PSI - Issue 58

J.R. Steengaard et al. / Procedia Structural Integrity 58 (2024) 61–67 J.R. Steengaard et al. / Structural Integrity Procedia 00 (2024) 000–000

64 4

a

b

Fig. 4. (a) Experimental strain over time for a strain gauge (15 on Fig. 3); (b) Experimental deflection over time for a wire potentiometer (27 on Fig. 3). The red lines approximately indicate the time interval for averaging. All results are normalized.

Left foot

Left leg

Beam

Discs

Right leg

Right foot

Fig. 5. Setup of the FE analysis in ANSYS Mechanical. Names of parts in the model are shown.

The cutterbar beam is hollow with bearings and gears inside. These are used to rotate the discs with knives that cut the crops. To significantly simplify the FE model, the bearings and gears have been modelled using spring elements connecting the lower and upper side of the cutterbar beam. This is deemed reasonable, as the bearings are fastened using a high clamping force. Fixed boundary conditions are applied to the legs, where they are welded to the steel table (A and C in Fig. 5). In addition, a symmetry constraint is applied in the symmetry plane (B and D in Fig. 5). The maximum deflection from the experiment is applied in the middle of the cutterbar beam (E in Fig. 5). The displacement is applied on a face corresponding to the bracket on which the hydraulic actuator is mounted. The strains at the locations of the strain gauges are exported to be used for FE model validation. To validate the accuracy of the FE model, the model is compared to the experiment in terms of strain values. Firstly, the strain values are graphed to illustrate how closely the FE model reflects the experiment, see Fig. 6. In Fig. 6, individual strain measurements are shown, but a line is added for easier comparison of the data sets. The strain values are normalized such that the lowest experimental strain is assigned the value of 0, and the highest experimental strain is assigned the value of 1. Secondly, the mean absolute error (MAE) is calculated to obtain a numerical measure describing the accuracy of the FE model, using Eq. 1. Where ε A i denotes the analytical strains from the FE model, ε X i denotes the experimental strains, and n is the total number of strain values. MAE is used, as all errors are weighted equally. Other methods such as the root mean square error (RMSE) prioritize the largest errors. The MAE of the FE model is 0.075. In relation to the span of experimental strain values, the MAE corresponds to 7.5 %. MAE = n i = 1 | ε A i − ε X i | n (1) 3.1. FE model validation