These two models (the highway and public transport assignment models) are the basis for the other models so they deserve first treatment. You do not have to have either of these assignment models if you are not concerned with networks and these other types of models are discussed later.
Highway Assignment Models Explained
The first component is the network which is a representation of the road network in a computer spreadsheet file. It contains a list of the road junction (nodes) and the road links which connect them together with details about the length, speed capacity and other characteristics. Only the main roads and junctions are normally represented. The study area is divided into zones which represent the places or geographical areas in the model. The trip matrix is a table which says how many car trips there were in a day between every pair of places (ie zones) in the study area. The highway assignment model software then is tasked with finding the shortest path through the network between every pair of zones. For each path it looks to see how much traffic travels between these two places from the matrix and loads this traffic onto each link. It deals with every pair of zones in the matrix and at the end of this process the total amount of traffic the model has put onto each link is the model's estimate of the traffic volume on each link. This process is called traffic assignment. The model's traffic volume estimates can be compared with the amount of traffic in real-life. For example if you were to go out and count the vehicles passing each day on average it should equal to the modelled traffic volume - in fact one of the steps taken when building the model is to make sure it does.
If you wanted to build a new road, you would want to know how much traffic would use it and how much traffic would be removed from other roads then the new road could be coded as new links in the model network and the assignment process re-run. The result would be the amount of traffic the assignment model estimates each link will carry. The amount of traffic estimated on the new links in the model is the amount of traffic the model has estimated that the new road will carry. For the existing roads the model's traffic volume estimate for a link can be compared with what it was without the new road and the difference is the traffic diverted from the link due to the new road. Bingo - we have our forecast!
Preparing the Highway Network
A highway assignment model takes a description of the road system comprising nodes at road junction connected together by links (the roads) so as to represent the complete network of roads and their interconnection. All the nodes are numbered and given a map reference (or geo-code) so that they can be located on the map. Each link in the network is listed together with the node number at each end of the link, its length, the traffic free flow speed, its width and capacity and some parameters which describe how the traffic speed decreases as the volume of traffic increases. These two (ie the node and link tables) are put into a spreadsheet. The network can be plotted on a map or given as a line diagram.
Preparing the Study Area Zones
The area for which the transport model is required is divided into study zones which area reasonably homogeneous in terms of the amount of traffic each area generates. Zone boundaries usually follow the lines used for other boundaries (eg national census, government, electoral or natural boundaries) because later on in the modelling process it may be necessary to relate the data collected from these other sources to the study zones.
Connecting the Zones to the Higway Network
A notional centre is designated for each zone which represents the centre-of-gravity of the traffic generated within it. The zone centre is connected to the adjacent network with special links called zone connectors which are coded with the usual link attributes approximated from the average road characteristics of the roads traffic would use to go between the zone centre and the network. The zone centre acts as a notional point from which the traffic generated by the zone is loaded onto the network via the zone connectors. The zone connector table forms part of the set of tables describing the network
What is a Roadside Interview Survey?
While driving your car, have you been stopped by a policeman and asked to pull into a bay behind other cars and an interviewer pops out and asks you where you have come from where you are going to (precise addresses or postcode please), what is the purpose of your trip? (also noting the type of vehicle and number of people in the car) - not many of us haven't by now. If you are observant you will also notice another person counting the number of cars, light, medium and heavy good vehicles, motorcycles, buses etc. Well that a roadside interview survey (rsi) site. A sample of the traffic is interviewed to find their origin and destination zones (from the addresses, postcode etc) and the counts are used to scale these up to represent all traffic on the road. For example if 12 times as much traffic uses the road as is interviewed the scaling-up (or expansion) factor would be 12. Usually a set of rsi survey sites are conducted round a town(s) or along natural boundaries (called screenlines) or on the slip roads of motorways (along screenlines).
Preparing the Trip Matrix
The interviews from all the roadside interview survey sites are built into a trip matrix. Each interview is examined in turn and if it represented say 12 counted vehicles and went from zone A to zone B then 12 would be added to the origin-destination matrix from zone A to zone B. Matrix building is a complicated subject and can be examined in more detail later. The result is a car vehicle trip matrix
Building The Base Year Highway Assignment Model
The highway network prepared earlier is submitted to the path building software with instructions to build paths between selected origin-destination pairs which the model builder defines. These paths are checked and any coding errors of link length speed etc made to ensure that the paths are reasonable according to the modellers local knowledge. When these are correct, the matrix is assigned to the network and the traffic volumes estimated by the model compared with counts made at selected points on the real road system for the same link. If there are any mismatches then the model inputs are adjusted to make sure the model matches the counts.
Forecasting Traffic Levels With The Highway Assignment Model
The transport model would be designed to test possible future improvements to the highway system. The improvements are then grouped into a set of network scenarios (a network scenario could be composed of one or many possible improvements) and coded into the network as a set of new files each representing one scenario.
If we interested in the amount of traffic carried on the road system in the future, then we need to take the trip matrix and make it represent the future origin-destination pattern. One simple way to do this is to measure how fast traffic is growing in general and apply this rate of growth to the future. If for example traffic is growing at a rate of 2% per annum then in ten years it would have grown by a little over 20%, so if your model base year was the year 2000 and your forecast year was 2010 then you could multiply all trips in the matrix by 1.2. (This is usually too simple for most purposes but we can look at this again later.) Your matrix now represents the traffic in the year 2010.
If you now assign the forecast scenario matrix to the forecast scenario network, the traffic volumes on the model's links represent the traffic volumes on these roads in the future. If you select the new links then the traffic volumes will be the amount of traffic on the new road. If your new links are in the region of 20,000 vehicles per day then you are approaching the levels needed for a dual carriageway. Bingo - you have your traffic model which can forecast traffic levels on new roads or changes to existing roads both now and into the future.
Strengths and Weaknesses
You have a basic model which works and can be used for a small town which is not likely to be developed. It does not cover the situation where public transport is likely to be improved nor where new modes of transport are envisaged. If you wish to develop a transport plan for all modes of transport then read the public transport section. In some countries there is a desire to attract car drivers onto other forms of transport including slow modes (cycle and walk). These issues need a mode choice model.
If extensive development is envisaged then you will need a better method of forecasting the trip matrix because new development is likely to occur in zones which are empty in the base year, ie there are no trips in the base year trip matrix so simply factoring these (zeros) by 1.2 still leaves you with zero whereas they should contain the trips which the new development sites will generate. There are methods for dealing with this which we will come to in trip generation.
If a major transport link is built, people may change the places they travel to so for example a local shopping trip can be replaced by one to a further-away supermarket because it is much more accessible with the new transport link. It also brings more job opportunities. It brings educational establishments closer and so on. In modelling terms the new link affects people's choice of destination and for this we need to change the distribution pattern of trips with a model called a trip distribution model. Our Highway assignment model does not model these distribution effects and we will look at distribution models later. But first let us turn to the next level of complexity - that of modelling public transport.
Public Transport Assignment Models Explained
What do we need a public transport assignment model for? To forecast what will happen if we improve public transport or if the demand for public transport changes, eg if it is in decline and you want to plan the reduction or if you want to see what happens if you reduce fares or change routes to make the system run more efficiently).
The Public Transport Network
Just as the highway assignment model models the choice of which path to take through the road network for all the travel demand in the study area, so the public transport assignment model models the choice of which path to take on public transport. This is similar to the highway assignment model except that cars can go anywhere, whereas public transport is confined to the services which the various bus and train companies run. On public transport you can only go between the places served by public transport, at the times they go there.
You will need to code the public transport routes. You code the nodes and links as for the highway network making sure you include the links used by public transport (including railway lines too!). You can use the nodes and links from the highway assignment model if you want although you may have to add some more. You also need an additional which describes each bus and train (and other public transport mode) routes. To do this trace the path of the route and list the nodes that the bus passes through. You may have to add a few bus stop nodes because each bus stop must be one of the nodes in the list. If the bus doesn't stop at a node then you can indicate this on the route table against the node. At the head of the list (of route-nodes) is a header record which gives the route number, its headcode (an alphanumeric code which the route is known by), the headway expressed as the time between services and perhaps some other information.
You will need to code the public transport fares. This is usually given as a fare table containing the fare per kilometre travelled on the particular mode of transport. First calculate the fare constant which is the fare for very short trip (of say 1 km) and enter this in the table. Then calculate the additional fare for a (say) 5km trip and divide by 5 (to get the additional fare per kilometre) and enter this in the table. Some modelling software lets you calculate different fare rates per kilometre for different distance bands but more of this later. For a particular trip the software calculates the public transport fare as the fare constant plus the distance travelled (in kilometres) multiplied by the fare rate per kilometre.
Apart from this the rest is similar to the highway assignment model. You have base year origin-destination trip matrices between study area zones representing all travel in a day by public transport. The matrices are assigned to shortest paths through the network (the path builder is a bit different but we will come on to that) to give the volume of people (instead of cars) on each public transport service (instead of link - although you can look at the public transport volumes on a link basis too if you want). You can forecast your matrix scenario and assign that to forecast volumes of people on each service or link, just as for the highway assignment model.
Strengths and Weaknesses
This has the same weaknesses as the highway assignment model (apart from the fact we are now modelling public transport). If you have a public transport assignment model it is usual to have a highway assignment model too - but you don't have to.