Fatigue and Fatigue Design

Cessna—"As a result of a recent fatal accident caused by fatigue failure of the front wing spar . . . " The majority of these AD's involved parts which had.
2MB taille 11 téléchargements 143 vues
Fatigue And Fatigue Design By E. L. Turner, EAA 3648 Aero Design Evaluation Engineer (Helicopter Specialist) Federal Aviation Agency, Region U Fort Worth, Texas he FAA encourages the amateur-built program genT erally; however, the fatality rate of homebuilt rotorcraft is alarmingly high—a point that cannot be overstressed. Since the FAA function in aviation is to insure safety along with promoting aviation, they have authorized this paper to be presented at the 9th Annual EAA Fly-in, in hopes that the presentation will help enable the homebuilt rotorcraft enthusiast to have a better understanding of one of the biggest problem areas in rotorcraft design—FATIGUE . . . the phenomena which takes place in materials causing deterioration and failure after a repetition of stress. The Airworthiness Directive, or AD as it is commonly referred to in the industry, is of no concern to the amateur builder. Generally, the homebuilt aircraft receives more attention or maintenance than the average standard category aircraft, thereby design discrepancies or system malfunctions are found and corrected before an accident or mishap occurs. Unfortunately, this "tender care" and accident preventive maintenance for "homebuilts" does not always apply to the amateur-built helicopters and autogyros. Perhaps the following excerpts from a few AD's will supply a clue as to why rotorcraft maintenance cannot, and does not, help the homebuilder from running into catastrophic accidents with their machines. These excerpts list only the make of helicopter and the reason for issuance of the AD: Vertol—"Fatigue failure of the lower directional bellcrank assembly". Hiller—"Fatigue cracking of metal main rotor blade". Sikorsky—"Fatigue failure of the main rotor blade spar". Sikorsky—"Fatigue cracks have been found in the area of the rear three bolt holes of the upper pylon folding hinge fitting". Brantly—"A fatigue crack has been found around the forward end of the weld joining the tail rotor guard to the sheet metal bracket at the upper tail rotor gear box". Sikorsky—"A fatigue failure of the RE5M7 shank rod end bearing has occurred in the S-55 main rotor upper controls". Bell—"As a result of several cases where fatigue cracks occurred in the tail rotor blades". In each of these AD's, the subject contains a common word—FATIGUE. Fatigue is the unpredictable factor in detail design of aircraft parts. From the above excerpts, one will note that fatigue problems plague the manufacturers of the nation's leading helicopters. It should also be noted that fixed wing aircraft are subjected to fatigue failures as well as rotorcraft. The following excerpts from three recent AD's will show this: Piper—"Fatigue originating at the AN7 attach bolt of the front spar hinge fitting". Aero Design—"Fatigue crack in engine mounts . . . " Cessna—"As a result of a recent fatal accident caused by fatigue failure of the front wing spar . . . " The majority of these AD's involved parts which had completed the FAA required fatigue evaluation testing for rotor and control parts. To combat fatigue failures the

helicopter manufacturers employ experienced design engineers, incorporate the best production methods and materials, and utilize the latest fatigue testing facilities in the development of their products. After the product has entered operational service, the manufacturer establishes procedures and time limits in which all moving parts are completely dismantled and thoroughly inspected by X-ray, Sonic and Ultrasonics, Magnetic Particle (Magnaflux), and liquid penetrant (Zyglo and Dy-Chek) methods. Usually, a fatigue crack has to be well formed before it can be detected with the naked eye. With this information known, it is only natural for the homebuilt rotorcraft enthusiast to ask, "Can I possibly hope to build a safe and successful helicopter or auto-gyro"? The answer to this question — The enthusiast should have knowledge on: 1. A brief history and theory of fatigue. 2. Design of structures to minimize the possibilities of fatigue failures. 3. Methods used for determining service life of fatigue critical parts. This paper will cover these three subjects more from the practical approach than from the theoretical aspects. A Brief History and Theory of Fatigue

A century ago, structural design was based almost entirely on concepts of static strength of materials. At that time, there were relatively few sources of vibration or repeated stressing in comparison with those existing today. Sources of motive power were limited, speeds were low, and many structural parts were designed with large safety factors. Under such circumstances, design on the static-strength properties was quite satisfactory. During the first half of the 19th century, development of the steam engine led to increasing sources of repeated stress on metal parts and structural elements. Shortly thereafter, "unexplainable" fractures—particularly in locomotive axles—became of great concern to engineers. Despite the fact that the axles were made of ductile iron, they were found to crack in an apparently brittle manner after varying periods of time in service. The term "fatigue" was attached to such failures in view of the observation that they usually occurred only after a considerable length of service. About the middle of the century experiments revealed the importance of the number of repetitions of stresses, rather than the duration of time, in causing failures of metals under relatively low stresses. Thus, the fatigue fractures of locomotive axles were explained in terms of the "repetition of stressing", and the term "fatigue failure" came to be associated with fractures from repeated stressing. (I have with me several parts which have failed from fatigue during service life substantiation tests. These parts will be circulated among you in order that you may see samples of actual fatigue fractures. These specimens were loaned to me through the courtesy of Bell Helicopter Company and Brantly Helicopter Corp.) From the appearcontinnp.d on next, page SPORT



FATIGUE AND FATIGUE DESIGN . . . continued from preceding page ance of fatigue fractures, it seems reasonable to consider the progress of fatigue as occurring in three stages: 1—The start of a crack, spreading out from some nucleus. 2—Propagation of the crack under successive cycles of loading. 3—Final rupture of the piece when the spreading crack has weakened the section. In efforts to derive a theory of fatigue, most attention has been devoted to the first state—the start of the crack. Moreover, attention has been focused on the primary question, "How can a crack start under repeated stress at levels lower than that stress needed for static rupture"? Almost every line of experimental study emphasizes the localized nature of the formation of a fatigue crack. It appears that, even though material near the nucleus of the crack (and along its path of propagation) may be highly distorted, material away from this region show very little visible damage. Accordingly, it is hardly to be expected that a simple model of metallic structure will afford an accurate representation of the phenomenon of fatigue failure However, facts which emerge from numerous studies provide clues to the basic nature of fatigue. Some observations which appear widely include the following: A—Fatigue failure is shown by most metals and metallic alloys, by some plastics, by woods and plywood, and by other materials that exhibit some ductility in static tests. B—Metallic fatigue fracture ordinarily depends on the number of repetitions of a given range of stress rather than upon the total time under load. Speed (at least from 100 cps to 10,000 cps) has almost no observable effect on the fatigue strength of a metal, except under special conditions of environment. C—Some metals (notably ferrous alloys) have "safe" ranges of stress. Below some stress amplitude, called the "fatigue limit", failure does not occur even after a very large number of cycles of repetition of stress. Other alloys do not seem to have such limiting stress ranges. D—For most materials, notches, grooves, or other discontinuities of section decrease greatly the average stress amplitude that can be withstood for a fixed number of cycles of stressing. E—The range of stress necessary to produce failure in a fixed number of cycles usually decreases as the mean tension stress of the loading cycle is increased. F—It appears that a fatigue crack, once started, follows a path of least resistance through the metal. A point of interest is that fatigue cracks tend not only to start at faults and inclusions in a metal but often will propagate along lines from one fault, or inclusion, to another. Design of Structures to Minimize the Possibilities of Fatigue Failures

The rotorcraft is perhaps more directly associated with fatigue failures than any other type of aircraft. The primary structural elements and systems are subject to large vibratory stresses in practically every regime of flight. In addition, being a highly maneuverable aircraft that is capable of forward, rearward, sideward, vertical and rotational flight, operating limitations due to fatigue are required in practically all flight situations. For these reasons, it is important that special attention be focused on the fatigue strength evaluation of design of detail parts and components. 16



Frequently, one will hear that fatigue failures are chargeable to "poor detail design" without any accompanying explanation as to what constitute poor detail design. The inference here is that a structure "properly designed" should not fail in fatigue any time. If one considers the fact a helicopter represents the assembly of thousands of parts, all of which are riveted, bolted, welded, or bonded together, it should become evident that it will be highly unlikely that no error in design judgment, or fabrication will take place. It is also highly unlikely that even the most careful workmen will assemble a structure without leaving a toolmark, nick, scratch, or gouge somewhere in the assembly, nor is it probable that the structure under normal operation and maintenance will not accumulate additional damage of one sort or another, all of which is detrimental to the fatigue life of the rotorcraft. A structure designed for static strength is usually reasonably complete and is based on certain clearly specified requirements. In helicopters, many of these requirements are empirical but are seasoned by years of experience. The methods of analysis are also in many cases empirical but are well substantiated by full-scale test components on the complex structure. In designing for fatigue, however, the requirements are not so clearly specified nor are the data available always directly applicable to the actual problems under consideration. What test data is available is mostly on small-scale laboratory specimens with stress concentration not always geometrically similar to those encountered in final design. Therefore, a direct transfer of knowledge available from laboratory data to full-scale design rarely is advisable. One should not be surprised to find that a fatigue failure occurs in a full-scale composite structure at about one-half to one-tenth the lifetime when compared to laboratory test data on small specimens. Designing for fatigue is not a matter of comparing "allowables" as is done in a structural design for static strength. It is a matter of securing guidance from test data to assist in making intelligent estimates for the lifetime of the part. Wherever possible, fatigue testing on the full-scale component or detail part is advisable. In the design and fabrication of details parts, the homebuilder should pay careful attention to small details such as sharp edges, small radii for fillets, fillets not tangent to the adjacent surfaces, mismatched surfaces, intersection of drilled, tapped and reamed holes, spotface in fillets leaving sharp edges and intersections, abrupt changes in section properties, angles and fittings with insufficient edge distance and bearing contact area to transfer the applied loads, and rounding of all sharp edges especially around undercuts and threads. In service fatigue cracks can, and will, start from small sharp edges, fillets, burrs, scratches in bolt holes, dimpled and countersunk rivet holes, tool marks, and machine marks. Structures in which a high degree of redundancy exists have a greater chance of surviving a fatigue failure without catastrophic results than structures with much lesser degree of redundancy; for instance, failure in the tension side of a single cell rotor blade would result in complete loss of the bending and centrifugal strength of the blade while a similar failure in a multicell rotor blade may only result in partial loss of bending and centrifugal strength. Where the design is such that oscillating leads act normal to the grain of the material (such as control rods to the swashplate), the swashplate should be designed for very low stress amplitudes. Although it is usually impossible to determine the magnitude of residual stresses in a member, they are known to exist to various degrees. In areas where fatigue

damage can be critical, tests should be made on duplicate full scale parts. Helicopter manufacturers generally test a minimum of four specimens. In cases where such tests are not possible, a high degree of conservation is the only substitute. For the homebuilder this statement should be repeated: In cases

where such tests are not possible, a high degree of conservation is the only substitute!

Where it is difficult to reduce stress-concentration

factors in main structural members and fittings and where the regions of abrupt changes in area occur, the surface should be reasonably well polished and all corners carefully dressed. Stress concentrations, and stress raisers, are the names which have been applied to surface irregularities such as fillets, grooves, holes, rivets, or sharp edges where fatigue cracks usually start. It may be shown, from the theory of elasticity, that the peak stress near

such a change in section is higher than the average or nominal stress in the surrounding neighborhood. A good example of a stress concentration factor is that of a round hole in a tension member: The increased stress around the hole can be as high as three times that of the nominal stress. Notch-sensitive materials should be avoided in tension members. When used, the design values for fatigue should be considered carefully. Again, full-scale tests are most desirable in this case. As a summary on design of fatigue critical parts, it would be sufficient to state: Without adequate fatigue testing facilities, the amateur cannot be sure of his rotorcraft design even though the basic configuration is sound. Methods Used for Determining Service Life of Fatigue Critical Parts

To determine the service life of a fatigue critical part requires an extensive program involving the knowledge of stresses and associated flight maneuvers to be expected in normal operation, knowledge of the frequency of occurrence of specific loadings and knowledge

of the fatigue strength characteristics of the structure.

These requirements necessitate a survey of components

that are expected to be fatigue critical, flight stress measurements, and testing of at least four specimens per fatigue critical part. This program is, of course, lengthy and expensive since each phase requires special instrumentation, testing facilities, and destruction of many test specimens. Although a uniform approach to rotor fatigue problems is desirable, it is recognized that in such a relatively new field, new design features, methods of fabrication or configuration may require variations and deviations from the methods now used. Engineering judgment should therefore be exercised in each case. Although there is some question as to whether a completely rational method exists for the prediction of the fatigue life of a builtup structure subject to random loading, it is believed that an engineering approach to the subject can be attained through the application of the Cumulative Damage Hypothesis. This hypothesis asserts that every cycle of stress above an "endurance limit" produces damage proportional to the ratio of cycles run at that stress to the fatigue life at that stress level. Laboratory tests of this hypothesis indicate that it is reasonably valid when the stress cycles are at random magnitude. Despite the approximations involved in the hypothesis and the lack of an adequate theory connecting the hypothesis with more basic properties of materials, it attempts to take more factors into account than any other method developed so far.

It is generally agreed that because of the approximations employed in rotor load and stress distribution analysis, it is not possible at present to determine analytically a reasonable approach to rotor fatigue stress levels. Rotor stress levels are therefore determined by means of carefully controlled, instrumented flight strain gauge testing. These tests are aimed at the determination of the magnitude of steady and oscillatory stresses associated with normal helicopter operations and the correlation of the occurrence of critical stresses with specific maneuvers. In other cases the information obtained can be of use in setting up a test program which would determine the fatigue life of the part. Prior to conducting a flight strain gauge testing program, some rational evaluation of the critical stress areas must be made in order to determine the proper distribution of gauges. A qualitative study is usually made by means of brittle coatings (such as Stresscoat), by photoelastic methods, or by random placement of gau;ges and cross plotting the results of several runs. In conducting flight strain measurements, besides the proper distribution of strain gauges on hubs, main rotor blades, blade attachments, directional control system, control members, provision is usually made for recording the collective pitch setting of the main rotor blades and the center of gravity acceleration during maneuvers. This is done so that it can be ascertained that for maneuvers in which a rapid control movement is utilized the severity of application of control is representative of that which can be encountered during actual service operation. Table 1, Appendix A, of CAM 6, contains a suggested list of flight maneuvers for investigation in a flight strain survey. This list also includes the total operating percentage of the total flight time associated with each flight maneuver. The operating percentage time is usually referred to as frequency of occurrence. Briefly, Table 1 gives the percent occurrence of the flight maneuvers expected during a normal flight from ground conditions through power-off autorotation. Following are a few of the percentages given in Table 1: 1.

Ground Conditions:

a) Rapid increase of rpm on ground to quickly engage clutch . . . . . . . . . . . . . . . . 2.


a) Steady hovering . . . . . . . . . . . . . . . . . . . . . d) Rudder reversal . . . . . . . . . . . . . . . . . . . . . . . . 1% 3.

Forward Flight Power on:

a) Level flight—20% Vne . . . . . . . . . . . . . . . . . . 5 %

d) Level flight—S0% Vne . . . . . . . . . . . . . . . . . 18%

m) Landing approach . . . . . . . . . . . . . . . . . . . . . . . 3 % 4.

Autorotation—Power off:

g) Cyclic and collective pull-ups . . . . . . . . . . . . 2 % h) Landing (Including Flares) . . . . . . . . . . . . . 2 V z %

The occurrence per flight, of course, totals 100 percent. The maneuvers are usually investigated from 95 percent of minimum power-off operational rpm to 105 percent of the maximum power on operational rpm, as well as the complete altitude, center of gravity and weight ranges up to 111 percent of Vne. (Never exceed speed). Determining the frequency of occurrence of total operating time associated with each flight maneuver is considered the second step in determining the service life. The third phase involves the determination of the fatigue strength of the actual structure. This can be accomplished by one of three ways: 1. Laboratory tests, which has been mentioned earlier. continued on next page SPORT



Education Thru Error CLEAN CAN BE MEAN

By Dick Forrest, EAA 956

here have been several fatal accidents in the recent T The causes of these accidents have not been the same

the maneuverability of the ship to his new passenger. Anyone who has had the pleasure of flying a Tailwind or even riding as a passenger knows that the control stick does not have to be moved very far, fast or hard to get an immediate, rapid and clean response from the airplane. Such control sensitivity can be dangerous if it is not handled properly. Abrupt or large movements at high speeds can run the G loads up to dangerous levels rapidly. Weather was the predominant factor in a Cougar accident recently, although the high speed characteristics of the aircraft no doubt complicated the matter. On the builder-pilot's first flight in his Cougar, he took off in the evening and was next seen coming out of the overcast 500 feet above the floor of a canyon. The aircraft was inverted and attempting a roll-out. The plane crashed into the trees on the canyon floor and the pilot was killed. Unfamiliarity with the aircraft and probably vertigo in the overcast contributed to this crash. The aircraft did not break up in the air. All the foregoing accidents are directly attributable to pilot error; the diving Cougar pushed beyond its structural limits by the pilot, the Tailwind also structurally pushed over the edge, and the last Cougar flown into IFR conditions by a VFR pilot unfamiliar with the aircraft. Clean, high speed homebuilts are here to stay. The Tailwinds, Cougars, racers and their like appeal to a great many homebuilders. These builder-pilots must always remember that they are flying aircraft in which it is not difficult to exceed the design limitations. In many cases of homebuilt aircraft design, the actual design limits are purely theoretical — or in simpler words, guesswork. It is the responsibility of the builder or owner, if he loans his ship to someone, to make sure that the other pilot understands the flight characteristics and the aircraft limits that should be applied to that particular plane. Any new builder, with Cub or Aeronca time only, should be doubly careful in his first hours of flight with an unfamiliar clean high-speed plane. Granted that 200 mph is no astounding speed in this day of missiles and cosmonauts but it is a long step from a Cub in terms of possible control overloads.

FATIGUE AND FATIGUE DESIGN . . . continued from page 17 2. Flight testing of the entire helicopter under very controlled conditions. 3. Whirl stand testing of complete rotor components. Of the three methods, laboratory tests are recommended because of the greater degree of control which can be maintained. The laboratory method requires a minimum of four test specimens to establish a mean oscillatory stress versus number of cycles curve (usually referred to as an S-N curve) for the range of oscillatory stress levels expected to occur in service. In order to compensate for scatter associated with fatigue testing, the S-N curve should be reduced by a factor of at least 20 percent. A separate S-N curve should be established for each critical steady stress level determined in the flight strain survey. From the flight stress measurements, the frequency

of occurrence of the flight maneuvers and the S-N curves, the fatigue life of the part can be calculated. An example of fatigue life determination from S-N data is given in Appendix "A" of CAM 6, therefore, no further discussion will be made here as the calculations and explanation of fatigue life determination is complex and lengthy.

past involving clean, high-speed homebuilt aircraft.

but they are basically alike due to a feature inherent in the design of such aircraft. By lightplane standards, ships like the Tailwind and Cougar are clean aerodynamically. They are fast in cruising flight and when the nose goes down into even a shallow dive, the airspeed needle wastes no time in winding up. These higher speeds put greater strains on the airplanes; any sudden or snappy control movement made to change the direction of flight increases the G load rapidly to a point where the design limits of the aircraft can be exceeded. In a clean ship with a stall speed of 50 mph a maneuver abruptly started out of cruise flight at 140 or 150 mph can pull 9 G's or more on the airframe. Various sections of the aircraft may decide to carry on alone because it is easier and gravity takes over. Excessive dive speeds were involved in an accident with a Cougar. This particular plane had been dived to high speeds many times by its builder-pilot during its 50-odd hours of flight. According to the pilot's friends, he seemed "fascinated by high speed dives and enjoyed doing them". He had been cautioned by other pilots in his area that faster and faster dives could only lead to trouble. Apparently in the fatal dive at a speed in excess of 200 mph the windshield collapsed. This led to a loss of control of the aircraft causing it to exceed the design limits and lose a wing. Although the pilot wore a parachute, his safety harness was still buckled when his body was found in the demolished plane. Again excessive speeds and control movements were involved in a Tailwind disaster. Although there were apparently no witnesses to the actual breakup of the plane in the air, the FAA investigation of this accident has given the plane a clean bill of health. The investigation indicates that the right stabilizer collapsed because of downward loads exceeding the design limitations, followed by the upward collapse of the right wing outboard of the strut connection. There were unfortunately two people involved, the non-owner pilot and a passenger. No one knows just what caused the excessive load but it is reasonable to assume that the pilot was demonstrating





It should be evident, at this point, that the problems of designing against fatigue are not straightforward and simple. No prescribed formulas, theoretical or empirical, will provide a solution that may be considered reliable. The best that can be hoped for is to fortify the "homebuilt" rotorcraft enthusiast with proper guidance by developing a comprehensive knowledge of fatigue. It is indeed unfortunate that such a vast and complex problem as fatigue stands in the way of safe and successful amateur-built rotorcraft. A