Unsolved Engineering Problem: Runaway Anchor Drops

[anorlunda]

I presume that numerous competent engineering organizations have addressed anchor drop brake designs many times in history. Nevertheless, runaway anchor drop accidents happen again and again.

Therefore, I conclude that the general problem remains unsolved. How would our PF engineers address the design problem to make sure that it never happens again, despite human error by the crew, despite malfunctions?

Runaways are dangerous as heck. Watching the videos makes me want to flee for my life.

By the way, it is apparent that the men manning the braking control wheel, have no feedback telling them the actual braking force applied to the chain. Lacking feedback, they turn the control wheel more and more if the chain stops, but the stop may not be caused by braking. However, improved feedback may not be enough to solve the problem.

 

[tech99]

In lifting operations generally I thought the safe procedure was to drive the load both up and down and not use a brake. The use of a neutral gear for the drive seems dangerous, and I also do not like the use of the devils claw for securing the chain, as it involves delicate human beings handling massive chunks of steel. I think electric drive could maybe provide automatic speed restriction by regenerative braking. A second safety device would be desirable, maybe a rotary fluid damper, but the issue is how to dissipate tens of Megajoules.
https://www.marineinsight.com/naval-architecture/anchor-windlass-understanding-design-and-operation/

 

[Rive]

Good challenge

I cannot actually confirm (yet), but I think those videos are not about regular anchor deployments, but about emergency/manual drop tests/drills. So there the point is exactly about the ability to drop the anchor (fast) without electricity or other power sources available.

 

[jack action]

Dropping an anchor seems to be an art. Trying to figure this out on PF by inexperienced people in that domain seems to be either pointless or against PF rules (Legal and Health Implications).

 

[anorlunda]

think electric drive could maybe provide automatic speed restriction by regenerative braking.

That would take a heck of a powerful electric drive. Remember that after the anchor touches bottom, the drive would need to decelerate the entire mass of the ship. Using the ships main engine, stopping at full speed can take up to half a mile. Granted, we are not at full speed, but you are talking about extremely short stopping distances.

I cannot actually confirm (yet), but I think those videos are not about regular anchor deployments, but about emergency/manual drop tests/drills.

I disagree. Regular deployments.

Dropping an anchor seems to be an art. Trying to figure this out on PF by inexperienced people in that domain seems to be either pointless or against PF rules (Legal and Health Implications).

I can't deny that you make good points. But in my defense, I view it as an academic thought exercise unlikely to come anywhere near trying ideas in real life. Tell you what, I'll self report this thread as potentially dangerous, and let the mentors decide.

 

[berkeman]

I'll self report this thread as potentially dangerous, and let the mentors decide.

After a Mentor discussion, we are okay with this thread continuing. It will not encourage folks to go out and try to design these large high-energy systems, and rather is a good discussion of real-world systems that involve high energy and how to try to make them as safe as possible.

That video was scary.

 

[jrmichler]

Those runaway anchor drops had some common problems.

1) The brake operator would slowly release the brake, the chain would accelerate, the brake operator would quickly start tightening the brake, and the chain would continue to accelerate while the operator continued to tighten. This implies some sort of positive feedback and/or backlash in the system. Positive feedback comes from the mass of the chain, which increases chain tension as more chain is let out. Backlash could be in the mechanism between the brake handwheel and the actual brake lining.

2) Smoking brakes happens when the brake thermal capacity is not sufficient for the job. It's easy to specify brake thermal capacity for situations where everything is working properly. Good engineers design the brake for the longest length of chain moving faster than it is supposed to move.

 

[russ_watters]

That would take a heck of a powerful electric drive. Remember that after the anchor touches bottom, the drive would need to decelerate the entire mass of the ship.

I don't understand what you are saying here: why would the anchor brake need to stop the ship? ...and why as soon as it touches bottom?

 

[Lnewqban]

Was not Elisha Graves Otis who found a practical solution (US patent No. 31,128) to a similar problem?

 

[russ_watters]

Therefore, I conclude that the general problem remains unsolved.

There's another possibility: While I think it's worth discussing, I suspect the issue is plenty solved, but the old way continues largely due to tradition.

 

[hutchphd]

Are these brakes subject to the same nasty positive feedback issues with heat (like famous overloaded El Cajon Pass train) ? Seems that cooling is one road of attack...boats do usually have a ready supply of cooling water handy when dropping anchor. Perhaps the cooling water could be driven by the anchor drop (and be auxiliary brake). Wouldn't need much precision.
Never dealt with anchor I couldn't (with effort) pull by hand.

 

[Office_Shredder]

1) The brake operator would slowly release the brake, the chain would accelerate, the brake operator would quickly start tightening the brake, and the chain would continue to accelerate while the operator continued to tighten. This implies some sort of positive feedback and/or backlash in the system. Positive feedback comes from the mass of the chain, which increases chain tension as more chain is let out. Backlash could be in the mechanism between the brake handwheel and the actual brake lining.

Isn't this just gravity? As something falls, it accelerates if you don't slow it down.

 

[erobz]

This may be a silly idea, but how about a parachute?

 

[berkeman]

Aquachutes?

 

[erobz]

Aquachutes?

Whatever floats your boat (or anchor in this case)!

 

[erobz]

Aquachute(s) that trigger at certain threshold accelerations using accelerometers.

 

[anorlunda]

OK let's try to make this good.

I don't understand what you are saying here: why would the anchor brake need to stop the ship? ...and why as soon as it touches bottom?

If the anchor catches on a rock or coral on the bottom and the ship is still in motion (forward or backward), then as soon as the chain becomes taught with ship's momentum versus anchor hold force producing tension. Unfortunately, chain has no elasticity to absorb shock. That shock force can be enough to yank the brake clear off the ship. Slippage in the brake in this case prevents the ship from trying to stop in near zero distance. So max brake force and slippage is a necessary property to allow for this contingency.

The ship should never be at a dead stop when the anchor is dropped, because after the anchor hits bottom and chain keeps coming, it would make a pile. Piles of chain can make knots or make loops around the anchor flukes; both very bad. The total length of chain, let's say 10 times water depth, needs to be played out as the ship moves. Ideally the chain should lie evenly distributed on the bottom in the catenary shape. So the first 1x water depth, the anchor falls by gravity. After that, 9x water depth more chain is pulled out by letting the ship drift with no propeller power. A runaway causes the chain to make a pile on the bottom.

Once full chain is deployed, the catenary curve give the whole system elasticity because chain needs to be lifted from the bottom before it forms a taught straight line. Most of the forces keeping the ship from further movement comes from the heavy chain lying on the bottom, not from the anchor. Vulnerability to sudden shock loads is highest when the anchor first touches down and the chain is nearly vertical.

Brakes:
Yes those brakes are subject to all the bad features as any brake. They could be cooled with water. But the momentum of a runaway chain is so great that an overly effective brake might cause a sudden stop to a runaway. The resulting forces could be so big, the brake housing and everything near it to tear right out of the deck, or conceivably tear off the whole bow of the ship. A runaway must be stopped by gradual deceleration. IMO, it should be much easier to prevent a runaway than to stop one.

The brake operator would slowly release the brake, the chain would accelerate, the brake operator would quickly start tightening the brake, and the chain would continue to accelerate while the operator continued to tighten.

Exactly. But the chain may also stop temporarily with no braking because of the motion of the ship due to wind and currents. Remember, that after the anchor reaches the bottom, more chain is pulled only by moving the ship. My theory is that stoppage fools the operators. They think it stopped because of braking, so they turn the wheel more and more to release the brake. Extra turns just move the brake pads further from the disc. They turn so many turns, that when the chain starts moving again, it takes many seconds for them to turn the control wheel in the other direction to start braking once again. In that interval, a runaway starts. That is very clear in more than one of those cases, that as long as the chain is stopped, the operators keep cranking that control wheel more and more turns. If the officer in charge had a GPS in his hand, he could better judge if the chain stopped because of braking or because the ship stopped moving.

There's another possibility: While I think it's worth discussing, I suspect the issue is plenty solved, but the old way continues largely due to tradition.

That must be at least partially true. The brake control wheel turns a threaded shaft to move the brake pads. That is selected to give sufficient leverage with zero auxiliary power. A hydraulic brake control could be much faster and have zero deadband. I suspect that it is just conservatism and tradition, to not depend on systems like hydraulics or electric that need aux power. After all, one reason to initiate an emergency anchor drop could be loss of onboard power.

Was not Elisha Graves Otis who found a practical solution (US patent No. 31,128) to a similar problem?

After seeing the stuff above, including the catenary, do you still think an elevator is a good analogy?

Good thinking everyone. This problem is not at all simple. Keep thinking. As I said, nobody anywhere has yet to devise a better system that fits all the contingencies.

 

[berkeman]

Quick Question -- why is the chain brake applied via pads instead of a sprocket assembly? There are holes in the chain (albeit 2-dimensional holes with each alternating link in the chain), and the hole diameters are on the same order as the diameter of the link metal material, so a close-fitting sprocket assembly would seem to be a much better way to apply any retarding forces, no? (sorry, land-lubber here...)

 

[Office_Shredder]

That is very clear in more than one of those cases, that as long as the chain is stopped, the operators keep cranking that control wheel more and more turns. If the officer in charge had a GPS in his hand, he could better judge if the chain stopped because of braking or because the ship stopped moving.

I feel like the solution here is automation. A gps attached to the wheel could lock it from opening the brakes if it detects motion in a bad direction. A sensor by the brake pad could detect if there is a physical gap between the chain and the pad. Neither of these sounds particularly complicated compared to things we currently do. I'm sure the solution is a bit more complicated than this, and perhaps runaway anchor just isn't that big a deal that it's worth building a version that actually works well.

 

[anorlunda]

Quick Question -- why is the chain brake applied via pads instead of a sprocket assembly? There are holes in the chain (albeit 2-dimensional holes with each alternating link in the chain), and the hole diameters are on the same order as the diameter of the link metal material, so a close-fitting sprocket assembly would seem to be a much better way to apply any retarding forces, no? (sorry, land-lubber here...)

So the sprocket turns with the chain (like a bicycle right?) but then braking is applied to the sprocket as if it were a brake disc.

I feel like the solution here is automation.

Maybe so. An emergency backup system need not be subject to the same tradition and conservatism as the primary function. If the backup fails, the primary function is still intact. A smart system could use modeling to predict the current location of all links in the chain as well as ship's position and velocities.

 

[berkeman]

So the sprocket turns with the chain (like a bicycle right?) but then braking is applied to the sprocket as if it were a brake disc.

The sprocket would be connected to a shaft of appropriate diameter (on the order of the chain link metal pieces' diameter), and that shaft would be connected to the water cooled electromagnetic brake and/or motor drive. Friction would not be used for braking (and metal dust/fire creation) in this system...

 

[anorlunda]

The sprocket

I forgot to mention. In existing systems, the chain goes through a capstan wheel. A much bigger version of the ones in the images. It plays the same role as a sprocket. The brakes act on the capstan wheel's shaft somehow.

@tech99 also suggested an electromechanical brake. It would have to be very powerful. How about an eddy current brake? It would need the same cooling as a friction brake, but it would be dustless.

Do eddy current brakes share the same positive feedback characteristics as friction brakes? The power of an eddy current brake could be continuously variable via the size of the air gap to the magnets; right? Why don't we see eddy current brakes on trucks, trains and planes?

Tomorrow, I'll try to calculate the required braking power in MW. Tonight, it is my bedtime.

 

[berkeman]

It plays the same role as a sprocket. The brakes act on the capstan wheel's shaft somehow.

Using friction or electromagnetic braking? Actively cooled with water? It doesn't look like it from the video...

 

[russ_watters]

If the anchor catches on a rock or coral on the bottom and the ship is still in motion (forward or backward), then as soon as the chain becomes taught with ship's momentum versus anchor hold force producing tension. Unfortunately, chain has no elasticity to absorb shock.

But in a properly executed, normal anchorage the ship is stationary when the anchor hits the bottom and only then starts backing down as the chain is played-out.

That shock force can be enough to yank the brake clear off the ship.

Right, so you wouldn't be sizing the anchor windlass/capstan motor to handle that scenario either.

 

[russ_watters]

This may be a silly idea, but how about a parachute?

Ahem; "sea anchor".

I think it could work, but the difficulty would be reeling it in if it is permanently installed on the chain.

 

[erobz]

Ahem; "sea anchor".

I think it could work, but the difficulty would be reeling it in if it is permanently installed on the chain.

My thought was in the case of an out of control decent, the retarding force would be large since the velocity of the chain seems quite high in those scenarios taking advantage of the $F_D \propto V^2$. I thought the force of drag hoisting it would be small in comparison to the weight of the chain because I suspect the lifting of anchor is a very slow (low chain velocity process). Also, my intentions were it to function as an emergency backup, not primary braking. So dealing with any additional force in that case would be an economical trade off to losing the entire chain to the depths.

 

[russ_watters]

My thought was in the case of an out of control decent, the retarding force would be large since the velocity of the chain seems quite high in those scenarios taking advantage of the $F_D \propto V^2$. I thought the force of drag hoisting it would be small in comparison to the weight of the chain because I suspect the lifting of anchor is a very slow (low chain velocity process). Also, my intentions were it to function as an emergency backup, not primary braking. So dealing with any additional force in that case would be an economical trade off to losing the entire chain to the depths.

I agree with all of that. To clarify my concern, it's the sea anchor being wrapped around the chain while going through the capstan/windlass. It could get tangled. However, as an emergency-only device, perhaps it could be mounted outside the ship and around the chain for deployment in emergency.

 

[erobz]

I agree with all of that. To clarify my concern, it's the sea anchor being wrapped around the chain while going through the capstan/windlass. It could get tangled. However, as an emergency-only device, perhaps it could be mounted outside the ship and around the chain for deployment in emergency.

I’m ignorant to how the system works. I was thinking you could mount it to the anchor itself, so it would not pass through the capstan. It would automatically deploy if an acceleration threshold was exceeded.

 

[russ_watters]

I’m ignorant to how the system works. I was thinking you could mount it to the anchor itself, so it would not pass through the capstan. It would automatically deploy if an acceleration threshold was exceeded.

That'd be a start, but the runaway typically continues after the anchor hits the bottom. The chain is most of the weight. Maybe deployable sea anchors at regular intervals along the chain...

 

[Rive]

I disagree. Regular deployments.

The first case is supposed to be a 'sea trial' of a new vessel, while the next was indeed an actual anchor drop. I could not identify the third and last ones, but the equipment looked fairly new. The fourth (in order) is not really about the equipment...

Anyway. An additional point about those brakes: since the chain is supposed to be 'laid down' on the seabed, the ship is usually moving (albeit slowly) during a regular anchor drop. If the anchor got stuck after it reaches the bottom then those brakes are dealing with the mass and momentum of the whole ship, and not just with the chain itself => there must be an upper limit for the braking force and must be able to provide some slip above that.
With that, I imagine it would be pretty difficult to bring an ongoing runaway under control.

 

[erobz]

That'd be a start, but the runaway typically continues after the anchor hits the bottom. The chain is most of the weight. Maybe deployable sea anchors at regular intervals along the chain...

But does the runaway start after the anchor hits bottom?

It would be great if you could have them at multiple points on the chain, but then we introduce the tangling possibility you mentioned before.

 

[hutchphd]

On smaller boats the anchor rode contains a relatively small section of chain and the rest is rope. Why is this not true for the big ships? The geometry of the anchor seems scale invariant to me. I would hate to haul up 100 ft of chain on a 25 ft boat.
The big ships seem to have all chain. Strength? Fire resistance?

 

[russ_watters]

Anyway. An additional point about those brakes: since the chain is supposed to be 'laid down' on the seabed, the ship is usually moving (albeit slowly) during a regular anchor drop. If the anchor got stuck after it reaches the bottom then those brakes are dealing with the mass and momentum of the whole ship, and not just with the chain itself...

No, after being set, the chain is simply being unspooled along the bottom, but stationary/with no tension beyond from its weight. There's no time when the ship has significant speed while setting/dropping the anchor, and no significant dragging occurs/is desired. The main problem would simply be dragging the anchor, not damaging the equipment. Also, the anchor can't be set if the angle is steep; it needs to be pulled horizontally. The scenario where an anchor is being dragged at significant speed on a short length of chain and then catches an obstruction would be very unusual.

And even for what it is, I don't see much chance for damage. It would require hard-setting the brake at a time when it has to be softly applied (otherwise the anchor wouldn't get to the bottom).

 

[anorlunda]

Wow! I asked for a good discussion and I'm getting it. More than I can deal with on a timely basis. I'll delay the power calculation to later.

Aquachute(s) that trigger at certain threshold accelerations using accelerometers.

As Russ Waters said, a "sea anchor." I don't think you can make one rugged enough to survive. Runaway can start both before and after the anchor hits bottom.

However, I think you're on the right track. My own thinking is leaning toward a "velocity limiter" as the best preventer of runaways. Something with a force versus velocity curve like this. It could be a backup added to existing system, not a replacement for the brakes.

A "sea anchor" is like that, but it is not very rugged. Perhaps an eddy current brake, or a hydraulic torque converter.

Using friction or electromagnetic braking? Actively cooled with water? It doesn't look like it from the video...

Sorry, I must have written that poorly. Existing systems don't have electrodynamic brakes or water cooling. I meant to say that they could have those things as you suggest.

But in a properly executed, normal anchorage the ship is stationary when the anchor hits the bottom and only then starts backing down as the chain is played-out.

True, but just like any vehicle, the design must be rugged enough to handle foreseeable abuse.

That'd be a start, but the runaway typically continues after the anchor hits the bottom. The chain is most of the weight. Maybe deployable sea anchors at regular intervals along the chain...

Yes, that's right. If the runaway speed exceeds ship speed, the fall becomes vertical, and the weight suspended depends on the depth of the water.

The big ships seem to have all chain. Strength? Fire resistance?

Durability. Coral in particular cuts rope like knives. In the tropics, you'll never see a boat with rope in the ground tackle.

But also holding force in rough weather. Small displacement boats depends on the holding force of the anchor. Heavier boats depend on the holding force of that long length of chain laying on the bottom. Ideally, after deployment the horizontal pulling force on the anchor is zero. That's another mystery of anchoring. To raise the anchor, you need to pull vertically up, not horizontally.

With that, I imagine it would be pretty difficult to bring an ongoing runaway under control.

Exactly. So runaway prevention should be the goal, not runaway halting.

---
By the way, this video illustrates the chain runaway scenario on a micro scale. It doesn't help much in our discussion other than to illustrate how the distributed masses and momenta play roles. Note that inertia of the ascending portion of the chain is also significant.



---
Great thinking everyone. PF has many talented engineers as members.

 

[Rive]

There's no time when the ship has significant speed while setting/dropping the anchor, and no significant dragging occurs/is desired.

May not be desired, but as far as I know this is how a failed anchoring attempt looks like.
That speed might look slow but if the anchor gets stuck somewhere you'll sure have some problem on your deck.