Modelling Safety-related Driving Behaviour: Impact of Parameters Values



Modelling Safety-related Driving Behaviour: Impact of Parameters Values

Authors

R Liu, P Bonsall, ITS, University of Leeds, UK

Description

Abstract

h4. Introduction

Traffic micro-simulation models make assumptions about the safety-related behaviour of drivers. The question is whether these assumptions should reflect safe behaviour or actual behaviour. It is clear that, in reality, drivers often engage in seemingly unsafe behaviour (e.g. running red lights at signalized intersections, or following cars too closely on motorways) in order, for example, to reduce their journey time.

Should models seek to replicate such behaviour? On the one hand it might be held that models should be as close to real life as possible. On the other hand, it could be thought unethical to design a scheme based on a tool which assumes unsafe behaviour if this could lead to the adoption of designs which are known to be unsafe.

In exploring the issue, we first question where the key parameters in well know traffic micro-simulation models have come from and whether they represent real behaviour or some idealised safe behaviour. We then investigate the sensitivity of model predictions to the value of key safety-related parameters and discuss the whole question of the representation of unsafe situations in traffic microsimulation models and their consequences. 2. Safety-related parameters in traffic simulation models Vehicle interactions and drivers? behaviour are generally represented in microsimulation models through car-following, gap-acceptance, lane-changing models.

Car-following models represent the longitudinal interaction between vehicles. The acceleration of the following vehicle is modified in the light of the relative speed and position of the preceding vehicle. The parameters required to determine such longitudinal progress include: the desired speed, the desired minimum headway, the reaction time, the rate of acceleration and the rate of deceleration.

Gap-acceptance models are used to determine how a vehicle from a low priority flow will cross, or merge into, a higher priority flow. The key parameter for such models is the critical acceptable gap for the manoeuvre being contemplated. Some models allow the critical gap to be reduced to represent the frustration of the waiting traffic in presence of heavy traffic ? with parameters used to indicate the stimulus required to induce use of the reduced gap and the reduced gap itself. Others allow for the fact that vehicles in the priority flow may deliberately slow down in order to create gaps for the low priority flow ? in which case at least one parameter will be required to indicate their willingness to create gaps.

Lane-changing models consider the individual driver?s intention and ability to change lanes. The driver?s intention to change lanes may be triggered when the time advantage to be gained by changing lanes exceeds some critical value. Some models may allow drivers to anticipate the need for a change of lane, with parameters required to determine how far ahead the drivers anticipate. The ability to change lanes is generally modelled in a way which is analogous to a gap-acceptance model.

The paper will present a full list of the parameters identified, indicate commonly adopted values and the sources of those values. 3. The consequences of changing the value of safety-related parameters 3.1 The consequences of ?unsafe? driving It is not difficult to think of a range of unsafe driving behaviours, such as:
* adoption of inadequate headways in fast moving traffic;
* speeding (in excess of the legal limit or in spite of local circumstances);
* excessive reliance on the vehicle?s brakes;
* reckless overtaking (e.g. where sight-lines are inadequate);
* passing traffic signals at amber (or even red).

We will consider the first of these in a little more detail. Behavioural research based on carefully observed experiments in laboratories and on the road suggests that drivers? minimum reaction times lie in the range 0.8-2.0 seconds. Allowing for typical reaction time and stopping distance, a safe headway on a 110 kph road would generally need to be in excess of 4 seconds and yet headways as low as 0.5 seconds have been observed on such roads. Recent research by Oates (1999) suggested that almost 50% of drivers on a congested stretches of a motorway were driving with headways at or below two seconds and that almost 25% were driving with headways at or below one second.

Simple calculation indicates that, in smooth conditions and constant speed, the flow achievable with a 0.5 second headway would be about eight times that achievable with a 4 second headway.

Similar arguments can be made in respect of each of the cases mentioned above. Unsafe driving will generally lead to enhanced system performance but when, as is inevitable, there is an incident, the results can be catastrophic.

However, provided that incidents remain relatively rare events, it is possible to conclude that if everyone were to drive in strict accord with guidelines and regulations, the effective capacity of the network would be reduced below the levels currently observed.

Various researchers have commented on this rather uncomfortable fact. For example, in his study of signalised intersections, Pretty (1974) found that traffic signals improved capacity at an intersection previously under police control only because drivers used the amber and all-red periods. 3.2 Analysis In order to illustrate the general argument made above, the DRACULA traffic micro-simulation model (Liu et al, 1995) was used to explore the impact that changes in key behavioural paramete rs might have on various model estimates of system performance. The results reported here relate primarily to the total travel time in the test network since this is the indicator of system performance most widely used to inform investment decisions.

One of the tests conducted was designed to show how assumptions about the distribution of one aspect of aggressive driving (in this case the rates of acceleration and deceleration) can affect the predicted performance of a scheme. The tests relate to the introduction of partial signalisation at a roundabout just off the M25 near Heathrow Terminal 5. The mean values of the acceleration/deceleration distributions used in the tests are shown in Table 1 (note that traffic at the site is 10% HGV and 90% car). The tests relate to two flow level scenarios; a current flow and a future flow at twice the current level ? as can be expected when the new Terminal opens.

The results of the tests are shown in Table 2. As expected, the effect of the signalisation scheme is dependent on the level of flow. At the current (low) flow levels, the signalisation would lead to 5-11% increase in journey times. With increased flow, however, the signalisation scheme leads to a very marked reduction in journey time (59-64%).

The assumed level of acceleration/deceleration affects the predicted impact of the scheme. If a more aggressive level of acceleration/deceleration is assumed, journey times are reduced (ranging from 2% to 25%). The reduction is more significant under normal priorities and under the high flow scenario (-25%).

The net result is that the assumption of more aggressive acceleration/deceleration causes a doubling of the disbenefit associated with signalisation under current flow conditions.

It is clear that the assumptions about levels of acceleration and deceleration can profoundly affect the prediction of scheme benefits and that this effect differs according to the flow level. 4. Summary The paper identifies the key parameters of traffic simulation models and notes that several of them have been derived from theory or informed guesswork rather than observation of real behaviour and that, even where they are based on observations, these may have been conducted in circumstances quite different to those which now apply.

Tests with DRACULA demonstrate the sensitivity of model predictions ? and perhaps policy decisions ? to the value of some of the key parameters. It is concluded that, depending on how the model predictions are being used, use of realistic-but-unsafe parameter values could result in the adoption of unsafe designs. However, it is noted that this problem can be overcome by paying attention to the safety aspects of designs and that, in general, the use of realistic values is to be preferred. The possibility of using traffic simulation models to produce estimates of accident potential is discussed and the difficulties involved in doing so are discussed.

h4. References

Liu, R., Van Vliet, D. and Watling, D. P. (1995) DRACULA: Dynamic Route Assignment Combining User Learning and microsimulation. Proceedings of PTRC Summer Annual Conference, Vol E, 143-152.

Oates, A. (1999) A study of Close Following on the M62. MSc Dissertation, ITS, University of Leeds.

Pretty, R. (1974) Police control and intersections. Proc. 7th Australian Road Research Board Conference, 7(4), 83-95.

Publisher

Association for European Transport