Designing For Lower Speeds

A key part of sustainable safety is speed management. In a situation where there is a significant speed differential, then separation is the order of the day. If traffic flows are low enough for integration, then low speed differentials are needed.

The Dutch will seek separate people cycling and driving if the latter is travelling at 50kph (30mph) or higher - if the UK did that, then we'd have a hell of a lot of 20mph streets and much safer main roads. Of course, 20mph with thundering lorries is also a matter for separation, but that's another discussion. 

Cycle separation from traffic on a main road in Amsterdam.

Additionally, the casual mixing of people walking and cycling creates conflict because of the speed differential and even in a low motor traffic situation, high cycle flows may well require separate space and shared paths become a nightmare. There is also the whole issue around through-cycle traffic in destination areas which I covered a while ago..

Cycle integration with traffic on a quiet city street in Utrecht.

Back to the speeds. Even on roads with a 50kph (30mph) limit, there is still the need to keep the speeds down because in a collision between two motor vehicles, lower speeds work towards keeping the outcomes in terms of severity of injury to a minimum. It's why opposing flows on high speed roads are separated (often with barriers) and why we might use traffic signals, roundabouts and grade separation.

So, what tools do we have to keep driver speeds down? One of the most useful is width. Manual for Streets considers this with a couple of graphs which were underpinned by research undertaken by the Transport Research Laboratory. The research comes from residential and mixed use sites

This research looked at forward visibility and road width and perhaps unsurprisingly, wide roads with lots of forward visibility leads to higher driver speeds. The graphs are split between average speed (left) and 85th percentile speed (right). The 85th percentile speed is the speed at which 85% of drivers travel at or below which has traditionally been used as a design speed in highway engineering and in fact, is often used in the USA to actually set speed limits.

This is also interesting because it also hints at the futility of trying to reduce speed limits on wide open roads without deploying engineering measures, notwithstanding the point that this research come from residential and mixed-use situations; although those are the streets many of us are interested in because that's where walking and cycling are taking place in urban areas.

Unfortunately, it isn't just about narrowing carriageways and reducing forward visibility in all cases, but it does push us to look at how space is allocated. For example, on a route used by buses and even a small percentage of HGVs, a 5 metre road is going to be an issue in terms of basic movement, but in urban situations, a 6 metre road should be sufficient (slightly wider on curves) and allows us to get a base design speed of maybe 25mph to 30mph, although in reality buses and HGVs become slower rolling road blocks. 

It is often a challenge to convince bus operators that a 6 metre road is appropriate because there are always concerns about poor car parking affecting traffic flow, but that's also a design consideration. For a retrofit situation on main roads, this means that once 6 metres has been set aside, there is often lots of space to play with. 

On wide streets, it also means that motor traffic can be physically separated by direction to give a median which can be used for general crossing of the street and planting (to reduce forward visibility). Keeping bus stops on-carriageway can also help slow traffic as well as making it more efficient for bus loading and yes, if it's separated lanes, then drivers are going to have to wait.

As an aside, you might be interested to learn that Manual for Streets and the companion Manual for Streets 2 are being reviewed for a single document planned for next year.

Away from the busier roads, going for narrower roads with limited forward visibility gives licence to the designer to provide layouts that can have a 20mph design speed which most drivers will have to physically stick to.

Car parking used to break up forward visibility. Utrecht.

The use of "horizontal deflection" both physically requires drivers to slow down and use their steering wheels as well as reducing the forward visibility and so it's no coincidence that the Dutch woonerf (living street) approach to residential street design uses horizontal deflection.

The image above is from Recommendations for traffic provisions in built up areas. This is less well known in the UK, but it's another manual from CROW. The English version (linked above) is from 1998 and so not only is it not UK guidance, it's a bit long in the tooth, but there are interesting principles to discover nonetheless and space is space.

In this example, the layout is for streets with fewer than 100 motor vehicles an hour (and not a bus or commercial traffic area) and in streets which have no through traffic function. The dimensions are interesting with 'b' being the traffic lane at less than 3 metres, 'a' being a buffer strip to frontages at 0.6 metre and 'd' being a walking of at least 1 metre over very short distances or 1.5 metres more generally. The "road" width is so narrow, it doesn't feature in the UK manual for streets, but we are talking very low speeds, maybe as slow as walking pace.

Then 'c' is variable width "obstacle" zone and 'e' being a parking bay width of 1.8 metres to 2 metres. Dimension 'L' is around 200 metres to 400 metres where there is one sided access and 400 metres to 600 metres with two sided access (in other words, if you have access junctions both sides of the street, then you can have a longer run of the feature).

Woonerven can also be used in shopping streets and despite their very narrow nature, they can be two-way for general traffic (above) with passing places of 4.5 metres to 5.5 metres in width, although 4.5 metres is enough for a fire engine. The dimensions of 'c' and 'd' are greater or equal to 20 metres and 40 metres respectively.

There's also the wonderful "shifting centre line" layout (above) which further helps with driver speed reduction. As a principle, woonerven have design speeds less that 20mph and with low traffic flows, they are places where people walk and cycle (2-way of course) with the occasional motor vehicle passing for access which is why there are no kerbed footways (although there is protected walking space for those needing it).

This street in Deventer uses some of the principles.

The approach can be costly to retrofit and so has fallen out of favour a little in the Netherlands and if used over too long a distance, some drivers may get frustrated which isn't a state of mind we want people in. This is why in larger low traffic neighbourhoods thought is needed about how the traffic cells are designed. In a UK context this also means that having traffic cells which are too large means some people might start driving within the cell areas.

It is possible to use road humps on main roads and residential streets, but for my mind they should be an integral part of a design and not simply "bolt-on". In other words, in a redesigned high street, they can coincide with the more formal crossing points that some people require to give a level crossing and in residential streets, they can be part of the layout. For example with the passing place in the woonerf example above, making the area a raised table will help discourage speeding through that section.

A more modern street on the ourskirts of Amsterdam. It's a low traffic neighbourhood, but the layout doesn't promote low driver speed. The yellow "Victor Veilig" figures (Safe Victor) show the residents unhappiness with the the speed of drivers in their street.

So, starting with a pretty simple relationship between road width, forward visbility and driver speed, we can play with space and layout to get our desired design speed, but we must always remember that network level considerations should be on our minds too.

Thanks to David Hembrow's blog for the explanation of Victor Veilig.