An IBEC working group session at ITS World Congress in Vienna in October was presented with an example of a very cost-effective means for reducing traffic travel time. There is no doubt that adaptive real-time traffic control is a very cost-effective ITS application”, Dr Ronald van Katwijk told an IBEC (International Benefits, Evaluation & Costs) working group session at the 2012 ITS World Congress in Vienna. The senior consultant with Netherlands consultant TNO and TrafficQuest, the Dutch Centre for Expert
An IBEC working group session at ITS World Congress in Vienna in October was presented with an example of a very cost-effective means for reducing traffic travel time.
There is no doubt that adaptive real-time traffic control is a very cost-effective ITS application”, Dr Ronald van Katwijk told an IBEC (International Benefits, Evaluation & Costs) working group session at the 20126456 ITS World Congress in Vienna. The senior consultant with Netherlands consultant TNO and TrafficQuest, the Dutch Centre for Expertise in Traffic Management, said over 85% of the country’s controlled intersections currently have traffic-actuated systems, using a movement-based approach.
This means that, although groups of signals are structured in stages, they can be controlled separately. This level of flexibility is necessary given the large number of separately controlled movements needed to manage the country’s changing multi-modal traffic flows (see Table 01).
The disadvantage is that traffic-actuated systems use historic information to determine crucial parameters, such as maximum time and sequencing of green phase and coordination of offsets between intersections. Past experience does not guarantee the quality of future results, stressed van Katwijk. In addition, traffic-actuated systems suffer from ‘tunnel vision’ (when a green phase is extended for a movement without taking into account demand for other conflicting ones) and ‘near-sightedness’ (when a green phase is ended as soon as it has served current demand, even though a platoon of traffic might be approaching).
Benefits:
• €373,554 (US$478,337) per five intersections annum of driving time savings based on the modelled peak hour exercise.
• Movement-based adaptive control improves intersection performance without increasing computational requirements.
Traffic-adapted systems in the Netherlands face the central problem of – given the large number of separately-controlled movements – how to come up with an optimal solution in real-time. To address this, TNO has carried out a number of modelling exercises, applying a traffic adaptive control algorithm tailored to Dutch movement-based control practice to a diverse range of intersection and demand configurations.
An illustrative example presented at Vienna covered a 1.3km stretch of road with five intersections. The modelling took place over a two-hour peak period. The results, in terms of time savings through successive junctions, appear in Table 02.
These results lead to the following calculation of savings over a year. The current total driving time spent by all drivers over the entire network during one hour of the two-hour period is 228.89 hours. Taking the relative benefit over the network modelled, as compared with the currently operating control system, as a conservative 17% gives an absolute benefit in each hour of driving of 38.91 total hours.
Assuming a four-hour total daily peak traffic period gives an absolute benefit per working day of 155.65 driving hours saved (31,130 per year of 200 working days). An average valuation of €12.00 (US$15.37) per hour produces annual savings of €373,554 (US$478,337).
“There is a lot to be gained by looking forwards instead of backwards when determining the signal timings of an intersection. But this is computationally demanding, given real-time constraints, requiring investments in detection and computing hardware.
Movement-based adaptive control significantly improves an intersection’s performance without further increasing the computational requirements.”
Table 01.
The changing modal split in Amsterdam
Source: DIVV, Traffic & Transport Infrastructure Service, Amsterdam.
Table 02.
Results of the modelling exercise
There is no doubt that adaptive real-time traffic control is a very cost-effective ITS application”, Dr Ronald van Katwijk told an IBEC (International Benefits, Evaluation & Costs) working group session at the 2012
This means that, although groups of signals are structured in stages, they can be controlled separately. This level of flexibility is necessary given the large number of separately controlled movements needed to manage the country’s changing multi-modal traffic flows (see Table 01).
The disadvantage is that traffic-actuated systems use historic information to determine crucial parameters, such as maximum time and sequencing of green phase and coordination of offsets between intersections. Past experience does not guarantee the quality of future results, stressed van Katwijk. In addition, traffic-actuated systems suffer from ‘tunnel vision’ (when a green phase is extended for a movement without taking into account demand for other conflicting ones) and ‘near-sightedness’ (when a green phase is ended as soon as it has served current demand, even though a platoon of traffic might be approaching).
Problems addressed
The situation can be improved, he suggested, by using a system equipped to take the entire intersection into account (counteracting tunnel vision); consider future or pending arrivals (counteracting near-sightedness); and make short-term decisions on the basis of longer-term analysis that considers all available alternatives in detail.Benefits:
• €373,554 (US$478,337) per five intersections annum of driving time savings based on the modelled peak hour exercise.
• Movement-based adaptive control improves intersection performance without increasing computational requirements.
Traffic-adapted systems in the Netherlands face the central problem of – given the large number of separately-controlled movements – how to come up with an optimal solution in real-time. To address this, TNO has carried out a number of modelling exercises, applying a traffic adaptive control algorithm tailored to Dutch movement-based control practice to a diverse range of intersection and demand configurations.
An illustrative example presented at Vienna covered a 1.3km stretch of road with five intersections. The modelling took place over a two-hour peak period. The results, in terms of time savings through successive junctions, appear in Table 02.
These results lead to the following calculation of savings over a year. The current total driving time spent by all drivers over the entire network during one hour of the two-hour period is 228.89 hours. Taking the relative benefit over the network modelled, as compared with the currently operating control system, as a conservative 17% gives an absolute benefit in each hour of driving of 38.91 total hours.
Assuming a four-hour total daily peak traffic period gives an absolute benefit per working day of 155.65 driving hours saved (31,130 per year of 200 working days). An average valuation of €12.00 (US$15.37) per hour produces annual savings of €373,554 (US$478,337).
Best of both worlds
Comments van Katwijk: “Adaptive real-time control systems look forward in time (taking into account predicted arrivals) as opposed to traffic-actuated ones which are tuned on the basis of historical information. Most adaptive real-time control systems, however, can only optimise the sequence and duration of stages, not each traffic movement. Our approach represents the best of both worlds.“There is a lot to be gained by looking forwards instead of backwards when determining the signal timings of an intersection. But this is computationally demanding, given real-time constraints, requiring investments in detection and computing hardware.
Movement-based adaptive control significantly improves an intersection’s performance without further increasing the computational requirements.”
Table 01.
The changing modal split in Amsterdam
| Period | Car % | Public transport % | Bicycle % |
|---|---|---|---|
| 1986-1991 | 39 | 28 | 33 |
| 2005-2008 | 31 | 22 | 47 |
Table 02.
Results of the modelling exercise
| Intersection no | Travel time current | Travel Time - new (seconds) | Reduction |
|---|---|---|---|
| K362 | 166658 | 136268 | 18% |
| K329 | 361209 | 283943 | 21% |
| K330 | 483359 | 368995 | 24% |
| K331 | 313973 | 266731 | 15% |
| K332 | 460170 | 159176 | 19% |
