Foray in computational design - Parachute Phone Case

Hi all,

I’d like to share with you a work in progress project I’ve been working on over the last couple of months in my spare time.

I have been very interested in computational design following the stunning works that have been made in architecture with the likes of Zaha Hadid and more recently in product design with the likes of Adidas and Nike applying very advanced algorithms as part of their design process. (Have a look at Adidas’ Futurecraft 4D if you’re not familiar)

This project, is as much of a learning experience and exploration as it is a design project for me. Market viability was a bit of an afterthought on this one. :laughing:

Drop Testing

Using anthropomorphic data, 4000 tests were devised representing a wide breadth of typical drop situations across the population. Testing parameters include elbow height, angle and length of forearm, initial angle of phone, initial speed of phone, as well as the possibility for an initial collision with the user’s hand.

The results from these 4000 drop tests were then compiled onto a single impact map.


In order to create the final mesh, a set of random points are located on the surface of the phone biased by the impact map.

A purpose built (soon to be shared!) mesh optimization algorithm refines the mesh into more regular triangles while keeping the mesh more dense in impact zones.

The left image being the initial random points seeded from the impact map and the right image after 12 iterations of the algorithm.

And finally, some renders of the final product:

As a bit of background, in my day to day work, I get to use those tools in the context of bespoke landmark playground structures, so my work is much closer to the Architectural uses. I’ve been meaning to learn and see how these tools can be applied product design. Yes this means some work in Grasshopper/Rhino but for me that meant going way beyond some of the low hanging fruits that get associated with that toolset.

This project is still a work in progress for me. I’m hoping to use some of the data I harvested to create some new shapes. Possibly going in a very different direction from the simple mesh. Next step for now is to get this version 3D printed and see how it feels and holds up.

I’d love to hear everyone’s thoughts on the project and maybe even have a bit of a discussion on the possibility of computational design methods becoming much more common.

Neat exercise. Computational design and merging with ID are always interesting since it addresses a specific piece of the design criteria (in this case drop/tumble), but areas which may be subjective (grip, aesthetics, ease of installation, manufacturability) are ultimately left out of the equation.

The auto industry seems to be using a lot of computational design and FEA work to better inform how they can build the strongest parts for the minimal amount of cost/material. You see this a lot for things like suspension arms, crumple zones and other structural components where the strongest, lightest and stiffest part is ideal.

Would be interesting to see the same approach taken for something more organic like a bike helmet.

Even if it mechanically doesn’t make much sense, it still is a beautiful case and in terms of business viability…sell these on Shapeways and you will have many customers. Many phone impacts are local indent impacts though so for true protection the entire sides need to be covered quite densely.

Hi Mike, thanks for the response!

What seems to be interesting right now is that these limitations are slowly lifting. These computational methods are no longer black boxes for the designer but rather something they can interact with and build up.

For the last few years, in product design, computational design methods have been mostly associated with shape optimization methods. IE give a rough shape, a static load and as the software runs FEA analysies, it removes material. Usually you’re left with a truss network that looks like a piece of swiss cheese. :laughing: As you mention, there’s not a lot of uses for this besides informing the designer as to where they can remove material. It doesn’t take into account any other parameter.

This is where things are changing. It seems like CAD was simply a step up from pen, paper, and a calculator. We pour a lot of thought into a product, research, sketch and talk about different avenues, then define its form in CAD, test our output with a mixture of computer tests and real world tests, make changes to our design, more testing and then we’re ready to ship. I see computational design as anywhere we can use computer power to enhance any part of this process. Coding little programs to help develop products is nothing new but it seems like there are finally ways of abstractly defining much more complex geometry and have the program interact with the shape of the product.

Say taking a look at this track and field spike from Nike for the 2016 Olympics,

Their algorithm didn’t simply ingest some data and spit out a model based on a minimum weight. They analyzed running data and interim FEA models to determine required stiffness on top of minimizing weight, keeping a certain aesthetic and ensuring manufacturability and comfort. It looks like they used their data to set some metrics such as spike placement, required stiffness at different area of the shoe and then this was fed into a definition that created the final shape of the sole. This definition probably had several input parameters that affected the algorithm which defined the shape. The designers created that definition in order to achieve a certain aesthetic keeping in mind things like comfort, manufacturability.

To illustrate this, here’s the output for multiples shoe sizes:

Different topology but certainly from the same family.

Ideas like this are taking place at a much bigger level in architecture where it seems reasonable for an architect to write a definition defining the overall aesthetic of the massing of the building and run it through optimization cycles to ensure compliance with urbanism code on top of optimizing for sunlight and radiant energy in the dwellings.

While certainly not the be all and end all, I think such strategies will become a very useful arrow in the designers and engineers quiver. Be it simplifying the creation of very complex geometry to achieve a specific goal, optimize a shape based on a specific set of parameters - even if you only end up using it as an inspiration for a new design direction.

You’re right a bike helmet would be super interesting! The wind resistance/cooling of the rider’s head could possibly be taken into account and make for something pretty neat. The phone was actually quite hard to deal with all the sharp edges and quite small radii.

Thanks for the response Ralph! I’m hoping to get the first one built at Shapeways soon, I’ll make it available for purchase if it’s up to snuff. You’re right about local impacts, I’m assuming the impacts are taking place against a flat surface. One thing the data did show me which is pretty obvious in hind sight is that the most likely point of failure is probably the corners but on the side of the screen. Not only is the glass fragile, most cases don’t have much protection there so not to impede on the use of the screen.

Love topology optimisation.

Years ago a designed a set of car wheels based on minimising their Moment of Inertia while maintaining strength. It’s probably a common process now, but at the time I couldn’t find any other record of people optimising for something more complex than minimum weight.