This article analyzes the reasons behind the sinking of the Titanic in the Atlantic Ocean in 1912. The Titanic struck an iceberg in the ocean and sank within hours. Equipped with only 20 lifeboats, the Titanic went down with the loss of 1523 passengers and crew. Recent engineering evidence suggests that the ship experienced a hull failure at the surface and broke into pieces before it went down. The analysis supports some witnesses’ testimony that the ship likely began to fracture at the surface, and that the fracture was completed at some unknown depth below the water’s surface. The resulting stress levels in the strength deck below the root of the second expansion joint (aft), and in the inner bottom structure directly below, were very high because of the unusual flooding occurring in the forward half of the ship. These patterns of stress support the argument that initial hull failure likely occurred at the surface.
When our boat had rowed about half a mile from the vessel, the Titanic—which was illuminated from stem to stern—was perfectly stationary, like some fantastic piece of stage scenery,” recalled Pierre Marechal, a French aviator and a surviving first-class passenger of the ill-fated liner. “Presently, the gigantic ship began to sink by the bows ... suddenly the lights went out, and an immense clamor filled the air. Little by little, the Titanic settled down ... and sank without noise ... In the final spasm the stern of the leviathan stood in the air and then the vessel finally disappeared.”
Elmer Z. Taylor, who watched from Lifeboat No. 5, close enough to the Titanic to observe its final demise, would later write, “The cracking sound, quite audible a quarter of a mile away, was due, in my opinion, to tearing of the ship’s plates apart, or that part of the hull below the expansion joints, thus breaking the back at a point almost midway the length of the ship.”
“At that time the band was playing a ragtime tune,” remembered Harold Sydney Bride, the surviving wireless operator of the Titanic. “I saw a collapsible boat on deck ... I went to help when a big wave swept it off, carrying me with it. The boat was overturned and I was beneath it, but I managed to get clear. I swam with all my might and I suppose I was 150 feet away when the Titanic, with her aft quarter sticking straight up, began to settle.”
One Hundred Years Ago this month, the Titanic struck an iceberg in the Atlantic Ocean and sank within hours. More than 1,500 people lost their lives. There were those who had claimed the ship was unsinkable.
In August 1998, Mechanical Engineering devoted its cover to a feature on the event. The article reported highlights of recent research into exactly how the ship sank. On the centennial of what may be the most famous shipwreck of the 20th century, we reprint that article here.
The paper, “Titanic, The Anatomy of a Disaster, a Report From the Marine Forensic Panel,” on which the article is based is still available through the website of the Society of Naval Architects and Marine Engineers.
“The orchestra belonging to the first cabin assembled on deck as the liner was going down and played ‘Nearer My God to Thee.’ By that time,” as Miss C. Bounnell, first-class survivor, relived the night, “most of the lifeboats were some distance away and only a faint sound of the strains of the hymn could be heard. As we pulled away from the ship, we noticed that she was hog- backed, showing she was already breaking in two.”
Four survivors with firsthand knowledge, remembering probably the most important—certainly the most traumatic—event in their lives, disagreed on one major point, and it has remained a mystery for more than 80 years: Did the Titanic break apart at the surface or sink intact?
Although all the officers testified that the ship sank intact, some survivors and crew testified to a hull failure at the surface. Even during the American and British inquiries into the disaster, few questions focused on the structural aspects of the ship. Despite survivors’ testimonies, it was concluded that the ship sank intact.
Evidence From the Depths
The mystery arose again when the wreck of the Titanic was discovered in 1985 and the hull was found in two pieces. Many theories were developed as to how the ship broke apart during the sinking process, and research was begun to determine how this could have happened. The speculation intensified further when the wreck site was revisited in 1986 and a third 17.4-meter section from the midship region of the ship was found.
To help solve this mystery, the Discovery Channel, in developing its award-winning “Titanic: Anatomy of a Disaster” television documentary, approached Gibbs & Cox, Inc., one of the oldest naval architecture and marine engineering firms in the world. Gibbs & Cox agreed to perform a stress analysis to help determine the possibility of hull fracture at the surface.
With funding provided jointly by the Discovery Channel and the Society of Naval Architects and Marine Engineers, Gibbs & Cox conducted a basic study of the breakup of RMS Titanic using linear finite-element-analysis (FEA) software. This study was done in conjunction with materials testing of the Titanic steel by the University of Missouri-Rolla, with advice from Prof. H.P. Leighly Jr., Timothy Foecke of NIST, and Harold Reemsnyder of the Bethlehem Steel Corp.’s Homer Research Laboratory in Bethlehem, Pa.
Important to the analysis effort was accurate weight and buoyancy data for the ship at the time it struck the iceberg, and then later while it was sinking. These data were provided via a recent study of the ship’s breakup undertaken for another technical paper, “The Titanic and Lusitania, A Final Forensic Analysis,” published in a 1996 issue of Marine Technology. The study provided the loading information needed to take “snapshots” of the ship’s state of stress during the sinking process. Tests conducted on the steel recovered from the wreck site were performed at the University of Missouri and the National Institute of Standards and Technology in Gaithersburg, Md. The results from these metallurgical tests of Titanic steel and rivets were also input as data for the finite element analysis.
Gibbs & Cox engineers selected MSC/NASTRAN, from the MacNeal-Schwendler Corp. in Los Angeles, to perform the analysis. FEMAP engineering-analysis modeling and visualization software from Enterprise Software Products in Exton, Pa., was used to perform the pre- and postprocessing of the analyses. Gibbs & Cox had been using MSC/NASTRAN for approximately five years. According to David Wood, the firm’s structures department manager, MSC worked closely with his team during the development of MSC/NASTRAN Version 70 to provide the special program solutions needed for use in their industry.
Engineers analyzed the stresses in the Titanic as the flooding progressed within the bow region, using modern FEA techniques that simply were not available until the 1960s, and certainly were not known to the structural designers of the ship in the first decades of the century. In the 1960s, engineers started to analyze the stresses in ship hulls using finite-element modeling (FEM). As a pioneer of FEA technology, MSC has been in the forefront of dramatically improving this technique to take advantage of advances in computer technology.
A full-ship model was graphically constructed, employing a modern approach similar to that used for U.S. Navy destroyers and cruisers today. Loadings for the model were developed based on one flooding scenario from the paper, “The Sinking of the Titanic,” by Chris Hackett and John C. Bedford.
The corresponding weight and buoyancy curves, developed by Arthur Sandiford and William H. Garzke, Jr., were used to model the critical flooding conditions believed to represent the hull loading just prior to hull fracture. Since the flooding process took place over several hours, a quasi-static analysis was considered appropriate. The initial modeling effort focused on the determination of the location and magnitude of high-stress regions that developed in the hull while she remained on the surface.
Engineers determined that stress levels in the midsection of the ship were at least up to the yield strength of the steel just prior to sinking. When considered alone, stresses at these levels do not indisputably imply catastrophic failure. Additional analyses, focusing on probable locations of initial hull fracture, are required to indicate that the ship sustained possible catastrophic failure at the surface and began to break apart.
Significant stresses were developed in the vicinity of the two expansion joints, and in the inner bottom of the ship between the forward end of Boiler Room No. 1 and the aft end of the Reciprocating Engine Room. Structural discontinuities, such as expansion joints, result in stress-concentration development. Typically, stress concentration levels are three to four times that of free-field stresses. While these structural discontinuities have not yet been thoroughly investigated, it is believed that stresses developed at these locations were significantly higher than the material yield stress.
The Death of a Ship
At 2:17 a.m., according to the various investigations after the disaster, the Titanic began to go under, her lights blazing in the cold of the sub-Arctic night and with more than 1,500 people still on board. With a rumbling, crashing noise, the bow of the ship sank deeper into the water and the stern rose into the air.
The stern section remained motionless and high out of the water for 30 seconds or more. The hull fracture was described as the sound of breaking chinaware, but as it continued, it was like a loud roar. A minute later, her lights flickered and then went out.
Then, at 2:20 a.m., the stern settled back into the water. Following a series of explosions, the submerged forward section began to pull away from the stern. As the forward section began its long descent, it drew the stern almost vertical again. Once this began, Titanic picked up speed as she sank below the surface of the pond-still waters of the North Atlantic. Some of the survivors on the stern stated that it was almost perpendicular as it slid silently and with hardly a ripple beneath the surface. William Garzke, staff naval architect at Gibbs & Cox, points out that, had the liner been elevated at 90 degrees, the huge boilers would have been ripped from their moorings, which was not the case. He suggests that the stern section likely rose from the surface to at least 20 degrees but not more than 35 degrees, as it filled with water or was dragged down by the bow section.
Chief baker Charlie Joughin, who was at the ensign staff at the stern end, later testified that it was like riding an elevator down to the water. With the absence of suction forces, he was able to swim away without even wetting his hair, so swift was the stern’s demise.
The failure of the main hull girder of the Titanic was the final phase of her sinking process. This began between 2:00 and 2:15 a.m., starting somewhere between stacks Nos. 2 and 4. The FEA results indicate that the plate failures might have started around the second expansion joint, or just behind it.
Stresses in the hull were increasing as the bow flooding continued and the stern rose from the water. Detailed examination of survivor testimony and underwater surveys has confirmed that the forward expansion joint was opened up while the ship was still on the surface, suggesting the significant stresses induced by the flooding of the forward part of the hull. An FEA review of the stresses in this area confirms that the nominal hull stresses were well above the material yield stress.
Most probably, significant stress developed in the way of the second expansion joint, between its root and the deck structure below it. As the flooding progressed aftward, the hull girder was strained beyond its design limitations, and the local stresses around this expansion joint soon reached the ultimate strength of the material. It is thought that, in the end, a critical structural failure in the hull or deck plates occurred in the area around the second expansion joint.
Once localized fracture began in the way of this joint, additional plate failures and associated fracturing likely radiated out from this joint, toward both port and starboard. The decks, however, with their finer grain structure, were most likely able to deform well into the plastic range of the material before failing in ductile tears. It is speculated, however, that the side shell plates suffered brittle fracture due to their coarser grain structure and manganese sulfide inclusions. This type of failure is evident on the wreck today.
Testing the Titanic’s Steel
In 1996, several samples of steel from the Titanic—a hull plate from the bow area and a plate from a major transverse bulkhead—were recovered from the wreck site and subjected to metallurgical testing by H.P. Leighly at the University of Missouri-Rolla, as well as at the laboratories of Bethlehem Steel and the National Institute of Standards and Technology.
Chemical testing revealed a low residual nitrogen and manganese content, and higher levels of sulfur, phosphorus, and oxygen than would be permitted today in mild steel plates or stiffeners. This indicates that the steel was produced by the open-hearth rather than the Bessemer process, most likely in an acid-lined furnace; the steel is of a type known as semi-killed, that is, partially deoxidized before casting into ingots. (Other fragments of the Titanic’s hull have yielded slightly different results, suggesting a degree of variability in the chemical and, hence, the mechanical properties of the steel used in the ship.)
Excess oxygen can form precipitates that can embrittle the steel, and will also raise transition temperatures. In the absence of sufficient manganese, sulfur reacts with the iron to form iron sulfide at the grain boundaries; it can also react with manganese, in either case creating paths of weakness for fractures. Sulfide particles under stress can nucleate microcracks, which further loading will cause to coalesce into larger cracks; in fact, this was found to have been the mode of failure in the shell plating of the Titanic. Phosphorus, even in small amounts, has been found to foster the initiation of fractures. Of course, much of this metallurgical information has only been learned in the years since the Titanic went down.
To determine the steel’s mechanical properties, it was subjected to tensile testing, as well as the Charpy V-notch test, used to simulate rapid loading phenomena; the test used samples oriented both parallel and perpendicular to the original direction of the hull plate. The ductile-brittle transition temperature (using 20 lbs.-ft. for the test) was found to be 20 °C in one direction and 30 °C in the other, compared with –15 °C for a reference sample of modern A 36 steel—and a water temperature of –2°C on the night the ship collided with the iceberg. The Titanic steel was also shown to have approximately one-third the impact strength of modern steel.
When the Titanic samples were also examined with a scanning electron microscope, the grain structure of the steel was found to be very large; this coarse structure made it easier for cracks to propagate. Rivet holes were cold-punched, a method no longer allowed (they must now be drilled), nor were they reamed to remove microcracks.
The steel grain size; the oxygen, sulfur, and phosphorus content of the steel; and the cold-punched, unreamed rivet holes were found to have contributed to the breakup of the Titanic, along with the steel’s relatively low ductility at the freezing point of water. The shell plates showed signs of brittle fracture, though some plates demonstrated significant plasticity.
Of course, the science of metallurgy has advanced considerably since the Titanic’s day, and William Garzke of Gibbs and Cox and his collaborators emphasized in their report that “the steel used in the Titanic was the best available in 1909–1914” when the ship was built. In fact, they add that when 39,000 tons of water entered the bow, “no modern ship, not even a welded one, could have withstood the forces that the Titanic experienced during her breakup.” Henry Baumgartner
Free field stresses, already at the yield point of the material, may have been increased by a factor of two to four in areas of structural discontinuities, such as large openings or those with small radii, or doubler plate edges. Fractures typically spread in random chaotic paths, following weaknesses in the plate and microcracks already present around rivet holes.
Assuming that the hull girder failed at the surface, then as Boiler Room No. 4 filled with water, the stern rose farther out of the water, resulting in some 76 meters of unsupported hull, which sharply increased the hull girder stresses, in turn accelerating the fracturing of the steel plates. The angle of trim grew to a maximum of 15 to 20 degrees, further increasing the stresses in the hull and deck plating near the aft expansion joint. The stresses continued to build in this area of the ship, where there were large openings for a main access, the machinery casing for the Reciprocating Engine Room, the uptakes and intakes for the boilers, the ash pit door on the port side of Boiler Room No. 1, and the turbine engine casing. As the hull girder continued to fail, the bow was first to begin its plunge toward the seabed.
As the bow and stern sections continued to separate, there were some local buckling failures in the inner bottom and bottom structure. This is what caused the stern section to settle back toward the water’s surface as the decks began to fail and the side shell fractured into many small plate sections. The MSC/NASTRAN finite element analysis indicates that the stresses in the region of Boiler Room No. 1 and the Reciprocating Engine Room were elevated.
An additional stress analysis, based on classical beam theory, indicates that the hull girder stresses exceeded the yield point of the steel. When the bow and stern began to separate, the two main transverse bulkheads bounding Boiler Room No. 1 collapsed as they were compressed by the downward movement of the deck structures. The decks, in turn, failed because of the lack of bulkhead support.
When this happened, the unsupported length of the inner bottom suddenly grew to 165 feet, encompassing Boiler Room No. 1 and No. 2, as well as the Reciprocating Engine Room. This condition allowed deformation of the inner bottom structure to extend up farther into the ship’s machinery spaces, while the deck structure failures continued. It is believed that this compression of the hull girder brought about the failure of the side shell plates, and also freed equipment inside the ship, such as the boilers in Boiler Room No. 1, from its foundations.
It cannot be established with any certainty what happened to the ship during its descent to the seabed. However, what is now known is that once the Titanic disappeared below the ocean’s surface, it broke into three pieces. The depth where these events occurred cannot be estimated with any precision. The buoyancy of the stern piece also appears to have resisted the downward pull of the bow. The extent of damage evident in the stern wreck implies that the bow section may have pulled the stern section quickly below the water’s surface, resulting in structural implosions that caused significant damage. Structural failures ultimately led to the separation of the bow portion, followed by the third or double bottom piece. It is interesting to note that the bow section did not suffer damage similar to that in the stern section. This was likely due to the gradual flooding of the bow section, and its stability during the descent to the bottom. It rests upright on the bottom with little apparent damage directly attributable to impact with the seabed.
The analysis supports some witnesses’ testimony that the ship likely began to fracture at the surface, and that the fracture was completed at some unknown depth below the water’s surface. The resulting stress levels in the strength deck below the root of the second expansion joint (aft), and in the inner bottom structure directly below, were very high because of the unusual flooding occurring in the forward half of the ship. These patterns of stress support the argument that initial hull failure likely occurred at the surface. Additional work is being performed to investigate this further.
These findings mirror the testimony of Seaman Edward John Buley at the U.S. Senate hearings. Stating that as the bow continued to slip below the surface, “She went down as far as the after funnel, and then there was a little roar, as though the engines had rushed forward, and she snapped in two, and the bow part went down and the afterpart came up and stayed up five minutes before it went down ... It was horizontal at first, and then went down.”
In response to what he meant by “snapped in two,” and how he knew this, Buley testified, “She parted in two ... Because we could see the afterpart afloat, and there was no forepart to it. I think she must have parted where the bunkers were. She parted at the last, because the afterpart of her settled out of the water horizontally after the other part went down. First of all, you could see her propellers and everything. Her rudder was clear out of the water. You could hear the rush of the machinery, and she parted in two, and the afterpart settled down again, and we thought the afterpart would float altogether. She uprighted herself for about five minutes, and then tipped over and disappeared ... You could see she went in two, because we were quite near to her and could see her quite plainly.”
RMS Titanic, the largest ship of its day, was built in Belfast, Ireland, and was said to be “unsinkable,” a belief so strong that it was to have tragic consequences. Having confidence in the ship’s “unsinkability,” many passengers chose to remain onboard. The first lifeboats to leave were only half, or one-third full.
The fallacy of the claim itself became tragically apparent during the ship’s maiden voyage. Just three hours after it collided with an iceberg, the majestic Titanic vanished beneath the cold waters of the North Atlantic. This ill-founded confidence led to the ignoring of at least 14 warnings of hazardous ice fields, six of which were received on the day of the disaster.
Equipped with only 20 lifeboats, the Titanic went down with the loss of 1,523 passengers and crew. This incredible disaster led to a number of investigations in Great Britain and the United States that resulted in sweeping changes in maritime safety law and ship construction.
The demise of the mighty Titanic was swift, sure, and terrible. Whatever could have gone wrong, did. The engineering marvel that heralded the beginning of the age of technology also displayed, all too clearly, its vulnerability and limits—as well as the need for prudence and safety.
“The analyses, and future analyses we hope to make employing both MSC/NASTRAN and MSC/DYTRAN, help us make critical design decisions about future marine structural features, such as deck openings and expansions joints.” Wood said.
“Today, we’re changing the way we design ships. In the past, nominal load conditions were averaged. Today, we design for the ultimate stress levels and strength,” says Robert Sielski, senior staff engineer at Gibbs & Cox. “MSC/NASTRAN helps us evaluate and design for increased survivability.”
This article is based on a paper, “Titanic, The Anatomy of a Disaster, A Report from the Marine Forensic Panel,” presented at the 1997 annual meeting of the Society of Naval Architects and Marine Engineers, that documents the work of William H. Garzke, Jr. and David Wood, Gibbs & Cox, Inc.; David K. Brown, RCNC; Paul K. Matthias, Polaris Imaging; Roy Cullimore, University of Regina; David Livingstone, Harland & Wolff; H.P. Leighly, Jr., University of Missouri-Rolla; Timothy Foecke, National Institute of Standards and Technology; and Arthur Sandiford, Consultant. Eyewitness accounts are from various sources, including the official transcripts of the 1912 U.S. Senate investigation.