Arctic route planning under ice uncertainty: A risk-averse stochastic shortest path problem

As global warming leads to reduced ice coverage, Arctic shipping routes are becoming increasingly accessible. These routes offer significant reductions in shipping time and may provide both economic and environmental advantages through lower fuel consumption and reduced emissions. The primary challenge in route planning comes from unpredictable ice conditions, which affect transit time. In this paper, we formulate a risk-averse stochastic shortest path problem for conservatively estimating transit times for Arctic shipping. This model estimates a risk-averse transit time for a ship prior to departure, considering the uncertainties in ice conditions. It minimises a weighted sum of expected and Conditional Value at Risk (CVaR) transit times, subject to a given risk-aversion level. We evaluate our model using projected ice data for the Northeast and Northwest Passages. Our conservative findings indicate that as sea ice retreats, shipping routes move northward, resulting in shorter CVaR transit times. While greater risk aversion can extend expected transit times by up to two days, this increase is relatively minor compared to the corresponding gain in reliability.