Trunk shakers perform a forced vibration that should be set according to tree species and architecture to efficiently detach fruits. However, manufacturing companies produce designs that are poorly customisable according to different trees. This work presents a prototype that is configurable in amplitude and frequency of vibration, which has been developed to generate the different vibration patterns required by various fruit trees. To study the feasibility of the developed technology, we built a test bench with a post, to which we fixed olive, almond and orange tree branches, and tested different configurations of the machine (frequencies from 12.25 to 20.75 Hz and eccentricities from 58 to 86 mm). Acceleration values were recorded at different points: the trunk shaker, the post in the grip, the top of the post, and in each branch. We recorded a large loss of acceleration transmissibility between the shaker and the post (20%–60%) due to existing frictions that dissipate energy through heat loss. There is a non-linear increase between vibration frequency and the resultant acceleration value recorded, as well as with increasing eccentricity. The shaker was able to provide a different vibration pattern in each configuration, affecting the transmission of vibration in each of the branches studied, for which mean acceleration transmissibility ranged between 82% (almond), 104% (olive) and 136% (orange).

The development of trunk shaker machines over the years has been based on test-error methods in field. Mathematical or computational models have been studied with great simplifications. This paper presents a method for modelling the dynamic behaviour of a trunk shaker with a test bench. Two mass configurations were used on the test bench as well as two different vibration frequencies on the trunk shaker. Acceleration values were recorded at different points of the system. The binomial shaker-post was computationally modelled, and its dynamic response was analysed based on a modal and transient study with a series of proposed simplifications. The results of the simulations were compared with experimentally recorded acceleration values. In both cases, a linear response to mass and frequency variation was observed in the acceleration that the shaker performed. There was a high correlation in the effective accelerations (error < 4%) between experimental and computational studies measured in the trunk shaker. However, there were higher errors when the post was used in the test in the post structure points. The greatest uncertainty in the model may lie in the assumption of contact between the attachment pad and the post, but if this is not carried out, it makes convergence in the computational calculations very difficult. The method has proved its worth in determining the dynamic behaviour of these machines.

No abstract