Demand events can be classified in:
- Attraction events.
- Demand events.
An attraction event is characterized by the arrival, at the destination point associated with the event, of an abnormal number of vehicles compared to what happens outside the validity interval of the event.
A generation event is characterized by the departure, from the point of origin associated with the event, of an abnormal number of vehicles compared to what happens outside the validity interval of the event.
Generally, you can consider these two types of events completely independent.
In the common use cases, during the interval time considered, an attraction event can become, symmetrically, a generation event.
For example, if you consider the footbal match for the final of the European Champions League, in the starting phase of the event the footbal stadium can be associated to an attraction event.
At the end of the match, the football stadium can be associated to a generation event.
For each event, the ratio of the extra generation or attraction is calculated with respect to the total generation or attraction of the relevant zone during the event.
At any instant:
- The origin flow is increased by a fixed amount determined by evenly distributing the extra vehicles over the entire event duration.
- The turn volumes are increased by adding to the baseline value of the origin or destination volume the event increment and a time relevance factor expressing how much of the current simulation interval is relevant to the event time window.
Important: TRE addresses the traffic simulation only, so demand events are considered by TRE in its traffic simulation.
The listed definitions presume:
- A generation event from origin zone O.
- An attraction event is symmetrical for a destination D.
| Parameter | Description |
|---|---|
|
Dod |
Total demand volume over the event duration, measured in [veh]. |
|
σe=Ne/∑dDod |
Fixed demand increase factor caused by the extra Ne vehicles from the event. |
|
Nod=σeDod |
Extra volume on the od pair caused by the event, measured in [veh]. |
|
Vt,o,e=Vt,oσeθt,i |
Extra volume on turn t caused by the event during interval i. |
|
θt,i |
It is the overlap between the current time interval, advanced by the travel time from the current turn t to the destination, and the arrival time window. |
Computation of OD shortest paths
For each time interval, execute a dynamic Dijkstra algorithm both forwards and backwards in time from each zone, to be used for generation and attraction events respectively.
Shortest paths are described by a sequence of turns, each coupled with an arrival time obtained as an offset with respect to the instant of departure|arrival at the event zone.
For each zone, the Dijkstra exploration should yield the costs and times of every turn in the tree.
Starting from each of the other zones (leaves) the shortest paths can then be computed by following the lowest cost back to the root and saving the travel times along the way.
Computation of the total OD demand
Generation Event
For each destination, compute the total number of vehicles leaving the origin during the event validity.It is the integral of the OD flows during the event.
∑dDod
Each matrix or matrix timeseries that is valid during the event, give a contribute according to the overlap between its validity and the event validity.
Attraction Event
For each origin, compute the total number of vehicles reaching the destination during the event validity.It requires computing, for each origin, the departure time window that satisfy the above → Definitions.
∑oDod
The departure time window is found by shifting the arrival window back in time according to the travel time from the origin to the destination.
By using equilibrium travel time values we assume:
- They are valid for any given point in the network.
- The computation is possible even if the arrival time is after the end of the current simulation.
- It is enabled compression and expansion of the departure/arrival time window.
Using offline OD shortest paths
The origin nodes are loaded with additional demand according to the event validity.
On each OD path, all turn flows are updated by adding a fixed number of vehicles Nod
spread over the simulation intervals that, according to the arrival-departure times saved with the shortest path, are relevant to the event validity.
If needed, the delta turn volumes are later modified by the rerouting events.
About how to configure TRE to manage demand events, see → Demand events configuration.