Investigation of Crack in Gyroplane Main Rotor Blade P E Irving N Smyth September 2010 Rept GS1

School of Applied Sciences 28th September 2010

Investigation of Crack in Gyroplane Main Rotor Blade

Page 1 of 20

BACKGROUND On the 9th of September 2010 Mr. G. Speich, of RotorSport UK visited Cranfield University to deliver part of the main rotor blade assembly of a RotorSport UK Gyroplane. During a pre flight inspection, a crack had been found on the bottom surface of the main rotor blade 5106A. Cranfield University were requested to investigate the failure and determine: 1. The extent of the crack, 2. If the crack originated from the inside or outside surface, 3. Whether an internal rectangular support bar, the termination of which was located close to the centre of the crack, played a role in crack initiation, 4. The length of service that the crack had been present in the structure.

DESCRIPTION OF ROTOR ASSEMBLY The main rotor consists of two blades each manufactured of extruded aluminium alloy 6005. The blades presented to Cranfield University have serial numbers 5106A (visible crack) and 5106B (still assembled). The dimensions of the blades are approximately 0.2m chord length, 0.024m maximum depth, and 4m long. A cross section through a blade is shown in Figure 1. Each blade is connected to the rotor hub assembly via a bolted joint with 9 bolts equidistant apart along a line adjacent to rib B. In the joint the blade extrusion is sandwiched between an aluminium doubler (0.3m long and of varying thickness) and two plates of aluminium (0.01m thick and 0.07m wide) on the top and bottom surfaces, blade 5106B is shown assembled in Figure 2. An aluminium bar is inserted and bonded into the internal space between ribs B and C shown in Figure 1. This extends the entire distance of the bolted connection, terminating at a point outboard of the final hole and approximately coincident with the end of the upper and lower connection plates. Figure 2 shows the disassembled connection area; the dark outline indicated marking the boundary of the clamped area between the lower connection plate and the blade extrusion. The mark outboard of the outermost connection hole is the approximate crack origin.

BLADE MATERIAL The blade was stated to be manufactured of 6005 aluminium. This is a low strength aluminium alloy with a minimum 0.2% proof strength of 240MPa, minimum ultimate tensile strength of 260MPa, and 8% ductility. Based on knowledge that the high cycle fatigue strength of aluminium alloys are approximately 0.3 of the UTS, the plane specimen fatigue strength of the alloy will be approximately ±80-90MPa, at a mean stress of zero. The lower surface of the rotor will experience fatigue loadings on Ground Air Ground (GAG) cycles with a minimum stress of approximately zero. Under these circumstances the fatigue strength for zero to tension loading applying a Goodman correction can be estimated as ±65MPa, with a maximum stress of 130MPa and a minimum of zero.

Investigation of Crack in Gyroplane Main Rotor Blade

Rib A

Page 2 of 20

Rib B

Rib C Rib D

Figure 1: Labelled Cross Section through Rotor Blade

Plates

Doubler

Crack End of Doubler

Figure 2: Main Rotor Blades 5106A (Bottom) and 5106B (Top)

CRACK LOCATIONS The location of the crack on blade 5106A is shown in Figure 3 and was measured at 27mm from the centre of the outermost bolt hole and to be 73mm in length. The blade 5106B was disassembled from the supporting structure and a small crack was revealed, dimensions shown in Figure 4. The crack was located on the underside of the blade in a similar location to the crack on blade 5106A. Both cracks on 5106A and 5106B showed evidence of fretting corrosion on the lower surface, as seen in Figure 3 and Figure 4. The origin process can be seen more clearly on 5106B (Figure 4). There are a number of dark patches caused by mating sites rubbing (local fretting). The crack has initiated from one of these outboard of the outer bolt hole. There are similar marks on 5106A (Figure 3) though the details of the origin are obscured by fretting product. The blades were visibly observed to have a slight bend upwards. The location of the crack on both blades relative to the doubler/plate end and internal reinforcing bar is shown in Figure 5. The cracks on both rotors were located inboard of the end of the doubler/plate and reinforcing bar.

Investigation of Crack in Gyroplane Main Rotor Blade

Page 3 of 20

The fretting corrosion residue (dark region on Figure 3) was only found on the outboard side of the crack. This may be because of the centrifugal force from the rotating rotor conveying the residue in the outboard direction only. A similar observation can be made on the internal side of the crack (Figure 6).

INBOARD

27mm

73mm

Figure 3: Crack Dimensions on Blade 5106A

INBOARD

6mm 23mm

Figure 4: Crack Dimensions on Blade 5106B

Investigation of Crack in Gyroplane Main Rotor Blade

Page 4 of 20

Upper and Lower Skin on Doubler

Internal Reinforcing Bar

Position of Crack Upper & Lower Attachment Plate

Upper and Lower Skin of Airfoil

Figure 5: Cross Section through Blade Assembly

Figure 6: Inside Surface of Blade 5106A after Cutting

The blade was cut further and pulled apart to reveal the fatigue fracture face, a photograph of which is shown in Figure 7. There was further evidence of fretting corrosion on the face. During the crack separation procedure new fractures were created, the ends of the original fatigue fracture region are highlighted in Figure 7 by means of white dotted lines. A representative SEM image is shown in Figure 8 clearly showing fatigue striations. The striations appear to be grouped into bands and it is conceived these bands may represent landing/take-off cycles with the intermediate striations representing in flight manoeuvre loads. The varying distance between the bands represent the varying durations of the flights.

Investigation of Crack in Gyroplane Main Rotor Blade

Page 5 of 20

Initiation Region

Figure 7: Fatigue Fracture Face of Blade 5106A

Manoeuvre Loads

Landing/Take-Off

Figure 8: Fatigue Striations on Fracture Face of Blade 5106A

Measured striation spacings for the GAG (indicating growth increments per flight) at three locations on the fatigue fracture face are given in Figure 9, Figure 10, and Figure 11. Figure 9 is measured 25mm from the initiation point, Figure 10 is 8mm, and Figure 11 is 4mm. The average GAG spacing at the three locations is 46µm, 13µm, and 31µm respectively.

Investigation of Crack in Gyroplane Main Rotor Blade

Figure 9: GAG Spacings 25 mm from Initiation Region

Page 6 of 20

Investigation of Crack in Gyroplane Main Rotor Blade

Figure 10: GAG Spacings 8 mm from Initiation Region

Figure 11: GAG Spacing’s 4 mm from Initiation Region

Page 7 of 20

Investigation of Crack in Gyroplane Main Rotor Blade

Page 8 of 20

DISCUSSION OF OBSERVATIONS Initiation Site and Conditions The crack in both the A and B blades has initiated at locations of fretting damage at almost identical positions, 20-30mm outboard of the outermost hole centre, approximately along an extension of the line of the holes and 4-5mm inboard of the termination of the inner aluminium rectangular bar. The crack origin is within the area clamped by the outer aluminium doubler and is therefore subject to compressive forces at the location point. From the origin the crack spread down through the outer skin and outwards in both directions towards the leading edge and the trailing edges and down into the ribs within the extrusion. The initiation process can be seen at an earlier stage in blade B (Figure 4) where there are a number of local fretting sites one of which has initiated a fatigue crack. Fretting conditions will reduce the material high cycle fatigue strength by a factor of 3, making initiation at these locations extremely probable. Role of Reinforcing Bar and Bending Overload in Crack Initiation The observation that the blades were slightly bent upwards, suggests that they may have been overloaded during service creating local plasticity. As 6005 is a very ductile alloy (minimum of 10% elongation to failure), it is unlikely that bending overload producing 2-3% plasticity would directly cause cracking. Strains in excess of 10% would promote local cracking, but there was no sign of deformations of this extent at the crack locations. The internal reinforcing bar may have influenced the load transfer from doubler to blade. A redesigned blade without a bar is reported to have failed via a crack in the outermost hole. This would be the expected site in the absence of an internal reinforcing bar as stresses will be at their greatest at the outermost hole edge. In the blade under investigation the internal bar may have acted to transfer local loads away from the hole and into the blade skin outboard of the bolt hole; leading to enhanced stresses at the observed crack location. During the crack extraction process it was observed that the bar/blade bonding was locally delaminated at the crack site. This will have reduced the effectiveness of the bar/blade load transfer making the region between the outermost hole and the bar end a region of enhanced stress, rather than the bar end itself. The role of the bend overload may have been to change the contact conditions making fretting more likely. When fretting fatigue conditions exist, the fatigue strength could be reduced from the ± 65MPa suggested earlier for normal fatigue, to the order of ±20MPa in aluminium alloys in tension. The stresses associated with the overloads will promote additional fatigue damage but they will not cause cracking directly. Fractographic Observations Observations on the fracture surface did not suggest any material defect which could have promoted the crack initiation. Indeed the observation of cracks in both blades and in other rotors suggests that the failure origin is not in material anomalies, but is to do with either design or operation of the rotor. The probable identification of GAG growth increments on the fracture surface allows very rough estimates to be made of how long the fatigue crack has existed. At the

Investigation of Crack in Gyroplane Main Rotor Blade

Page 9 of 20

largest increments observed of 50µm/GAG, growth will be at 1mm every 20 flights. Taking a half crack length (crack is growing both ways and has 2 fronts) of 36mm, this suggests that the crack has existed for at least 700 flights. Service load measurements The initial service load measurements presented in Appendix A and interpreted in Appendix B are very limited and brief but they do allow initial calculations to be made of the likely fatigue and fatigue crack growth behaviour of the blade under the measured service loads. These are described in Appendix C. It is unlikely that the stresses measured fully reflect the extremes of loading placed on the failed blades, as a maximum stress of 179MPa has been recorded. This is insufficient to cause plasticity and permanently bend the blade. The 6005 alloy has a specified minimum proof strength of 240MPa, possibly the blade strength level is greater still, and stresses larger than this will be required to cause plastic deformation. Nevertheless, the recorded fatigue cyclic stresses are sufficient to cause very early crack initiation in fretting. Data developed in Cranfield fatigue laboratories suggest that only a few hundred cycles of stresses of this range would be necessary to create a fatigue crack in aluminium alloys under fretting conditions. Use of the service load measurements to calculate crack growth life using AFGROW Appendix C describes fatigue crack growth life calculations made using a conservative flight service loading spectrum constructed using the supplied service loading information as input to the software fatigue crack growth calculation package AFGROW. The constructed spectrum assumed that each flight consisted of 60 repeats of the worst case manoeuvre loading recorded in the data supplied. Material crack growth data for 6061 aluminium sourced from the AFGROW database was used in the analysis. The results show that under the conditions assumed, the number of flights to grow the crack from 1mm starting half crack size up to a crack length of 72mm total length is approximately 297 flights. The analysis was repeated using different sets of assumptions about the material data, this resulted in a life of 700 flights. There are both conservative and non conservative aspects to this calculation. On the one hand, the constructed spectrum represents a severe worst case load history derived from the data supplied. On the other hand, it is unlikely that this limited data set fully represents the extremes of the possible loads which as noted above can be so severe as to cause blade bending. The best course of action would be to derive a spectrum fully representing rotor stresses on a comprehensive service loads measurement data set. This would enable an accurate calculation which could be given with full confidence. The present analysis can only be regarded as preliminary.

Investigation of Crack in Gyroplane Main Rotor Blade

Page 10 of 20

CONCLUSIONS 1. Visual and scanning microscope observations of fatigue cracks propagating in both blades of a gyroplane rotor suggest that they originated in fretting fatigue at locations in the external surface within the clamped area of the doubler connecting the blade to the hub. The blade was bent suggesting it had been overloaded; this may have promoted fretting fatigue conditions. 2. The crack subsequently propagated transversely towards both the leading and trailing edges and down through the ribs within the extrusion. 3. Fractographic observations suggest that GAG growth increments can be identified. These have a maximum spacing of approximately 50µm and suggest therefore that 20 flights are required to grow the crack 1mm. The blade with the longest 73mm crack had therefore approximately been cracked for at least 700 flights. 4. Preliminary service load information used as input into a fracture mechanics analysis of fatigue crack growth behaviour shows that service life to grow the crack from 1 mm half crack length to 36mm final half crack length is between 300 and 700 flights.

Investigation of Crack in Gyroplane Main Rotor Blade

Page 11 of 20

APPENDIX A – RotorSport UK Service Strain Measurement Report

Investigation of Crack in Gyroplane Main Rotor Blade

Page 12 of 20

Investigation of Crack in Gyroplane Main Rotor Blade

Page 13 of 20

Investigation of Crack in Gyroplane Main Rotor Blade

Page 14 of 20

Investigation of Crack in Gyroplane Main Rotor Blade

Page 15 of 20

Investigation of Crack in Gyroplane Main Rotor Blade

Page 16 of 20

Investigation of Crack in Gyroplane Main Rotor Blade

Page 17 of 20

Investigation of Crack in Gyroplane Main Rotor Blade

Page 18 of 20

APPENDIX B – RotorSport UK Service Strain Measurement Interpretation The interpretation of the service strain measurement report is detailed here. It is stated in the report (see Appendix A) that “the strain gauges are calibrated in microns per m” which was assumed to be micro-strain. It is also stated that the blades were “fitted with GS3BT strain gauges, one on the lower surface of the blade adjacent to the blade at the end of the hub bar...and one outboard approx 1m from the end of the hub bar to measure the centripetal force in the blade”. However it should be noted that using only one strain gauge on the lower surface it is not possible to differentiate between bending and tension induced strain. However for the purposes of the current analysis a combined tension/bending strain measurement is sufficient. For each of the graphs of the RotorSport UK service strain measurement report the maximum and minimum strains were noted for both the blue and green lines. The report mentioned that “the scaling factor within the system are such that the blue line MUST be read double scale”, therefore strains were doubled for the blue line. To convert strain to stress, the measured micro-strain was multiplied by the Young’s Modulus (69GPa). The noted strains and resulting stresses are shown in Table 1. Table 1: Calculated Stresses from Strain Data

Graph Page No. 2 3, 5, 7 4 6

Line Blue Green Blue Green Blue Green Blue Green

Strain (µε) Min Max 1200-1600 650-750 1200-2150 625-925 1150-2600 600-1075 950-2150 575-900

Stress (MPa) Min Max 83-110 45-52 83-148 43-64 79-179 41-74 65-148 40-62

The greatest minimum and maximum stresses at the location of the crack (blue line) were 65 and 179MPa respectively.

Investigation of Crack in Gyroplane Main Rotor Blade

Page 19 of 20

APPENDIX C – AVGROW ANALYSIS AFGROW is a software package which simulates fatigue crack development in metallic materials using fracture mechanics approaches to fatigue crack growth. It was developed by the US Airforce over a 10 year period to solve fatigue crack growth problems on military aircraft and is publicly available. AFGROW contains a database of material fatigue crack growth properties necessary for performing the fracture mechanics calculations. The alloy 6005 used for the rotor blade is not available in the database, but the closely related 6061 alloy is in the data base. It is unlikely that there will be significant differences in the crack growth characteristics of 6061 and 6005 as they are of similar strength and similar aluminium alloy series. Apart from the material fatigue crack growth properties and the static fracture toughness of the material, also in the data base, the other data required for the crack growth prediction are the stress spectrum experienced by the blade in service. These data were extracted from the preliminary data contained in Appendix A, and a simple spectrum constructed containing two components, based on the worst case stresses at the maxima and minima for the Ground Air Ground cycles and the worst case stresses associated with the manoeuvres. These should give a conservative calculated number of flights to static fracture. The spectrum derived from the worst case data in Appendix B is shown in Figure 12 below. It was assumed that each flight consisted after takeoff of 60 worst case manoeuvres in a 1 hour flight, followed by landing. In the analysis, this spectrum was repeated until failure of the rotor was predicted. The final item of data required for the analysis is a starting crack size. This is the crack size at the end of the initiation stage of the failure. In fretting fatigue we believe that the initiation stage took a very small number of cycles at the GAG stress level of zero – 179 MPa.- perhaps only a few hundred cycles. This would create a crack of around 1 mm size. To perform the fracture mechanics calculation of life it was assumed that the blade lower surface could be approximated to a plate 200 mm wide, 1.4 mm thick and of length in excess of 400 mm. The analysis was repeated using different assumptions about material properties and the form of the crack growth data. The result of the analysis showing a plot of cycles Vs Crack length describing the crack growth process is shown in Figure 13 below.

Investigation of Crack in Gyroplane Main Rotor Blade

Figure 12: Stress Spectrum used in AVGROW Analysis

Figure 13: Life Vs Crack Length

Page 20 of 20