Underwater Chariots
The Nautilus prior to its first run
The Nautilus
Design Pitfalls to avoid Shopping List Construction Begins Six Hours
Test Day Construction Credits Analysis (hindsight) Cutaway Views Alternative Designs

The occasion: An episode of the British engineering TV show called Scrapheap Challenge.  (In the US, it's called Junkyard Wars.)

The Challenge: Build a two person "wet sub" that could propel itself, and two members of the team, maneuvering thru a  complex underwater "show jumping" course.  Explicitly not a race, the subs would be on the course one at a time, and scored based on how many gates each machine could cleanly maneuver through. The win should go to the more agile machine.  The machine would need at least a 200 meter range, and no human power was allowed.

The time limit for construction: 10 hours.

The Team: Jeff (AKA dp), the organizer, team leader, and author of this treatise. Crash (AKA Bill)  chief designer (captain), and Geo, lead scrounge.  Making a guest appearance was Duncan Maycroft, sometime builder of diver assist units.

The place: A corner of a working scrapyard in the Canning Town part of London.

Note: The TV show was very heavily edited, this document contains a lot of detail they didn't have time to show.  They also re-arranged many steps for dramatic effect.  This account was made from notes made shortly after the build day, without having seen the edited show.

The Design

Rendered drawing of The Nautlius

General Theme: A diver assist tug  with electric propulsion.

Our reasoning:  A wet sub is a lot simpler than a dry one, and one that uses the divers' bodies as the control surfaces will be simpler still.  The course is a tight one, so we will also benefit  from the reduced "rigid" length that such a design will offer.  Electric propulsion is fairly simple, especially compared to the semi-diesel or chemically generated steam  engines that some torpedoes use.  (It would be a real pain to provide an air supply to run an internal combustion engine, we had better not be able to find the chemicals needed for chemically heating steam.  As hazardous waste, they need proper disposal, and should not be found in general scrap. Additionally, we would be unlikely to find an engine that could be adapted to run from steam.)

Risks specific to this design: Using the divers' legs for propulsion was going to be penalized. With their legs trailing behind the divers, the temptation to use them would be great (and might even be an automatic reflex).  Since the UK broadcast, some divers have contacted me, and said that some amount of kicking is required to control such machines. You have to kick to restore body position after you are deflected by the prop wash.

Design Issues
(Or: Things Crash was Really Scared Of )

Buoyancy and Trim:

Crash has read enough books by real submariners to know that, unlike a surface ship, a submarine's buoyancy and trim are absolutely critical for operation. Without precise buoyancy control, a sub will go straight to the bottom and stay there forever. Likewise, with improper trim, the sub will point off either nose-high or nose-low, and be uncontrollable. Real subs go through very complex exercises to test the buoyancy and balance at the start of every cruise. This exercise is called "angles and dangles", and it's absolutely critical for the maneuverability and safety of the vessel.

But we weren't building a sub, we were building a diver tug. We got to skip a good part of this.  We still had to be concerned about achieving very near neutral buoyancy, but we got a break with trimming things level.  A real submarine doesn't have a lot of mass that it can shift around. Most of a real sub's mass is bolted or welded into a fixed location. They can't easily shift 20% of the total vessel mass fore and aft to level things. We could.  They have to get it right, and pay careful attention to it. As long as we had left a bit of a lift margin, we could fix it in the tank.

A real sub's control surfaces are fixed; ours would be free to shift where needed. Again, and very much unlike real subs, our vehicle could have turned turtle, without affecting its function.  Our control surfaces (the crew's bodies) were not  firmly attached to the hull. The grab ring ran a full 360 degrees, so orientation didn't make a huge difference.  As long as it didn't twist under power, it would be fine.  Since I don't dive, I have one obvious question to divers: do the commercial diver propulsion gadgets have a this-side-up? Or can you just grab the handles whichever way is most useful?

While we could fix the trim "after the fact", we had to get buoyancy right.  Full sized submarines have ballast tanks, and a means to alter the volume of water that they contain. They even control their vertical position using this system.  Instead, our plan was to use fixed buoyancy, with thrust alone to change vertical position.  With sufficient thrust, we could even cope with small deviations from strict neutral buoyancy.  But we had to be close, and more importantly, buoyancy had to stay the same with changes in depth.  Without a pressurized air tank that could be used to displace some water (we weren't allowed to use the air in the divers' tanks), all floatation would have to be rigid. Things like bags or open bottomed air chambers, which would change volume with a change of depth, wouldn't work.

To get buoyancy right is in theory fairly simple.  All you have to do is get the contraption to weigh the same as the volume of water it displaces.  So we had to build a chamber of known volume, and get its weight to match.  This meant a lot of measurement and calculation.  We had to weigh every part we used, and then measure the volume of the shell we constructed.  One lucky thing: we had the option of adding stuff that floated, we weren't limited to ballast.

We planned to make a very buoyant hull, then add iron to balance out the lift. A ballast rail added to the initial sketch immediately morphed into a pair of rails, for easy handling and transportation. Our sub provided its own transport skid  (and landing gear). In hindsight, we should have fitted upturned tips at the front or even some wheels to the skid, given the time we spent moving on the bottom.

Pressure Hull:

5.5 meters down, at the bottom of the pool, every square foot (30cm) of our sub was subjected to about a ton of force squeezing it inward (ten metric tons per square meter). Whatever we built for a pressure hull had to be able to withstand the weight of a small car on an area the size of a sheet of paper.

One way we could have mitigated this, is by pressurizing the hull.  If we increased the pressure of air inside the hull, it would push outwards, and taken up some or all of the force attempting to crush the hull.  The reduction of crushing depended on the pressure choosen.  In addition, pressure in the hull meant that if there were any leaks, they would be air leaking outward rather than water leaking inward.

Nothing in design is free.  The problem with filling the hull with air is that, at the surface, the equivalent pressure would be trying to blow the hull apart. We thought about it a while and decided that it was easier to build something to resist crushing than to resist being blown apart.

The problem was at the joints. Finding tube that could resist the crush wasn't that hard, and any tube that could handle the crush, would have no trouble containing the pressure.  [The plastic tube we used could hold more than 10 times the pressures we would see.]  While the nose joint took a fairly big impact to install, it was not that hard to pull apart.  The joint worked fine in compression, so we only loaded it in that way.  [But check out alternate designs further down.]


We had a lot of joints in our sub. As it turned out, because we used plastic pipe, there weren't any welded joints at all. (It had a couple of welded-up holes in the front cap. One of them leaked the first time we tried it.)  If water did get into the sub, its effective volume would change, and descent would  be swift.  Despite various comments to the contrary (including the opinions of both teams' experts, and Steph the judge) we knew that any water getting to our electrical system was not going to be a problem.

Another water seal problem area was the prop shaft seal. There was no way to make it perfectly waterproof, without a lot of precision machining that we didn't have time to do. If the motor was mounted inside the hull, we had to make sure that any leak we had was as small as possible.  There was an alternative:  We could have run the motor "wet", outside the pressure hull, eliminating the motor shaft as a path for leaks.

In fact, one design variation that we discussed had the entire electrical system wet.  The pressure vessel would have just been a sealed lifting unit, without any holes that might leak.  It didn't need to be a hollow vessel.  If we had happened on a large enough chunk of closed cell foam, we could have made a very minimalist machine.

Yes, you can get electric motors and batteries wet.  With proper preparation, a lead acid battery will be completely unfazed by a bath, even in salt water.  It's possible to construct a motor that is happy with a salt water bath,  but even an unadapted motor would have held up long enough to complete our fresh water excursions.  Long term operation while immersed can shorten the life of the brushes, but that wasn't something we had to worry about.  Depending on the manufacturers choice of lubricant, bearing life could have been affected.  With water rather than air in the gap between rotor and stator, some power loss was likely.  Cooling however, was much improved.

In fact, most automobiles contain several motors that commonly operate wet.  The motors that operate the electric radiator cooling fans get drenched while driving thru rain.  Automotive alternators also shrug off any water that gets splashed into them.  When I removed the automotive starter that I harvested to be our contingency motor, I poured out the water that had collected while it was in the yard.  When smoke tested, the motor ran just fine.  The designers of such motors don't try to keep them tightly sealed. Effective seals cost too much, and ineffective seals are worse than nothing.  They do a better job at making sure any water that happened to  get past them,  stays inside, where it can rust things.  Instead, they don't bother with an elaborate seal.  They simply use a waterproof lubricant in the bearings, and design in drains, so any water  that gets in, doesn't linger to rust stuff.

If we had used a "wet" electrical system, keeping the batteries happy when submerged wasn't hard. We could have borrowed a trick from deep diving vessels, in particular, the bathyscaph Trieste. Just fill the "air space" at the top of the battery with oil (almost any kind would do, even petrol) and seal the vents with caulk.  A liquid, the oil would not compress, and since oil is lighter than water, it would stay put, at the top of the battery, away from the plates.  It would not have affected the operation of the battery.  Besides preventing the collapse of the air space, the layer of oil would have also keept the tank's water away from the electrolyte.

Once the vents had been sealed, recharging would have been at best, difficult.  If you want to use this system,  you have to have fully charged your batteries before sealing them up.  Since we were unlikely to have been given  enough time to recharge a battery between runs, we would had to either size our batteries to do both runs, or swap them for prepared, fresh ones.

Besides minimizing holes in the hull and sealing them, we could have minimized the consequences of leakage.  All we would have needed to do was eliminate any air or other compressible fluid inside our hull.  Again, like the float of the Trieste, we could have filled the hull with some kind of petroleum, and restricted all openings to the bottom of the machine. But a petroleum filled space, while lighter than water, isn't  that much lighter than water.  Compared to an air filled volume, an oil filled one would need to be much larger to get comparable amounts of lift.  Besides harder to make, the bigger float would have been more of a challenge to fit thru the gates.  Also, if some of the oil did escape, it would contaminate the test tank, something we didn't want to do.

We could have foam filled the open spaces, but the behaviour of liquid inflate-in-place foam is hard to predict.  Crash vetoed the idea because it would have been to easy for it to get into places we didn't want it to. (Like in the motors commutator area, where it would have provided electrical insulation between the commutator and the brushes, which would have kept the motor from running.)

Hydrogen buildup:

Besides having to keep water out of our hull, if we had put the batteries and the motor inside it, we would have another sealing problem -- keeping hydrogen away from the motor. We expected to use an electric motor and lead-acid car batteries for our propulsion system. The problem is that such batteries produce hydrogen. Hydrogen, when mixed with air, is explosive. It could have been set off by a small spark, like the sparks that are constantly produced by the commutator of a running electric motor. So, we had to come up with a way to keep the whole sub from exploding.  It was OK for the sub to look like a torpedo, but if it acted like one, we wouldn't be doing another round.

Motor overheating:

Our likely source for batteries was automotive, so an obvious possible motor was a car starter. While powerful (and some include a gear reduction drive) they aren't meant to run very long.  The limit on run time comes from their inability to get rid of heat.  Run a loaded starter for more than a few minutes, and it will catch fire.  But we had an out:  Water cools a lot better than air. If we arranged for our motor to make direct contact with water, heat would not be a problem.

If we could find one, a continuous duty motor would be best, but we could not be sure that there were any suitable ones in the yard.  A typical place to find such a motor would be an electric forklift, or other industrial machine.

Compiling the shopping list

At this point, we hadn't seen very much of the scrap pile and had no idea what we might find there.  (They are very careful to limit your view of the pile during training and team photo sessions.) We knew there were cars in there, so our drive system was designed around stuff that could be pulled off cars.  Minimum requirements were a lead acid car battery for energy storage, a car starter  for a motor, and a radiator fan for a propellor.   The best we could expect was heavy duty truck batteries, a continuous duty motor, and an actual marine propellor that was a good match to our motor.  We could get wire and other incidentals  from a car (might be a little light) or from various industrial electrical scrap.

We wanted some kind of strong, light tube for the body,  or a large tank of some kind.  We also needed a way to cap the ends of the tube.  The design sketch that Duncan had submitted suggested large metal milk cans.  An oil drum could do, but it was a lot larger than we wanted.

The tube or tank had to have an opening large enough to admit the battery. (I had real doubts that we would be able to cut open a (likely metal) can or tank, put the battery inside, then rejoin it, in a water and pressure tight manner, with the battery still alive. But again, see alternate designs below.)

The critical list:

A hull.  Something that would displace 100  to 200 liters, while weighing at least 65 kg less than the hulls water displacement.  If it was hollow, it had to be strong enough to take the crush load at the bottom of the pool. Bonus points if the batteries could be sealed  inside.

A low voltage DC motor of 1 to 4 horsepower. If we couldn't do better, an automotive starter (preferably the long shaft style, and from a big truck).  Even better would be something rated for continuous operation.  Obvious places to look for that sort were electric forklifts, and portable industrial machinery. Geo was convinced that front loading washing machines also used DC motors (and it took finding several to convince him otherwise).

A propellor.  An easy to find but far from optimum answer, would have been a rigid fan from an automotive cooling system.  Since marine propellors had emerged from the yard in prior challenges, we figured there might be a few to find.

Car batteries, in good enough condition to be recharged.

Other bits on the list, that we expected would be easy to find.

Heavy wire -- #6 or heavier.  (Auto battery cables are usually #6 gauge.)

A heavy duty switch.    While the voltages weren't a shock hazard, the combination of large currents and an inductive load meant that there would be a healthy spark when we opened the switch.  An automotive starter solenoid would have been a good solution to the problem.   We could have even provided multiple control positions.  But a solenoid would have been a  real reliability issue. (All the separate solenoids that we could expect to find would be Lucas products.)  The risk with a light duty switch would be it welding itself in the ON position.

A grab ring.

A propellor guard.

A ballast rack, and some iron for ballast.

The Build - Step by Step

Ok, we had all the details of the course, a basic design, and a shopping list.  We had enough information to refine our design, once we saw what we had to work with.  The scroungers were released into the yard.  We roared off, Geo driving, me in the trailer.  Geo had spotted a fan, that was a candidate prop (but cracks meant we discarded it).  I grabbed some bright yellow, thick walled plastic pipe, and Geo found a"slightly shopworn" air compressor tank, .  (It had a large dent, and its wheels were destroyed.  On the show, they mis-identify it as a propane tank.)  He also found a very large, and very heavy (85 kg) propane tank.  Everything we found made a stop at the scale for a weigh in, before it got into the shop.  After we weighed the trailers contntens, Geo headed back to drop the first round of treasure in the shop.

I continued on foot, looking for the propulsion department.  I spotted an easy to remove starter on an upside-down car, and grabbed the wrenches.  A few minutes later we had a contingency plan motor -- we really would have liked something bigger, and rated for continuos operation.  Next step was a propellor.  The first thing that caught my eye was a molded plastic, multi-blade cooling fan.  It had a large number of fairly short blades, so we could have reduced their number to better match its power requirements with our motors output.  Designed for moving air, its coupling to water might be a bit less than efficient. We would use it only if we didn't find better.

I dropped it off, and went looking for batteries. Geo and I dragged a fair number back to the shop, including a couple of truck batteries.  Most were beyond redemption, but one of the larger car batteries, and two of the truck ones were as far as we could tell, fully charged.  Like all things with the word "lead" in their name they were heavy, the truck batteries weighed 26 kg EACH.

The plastic pipe was, we decided,  the best ticket to a pressure hull.  The compressor tank didn't have enough lift (it was heavy for its volume, with only about 40 kg of lift once you subtracted its weight), and the propane tanks were just too big and heavy.  The big tanks had plenty of lift (roughly a 240-250 liter volume minus an 85 kg mass leaving aprox. 150 kg of lift for batteries and motor), but really didn't fit our needs that well.  The opening in the end was too small to pass batteries thru, which would have meant cutting them open, then welding them shut. Yeah, we could have taken a slice out to make them smaller, but in the end, their general awkwardness got them ruled out.  The compressor tank still proved worthy, Geo cut off one of the end caps, and  it (eventually) got the nod as our nose.

Geo came up with a grab ring, the rim from a moped.  Some more yellow pipe got dragged in.  Crash and Duncan removed the heavy clamping ring that was a captive part of the largest pipe, the one that we intended to use as the hull.

Meanwhile Geo went looking for more motors, targeting the larger trucks for their starters. There was a forklift, but its smaller motor had an integral hydraulic pump, and was thus clearly too heavy.  While trying to find a route thru the tangle of hoses so he could pull the Army truck's starter, Geo made a delightful find, a 2.5 hp 36 volt DC motor.  It would work on 12 or 24 volts, but not produce its full rated power.  It landed on the bench, and passed the smoke test.  We had our motor, and we were still an hour to lunch.  On another run, a proper propellor turned up.  The motor even had a reduction gearbox, but according to the propellor tables, the one we found was a better match at high speed.

Several variations on the plastic tubes were considered.  We wanted to use the one with multiple diameters, as its small end was a good fit for the motors we found.  By itself, it didn't displace enough to lift the batteries we wanted to use. (26 kg each).  We decided on a combination of the biggest tube, and the one with the step down to the narrow bit.

They didn't fit together "as found", but with an improvised "lathe" (the floor vise and the electric plane)  we tapered the end, for an impact fit.  The same sort of tapering was needed on the largest tube,  before the nose cone would fit.  Before Geo's discovery that the tubes could be made to telescope, an alternative nose cap got some effort expended on it.  (Not a total loss, the bit of chipboard that the sink mounted into got used as the battery tray.)

After planeing a taper onto the smaller, stepped tube, it was "encouraged" into the large tube, then about 35 drywall screws served to pin the two tubes from moving against each other.  The flange at the front was also cut back and the seating surface for the nose also received a taper.  I attached the moped rim with a couple of screws, and we hauled the whole assembly out to the scale to get its weight. (35 kilos).

Along about this time, our initial work assignments broke down.  The original plan was to have Crash as speaker-to-expert and chief designer.  Me as a first round scrounge and lead harvester, then I was to settle in and do the heavy construction.  Geo was to be our primary scrounge.  Well Geo got dragged into the design process sometime before lunch, and spent the next several hours calculator in hand.  I wound up doing essentially all the scrounging from then on.  (My particular talent on the pile was harvesting  -- detaching something from the greater whole. Geo's younger eyes are better at spotting stuff, but he is slower when it comes time to detach it.  I have a "no prisoners" style, that has me thinking "where should I cut" rather than "have I found all the bolts".)

Back to propulsion.  While the shaft key on the motor matched the groove in the propellor, the propellor followed marine practice, and was intended to mount on a tapered shaft.  It wasn't a perfect fit on our motors shaft, but after Duncan and I drilled and tapped the end of the shaft for a retaining bolt, Crash did some judicious shimming,  and with a few carefully applied blows from the sledgehammer (also from Crash), the prop would hold well enough for our purposes.

There were some small phenolic blocks, originally used to hold the cooling fan's cover in place.  They kept the motor from fitting in the smallest plastic tube, the place we were considering for its home.  I made a couple of quick taps with the small sledge, and out they popped.  The motor now slid into the tube (until you got to the power studs).

Lunch was called.  Lunch is tools down, camera's still running.  The whiteboard isn't considered an off limits tool, so we set to work refining our design, and answering questions about how the team came to be.  I did make one sneak pass thru the yard -- to snag a set of ordinary bicycle pedals, so others could give frankenbike a try.  (I forgot to throw a set in the suitcase, and only had the pedals that required a cleat on the bottom of the shoe with me)

After lunch I went looking for wire, and switches.  The crane cab looked promising, but it was chained shut.  I had to go back for the recip saw. (It was my first time to use a cordless model.  I WANT ONE!)  I brought the crowbar along just in case.  The saw made quick work of the chain, but the crane was heavy. It had sunk into the soil a bit and the door still wouldn't open.  The crowbar made an impromptu pickaxe and shovel. I dug out the earth that was holding it closed and I was able to slip inside after about 5 minutes work.  The crowbar also proved useful when it came time to remove the clear plastic covers over most of the wiring.

While the switches wouldn't help (they were all low current models, all the real power was switched by solid state relays and high voltage electromagnetic contactors), it proved to be a real treasure trove for wire and electrical fittings.  I removed several lengths of "double ought" (size #00, with insulation it was over 20 mm in diameter) fitted with lugs.  I also grabbed various bolts and joining plates.

It turned out that the crane was "set dressing", and not supposed to be harvested from.  We discovered this the next day, long after the damage was done.  Someone asked where we found the heavy wire. The reply: "the cab of the blue crane".  One of the producers groaned "that wasn't ours".  We pointed out that it would have helped to have mentioned that, we just considered the chain a relic from its prior life.  They should know that the kind of people they recruit are capable of dealing quickly with such things.  If you want us to leave something alone, you have to mark it somehow.

Building the propellor guard was a task originally assigned to Geo and Duncan, on the grounds that they would be motivated to do a good job.  (Our name for it was "Cojones Guard".)  We looked at the perforated metal of the inner drum from a washing machine, but decided it was too short.  The milk jug Duncan had asked for turned up, it was also considered for the guard job, but while long enough, wasn't large enough in diameter. So I cut a bit of wire fence from what might have actually been one of the set security fences, and rolled it around one of the plastic tubes to form it.  A bit of wire to hold it together, and we had a nice lightweight guard.  It was a good diameter match with the moped rim, and thus could be held in place with zip ties.  We later welded in some short metal rods, to make sure it didn't tangle with the propellor.  (In hindsight, I would have grabbed the other moped rim, and fastened it to the far end, along with a bit of screen over the open end.  The divers used the end of the guard as an additional grab ring.)

To attach the nose, 4 metal straps were welded on.  With a good kick from Crash, it was seated on its taper, and the tabs were drilled to take drywall screws to keep it in place.  After the holes were drilled, it was removed, so the batteries could get loaded.

According to our calculations, there was a possibility that the batteries would liberate enough hydrogen, such that separating the motor with its sparking commutator from the battery bank was a good idea. It was also considered good insurance against leaks.  Geo manufactured a bulkhead from plywood, and it was installed at the aft end of the battery compartment, just before the tapered section.  (In hindsight, it would have been better if it had been located in the small diameter tube we mounted the motor into.)  It was secured with screws and silicone caulk.

While Crash and Geo were busy with calculators, the nose and bulkhead, Duncan was working on a rear motor seal.  Geo described the seals used by a friend of his on "sharkey" an electric boat motor with an amusing appearance.  Duncan fabricated one based on this design.  This involved a metal plate, and the rubber of a truck inner tube, various hose clamps, and at one point about 6 hands to hold everything together while the clamps and bolts got applied.

I went on another search for switches.  For all the industrial gear in the heap, there were surprisingly few manual motor switches.  We had ruled out use of Lucas products, so no starter solenoids.  (As we planned to run on 24 volts, a solenoid would have been particularly unhappy.) So I looked for a single pole double throw, 200 amp rated switch, with a center off position.

Not a chance.  Nothing even close.  I grabbed a heavy fuse assembly from the crane, figuring that its contact should be able to take the current.  I pulled the fuse wire, and installed one of our heavy cables in its place.  It was ready to install, but my hand cramps keep me from doing so.  (The tv show takes "artistic license" here, I didn't continue to work on the switch (save some disassembly) after our meeting.  That bit, like our motor "crisis" and the broken welds they claimed the leap left, were just to add drama. There is a fair amount of TV in this reality.)

Status report: 6 hour point

We had the following items:

We needed to do the following things
The team is in a frenzy of construction near the end of the day

Building goes into high gear

Geo and Duncan started carving out the various slots and openings to mount the motor.  We discussed getting power out of the battery compartment, and I suggested feedthrus.  I handed Geo the bolts, and he started boring holes for them to pass thru.  (I had to go find stuff to hold the batteries in place, as well as some battery terminal clamps.)   The motor got bolted and duct taped into position.  Crash and Duncan mounted the propellor guard, then wired the motor itself, and sealed the motor wire opening.  (There wasn't room to use feedthrus on the motor tube.)

After I found the connectors I built the ballast rack.  Crash did the first round of contouring the uprights.  The rear set fit well, the front ones needed some opening up.  With prompting from the tv folk, after others had tried sitting on the tube to shape the uprights, I used a bigger hammer (both feet).   No, my leap didn't crack any welds.  ("artistic license" again)  Some screws, and it's firmly attached.  Our sub had an integral shipping skid.

With the feedthru's ready, I made the wire harness, and Geo connected it to the feedthrus.  Geo built the spacers needed to keep the batteries from shifting around, and I loaded them inside.  We decided that for the first run, we would just use 12 volts, saving the second battery for the second run, and wired the moped rim (part of our switch now) accordingly.

(We didn't know how long we would be running the motor, and we didn't expect to be able to recharge the batteries between runs. Besides having both voltages available provided a great bit of silliness, as the 24 volt terminal got dubbed "turbo" and "secret weapon".)

In the closing minutes, Crash installed the nose with a good kick, some drywall screws and a wrap of duct tape.  A rubber apron keept the the battery terminals from shorting.  Geo fashioned a pair of fins and attached them to the sides.  Crash welded the supporting rods into the propellor guard.  I cut ballast.  Robert called time.  We put our tools down.

We were exhausted.  They opened the door, so we could have our first look at the other teams creation. We got another shock -- the sheer size of the machine, and the unusual propulsion system that the dipsticks employed.  They looked at our torpedo.  We demonstrated our advanced switching system, and showed off the secret weapon.  They hauled us back to the hotel, where we continued the discussion over some beer.

Test Day
A view of the test tank, along with yours truly, and the buisness end of our machine.

The testing took place in the test tank of "DERA" in a town near Portsmouth.  Its hull testing tank is the largest indoor swimming pool in Europe:  60 meters wide, 120 meters long, and 5.5 meters deep.  It has a hull towing machine that can reach an actual 50 km/h.  There are scale model hulls, and a huge rack of propellors lining the edges of the room.

Since the pipe we used was in the other teams colors, the TV crew provided some self adhesive plastic to match our costume.  We apply it, and also improvised a fix for a problem that occurred to Geo during the night: possibly insufficient motor retention.  If the rear seals had really worked, it would have been possible for the motor to be forced to slide into the tube, like a piston in a cylinder.  With a clamp scavenged from the dive shop, we fixed the problem, and applied fresh duct tape.

We finally got to put our machine into the water, and adjust the trim.  We discovered a missing screw (easily fixed) and one of the nose cone plugs had a minor leak (some caulk).  We attached foam and iron to trim it neutrally buoyant, floating level, just below the surface.  They haul it out of the tank, and let the Dipsticks have a go.

While it was out,  I attached the propellor from my bike helmet to the nose (telling Robert that it was there to arm the warhead).  I also put some tie wraps on the reverse wire, making it "prickly", so the wires for the two running directions can be distinguished by touch. (Both teams had reverse.  I don't think we used ours)

The main hull is watertight now.  The holes that worried Steph weren't a problem as long as a fastener was in place.  The motor sealing, was a problem. Where the wires come thru is the trouble spot.  The seal was just silicone rubber, and it hadn't cured fully overnight. The problem was made worse by the designed in flexing of the motor switching wires.  They were stiff, and bending them into contact with the moped rim triggered leaking.  Well, we had removable ballast for just that reason.  We attached some chunks of closed cell foam to the outside of the machine.  They were to provide lift, that would take the place of the lift in the rear section. Since it wouldn't start out with the rear section flooded, we attached what we hoped was a comparable amount of ballast.

The machine wass lowered into the water for our first real run.  Crash and I watched from the control booth.  (Only licensed divers may use the tank.)  Our machine was placed on the launching pad.  Duncan grabs the switch wire, and contact was made. Off we went.  Mostly down.  The rear section took on a fairly small amount of water, and it was enough to upset our trim.  They droped some ballast, and despite that, did a tour of the bottom.

We hauled it out out, and were greeted by a hissing machine.  The hull was under some pressure, from water intrusion.  (No air escaped, so the pressure had to go up.)  We pulled our drain plugs, and let out water from the rear section, and nothing but air from the front compartment.

The test run was actually fairly informative.  We learned quite a bit about the machine from the short trip.  Much of what was wrong could actually be fixed.  Our faith in the idea of a wet motor was vindicated.   The motor demonstrated that it didn't care that it was in a pool of water. The watertight bulkhead actually did its job.  It even resisted the pressure at the bottom.  The motor didn't move. (But with a leak behind it, it hadn't see the forces we feared.)

There are problems, kicking is needed to orient the divers bodies, and while it doesn't add to forward motion to speak of (the divers aren't wearing flippers), it does carry a huge penalty.  Had they scored our first attempt, we would have been left at -100 points.  With only 12 volts, the motor wasn't powerful enough to push past trim problems.

Ah, trimp problems.  We needed more floating stuff.  Duncan scared up a buoyancy compensator from the dive room, and I found some more bits of foam (two in the form of life preservers).  Unfortunately the cans of DIY foam that the tech crew had brought along had all went to filling the dipsticks  collapsing tanks. There was none left for us. (I had been trying to convince Crash that we could foam the rear section, as long as we ran the motor while the stuff cured.)  They made me put the life preservers back, but the BC and chunks of foam were fair game.

The rules have an interesting twist: We can't add any air to the BC once we are on course, as all trim adjustments must be made with stuff attached to the vehicle, not to the diver.  We could have let some air out wile running, but we couldn't add any once on course.  This makes the BC much less than perfect, as it couldn't help but be depth sensitive in the lift department (as presumably our leak is).   We had to make multiple ascents and descents, so we couldn't have trimmed its volume as we need it.

It was clearly time to haul out our secret weapon, 24 volts.  We moved the cable to the second battery terminal.  With 4x the power, minor trim problems can be overridden with thrust (we hope).

The strategy for the points run.  Attach anything that floats, and a couple of heavy objects so the thing would sink. Send the works to the bottom  of the tank when it was dropped in.  Time for a little paitence, I told Duncan to wait, and let the rear compartment fully flood. Only then was he to remove pig iron.  I asked that he lighten it  till it almost floats, and make the precision adjustment with the  bag of air.  While they were still able to adjust the bag, lift it a little to see how it responded to depth changes, a little heavy while sitting on the bottom might be a good thing.  He was to try to leave some ballast so if it got too light near the surface, he could have let out a very tiny amount of air, but still had a way to lighten things should it get too heavy at the bottom.

Despite the pig iron, the divers almost had to stand on the hull to get it to sink. Once it reached about a meter below the surface, it began to sink on its own.  They joined it on its ride to the bottom,  then let it sit a minute or so, and only then started to adjust.  With some air in the BC, and a little ballast left, it was in good trim. It was time to haul it up to the launching platform.

Run two.  We're off.  The power boost was impressive.  The machine moved with real authority.  There was even a hint of cavitation from the propellor.  It was moving faster than the safety divers.  The nose ornament was spinning like mad.  They took the gates in turn, unfortunately racking up significant penalties for reflexive kicking.  (Divers have since told me that the penalties weren't particularly fair.  You apparently need to kick to use such machines).  Some comments over the underwater PA from Crash stills their legs, and with 6 clean gates, less 3 for the kicking penalty, to the other teams total of only 2 gates, with a fault on one of them, we were handed the champagne.

Credits: Who did what

Scrounged items: (used and unused)

Built Items: (used and unused)

Analysis and Alternatives

Surprises: Things we could not have predicted

20:20 Foresight: Things that worked well

20:20 hindsight: Changes to the design we used, assorted mistakes summarized.

A view inside the skin

First, a picture of what got built, with some of the internal details drawn in.  This was taken prior to decoration, and the addition of extra floatation.

The Nautilus -- the sub, with the bits labeled (including some of the internals)

There are two important details that the TV show didn't talk about.  The first is in the power department.  There were two batteries inside the hull.  We had a choice of 12 or 24 volts to our propellor.  On the first run, we used 12, for the second we took out our "secret weapon", and connected to the 24 volt tap.  This gave us FOUR times the power from our motor, and let us "power out" when our trim wasn't perfect.  To use one of Crash's lines "Twinkle Twinkle little star. Power equals I squared R".  Had we had lift to allow a third battery (our motor was rated for 36 volts) we would have had 9 times the power.

The second important detail was an internal, watertight, bulkhead.  Just a plywood disk, it was mounted at the end of big compartment. It had two important functions.  The first: it kept any hydrogen from the batteries well away from the motor brushes. The second (and more important)  it kept water that had gotten past the motor seals, out of the main lift compartment.  We didn't trust the improvised motor seal, and took stepts to limit our lift losses.

Cutaway drawing of The Nautilus, showing batteries and bulkhead.

As it turned out, the motor seal wasn't our big problem, the  wire routing was.  The snug fit of the motor in the tube meant we didn't use feedthru's like we did up front.  If water had gotten into the battery compartment, the foam on the outside would not have come anywhere close to getting things to float

Alternate designs

One close variation of the solution we chose, that might have been less problematic.

Rendering of an alternate design, using both ends of the compressor tank, and the largest pipe

The air compressor tank wasn't big enough, and its metal walls were heavy.  Without an opening, we would have needed to cut it open to load the batteries. With the dent, batteries wouldn't fit very well, and welding it back together would have been tedious.  We would have needed to make an insulated plug, to get power outside the can, and we likely would have been stuck with a single voltage as a result.

Having a metal center section wouldn't do.  But we had the means to replace it with something much more tractable.  Just remove both end caps from the tank, and insert a big chunk of the yellow pipe between them.   To hold them on, I suggest running long bolts between the two caps.  (using tabs cut from the clamp ring)  The motor would have been run wet.  Batteries would live inside the tube, with the same feedthru system to get the power out.

Photomontage of a possible alternate design, using the compressor end caps.

Something very different:

This tug embraces wet electric's completely.  The greenish-yellow blocks are closed cell foam, that combined can lift more than 150 kg.  In the space between the two blocks of foam, live the batteries.  The batteries would have been prepared for this use by filling the airspace with petrol.  The motor, running wet, has been drawn hanging out the back, but depending on the size of batteries, could also get located between the foam blocks.  (Which would make guarding the propellor easier.)

An alternate design that doesn't use a pressure hull

To show scale, the blocks are each 20x40x120 cm long. They are separated by 30 cm, which would allow the propellor we found to just clear the foam, should we locate it within the framework.  Otherwise, the open spaces width should be dictated by the size of the batteries.  The frame drawn is supposed to be made from a mixture of 25 mm angle iron, and 20x40mm steel box section. (Both materials are common in the heap.)  Its also not that critical.  If we didn't find enough foam, and weight became an issue, wood could be substituted for much of the frame, without compromising the final result. (I would make plywood lids for the foam, and connect them with metal links, to a part metal lower frame -- the battery tray, the motor mount, and the front and rear lower crossbars.)

This tug is actually shorter than the Nautilus, and not as tall.  It's a bit wider, but vehicle width wasn't an issue on the course we ran. A wider sub actually makes the divers position a bit easier to manage.  Instead of a grab ring near the motor, I would attach handholds along the sides, starting with the front of the tug.   I would be tempted to fit adjustable dive planes, or perhaps a small motor/propellor (say from an auto radiator) to the nose to make it easier to control the pitch without use of the divers legs to reposition their bodies.

 The NERDS home page