Traffic Forecasting

Introduction

The modification of a highway network either by new road construction or operating improvements on existing roads(e.g., use or retiming of traffic signalization) must be predicated upon some expectation or forecast of roadway traffic volumes. For new road construction, forecast of traffic are needed to determine an appropriate pavement as well as a appropriate geometric design that will provide an acceptable roadway level of service .For operational improvements, traffic forecast are needed to estimate the effectiveness of alternate improvement option.

In forecasting vehicular traffic, two interrelated elements must be considered: Overall regional traffic growth/decline. -In the design of these projects, the engineer must seek to provide a sufficient roadway level of service and an acceptable pavement ride quality for future traffic volumes. One would expect that factors affecting long-term regional traffic growth/decline trends are primarily economic and, to an historically lesser extent, social in nature. For example, some regional populations may have social characteristics that make them more likely, than other regional populations to make fewer trips, to carpool, to vanpool or to take public transportation ,all of which significally impact the actual amount of roadway traffic.

Traffic

Diversion -in addition to overall regional traffic growth/decline, there is the more microscopic , short-term phenomenon of traffic diversion. As new roads are constructed, as operational improvements are made, and/or as roads gradually become more congested, traffic will divert as drivers change routes or tripdeparture time in an effort to avoid congestions and improve the level of service that they experience.

Traffic Decisions
- Forecast of highway traffic should, at least in theory be predicated upon some understanding of traveller decisions, since the various decisions that travellers make regarding trips will ultimately determine the quantity, spatial distribution, and temporal distribution of vehicles on a highway network. Within this context, traveller can be viewed as making four distinct but interrelated decisions regarding trips:

Temporal decisions The temporal decision includes the decision to travel and, more important, when to travel. destination decisions The destination decision is concerned with the selection of a specific destination. modal decisions Modal decisions relates to how trip is to be made spatial or route decisions spatial decisions focus on which route is to be taken from the traveller's origin initial location.

SCOPE OF THE TRAFFIC FORECASTING PROBLEM

since traffic forecasting is predicated upon the accurate forecasting of traveller decisions, two factors must be addressed in the development of an effective traffic forecasting methodology: -The complexity of the traveller decision-making process. -System equilibration.

Figure 7.1

If we look more directly at the decision to travel mode/destination choice, and highway route choice, fig 7.1 indicates that both long-term and short-term factors affect these decisions. For the decision to travel, as well as mode/destination choice, the long term factor of modal availability, residential and commercial distribution, and modal infrastructure play a significant role. These factors are considered longterm because they change relatively slowly over time. A short-term factor, such a modal traffic, is one that can vary in short period of time.

The outcome of the combination of these traveller decisions is of course, highway traffic, the prediction of which is objective traffic forecasting. Aside from the complexities involved in the traveller decision-making process, the issue of system equilibration(mentioned in the beginning of the section must also considered. Note that fig 7.1 not only indicates long term and short term factors affect traveller decision and choices, but also that these decisions and choice in turn affect the long and short-term factors. Such a simultaneous relationship is most apparent when considering the relationship between traveller choice and short-term factors).Travellers decisions to take specific routes ultimately determine the route traffic upon which their route decisions are based. This interdependence between traveller decisions and modal traffic is schematically presented in Fig 7.2 In addition to these shirt-term effects, persistently high traffic volumes may result in a change in the highway infrastructures. Again resulting in an interdependence. This interdependence creates the problem of equilibrium that is common to many modelling application. Perhaps the most recognizable equilibration problem is the determination of price in a classic model of economic supply and demand for a product. This evolution of traffic forecasting methodology has led to the popular approach of viewing traveller decisions as a sequence of three distinct decisions, as shown in Fig 7.3 the result of which is forecasted traffic flow

Clearly, the sequential structures of traveller decisions is a considerable simplification of the actual decisionmaking process in which all trip-related decisions are considered simultaneously by the traveller. However, this sequential simplification permits the development of a sequence of mathematical models of travel behaviour that can be applied to forecast traffic flow.

Trip Generation
-The first traveller decision to be modelled in the sequential approach to traffic forecasting is trip generation. The objective of trip generation modelling is to develop an expression that predicts exactly when a trip is to be made. This is an inherently difficult task due to the wide variety of trip types.

Traveller in a sample day, as is schematically shown in fig 7.4 to address the complexity of the trip generation decision, the following approach is typically taken: Aggregation of decision-making units predicting the trip generation behaviour is simplified by considering the trip generation behaviour of a household as opposed to the behaviour of individual travellers. Such an aggregation of travellers decisions is justified on the basis of the comparatively homogeneous nature of household members and their often intertwined trip generating activities. Segmentation of trips by type Different types of trips have different characteristics that make them more or less likely to be taken at various times of the day. For example, work trips are more to be taken in the morning hours than are shopping trips, which are more likely to be taken during the evening hours. Also, it is more likely that the traveller will have two or three shopping trips during the course of a day than two or three work trips. To account for this, three distinct trip types are used: -work trips, including trips to and from work -shopping trips -social/recreational trips Temporal aggregation although research has been undertaken to develop mathematical expression that predict when traveller is likely to make a trip more often . Trip generation focuses on the number of trips made over some period of time. Thus trips are aggregated temporally, and trip generation models seek to predict the number of trips per hour on per day.

Trip Generation Models

Example Problem (7.1) A simple linear regression model is estimated from shopping trip generation during a shopping-trip peak hour(e.g., Saturday afternoon). The model is number of peak-hour vehicle-based shopping trips per household

=0.12+0.09(household size) +0.011(annual household income (in thousand of dollars )) -0.15 (employment in the households neighbourhood (in hundreds))

A particular household has six members, and an annual income of $50,000.they live in a neighbourhood with 450 retail employees. Calculate the predicted number of peak-hour shopping trips the household makes before and after the move.
Solution Note that the signs of the model coefficients ( bs, +0.09, and +0.011) indicate that as household size and income increase, the number of shopping trips also increase. This is reasonable since wealthier, larger household can be expected to make more vehicle-based shopping trips. The negative sign of the employment coefficient (-0.15) indicates that as retail employment in a household neighbourhood increase, fewer vehicle-based shopping trips will be generated. This reflects the fact that a larger neighbourhood retail employment implies more shopping opportunities nearer to the household, thereby increasing the possibility that a shopping trip can be conducted without the use of a vehicle turning to the problem solution, before the household moves, we find that number of trips=.12+.09(6)+.011(50)-.15(4.5)=.535 vehicle trips after the household moves, we obtain number of trips=.12+.09(6)+.011(50)-.15(1.5)=.985 vehicles trips Thus the model predicts that the move will result in .45 additional peak-hour vehicle-based shopping tips due to the decline in neighbourhood shopping opportunities as reflected by the decline in neighbourhood retail employment

Mode And Destination Choice

-After the number of trips generated per unit time is known, the next step in the sequential approach to traffic forecasting is the determination of traveller mode and destination. As was the case with trip generation, trips are classified as work, shopping, and social/recreational. For both shopping and social/recreational trips, the traveller will have the opinion to choose a mode of travel as well as a destination. In contrast, the work trip offers only the mode option, since the choice of work location is usually a long-term decision that is beyond the time range of most traffic forecast.

Theoretical Approach

Highway Route Choice

To summarize, the trip generation and mode/destination choice model give total highway traffic demand between a specified origin and destination, in terms of vehicle per some time period. With this information in hand, the final step in sequential approach to traffic forecasting, route choice, can be addressed. The result of the route choice decision will be traffic flow on specific highway routes, which is the desired output from the traffic forecasting process.

Highway Performance Function

Route choice presents itself as a classic equilibrium problem, since travellers route choice decisions are primarily a function of route travel times that are determined by traffic flow, which is itself a product of route choice decisions. This interrelationship between route choice decisions and traffic flow forms the basis of route choice theory and model development. To begin modeling traveller route choice, a mathematical relationship between route travel time and route traffic flow is needed. Such a relationship is commonly referred to as a highway performance function. The most simplistic approach to formalizing this relationship is to assume a linear highway performance function in which travel time increase linearly with speed. An example of such a function is illustrated in fig. 7.5 in this figure, the free-flow travel time

Fig(7.5)

Fig(7.6)

Refers to the travel time that a traveller would experience if no other vehicle were present to impede travel speed. This free-flow speed is generally computer assuming that a vehicle travels at the speed limit of the route. although the linear highway performance function has the appeal of simplicity, it is not a particularly realistic representation of the travel time/traffic flow relationship. This parabolic speed/flow relationship suggest a nonlinear highway performance function, such as that illustrated in fig 7.6. this figure shows route travel time increasing more quickly as traffic flow approaches capacity, which is consistent with parabolic relationship presented in Chapter 5. Both linear and nonlinear highway performance function will be demonstrated, through example, using two theories of travel route choice; user equilibrium and system optimization.

Theory Of Equilibrium
In developing theories of traveller route choice. Two important assumptions are usually made. First, it is assumed that travellers will select routes between. Origins and destinations on the basis of route travel time only. This assumption is not terribly restrictive, since travel time obviously plays the dominant role in route choice, but other more subtle factors that may influence route choice(e.g., scenery), are not accounted for. The second assumption is that travellers know the travel times that would be encountered on all available routes between their origin and destinations. This is, potentially , a strong assumption, since the traveller may not have actually travelled on all available routes between an origin and destination and may repeatedly(day after day) choose one route based only on the perception that the travel times on alternate routes are higher. However, in support of this assumption, studies have shown that travellers perception of alternate route travel times are reasonably close to actual observed travel times with these assumptions, the theory of user equilibrium route choice can be operationalized. The rule of choice underlying user equilibrium is that travellers will select a route so as to minimize their personal travel time between their origin and destination. User equilibrium is said to exist when individual travellers cannot improve their travel times by unilaterally changing routes. Stated differently user equilibrium can be defined as the travel time between a specified origin and destination on all used routes is equal, and less than or equal to the travel time that would be experienced by a traveller on any unused route.

Example (7.9) Two routes connect a city and a suburb. During the peak-hour morning commute, a total of 4500 vehicle travel from the suburb to the city. Route 1 has a 60mph speed limit and is 6 miles in length route 2 is 3 miles in length with a 45-mph speed limit. Studies show that the total travel time on route 1 increases two minutes for every additional 500 vehicles added. Minutes of travel time on route 2 increase with square of the number of vehicle expressed in thousands of vehicle per hour. Determine user equilibrium travel times.

Fig(7.7)

Mathematical Programming Approach to User Equilibrium

Theory of System Optimal Route Choice

PRACTICAL ISSUE IN TRAFFIC FORECASTNG

As mentioned earlier, the objective of this chapter was to convey the basic principles underlying traffic forecasting. One can readily imagine that the actual application of the various method presented in this chapter to a real-world traffic forecasting problem requires that a number of important practical issue be addressed. Such issue include:

The geographical definition or origin and destination areas, The definition of a household as a decision-making unit, The possibility of trip chaining, The large number of alternate routes available between most origin and destination, Specifying appropriate route performance function, The equilibrium problem in nonroute choice models The long-term effect of travel decision on residential and commercial location

METHOD OF LEAST SQUARES REGRESSION