Tactical Evaluator Pipeline¶
The tacical evaluator enables the simulation of the realisation of the (re)planned day of operations. It uses Mercury ABM simulator in its MultiModX branch. The MultiModX repository has this imported as an external library (tactical evaluator).
This documentation describes the capabilites of the Tactical Evaluator and the scripts within MultiModX for the creation of the required inputs and the computation of associated performance indicators.
See additional documentation for information on:
1. Intended Audience¶
This documentation targets:
Researchers¶
- Study robustness in passenger-centric metrics of (re)planned networks under different levels of stress (delays)
- Analyse the effectiveness of mechanisms (e.g. fast-track to process multimodal passengers at airports) to support multimodal connectivity
Scenario Analysts¶
- Evaluate changes in infrastructure or operations that support multimodal connectivity
- Assess the number of expected missed connections under different environments
Developers¶
- Prepare input datasets for Tactical Evaluator
2. Capabilities¶
Mercury can simulate passengers with their itineraries (including their connections) for flights (and flight-to-flight connection ✈️ ↔️ ✈️) and multimodal itineraries (✈️ ↔️ 🚆) when passengers are connecting with a rail service (rail-to-flight, flight-to-rail).
The model also simulates the required flight, rail services and ground mobility (🚇, 🚌) (from rail station to airport and from airport to rail station).
Figure A shows the different type of multimodal connections considered.
The Evaluator can simulate any network, but focus on a given hub with its multimodal connectivity is a particularly well-suited use case.
The Evaluator can also assess different levels of delays (e.g. in the ground mobility layer (connectivity between airports and rail stations)) and mechanisms that modify some connectivity parameters (e.g. fast-track at airport for multimodal passengers or dedicated bus line between airport and rail stations). These elements were evaluated for example in Weiszer, M., Delgado, L. and Gurtner, G. (2024), Evaluation of Passenger Connections in Air-rail Multimodal Operations, SESAR Innovation Days 2024.
3. Limitations¶
The tactical evaluation of rail-rail connectivity is not explicitly modelled. Passenger itineraries with only rail itineraries are not simulated. Only rail services involved in multimodal journeys are modelled.
These limitations do not hinder the evaluation of multimodal connections, and the delay for rail passengers can be estimated
as post-processing using the postprocessing_tactical.py
provided as part of the Performance Indicators.
This script post-process the execution of the tactical evaluator to add the modelling of the passenger itineraries which
are not directly supported by the Tactical Evaluator.
However, due to these modelling aspects, the Strategic Evaluator Pipeline
divides, in its final computation stage, the passenger itineraries between pax_assigned_tactical_#.csv and
pax_assigned_tactical_not_supported_#.csv.
4. MultiModX Scripts¶
As mentioned, the final stage of the Strategic Evaluator Pipeline is the generation of tactical inputs. In this stage, the output of the strategic analysis is transformed into:
flight_schedules_tactical.csv, which reformat the flight schedules into the format used by the Tactical Evaluator.pax_assigned_tacitcal.csv, creating passenger itineraries to be used by the Tactical Evaluator.pax_assigned_tactical_not_supported.csv, with the information on the passenger itineraries that are not supported by the Tactical Evaluator (rail-to-rail connections and rail only itineraries).
In the GitHub repository, script/pre-tactical/, contains
two additional scripts are used to finalise the creation of the input for the Tactical Evaluator:
generate_input_for_mercury.py (to finalise creating all required inputs) and routes_ddr.py (to ensure all flights have available fligth plans).
Note that the Mercury requires, in addition to the flight and rail schedules and passenger itineraries, other paramters and distributions, such as: flight plans, ATFM regulations, minimum turnaround times, etc. All these factors and parameters are beyond the outcome of MultiModX.
Finally, the outcome of the Tactical Evaluator is processed to generate multimodal passenger-centric indicators as described in performance indicators computation.