Introduction: Strewn Fields: Waterjet Etching Into Stone

About: Scott Kildall is an new media artist and researcher. He works at Autodesk, Pier 9 and is an artist-in-residence with the SETI Institute

Strewn Fields maps the impact of asteroids after they enter Earth’s atmosphere and disperse into hundreds of meteorites at four distinct locations: Sudan (2008), Sutter’s Mill in California (2012), Chelyabinsk in Russia (2013) and Sariçiçek in Turkey (2015). At each site, scientists have geolocated and weighed hundreds of meteorites, which I then translated into a physical data visualization that shows the patterns of ejecta.

Using a waterjet, which emits a high-pressure (55,000 psi) stream of water usually used to cut various materials, I repurposed this machine as an etching device such that the impact data gets digitally-carved into various stones, from each site.

The violence of the waterjet machine gouges the surface of each stone, mirroring the raw kinetic energy of a planetoid colliding with the surface of the Earth. The static etching captures the act of impact, and survives as an antithetical gesture to the event itself. The actual remnants and debris has been collected, sold and scattered and what remains is just a dataset, which I have translated into a physical form.

This Instructable will detail how I made this artwork.

Step 1: Gather Data

I made this project, while being an artist-in-residence with the SETI Institute. I specifically worked with Dr. Peter Jenniskens, who is an expert on meteors and meteorites to create this artwork.

When a large asteroid enters the earths atmosphere, it does so at high velocity of approximately 30,000km/hour. Before impact, it breaks up into thousands of small fragments, which are meteorites. Usually they hit our planet in the ocean or at remote locations, so that scientists cannot find the debris.

Only recently have scientists been able to use GPS technology to geolocate the spread patterns, called strewn fields.

Dr. Jenniskens provided me with four datasets: Almahata Sitta (Sudan), Chelyabinsk, Sutter's Mill and Sariçiçek. Some of these datasets are his own cleaned data and some are online data such as the Sutter's Mill impact.

The data points on the meteorite impact are as follows:

Almahata Sitta (2008) had 639 found meteorites
Chelyabinsk (2013) has 177 meteorites
Sutter's Mill (2012) has 76 meteorites
Sariçiçek (2015) has 179 meteorites

This is a lot of meteorites to find at a single location, but the number of data points is not at all overwhelming.

Step 2: Clean Data

The datasets I got from Dr. Jenniskens were text files and not formatted in a way that was easily computer-readable. What I prefer is something such as CSV files, which I can easily read in with software code.

Since the datasets were relatively small, I cleaned them up using a text editor — Sublime is my weapon of choice — and I inserted commas between the fields.

I standardized of the four files to use have the same columns of data. The important fields that I wanted to capture were longitude (x), latitude (y) and weight (size of the data), which you can see here would be column #3, #4, and #5.

When I was done, I cross-checked them in a spreadsheet to make sure it was properly-formatted.

Step 3: Map the Data

With my recent artwork, my focus is on writing software code that transforms datasets into physical objects.

I work with Open Frameworks — an open source C++ toolkit — which can be tedious to set up, but once it is working, is incredibly fast, since you are working with compiled code.

It has some strong features for exporting to post-script (.ps) files, which I can then edit and clean-up using Illustrator. Another front-end tool that can be powerful is Processing, but I find that vector file output never to work quite the way I would like, so I use OF instead.

I did some preliminary mapping of the data as (x, y) positions, showing the latitude and longitude. I then wrote code to make the size of the circle proportional to the mass of each meteorite.

Step 4: Make a Prototype

This is the part of the project, that I try to get to as quickly as possible. Using rough output from my Open Frameworks code, I laser-etched a panel to show the data.

Burnt wood looks cool, but is very much overused. But, what I found from this early experiment was that there is a visual appeal to this data.

Ultimately, however, I knew that I'd have to work with a different material, and probably a different digital fabrication machine.

Step 5: Etching on the Waterjet

I decided to test etching into materials using the waterjet at Autodesk, which opened up a variety of materials that you can't effectively use on the laser-cutter such as plastics, metals and stone.

I setup a tool-path file of a spiral etch pattern with 8 different widths for the spiral path. The shorter the width, the deeper the etch would be.

I initially tried it with some plastic composite scrap that I had, and then some stainless steel scrap. And after that, I worked then some stone samples I got at a scrapyard.

This was the a-ha moment: the stone was a perfect expression of the idea of a meteorite impact.

The crust of the Earth is composed of stone and the high-pressure stream (55,000 psi) from the waterjet would reflect the raw kinetic energy of a planetoid colliding with the surface of the Earth. The static etchings would then capture the act of impact and survive as an antithetical gesture to the event itself.

And so, the actual remnants and debris (the meteorites) have been collected, sold and scattered and what remains is just a dataset, which I would translate into a physical form.


Step 6: Iterative Stone-etching

I spent many weeks iterating with different stone samples. I would visit various stone yards and tile manufacturers and purchase samples, then do the same spiral etch tests. Over and over and over and over.

It wasn't until actually doing the etching that I could figure out what the result would be. How would the stone look inside? Would it shatter or would it be too tough?

What I quickly found out is that a limestone or travertine etched well. Harder stones like marble didn't produce good-looking results. A lot of stones shattered in various ways.

What I decided — and this is from my years of training is that the material you select always needs to match the concept — thus, the stone should align with the site. It's nearly impossible, well, without a huge budget, to acquire the exact stone at each site, but I could come close.

Step 7: Sourcing the Final Stone

After a couple of months and many etching tests, I came up with my selected stones, which I sourced from two local stone yards.

The important thing was that the stone be both affordable and not-honed. The honed stone was expensive and looked like tiles rather than rock.

My final choices were:

Sutter's Mill, located in the Sierra Nevada foothills: quartzite
Almahata Sitta, in Sudan: French limestone
Chelyabinsk: green granite
Sariçiçek, in Turkey: tumbled corsica (travertine)

Step 8: Tool-pathing Tests

The custom spiral tool-paths were fun to work with and good for effective testing, however, they had three flaws that made me abandon doing my own tool-pathing in code.

First, the spiral paths were easy enough for me to write code to generate. But, many of the circles overlap with one another, creating complex 2D shapes, which were beyond my coding skills to generate tool-paths for.

Secondly, and perhaps more importantly, I had problems of lead-in and lead-outs with the waterjet, which created undesirable results at the start and the end of the spiral. This is something I could never tame.

Finally, and also an insurmountable problem, as the spiral neared its center, the head of the waterjet itself would move too slowly, gouging a small hole in the material.

Step 9: Brute-force Etching and More Testing

I came up with a different approach: cut a metal mask which would have the shapes of the meteorites I wanted and then put this on top of the stone to get a clean etching.

This approach was certainly more time-consuming because you have to first cut out the shape from the metal mask. And, the mask had to be thick. The waterjet would spend time on the left and right edges of the traverse and cut out a slight slit in the mask, if it wasn't thick enough.

The answer was 1/4" steel...ouch.

I setup back-and-forth traverses in Illustrator and then did a series of tests for each of my four stones. Super-tedious, but these had to be dialed in *just* right.

I came up with a test sample (the nice-looking quartzite, above), which I was very happy with.

Here are my final settings

Sutters Mill (Quartzite), data etch speed = 150 @ .5mm resolution
Chelybinsk (Granite), data etch speed = 90 @ .5mm resolution
Sariçiçek (Tumbled Corsica), data etch speed = 150 @ .6mm resolution
Sudan (French Limestone), data etch speed = 150 @ .6mm resolution

Step 10: Adjust Code for Minimum Etch Circles

Each data point has to have at least two etch lines in it, so I then spent time re-coding for each of my four data sets. I wanted to work with the minimum for each stone so that I could get a distinct hole in the surface

Step 11: Cut the Stone

I was now facing looming deadlines for an upcoming art show, which was to feature my Strewn Fields.

Enough testing, and time for production!

I spent a whole day just cutting each stone — 5 of each kind. The edition would be 3, but I wanted a couple of extras for slip-ups, which I knew would be inevitable.

I split up the tasks into hours of waterjet-cutting and hours of file-preparation.

Step 12: Prepare the Data Files

This was a fastidious task. Each data file would have a metal mask, which I would prepare in Illustrator. I made extra tabs for clamping the material down.

Then, I would prepare a second file for etching, which would be for the stone itself. There were some alignment shennanigans, as you will see later, but the important thing, would be to make sure that both files had the exact same origin point.

Step 13: Cut Aluminum Backings

Mounting the stone etchings on the wall would be tricky, since these were to be hung and the stones are heavy, about 35 pounds each. Upon the advice of one of my co-workers, the technique that I ended up using was to:

(1) cut aluminum backings, which would be 1/2" smaller on each side than the stone (i.e. for a 18" x 10" stone, the backing is 17" x 9"). The aluminum plate would have holes in it for press-fit pins and also holes for screws

(2) press-fit pins through aluminum plate

(3) screw a french cleat onto the back of the aluminum backing

(4) waterjet etch pin holes on the backside of the stone, then flip and do the data-etching on the front side

(5) do a glue-up with epoxy of the stone onto the pins, fitting the pins carefully in

I did a series of test with pressing the pins in through a sample piece of aluminum to figure out the exact tolerance.

Some details:

The pins themselves are 5/8" long by 1/4"

The aluminum is 1/4” inch thick
I cut out .255” pin holes in the waterjet and pressed them, starting from the chamfered side, pushing them all the way through

The screw holes are .18” in diameter

As you can see, I cut out the aluminum backings en masse.

Step 14: Countersink Screw Holes

Using a countersink drill bit on a drill press, I drilled countersinks for all of the screw holes.

I wanted to do this step before press-fitting the pins in, since having the pins poking out while using the drill press would be harder to work with.

Step 15: Press Pins

Using a manual press, I press-fit pins into the frontside of the aluminum plate.

Step 16: Cut French Cleats and Spacers

On the table saw, I cut a number a set of French cleats and a spacer for each of the stones. Some of the cuts were a bit nerve-wracking. I definitely felt more comfortable with the table saw when I was done.

Step 17: Adhere French Cleat and Spacer

I used a drilling template and a transfer punch on the French cleat itself to align it level with the mounting.

Using #2, 3/4" Phillips head screws, I attached one French cleat and a spacer to the back side of the aluminum backing.

This will make it so the artwork will be properly leveled in the vertical plane. The cleats need to be attached before epoxying the stone.

Of course, you need to be careful to get the orientation done correctly.

The mountings are done. We'll come back to these later.

Step 18: Figure Out Pin Hole Depths

The backs of the stones will have holes for the pins in them for the aluminum backings.

Since each stone has different properties, like with the data masks, I ran tests with a pin hole mask to determine the exact settings for the waterjet etching.

I etched a .32" hole in each one — this is slightly oversized for the .25" pin

The settings ended up figuring out were:

Sutters Mill (Quartzite): speed = 100, (110 is probably ok, too)
Chelybinsk (Granite): speed = 50 (60 is ok)
Sariçiçek (Tumbled Corsica): speed = 150
Sudan (French Limestone): speed = 150

All of this was using a .7mm resolution on the etch lines

Step 19: Cut Out Metal Masks

Using the data file from the earlier step, I cut out metal masks for each of the stones. The metal masks can be reused for a few etchings.

Step 20: Cut Out Pin Hole Masks

Using the same technique as the data-masks, I cut out pin hole masks for the backside of the stone, for etching later.

Step 21: Cut Out Plywood for Aligment

I had to make the stone align exactly to the data and the pin holes.

Using a piece of scrap plywood, I cut a rectangle that is exactly the size of the stone. I used this self-cut as a squaring mechanism for the pin holes and the data-etchings.

Step 22: Etch Pin Holes

I laser-cut an wood piece exactly the size of the stone and place it inside the plywood piece. This is so the stone doesn't get scuffed up in the waterjet.

Then, I placed the stone, put the pin mask on it and ran the etching on the pins.

The result looks solid. There was a small vertical line where we etch through the mask itself, which is due to the fact that the pin-etching is on a very slow setting. But, this won't affect the piece at all and is on the backside, not visible

I tested it with a spare aluminum plate with pins to make sure it fits easily.

Step 23: Flip and Etch the Data

I flipped the stone, put some cardboard on it to prevent scratches and then etched the data itself. Each one took about an hour to etch.

Step 24: Mount With Epoxy

Using a two-part epoxy, I did glue-ups for each of the 4 stones.

Step 25: Done!

Here are some photos of the final pieces. After many months of work, I was super-pleased with the results.

Dr. Jenniskens is on my left showing a fragment of one of the 2008 TC3 meteorites.

I hope this was helpful!
Scott Kildall

For more on various projects, you can find me here:@kildall or