3D Printed Bridge Goes Dutch

If you’ve ever been to Amsterdam, you know there are plenty of canals and, therefore, plenty of bridges. Next year, a unique pedestrian bridge in the old city center will go into service. The stainless steel bridge will be 3D printed and also embed a number of sensors that will collect data that the printer — MX3D — and their partners Autodesk, the Alan Turing Institute, and the Amsterdam Institute for Advanced Metropolitan Studies, hope will help produce better 3D printed structures in the future. The bridge will cross the Oudezijds Achterburgwal which is near the city’s infamous red light district.

Since the bridge matches exactly with the model used to print it, scientists hope to be able to map the sensor data to a virtual twin of the bridge very easily. You can see a few videos about the bridge’s construction below. This month, during Dutch Design Week, visitors had a chance to walk across the bridge to generate some of the first live datasets.

If you own a 3D printer, you know how a lot of cool things take a long time to print. This bridge took 6 months, 1,100 km of wire and 4,500 kg of steel. You can imagine telling a friend, “We checked on the print and the filament jammed back in March.”

We’ve seen MX3D before, most recently printing bicycles. If you want to build tiny bridges, you can always use Technic.

17 thoughts on “3D Printed Bridge Goes Dutch

          1. Being pedantic here, but the I in MIG stands for inert gas, which CO2 is not.
            There are some that call the current consumer level steel wire feed welding MAG for active gas, but it’s rare.

            The higher powered wire feed welders can run pure Argon shield, since the CO2 is used to lower the power requirement to sustain the arc, on the lower powered machines.
            Once you get up into the +200A range, the requirement is no longer there, as the machines can sustain the arc without the shield gas assisting.

            That said, using pure CO2 is an option as stated, but it spatters almost as badly as stick welding does.

          2. @Josh: Being pedantic, the I in MIG only stands for inert.

            You shouldn’t be using pure CO2 to weld stainless unless you’re using flux-core wire. You want to keep the weld from absorbing too much carbon, and the slag coating from the flux lets you get away with it.

    1. This thing is only done as an experiment / “because they can” – I’ll bet in basically every way you care to measure it absolutely sucks as a construction method – energy use, construction time, material strength, flaw detection, repairability…

  1. i guess people forgot that this is more of a “can it be done” project then a “can it be done cheaper” project. no reason to talk about the cost when the project is more about concept then anything else

  2. The first idea for a bridge like this in Amsterdam was to use the printer on site and build the bridge from both sides of the channel to the other.
    A few ideas based on this concept have been presented, for example this one https://www.remieconsultants.nl/nieuws/3d-geprinte-brug-van-metaal/
    It seems that none of them ever worked outside a lab or factory.
    This last example is far from any of these ideas, its just a combination of large 3D printed structures, not a single structure, and in my opinion just an advertisement for a startup.
    That doesn’t mean its not interesting, the sensor data is really worth, specially the data just before the bridge collapses :-)

  3. I suspect that the value of the data and metadata collected will more than offset any additional cost. There is a list of 20 companies/institutes which will have access to all data collected.
    It kind of has me wondering can you extract and identify multiple people crossing the bridge simultaneously by the timing of their strides (with enough processing power and a large enough dataset). I’m just curious more than anything else in this instance. But I’m not just thinking about this one bridge, I’m projecting ahead to smart pavements/footpaths silently harvesting data.

    1. This sounds like a straightforward application of signal processing.

      The amount of time and number of footsteps used to cross the bridge gives you a fairly large dataset per person. If you assume that the footstep cadence has minor variations per person it seems like it would only take a combination of self-correlation (to match footsteps to each other), Fourrier transform (to select different people by their cadence frequencies), and a bit of AI processing to combine probabilities for prediction.

      The result would be a set of prediction probabilities, such as 81% chance of 3 people crossing right now, 10% chance of 4 and 9% chance of 3.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.