Component stresses are determined in order to perform crack growth analysis. The disk is assumed to be at constant temperature. There are no thermal stresses. The stress model must include stress concentration at the hole.
NOTE: Typically a Kt would be applied to the rim stress due to the blade slot or a 3D model would include the blade attachment slot. However, it has not been included in the test case.
b. Hole set definition including surface area and number of holes. Hole sets are defined further in Appendix 5.
c. Crack growth model definition. Construct crack growth models for each of the locations defined in figure A1-2.
Appendix 1
(1) Assume a 2:1 crack aspect ratio semi-circular crack for surface cracks, with surface length (2c) equal to twice the depth.
(2) Position one surface crack at the center of the hole at the inner diameter and one at the center of the hole at the outer diameter.
NOTE: In design analysis of engine hardware, position the crack in the lifelimiting location of the entire hole.
(3) Assume a 1:1 crack aspect ratio quarter-circle crack for corner cracks, with surface length (c) equal to depth (a).
(4) The effect of surface enhancement, such as shot-peen, is not considered in the test case analysis.
(5) Position one corner crack at the life-limiting corner of the hole at the inner diameter and another crack at the life-limiting corner of the hole at the outer diameter. See Note in paragraph 3.c.(2) above.
d. Crack growth calculations. Perform cr ack growth calculations using the predicted stresses and crack growth rate data to determine the residual life associated with each location.
Conduct the calculations for a range of initial crack sizes to ensure that the component service life is covered. The following guidelines apply to the test case crack growth calculations:
(1) All anomalies act as sharp propagating cracks (no incubation life assumed).
(2) The cracks are orientated normally to the maximum principal stress at the crack origin.
(3) Consider the impact of stress gradients. The stress direction should remain constant along the stress gradient. Use direct ional stresses normal to the max principal stress plane at the crack origin for the stress gradients.
(4) Use average crack growth data obtained in air (instead of in a vacuum).
(5) No surface enhancement effects.
(6) For surface cracks, consider a surface crack growth correction factor in the stress intensity (K) solution.
(7) For corner cracks, consider a corner crack growth correction factor in the stress intensity (K) solution.
e. Relative risk calculation.
(1) The POF for each location is calculated by integrating the total surface area of all the holes in the feature, anomaly distribution, residual life, and inspection POD (if applicable) from the previous steps (see figure A1-3). The POF of each hole can be calculated by either an 16
Appendix 1 integrated probabilistic method or a “Monte Carlo” method. In the Monte Carlo method, the number of simulations required is related to the computed risk. The general rule is that the number of simulations should be at least 2 orders of magnitude higher than the computed risk.
For example, if risk is 1 failure in 104 parts, the number of samples required is 106. This ensures that about 100 “failed” parts are involved in the assessment. The results for the limiting location are multiplied by the number of holes in the feature to determine the total feature POF.
(2) The disk probability of fracture will be calculated twice assuming: (a) no inservice inspection, and (b) two in-service ECIs; one at 4,000 cycles and another at 8,000 cycles.
The “with inspection” POF calculations are performed in the same manner as “without inspection,” except the inspection POD data and cycles to inspection are included in the risk integration (see figure A1-4). The inspection POD curve applies to 100 percent of the hole and the front and back surface area. Assume that 90 percent of the parts are inspected at 4,000 cycles and an independent 90 percent of the parts are inspected at 8,000 cycles. These inspection intervals are provided for example purposes only. Inspection intervals for actual components may vary.