What Caused Titanic To Sink?
[Author's note: the following is not meant as a scientific test report, but rather as a layman's narrative of events for our experiment]
The test stand used to conduct the experiment was the hydraulically-operated Baldwin 300 kip (1 kip = 1,000 pounds-force) machine in the UW Civil and Environmental Engineering lab, with the Laboratory Virtual Instrumentation Engineering Workbench (LabVIEW) recording the data. It was not our intent to calculate the precise loading numbers of the iceberg collision, because as accurate as our samples were, they were not large enough to substitute for the actual Titanic hull; for example, our steel plating was not wide enough to be supported vertically by frames. When loads were applied, we could only go as high as the unsupported steel plate would allow, so the numbers yielded from our experiment would be far lower than the actual event. What we were looking for instead was proportional reaction to applied loading.
The experiment was held on the UW campus during the first week of November, 2011. As we were discussing the test procedure, the issue of freezing temperatures was raised. Since our replicated plating seam was directly outside Boiler Room #6, we felt that the mixture in the steel of 100-degree heat from the boiler room and the near-freezing temperatures of the water outside would result in roughly the same (or higher) temperature as we were experiencing in the unheated building (which was cold enough for us to see our breath) in which we were conducting the test. We deliberately chose not to freeze our specimens.
[Author's note: since our experiment was made public, almost every critic has accused us of running a faulty test because we didn't freeze the samples. In my opinion, our critics failed to comprehend the effect of the heated boiler room on the inside of the steel plating.]
Our first test was conducted with the "Olympic" specimen. A horizontal load was applied to the plates, as would have happened in a sideways brush against the iceberg. The steel-rivet specimen withstood the load well, with only one rivet audibly popping as the steel plate above (K strake) buckled at 25,000 pounds-force (25 kip). Once the strake buckled, the loading bled off the riveted seam and the usefulness of the test run was essentially over at that point. Even when the rivet (found after the test to be in the uppermost row) failed, we didn't observe any "unzipping" effect that some scientists have offered up as a means by which a seam could have opened up. Upon post-test ultrasonic examination, we found the "popped" rivet was in the uppermost row (the head had remained in place, even when it had separated from the shaft). We didn't observe any "unzipping" effect that some scientists have offered up as a means by which a seam could have opened up. It appeared that when the one rivet failed, the surrounding rivets immediately picked up their respective share of the load. The riveted seam in the "Olympic" specimen seemingly retained its watertight integrity under heavy loading and the uppermost row of rivets appeared to act as a buffer of sorts, protecting from the buckling plate the two rows holding closed the seam itself.
The first test on a "Titanic" specimen was with the same horizontal loading. Again, everything held tight until about 25 kip when the K strake started to buckle. At around 50 kip, the top row of rivets failed (the one holding the spacer bar for the reinforcing strap), but they did not go all at once. One failed, and the load transferred to the remaining two in the row. Then the next one went, and the load transferred to the remaining rivet. Finally, at 60 kip, the last one failed. At that point, we stopped the test. As was the case with the "Olympic" specimen, the rivet heads did not pop dramatically out…they stayed right where they were. It wasn't until we took the specimen out of the machine that one of the failed rivet heads was loose enough to pry out of its hole. The inside of the decapitated rivet head showed signs of both compressive (the rivet shank was squeezed by the changing geometry between the separating plates) and tensile (the rivet was stretched by the separating plates) failure. Post-test ultrasonic examination revealed that the second row of rivets appeared to have stretched slightly, but were still intact, while the third and fourth row of rivets (the ones straddling the seam) appeared undamaged. I did notice visible movement of the spacer during the test and speculate that it worked against the uppermost row of rivets, like I feared. It was apparent that if the same had happened in Titanic, then water could have forced its way through the holes left open by displaced rivet heads and around the spacer to flood the compartment within, especially when driven by the pressure at the depth of the damaged seam (approximately 30 feet, or almost 2 atm hydrostatic pressure). In other words, the actual seam didn't have to open to allow water into the hull.
For the final run, we used our remaining "Titanic" specimen in a vertical orientation, to test the kind of vertical shearing stress that might have been encountered in a grounding event. The specimen at first appeared to stay intact under an extremely heavy load. At around 100 kip, the J and K strakes appeared to twist slightly and two distinct pops were heard. At around 142 kip pounds, the K strake quickly and dramatically bent, bleeding off pressure through strake deformation. At that point, the test was stopped. Post-test ultrasonic examination revealed that the centre rivet in both of the bottom two rows experienced a failure, while the rest of the rivets appeared undamaged. The seam in the specimen looked undamaged, but with failed rivets in the rows straddling the seam (which none of the other specimens – under horizontal loading – had experienced), I could not positively state that the seam would remain watertight at 2 atm pressure.
We were unable to replicate the complete "unzipping" effect of the rivets that has been suggested by numerous experts. In our first "Titanic" run, the uppermost row of rivets did fail, one after the other in a sequence, but the other three rows in our quadruple-riveted seam did not follow suit. Certainly, an "unzipping" can happen under certain conditions – especially in those cases where the stresses far exceed the tensile strength of the rivets – in which case the quality of the rivet becomes less important. There are many riveted seams at the wrecksite (some in hull plating, but mostly in other applications) that show complete separation with all rivets missing. Most of these, though, are single- or double-row configurations. It is difficult to find rivets missing from treble- or quadruple-row configurations, especially where the heavy hull rivets were used. The point is, when multiple rows of rivets are involved, our experiment indicated that they were more prone to share loads among neighbouring rivets than simply failing in one swarm.
Most surprising of all was the fact that the iron rivets, despite the impurities from a century ago and lack of proper peening, performed functionally equal to the steel rivets used in the comparison specimen. Whether steel or iron, the rivets themselves held their strength under loading much longer than did the steel plating.
In my admittedly layman's opinion, our experiment suggested that the iron rivets used in Titanic's hull were not as weak as has been alleged. Certainly, the rivets had flaws and had to fail for water to enter Titanic. I just don't see – based on my study of the history, my observations at the wreck, and our experiment on replica specimens – where anyone has more powerful evidence to support an argument to the effect that Titanic's rivets were a deciding factor in the loss of the ship.
Did the experiment give me evidence to use in support of the grounding theory? At this point, I think that only a finite element analysis (FEA) model – like the one that was commissioned by James Cameron for his 2012 analysis – will provide the insight that we need to make that determination. My conclusion from the vertical sheer test run is that we did not find the easy corroboration for the grounding theory for which I had hoped but at the same time, we found nothing to disprove it, either. It was disconcerting to see the rivets straddling the seam fail because of flexing in the plate caused by the vertical forces. As has been the case since I first adopted the theory, there is much to support consideration of the grounding, but not enough to prove it. Personally, I believe that the ship did ground on the iceberg because there is too much evidence that cannot be explained by the conventional sideswipe theory. The reinforcing strap applied over the J/K landing should have resisted a sideways bump against an iceberg (horizontal load), but it was that very seam that we know failed. I suspect, based on my naval architectural training, that the strap actually created a stress riser – a concentration point, of sorts – in the side plating forming Titanic's bow that served to focus the vertical shearing stresses on that particular seam when Titanic rode up on, and off of, an underwater shelf of the iceberg. But I would need to run that scenario through an FEA model before I could even begin to prove it.
One more perspective to consider before leaving the issue of inferior steel and rivets. I asked Mark Chirnside to compare surveys of Olympic's hull conducted during the 1920s and 1930s to her contemporaries. Without going into detail, Olympic required no more repair work to stop crack propagation and replace loosened rivets than did her British- and German-built equivalents. Going by the lifecycle maintenance record alone, it appears that the rolling and bar stock purchased by Harland & Wolff to build Olympic and Titanic was not substantially inferior than that used by the major shipyards in two countries to build other liners of the period.
That leaves us with the charge of faulty design. As was clearly stated in History's 2012 documentary, Titanic at 100: Mystery Solved, the manner in which the ship flooded and later broke apart tells us that Titanic was indeed a strong ship, a feat made even more remarkable by the fact that she was designed without the rigorous computer modelling and testing that is standard today. For additional detail, please refer to that documentary. The story of the disaster is defined by the fact that Titanic settled slowly on an even keel, instead of capsizing and thereby denying the use of her lifeboats to the souls aboard. Similar ship disasters, including the Costa Concordia in our own time, are more often than not characterised by a capsizing ship that cannot launch its lifeboats in time to be of any use. I believe that I know why Titanic refused to capsize, but that is a story for another day.
By the way, I do believe that Titanic had a faulty design – her watertight door tracks – but that wouldn't play a significant role until Britannic met her fate four years later. I would cite the tendency of the watertight door tracks to warp when a violent shock transmits through the steel hull as my main response to the impractical suggestion that First Officer Murdoch should have steered into the berg in order to hit it square on the bow.
So, if not for brittle steel, weak rivets or faulty design, what – in this layman's opinion – DID cause Titanic to sink? We've known the answer all along: she hit an iceberg. It may just be a matter of simple physics…any steel structure, hit with a greater mass at a certain momentum in just the wrong way, will suffer consequences. In the grand scheme of things, it didn't take a massive gash in the hull to sink the ship; in fact, our experiment showed that a seam need not be opened for water – under pressure – to find its way inside the hull. The spacer bar, installed behind the reinforcing strap as a last-minute change, might have moved enough to increase the stress on rivets and provide a channel for water to exploit, two rivet rows above the actual seam. It could very well be that no actual seams opened and that, as Bill Sauder remarked, Titanic leaked herself to death.
In the meantime, none of what we explored in our experiment addresses what I see as the final Titanic mystery…the as-yet unexplained flooding in Boiler Room #4. I cannot be done with my Titanic research until that gets solved.
All images courtesy of the author unless noted otherwise.