Home / Summaries / Class notes - Transportation and Spatial Modelling / model-models-trip

Overall Summary

Q: Modelling approach for Time-of-day modeling

The main technique for time-of-day modelling is discrete choice or logit modelling . Usually: 1. the utility for travelling in the peak is set to 0 2. while the utility for travelling off peak is described by a disutility for arriving in an earlier or later period plus a utility of the reduction in travel time due to travelling in the off-peak period.

Q: Specify the variables in the formula for the multiplier

The multiplier xk is simply the count (Sk) divided by the computed flow: - The latter is the result of the assignment map (or route incidence matrix) α k , the ijr route fraction φijr and the flow Tij. - The variables i, j and r are indices for the origins, destinations and routes respectively.

Q: Why can't we update the matrix with just a multiplier when more counts are being used?

It is very likely that for OD-pairs passing multiple counts the information on the direction and the magnitude of each multiplier will be inconsistent. As there are more matrix cells than constraints the number of possible solutions could be very large, so what’s the best solution then?

Q: How would you adapt a model if the through traffic was set at 1000 trips in each direction?

The values for the deterrence function for OD-pairs 5-6 and 6-5 (cordon zones) should be set to 0 as well and the departures and arrivals for zones 5 and 6 should be reduced with 1000 trips. After the trip distribution has been computed the through traffic can be added to the matrix.

Q: List a descriptive method to model mode choice as a separate step in the 4 stage model?

Descriptive approach using empirical functions such as travel time ratio: illstrated by the VF-curve

Q: What is the assumption for Wardrop's equilibrium

This equilibrium states that for each OD-pair the travel time for all routes used is equal and shorter or equal to all other routes . In that case the duality gap should be equal to 0. (In case of a SUE travellers will also opt for routes they perceive to be shorter, yet having a longer travel time in reality. In that case the duality gap will converge tot a certain value, larger than 0).

Q: Difference 4-stage model and LRS/NRM structure?

1. The steps destination choice and mode choice are combined 2. A 5th step is added: time-of-day choice.

35 important questions on Overall Summary

Difference time-of-day model and departure time modeling?

Time-of-Day: choice to make a trip in peak hour OR in a earlier or later period. [period scale is at least an hour]
Departure time: determines the actual departure time based at the level of minutes based on a preferred arrival time

Modelling approach for Time-of-day modeling

The main technique for time-of-day modelling is discrete choice or logit modelling.
Usually:
1. the utility for travelling in the peak is set to 0
2. while the utility for travelling off peak is described by a disutility for arriving in an earlier or later period plus a utility of the reduction in travel time due to travelling in the off-peak period.

Specify the variables in the formula for the multiplier

The multiplier xk is simply the count (Sk) divided by the computed flow:

- The latter is the result of the assignment map (or route incidence matrix) α k , the

ijr route fraction φijr and the flow Tij.
- The variables i, j and r are indices for the

origins, destinations and routes respectively.

Specification Cordon Model Trip Generation

1. Traffic counts
2. Derived from regional models

Why can't we update the matrix with just a multiplier when more counts are being used?

It is very likely that for OD-pairs passing multiple counts the information on the direction and the magnitude of each multiplier will be inconsistent.
As there are more matrix cells than constraints the number of possible solutions could be very large, so what’s the best solution then?

How would you adapt a model if the through traffic was set at 1000 trips in each direction?

The values for the deterrence function for OD-pairs 5-6 and 6-5 (cordon zones) should be set to 0 as well and the departures and arrivals for zones 5 and 6 should be reduced with 1000 trips. After the trip distribution has been computed the through traffic can be added to the matrix.

List a descriptive method to model mode choice as a separate step in the 4 stage model?

Descriptive approach using empirical functions such as travel time ratio:
illstrated by the VF-curve

What is the assumption for Wardrop's equilibrium

This equilibrium states that for each OD-pair the travel time for all routes used is equal and shorter or equal to all other routes. In that case the duality gap should be equal to 0.
(In case of a SUE travellers will also opt for routes they perceive to be shorter, yet having a longer travel time in reality. In that case the duality gap will converge tot a certain value, larger than 0).

What is a tour?

A chain of trips that start and end at the same address. LMS considers tours of two trips.
Advantage: Modeling tours is more consistent with actual behavior.
--> Consistency in mode choice

Difference 4-stage model and LRS/NRM structure?

1. The steps destination choice and mode choice are combined
2. A 5th step is added: time-of-day choice.

What is the pivot point method?

Pivot point method determines growth factors on a base year matrix.
1. A synthetic model is applied using the input for the base year and using the input for the future year
2. Based on these model runs: growth factors per OD-pairs are determined.
3. Growth factors are applied to the base year matrix.

The future OD matrix is therefore pivoted around the base year matrix with growth factors.

Why is the Pivot Point method applied? + Major disadvantage

The pivot point method tries to combine the best of both worlds:

The (choice) models of LMS/NRM provide a detailed description of the sensitivity of all modelled choices for changes in the input, while the base year matrix gives the best possible description of the spatial demand pattern.

Disadvantage: If a zone did not exist yet in the base year, but new developments exist in a future year, then the pivot point method is not able to determine this growth factor (as the cell in the OD matrix is zero).

What are the main characteristics of descriptive models and models based on travel choice theory?

Descriptive models: based on analogies or statistical relationships (input and output only)
Choice based models: such as logit models are based on modeling travel behaviour using utility theory.

Differences LMS/NRM with 4stage:

1. Use of tours instead of trips
2. Focus on choice modeling
3. A dedicated assignment method

What is the stop-and-go model for tour generation

A binary choice model, to determine the probability of making a tour or not.
For the possibility that a tour is made, the choice model is used to determine whether another tour is made.

This process is repeated until a maximum of tours is reached or that the probabilities become too small. Note that there’s no continuation for the options that no tour is made.

Typical limitation of an exponential deterrence function?

It overestimates shorts distance trips --> Top-lognormal function might be useful.

Define the demand models that are distinguished in LMS/NRM framework.

Trip generation (Stop and Go model)
o This model might be split into car ownership model, driving
license model and a tour frequency model
Destination choice
o For car driver this is extended with models for intermediate
destination in order to include complex tours

Mode choice
Time of day choice
The latter three models are integrated in nested logit models having different nesting structures per trip purpose. The logsums of these nested logit models are used in the trip generation model.

Dutch National Model system : feedback loops

LMS: distinguishes car driver and car passenger as separate modes.

Feedback loops:

Level of service of the road network is fed back to the simultaneous distribution choice, mode choice and time of day choice model. The recent version also includes a feedback of level of service to trip frequency (via the simultaneous choice model)

Criticism for every component of the classic 4 stage model

Trip generation: no real behaviour, no impact of accessibility, no link with actual activities, no trip chains
Trip distribution: no real behavior, limited level of detail given the variety in reality, no trip chains (could be combined with modal split)
Mode choice: no real behaviour, strict split in modes, inconsistency between peak hour models (could be combined with modal split)

Assignment: no route choice, static assignment (where’s the congestion?), no queueing

2 reasons why the classic 4-stage model is still applied?

Simplicity
Limited need for data
Commercial software available
Suitable/sufficient for small study areas (cost effectiveness)

Improvements for components of the 4 stage model

A. Trip generation: modeling of tours, accessibility is considered, flexibility w.r.t. time periods (TOD-model)
B. Trip distribution: (tours, including complex tours for car driver), use of base- year matrices (and thus pivot-point)
C. Mode choice: more modes are considered and for train different access and egress modes are considered, use of tours (consistency outbound-homebound)
D. Assignment: Explicit modelling of blocking back (Qblok) albeit still a static assignment

Promising future developments for LMS/NRM?

Trip generation: activity based modelling
Trip distribution: use of big data for OD-estimation
Mode choice: integrated mode choice and route choice/multimodal assignment
Assignment: dynamic assignment

Difference between a tour and a trip

A trip is defined as a movement of an individual from one activity location (origin) to another activity location (destination).

A tour is a chain of trips starting at ‘fixed’ location (usually home, but work can be a starting point as well) and returning at that location. The most common tours consist of two trips: an outbound trip (from home to work) and a homebound trip (work to home).

Benefits tour based approach?

Consistency in mode choice modeling. People tend to go somewhere by one mode and return by the same mode.

Schematic overview of the pivot point method

The pivot-point method uses the transport model to predict relative changes between a future year and the base year. These relative changes (or growth factors) are then applied to the base year matrix to obtain a future matrix.

Explain the key characteristics of a disaggregate model approach (from LRS/NRM)

The basic unit of analysis in disaggregate models is the individual or household (or a rich variety of person or household types). The travel behaviour of the persons or households is modelled using discrete choice models using among other matters typical person or household attributes.

Mathematically formulate the potential value for a leisure centre that you would like to be accessible for potential clients

Potential value is based on the incoming customers (from zone i)
Coming towards the leisure centre in zone j
F(Cij) should be a decreasing function, since a zone is less accessible if costs/distance is higher. --> F(Cij) = 1/Cij

What is Potential Value

The accessibility of a zone j

How is the potential value used in classic models such as the Hansen-model or the Lowry-model?

Hansen model: distributes a volume (e.g. extra inhabitants) based on the ratio of the product of the potential value of a zone and the attractiveness (or capacity) of that zone, and the sum of this product for all zones.

Lowry mode: distributes a volume related to a zone j based on the ratio of the contribution of a zone I to the potential value of zone j and the potential value of zone j itself.

Wegener's circle: What is input and what is output of the transport model part?

1. The input is the block with activities.
2. Output: travel times and costs, which can be used to determine the accessibility of locations.

Land-use system part: main components?

1. The accessibility of locations affects the attractiveness of locations.

2. The attractiveness is an important factor for e.g. project developers (projects that have been decided upon, have to be built)
3. The characteristics of the project and the attractiveness of the location are important factors for firms and households
(if they choose to opt for a new location, they have to move their firm of household)
Once the firms and households have settled, they start their activities

Define at least three markets for the land-use part and indicate where they fit in Wegener’s circle.

- Land market: directly relates to choices of investors
- Real estate market: relates to the choice of users, i.e. firms
- Housing market: relates to the choice of users, i.e. households
- Labour market: fits less straightforward in Wegener’s circle, however it ‘balances’ the Real estate market and the Housing market.

Lowry Model: main principles of the algorithm

This graph illustrates the Lowry Algorithm.

TIGRIS XL Land use models, which markets are distinguished?

Land market: allocation of space
Housing market: interaction houses and households
Real-estate market: interaction between real-estate and firms
Labour market: interaction between firms and households

Main modelling principles applied for TIGRIS XL?

Land market and real estate market: simplistic allowing for different policies ranging from fully regulated to free market
Housing market: choice modelling in two steps: to move or to stay, and if it is decided to move the choice of the new location
Labour market: regression models for different sectors

The question on the page originate from the summary of the following study material:

Transportation and Spatial Modelling

View summary

A unique study and practice tool
Never study anything twice again
Get the grades you hope for
100% sure, 100% understanding

Remember faster, study better. Scientifically proven.