Pump Tower Loads in Spherical Tanks Onboard LNG Carriers Scaled by Characteristic Vessel Response

Fatigue of the pump tower structure is an important design aspect for spherical cargo tanks onboard liquid natural gas carriers. Current state-of-the art methodology for assessing these loads is based on a series of model tests carried out about 10 and 40 years ago. The last campaign recorded tower loads for several tank filling ratios, vessel headings and sea states. Long-term distributions of loads were established, and this required the development of a scaling methodology to estimate short-term load distributions for sea states outside the range of tested conditions. However, the scaling approach proved to be inadequate when put into use in projects. The present paper proposes a new and enhanced scaling method for short-term distribution of loads on the pump tower involving re-assessment of the model test results combined with new statistical treatment. The dynamic loads on the pump tower causing fatigue damage can be separated into loads due to sloshing and inertial loads. The sloshing loads are seen to correlate well with the amount of energy in the acceleration response spectrum of the tank in a frequency range in vicinity of the lowest natural sloshing frequency. A characteristic response variable is defined as the integrated energy in the acceleration response spectrum over one half octave band around the sloshing resonance frequency, and a functional relationship between the loads on the pump tower and this parameter is established. The results show that this model describes the observed pump tower loads well, especially for filling heights 30 to 70%. Some more scatter is observed for very low or high filling levels, but as these cause lower loads this is not of concern. The underlying model tests primarily focused on high sea states in order to reflect global trade. The proposed scaling methodology allows for robust extrapolation to lower sea states in a more accurate way than was previously possible. This is particularly useful to avoid overly conservative fatigue-load estimates for other applications such as FSRUs and FLNGs in more benign conditions.