National Highways has refused to disclose how many of its failing smart motorway safety systems have nevertheless been cleared for permanent operation, increasing the risk for drivers on all lane running (ALR) routes.
When National Highways first planned the roll-out of radar-based stopped vehicle detection (SVD) for ALR smart motorways it set a clear testing process for each individual installation to pass before it could be officially declared 'business as usual'.
It now says it has 'determined that the system has improved sufficiently' at a national level to be passed into business as usual, despite the fact that individual schemes have failed to hit benchmarks.
The testing framework is important as it allows ALR schemes to operate at 70mph indefinitely – and a failure to pass the performance benchmarks has previously been cited as a relevant factor in a fatal crash.
In 2022 National Highways’ executive director for operations, Duncan Smith, told Highways that a fatal crash on the M4 – when the stopped vehicle detection (SVD) system had failed to alert operators to a stopped vehicle – had happened when ‘the scheme was still in operational acceptance’ and hadn’t yet ‘been handed into business as usual’.
Core requirements for SVD are:
- to detect 80% of stopped vehicle ‘events’
- for detection to take place within 20 seconds
- for the proportion of false detections to be lower than 15%.
If these requirements are not met, testing should continue 'with a cycle of root cause analysis, implementing improvements and re-testing' until they are.
In January 2023, the company told Highways that none of its SVD schemes had transitioned to business as usual ‘as they are not yet meeting all of the core performance requirements’.
Last year, following a software upgrade, National Highways retested all 21 schemes having changed its methodology to the point where it based its calculations on just 15% of stopped vehicle events.
Its monitor, the Office of Road and Rail (ORR), reported in December that the system was achieving an 89% detection rate nationally.
However, one region fell below the 80% benchmark and last week, Highways revealed that five of the 21 individual schemes had fallen below the detection benchmark. If you factor in the requirement to detect incidents within 20 seconds only around half make the grade.
Following the ORR report, Highways asked National Highways how many individual schemes had passed into business as usual.
It did not answer the question directly but stated that ‘the treatment of alerts, automatic sign setting and our approach to faults is all in line with our normal ways of working’, suggesting that SVD systems that have not passed testing are being operated at 70mph – apparently indefinitely.
National Highways has said that the testing process has not changed. However, it appears to have moved from a scheme-by-scheme basis to a system basis.
A spokesperson told Highways: ‘We’ve determined that the system has improved sufficiently at national level to allow [handover to business as usual]. This has been the message from ORR also.’
The spokesperson argued that ‘individual tests are indicators at a moment in time’ and that ‘with the small sample sizes, the results would be different tomorrow’. National Highways insisted that the sample size for individual schemes is too small to make an informed decision on efficacy.
A National Highways spokesperson added that testing will be repeated annually 'to continuously monitor and improve'.
The sample sizes of stopped vehicle events in the 24-hour window for ground truthing range from nine to 48 incidents for each of the 21 schemes, giving a total of 564. However, this is only after 3,286 events were thrown out under the new methodology.
While the ORR appears to agree on the point about the sample size, it told Highways: ‘[The] ORR has no role in approving how or when National Highways passes technology for business as usual.’
It declined to say how it decided it had no role in the process, despite its significant responsibility for overseeing safety on the strategic road network.
The ORR also pointed out that its December 2023 report had ‘called on the company to consider sample size and frequency of testing’.
Alan Hames, an experienced traffic engineer who gave evidence to a parliamentary inquiry into smart motorways, told Highways that in his opinion, the testing procedure must be fully complied with and signed off before any scheme can be declared as business as usual.
He described the process of declaring business as usual status by aggregating the results of schemes as ‘of great concern as it does not make clear and public this system is hiding failures, then allowing the signing off of a proportion of locations known to be under par’.
Recognising that the effectiveness of radar-based systems is site specific and subject to geographical and topographical features, he added: ‘All locations should only reach business as usual on an individual basis, site by site.’
