According to the problem of uncontrollable safe shutdown time when the insulation layer was partial failure under the specific conditions, a two-dimension, unsteady-state model of hot oil pipe was established by the finite volume method. In the model the influence of the latent heat of solidification was considered and the law of temperature drop in the hot oil pipe was compared and analyzed in five cases (i.e. no insulation failure, 1/8 insulation failure, 1/4 insulation failure, 1/2 insulation failure and complete insulation failure). Based on the above, the relative curve between the shutdown time and the average temperature of hot oil after line turndown was fitted by SPSS 18.0, and the safe shutdown time of the above five cases were determined finally. The results show that the laws of temperature drop of hot oil pipe is similar in the above five cases, and the safe shutdown time is 205, 148, 118, 99 and 74 hours. Due to the latent heat of solidification could partly make up for the heat loss, safe shutdown time of 1/2 insulation failure was only 19 hours less than 1/4 insulation failure’s.

Considering the difficulty in the estimation of safe shutdown time of oil pipeline due to the complex safe shutdown of submarine oil pipeline influencing factors, the radial basis function neural network model was proposed for predicting safe shutdown time of submarine oil pipeline and the various influence factors on safe shutdown of oil pipeline were analyzed. The model was validated on the basis of actual data and the network was trained and the accuracy was verified. The results show that the fitting and simulation precision for training and testing samples is 98.40% and 97.33%, respectively. So the safe shutdown time of submarine oil pipeline can be predicted validly, and the important basis of safe transportation of submarine oil pipeline was provided.