Trains are planned with margins compared to their fastest possible drive. Without them, they would not be able to catch up in case of a delay. A train departing 5 minutes late would arrive at least 5 minutes late. A driver accelerating or braking a bit slower than planned would inevitably be late. The entire transportation plan would fail.

Those margins/allowances consist of adding extra time to the total running time. This implies planning trains with a lower speed than the maximum they could physically drive.

There are many ways to lower the speed of a train along its trip, but one of them is particularely efficient in terms of energy savings. The algorithm MARECO, developped by SNCF, details an entire strategy for this purpose: how to plan a train that would intelligently drives slower in a way that consume the least amount of energy ?

In this talk we will go though the main ideas of the algorithm, then we will compare its efficiency to other margin distribution stratgies on a few relevant examples, and finally we will discuss the impact of this algorithm on train planning and operation.