I read an article recently discussing the merits of 50 mph speed restrictions within road works, as opposed to 60 mph, arguing that drivers on average would be less stressed in the higher speeds, so may be safer, and that got me thinking.
The Highways Agency are responsible for motorways and major highways in the UK, to be run economically, safely and to promote the well-being of those that use and maintain those roads.
I’m a Civil Engineer by training, a Highways Manager by initial experience, and a Transformation Specialist, anchored on W E Deming’s ‘Continual Improvement’ (CI) philosophy.
CI has two core components, a SoPK and Deming’s Chain Reaction, where SoPK embraces four core requirements: an Appreciation of a System, Knowledge about Variation, a Theory of Knowledge and a Knowledge of Psychology. Deming stated that: “The various segments of a system of profound knowledge proposed here cannot be separated. Thus, knowledge of psychology is incomplete without knowledge of variation.”
A highway network is incredibly complex, consisting of every combination possible from brand new / well maintained to extremely old / falling apart. From very well designed and built to very poorly designed and built.
In general; we use single status rules to run these assets, 70 mph on all dual carriageways, 60 mph on most other major highways and 50 mph where there are road works or hazards.
Are these generic concepts correct for all purposes? Why only think in tens? What’s wrong with 55 mph? If 70 mph is good for fine sunny days, is it good for icy cold nights? If the speed limit is 50 mph how many motorists comply with that with and without average speed cameras.
From a ‘Knowledge of Variation’ point of view, there will be differing statistical realities, for the same highway sections, at different times of the day, week and the year.
During commuter periods on Motorways, there tends to be far more vehicles per hour, travelling pretty fast, technically too close together, with most people being predictable about ‘road sense’. Overnight tends to have a greater ratio of HGV’s with cars travelling a little faster than average, with a few outlier very fast cars. The greatest risks coming from inattention due to tiredness.
During holidays and weekends, there will be the greatest variation of driver styles, with some occasional drivers being less predictable in their road-craft.
A Knowledge of Psychology is also interesting in these scenarios, for instance the relatively new highway laws regarding ‘lane hogging’ has produced two equally dangerous thinking patterns, the first is for people who believe their speedo is accurate, and once they are showing 70 mph they simply will not change lanes.
I’ve calibrated my speedo’s using sat nav equipment, my previous car would have a road speed of 64 mph when the speedo showed 70, where-as my current car is doing 68 when it shows 70. Too few people understand that a speedo is a guide, it’s not statistically reliable.
Then there’s the ‘lane-swervers’, driving at speeds in excess of 85 mph quite often, then moving from the outside lane, to middle to left hand lane, then back again, as they consciously comply with the ‘non-blocking’ requirements, despite being 20% above the speed limit.
Slow blockers cause a string of drivers to change lanes to get past, while reducing the capacity of the highway, while the high speed swervers are difficult to see and predict if changing lanes your-self.
To be able to drive safely, reliably, economically and environmentally soundly, requires slightly different driving styles in those different driving situations, prior to considering weather conditions.
Road designs.
The capacity of a system depends upon the sum of all the parts in a system, and how they interact at different times.
Understanding a complex system requires an appreciation of theories, modelling of options, and an appreciation of the impact of changes upon the psychology of the road users.
Take the M25 for instance: When the designs for the M25 were being completed, the transport modelling team I was working with were asked to test out the capacity to carry traffic. We modelled the proposed highways, against the existing transport network, and discovered that with no increase in overall traffic, many sections of the M25 would NOT be able to cope with demands that existed at that time.
The government refused to accept our calculations, and built the M25 as designed. The reason they wouldn’t accept our designs was because they had a flawed theory (or concept), which was the purpose of Motorways is for ‘long distance travel’ while other highways are for shorter distances.
Our predictions, showed that many vehicles would use the Motorways for relatively short runs, covering one to three junction lengths then return to more local roads. The net impact of increasing capacity would be to attract vehicles into those areas, up to taking the same net time to travel, enabling other vehicles to then utilise the slightly better options, leading to the same amount of vehicles being used a little more.
Had we been believed, many parts of the M25 would have been built as 4 lanes or more, which would have been massively cheaper than retrospectively increasing capacity, while that traffic is in flow.
The capacity of the majority of a system is greater than a few pinch points.
If you look at some local networks, such as the A43 south from Brackley, via the M40 to Bicester, onwards to the A34 towards Oxford.
The A43 and A34 are both two lane dual carriageways, the A43 has relatively few roundabouts, while the A34 has no roundabouts on the main through-way, as junctions are grade separated. These roads travel roughly North East to South West and have two lanes in each direction. They merge with the M40 which is nearer North West to South East, and is three lanes wide.
For about five miles the five lanes of traffic are merged onto three lanes of Motorway. The A roads are nearer full capacity than the motorway either side of this joint flow. However, five into three doesn’t go and that stretch of Motorway is always congested during peak hours, with a very high rate of accidents.
The A43 roundabout link to the M40 (junction 10) has been redesigned twice in recent years, to a system that had a high proportion of over-lapping right turners causing congestion two ways, to a complex design causing lost capacity south-bound. A simple approach that was grade separated would have enabled a far greater M40 access capacity, thereby attracting more traffic onto the A43 and M40.
In addition to carrying ‘five-lanes-worth’ of traffic on the M40 south from J10 (Brackley) to J9 (Bicester) on only three lanes, there’s the additional problem of the A43 roundabout over the M40. It simply hasn’t the capacity to take a very high proportion of right turn traffic.
Thus the inside lane of the M40 often tails back over two miles along the M40 towards J10, leaving at best two lanes for transport aiming onwards towards London. The queue capacity could be increased by lane widening, or the queue averted by a flyover link straight onto the A34, which has already the indignity of being near capacity.
On bad days the tail-back is much greater, leading to middle lane queuing, with only one lane available for straight through traffic towards London. On those days, the whole Motorway backs up.
The psychology of some drivers is to always seek to ‘beat the system’, so when the near side lane is backing up to get off at J9, some will stay in the middle lane until only a 100m from the junction, then slow down and merge into the slow lane. Thus running the risks of being either rear-ended in lanes one or two, or indeed running into the back of a vehicle in lane one.
Clearly a system which is not designed to optimise flow or capacity. Each part of the system is poorly designed, not considering the whole network.
There are a few junctions on the M40 where a 3 lane motorway reduces to 2 lanes between the exit slip and the entry slip roads. If 1/3rd of the traffic exits reliably at those junctions during peak-flows, then all is well. However, all too often only 10% goes off, leaving 90% of three lanes with only two lanes capacity. This is a real penny pinching stupidity, saving a little during construction, then costing millions in delays over the next 50 years!
Roadworks designs.
The M1 between J15a (Northampton) and J19 (A14 – M6) has been in a near state of permanent roadworks for the last four years. The first phase was to replace the previous Armco barriers with the now in-vogue triangular section reinforced concrete, followed by the introduction of ‘smart-motorways’ which will add another lane for most peak travel times, with overhead gantries for lane and speed management.
Hence about sixteen miles of both North and South-bound carriageways have had 50 mph restrictions, with lane widths reduced, and far more accidents and congestion than normal. So from ‘Economic, Health and Safety and Wellbeing’ (EHSW) points of view, is this optimal design, or one that feels right?
For instance, why was central reservation changed, Motorways re-opened and then Smart Motorways introduced? Would both together had been less costly from all four perspectives? E-H-S-W.
Economically doing both at the same time would have had shared use of the highway management systems, probably taking only 66% of the time that doing them in series takes, with perhaps only 75% of the total cost to construct. Then add in the cost of delays on the economy and two separate is probably twice as dear as one combined.
Then: is three narrow lanes over 16 miles all in one go REALLY VfM, or an illusion of cheaper? Evidentially most of the system is waiting for work over that ’32 mile length’. At times there have been less than 3 people actually seen working over a 16 mile single run. Thus most of the signage and 95% of the slow narrow constraint, is of no value to the motorist at that time.
Would more radical temporary traffic management regimes work better? For instance, closing one side for 1/3rd of the length, say J16 to J18 . Have two lanes width of free-access high intensity works, with two narrow same direction flow, plus one lane of traffic contra-flowing the opposite direction.
End to end times are better, less congestion, less narrow lane driving, lower traffic management costs….
One aspect to cover is ‘a Theory of Knowledge’. Borrowing from Professor Russell Ackoff, there’s a hierarchy of Knowledge: Data, Information, Knowledge, Insight – Wisdom. I’ve utilised that to create D>I>K> Tactics.
If you have excellent ‘Knowledge’ of a system, then you can model and make predictions to underpin future state designs, not only of the final design method, but also how to achieve that.
To design the ‘future state’ of a Motorway, many, many aspects are weighed up in an Options Appraisal, starting from the current designs, land ownership, constraints etc., then the issues to overcome or the outcomes to be achieved.
Swapping the focus around to ‘Safety’, what if the purpose of the works were to transition from a peak traffic capacity of 10,500 vehicles per hour, to 13,500, with fewer accidents, and less congestion, with the works minimising risks to the workforce and travellers during work.
There’s a lot of ICT in all this: Computer Aided Design (CAD) for the construction designs and temporary works, Average Speed Cameras to control excessive drivers, Automatic Number Plate Recognition, (ANPR) to identify individual vehicles, including their legal speeds (i.e. Large Good Vehicles maximum = 56 mph normally not 70 mph), Urban Traffic Control (UTC) where traffic congestion is monitored and managed from.
All these data sources start to weave together a ‘rich picture’ of knowledge, including the abilities to automatically model current traffic demands, origin and destination demands, speeds per lane, percentages of traffic between speed X and Y, creating awareness of reality against legality.
However, what if we added some aspects together, perhaps utilising Artificial Intelligence (AI) into the picture, to morph information together?
Consider the possibilities for both designing how features of the network will be used, and then the implications during different states of the construction.
Let’s say this was to create a ‘smart motorway’ using a split two lane – one lane contraflow.
Starting with lane awareness utilisation information, the number of ‘standard axles’ per day, year or 25 years could be calculated for each lane, demonstrating that the nearside lane needs to be far stronger than the off-side one.
Lane widths could be greater for the nearside, and narrower for the offside, reflecting vehicle types.
Transitions from lane to lane could be mapped, to give frequency activities which vary between ‘featureless lengths’ compared to those with junctions and services.
From real time traffic observations, mixed with accident and near miss information, the root causes of accidents could be modelled in differing traffic scenarios, such as peak flow, off-peak, over-night and ‘Sunday drivers’.
Risks change with use and demands.
In this way, not only could the designs have optimal costs / safety options designed in, but also fine tuned in many ways.
Then the build cycle could have differing factors of risk assessed, for that type / element of construction, at that time of week. Some temporary works will be far more effective overnight or at weekends, than at peak times, while some others are potentially cost / safety neutral.
Thus, instead of a one size fits all safety design, variable designs would be utilised at varying times of the week.
Put quite simply, a vehicle usage model, overlaying the CAD permanent and temporary works design, could be interfaced via AI to predict the least worst options for any scenario that can be thought of.
Then modelling of road-work traffic speeds could be applied, utilising known traffic characteristics, perhaps having stretches of highway with virtually no current works at 60 mph, while high intensity works could be reduced to 45 mph. Thus increasing average speeds, while reducing accidents.
During transitions, that is a change in temporary traffic management, there’s no reason why a 40 mph speed shouldn’t be applied, with say 55 mph reinstated once the risk has reduced.
Safety by design?
Considering all that ICT that is gathering data, creating information all the time, what if another function of AI were applied to vehicle movement monitoring?
We’ve all seen vehicles weaving due to either a distraction in the vehicle, or the driver nearly falling asleep, and yet there’s little we can do on Motorways other than give them a wide berth.
What if you used similar systems to Face Recognition or ‘Suspected terrorists’, that is you used the vehicle monitoring data to detect suspicious driving characteristics, the AI could probably differentiate between novice driver, sleepy driver and distracted.
If the ‘body language’ of the vehicle gives cause for concern for either their own safety or that of others, then overhead signs and road side signs, plus the transport police could be used to direct the driver off the highway at the first convenient opportunity.
The recent two lorry and one minibus catastrophe near Luton on the M1 would have probably have been avoided if this technology was in place.
Have you noticed new junction designs sometimes being set up, with more lanes and far higher capacities, with far lower congestion, being in place for a few weeks, then being altered to have less lanes, some hatched out areas, and then a return to greater congestion?
This is due to a two-stage design! 1) Design, build, normalise traffic; 2) Carry out safety audit, alter design.
I’ve seen some of these create over 800M2 of new carriageway, which then simply becomes hatched white line areas. That’s 800M2 of brand new carriageway built and paid for that will never get used.
If you integrated the CAD designs with some simple traffic flow models, then you could build what will get safely used, saving 800M2 of needless costs.
Better designs are not only safer to use and safer to build, but also sensible for the economy!
Thinking a little wider.
Traffic is often slowed around schools to 20 mph, thus a special cause required for 192 days a year, for 1½ hours a day, is put in place 24 / 7. That’s a requirement for 3.3% of the year, applied for the whole year. In Scotland, they have 20 mph ‘when the yellow lights are flashing’, the rest of the day is 30 mph or more!
Lastly, the Deming Chain Reaction: Improve quality, costs reduce…. More can be done!
Dave Gaster
October 2017