- Precisions about trains order after junction/disjunction (new sections 4.3.9 and 4.3.10), (questions from Martin Geiger in the forum, 30th Aug).

- Precision related to the header line inside solution file (in section 6.2.2), (question from Nicolas Catusse in the forum 30th Sep).

- Style modification in section 4.3.8 : enumerated list changed into a table to be more synthetic.

- Small correction in solution example in Table 1 in section 6.2.2 (added a ressource associated with event EnterSystem for Train 1).

- Correction in the symbols list (inversion between km and second in remDCost and remTCost).

- Upper Bound for category number increased from 10 to 30 (section 6.1).

- Correction in the expression of the way conflicts are detected on trackGroups in section 5.2.1 (message from Nicolas Catusse in the forum 8th Oct).

- Added the following precision to Section 4.3.8 (Train Order on Individual Tracks) : “The entrance times beg1 and beg2 must be different. Consequently, either t1 or t2 enters first on r, without ambiguity.”

As announced during the qualification phase, modifications are brought for the final phase. In short:

- Penalties associated with objective f2 in qualification phase are turned into constraints

- Uncovered arrivals and unused initial trains are now feasible, as it was already the case for departures which could be uncovered in the qualification phase

- Objectives f1 and f3 in qualification phase are merged into a regular (weighted-sum) objective function instead of the lexicographic approach