# Makable recipes AT6012 Design Research: Technology Transformations Cork Centre for Architectural Education, University College Cork 30 August 2026 A recipe is the last stratum of the pipeline. Above it sit the ontology, the knowledge graph, the generalised principle and the case study. A recipe is what somebody actually does on a Tuesday, with named inputs, a stated output, and a statement of what the result can and cannot support. Each one is runnable today with the files in this folder. Each names its own limit, because a recipe that hides its limit produces a number somebody else will quote without it. --- ## Recipe 1. Read a specification as a demand on a forest **Takes** a timber bill of quantities: a member, a species, a volume. **Gives** the demand in cubic metres against seven building groups, and the ecological groups that would have to supply them. **About** twenty minutes the first time. ``` cd scripts python boq_to_forest.py ../data/example_boq.csv # to see the shape python boq_to_forest.py ../data/my_project.csv # then your own ``` Write your own CSV with the columns `member`, `species`, `volume_m3`. The species has to be one of the 35 in the study landscape, and the script refuses anything else, because guessing a group would invent one. **The reading.** Look at the share in BG1 and BG2 first. If it is above about 80%, your specification is a conifer specification, and it is pulling the forest toward the composition the FDN scenarios are trying to diversify away from. **What it cannot support.** It says nothing about carbon, cost, distance or availability, and it does not know whether the forest can meet the demand. It is a description of what the drawing asks for. --- ## Recipe 2. Find where the supply actually is **Takes** the published landscape data. **Gives** a map, at 100 m, of how many building groups each cell could supply. **About** ten minutes, plus the download. ``` cd scripts python fetch_landis_data.py # from Zenodo, and read the notice it prints python build_coupling.py # derive the seven-by-seven coupling python supply_map.py # build the raster ``` Then open QGIS, Plugins > Python Console > Show Editor, and run `qgis/make_project.py`. **The reading.** The landscape is 1,649,532 cells. 79.6% of them are not forest at all, which is what "fragmented rural landscape" means when you look at it. Of the forested fifth, **337 cells can supply all seven building groups**. That is 337 hectares in a territory of 136 by 121 km. **What it cannot support.** It is the landscape as recorded at the start of the simulation, in 2010. It is not a yield, not a projection, and not a harvest plan. Nothing here runs LANDIS-II. --- ## Recipe 3. Put the two together **Takes** recipes 1 and 2. **Gives** the question the module is actually about. **About** an afternoon, and it is the one worth doing. Run your own bill through recipe 1. Note which building groups it reaches. Then open the map from recipe 2 and use the QGIS raster calculator to isolate the cells that can supply exactly those groups. ``` Raster calculator, on bg_supply_breadth: "bg_supply_breadth@1" >= ``` That gives you the fraction of the territory your drawing can be built from. **The reading.** Almost every real specification comes back with a large supplying area and a narrow group spread, or a small supplying area and a wide spread. Which one you get is a design decision that was made without anybody noticing it was one. **What it cannot support.** A cell shows a supply possibility. Ownership, access, road distance, protection status and whether anybody will sell you the timber are all outside this data. --- ## Recipe 4. Name the gap for a landscape that is not Quebec **Takes** an honest hour. **Gives** a written statement of what applying this framework somewhere else would require. **This is the recipe most likely to end up in a thesis.** The trait data covers 75 species and the study landscape 35, all eastern North American. Check for yourself: ``` cd scripts python -c "import csv,io; print(sorted({r['spp.name'] for r in csv.DictReader(io.open('../qgis/landis/Osborne-et-al_TRAITS DATA.txt',encoding='utf-8'),delimiter='\t')}))" ``` **There are no Mediterranean species in it.** No *Quercus ilex*, no *Pinus pinea*, no *Quercus suber*, no *Castanea sativa*, no *Fagus sylvatica*. So the seven-by-seven coupling does not transfer to Lazio, and neither does it transfer to Ireland: *Picea sitchensis*, which is most of the Irish estate, is not in the table either. The exercise is not to apply the framework anyway. It is to write down what applying it would take: 1. Which species make up the estate you care about, from a national forest inventory, and name the inventory. 2. Which of the 29 trait columns in the Zenodo file exist for those species, and which do not. 3. Which of the seven ecological groups have a plausible counterpart, and which have none. 4. What you would have to measure, and who would have to measure it. **What it cannot support.** Nothing, and that is the point. Step 4 is a research proposal, and saying so is the honest end of the exercise. A gap named precisely is worth more than a framework applied where its data does not reach. --- ## Recipe 5. The climate half, next door **Takes** the Ladybug submodule. **Gives** the climate the building sits in, beside the forest it comes from. ``` cd ../ladybug python compare_locations.py weather/*.epw ``` Cork holds 30 comfortable hours a year, Rome 877. Run your own site's weather file, then ask recipe 1 what the building is made of and recipe 2 where that came from. **The reading.** These are two different coordinate frames on one building. The climate reading is Eulerian: a fixed site, conditions flowing through it. The forest reading is Lagrangian: material followed through space over time. Kiel Moe names those frames in *Convergence* (2013) and is a co-author on the paper this folder is built from, which is why the two halves belong in one module. **What it cannot support.** Neither half predicts. Both describe. --- ## What every recipe shares **State the settings.** The file, the year span, the species list, the assumptions. A figure without them is not checkable by anybody, including you in six weeks. **Every result here is structural and none is predictive.** These recipes show what a change of frame reveals. None of them says a number will move by an amount, because nothing in this data can support that claim. **Where a question cannot be answered from the record, say so and name what answering it would take.** That gap is the teaching, and recipe 4 exists entirely to practise it. ## Sources Osborne, P., Aquilué, N., Mina, M., Moe, K., Jemtrud, M., and Messier, C. (2023). A trait-based approach to both forestry and timber building can synchronize forest harvest and resilience. *PNAS Nexus*, 2(8), pgad254. https://doi.org/10.1093/pnasnexus/pgad254 Supporting data: https://doi.org/10.5281/zenodo.8184010 Moe, K. (2013). *Convergence: An Architectural Agenda for Energy*. Routledge. https://doi.org/10.4324/9780203489444