Good to see you back posting!

I don't think the 'real' issue with hot crownsheets was ever "prompt-critical" (as it were) steam generation. It's the softened sheet either failing under pressure (above the waterline but catastrophically reducing pressure) upon which you either get the rocket effect from all the supercritical water "going off" or the shaped-charge water-hammer effect from the nucleate boiling accelerating the water mass, including around the inside shell of the boiler to meet 'at the top'.

As noted, the Eisenhoffer/Leidenfrost effect should protect against prompt quench, too... at least at the rate injector makeup 'adds' water to the volume in the legs. One place I see potential trouble is when the engine has been running downgrade, perhaps with low water compounding crown exposure, then starts up the 'other side' with the throttle opened up and the water surges rearward, perhaps in volume over a hot crown. It might not take long at 'relatively' saturation pressure (the pop volume having to be compared to the collapse of the insulating steam film, which I suspect would be adequate venting mass flow to prevent wild overpressurization).

But I still find the Nicholson syphon issue to be significant with respect to crown overheating -- there's a lavish flow of relatively cold water being somewhat stochastically dribbled across the hot surface, potentially around the 'corner' where the syphons join the crown, for any given small adjacent area of crown 'here on4 moment and gone the next' in what might be rapid alternation. That seems like a formula for induced metallurgical failure.

When the Santa Fe #5011’s (IIRC, the world’s largest two cylinder locomotives) were bult, they were lacking circulators due to wartime shortages, and in operation, continually suffered from leaking crown stays. When shopped, circulators were installed, and the leaking crown stay issue stopped. The implication from this is that the crown sheets on these locomotives were so large that when under heavy load, the crown sheet was evaporating water so fast that the perimeter of the firebox couldn’t supply water fast enough to keep it covered (!). The additional supply of water from the circulators directly to the center of the crown sheet was enough to keep the crown sheets continually covered.

Note that in the old ICC investigations of low water explosions, the crown sheet clearly shows how low the water got prior to the boiler launching. It is virtually always several inches, and sometimes over a foot below the top of the crown sheet. We are talking about several minutes of operation since the crown sheet was uncovered, after several minutes of operation since water was last seen in the glass. I recently read about an explosion where the burned portion of the firebox reached to 16” below the bottom of the glass. Where the hell were the crew’s heads during all that time?

A friend fired #611 during the old NS steam program. He related that it was SOP to rely on the low water alarm on down grade runs, "We would crest a summit, the water would disappear, and sometimes we wouldn't see it for five miles. As long as the low water alarm didn't go off, there was no concern." Of course, the injector would be running the entire time that the water was out of sight.

Crown sheets are traditionally built with the highest point being at the tubesheet, then sloping down toward the backhead. This keeps the entire crown sheet covered when going down normal grades.

On #475, the glasses are at the front of the cab, near the front of the crown sheet, one on each side. Being closer to the longitudinal center of the boiler, they are much less effected by grades, but being roughly seven feet apart side to side, they are greatly effected by superelevation. #90 on the other hand, has the glasses on the backhead, and shows the greatest effect when going up or down grades, but being only about two feet apart show very little difference on superelevated track. Then again, one of #90’s glasses is mounted on a water column, and the other directly on the boiler. The glass on the column habitually reads a lower level due to the water in the column being cooler and therefore heavier than the water in the boiler.

RGS #20’s boiler came to Strasburg with four gauge cocks set in her backhead. The true purpose for having four didn’t come to light until we were setting up the new water gauges to read empty at 3” above the crown. The top gauge cock was approximately even with the top of the glasses, the second gauge cock was approximately even with the middle of the glasses, and the third gauge cock was dead even with the bottom of the glasses at 3” above the crown sheet. The fourth gauge cock? It was even with the highest point on the crown sheet. Apparently, RGS had added it as a final check to see if they were truly in the danger zone in an empty glass situation. Totally illegal, but handy information to have on a bankrupt line with lots of steep grades. Of course, we had no option but to plug that fourth gauge cock hole before returning #20 to CRRM.

The little logging mike #100 at the old Heber Creeper has the same kind of plate, except it was for a 6% grade!

Here is some interesting reading:

https://archive.org/details/cu31924004616839

It is rather lengthy, but the really pertinent bit to this discussion is chapter 11, starting on page 81 through page 117. The basics of the test were to take two new boilers, one with a Jacobs Schupert Firebox, and one with a contemporary radial stayed firebox, set them up in a field much as Matt described, and fire them at a high steaming rate equivalent to a hard working locomotive, then intentionally stop the water feed and attempt to blow them up.

The whole thing is admittedly a sales paper for the Jacobs-Schupert Firebox company, (and the fact you don't see them should tell you something) and their method of firebox construction, which they advertised as being explosion-proof. BUT, Jacobs-Schupert felt strongly enough about their product they enlisted the services of W.F.M. Goss, then the head of the Railroad Engineering Department at the University of Illinois to establish the testing criteria, set up and run the test as even-handed as possible.

The interesting takeaway in light of some of this discussion, is the fact that during the test, the radial stayed boiler, from the moment the water dropped off the crown sheet, took 14 minutes before the crown failed. As Matt noted though, you don't really know how long it's been off the crown sheet since the glass is well above that. On the other hand, if the water winks out of the bottom of the glass, (likely by account of a grade change, or brake application) and it is not accompanied by a loud roaring sound like a blowdown, the water is still in the boiler, and it'll be back. (I only say this due to some folks I've noted of late who appear to have been told that "If the water goes out of the glass, the boiler will explode! Leading them to believe the two things happen nearly simultaneously.)

Kent

I wonder, do any Jacobs-Schupert fireboxes survive?

What I know about steam locomotive repair, operation, and maintenance I can fit in a thimble. That said, I am posting these thoughts so that the experts who have been posting may clarify my thinking.

In my mind, what has been lacking in this discussion is the separation of the concepts of crack nucleation and crack propagation. Addressing the latter briefly for now, the primary driving force for propagation is the pressure differential between the interior of the boiler and the interior of the firebox.

It has already been discussed that the local pressure increase due to hitting an overheated crown with water is likely non-existent or negligible at best. So why the sudden failure or, at least, the legend of the sudden failure? The answer may lie in looking at crack nucleation.

There are many temperature gradients in a firebox and across the crown which get more complex and are put in a non-equilibrium state in the hypothetical under discussion. On a macro scale we have a lateral gradient from one leg to the other, a longitudinal gradient from the tube sheet to the backhead, and the cross-sectional gradient between the interior of the firebox and the interior of the boiler.

Under normal full water conditions with the crown under water, the lateral and longitudinal gradient should be negligible, and the cross-sectional gradient is relatively uniform between interior firebox temperature (about 700 F at high fire as per Mr. Austin) and a bit higher than the boiling temperature of water at the operating boiler pressure.

When the water drops this equilibrium is disturbed. Again, as Mr. Austin noted, the firebox gets to about 1200 F, but perhaps more importantly, what happens to the thermal gradients in the crown sheet material? The backhead's thermal transmission characteristics should remain constant. The thermal transmission of excess crown sheet thermal energy by the side sheets and the flue sheet will vary somewhat on how much water remains in the boiler but is limited by the thickness of the material as we are looking at thermal conduction only, so a lateral and longitudinal thermal gradient builds rapidly. The steam above the crown is an excellent insulator so, besides gaining temperature as Mr. Austin has noted, the cross-sectional thermal gradient flattens out such that the material is a more uniform higher temperature from one side to another.

The system is now out of its designed operating equilibrium. The crown is heated, so it has expanded and is putting a greater pressure on the side sheets, backhead, and flue sheet and so there is a shear at the points where the relatively horizontal crown meets the other vertical sheets. These same points now have a much larger temperature gradient because they are either the same, or in very close proximity, to the gradient resulting from the cooling of the water-cooled side sheets and flue sheets.

We now have operating boiler pressure, a higher temperature crown, thermally induced stresses where the crown meets the vertical sheets, temperature gradients laterally and longitudinally which are more pronounced in a similar location to the thermally induced stresses, and a minimized cross-sectional temperature gradient. So let's throw some water on it and see what happens.

Hitting the overheated crown with water in these conditions is an excellent recipe for crack nucleation at an impurity, a high stress area such as where the crown meets the vertical sheets, or a dislocation, the last of which is more likely the more the boiler has been stressed in normal operation. Rapidly cooling the crown will cause it to contract rapidly, introducing shear stress at the vertical sheet interfaces which is in the opposite direction to the shear of the overheated crown. Rapidly cooling the crown on the boiler side only will cause the boiler side to contract more rapidly than the firebox side, resulting in tensile stress on the boiler side of the crown. These stresses can provide the energy necessary to nucleate a crack and facilitate initial propagation. Once the crack is nucleated and propagating, the regular boiler pressure can easily take care of the rest, especially in an overheated crown where the thermal energy of the material makes the required energy for crack propagation lower.

Thanks for your time.

It would be interesting to hear some input from the crews at Cass, WV.

The grade from Cass up the mountain is anywhere from 4%-8% more or less, with about 7% being the normal. Some portions are steeper, and of note is the track between the switchbacks. This track has a 9% grade and tight curves.

While most of the route the locomotives are front end uphill, the locomotives run up this 9% grade in reverse.

The older engines (Shay #5, Shay #4, Heisler #6) are saturated, and have longer "straight" boilers. #5 and #6 have the longest, straightest boilers, while #4 is a bit smaller in the front end.

The newer engines (Pacific Coast #2, Shay #11, Western Maryland "Big 6") are superheated.

While all of the engines work hard up this grade, the saturated engines are a bit trickier to run. The water must be kept as low as possible, to keep from "flooding" the engine, as they call it, but high enough to keep the crown sheet happy. The superheated engines are more forgiving.

I have ridden in #5 up this grade, and the water is at the bottom of the glass. #5 has the straightest boiler, meaning that all of the water can run to the front. The crews know where their water is and watch it like a hawk, but it is still nerve racking.

I would speculate that with the normal bouncing around of the locomotive in this situation, the water does slosh on and off of the crown. ??

Kelly,

This statement sparked a memory of a conversation I had with one of SRC's crew a short while back about the sight glasses placed on water columns - specifically #90's. He mentioned that due to placement of the glass on the column and how the siphon was connected at the top, the sight glass would routinely fill with condensation thus giving a false reading. In order to gauge an accurate reading, the glass had to be blown down. This is not the only time I've seen or heard of this phenomena. On paper, this seems an impossible scenario but obviously, it is legitimate. Is this more of how the piping to the glasses are arranged or does it center more around sight glasses attached to a water column?

DC

A point that might be remembered is that, while water cooling is subject to Eisenhoffer/Leidenfrost insulation, "steam cooling" might not be. This might explain how protracted exposure of parts of the crown wouldn't lead to prompt overheat failure. Presumably there is some higher temperature at which steam kinetics won't abstract enough heat to keep the waterside from reaching softening temperature...

The Jacobs-Shupert (note sp.) was made out of alternately riveted rings, and these riveted joints gradually worked loose with thermal cycling until leaks were chronic. I have a high suspicion that this produced both accelerated crevice corrosion and solids deposition, making the separation worse over time. There was a truly epic amount of 'plate truss' structure in the water space (in lieu of staybolting) which could only be chemically cleaned...

That raises the question of how rapidly are you cooling it? We're not spraying cold water on it. The water that is being added is going to be pretty hot if you're using an injector, and it will also mix with the rest of the water in the boiler. Even with a feedwater pump, the water gets heated and you're not putting it directly on the crown sheet.

Should we have a mental picture of a bucket of cold water being poured directly on it, or is it more like a slowly rising tide of slightly cooler water gradually moving up the crownsheet?

I don't doubt what you're saying at all. Like you, I'm curious though. I can definitely see condensation forming as the steam on top of the column condenses due to the relatively cool air in the cab. That part's simple enough.

What I don't understand is how it wouldn't simply flow out the bottom pipe until it reached the level equal to the level in the boiler. If there's no restrictions, shouldn't it simply balance? The pressure is equal, and if you add more water to the column via evaporation, shouldn't that force a bit of the water out and back into the boiler?

Again, not doubting what you were told, just trying to understand. Obviously the ability for that level to freely change and be accurate is kind of critical.

Very careful location of the boiler feed 'clack' valves was made to introduce the feedwater far forward in the shell, originally at the side and then up top where it would more readily 'downcome' into the convection-section circulation.
By the time any of that water made its way back to the crown, if would likely be at or close to saturation temperature.

That is not the same thing as pressure relief on the crown when relatively cold feedwater is injected through the steam space. That might cause 'swell' in the legs and some wobbling wash over the corners of the crown, adding to any longitudinal slosh.

It would sure seem like it does, at least for a moment. Having fired on fairly steep grades (2% to 3%) I can confirm the water will slosh around. It won't take too much time firing before you glance at your waterglass, which had maybe 1/3 glass a few moments ago, and find it totally empty. You stare, trying not to panic, and see the water bob back up into sight. Then you realize that the engineer had just taken a heavy set on the train and made it slosh. But it sure gets your attention. Whether the slosh is low enough to expose the crownsheet for a moment or not, I'm not sure. But it could well be.

A common outlet for a dry crown sheet’s expansion is upward, at least as long as it retains enough strength to rise against the steam pressure, mainly in the rounded radial areas. I understand that EBT #12’s crown sheet displays this phenomenon in that during an apparent low water excursion in the dim distant past, portions of the crown sheet stripped itself off the threads of the button head crown bolts, and then restabilized with the button heads some fraction of an inch below the bottom surface of the sheet. I understand that the resulting leaks were then sealed by electric welding under the button heads.

I have never experienced this with #90, and I hope that it isn’t the case now. I would think that this is indicative of a partially clogged lower connection, allowing condensation from the steam space to gradually fill the glass or column, or a slight leak in the steam space, lowering its pressure slightly below that of the boiler, allowing more water to enter the glass, and not be pushed back out. Regardless, that needs to be written up and repaired.

Everything you need to know about firebox expansion(Link)

Insert 1200F for expansion during dry firing conditions. Above 1200F the steel will likely deform before causing forced expansion.

Download Full Book(2MB)

Thanks for responding to that. A partially clogged lower connection was the only thing I could come up with that wouldn't allow the glass to equalize properly. Whether it's condensation or water bobbing in the glass when you blow it out, things should very quickly equalize to the proper level. If they don't, as Kelly says, that needs written up.