With the help of generative, space can be maximised and parts can be optimised with ease. The future of design – the future of making things – stands to look a little bit different with the help of artificially intelligent CAD programs.

Engineers, architects, and designers all work on finding new innovative ways of making things within the realm of their own discipline. Imagine an architect working on designing the layout of different structures, say an office building. In the traditional realm, the architect would undergo a series of iterative design rounds and use his intuition to analyse each concept. 

He or she might draw one that allows more light into the space and he might draw another that maximises the views of a nearby pond. At the end of the day, what the architect is doing here is a matter of optimisation. A determination of which design is the best for his belief of what the end product should be. Enter generative design.

Generative design in architecture

Generative design is the concept of allowing a smart AI that works inside of CAD and CAE software some control over the design process. It takes what computers are good at, making a multitude of calculations and developing many possibilities all at once – maths problems – and mixes that with human choice and influence. 

CAD software that works with generative design tools, such as Autodesk Revit or Autodesk Fusion 360, allows the user to input a number of constraints to a design and the generative algorithms work out a plethora of choices.

One might be the best scenario for a certain load. The other might be the best building for high-occupancy offices. Literally thousands of design options can be developed based on the constraints the designer inputs. 

So, back to the original example. 

The architect designing the office building would input the constraints he wanted to work around into the generative tools of the CAD program: how much light, the pad size, the height, the proximity to other structures, etc. The generative algorithms would then take that and develop fully modelled frameworks of different options for the architect to peruse through.

This complex work takes a significant amount of computing power, which is why most generative CAD tools will leverage the cloud for computing power. This is the model that Autodesk employs, allowing users access to their cloud platform for a variety of applications, generative being one of them. 

The biggest advantage to this workflow is that the architect goes from manually sketching a few different options to now being presented with every possible option for the structure. Not just some; every possible option. That's the power of generative. 

At this point, anybody familiar with the process or the buzzwordiness of generative design is likely scoffing. "Generative is just the pipedream of some bull-nosed computer programmers who don't understand how things actually get made," you might say. While you have every right to think that, and while that may have been true in the past, it simply isn't any more.

An example of the different design options generative algorithms can provide inside of Autodesk Revit, Source: Autodesk

Generative design algorithms can take into account manufacturing processes, building costs, material costs, material weight, structural design, easements for roadways, clearances for parts. This isn't your grandmother's generative any more, we're in the age of the future. 

Closing out the initial example we posed, generative design would allow the architect to look through a plethora of different building options, pick which one they like the best, and then start designing the more refined structures on top of the model. Generative spits out editable CAD geometry, so working alongside the algorithms is easier than ever. 

I'm getting ahead of myself here, though. While it's not hard to see through that example and understand how transformative generative design software can be for engineers and designers, first we need to understand the scope of generative design. 

Generative design as a methodology

When we break down the word 'generative', we can understand a little bit more about the concept behind the buzzy title.

In the field of linguistics, generative is understood as a process of applying rules to inputs in order to produce well-formed items of language. In a more general sense, generative relates to the capability of production. 

Drawing from those two root definitions and into the design space, we can start to see this AI-led technology as more of a methodology and design workflow, rather than just a tool.

Generative design embodies the design process, aiding the flow of engineers and architects throughout the process. It isn't meant to replace their roles, rather augment them and make the human element even stronger.

At the heart of generative design is the vision of helping makers explore, optimise and make informed decisions on their design process. A generative methodology is a mindset taken by makers at the onset, to fully leverage all the technical tools at their fingertips, to fully explore the possibilities of design, to ultimately make the best product they could ever possibly create.

All of this is augmenting the design process for engineers and architects. It augments our design logic, letting us consider more use cases. We define the parameters still, though, but rather than having to model all different options that fit with those in the parameters, we can let the computer do that for us. 

After all, humans are perfect for creative tasks. Figuring out all the ways a building or series of buildings could fit within a certain plat, well, that's a little boring. I know I've angered a few architects and civil engineers out there in saying that, but I'll pose this.

Rather than exploring 25 different ways to stack and orient buildings on a landscape manually, why not let the computer give you the options, and then you can pick between them. The time savings should be evident at the forefront.

The time you do save in that process can then be used to make the structure more unique, it can be better spent on human tasks, all while still fitting the bid and scope of your client. 

A generative methodology is the idea that while designers may enjoy the quicker more menial processes of figuring out where to put all the windows in an office building, why not let the computer develop the choices while we work on what humans do best.  

According to various studies, the global engineering and construction market is set to grow 11.8% in the next five years. The market is already seeing a labour shortage, which means that existing architects and engineers are facing a dilemma. More work coming in than they have time for.

A generative approach to design seems to help solve this problem. Not only can the generative algorithms working inside of Autodesk CAD already take away more menial design tasks from you, the maker, but you can also be left knowing that the design you chose is the most optimal. Good old mathematical validation. 

Generative today

If you've made it this far in the article, hopefully, I've been able to convince you to some degree that generative is the way of the future. It doesn't have to mean that your buildings look like futuristic spaceships.

It doesn't have to mean that your parts can only be made through additive manufacturing. Rather, it can simply mean that your building is optimised for light, optimised for HVAC efficiency. Or, it can simply mean that your part is as light as it needs to be and as strong as it needs to be. 

Don't look at modern generative as a way to create futuristic parts, look at it as a way to aid in the creation of your parts now, ensuring they're fully optimised, and removing the more menial workflows from your standard design process. 

Ever since the onset of accessible computation, it's been used to alleviate the drudgery of tasks. Taking away the repetitive, the tedious, and letting humans focus on the creative, the provoking. 

Generative design is just the application of intense computing algorithms within the realm of engineering and design. Taking away the repetitive tasks of today, the tediousness of the process; letting you create the next provoking structure.

Generative design and its natural aptitude at optimisation also means that it can play a huge role in a variety of engineering and design processes, far beyond what we've discussed here.

In the realm of sustainability, you can use generative to maximise the thermal efficiency of a building all the while maximising roof space and solar coverage for photovoltaic cells. The possibilities are endless if you're creative enough to leverage the tools for your own power. 

Currently, it's estimated that 37% of modern design firms utilise generative design in some form or fashion. While that does represent a third of design firms, it still means that the adoption is in the early phase.

Any depth of research into generative design and its potential impact on infrastructure and design should leave you understanding that it isn't going away anytime soon.

Now appears to be the perfect time to begin adopting it. Right as it's reaching a phase of practicality and useability while still being a futuristic technology of tomorrow, available today. 

This article was written by Trevor English and first appeared in Interesting Engineering.