The Case for Less Communication in Building Design

The goal of any new business should be to provide a good or service that is 10 times better than existing alternatives.? That has always been the north star at Ripple: we want engineers who use Ripple's software to be 10x more productive than those who don't.

After two years of building software for HVAC engineers, I now realize that this goal is impossible on our current path. Even when the software is at a point where a complete HVAC design can be produced at the click of a button (we’re getting closer every day), that still does not improve productivity by 10x.?

Engineers spend too much time on tasks not related to design work for an automated design tool to increase productivity by 10x.? First and foremost is communication and coordination.? If an engineer spends 25% of their time communicating and coordinating, the best we could hope to accomplish with an automated design tool is a 4x productivity multiplier.? That is simply not good enough.

Coordination and communication between an architect and the half dozen or so engineers required to design a building is a vicious and painful cycle.?

Building engineering systems are only a function of the architecture, which is to say that the architecture totally drives the design of the engineering systems (MEP, structural, civil, etc.).?? The architects, whether they know it or not, are deciding what the optimal engineering system is when they are designing the architecture of a building.? 100% window? Big HVAC. 20% windows? Smaller HVAC.? Cantilevered pool? Big structure. And so on and so forth.

The engineer's job, then, is to simply understand the architect's building and come up with an optimal engineering system that supports that building. This seems simple enough, but there is a glaring problem: the design of each engineering system affects the design of the architecture and the other engineering systems.

For example, the architect must know how big the mechanical rooms are before they can finalize a floorplan, the structural engineer must have a final floorplan before they can finalize the structural design, and the HVAC engineer must know the depth of the structural beams before they can design their ductwork.?

Therefore, building architectural design and engineering must occur in parallel, which sets up a complex web of dependencies. A dependency is something that one discipline needs, but can't supply itself. Every dependency requires communication and coordination, which drags the design speed down and forces compromises.? This shouldn't be catastrophic for the industry, good design takes time, we all know that.?? What makes it catastrophic for the industry is the perverse economic incentives that the dependencies introduce.

To an engineer, time is money, so designing the HVAC system, only to have to redesign it when the architecture changes, and then again when the structural engineer gets done, costs the engineer a lot of money. This causes engineers to wait as long as possible to do work on a project.? The longer the engineer can wait, the fewer times they'll have to redesign the system. But by waiting, the engineer forces compromise on the other disciplines.? For example, the HVAC engineer waits until the structural engineer is done with beams before laying out ductwork.? The ductwork is larger than expected due to higher-than-anticipated window-to-wall ratios, this forces the ceiling to drop from 9’ to 8’.? The architect then must decide to compromise on the ceiling height or rework the height of the building, which will then require new HVAC load calculations to be run and all HVAC components to be resized.? Because the HVAC engineer waited so long, the architect chooses to compromise and live with 8’ ceilings, the HVAC engineer then doesn’t have to do rework and is therefore perversely rewarded by waiting so long to do any work.

I call this the A/E game of ‘Chicken!’ and up to this point, AEC software has made this problem worse. The technology used to force us to take our foot on the gas as we headed toward mutual disaster.? It used to take architects updating mylar prints and couriering them over to the engineer to make design changes, then it was just an emailed CAD file, and now changes made to the architectural model update automatically on the engineer's background whether they've reviewed the changes or not.

The standard way to resolve dependencies is through communication and coordination, but we've enjoyed a communication renaissance over the past 30 years: email, cell phones, electronic document transfer, cloud documents, we’ve never communicated more, and yet construction productivity hasn't increased at all, so I find it hard to believe that more or better communication and coordination is going to increase engineering productivity.?

We are not the only industry that has had to solve a complex web of dependencies.? Amazon's CEO Jeff Bezos said, "If we want Amazon to be a place where builders can build we need to eliminate communication, not encourage it", and Amazon went off to build AWS, one of the most complex businesses to succeed at scale.? How did they do it? Jeff Bezos simply mandated less communication, and that all communication would flow through specific interfaces, and those who didn't follow these communication chains would be fired. For example, teams should not email or set up a meeting to ask how to get another team's data, they should go to the interface, read the documentation, and consume the data as necessary.? If that couldn't happen, something was broken.?

We need a corollary in the AEC space, and I propose: "Communicate through the model."? The architect shouldn’t make the HVAC engineer search through the specs for window U-values, they should put them in the model. The HVAC engineer shouldn’t email the architect that a 20’ x 15’ mechanical room is required, they should put a mechanical room placement block in the model and let the architect adjust it to fit their building.??

Architects have the most to gain from communicating through the model.? As mentioned earlier, all building engineering is just a function of the architecture.? If the architect can communicate perfectly through the model, they can get instantaneous engineering.? This would allow them to fulfill their vision of the building without compromise.? They can finally get the 9' ceilings they want without last-minute contortions.? But there is no Jeff Bezos in the construction industry to tell everyone "Communicate through the model or you're fired".? The architects and engineers probably don't even work for the same company, and even if they do, they report to different managers who have different budgets to protect. So, we have a cold-start problem.? Why would architects spend all the time and effort to communicate through the model if they're not even sure if the engineers will consume the data? Even if the engineers consume the data, what benefit does that offer the architect?

Ripple is in a unique position to solve this cold-start problem.? Ripple can use the software it has already developed over the past two years to provide architects with instantaneous feedback on the HVAC space required (shafts, mechanical rooms, plenums, etc.).? Architects will then be more inclined to communicate through the model because they can get more accurate engineering space requirements.? Through this process, we can reduce the whole industry's communication and coordination overhead.? This will lead to more complete and coordinated designs which will lead to more productive construction.

We’re going to call this product for architects “Engineering On Demand”.? Look for a demo in the following days.