There is one task which seems to defeat urban planners around the world. Having been schooled on schemes of fractions of a Sq KM they have to design a city at 40-50+ sqkm in area and apply the same principles. The result a traffic clogged disaster. The city keeps growing and the same principles are applied but the costs of maintaining the infrastructure of the city grow exponentially. 3rd Ring Road, 4th ring road, all comprised of mutually inaccessible superblocks of hyper-dense development all looking the same and having similar problems. What has gone wrong?
Although in most of the world informal settlement predominates in this ‘Asian’ model of urbanization planned development predominates. So planning by itself is not the answer.
Similar in most Asian megacities high rise buildings predominate – so building at high densities by itself is not the answer.
Finally in China most land ownership and gains from uplift of land value are captured by local government, so this is not by itself a solution to urban dysfunction. Something else is going on.
We have in recent years gained an increased appreciation of the economics of agglomeration from rapid urbanisation. Lets look at the classic Alonso model of ‘ideal’ city size with fresh eyes. Here cities have increasing economics of agglomeration with size, realised by the firm (access to labour markets, innovations etc.), however there are also increasing diseconomies of scale borne by society (pollution, congestion etc.) hence it may individually advantage a land developer to capture the proportion of agglomeration value captured as rent even though the point of net loss (economies-diseconomies) has been exceeded. At this point the optimum city size for a societal perspective has been breached. I would add to that the caveat ‘for any given level of infrastructure’ by investing in infrastructure such as transit the benefits of agglomeration can be achieved for longer as the disbenefits curve is flattened. The key factor then for planning on a city scale is how this infrastructure is designed. At some point however, as we may be seeing with Beijing, the costs of developing the infrastructure to the same pattern may become prohibitively expensive and (setting aside other factors for controlling sprawl) there may be a purely economic case for limiting city size.
It is this issue of city scale form and supporting infrastructure that is the focus of this article. Although many urban planning principles are scale independent, as Andreas Duany has argued, some importantly only come into play when dealing with scales of several SQkm or above. As urban planning students are not given exercises of this scale they will not have been taught them. Hence the propensity to simply scale up design features of s scheme scale.
One. The higher the classification of a road the less rectilinear it should be.
Because the top end of road hierarchy’s, such as expressways and major arterials- have the widest right of ways they disproportionately consume developable (and sellable) land. Also the top end of the hierarchy is designed to get from A to B, through and between cities, rather than collecting traffic between districts. Therefore they should be straight, whilst allowing for gentle curves as per good practice in design of such roads in order to keep drivers alert and awake. The only exception to such a rule should be for bypasses, where the same rule in adapted form applies. The shortest route around the feature to be bypassed.
Two -The level of service of roads should be restricted to that required to maintain efficient circulation for all users over the medium term and no more, no less
The skinnier roads are the better else they eat into the land budget of everything else, dominate urban form and restrict pedestrian and cycle circulation. The spare capacity of roads should be that to meet the planned growth of the quarter not the unspecified growth of the city, or else the high speeds will simply induce traffic onto the system. Slow and steady speeds and a degree of congestion is to be expected in a city as a sign of economic activity and people wanting to travel to and from places that generate wealth. Problematic congestion is not to be measured by length of congested roads or the speed on those roads but by the aggregate time spent commuting – a metric determined by the alternative transit capacity available. The true way to measure cost of congestion is between generalised cost of travel in a car commute against the discounted generalised cost of travel of a transit or cycling investment; otherwise there is no opportunity cost of savings to be made. To illustrate in Central London roads speeds have been falling but this does not indicate a rising cost of congestion, it has been falling because of the decreased traffic volumes induced by switching to cycling and buses as road space has been reallocated. Buses and cycles use up far less land area per person traveled so overall raise the capacity of the city to house people and jobs.
Once the growth of a city exceeds the level of service limits of arterial routes growth of the city should be facilitated by transit networks not new roads.
As a city expands its lengths of arterial roads expands in a linear manner, but the length of its rings roads and its surface area, and hence population, expands in a geometric manner. These two geometrical constraints considered together means that arterial roads will begin to clog up as cities expand. This we can see most dramatically in emerging economies cities like Jakarta and Manilla where public transport investment has not kept up with the growth in cities. We can see it even more dramatically in many African cities where arterial roads have been expanded outwards maintaining the dimensions from colonial times and hence being completely clogged.
Hence once a city has expanded beyond the level of service limits of its arterial roads, and once ring roads become ever more expensive to construct because of their increased length cities should expand as a series of sub-cities in a polycentric way – primarily connected by transit.
Very high traffic generators should be dispersed and not concentrated, high traffic generators should be concentrated and not dispersed
Very high traffic generators such as hospitals, Malls, schools, universities etc. should be dispersed around a city rather than concentrated in one area to spread traffic flow. Conversely high traffic generators such as shopping, high density housing and offices should be concentrated around specific nodes, rising in density towards the centre, to provide the farebox for viable transit systems. The joint effect of both rules will be to spread the load proportionately so that any one ‘ring’around a city centre has roughly equal traffic volumes in each of its sectors.
Commercial uses should be located close to but not on or parallel to major routes
Locating major commercial uses in strips along major highways concentrates major trip attractors onto the busiest parts of the network. It also creates barriers between housing and main routes for transit along those corridors and creates a very dull townscape with space consuming service roads that only service units on one side of teh street. The answer is to create a ‘tartan grid’ where shops and services are offset from the main route grid focused around town centres where walking, cycling and public transport have priority. Such centres should not be at the centre of neighbourhoods – that is the function of a neighbourhood centre, rather District and town centres should be at the boundary of two or more neighbourhoods on the offset tartan grid.
There is broad consensus now, influenced by the work of Shlomo Angel, that flexible grids have many advantages, especially for rapidly growing cities.
They allow for easy pacrelisation and laying out of districts, they allow for distribution of traffic and reduced route distances, especially if introducing diagonal elements, and can easily be extended as the city grows. Grids can also be distorted and allow for a variety of tesselations, not just the rectilinear bloc. Canberra and New Dehli for example have hexagonal grids. However the biggest problem that large cities have is basing the geometry of the grid on major collectors spaced too far apart, such as the 1 mile grid in Detroit or the 1-2km grid in Riyadh. Then major roads isolate neighbourhoods with ROWs too wide to develop nclosure, street character or allow for easy pedestrian crossings. By contrast successful historical gridded cities typically have 150m-600m grids, and avoid to great a land take from roads by having rectilinear rather than square grid blocks. Which leads onto the next point.
Neighbourhood character should be defined by block form and not road grid form
The best gridded plans like Oglethorpes plan for Savannah (below)
and Cerda’s for Barcelona
are based on blocks forming streets and squares. which then form neighborhoods and define major routes, by contrast in most large scale subdivisions today the process is reversed with the definition of the grid starting with major roads. Such an approach produces organised sprawl, vertical or horizontal whatever the density. By starting with appropriate sized blocks you are defining places and building community at human scale.
The Superblock fallacy should be avoided;- major Roads divide local community and connect remote communities; major roads are only needed in the predominant direction of travel and should not be used on all sides of a grid unit
The superblock has been described by Peter Calthorpe as a ‘weapon of mass urban destruction’, it is the almost universal form of urbanism in China and now copied all across asia and even now in Africa where Chinese companies have built. Several hundred metres long a series of near identical high rise block are formed within a grid square surrounded and separated by wide collector roads. Globally it is now the predominant form of urbanisation.
The superblock is a major departure from traditional chinese urbanism of cluster housing forming a small scale grid. originally adopted from the Soviet version of the modernist superblock.
As I have written before the superblock was originally conceived as a grouping of apartments along a narrow strip to maximise natural lighting from double aspect units. originally as conceived by Ernst May in the Zeilenbau (objective) Block tended to recreate traditional urban forms of perimeter blocks, enclosed spaces, srett wallss, lot lines squares and streets with little penetrating traffic.
This was firmly rejected by CIAM and especially La Corbusier and Walter Gropius. It is little realised that the Athens Charter was especially designed to abolish the street. They wanted to see He wanted an evolution away from the perimeter block to the slab blocks, longer (requiring deck access) and taller.
The focus on sunlight meant taller blocks had to be set further apart. The green space was incidental – it was not to be used but to be looked at from flats and provided access to the sky. The corridor street was to be abolished, in part because it discouraged car use.
Whilst these modernist ideas were disfunctional in terms of attractive open spaces and vibrant neighbourhood interaction they had at least some saving grace in terms of spacing, light local sports facilities and protecting residential areas from street level traffic. In the chinese superblock however these advantages are lost we have instead domino cities of ‘tombstone urbanism’ where repetitive tall buildings outlook onto tall buildings and car dominating at street level. Superblocks typically do not offer smaller connecting roads. Without secondary roads, all traffic — local and longer distance — intermingles up on connectors causing congestion .
These often gated and inaccessible to non resident superblocks are effectively high density subdivisions internally well connected but externally poorly connected.
The great advantage of this form of superblock however is the rapidity at which peripheral expansion can be planned and developed by private developers handed large lots with a few simple planning standards. These advantages should not be dismissed or underestimated.
The advantages of the flexible grid can however be gained without its disadvantages. There are two sides to this, planning the city grid and site planning.
The city grid should avoid isolating communities into large superblocks. As well as limiting grid and block size and ROW land take connectivity between grid squares can be enhanced by ensuring that there are not connector grade traffic on all sides of a grid square. A grid square only needs a means of access, capacity in the predominant direction of travel and a choice of routes. That can be done with connector roads on one or two or at most three sides and not 4, the other roads can be distributor or local roads providing pedestrian and cycle circulation between neighbourhoods. The repition of this across districts can achieve large and relatively traffic free areas across large areas and without the need for full scale Radburn like segregation of cars, as traffic would be disbursed and local roads would only each receive limited traffic.
Within the block the inclusion of local roads and alleys, with perimeter blocks and cluster and courtyard housing with a variety of height provides for variation and defined open spaces enclosed by housing rather than a bare expanse of ill defined semi-private space. Sunlight or shade depending on climate can still be achieved by street orientation and variations in aspect and building height.
It is a false dogma that the road grid, the community grid, the pedestrian and cycle grid the Green Grid and the Building footprint grid should all follow the same topology – where these divert and clash you gain a sense of place
The road grid is the road grid, defining the strategic layout of a cities interconnections. This need not defined other networks such as open spaces etc. These need to follow there own logic of interconenctivity and matters such as water flow and storage. Pedestrian and cycle networks flow within and between road networks and ideally should be able to follow more direct routes through links closed to motor traffic
Within a neighbourhood it is the layout of buikdings that should define streets and spaces and not vice versa.
It is where these varying networks come together and then diverge which create interesting places.
Block and Neighborhood size should vary by population density and not remain fixed.
The problem with uniform grid size is that population density will inevitably increase in the more accessible areas, which means that both density and spacing of local facilities and public services need to reduce towards the centre. To avoid roads taking up too great a proportion of land area as the grid size decreases most roads also need to reduce in size, which also means diverting most traffic around and outside the central areas.
Major buildings in a neighborhood and connector roads define the townscape and visual relationships of one neighborhood to another.
Most views in a city are local, built form restricting views to one street or two.
Parks, connector roads and local landmarks define views at a district scale and give a city much of its legibility and character. Rarely open areas like rivers and sea and major open spaces, combined sometimes with major topography shifts and major taller buildings can define a city to the outside world, great examples being Rio and Capetown, though in both cases what is perceived internationally as the City is in reality just a small Central Part.
Defining the same visual relationships at a City Scale is much neglected in modern City Planning, with visual axes and skylines seemingly merging by accident.