With how far technology has come it’s obvious that designing with an algorithm is very much like shaping clay. CAD software has seen several breakthroughs like Parametric Designing, Additive Manufacturing, Topological optimization, and Generative Design. No doubt all these improved facets have helped designers to leverage on the benefits of machine learning, especially with generative design which now allows designers to create parts that they’ve never imagined possible, parts that are stronger, lighter, and more efficient than their predecessors.
What is Generative Design
According to The AEC, generative design is the automated algorithmic combination of goals and constraints to reveal solutions. The generative process of design allows designers to instantaneously generate several iterations of a design that meets all the requirements stated by the designer. Anthony Hauck gives an in-depth analysis of the meaning of generative design here.
Generative design mimics nature’s processes and capability as it relates to design. And just like parametric design, all a designer has to do is specify several technical constraints like strength, weight, dimensional constraints, and several manufacturing specifications.
After the designer has set the constraints, the generative design software takes all the goals of the designer, with the help of cloud computing, and cycles through millions of design options and test configurations. The generative software learns from these iterations, selects the best performing designs that conform to the designer’s constraints, and presents them.
With generative design software, designers can easily create new and effective designs and one of the best parts of the process is that the end result is very organic and gives a sense of harmony between man and nature. It’s a sweet union between A.I. and CAD.
Generative Design and Additive Manufacturing
As the world quickly turns toward additive manufacturing, it becomes even clearer that generative design is a perfect complement to the versatility and low complexity of additive manufacturing. With generative designs, manufacturers, within a short time can quickly get lots of ingenious designs that capture all the possibilities they hope to achieve and build. So the process of generative design is well suited to all the forms of product manufacturing and not just additive manufacturing alone.
Generative Design and Parametric Design
Like we’ve discussed here, parametric design is a feature of CAD software that lets the designer effectively modify the shape, geometry, and curvature of a designed model just by changing the value (parametric) of one part of the model’s dimension. Parametric design has become the norm for all CAD software seeing that it makes the entire design process faster and more efficient.
But if there’s anything generative design has over parametric design is the fact that it doesn’t only offer an easy way for designers to modify their designs, it also co-creates the design. Designers using generative design can rest assured knowing that the algorithm would generate the best design solution for all their problems.
While parametric design is limited to altering the shape of a design based on the parametric values of the model’s length, curvature, depth, etc. generative design takes it a step further and creates a design based on goals like; type of materials, how the product would be manufactured, performance requirements, the strength of the material, the weight it must support, etc. and then the generative design algorithm literally spits out thousands of design iterations that match the goals of the designer.
So in a way, both parametric and generative designs are similar seeing that they both work with design parameters set by the designer, but that’s basically where their similarities end.
Generative Design and Topology Optimization
When it comes to design optimization, topological optimization allows designers to transform an already existing machine part into the most efficient shape by removing excessive materials from areas of the design that doesn’t have any stresses acting on them. Topological optimization has allowed designers to create efficient parts quickly while reducing the manufacturing cost by removing excessive materials from a machine part.
But unlike topological optimization, generative design, based on the given constraints, allows designers to creatively generate an efficient design from scratch. There’s no need to create a model before generatively designing it because the algorithm would create them and optimize them at the same time.
So in terms of practicality and efficiency, generative design is a step ahead of topology optimization, generative design offers much more than what topology optimization offers. And that’s because topology optimization only seeks to mathematically optimize a model to adhere to certain constraints imposed by the user.
Generative Design Software
This method of designing is the near frontier in CAD design and there’s no doubt that in a couple of years the system would be so well established that we’d wonder why it took us so long to find this path.
It gives a picture of what the human mind can achieve when it’s paired together with the far-reaching capacities of A.I. and for now the medium that allows such a partnership to work are generative design software. For now, most of these softwares are usually provided as plugins to other CAD software. So let’s check a few of them out.
Autodesk’s Dreamcatcher
Autodesk is a big name in the world of generative design. Their generative design software Dreamcatcher, a cool name by the way, is cloud-based and lets its users generate a 3D design by following a simple process.
It all begins when after the user inputs all the required goals the end design must meet, as we’ve discussed earlier. Dreamcatcher then evaluates an immense amount of designs generated by the software based on the particular niche the user wants. The software then presents lots of design options to the designer who has the freedom to either tweak the design criteria so the software generates more design iterations or just select any of the designs displayed by the algorithm.
Although Autodesk’s Dreamcatcher is still under development, its been used to generate a lot of impressive design models that have found applications in architectural, automotive, machine and aerospace designs.
Rhino’s Grasshopper
Rhino Grasshopper is a preinstalled plugin that allows users to generate models with the Rhino3D software. Grasshopper is seamlessly integrated into the multipurpose Rhino3D environment that has served various professionals across different fields.
Grasshopper gives its user complete parametric and topological control of a model and also allows designers to explore the generative design process. It’s also worth adding that users would need to possess an understanding of high-level computer programming logic.
Siemens NX
Siemens’ NX, like every other CAD software, excels both in topological optimization and generative design. NX’s generative design ability has allowed designers to create complex error-free products in the shortest time possible.
Other Generative Design Software
- PTC’s Creo
- Inventor
- Fusion 360 etc.
What Are The Challenges of Generative Design
Generative design is a strong leap in both CAD and CAM, and the world is gradually embracing all that it has to offer. But there’s no doubt that it would face its own fair share of challenges. Some of these challenges would arise either from A.I., designers, or even the end-users of the product.
Lots of Confusing Iteration
It’s easy to select the right design from a group of three, seven maybe even ten models. But it gets a lot more complicated when a designer has to select one design from 10,000 iterations that match all his design constraints. How does the designer get to pick the correct one that wouldn’t fail when used, and the process is further complicated in a situation where the designer has no idea of what the design he’s creating could look like, so he might end up sitting in front of his computer for several days, maybe weeks, trying to decide which design to choose when he could’ve easily modeled it himself.
Weirdly Shaped Iterations
Since generative designs require the designer to input design constraints that help guide the algorithm in the search for a model that meets all these goals, it becomes obvious that the more constraints we add to a design, the more misshapen and visually impractical a design model becomes. But we believe that with time, this challenge would be straightened out.
Also, it’s almost too easy for designers to get their hopes up, imagining that generative design would be able to generate multiple design ideas from just the description of what the designer wants. But in the real sense, generative design demands much more inputs from the designers than expected, so in the end, the design might’ve been generated by algorithms but much of the design work was done by people who just took advantage of the software’s capabilities.
A simple look at some of the design iterations generated by the algorithm would reveal how mangled and oddly shaped the tools suggested by the A.I. were before designers had to step in and generally fine-tune them. One weird part about generatively designed chairs is that they aren’t stackable so designers would have to ensure that the finished design for the chairs would be stackable.
Cultural Change
One of the biggest challenges generative design would face would be cultural. Cultural change has been an initial drawback to the advancement of innovation and technology. A simple example of cultural change occurred during the introduction of modern knitting machines in Nottinghamshire where Ned Ludd led the burning of factories in Manchester in fear that the machines would make them redundant. Luddites even went as far as asking the British Parliament to pass a bill that prevented manufacturers from using modern machinery in their factories. Thankfully the British Parliament didn’t grant the offer. But that was in the past.
But when we come right down to it, mechanical designers all over the world would be asked to do things that they aren’t traditionally used to doing, and create products that wouldn’t take the amount of time they’ve gotten used to spending on designing. There’s a certain fear that A.I. based generative design software might lead to lower pay for design technicians and eventual redundancy.
Not to forget that most people still complain that the income level has remained stagnant for quite a while now, and a rise in technology is partly responsible for that. But then, these traditional economic responses are all temporary and with time everyone would realize that technology makes everyone better than they were before.
Legal Implications
Imagine a product generated by an algorithm fails on the field due to a fatal flaw, a legal case would most likely be made that the designer had thousands of designs to pick from and still chose the one that would take the life of trusting customers. It gets worse if some of the unchosen designs don’t have this particular flaw. How do designers defend themselves in court, if it can be proven that they chose the wrong design among thousands of better options? It even begs the question; how do manufacturers ascertain that a part is structurally dependable and wouldn’t fail while on the job.
Why Generative Design is Important
Although generative design is still in its early stages, it has already begun playing an important role in design and manufacturing, and there are several examples of how generative design has helped several industries. Here are some ways companies have leveraged on the power of generative design.
The Aero Industry
Airbus, through the aid of generative design, has begun exploring the possibilities of creating a next-gen. aircraft. They applied generative design in creating a new A320 partition, a part that separates the passenger section from the galley of an airplane.
To design an efficient A320 partition, the new part would have to be attached to the plane in four separate locations, have lighter weight and still be able to support loads during landing and takeoff.
With generative design, Airbus was able to apply several design constraints like loads, type of materials, etc. and the generative design algorithm was able to generate the best design model after cycling through countless design iterations. The best design suggested by the software is based on mammal bones and is 45 percent lighter than the previous partition, and since it was produced through additive manufacturing, it cost a minute amount of raw material when compared to other methods of production. In the end, Airbus was able to 3,180 kg of fuel on each partition produced every year.
In Architecture
When it comes to architecture, generative design works the same as it does for manufacturing, but in this case, more data needs to be collected from both the architect and all the stakeholders involved. Before feeding the software with constraints, the designer would need to understand the location preferences and general style of the building he intends to create.
The designer then creates a geometric system that encompasses several configurations of amenities, neighborhoods, circulation, etc. After that, the algorithm explores countless design configurations, chooses the best ones that satisfy the constraints, and presents them to the architect.
This method was used in the design of new research and office space in the MaRS Innovation District of Toronto, architects used generative design to expand how architecture. As stated above, they began by setting high-level constraints and goals for the algorithm, which then evaluated and generated thousands of design options for the MaRS Innovation District. In the end, the architects were pleased and excited with the top-level design that the algorithm generated.
In The Automotive Industry
Generative design has helped GM create a new seat bracket for their electric motorcycle. The generative design algorithm was able to produce a new bracket that wasn’t boxy or welded together, it was made of stainless steel and although it weighed 40 percent less than the previous model, it was 20 percent stronger.
The entire process was more than enough to convince Kevin Quinn that generative design is a must-have tool for future manufacturing. Quinn even commented that “The motivation to consolidate eight parts into one is twofold. One, we can optimize for mass. But another ancillary benefit is that you’re reducing all the supply-chain costs associated with having many different parts that may be made by many different suppliers, which then have to all be joined together.” And that brings us to the next point.
Raw Materials
Design models created by generative design have always performed better than their predecessors in terms of weight, strength, durability, product, and cost-efficiency. Since generative design creates a model that consolidates several parts into one part, less material would also be needed in manufacturing models that have fewer parts, so in the long-run material cost and waste heavily reduced.
The Furniture Industry
The beauty of this design process isn’t limited to just a few aspects of manufacturing. Generative design can be applied even in the generation of furniture like tables and chairs. Early last year, Philippe Starck had a chair built with the help of a generative design algorithm. The chair uses as little material as possible but still delivers a great product that displays Starck’s creativity and expertise. With time generative design would be used by designers all over the world to create brilliant designs.
In Art
Well if you thought an A.I. couldn’t create art then you’d be surprised to learn that a piece of art generated by an algorithm was sold for almost half a million dollars. The abstract portrait was titled Edward de Belamy. The work was created using GAN, Generative Adversarial Network Algorithm, a two-part algorithm that comprises of the generator and discriminator. It created the piece of art after collecting information from 15,000 units of 14th and 20th-century artworks.
The generator created new images based on the information it had derived from the 15,000 units of art it had analyzed, while the discriminator studied each output, and compared the images it’d generated with the images painted by human hands until there was no difference between the two. The generated results were then printed on a canvas, and the mathematical formula of the A.I. used in creating it was signed on the canvas of Edward de Belamy.
With A.I stepping into art, there’s just no telling how far generative design would go when it comes to helping mankind achieve their design goals easily.
The Future of Design
Without a doubt, when it comes to designing, the future lies with generative design software. Very soon, everyday items, the cars we travel in, the design and layout of our work environment and countless things we’re used to doing would all be created with generative design. And that might lead to several products being designed with new materials altogether and wearing unique shapes suggested by algorithms.
So, as generative design becomes an active part of all the working processes and artificial intelligence quickly rises to prominence, it’d be really great to see the new technology that’d be created in less time, with less material consumption, less fuel wasted, little negative impact on our environment.
Questions on Generative Design
People are still adapting to this method of designing and it might be a while before A.I. is completely accepted in the lives of mankind.
Can Generative Design Make Me Lose My Job?
Well no. According to Jesse Blankenship, the senior vice president of technology for PTC, “It may seem counterintuitive, but AI will increase jobs in manufacturing, the learning curve for generative design is so much less than it is to acquire all the manufacturing know-how to design a good part. AI provides a quicker path to a better part, regardless of whether that part is cast, molded, or printed, and it does so without the designer needing to know very much about how those processes work.”
If you’re not satisfied with his answer, then take it from us that generative design cannot be substituted for the judgment of a designer. Instead, it offers a lot of practical design solutions for the designer to pick from depending on the constraints the designer has placed earlier on. The process is completely dependent on designers and engineers of that particular field to select which model best suits them.
Was It Built For Large Companies Alone?
It isn’t limited to large corporations only. Generative design can be used by anyone anywhere so long as they have access to CAD software. When it comes to generative design, we’ve only scratched the surface of its, and our potential.
Can The Algorithms Design All-By-Themselves?
No, generative algorithms cannot design anything on its own. It needs lots of inputs from the product designers (constraints, goals, or parameters) before it can generate anything.
