Computing delayresistantrailwaytimetables

In thepast,muchresearchhasbeendedicatedtocomputeoptimumrailwaytimetables.Atypicalobjec- tive hasbeentheminimizationofpassengerwaitingtimes.Butonlytheplannednominalwaitingtimes have beenaddressed,whereasdelaysastheyoccurindailyoperationshavebeenneglected.Delayshave been rathertreatedmainlyinanonlinecontextandsolvedasaseparateoptimizationproblem,called delay management. We providethefirstcomputationalstudywhichaimsatcomputingdelayresistantperiodictimeta- bles. Inparticularweassessthedelayresistanceofatimetablebyevaluatingitsubjecttoseveraldelay scenarios towhichoptimumdelaymanagementwillbeapplied. We arriveatcomputingdelayresistanttimetablesbyselectinganewobjectivefunctionwhichwe design tobesomehow in themiddle of thetraditionalsimpletimetablingobjectiveandthesophisticated delay managementobjective.Thisisaslightextensionoftheconceptof“lightrobustness”(LR)asit has beenproposedbyFischettiandMonaci[2006.Robustoptimizationthroughbranch-and-price.In: Proceedings ofAIRO].Moreover,inourapplicationweareabletoprovideaccurateinterpretationsfor the ingredientsofLR.Weapplythisnewtechniquetoreal-worlddataofapartoftheGermanrailway network ofDeutscheBahnAG.Ourcomputationalresultssuggestthatasignificantdecreaseofpassenger delays canbeobtainedatarelativelysmall price ofrobustness, i.e.byincreasingthenominaltraveltimes of thepassengers.