Helicopter Lift Frame Test Specimen
Sample Applications
Helicopter Lift Frame Test Specimen
Crack growth in a helicopter lift frame test specimen
This example is based on the specification provided for a round-robin crack growth problem proposed at a workshop on fatigue design of helicopters in Pisa, Italy in September 2002. The problem was devised by R.Cansdale and B.Perrett of QinetiQ and P.E.Irving of Cranfield University. The purpose of the round-robin was to investigate the accuracy of different damage tolerance analysis methods and software in determining the crack growth time for a defect in a typical helicopter dynamic component. The crack was grown from an initial detectable size of 2mm to 25mm under the action of a typical helicopter load spectrum. The round-robin provided experimental results for comparison of the analytical methods.
Although Zencrack was not used as part of the formal round-robin process, the analysis was subsequently undertaken as a benchmark without knowledge of the experimental results. The original Zencrack analyses were conducted with Zencrack 7.1 and Abaqus 6.3. Some features of a revised Zencrack analysis, conducted with Zencrack 7.8-1 and Abaqus 6.12-1, are shown here.
The specimen is aluminium alloy 7010 T73651. Raw crack growth data was provided for a number of stress ratios between 0.1 and 0.9. The peak load was defined to be a load giving a net section stress of 130MPa in the symmetry plane. The specimen was loaded by the ASTRIX load spectrum. This is a standard spectrum which contains 743240 load points.
The geometry for the problem is a flanged plate with a central lightening hole. This is typical of many features in a helicopter lift frame. The test specimen is symmetric and only a half model is required for the analysis. The geometry half model is shown in Figure 1. The initial crack is a 2mm corner crack in the symmetry plane and located at the inner edge of the hole (see Figure 5 for clarification). To model the crack, the geometry is partitioned into two regions; tets for the bulk of the model and hex elements near the crack, as shown in Figure 2. The uncracked mesh is shown in Figure 3.
Figure 1 - Geometry half model cut along the symmetry plane that contains the crack
Figure 2 - Geometry with partition to define tet and hex regions
Figure 3 - Uncracked finite element mesh - mainly tets with hex elements near the crack region
Due to the geometry and the region in which crack growth is to be conducted, the analysis is carried out two stages. The first stage is the growth of the crack in the corner crack phase and through the thickened square section near the hole. A close-up of the uncracked mesh for this part of the analysis is shown in Figure 4. For this phase, several standard crack-blocks are lined up to create the desired initial crack. The initial cracked mesh is shown in Figure 5. Due to the reduction in crack length as the crack grows out of the thickened section, a second analysis is carried out for the growth through the thin section of the plate. The same uncracked mesh is used with a starter crack position and position within the load spectrum based on the results of the first phase of the analysis. The calculated profiles for both phases of the analysis are shown in orange and red in Figure 6. An animation of the crack growth for both analysis phases is shown in Figure 7.
Figure 4 - Uncracked finite element mesh close to the crack region
Figure 5 - Cracked finite element mesh for the initial crack
Figure 6 - Calculated crack profiles
The Zencrack results for crack growth are shown in Figure 8. Crack size in this plot is defined as distance from the bore of the hole to the end of the crack on the top surface of the specimen. The 25mm crack length was reached in approximately 160 flight hours compared to around 410 flight hours for the experimental test (whose data is not included here). Zencrack reached a 5mm crack size in about 113 flight hours compared to around 180 hours from the test data.
Investigation into the reasons for these differences led to some questions over the load levels that were applied in the tests and whether the quoted stress figure of 130MPa was correct. A number of sensitivity analyses were conducted for the corner growth phase of the analysis. The results of these studies (from the original runs with Zencrack 7.1) are shown in Figure 9. It is most notable that a decrease in load level of 10% increases the time for this phase of the growth by 76%. This type of sensitivity analysis is readily performed in Zencrack - for example the load level modification requires change of just a single number in the analysis input file.
It is clear that the overall accuracy of the life prediction depends heavily on the initial corner phase. In addition to any issue over the accuracy of the load levels, it is believed that additional crack growth data close to the threshold region would have had a significant impact on the results of the analysis.
Figure 8 - Crack growth vs cycles
Figure 9 - Effect of changing input parameters on the duration of the corner crack phase
References
The Helicopter Damage Tolerance Round-Robin Challenge, R.Cansdale, B.Perrett, QinetiQ, UK, Workshop on Fatigue Design of Helicopters, University of Pisa, Sep 12-13 2002.
Life Predictions for High Cycle Dynamic Components Using Damage Tolerance and Small Threshold Cracks, Robert E. Vaughan, Jung Hua Chang, Structures and Materials Division, US Army, AMRDEC, Redstone Arsenal, AL 35898, USA American Helicopter Society 59th Annual Forum, Phoenix, Arizona, May 6-8 2003.