Search this Site
Hidden Stuff
« Join us for #BlackFridayParking | Main | Understanding Fresno's lack of cash »

Optimizing the Street Grid

Does block size and street width really matter? 

The great American street grid; 


Some people love it, some people hate it. But you have to admire its simplicity, and it certainly does have some merit; 

  1. It is incredibly easy to plan out and survey (everything is an aligned rectangle.)
  2. It is highly connected (there are no dead ends.)
  3. It is very hard to get lost in (especially if you number your streets like in Manhattan.)

Of course, there are disadvantages. Particularly, it can become repetitive and uniform (a strict grid usually does not offer much variation in block sizes and street widths), and it does not accomidate diagonal travel very well. But, I have chosen to focus on the street grid, as it is not the worst street layout there is, and it is very easy to mathematically model and study. 

Within the basic street grid, you have some flexibility, namely, in the size of the blocks and the widths of the streets. But, just how much of an effect do these variables have in affecting the walkability and efficient land usage of a city? 

To help me in my analysis of the street grid, I developed a simple Street Area Calculator tool. You can input a street width, block size and it will generate some simple statistics about the land usage and walkability of a particular grid. I have used it to generate the statistics and diagrams in this post. 

Sample Grids

To study the properties of different street grids, I needed some data to work with. I went on a tour around the United States in Google Maps, and used the measuring tool to measure the block and street sizes in various cities; 


The measurements are pretty crude but they are accurate enough to get a general idea of each city's grid. I took measurements for the following cities; 

CityStreet WidthBlock Size0.5 Mile Section
Austin, TX 115 ft 345 ft
Little Rock, AR 95 ft 300 ft
Manhattan, NY 58 ft 600 x 250 ft
Portland, OR 60 ft 200 ft
Salt Lake City, UT 130 ft 660 ft
San Francisco, CA 65 ft 420 x 275 ft

Already, we can see some massive differences between the grids. We can tell just by looking at it, that the streets and blocks of Portland (left) and Salt Lake City (right) are vastly different; 


Land Usage

One metric we can pull from the grid is how much land is dedicated to the streets versus the actual block. This is an important metric, because essentially, the area used by the block is productive (it can be privately owned, and generates revenue for the city), while the area used by the street places a burden on the city (both due to physically maintaining the surface of the street, and due the lack of land that could potentially be privately owned and generating tax revenue.) 

Here are our above cities, sorted by their land usage. From a tax perspective, the less area dedicated to streets, the better; 

CityStreet WidthBlock SizeStreet AreaBlock Area
Manhattan, NY 58 ft 600 x 250 ft 25.99% 74.01%
San Francisco, CA 65 ft 420 x 275 ft 29.96% 70.04%
Salt Lake City, UT 130 ft 660 ft 30.2% 69.8%
Portland, OR 60 ft 200 ft 40.83% 59.17%
Little Rock, AR 95 ft 300 ft 42.32% 57.68%
Austin, TX 115 ft 345 ft 43.75% 56.25%


This is a great metric for measuring how much land has the potential to be put to productive use - for example, I would consider the grid wasteful in Austin, TX wherenearly half (43.75%) of land is dedicated to streets, removing the potential for it to be put to productive use. 

However, it is not such a great metric for measuring walkability. Walkability is correlated with scale - the closer destinations are, the more destinations that are within walking distance, and the more likely you are to walk. A simple Street:Block ratio does not account for scale differences, and hence, are not good indicators of walkability. This can be observed when comparing Portland's grid to that of Salt Lake City, despite Portland being 'less efficient' by allocating more land to streets (40.83%) than Salt Lake City (30.2%); 


Street Frontage

Another metric we can gather from the street grid is how much of the street frontage is within walking distance (street frontage being defined as the perimeter of a block that faces the street.) For example purposes, I will define "walking distance" as refering to half a mile - which may take an average person 10 minutes or so to walk. 

An interesting property I have noticed about the street grid is that in most circumstances, 45% to 50% of street frontage of a surrounding square mile is within walking distance, regardless of the scale of the grid. Both of the grids below represent a square mile. Let's assume that we have been dropped into the middle of this grid, with the red areas representing the street frontage that is within a half a mile walk; 


You will notice how our walking distance takes the shape of a diamond pattern. This is the result of the grid layout forcing us to use Manhattan distances (we can only walk at 90 degree angles, not diagonally) compared to Euclidean distances (in an organic street layout with diagonal roads, the red area may look more like a circle rather than a diamond.) 

Measuring the amount of street frontage within walking distance is an interesting metric, because you could easily correlate street frontage with 'destinations.' A destination is along a street, and takes up street frontage; 


So we can reasonably assume that the more street frontage that is within walking distance, the more destinations that are within walking distance, and the more walkable the environment is. (This is a flawed assumption, as the street front may not actually be a destination, and instead a parking lot, or a blank side wall, but it will work as a generalization.) With that in mind, here are our cities sorted by the street frontage that is within a half a mile walk - in this case, larger is better; 

CityStreet WidthBlock SizeStreet Frontage Within Half A Mile
Portland, OR 60 ft 200 ft 158400 ft
San Francisco, CA 65 ft 420 x 275 ft 110920 ft
Manhattan, NY 58 ft 600 x 250 ft 106696 ft
Little Rock, AR 95 ft 300 ft 98496 ft
Austin, TX 115 ft 345 ft 83280 ft
Salt Lake City, UT 130 ft 660 ft 56960 ft


This appears to be a much better metric for measuring walkability. Despite Salt Lake City's grid having less land allocated to streets than Portland, Portland's grid has 2.78x the street frontage within walking distance. 


The smaller the blocks are, the more there is within walking distance. Let's assume that we have 60 ft wide streets, and we will play with the grid sizes; 


The smaller our blocks are, the more they are broken up, and the more street frontage we have. The trade-off is that this greatly increases the amount of land we then dedicate to streets; 


The reason behind this is simple - when we break up a block, we need to run a new street through it. Eventually, we reach a point where we are simply wasting space, by allocating more than 50% of our land to streets; 

Grid SizeStreet Frontage Within Half A MileStreet Area
100 ft 207,360 ft 60.94%
200 ft 158,400 ft 40.83%
300 ft 120,960 ft 30.56%
400 ft 101,760 ft 24.39%
500 ft 86,080 ft 20.28%

If we are to downscale our blocks to make our grid more walkable, we also need to downscale our streets, in order to keep the ratio of Street Area:Block Area down. We can have 150 ft blocks and keep our street area down to 22.15%, if we also build 20 ft streets (which would result in 284,800 ft of street frontage being within a half a mile walk - far greater than that of Portland!) 

However, when we start talking about 150 ft blocks and 20 ft streets, we begin to get into the realm of traditional cities; 


Traditional cities are naturally highly walkable, human-scale environments. 


The grid layout is great to study because it is very simple to model mathematically, which allows us to see the effect of different street widths and block sizes on land usage and walkability. The trade-off with the grid layout is choosing between walkability (small, finely grained blocks) or efficiency (large blocks, with very few streets). 

In this blog post, I presented a purely mathematical model where all blocks and streets were equal in size. In real life, this is not always the case, and there are plenty of cities that have added their own variation to the grid layout. The best implementation of a street grid that I have seen is Melbourne's. Melbourne has a great mix of wide streets; 


And narrow laneways; 


In the future, I will discuss other street layouts - such as radial grids, organic streets, cul-de-sacs, and hierarchical systems. 

PrintView Printer Friendly Version

EmailEmail Article to Friend

Reader Comments (14)

Great analysis! Have you taken into the account the effect of street width on street frontage? ie 100 ft blocks with 90 ft streets would only have 40ft of frontage per block compared to 360 ft of frontage for 10 ft streets.

November 27, 2013 | Unregistered CommenterNicolas Derome

Cool idea, and it does start asking some interesting questions. For example, South Philly as a rule has 50 ft. streets and ~400 ft. block lengths*. Input these into the calculator, and you get 20.99% street area over a square mile, and 48.81% accessible in a half-mile walk. But South Philly also has a high count of interstitial streets--two, on average, per square block, and these streets are usually 30 ft. wide, which yields

-Average street width of 40 ft. ((50+30)/2)
-Average block width remains 400 ft.
-Average block height is (400-(30*2))/3=340/3=110 ft,

yielding statistics of 33.33% street area and 45.79% accessibility.

The issue with this, however, is that I've had to conflate the clear primary-secondary hierarchy in Philadelphia's street system to obtain this result. Is there a way to, say, construct a model which allows you to overlay a secondary grid on a primary grid--the secondary grid being that of the city's interstitial streets--to obtain a "better" combined result?

*Average of ten random measurements on Google Earth.

November 27, 2013 | Unregistered CommenterSteve S.

This is really neat. With enough test application to several cities it should be possible to establish Price's Proportion, and discover the ideal for a Strong Town, yes? Charlie at the Old Urbanist did an exercise something like this a couple of years ago:

November 27, 2013 | Unregistered CommenterSophia Katt

Thanks, people!

Nicolas - yes it does take street width into account for all calculations.

November 27, 2013 | Unregistered CommenterAndrew Price

Great post Andrew. I wrote A Strong Towns inspired post about Melbourne block sizes and laneways here:

November 28, 2013 | Unregistered CommenterKent Lundberg

Street grids are great for flat areas, but where there are hills, they place an undue burden on bicyclists. See The San Francisco Wiggle as an example of a workaround that bicyclists have devised. It would have been more efficient for them if the roads had been laid out in a way that minimizes slopes.

November 28, 2013 | Unregistered CommenterDerek


You got the blocks for Austin completely wrong. In downtown, each block is 276' on either side and the streets are 80' wide. That would mean that 28.99% of land is devoted to streets and 71.01% to land, placing it between Manhattan and San Francisco. I do get that the measure tool in Google Maps is imprecise, but simple fact-checking would help you tremendously, because factual errors like this one undermine the whole point of the post.

November 29, 2013 | Unregistered CommenterJeremy Atkinson

Great post Andrew. Have you looked at overlaying large and small grids? One concern I have with most grids is that they tend to encourage a fair amount of fairly fast traffic and slowing that traffic down to a speed, like 15-20 mph, where cars/bikes can safely share street space and where the space overall is more human friendly is difficult and becomes more difficult the greater the grid area.

A larger grid of wider & higher speed streets with each of it's grids comprised of numerous slower and narrower streets (10 blocks x 15 blocks for instance) would reduce traffic volume and the need for higher speeds in the smaller grid area which would make it more human friendly (distance is a minor element of something being 'walkable') and eliminate the space needed for segregated bicycle facilities. The smaller grids could be pedestrian/bicycle permeable while blocking all thru motor traffic creating pseudo-woonerfs (necessary if bicycles and motor vehicles are to share the road space).

December 1, 2013 | Unregistered CommenterWalker Angell

Ah, Steve beat me to it, but I was also going to remark that the Philly street grid seems to resemble Melbourne's in its distribution of arterials and laneways (and every kind of street in between) in a comfortably-familiar grid.

However, I'm not sure if new "interstitial" streets can be created anymore; most seem to be holdovers from the prewar era. While there have been a few developments that explicitly emulated old-world public spaces (the Piazza in Northern Liberties, for example), it's not clear whether interstitial streets and alleys can still be freely added to blocks undergoing redevelopment (i.e. spontaneous subdivision), as was once possible.

One criticism of grids (voiced by Jacobs) is that they tend to run off into a drab unsatisfying horizon, with few focal points, enclosing T-intersections, and other interruptions to make the grid feel less monotonous and more personable. This might not be an issue for some, but Jacobs wondered if this was why some European visitors - accustomed to irregular street networks even in 19th century districts - complained about the monotony of American urbanism.

Some cities, like San Francisco, overcome this via sheer topography, but otherwise it seems to be a common complaint in American cities - "the grid is boring." Again, Philly's City Hall is a notable exception since it interrupts the city's two axial spines to offer an orienting focal point.

December 2, 2013 | Unregistered CommenterMarc

Hey! Super interesting piece! I did just want to make a point on your comment about the "flawed" concept of walkability and frontage space, specifically in relation to Portland, OR. In fact, the problematic issues you outlined with measuring available frontage (the fact that some of the available space may be parking lots and blank store space) were issues actually actively countered by Portland planners in the '70s and '80s. Some specific municipal regulations were passed to cap the number of parking lot spaces allotted the to the downtown area, stores were required to include 1st floor display windows, while at the same time the city designed it's now coveted light rail system. James H. Kunstler outlines these measures in his book "The Geography of Nowhere" although he fails to cite most of his material, so please, correct me if I'm wrong here. I've always read Kunster with a grain of salt. In any regard, most urbanaphiles would agree that Portland is doing something right. Cheers!

December 2, 2013 | Unregistered CommenterAJ

Thanks for all your comments!

To answer some of your questions:

- I apologize that I have some of the street dimensions wrong. I did start out by researching online for the grid and street dimensions, but many cities didn't publish them, so I resorted to measuring them via Google Maps.

- Overlaying grids: Something like a large grid of wide roads, containing a smaller grid of narrow roads? That's certainly possible. For this post I decided to stick to a plain grid that is simple to model mathematically. Hybrid approaches would add a lot of complexity to the calculations, but are certainly possible.

- Several people have pointed out to me that my simple metrics such as land usage and street frontage are not good indicators of walkability. The street frontage metric does not actually say what the street frontage is being used for (building fronts? blank walls of a big box store/factory? a parking lot?) I agree with them - it is just one of many other factors, such as what you actually build in the blocks, and the configuration of the street, that also affects walkability.

December 3, 2013 | Unregistered CommenterAndrew Price

Kent: Great post. I love Melbourne - it's an amazing city!

December 3, 2013 | Unregistered CommenterAndrew Price

I would be very curious if you have done a double analysis- not only the street grid, but also the height of the buildings along the street. Does this make a difference? It should, given that the higher the buildings, the more foot traffic can be generated. Also, the number of entrances along a block should be an important factor.

December 4, 2013 | Unregistered CommenterDavid Stein

To build on David's question, I think where the frontages (most entrances) are oriented toward on the blocks is an important factor, and it is not at all incidental to land productivity and access. This analysis may be an adequate simplification for square grids like Portland's but not rectangular blocks as in San Francisco's. The frontage grain in rectangular block cities is organized along the longer block dimension primarily, and the majority of parcels only have frontage on these sides of the block. Maybe, in these cases, what you want to increase, in trade-off to access, is the length of frontage streets.

I also know that Savannah happens to be in the 50% (and more!) street area range, but the way that land is utilized in Savannah for buildings and open space is very different, we all know, from the way productive land is utlized in Austin. Urbanistically, that Savannah fares so badly in this analysis is interesting to me. In terms of a form analysis comparison, it would be useful (at least for urbanists) to compare fabrics in terms of how much land *tends* to be conceded to automobiles versus other productive uses such as open space (space for people) and buildings.

December 8, 2013 | Unregistered CommenterEric O
Comments for this entry have been disabled. Additional comments may not be added to this entry at this time.