[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"blog-how-good-is-spacios-wind-tool":3},{"id":4,"title":5,"author":6,"authorAvatar":7,"authorEmail":8,"authorLink":9,"body":10,"category":243,"description":244,"extension":245,"gradient":246,"mainImage":120,"meta":247,"navigation":257,"path":258,"pinned":259,"publishedAt":260,"readTime":261,"seo":262,"stem":263,"__hash__":264},"blog\u002Fblog\u002Fhow-good-is-spacios-wind-tool.md","How good is the wind tool? We ran the AIJ benchmarks to find out","Franz Forsberg","\u002Fimg\u002Favatar_franz.webp","franz@spacio.ai","https:\u002F\u002Fwww.instagram.com\u002Ffranz.forsberg\u002F",{"type":11,"value":12,"toc":232},"minimark",[13,21,28,43,48,64,71,77,81,94,100,104,115,121,125,131,137,143,149,155,158,162,165,172,178,184,188,201,204,208,214,229],[14,15,16,20],"p",{},[17,18,19],"strong",{},"Spacio's wind tool tells you how windy it is for a person standing on your site",", at 1.5 metres, around the massing you are drawing right now. That is a claim about the real world, and claims of that kind should be checked against something that is not a computer.",[14,22,23,24,27],{},"So we ran the shipped tool, unmodified, in a browser tab, against four wind-tunnel experiments published by the Architectural Institute of Japan: ",[17,25,26],{},"an isolated block, a thin slab, a tower standing in an array of low blocks, and an empty domain"," that checks the tunnel reproduces the wind you feed it before anything is put in the way. Public experiments, public measurements. The score is not ours to choose.",[14,29,30,31,34,35,42],{},"The results are mixed. ",[17,32,33],{},"What follows is what worked, what did not, and where the tool cannot answer",". The full technical record follows separately. The companion post, ",[36,37,41],"a",{"href":38,"rel":39},"https:\u002F\u002Fspacio.ai\u002Fblog\u002Fwind-comfort-design-decision",[40],"nofollow","how to read a pedestrian wind map",", covers what the map shows and why Comfort, Risk and Safety are three different questions.",[44,45,47],"h2",{"id":46},"the-tunnel-we-build-around-your-site","The tunnel we build around your site",[14,49,50,51,54,55,60,61,60],{},"Every run builds a ",[17,52,53],{},"virtual wind tunnel"," around the site. Buildings, context, trees and terrain are cut into a grid of cubic cells, 1 metre near your massing, 2 metres further out, 4 metres at the edges. The fine cells sit where people walk. The map is read at 1.5 metres, roughly at a pedestrian's face.\n",[56,57],"img",{"alt":58,"src":59},"The virtual wind tunnel, and where the fine cells go","\u002Fimg\u002Fblog\u002Fhow-good-is-spacios-wind-tool\u002Fvirtual-wind-tunnel-grid.webp"," ",[56,62],{"alt":58,"src":63},"\u002Fimg\u002Fblog\u002Fhow-good-is-spacios-wind-tool\u002F1.webp",[14,65,66,67,70],{},"One constraint arrives immediately: ",[17,68,69],{},"buildings above 100 metres are cut to 100 metres"," in the solve, because the cells are deliberately spent near the ground. A 200 metre tower is solved as though it were half as tall, and the downwash it throws into the street is not fully in the picture.",[14,72,73,60],{},[56,74],{"alt":75,"src":76},"From open-country weather data down to street level, in twelve directions","\u002Fimg\u002Fblog\u002Fhow-good-is-spacios-wind-tool\u002F2.webp",[44,78,80],{"id":79},"where-the-wind-comes-from","Where the wind comes from",[14,82,83,86,87,90,91,60],{},[17,84,85],{},"The wind comes from the weather record for your coordinates",", which describes open, flat country rather than a street. Getting from one to the other is a two-step transfer: up to the height where terrain stops mattering, then back down through whatever your site sits in, ",[17,88,89],{},"open, suburban, urban or dense",". Spacio reads that category off the buildings upwind of you, per direction, and you can override it.\n",[56,92],{"alt":75,"src":93},"\u002Fimg\u002Fblog\u002Fhow-good-is-spacios-wind-tool\u002Fweather-to-street-translation.webp",[14,95,96,99],{},[17,97,98],{},"The year is solved in twelve directions"," of 30 degrees each and recombined hour by hour, so the map shows the wind your site gets rather than the wind a compass rose suggests it might.",[44,101,103],{"id":102},"what-the-benchmarks-showed","What the benchmarks showed",[14,105,106,107,110,111,114],{},"The good part first. ",[17,108,109],{},"Around isolated buildings"," the tool puts the pattern in the right place: ",[17,112,113],{},"the corners that accelerate, accelerate; the pockets that shelter, shelter."," At the fast points, the ones that decide which comfort class a spot lands in, the speeds sit inside the tolerance the CFD literature accepts for this kind of work. And runs are deterministic, so two options can be compared without wondering whether the tool simply had a different day.",[14,116,117,60],{},[56,118],{"alt":119,"src":120},"How a wind-tunnel benchmark is scored against a simulation","\u002Fimg\u002Fblog\u002Fhow-good-is-spacios-wind-tool\u002F3.webp",[44,122,124],{"id":123},"what-it-gets-wrong-and-what-it-has-never-been-asked","What it gets wrong, and what it has never been asked",[14,126,127,130],{},[17,128,129],{},"Sheltered spots read windier on the map than they are in reality."," A courtyard the map calls borderline is probably calmer than that, and the wake behind a building is drawn shorter than it should be.",[14,132,133,136],{},[17,134,135],{},"In dense fabric the streets read too calm."," Ranking still holds in a block array, so the windiest street is still the windiest street, but the speeds in the interior under-read. A marginal street deep in a block is likely worse than shown.",[14,138,139,142],{},[17,140,141],{},"The flow behind and over a building is simplified."," The wake settles back down too soon, and the recirculating pocket on a flat roof is not reproduced at all. Separation off a sharp edge is right; what happens downstream of it is approximate.",[14,144,145,148],{},[17,146,147],{},"Some things have never been tested."," Terrain is in the solve, but no wind-tunnel experiment with relief exists in the suite, so behaviour over a hillside is unvalidated. Nor are the tree canopy model, the facade pressures, the default ground roughness, or oblique winds. Dense high-rise districts are the largest gap, and the 100 metre clip is why.",[14,150,151,60],{},[56,152],{"alt":153,"src":154},"What the tool sees, what it approximates, and what it has never been shown","\u002Fimg\u002Fblog\u002Fhow-good-is-spacios-wind-tool\u002Fsimulation-limitations.webp",[14,156,157],{},"The tunnel rebuilds itself around whatever scene you put in it, which is where that last gap starts: its height comes from your buildings, not from the ground, so a hill eats into the air above it.",[44,159,161],{"id":160},"three-sentences-that-bound-all-of-it","Three sentences that bound all of it",[14,163,164],{},"They are in the product's own on-screen copy, not in fine print.",[14,166,167,168,171],{},"• ",[17,169,170],{},"It is comparative."," It supports ranking: which corner is windier, which massing shelters better, which option puts the entrance in the wrong place. Absolute speeds carry the biases above.",[14,173,167,174,177],{},[17,175,176],{},"Buildings above 100 metres are clipped in the solve",", and dense high-rise districts have not been benchmarked.",[14,179,167,180,183],{},[17,181,182],{},"It is not a certified assessment."," Where a planning authority requires a wind study, this informs the design before that study. It does not replace it.",[44,185,187],{"id":186},"how-to-actually-use-it","How to actually use it",[14,189,190,193,194,197,198],{},[17,191,192],{},"Read differences, not absolutes."," Two massing options on the same site is the comparison the benchmarks support; a single quoted speed in a dense street interior is the weakest thing the tool produces. ",[17,195,196],{},"Treat sheltered readings as pessimistic and fabric interiors as optimistic."," City Lawson is the default classification, and Safety is a separate test on a separate record, so ",[17,199,200],{},"never read a comfort class as a safety result.",[14,202,203],{},"And run it early. A direction settles in minutes, not hours, which is the point: it belongs in the sketch, while the massing can still move.",[44,205,207],{"id":206},"why-publish-the-misses","Why publish the misses",[14,209,210,211],{},"Because a wind number you cannot audit is worth nothing to the person carrying professional liability for the massing. If you are going to move a building five metres on the strength of our map, ",[17,212,213],{},"you are entitled to know where it is right, where it is optimistic, and where it has nothing to say.",[14,215,216,217,222,223,228],{},"See where your own site lands at ",[36,218,221],{"href":219,"rel":220},"https:\u002F\u002Fapp.spacio.ai",[40],"app.spacio.ai",". If you want to argue with the method, we are available on ",[36,224,227],{"href":225,"rel":226},"https:\u002F\u002Fdiscord.gg\u002FjdAGrfUBbY",[40],"discord",".",[14,230,231],{},"We will continuously research and improve the tool, so you can expect improvements in the near future.",{"title":233,"searchDepth":234,"depth":234,"links":235},"",2,[236,237,238,239,240,241,242],{"id":46,"depth":234,"text":47},{"id":79,"depth":234,"text":80},{"id":102,"depth":234,"text":103},{"id":123,"depth":234,"text":124},{"id":160,"depth":234,"text":161},{"id":186,"depth":234,"text":187},{"id":206,"depth":234,"text":207},"Product","We scored Spacio's wind tool against four AIJ wind-tunnel experiments. What it gets right, where it is biased, and what it has never been tested on.","md","linear-gradient(135deg, #1A1F36 0%, #2D3561 100%)",{"slug":248,"pillar":249,"tags":250,"status":255,"sources":256},"how-good-is-spacios-wind-tool","Core Product",[251,252,253,254],"Wind Comfort","Validation","CFD","Feasibility & Massing","READY",null,true,"\u002Fblog\u002Fhow-good-is-spacios-wind-tool",false,"2026-09-09","6 min read",{"title":5,"description":244},"blog\u002Fhow-good-is-spacios-wind-tool","2ioT27uvzwdVodwACB73frBApo6n0h0B7okFXP2dFRM"]