As an archipelagic country, Indonesian government gives subsidy and public service obligation (PSO) to Sea Tollway, Pelni and Pioneer vessels to distribute freights to small islands. These three types of vessel are independently managing their own time and network. As the result, the distribution network becomes unintegrated and not optimal in suppressing transport cost and subsidy-PSO. The aim of this paper is to develop and implement bi-level model in sea transport network integration on freight distribution system on archipelagic territory. Upper level model is intended to minimalize the number of operated vessels, while lower level model is intended to determine the route that maximize the profit in freight distribution that involves Sea Tollway, Pelni and Pioneer vessels. The application of genetic algorithm (GA) in problem solving on milk run time windows (MRTW) network can accommodate the uncertainty variable, namely load and wave height. The result of model validation with empirical test on the case in Indonesia shows that the model can gives optimal value in solving the network integration problem. The sensitivity analysis shows that the number of operating vessels and the uncertainty variable intervention on the determination of vessel route affect the clustering. The management of vessel route should be integrated in real time by factoring the wave height variable, with consequence that the received gross profit is decreasing by 11, 8% when compared to the condition without the wave height consideration.