Uncertainty Scenarios

There are many points in the simulation model where we can inject uncertainty to mimic the real world scenario. Below we list out the various points and the type of uncertainty model which can be applied to each:

• Aircraft speed - The speed of the aircraft can be expected to differ around its expected value with some variance. (Gaussian uncertainty)

• Time uncertainty - All times in the scenario i.e arrival and departure time can be affected with gaussian uncertainty around expected value.

• Location - Unlike speed and time which can be changed in a continuous range, location in our model can only hold certain discrete nodes and for this we can model the uncertainty as a binary parameter i.e either the vehicle changes node location with 80% probability or waits with 20% probability to mimic a scenario with 20% uncertainty.

• Pilot decision - This could mean pilot decision to wait at a certain point which could affect flights before and after it. In any case, this can be modeled as point uncertainty or binary parameter which would mimic if the pilot did move the vehicle or not.

• Uncertainty at terminal nodes - Distinction between appear time and actual appear time for scheduler as terminal nodes are more prone to uncertainty.

Factors to consider while determining how much uncertainty to apply.

• Rescheduling frequency - Inversely proportional
• Simulation time - Directly proportional
• Time gap in start time - Directly proportional

Other scenarios: We could also make certain scenarios to combine many of these factors and uncertainties together which the simulator can use. Possible scenarios include:

• Bad weather/Storm/Fog Aircraft speed reduced, pilot decision and time uncertainty increased due to bad visibility.

• Construction/Repair Certain random edges will not be available with total uncertainty.