Dear autonomous sailing fans,
Ada, our autonomous sailboat, appears to have experienced a rudder control failure, probably mechanical in nature (servo or linkage), and is currently headed south on a close-hauled bearing. The failure occurred about 800km into her journey, in over 20 knots of wind, while she appeared to be planing at 12.41 knots. Up until that point, everything was working beautifully, and this trip had been a major success! Ada has already set multiple records on this journey, sailing further east than any boat ever has autonomously crossed the Atlantic Ocean.
We have fielded several questions now about potentially recovering Ada. Ada is over 800km from the nearest coastline (St. John’s, Newfoundland), and heading south into an even more desolate area of the Atlantic. Frommarinetraffic.com, we can see that the only ships in the vicinity of several hundreds of kilometres are tankers and cargo ships.
Previous boats that have attempted this trans-Atlantic voyage have been recovered: by the Canadian Navy, the Coast Guard, and fishing vessels. Ada isn’t near any of these types of vessels. She’s simply too far from the coast, and we think that these are the only types of vessels from which we can reasonably hope for assistance.
For now, Ada is truly on her own. We will daily consider the feasibility of a recovery given traffic in the region, but the opportunity for recovery might take several weeks or even months. In the meantime, Ada’s electrical and satellite systems seem to be working fine, and those will remain on to aide a recovery.
This has been an incredible journey with all of you, and we can’t thank you enough for your support! When we first began this project three years ago, we knew that we were attempting a crazy engineering challenge that would involve tens of thousands of volunteer hours, the support and true dedication of so many incredible individuals, and many all-nighters. This entire project wouldn’t have been possible without the support of the University of British Columbia, our many sponsors, our amazing industry mentors and advisors, and the incredible people of Newfoundland who helped us every step along the way in the lead-up to saying good-bye to Ada as she began her trans-Atlantic voyage.
Just launching Ada from St. John’s and seeing her so successfully sail those first 800km has given our team the conviction that we really did accomplish something, and this failure is a great lesson for our team in our need to focus on iterative design, mechanical reliability of critical components, and component testing for future autonomous ocean-going boats!! We do hope to recover Ada, and we will monitor traffic daily in the region which could facilitate this. If recovery does take place, we would expect it to take place once Ada is much nearer to shore. Ada was just too good a sailor for the first 800km of her journey for us to realistically reach her right now!
We will continue to keep you updated on Ada, and will also soon be excited to share how this project will shape our next one in a few weeks! Our team’s dream for autonomous sailing continues, and we can’t wait for you to continue this voyage with us!
Our very own Ada testing lead and team mother, Vivian, spoke to CBC about our launch plans and progress so far. Watch the full video here.
The winds picked up earlier in the day, and we had a very successful sailing session to test updates to our routemaking algorithm. Testing went well, and high fives were thrown around. In the evening, the electrical team will be back at Royal Newfoundland Yacht Club adding safety features to some of the power electronics aboard Ada.
Winds in St. John’s did not pick up until around 6:00 PM, which gave the electrical team had some time to get some valuable diagnostics on our power control as well as communication between the micro controllers and the Battery Management System to enable live diagnostics and keep the team updated on Ada’s vitals. We hit a snag with a non-critical function earlier, so the lack of wind gave us a chance to work on bug squashing.
Days 19-24: Sailing and Testing in Conception Bay, Newfoundland
As a bit of background info, the RNYC docks are separated from the open waters of Conception Bay by a long breakwater. This creates a narrow channel that the boats go through under motor power, which we have to tow Ada through to reach the bay. However, right next to the docks is a more sheltered area, about the size of a large pond. On weekdays, it’s filled with kids learning how to sail, but it’s a great place to run system checks for Ada before getting into more dangerous environments.
Back to our testing efforts, on Sunday (Day 19) we were able to set sail in Newfoundland waters for the first time ever! It was an extremely rainy and windy day, so we stuck to the pond to test all of Ada’s systems and make sure our point to point sailing was still working. Having never sailed in 20 knot winds before, we were rather unprepared for how far Ada would heel over in such strong winds. Her 75kg keel meant that she recovered beautifully, and she passed the day’s tests with flying colors.
In the following days our cellular modem became extremely unreliable, so days 20-24 alternated between staying at the dock to resolve the issues with the modem and going out for hours to test Ada in Conception Bay. We got a little over-excited on our first day of sailing in open waters and covered over 30 km in one afternoon/evening, proving Ada’s long distance sailing abilities! The subsequent testing days focused more on fine details. Some highlights include:
- Sailing in octagons to test sailing abilities for every point of sail
- Successfully sailing to waypoints directly upwind and downwind
- Enabling routemaking code (Ada plotting her own short-term waypoints between longer distance ones we send her) for the first time and having it actually work without weeks of debugging!
- Spotting whales and sea otters out on the water
- Swearing that we wouldn’t open up Ada’s hatches because she is being stored in water, and then opening them up twice for cellular modem debugging and updating code
- Staying out until midnight working with headlamps to test new software for our control boxes
- Beginning to test Ada’s ability to avoid other ships using AIS (Automatic Identification System)
- Show and tell day with the young sailors
- The testing team getting really excited about burrito day at lunch
While our testing team rotated people on and off the chase boat, our home team kept everyone fed, made many more trips to hardware, marine, and grocery stores, and scouted out potential ships for launching Ada. We have been very encouraged so far with the progress we’ve made on the sailing days we’ve had. Ada has been passing the tests our software team throws at her with flying colors, so we are well on our way to launching her in time to get our team back to UBC for the start of school!
Day 17: Moving out of MUN
Soon enough, we arrived at the day we planned to move out of MUN. There was a rush that morning to finish our remaining tasks to be done in the workshop. The members of the electrical and software teams coordinated to finish integrating the relay circuit for the motors. Meanwhile, Madie and Riley made friends in the tech machine shop who were kind enough to help bend a stainless steel plate into place to complete the rudder mechanism.
Mechanically, Ada was ready to move out, and with this being a Friday, we didn’t want to be trapped in MUN for the rest of the weekend given that we needed Craig, the lab supervisor, to open a bay door for us. So, we moved Ada outside of the lab and continued working in the parking lot for the rest of the evening, watching the cloudy sky carefully in case of rain.
Luckily, it stayed dry, and Ada happily spent the night in her trailer outside of the team house.
Day 18: Moving into RNYC
Things went fairly smoothly with moving Ada into her new (and last) home at the RNYC. We put the keel on, craned her into the water, and towed her to her dock space by mid-afternoon. A big shout out to Steve for coming in on his day off to operate the crane! Ada was visited by many members of the RNYC that day, and continues to be throughout her stay here. Everyone here is so enthusiastic about our project and wants to help in anyway possible. It’s amazing!
The next three days were a flurry of working in the MUN lab, late nights, shopping trips, and many meals brought to MUN. Some highlights include:
- Meeting many supportive people at the Royal Newfoundland Yacht Club: the Commodore, Leo Quigley, the Manager, Lori Anderson, the crane operator, Steve, and the President of Sail Newfoundland and provincial team coach, RyanKelly who sponsored us in kind with an amazing Zodiac chase boat
- Connor, Madie, and Oscar joining Dave, one of the members of RNYC, on his sailboat Nomad for Wednesday evening racing, winning their division and receiving compliments on their spinnaker skills
- Head chefs Jordan and Connor arranging a fantastic team dinner featuring bangers and mash late on Thursday evening at MUN to keep the workshop team fueled up
Otherwise, these days settled into a pattern of arriving at MUN in the morning and working through the day until we had to leave the lab at 11pm. The team only stopped for lunch and dinner when our home team brought meals to MUN to enable short mealtime breaks, and thus more working hours. Sprinkled throughout were lots of runs to grocery, hardware, and marine stores.
The electrical team did a great job of reinstalling Ada’s guts and catching any errors along the way. A new rudder motor was installed, more testing was done on our charge controllers, a relay circuit to switch power to the motors (which was previously installed but not used) was re-implemented due to power considerations and we verified that no connectors were failing.
The mech team overhauled a few components on Ada: replacing the rubber bow skin, reassembling the rudder mechanism, applying Loctite to many fasteners, replacing the rudder motor and making clamps hold the port side control box securely in place inside the main hatch. They also added a foul weather retrieval system, in case we encountered conditions where it wouldn’t be safe to come up right alongside Ada to attach a tow line
Finally, the software team worked night and day, not only from the team house and MUN but also remotely from Montreal and Vancouver, continuing to improve existing functionality and add new intelligent behaviors. All components of the system benefited from significant improvements; most notably satellite communication and routemaking since all the individual components were ready for integration. Despite the countless hours of hard work that was put into obstacle detection and avoidance using our infrared cameras, it was decided that our machine learning algorithms had not been exposed to enough data that simulated Atlantic sailing. Even though the code is finished, there is not sufficient time to test it to make sure that it actually helps Ada more than hurts her. As such, Sailbot software decided to implement an image service that will record data for the duration of Ada’s Atlantic journey, training our obstacle detection for our future projects. The control systems are now very stable and are able to account for a wide variety of scenarios, including sensor failures, power failures, and more.
Picking up where we left off on Day 10, we find ourselves on Saturday, July 30. The team members in Newfoundland at that point spent the day sleeping (recovering from the marathon sprint across Canada), getting to know our landlords, and scoping out the nearest Tim Hortons, Subway, and Domino’s.
We were very interested to learn that Wayne and Julie, our landlords, run a whale rescue program out of the St. John’s area. Like all of the Newfies we met, they are super nice and helpful people, going out of their way to find more mattresses for the 10 people staying in the team house.
After spending a day recovering, we finally found the energy to get some real work done on Day 12. The mech team made a temporary fix for the broken back axle on our cradle so that we would be able to move Ada out of the trailer into the Memorial University of Newfoundland (MUN).
Lots of food was purchased to feed the bottomless pits found in many of our team members’ stomachs—dinner included a curry that was delicious, but perhaps a little too spicy for some members:
“It’s so good, but it tastes like pain!” –Chantelle
Riley and Connor also performed city reconnaissance by going down to the George street festival to visit UBC Engineering Physics class mate, Theoren. He has kindly allowed us to ship various packages to his address before we arrived in Newfoundland, as well as assisting us in finding resources in the city.
UBC Sailbot owes Justin Royce, Emma Williams, Dennis Peters, Craig Mitchell, MUN Sailbot Team and affiliated members of MUN big time for arranging some space in the fluids/hydraulics lab in the engineering building and other resources at MUN for us! We moved into our corner of the lab on Monday (Day 13) and met Craig, the lab’s coordinator. After receiving the safety orientation, we could finally get to work!
The mech team used the 5’ aluminum bar Oscar flew over with him to replace the back axle of Ada’s cradle and removed the torn rubber skin on Ada’s bow. The electrical team started reinstalling all of the electronics into Ada. Software resumed their efforts to finish and debug code for routemaking, satellite communication, and various other functions.
The launch team gained two members this day with the arrival of Jordan in the morning and Madie in the evening by flight. After a loud team family dinner, Jemima and Madie moved into a Couchsurfing apartment with Dan, a systems engineer at MUN. They certainly appreciated having easy access to a shower and a quiet environment whenever they wanted to sleep, instead of the rather bustling house team.
UBC Sailbot is a team of engineering students who design and build autonomous sailboats. In 2004, our team began building what would be our first vessel to enter the International Robotic Sailing Regatta. After claiming first place three years in a row from 2012-2014, our team decided we were ready for a bigger challenge: we would build a sailboat to cross the Atlantic by itself. After three years of work our boat Ada is finally ready for our journey from Vancouver, British Columbia, to St. John’s, Newfoundland, and, we hope, for her journey across the ocean.
The last day of preparation in Rusty Hut (the UBC warehouse where Ada has been kept during her construction) consisted mostly of packing. To celebrate the beginning of our final journey, our team went for a celebratory meal at a local pie shop.
Tools were moved, packed and put into two vehicles that will be crossing Canada. Ada was carefully wheeled into the trailer, and her keel was strapped onto the cradle with fire hose. A Suburban SUV served as the main transportation vehicle, which housed six passengers and towed Ada in the trailer. Drivers of this car were well-trained for tough days of trailer-driving. A secondary vehicle, nicknamed Millie, fit the other four members of the launch team.
Once the tools were packed and the checklists checked, the crew gathered together for one last celebration in Rusty Hut. To mark the occasion, several members of the team dyed their hair blue. Blue is the colour of the ocean, and of Sailbot’s main logo. Everything was done, and we were ready to set off.
The launch team started early, with team Millie heading off first, followed shortly by the Suburban.
We travelled through the Okanagan, stopping for fresh fruit and produce. Shortly after, we hit the Rockies.
After nine hours of driving, the launch team finally gathered and rested in Banff. There, we had a meal of smokies, chicken, corn and tofu.
The team got up early to prep a great campsite breakfast of coffee, tea and homemade parfaits.
After a bit of deliberation, the team decided that our secondary vehicle, Millie, might not be able withstand another six thousand kilometres on top of the 400 000 km’s already on her. So, we went to Enterprise, which gave us a student discount on a Chevy Malibu which we used for the rest of our trip. We promptly dubbed the new car Phyllis.
Halfway into our drive, Riley got a phone call from his radiologist. As it turns out, his ankle was broken instead of sprained. With bone fragments floating in Riley’s ankle team Phyllis pushed through to Regina where they were able to drop Riley off in the hospital.
Meanwhile, the Suburban team installed sway bars for the trailer. Endless fields of Canola later, both cars spent the night in a RV campground 10 minutes down from the hospital. This was our second late night in a row. With the exception of Riley, who spent the night in the hospital, everyone slept deeply and soundly.
We started off the third day with blueberry pancakes.
Riley, now freshly in an air cast for his ankle, was picked up from the hospital in the morning. Refreshed and renewed, we headed off for Ontario. On the third day, our vehicles travelled through three provinces. Obviously, it was quite a long driving day. Luckily, the Malibu ran ahead of the suburban and trailer and prepped a meal of butter chicken for the team.
We planned for a quick breakfast and an early start. Since day four was to be our shortest driving day, we had hoped to be at the campsite for about 4pm. However, as we were packing up for our departure, we discovered that Ada’s cradle had broken.
Our plans for an early start had been dashed; instead we stayed until 1pm at the campsite to put together a fix for the cradle. The wooden axle beam broke from vibrations on the road. To fix it, the mechanical team came up with an workable plan: they created a gusset to patch the broken beam to last only until arrival at Memorial University in St. John’s, at which point the entire beam would be replaced with an aluminium version. This aluminium beam was designed by Dave Tiessen, one of our former mechanical leads, then was generously manufactured on short notice in Vancouver at Fluxwerx Illumination. Oscar Janzen, one of our current members, packed the beam in with flew out to St. John’s with the beam.
We did get to pit stop by Canada’s largest goose in Wawa, Ontario.
After a refreshing sleep and a quick breakfast by the water, we headed off to our last campsite in Ontario.
Due to good traffic conditions, Phyllis’ team made it the next campsite in Ontario for 6pm, and began cooking black bean chowder. Meanwhile, the Suburban was slogging its way through a torrential rain storm roughly an hour behind the Phyllis team. Environment Canada had released a severe storm warning, and team Phyllis was blissfully unaware. A phone call from the Suburban later, team Phyllis was madly packing up camp to beat the rain. We drove to the next town over, Sudbury, and found a Best Western for the night.
We definitely took advantage of the continental breakfast.
Since the team had enjoyed a great rest at the hotel, the Suburban crew went for a visit to Parliament Hill in Ottawa. Four launch team members took Phyllis directly to the next campsite in Quebec, to enjoy a swim and a hike in the wilderness instead.
About an hour after the group in Phyllis entered the campground, we received another rain warning from Environment Canada. It would very likely be stormy again that evening. We scoured AirBnb and found a nice little apartment for our two days in Montreal.
It was located in the 10th safest district, next to a park with a swimming pool. Tiredly, we piled in, reheated leftovers and went to bed.
The entire team took a rest day in Montreal. Alan, who had been sick for the last few days, left the house alarmingly early to tour all the metro stations in the city, and bring back a full detailed report of architecture in every station. Kurtis followed soon after.
The rest of us had a lazy day in the apartment, and left to see some of the city around 2pm. We stopped by a beautiful church, then headed down to a park by the water. There, we found small remote controlled boats racing in a pond. The team enjoyed some ice cream and took a nap.
That evening, we headed to Google’s Montreal office for a tour from our software subteam lead, Arek. We worked up an appetite after seeing all the joys that working for Google had to offer, and promptly went for poutine at a nearby pub.
Riley had an exam to write in Montreal, so Team Phyllis cleaned the apartment and packed up the remaining camping supplies. As Riley had been unable to experience all the joys Montreal had to offer the day before, we took a dinner stop in a very small town just before leaving Quebec. The primary language was indisputably French, so we were very glad to have the bilingual Connor Vandenberg with us to order our meals.
On the final stretch of our drive for the day, we were once again plagued by a hailstorm. For safety’s sake, we pulled over beneath an underpass to protect our rental vehicle until the worst of the storm passed.
In the morning, we packed up at our final campsite and took drove eight long hours to the ferry from North Sydney, Nova Scotia to Port aux Basques, Newfoundland. Along the way, we saw many small towns advertising lobster dinners. Salivating, we decided to wait until we hit the city with the ferry port to eat. Upon our long awaited arrival in Port aux Basuqes, we discovered that the only lobster joint in town was booked fully. We just made it on the evening ferry, and the team immediately went to sleep on the extremely comfortable recliners.
Upon awakening at 7am, the ferry had docked. We promptly got off the boat and straight into Tim Horton’s. We were surprised and charmed by the Newfoundland accents from the Timmy’s cashiers. Our drivers then took shifts in both vehicles until the launch team and Ada arrived safe and sound in St. John’s. We are graciously hosted here in St. John’s by Julie and Wayne Ledwell, who heard about our project from this very blog!
by Madie Melcer, Mechanical Co-Lead
The last, but certainly not least, major part of Ada to be completed was the bow, or the front of the boat. Work started on the design of the bow back in Fall 2014 and proceeded intermittently until the task was picked up again in earnest in late January 2016. The past few months have been a flurry of designing, building, rethinking, and more building! This post will focus on explaining what we’ve done to manufacture and complete the bow since January and why we’ve done it.
But first, let’s take a few steps back. Why was Ada’s bow designed and build separately from the rest of her hull? The short answer is that since the bow’s purpose is to protect Ada in the event of a collision, it needs to handle much more of a beating than the rest of the hull. In terms of design considerations, that means the foremost focus is energy absorption.
Some key solid mechanics concepts to understand the more technical side of the bow design and requirements:
- Stress: the force per area on a material
- Shear: when a material fails by having planes slide against each other; think of bending a paperback book and watching the unbound edge split from one vertical edge to chunks of pages that eventually slip over each other.
- Plastic vs elastic deformation: When you apply a force to compress a material, it can deform elastically or plastically.
- Elastic deformation is temporary; materials will return to their previous uncompressed state when the force is removed. Elastic deformation requires less energy, so elastic materials are more “squishy” and easier to deform
- Plastic deformation is permanent; materials stay in their compressed state when the force is removed. Plastic deformation requires more energy, so plastic materials are harder to deform.
- Compressive strength: a measure of how hard it is to compress something, or the force per area required to deform a material. The compressive strength of plastic materials is much higher than elastic materials, and higher compressive strength requires more energy to deform a material.
- Deceleration: Just like the human body, Ada can only tolerate a certain rate of deceleration before something breaks. The higher the compressive strength of a material, the more energy it will absorb and higher the deceleration it will provide. This is important because the bow has to absorb Ada’s entire kinetic energy, but has to do it slowly enough that we don’t overwhelm Ada and break her.
There have also been some key concepts carried forward from preliminary design in the 2014-2015 school year and the work completed last summer:
- How much energy the bow would need to absorb in order to protect Ada in a collision, which was calculated as the kinetic energy of the boat at full speed. Ada’s mass was estimated to be 250kg, and the top velocity we expect to achieve is approximately 7m/s:
- The idea of a multi-phase bow with some parts deforming more easily than others to cover a wider variety of collision severities was carried through to this year, albeit in a modified form.
- Summer of 2015 saw the construction of a male model of the shape of the bow using foam slices (more detail about that below). This model stepped into a placeholder role as the bow on Ada from last summer until May 2016.
- Summer of 2015 also saw the construction of a female mould of the bow using the male foam model.
- Some Impaxx foam was purchased last summer to use as the plastic material in the bow. Impaxx is a foam used as an impact attenuator in Formula One cars, so it seemed like a good candidate to absorb all of Ada’s energy in the event of a collision.
We started again in January with these solid mechanics concepts and previous knowledge in mind. Perhaps unsurprisingly, there is a bit more to the bow than just absorbing Ada’s energy in a collision. We had four main concerns to address with our design:
- Selecting an elastic material for the front portion of the bow
- Determining the ratio of plastic to elastic material to ensure that sufficient energy could be absorbed while keeping deceleration low enough to keep the boat intact
- Determining the maximum allowed deceleration
- Minimizing the weight of the final design
- Accounting for the possibility of friction during a collision
Even taking all of the previous work over the past year and a half into account, these concerns could have made up a capstone project for a group of fourth year engineers. As it was, we did the best we could with limited time and resources, and are extremely grateful to UBC professors, industry experts, and graduated UBC SailBot team alumni Dave Tiessen and Kristoffer Vik Hansen for their help and advice.
Selecting Materials & Minimizing Weight
As mentioned above, there was some Impaxx foam purchased last summer to be the plastic portion of the bow, but we still needed to pick the elastic material. Keeping weight in mind, we investigated the possibility of having front portion of the bow be inflatable.
We quickly learned that an air bag, such as you would find in a car, is only meant to be inflated for short periods of time. We wanted ours to be inflated at a consistent pressure for at least as long as it will take to cross the Atlantic, so at least three weeks to be safe. Thus, we turned to an air bladder. Air bladders are intended to stay inflated for years; they can be found in rubber inflatable boats and can be extremely durable. I consulted with an engineer from Zodiac Hurricane Technologies, the Vancouver branch that manufactures Zodiac inflatable boats. After considering a series of options, it became apparent that any inflatable option would not be able to hold a shape, so we reluctantly abandoned the idea of air bladders. Instead, Zodiac referred us to a foam specialist at Norseman, who helped us find an appropriate low-density lightweight elastic foam called Plastazote.
By the time we abandoned inflatable options, weight had become less of a priority for our team as a whole, so we accepted the trade-off of using heavier foam air in order to have a bow with a better shape.
Determining Ratio of Plastic to Elastic Material
Once we determined we wanted to use Plastazote and Impaxx foams, we had to figure out how much of each foam was optimal to absorb as much energy as possible without damaging Ada with too high of a deceleration. In order to do this, we collaborated with Dr. Chris McKesson, a professor of naval architecture at UBC, to first establish what Ada’s behavior would look like in a collision, and then to model that behavior to optimize it.
We focused on a situation consisting of a head-on collision with Ada moving at her maximum speed, crashing into a stationary object. In this idealized situation, we expect Ada to rotate around her transverse axis, bringing the nose up. This would happen because all forward motion of the hull would come to a halt, while the 75kg keel bulb would want to continue its motion, causing the entire boat to rotate.
Ultimately, it was this rotation that helped us decide that the first point of failure on the boat in the event of a collision (and thus the point that we should focus on) would be the bolts that hold the keel in the boat. Those bolts were designed to have a safety factor of 3 if (theoretically) Ada was standing on her nose, under the effects of about 10m/s^2 of gravity. With a safety factor of three, we set the deceleration limit to 30m/s^2.
Dr. McKesson then assisted us with modeling the keel’s rotation about the center of buoyancy as a spring system; the spring force served as an analogy for the righting moment of the boat, which we could use to find the maximum displacement of the spring (or keel). For those not familiar with the naval architecture concept of righting moment, it is essentially the resistance of the submerged portion of a boat to rotating away from a neutral vertical position. This surprisingly helpful Wikipedia article explains it far better than I can: https://en.wikipedia.org/wiki/Metacentric_height.
For the sake of brevity, I am going to skip over the majority of our iterations of what turned into a very tedious calculation. Anyone who is interested in more detail, please feel free to leave a comment and I will be more than happy to explain every step of the process! But in short, using a small angle approximation, the righting moment of a boat for a given angle of rotation is defined as:
where GM is the radius of application of the righting moment. After some algebra with spring equations, we derived the following equation for the analogous spring constant (x is the arc length the keel bulb would travel while rotating around the center of buoyancy)
and substituted the above expression for the righting moment into it:
Once we obtained a value for the spring constant, we plugged that into a rearranged spring equation:
where x0 is the maximum displacement of the keel bulb, vi is the initial velocity of the keel (assumed to be Ada’s maximum velocity), and m is the mass of the keel bulb. From there, it is a simple matter to use kinematics to find the deceleration of the keel bulb for the given initial and final velocities and the displacement we just found above. Our initial attempt with this calculation gave us a deceleration of approximately 40m/s^2, which is higher than the allowed 30m/s^2. However, we didn’t account for the damping effects of the bow, so that was a reasonable result.
Of course, the small angle approximation is hardly accurate for our worst case scenario collision, and we simplified the rest of the situation severely by assuming that Ada would only rotate in one direction, that the object she hit would be stationary, and that the keel bulb would rotate around the center of buoyancy. We ran through many variations of this calculation, first trying to eliminate the small angle approximation, and then attempting to address other troubles. Our results all pointed towards the fact that the keel would separate from Ada without the Impaxx foam in the bow, and we wanted to create as low a deceleration as possible.
Somewhat limited by the quantity of Impaxx available to us, we decided that 40cm of the length of the bow would be Impaxx. Another 40cm in front of that would be the elastic Plastazote. Finally, the remaining space behind the Impaxx would be filled with lower cost insulation foam with similar properties. This substitution seemed reasonable to us because the Impaxx should absorb the entire collision, so the insulation foam should not need to absorb any energy.
Now for the last step: ensuring that the bow wouldn’t create too large of a deceleration. We assumed our elastic foam would have negligible contributions to energy absorption during a collision, and just focused on the Impaxx foam. The more of a material you have, the more force it will take to crush the material, and thus the more energy it will absorb. However, according to Newton’s second law of motion, that will also result in a higher deceleration. So we turned to algebra once more and rearranged the equation for stress (σ), substituting out force:
The compressive strength of Impaxx at 50% compression is 434kPa, so substituting that in along with Ada’s mass and maximum deceleration gave a maximum allowed surface area of the Impaxx foam of 0.0576m^2. Unfortunately, the frontal area of the Impaxx portion of the bow was above that value. Our solution was to cut holes through the Impaxx to remove material, decreasing the surface area, which decreases the force used to crush the bow, and finally decreases the deceleration of the boat to a manageable level.
Accounting for Friction
We already established that in a collision, Ada would rotate such that the bow rises up into the air. This means that the bow might drag along the surface of whatever object we collide with. The force of that friction was estimated to be roughly equivalent to hanging a Smart car off the bow, so there was concern that the foam in the bow would fail by shearing apart.
By this point in the process, we had very little room to modify our design. We ultimately settled for a Kevlar lay-up done vertically along the longitudinal axis of the plastic portion of the bow. This Kevlar will aid in withstanding any possibly shear forces through the foam; it will not have a significant impact on the crush properties of the foam because materials such as carbon fiber and Kevlar are very weak in compression and will crumple in a head-on collision.
We had already been planning to wrap the outside of the plastic portion of the bow with fiberglass fabric. Hopefully this will provide some similar strength against shear forces in the event of a collision, while also guarding the outside of the foam from dents and scratches.
This design process took us well into March, by which time we were eager to begin construction. The basic shape of our bow was made using the same method as the placeholder bow from last summer:
- Take a 3D model of the bow and chop it up into 2 or 4 inch slices (4 inches for Impaxx, 2 inches for the other foams), sometimes extrapolating those slices forwards or backwards to ensure that each slice covered the outer boundary of the bow throughout the length of the slice. Add holes to the Impaxx slices to minimize surface area
- Cut the foam slices using a waterjet cutter
- Glue together the plastic slices and remove material carefully with lots of sanding to bring the slices into one rough bow shape
At this point, we departed from the procedure used last summer to accommodate our Kevlar addition. We had glued the slices into two separate halves, and then attached those halves with the Kevlar lay-up in between.
The rest of the process continued on more or less as follows:
- Sand more to bring the plastic foam all the way to the intended bow shape
- Smooth the outside of the plastic portion, and then fiberglass lay-up around the outside
- Sand and apply sealant to the fiberglass
- Glue and shape the elastic portion (not cut in half) together separately
- Epoxy the elastic portion to the plastic portion, and then the entire bow to Ada’s hull
- Fill any gaps between the hull and bow, and then apply carbon tape over the boundary between the hull and bow
- Paint the plastic portion and carbon tape
- Cover the elastic portion with a urethane rubber skin (with a UV resistant additive)
- Celebrate the final product!!
Any team member who contributed to the bow would probably voice an objection stating that the amount of sanding has been severely understated. But our efforts certainly paid off!
The bow has already withstood one minor (accidental) collision during testing, so we are confident in its ability to protect Ada as she journeys across the Atlantic. There may be things we would do differently if we ever repeat this process, but designing and constructing Ada’s bow was an incredible learning experience, and ultimately a success. We are looking forward to seeing the bow on the other side of the Atlantic (hopefully unused!), and applying our newly gained knowledge to any future Sailbots.