Room Based BIM I

Previously, Furniture Clearance Configurator gave every object three envelopes: the footprint it covers, the walking room around it, the space it sweeps when you open it. All three are about people. A bed stops there; it asks nothing of the building. A sink asks for water and a drain. A range asks for power, gas and air. In nearly every case, when a user is drawing a wall they will eventually end up with a room. Why not start with the room as the first order of thinking, and let the objects inside it say where the services have to arrive? The drawing below puts up to eight objects in one room, and each one brings its service points with it.

Why start with the room

Because the tools start with walls. Type “how to draw a wall in” into a search bar and the internet finishes the sentence with every drawing program ever sold.

Here is the problem with that. In elemental systems like CAD and BIM there is no concept of embedded intelligence. These systems are records of design having been done hundreds of thousands of times, but only with the final result of what the designer intended. The wall is recorded. Why the wall is there is not.

I posit that as an architect, designing the same bedroom the same way 2,000 times might get a bit repetitive. What about the same vanity – toilet – tub arrangement that I’d guess shows up in 90% of bathrooms? Surely a bathroom can be fully intelligent, since building codes regulate nearly all of a bathroom’s makeup. Why did I, for example, need to design the same kitchen 25 times, when the only difference was slight dimensional adjustments to the kitchen’s bounding box? I wish I had recorded every possible combination of casework-sink-dishwasher-fridge there is, so future me could call that logic, like Thor, the next time I need to design a kitchen.

If you start with rooms, then I think the problem of “Automating Architecture” becomes an order of magnitude easier. In fact one could build a room-based configurator, per room.

Clifton’s Top 10 Room Configurators (the top two are the most valuable):

1. Bathroom – Powder rooms with vanities on top of the toilet tank -> 100 persons capacity at a stadium
2. Kitchen – Ranging from kitchenette to fully fledged commercial kitchens
3. Bedroom – Crib and standard all the way to King sized beds
4. Living Rooms – Solve for couch yield – how much couch can I get, given I need a table and a TV?
5. Dining Room – Like every dynamic block ever has chairs arraying into place as the user makes the table bigger.
6. Office – WFH study all the way to WeWork levels of space
7. Garage – Cars, gotta actually place the cars
8. Closet – Storage can go from shallow pantry to luxury walk in closets
9. Utility Room – In your home it might have a laundry machine. In a skyscraper it will have a massive generator so the elevators won’t get stranded
10. Entry – The portal into a hobbit hole all the way to the Iris device that covers the stargate. Sequence and scale held high.

The system would run on requirements, and would need to solve all kinds of unique conditions within all of the room-based bounding boxes.

Objects carry their service points

Take the second most valuable room on that list. A kitchen’s requirements are a fridge, a range, a sink and 48 inches of casework, all inside four bounding points.

Now ask each object what it needs. The fridge gets cold water access. The range gets a natural gas line. The sink needs hot water, cold water and waste, and so the wall behind it is no longer just a wall: it has to be a plumbing wall. The casework needs nothing. It is still lorem ipsum.

That is the whole idea. If this is the way we build our database, then I could, say, with software make sure that walls attached to fire-rated spaces like a garage would always be fire-rated. Not because a user manually set a wall type to be fire-rated, but because the user set the garages to be fire-rated. Or in a kitchen, the wall type behind a sink automatically grows into a plumbing wall, because the user put a sink there. The room requirements are linked to services, and the services are linked to specific objects at specific places.

Looks like we are forgetting electrical services too, which is unfortunately (or fortunately) governed by code, with ways of mitigating plugmageddon. In general, this is how bad it could get:

Six plugs on one wall, each one there because of the object in front of it. A powered wall, like a plumbing wall, is a consequence of what is in the room.

What the drawing does

The study above turns those diagrams into a rule you can push on. It has eight objects, each with a fixed footprint and a fixed list of needs, coded W water, P power, D waste, G gas and H air:

• Sink (2.5 × 1.5 ft): W, D
• Range (2.5 × 1.5 ft): P, G, H
• Refrigerator (2.2 × 2 ft): P, H
• Washer (2 × 2 ft): W, P, D
• Desk (2.5 × 1.5 ft): P, H
• Bath (2.5 × 2 ft): W, D, H
• HVAC (1.5 × 1.5 ft): P, H
• Lighting (2 × 1 ft): P

Those lists never change. What you change is which objects are in the room (tap one to add or remove it) and the room itself: drag the cyan corner handle for width, from 8 to 24 feet, and the orange one for depth, from 8 to 18. Each need gets a candidate point on its object, on the side nearest a room edge, and a straight run to that edge. If the run crosses another placed object it turns orange and dashed, and counts as blocked. The four cards count objects placed, objects that didn’t fit, service connections and blocked routes. These sizes and need lists are chosen study values. My original drawing gave the fridge cold water; the study’s refrigerator carries only power and air, which tells you the need list is a design decision, not a fact about refrigerators.

What to watch

The first example is a sink alone in a 14 × 10 ft room: one object, two connections, water and waste, both from the same point six inches off the left wall. That point is a plumbing wall in embryo.

Add the washer and the bath and you have a wet room: three objects, eight connections, three of them water and three waste. Every wet object needs both, and every one of them drags its pair to a wall.

The kitchen preset (sink, range, refrigerator) makes seven connections: one water, one waste, two power, two air and one gas. The range is the only object of the eight that needs gas, so it carries the room’s only G. Move the range and the gas moves with it. The desk-and-lighting room makes three connections and no water at all; it asks for power and air and nothing else.

The last example is the interesting one. All eight objects in an 18 × 12 ft room make 18 connections (3 water, 6 power, 3 waste, 1 gas, 5 air), and 6 of the 18 routes are blocked. The bath’s three run 3.75 feet left, straight through the desk. The HVAC’s two run 5 feet up through the refrigerator. The lighting’s one runs 5 feet up through the washer.

Now drag the room down to 14 × 10. Same eight objects, same 18 connections, and zero blocked. The bottom wall is now 3 feet from the bath and 3.5 from the HVAC, so their routes turn down instead of across the room, and the lighting turns right, 2 feet to the side wall. Add up the runs the model draws and the total falls from 34.25 feet to 26. The smaller room is the easier one to service.

So notice what moved and what didn’t. The service points belong to the objects: the sink’s W and D are the same two needs at every room size. The routes belong to the room, and they change with every drag. The rule that picks them is greedy. It takes the nearest edge and never looks at what is in the way, which is how 6 of 18 routes in a slightly bigger room go straight through the furniture.

What this is not

These sketches place candidate service connections. They are not installation design. The model draws a straight line to the nearest room edge; it knows nothing about pipe or wire sizes, circuits, venting, fire ratings, sound or thermal performance, and it makes no plumbing or electrical specification. A blocked route means the shortest guess failed, not that the room can’t be serviced. What the drawing does establish is the bookkeeping: which objects ask for which services, and roughly which edges of the room those requests land on.

The judgment

Store the services on the objects, not on the walls. A wall drawn first records the result and loses the reason; a sink placed in a room carries the reason with it, and the plumbing wall, the powered wall and the gas line become things the software can derive. The same move works at building scale. If there is a combustible element, give me a fire rating. If there is a vertical exit component to this room, give me a fire rating. If I label the room “sound studio,” automatically juice my walls for larger STC ratings. If my building is geolocated in the southwest, give me a higher R rating. If my windows are placed to the west, push them as low on the wall as possible. The ideas are endless. Each is the room telling the wall what it has to be. Making sure the wall around the garage actually is a fire-rated assembly waits for a later post. The route is a separate and harder problem, and the shortest line is not an answer to it.