This paper proposes a novel model that can predict the displacement of suction caisson anchors under monotonic and cyclic loading. Failure is assumed to occur when the accumulative monotonic and cyclic displacement along the load attachment point is over 60% of the diameter of the anchor. The anchors will go through lateral failure when the accumulative monotonic and cyclic displacement along the loading direction at the load attachment point is over 30% of the diameter. Hence, it is important to predict this displacement and therefore determine the expected failure of the anchor. However, it is difficult to predict displacement using the modern software without knowing the material properties of the soil and piles. Hence a new model that relies only on the normalized static load (Fa/Ff), normalized cyclic load (Fcy/Ff ), loading angle (Θ), and the number of cycles (N) is proposed. The inputs for training of the proposed model are (Fa/Ff), (Fcy/Ff), (Θ), (α) and (N). The output of the model will be the displacement normalized by the diameter of the anchor. To generalize the trained model, unused sets of data are used to validate the model. Furthermore, a comparative study is performed to evaluate the effectiveness of the proposed model. It is shown from extensive simulation that the model can accurately predict the normalized displacement of suction caisson anchors.

This paper proposes a novel model that can predict the displacement of suction caisson anchors under monotonic and cyclic loading. Failure is assumed to occur when the accumulative monotonic and cyclic displacement along the load attachment point is over 60% of the diameter of the anchor. The anchors will go through lateral failure when the accumulative monotonic and cyclic displacement along the loading direction at the load attachment point is over 30% of the diameter. Hence, it is important to predict this displacement and therefore determine the expected failure of the anchor. However, it is difficult to predict displacement using the modern software without knowing the material properties of the soil and piles. Hence a new model that relies only on the normalized static load (Fa/Ff), normalized cyclic load (Fcy/Ff ), loading angle (Θ), and the number of cycles (N) is proposed. The inputs for training of the proposed model are (Fa/Ff), (Fcy/Ff), (Θ), (α) and (N). The output of the model will be the displacement normalized by the diameter of the anchor. To generalize the trained model, unused sets of data are used to validate the model. Furthermore, a comparative study is performed to evaluate the effectiveness of the proposed model. It is shown from extensive simulation that the model can accurately predict the normalized displacement of suction caisson anchors.