SCOOT Introduction How SCOOT works SCOOT Versions Where is SCOOT used Results and case studies SCOOT Reference SCOOT Training User stories Contact Us

Frequently Asked Questions

The aim of this page is to answer any questions that you might have about SCOOT, it's facilities, benefits or uses. If you have a specific question, please check those listed below. If your question has not been answered please fill out the form at the bottom of this page. Questions will be answered directly and also included on this page, if felt appropriate.

  1. How does SCOOT decide how much to increment or decrement the cycle time by?
  2. And, how does SCOOT decide whether to advance, stay or retard the offset?
  3. Can two lane loops be connected in series to a single feeder? If not, can one large two lane loop be used in place of two single lane loops?
  4. What is the difference between SCOOT and SCATS in terms of working principle, implementation, efficiency etc.?
  5. Which is optimised first, offset or split?
  6. What faults can occur in the system, how are they dealt with and does it affect the traffic? How is the system protected?
  7. How does SCOOT measure vehicle emissions? And how is that data presented?
  8. Is there a maximum distance for the detection from the stopbar? And is there a minimum (very close spaced intersections)?
  9. If we have 3 lane road is this still one loop for all?
  10. How do you handle sink/sources that are mid-block? Mall/movie theatre entrance/exit, etc.
  11. Are regions static throughout the day?
  12. Are cyclic flow profiles static or dynamic? Does it learn them and keep them for next Monday, etc? Are Holiday, school out, exceptions accounted for?
  13. Is validated saturation flow static?
  14. What happens if you can't get the saturation to 90%? Is there an upper limit on cycle and how does it then choose splits and offset?

How does SCOOT decide how much to increment or decrement the cycle time by?
Kruno Peric (Rutgers University)

SCOOT changes the cycle time in two steps. It first decides whether to increase or decrease the cycle time and then it moves it to the next SCOOT cycle time in the correct direction.

SCOOT aims to run each node at a saturation of 90% or less. In general SCOOT will run at the minimum cycle time compatible with this objective.

SCOOT accepts a discrete range of cycle time values. These are from 32 to 64 in steps of 4 seconds, from 64 to 128 in steps of 8 seconds and from 128 to 240 in steps of 16 seconds. So 60 and 120 are valid SCOOT cycle times, for example.

And, how does SCOOT decide whether to advance, stay or retard the offset?
Kruno Peric (Rutgers University)

The SCOOT offset optimiser models the effect of an advance, stay and retard decision on the total delay and stops at the node and makes the decision which minimises these.

Can two lane loops be connected in series to a single feeder? If not, can one large two lane loop be used in place of two single lane loops?
D. Ireland (eThekwini Municipality)

Yes it is possible to connect two loops to a single feeder. It is also possible to either use one loop per lane, or one large loop over two lanes. It should not have any effect on the operation of SCOOT. In either case, the correct guidelines need to be followed on the size of the loop.

What is the difference between SCOOT and SCATS in terms of working principle, implementation, efficiency etc.?
Mallikarjuna Setty (Gherzi Eastern Limited)

In order to answer this complicated question, a table has been created outlining the differences between SCATS Version 7 and SCOOT Version 4.2 (the most recent versions of each, as of the 16th January 2002).

Topic

SCATS Version 7

SCOOT Version 4.2

Working principle

SCATS relies upon plan selection with some local adaptation.

With SCOOT, the entire network is modelled, and therefore, requires no fixed plans. SCOOT uses current detector data to model the cyclic flow profile, similar to TRANSYT. The cyclic flow profiles are created on-line and are updated every four seconds.

System layout

The system is set up as a series of regional computers, which control sub-systems. Each sub-system is a collection of junctions, of which, one is designed to be the critical junction. The cycle time and splits are calculated for this critical node and then plans are selected for the other junctions, based on these calculations.

A SCOOT system consists of many regions, all controlled on one central computer, with each region containing a number of junctions.

Cycle time optimesed

Each cycle.

Every 2.5 or 5 minutes.

Split optimised

Each cycle.

Each stage.

Offset optimised

Not optimised - offset is selected from plans calculated off-line.

Each cycle.

Optimisation performance

SCATS calculates green splits based on the flow in the previous cycle and so, may be unresponsive to unpredictable arrival flows.

Using the normal SCOOT loops, optimisation is based on the real queue at the signal and therefore more responsive to unpredictable arrival flows.

Detection

Vehicles are detected by two types of detector - one near the stopline to measure flow and occupancy and (sometimes) one upstream, to measure queuing.

SCOOT only needs one detector per link, in an upstream location, to detect flow, occupancy and queuing. This allows SCOOT to monitor congestion in the network.

Traffic data available

Very little traffic data.

A large amount of data is available, including flows, delays and occupancy.

Setting up the system

SCATS needs the critical node in each subsystem identified along with plans to use at these nodes. These plans can be extremely time consuming to derive and need to be updated regularly in order to maintain satisfactory operation. Typically subsystems contain 1 to 5 intersections.

SCOOT needs model information, such as the cruise time from detector to stopline. These parameters, once entered, do not need to be updated.

Trial results

SCATS has been installed in many locations, but it has been suggested that the success of the system relies on the skill of the traffic engineers that set up the system.

Many independent trials, including five cities in the UK. Many other authorities have also carried out their own studies.

Benefits

A survey carried out by the Australian Research board showed no significant reduction in journey times compared with using TRANSYT. There was, however, a large reduction in the number of stops (around 9% in the central area and 25% on the arterial roads).

SCOOT has been installed in many cities. The effectiveness of SCOOT has been assessed in many different ways, both by TRRL (now TRL) and by other authorities, testing their own system. Five cities took part in assessment of SCOOT. These results show overall a reduction in delay of about 12% compared with good fixed time plans. The benefit increases as fixed time plans become out of date.

Which is optimised first, offset or split?
H First (Tonji University, China)

Since SCOOT works on a continuous basis, a value for split and offset is always available. SCOOT simply uses the current value and updates them for each junction. It updates the offset every region cycle time and the split, every stage change.

What faults can occur in the system, how are they dealt with and does it affect the traffic? How is the system protected?
Samantha Riches

TRL maintains a database of software faults reported, with a record of the faults and dates fixed. No major faults have been reported since the database was created, many years ago. Siemens and Peek are involved in reporting faults to TRL, who are responsible for fixing any bugs in the SCOOT kernel.

The majority of system faults, as in any adaptive system, tend to be detector failures. These can effect traffic, however SCOOT has very good default logic which should minimise any adverse effects.

How does SCOOT measure vehicle emissions? And how is that data presented?
Olivia Henderson (Babtie)

The process is that for each link SCOOT models delay as normal. The delay is added to the journey time and the result used to calculate the average speed using the link length. The speed is used to calculate the mean emission rate/vehicle/distance/time along the link for each of four categories of vehicle (petrol no catalyst, with catalyst, light and heavy diesels). The flow together with standard, or link specific, proportions of vehicles in each class provide the number of vehicles in each class. Multiplying the rate, by the number, by the full link length and summing over classes produces the result in mass per unit time for the link.

Emissions as mass per hour are output by link, node or region as desired by the user. The emissions estimated are carbon dioxide, carbon monoxide, oxides of nitrogen, particulate matter and volatile organic compounds.

Is there a maximum distance for the detection from the stopbar? And is there a minimum (very close spaced intersections)?
Mike Whiteaker (City of Bellevue, WA)

There is not a fixed maximum distance for detection from the stopline (stopbar). The effective limit is that junctions should be close enough together that there is a benefit from coordination. If there is, then detection at the upstream end of the link will provide that benefit.

If junctions are very close together then special arrangements may be necessary. They can be treated as one node with a rigidly fixed offset for the internal link, or they can be modelled separately with a "biased" offset holding the internal offset close to a fixed value. Closely spaced junctions are normally assessed individually, but as a guide, if they are closer than 50 metres then generally they would be assessed to see whether they should be treated normally or considered as a special pair. SCOOT will work well in compact networks; it was developed in Glasgow where the central grid has an average link length of 110m.

If we have 3 lane road is this still one loop for all?
Mike Whiteaker (City of Bellevue, WA)

A single detector is not sufficient on a 3-lane road. SCOOT is designed to deal with masking from using a single detector over two lanes, but the masking problem with more than two lanes is too great. The rule is a minimum of one detector per two lanes. It will work well with a detector for every lane, but the cost goes up!

How do you handle sink/sources that are mid-block? Mall/movie theatre entrance/exit, etc.
Mike Whiteaker (City of Bellevue, WA)

If there are major sources or sinks of traffic mid-block then it will be necessary to site the detectors downstream of the sources and sinks. The prediction will not be as good as from a detector at the start of the link, but neither will the coordination because of the disruption from additional uncoordinated traffic from a source, or gaps in the platoon from a sink. Because mid-block detectors are further from the nearest controller, feeder costs will increase. There is not a fixed rule about when to relocate detectors. If the sources and sinks account for less than 10% of traffic then no action is likely to be needed, if they account for more than 30% then the detectors will very likely need to be moved. In between it depends how uniform the effect is. If traffic enters or leaves at about the same proportion throughout the day, then it is less important than if the effect varies markedly at different times of day.

Are regions static throughout the day?
Mike Whiteaker (City of Bellevue, WA)

Regions can be changed by timetable.

Are cyclic flow profiles static or dynamic? Does it learn them and keep them for next Monday, etc? Are Holiday, school out, exceptions accounted for?
Mike Whiteaker (City of Bellevue, WA)

The cyclic flow profiles are dynamic they are updated every 4 seconds (1 second for the most recent data). The short term profile holds data from the last cycle only, the long term profile has more smoothed data, but again updated every 4 seconds. They are not learnt and retained, SCOOT uses current data. However, historic data can help predict the onset of peak periods etc. by showing how the demand will increase in the immediate future. To do this, the optional ASTRID database will hold a profile of demand throughout each day of the week and the cycle time optimiser will use this information as well as current demand on the link to predict the need to increase the cycle time. If the current demand is much lower than the historic one at this time of day, then SCOOT factors down the historic data for future times on the assumption that today is not a typical Monday, e.g. a public holiday.

Is validated saturation flow static?
Mike Whiteaker (City of Bellevue, WA)

Validated saturation flow (saturation occupancy in SCOOT units) is normally static, although it can be changed by timetable, for instance to allow for a mid-block source of traffic that is only active at one time. There is a facility, SOFT, that can be used to dynamically adjust the saturation occupancy where all the downstream links have detectors in the normal SCOOT position and are not subject to congestion. SOFT relies on all the traffic from the link being detected on downstream links.

We have found that normal rain and sun do not affect traffic behaviour sufficiently to need special action. Snow could require special action, however, if unexpected queues build up, SCOOT will respond to the resulting congestion. The simplest way to respond to severe conditions caused by snow would be to have a special data set. If you know that typically in snow cruise speeds are x% of normal and saturation flows are y%, then it would be simple to produce a data set with all journey times reduced by x% and saturation flows by y%; an easier task than preparing special fixed time plans. Similarly construction works can affect saturation flows. If they are sufficiently serious and long term, then a simple measurement (re-validation) of the saturation flow on the affected link is all that is required; much simpler than recalculating fixed-time plans for all periods. SCOOT does all the re-optimisation.

What happens if you can't get the saturation to 90%? Is there an upper limit on cycle and how does it then choose splits and offset?
Mike Whiteaker (City of Bellevue, WA)

The objective of the cycle time optimiser is to run the most saturated junction in a region at 90% saturated. The reason is to leave a bit of spare capacity to cope with increases in demand between runs of the optimiser, but not to run unnecessarily long cycle times and waste green, keep pedestrians waiting longer than necessary etc. If overnight conditions are very quiet and no junction is 90% saturated on its minimum cycle time, then SCOOT will simply run the region at the minimum cycle time (the highest of the individual junctions' minimum cycle times). At very busy times it is often not possible to avoid going over 90% saturated. A policy decision is taken, as with calculating a fixed-time plan for the period, of what is the maximum acceptable cycle time and this is input as an upper limit, which the cycle time optimiser will not exceed.

Hitting the maximum cycle time has no effect on the operation of the split and offset optimisers. The split optimiser will share out the available green equitably as normal and the offset optimiser will choose the offset to minimise delay.


Has your question been answered? If not, fill in the form below to get an answer direct from the SCOOT team. Click "Submit" when you have entered your question. You must enter your name and e-mail address or we will not be able to answer you!

Please note that your name and organisation will be included on any posted reply on this page. Please inform us if you would like to remain anonymous.

Name:
Organisation:
Email:

Your question:

Please type the word "orange" to help us prevent spam