From the editors of

An Introduction to
Generative Design

Generative design is opening new opportunities for solving design problems —

and turning traditional workflows upside down, moving simulation ahead of
model creation. Understanding the basics of the technology can help you decide
if it’s time to share your AEC or MCAD workload with generative design tools.

Includes links to in-CAD and stand-alone

generative design tools on the market today
An Introduction to
Generative Design
This latest design
A
s a CAD professional, you’ve used software to create many
designs — but what if the software itself could shoulder
technology trend taps some of the design work? Just imagine: You enter the
computer power to general criteria for your design, a four-legged chair with a seat and
help architects and a backrest. The chair must be able to support a 300-lb load, but
it can’t weigh more than 15 pounds. Your sales department tells
product designers you this chair can retail for $150, so you figure your budget for
automatically optimize materials and production is $50. Marketing would like it in walnut.
products, spaces, and And your boss wants 25 design alternatives to review by Friday. If
you were relying on traditional tools and methods, that would be
processes. a tall order — but an emerging class of software can help get the
job done with generative design technology.
By Monica Schnitger
Defining Generative Design
»» Cover image: Let’s start at the beginning: What is generative design? It’s a
Part iterations created using category of technologies that suggests design options, or optimizes
generative design in Solid Edge. an existing design, to meet criteria defined by the user. The best-
known of these technologies is topology optimization, which
uses physics to find the most appropriate shape for an object,
given material specifications, loads, weight, and other conditions.
Generative design also includes rules-driven parametric CAD model
creation, which automates the design of complex but repetitive
shapes (and can incorporate non-geometric aspects as well, such as
cost targeting and manufacturing optimization).
Most designers are used to creating a CAD model first,
then subjecting it to simulation just before production,
to predict how the final product will perform. Generative
design upends this sequence, using simulation during the
earliest concept stages to identify alternatives that will
satisfy design criteria. The human designer then selects
the most aesthetically pleasing or otherwise desirable
alternative, and completes the design process.
Why do this? Because it’s easier for humans to make
minor design changes early in the process rather than
major ones later, and because they can only process so
many alternatives at once. Computers aren’t biased

»» The Generico chair by designer Marco Hemmerling of the SPADE Spatial

Design Studio was designed using generative techniques to combine comfort
with response to load forces. The chair is 3D-printed to allow for optimal
material distribution.
 An Introduction to Generative Design  3

toward any one concept

nTopology or design, and can create
hundreds or thousands of
ideas in the time we take
to create one. Enabling
the designer to only select
from alternatives that meet
operating parameters, rather
than considering options
that might not meet these
criteria, prevents wasting
time on nonviable options
and potentially increases
innovation. In a few short
hours, the designer can
examine many alternatives,
»» Element from nTopology was used select one, and finally create the detailed CAD model needed for
to create a lattice inside this pedal, manufacturing and other downstream steps.
reducing weight while maintaining
Generative design has many practical benefits beyond time savings
structural integrity and industrial
and innovation. Manufacturers can optimize designs to use less
manufacturability.
material and weigh less, saving money on materials and transport.
They can replace expensive materials and still meet the design
criteria. Where appropriate, they can move between additive
and traditional subtractive manufacturing processes, facilitating
custom and short-run product development alongside traditional
manufacturing. And as the technology continues to evolve, they’ll
be able to trade off between business and technical aspects to find
the best alternative in a cost–timeline–function analytical envelope.

Types of Software Tools

There are two broad categories of generative design tools:
specialist, or stand-alone tools, and in-CAD tools, which are
incorporated within CAD solutions. Several of the in-CAD tools
specifically cater to the design needs of additive manufacturing
(often referred to with the umbrella term 3D printing). This type of
manufacturing enables the creation of shapes that can’t be milled
or formed using traditional methods, and is especially well suited
to lattices, internal voids, and other structures that reduce weight.
But generative design isn’t limited to product development; there
are AEC-focused tools as well (see the table on pages 5 and 6 for
examples of the various types).
Each software developer has taken its own approach to generative
design, and the solutions are evolving quickly.

Topology Optimization
The best-known technology in the generative design bucket is
topology optimization, which uses finite-element analysis (FEA)
to optimize material layout within a design space. A typical
topology optimization process starts by defining the shape of a
part to be optimized. (If it is part of an assembly, the designer
can also define what not to optimize.) Next, the designer sets
the loads, constraints, and goals for the optimization. Lastly, he
or she defines how and where the optimization will run. This
 An Introduction to Generative Design  4

Altair
»» This engine mount step varies depending on the software used, and might include
was designed using Altair which algorithm to use, and whether calculations will be run on
solidThinking Inspire. Starting a local workstation, in the cloud, or elsewhere. The result will be
with a block of material, one or more design options that meet criteria. Today, the results
factoring in constraints, forces, are likely to be shapes made up of mesh elements, plus visual
and material type, the software representations of physical forces on components. From these, the
creates optimized geometry designer identifies a single, best option.
for the part.
What happens next depends on the use case. For an object
destined for additive manufacturing, a designer might add fillets,
etch part numbers, and perform the other final steps necessary to
support 3D printing before running one last simulation to ensure
that the manipulations haven’t affected the part’s viability.
If the part is destined for traditional manufacturing or other
downstream processes that require adding solids or surfaces, there
are two alternatives: Use the optimization output as a template
and model solids/surfaces over the mesh, or use specialized
software to automate this task. The final part or assembly should
undergo simulation again before moving to manufacturing.
Appearances matter. Topology optimization algorithms can be
based on many different principles; the one most familiar to us is
based on research into bone growth patterns. Bone responds to
external stimuli by only growing as needed — the body adds no
excess material.
Bone-like and other visually unusual structures are often beautiful,
in a way — and, because of the optimization, we know they’re
functional — but they’re certainly different from traditional
designs. Many designs created via topology optimization don’t
meet the accepted design aesthetic, and must be modified to
appeal to buyers’ sensibilities, especially in the case of consumer
goods. A designer might use the structure created by the topology
optimization tool as a starting point, and edit that geometry or
overlay it with more traditional-looking solids to create and end
result that’s more conventional.
 An Introduction to Generative Design  5

A Sampling of Generative Design Tools

Stand-Alone Tools for AEC and MCAD
CAE-focused stand-alone tools can be used in many design applications, for manufacturing or AEC designs.
Company Name Product Examples Technology Applications
Altair OptiStruct was one of the first solutions of this type, and has been used across
Altair OptiStruct industries for lightweighting. OptiStruct is parallelized, meaning that it can be executed
on hundreds of cores to quickly perform large-scale optimization runs.

ANSYS ANSYS ANSYS Topology Optimization technology is part of the ANSYS Mechanical offering, and
Mechanical can be launched from within the Workbench to fit into a simulation workflow.

Dassault Tosca Structure, Dassault Systèmes Tosca is a structural optimization suite that that works with
Systèmes Tosca Fluid ABAQUS, ANSYS, and MSC Nastran finite-element solvers to integrate with an existing
CAE environment.

ESI Group PAM-STAMP, ESI Group embeds generative design techniques into its PAM-STAMP, ProCAST, and
ProCAST, SYSTUS other manufacturing solutions, and offers specific shape optimization in its SYSTUS
simulation suite.

MSC Software MSC Nastran MSC Software offers several types of design optimization solutions, from shape and
Optimization topology optimization in MSC Nastran Optimization to process management solutions
that optimize designs across multiple disciplines.

In-CAD Generative Design Tools

In-CAD tools can be CAE solutions from one of the suppliers listed above, accessed from within the CAD design environment and
then returning to it with the suggested design, or designer-specific tools that are part of the CAD workflow. The user defines a few
inputs — usually loads, the design space, constraints or boundary conditions and, in some cases, manufacturing methods — and
then runs an iterative algorithm to compute the geometry. The solutions listed below are examples of CAD tools with designer-
specific generative design capabilities.
Company Name Product Examples Technology Applications
Autodesk Fusion 360 and Inventor, in premium subscriptions, include access to
Fusion 360, optimization setup and calculation capabilities, as well as to cloud computing
Autodesk
Inventor resources.

Dassault TOSCA suite Dassault Systèmes offers access to its TOSCA optimization suite from within its CATIA
Systèmes and SOLIDWORKS CAD offerings.

Robert McNeel & Rhino Rhino, a CAD package used for both manufacturing and AEC, relies on the Grasshopper
Associates visual programming language and environment to automate design. Users drag
components onto a canvas and connect them logically to specify how a design is
created. They can vary parameters to generate design variants for further exploration.

PTC Creo Simulate PTC Creo Simulate uses Vanderplaats R&D GENESIS to power its optimization
technology. It uses the familiar Creo interaction and workflow, and results are converted
to freestyle (B-rep) objects, which eliminates the need to rebuild or recreate faceted
geometry.

Siemens NX, Solid Edge Siemens NX and Solid Edge integrate Frustum’s Generate kernel to generate optimized
design alternatives. Users can then manipulate the geometry with Siemens’s proprietary
Convergent Modeling technology.

Altair solidThinking SolidThinking Inspire, Altair’s brand for mid-market CAD and simulation, integrates
Inspire topology optimization as part of the CAD modeling process.

Table continues on page 6

 An Introduction to Generative Design  6

A Sampling of Generative Design Tools Continued from page 5

Optimization-Focused CAD
Several tools combine CAD and topology optimization to emphasize design and/or manufacturing optimization over other
design goals.
Company Name Product Examples Technology Applications
Frustum’s geometry kernel and cloud-based app, Generate, combines the company’s
voxel-based design algorithm with finite-element analysis to generate iterations of a
Frustum Generate topology-optimized design. It is available as a stand-alone product or integrated into
Siemens NX and Solid Edge.

nTopology Element NTopology’s Element is a generative, function-based engineering application that

provides instant feedback during design, to optimize both an object’s shape and the
industrial manufacturing process.

ParaMatters CogniCAD The CogniCAD platform from ParaMatters is a cloud-based design platform targeted at
additive manufacturing processes. Users import CAD files into the platform, define load
and design criteria, and set it to calculate.

Optimization Tools for AEC

Many AEC workflows use standard CAE tools to optimize structural members, facades, and other features. In addition, the
repetitive nature of buildings and other types of infrastructure has led to the creation of computational parametric tools.
Company Name Product Examples Technology Applications
Autodesk Dynamo Studio is a rule-based parametric scripting language that works with
Autodesk Revit to extend building information modeling with the data and logic of a
Autodesk Dynamo Studio graphical algorithm editor. Dynamo enables designers to create visual logic to explore
parametric conceptual designs and automate tasks.

Bentley Systems Generative- Bentley Systems GenerativeComponents is computational design software that
Components automates the design process by defining and controlling parametric object attributes
and relationships. This enables designers to explore alternative building forms without
manually building a detailed design model for each possible scenario.

Nemetschek ARCHICAD Nemetschek Allplan and Graphisoft, and other AEC design software suppliers, offer
Allplan and algorithmic design capabilities via partnerships with Rhino/Grasshopper. Graphisoft
Graphisoft ARCHICAD, for example, has a bidirectional connection to Grasshopper that allows
architects and designers to define the parameters of a design and quickly explore many
alternatives.

Algorithmic Modeling and Computational

Geometric Design
Generative design isn’t, however, limited to physics-based concepts
or, for that matter, to manufactured objects. Architects long ago
discovered that computational design (also called algorithmic
modeling) can replace the drudgery of modeling repeated
components, such as the individual stairs in a staircase, and they
now use parametrically driven tools to create such designs. This
labor-saving approach is spreading from AEC to other industries
where complex but repeated patterns benefit from automation.
In the staircase example, building codes govern the vertical
distance between stair treads, the required depth of the tread,
and other aspects of the design. Computational design techniques
use these rules and surrounding geometry to drive design. Rather
 An Introduction to Generative Design  7

Bentley Systems

»» This is a designer’s than modeling each staircase step by step, software automatically
desktop in Bentley Systems creates the appropriate geometry within the building CAD model.
GenerativeComponents software.
But the software’s capabilities are not limited to geometry in a
At the bottom is the flow of rules
traditional “CAD model” sense. Computational geometric design
that govern the design of the various
can also be used in space planning to determine optimum size,
elements. The user can adjust values
shape, purpose, and adjacency of rooms in a building, or the layout
on the left side of the desktop to
of homes in a subdivision. Feed room requirements and rules
vary the design (note the different
into the computational engine, and it creates design variations
roofline from one image to the next),
for building, campus, or neighborhood planning. (Users focused
automatically updating the building
on product development can set up similar rules and calculations
information modeling (BIM) model to
for the placement of weight-reducing holes in tire rims, or for
add longer supports for the roof.
aesthetic considerations in furniture surfaces.)
 An Introduction to Generative Design  8

Getting Started with Generative Design Tools

Generative design tools create alternatives that humans might
not have come up with, and they save time and money. But the
technology is not right for everyone: a bit of familiarity with
physics helps, as does the willingness to explore designs that might
not be immediately obvious or fit into prior generations’ aesthetics.
You should give it a try, however, even if you’re not sure. The odds
are that your existing CAD tool already provides some access to
these capabilities — check with your software supplier.
If generative design is something you’d like to try, consider the
following:
• The conceptual design team probably doesn’t include
simulation specialists, and that’s fine. Many CAD-embedded
solutions are intended for designers: The setup is fast and
easy, as is interpretation of the results.
• Computational geometry can be more difficult to use than
topology optimization, because of the number and types
of rules one can create. Computational geometry engines
are usually set up using Python or another scripting tool as
Generative well as an understanding the rules of the design and which
parameters can be varied — in other words, the ideal user
design is a designer who can code (or a coder who understands the
design).
technology is • Generative design optimizations rely on accurate inputs for
not right for materials, loads, and other factors very early in a design
process. That may mean changing how information flows,
everyone, but and perhaps a more open collaboration with suppliers.
Consider how you share and gather this data to protect your
you should intellectual property.
• It’s likely that a computer-generated design mesh will need
give it a try, to be manipulated by a human for aesthetics or downstream
uses, or to add in elements that were not part of the
even if you’re optimization. Some CAD tools make this easier than others,
but it is always possible, and developers are working hard to
not sure incorporate mesh modeling into their CAD tools. Allow time
for this editing in project plans and schedules.
whether it’s • Think big: Generative design isn’t just about geometry and
appropriate physics. Your simulation could include manufacturing process
type, cost, and availability; material cost, as well as physical
for your needs characteristics; supplier tradeoffs; and many other factors.
• Remember that physics-based simulation early in a design
and abilities. cycle doesn’t guarantee that the final design meets criteria.
Always run a validation simulation before deciding the
design is complete.
 An Introduction to Generative Design  9

Frustum
»» Frustum’s Generate was used to optimize this part for 3D printing. The designer was able to precisely control how this
complex geometry was built, to take full advantage of the latest additive manufacturing capabilities.

The Bottom Line

Let’s be clear: Computers will not replace designers. Humans
still need to validate computer-generated designs and perform
the myriad tasks that take place downstream from ideation.
Generative design solutions can, however, propose concepts that
humans might not have considered otherwise, and save time by
pointing designers toward workable alternatives.
Computational techniques such as topology optimization enable
designers to explore dozens of alternatives, all of which meet the
requirements. We go from testing a single design toward the end
of the process to creating capable designs right from the start.
Seeing the benefit from these solutions, however, requires
organizational change and technology adoption — start now to
think about how you can take advantage of these emerging tools.

Monica Schnitger is the president and principal analyst of Schnitger Corp.

This publication was developed by Cadalyst, the leading source of information

about computer-aided design and related software and hardware
technologies for the AEC, civil engineering, and manufacturing markets.
©2018 Longitude Media, LLC. Reproduction in whole or in part is strictly
prohibited without written permission of the publisher.