Overview: Node Types
Adjacencies in our urban-agricultural landscapes are only as performative as our food distribution system. Network performance will be evaluated on the energy expended linking consumers to their food sources.
To begin structuring the movement of food from source to consumer we identified three distribution types:
- Wholesale Nodes: Farmers will take advantage of their urban locations by setting up onsite Farm Shops which sell to retailers as well as direct to consumers via subscription services such as a CSA(Community Supported Agriculture) programmes and ‘veg-boxes’. (open daily - visited weekly).
- Retail Nodes: Small grocers and convenience stores will be distributed throughout the neighbourhood with a high level of redundancy. These retailers purchase direct from farmers and offer consumers ease of access and longer hours. (open daily - visited daily)
- Wholesale Cooperative Markets: Farmer’s Markets where several farmers come together to sell in a market setting where consumers are offered ‘one stop shopping.‘ These should be located in the denses dwelling clusters in order to serve the largest audience and reduce travel barriers to those inhabitants. Farmers markets also serve to activate other commercial activity therefore they should be located on a high-use route within the network structure. (weekly event - visted weekly)
Computational Rules
Therefore, wholesale farm-stands will be located on large areas of productive surface (minimum of 1-2 hectares).
Retail nodes will be distributed throughout dwelling clusters such that any dwelling should be within a one minute radius of a retail node which at 5km/hr is equivalent to 83m.
Retail nodes will be connected back to farm-stands using a branching structure.
Farmers Markets (wholesale cooperatives) will be located once network topology is established and highly trafficked branches of the network are identified as ‘market streets’.
In order to begin locating these nodes, we must first identify the clusters of dwellings and commercial sized aggregations of productive cells.
Cluster Detection
After the algorithm distributed the consumer and producer cells, clusters need to be defined in order to place the Wholesale and Retail Nodes . Different thresholds define how many of each will be present in the tissue. These nodes will be placed when the different clusters are defined.
Computational Rules
Recognition of clusters for urban or productive cells slightly differ from each other. The rules are as follows:
1. To select the productive clusters:
- - From all the productive cells, the algorithm picks a central cell in a cluster, meaning the cell with the lowest productive intensity.
- - From there it crawls outwards to find connected cells. The criterium for the productive cells is that they need to share a common edge to belong to the same cluster. Therefore only four neighbours are taken into consideration.
- - From these neighbours, the system looks for the next ring of neighbours to join to the cluster.
- - If there are no connected neighbours present anymore, the algorithm switches to the next cluster, by looking for the next cell to start the search.
2. To select the dwelling clusters:
- - From all the dwelling cells, the algorithm picks the tallest cell available to start crawling outwards to find connected cells.
- - If there are dwellings available within the eight neighbours, these cells are added to the current cluster.
- - From these neighbours, it will look for the next ring of neighbours to join to the cluster.
- - If there are no connected neighbours present anymore, the algorithm switches to the next cluster, by looking to the next tallest cell available.
Parameters
(A1) Minimum Area of productive cluster: 1 hA.
Threshold for placing a Wholesale Node when a farm cluster is ought to be commercially viable.
(A2) Area of a large-size commercial farm: 7 hA.
(W) Maximum walking distance: 1 min. = 83 m.
Used as a threshold to place the Retail Nodes.
When the algorithm finished the search for productive clusters, it will add a Wholesale node when the cluster is larger then A1. When the farm cluster is larger then A2, it will be subdivided in sectors of A2 and a nodes will be added for each sector.
Place Retail Nodes
When the dwelling clusters are defined and Wholesale Nodes are present, Retail Nodes are distributed throughout the cluster. If there are no Wholesale Nodes present, the urban tissue generated is of such fine grain, that there is no reason to exist for a food distribution network and therefore no Retail Node will be placed.
When the cluster is large enough, which means that there are dwellings further away from the central node then W, Retail Nodes are placed. The algorithm looks for the central cell to add the first node. From this node it looks for a hexagonal pattern to place nodes in order to ensure that all the cells in the cluster are within the set range W. The steps are more precisely described in the diagrams on the left page.
Qualities Measured
(N) Number of Nodes:
Shows how many nodes are present in the tissue.
(Ar) Area distribution of productive plots.
The distribution shows how large the farms are.
Method
For the nine tissues generated in the previous PI evaluation, firstly Productive and Dwelling Clusters are defined and then Wholesale and Retail Nodes are placed according to the given parameters. Area Distribution of Productive Plots is measured to define how each category of farm is represented in the tissue. For the dwelling clusters trip time to the nearest Retail Node is calculated to measure how successful the placement of the Retail Nodes was.
Observations: Wholesale Farms
The most extreme tissue is generated by F=0.95 and D=140, where there is primarily one large farm. Here the productive area was divided in 9 7ha sectors each with a Wholesale Node.
The other end shows tissues with only plots smaller than 1 hA and no Wholesale Nodes.
This means that by current economics, the productive plots are likely not large enough to be viable, requiring a higher level of participation in individual production on the part of the population.
When we look at the diagonal from top left to bottom right, we see the threshold where the tissues generate farms larger than 1 hA allowing for Wholesale Nodes. Around 50% of the plots here are small urban gardens, between 300 and 3000 m2. The distribution of categories show a mix of categories which could be compared with the situation we have seen before in Havana.
Observations: Retail Shops
Compared to the farm clusters, the dwelling clusters are working inverse. The tissue for F=0.95 and D=140 shows only a few clusters large enough to contain Retail Nodes, with more than 50% of the dwellings further then 1 minute away from the closest Retail Node.
The tissues on the other end of the matrix, like F=0.45 | D=280, show one continuous built cluster. Because no Wholesale Nodes are present in these tissues, the algorithm in the next step will not produce a network to distribute food from producer to consumer. Therefore no Retail Nodes are placed. Again we see a situation in which the inhabitants would be producing the majority of their own fruits and vegetables.
Conclusion & Discussion
What we see in the previous two sets of experiments is that:
varying combinations of the parameters Density (D) and Friendliness (F) result in radically different distribution structures with social implications.
When the Density is high and Friendliness is low, we lack the continuous productive surface required for commercial-scale growers. Instead productive area is fragmented into small urban gardens.
This type of pattern may not be politically feasible in a contemporary western city, where urban dwellers have come to rely on retailers for their food supply.
However, One could imagine a social situation in which such an arrangement could be implemented; for example a new city for climate refugees, in which the displaced population arrives with agrarian skills and the desire to maintain this cultural knowledge in an urban environment.
For our next step of development, we will take forward two of our tissues (F=0.85 | D=220 and F=0.95 | D=220) shown above, which are similar in input values but demonstrating distinctly different distributions of large continuous productive areas.
Distribution paths will be generated for each, connecting Wholesale Nodes and Retail Nodes while evaluating relative energy expenditures.
