Failure Analysis of Cheonan Sinking

Abstract

The Cheonan, a patrol combat corvette of the South Korean (SK) Navy, sank during a routine patrol mission in the Yellow Sea in March 2010. An SK government report concluded that a North Korean (NK) submarine torpedo attack was responsible for the Cheonan sinking. This report presented a torpedo propulsion section as the conclusive evidence of an NK torpedo attack on the Cheonan. However, a forensic corrosion analysis of this torpedo propulsion section proved it could not have been involved in the Cheonan sinking. Instead, the deformation characteristics of Cheonan’s hull plate and main deck are more consistent with damage due to a collision with a submarine than a torpedo explosion. Evidence that led to this conclusion is presented and discussed.

Introduction

The Cheonan was one of the patrol combat corvettes (PCC) that the Republic of Korea (ROK), commonly known as South Korea (SK), constructed in 1987 for its navy (Fig. 1a). During a routine mission in the Yellow Sea on the night of March 26, 2010, the Cheonan broke up suddenly into two halves and sank. Of a crew of 104 sailors, 40 died and 6 went missing. The two halves of the Cheonan were salvaged out of the seabed of the incident site (Fig. 1b).

Fig. 1

The Cheonan, (a) before and (b) after being sunken and salvaged

To determine the cause of the Cheonan sinking, the SK government formed the Joint Investigation Group (JIG). It consisted of 73 members, including 15 from the US and 3 each from the UK, Australia, and Sweden. On the 50^th day of the incident, JIG presented a torpedo propulsion section (Fig. 2). Allegedly, two fishing boats, the Daepyeong No. 1 and 2 (each 135 tons), found it by dragging a net over the seabed of the incident site. Five days later, JIG claimed it was “a smoking gun” or conclusive evidence that a North Korean (NK) submarine torpedo attack sank the Cheonan.

Fig. 2

The torpedo propulsion section, the very first time it was revealed when a green net cover over it was removed on the Daepyeong, the fishing boat that salvaged it out of the seabed

In September 2010, JIG issued the report, “On the Attack Against ROK Ship Cheonan” [1]. As the report title indicates, it concluded an NK submarine torpedo attack sank the Cheonan. JIG estimated an NK torpedo detonated on the port side, below the keel of the Cheonan, 6–9 m from the sea surface (Fig. 3).

Fig. 3

The torpedo detonation location according to the Joint Investigation Group (JIG)

The purpose of this article is to show why JIG’s conclusion about an NK torpedo attack on the Cheonan may be unfounded as it contradicts the evidence. Instead, this article will show why the fracture characteristics of the Cheonan are consistent with an alternative cause of the Cheonan sinking: a collision with a submarine.

Torpedo Propulsion Section

JIG found white material on the torpedo propulsion section and several locations of the superstructure of the Cheonan. Figure 4a shows a torpedo propulsion section in a display case at the Cheonan Memorial Hall, near Seoul. It has white material mainly on the two propellers and a cover plate (Fig. 4b). JIG stated that this white material was amorphous (i.e., non-crystalline) aluminum oxide, Al₂O₃, based on energy dispersive spectroscopy (EDS) and x-ray diffraction (XRD) analysis data [1].

Fig. 4

(a) A torpedo propulsion section in a display case as the critical evidence of a North Korean (NK) submarine torpedo attack on the Cheonan (b) As viewed from the opposite side of (a). The propeller blades are numbered: F1-F5 and R1-R5

JIG contended:

• (1) The white material came from the aluminum metal powder in the explosive of the NK torpedo that sank the Cheonan.

• (2) The aluminum metal powder in the explosive converted to Al₂O₃ particles, white, during an underwater explosion (UNDEX).

• (3) The Al₂O₃ particles are “adsorbed” by the torpedo propulsion section and the superstructure of the Cheonan. (JIG used the term “adsorbed” interchangeably with “adhered.” The former is, however, a misnomer.)

• (4) The white material was the link between the NK torpedo attack and the Cheonan sinking.

Several scientists in SK and the US pointed out, however, that JIG misidentified the white material as Al₂O₃ by misinterpreting its EDS/XRD data [2]. These scientists believed that the white material was amorphous Al(OH)₃, not Al₂O₃. Chung determined the white material was an aluminum corrosion product, not a detonation byproduct of an explosive [3]. Aluminum forms Al(OH)₃ as corrosion products in water [4].

Even the morphology of the white material on the torpedo propulsion section has no resemblance to an adhered material. For example, no adhered white particles can form a snowflake pattern (Fig. 5a). When enlarged, the snowflakes were filamentous (Fig. 5b), indicative of filiform corrosion of aluminum underneath a coating.

Fig. 5

(a) The white material in a snowflake pattern on the front face of F5 in Fig. 4b. (b) Enlarged view of the white filamentous materials at arrow (b) in (a), indicative of filiform corrosion

Figure 6 is an enlarged view of the area marked by the red arrow in Fig. 4b. The red arrow in Fig. 6 points to the white material, which is underneath the black coating. No white material is on the top surface of the black coating. Therefore, the white material cannot be an adhered material. It has to be the corrosion product of the substrate, aluminum in this case. The black coating in Fig. 6 cracked because the volume of the white material as a corrosion product is much greater than the aluminum metal that has corroded.

Fig. 6

Enlarged view of the white material on the aluminum cover plate at the red arrow in Fig. 4b. The white material formed as a corrosion product underneath the black coating (Color figure online)

Figure 7a shows the white material ⑤ on the aluminum cover plate as it was first found on the salvage ship (Fig. 2). If the white material ⑤ was an adhered material, it should have adhered to the steel gearbox as well as the aluminum cover plate. However, the aluminum cover plate is completely covered with a lumpy white material; yet no white material is on the adjacent steel parts. Only a small amount of the white material encroached upon the edges of the gearbox case (Fig. 7a, yellow arrows). This occurred also because the volume of the white material as a corrosion product is many times greater than the volume of the metal that has corroded.

Fig. 7

(a) Enlarged view of the white material on aluminum in Fig. 2 Cover plate, marked ⑤. (b) Rear propeller, marked ④

The JIG’s claim that the white material was Al₂O₃ particles and it was the link between an NK torpedo and the Cheonan sinking was flawed from the beginning for the following reasons. The white material covered not only the propeller blades but also evenly all around the recessed surfaces of the propeller hub ④ in Fig. 7b. It even caked up on it right up to the steel washer. This would be impossible if the white material had adhered there, but possible only if it was an aluminum corrosion product that precipitated in situ. Therefore, the white material on the torpedo propulsion section is the aluminum corrosion product, Al(OH)₃, not a torpedo detonation byproduct, Al₂O₃, that adhered.

Furthermore, a torpedo UNDEX would cause the seawater to become so turbulent that it would be mechanistically improbable for any Al₂O₃ particles to adhere to any surfaces of the torpedo propulsion section or Cheonan’s superstructure. JIG provided no technical data or references that Al₂O₃ particles as the detonation byproduct of a torpedo UNDEX would adhere to its own propulsion section.

According to JIG, the torpedo propulsion section with a black coating in Fig. 2 was in frigid seawater for 50 days. Then, it should have been practically corrosion-free as the black coating should have protected the substrate from corroding for 50 days or longer. The profuse corrosion products on the aluminum propellers and cover plate (Fig. 7) indicate it was in seawater for much longer than 50 days, perhaps two years or longer. The torpedo propulsion section in Fig. 2 is inconsistent with JIG’s description of its background and could not have been involved in the Cheonan sinking. This judgment is supported by the following observations.

• (1) A loose washer between the two propeller hubs (Fig. 8a).

• (2) A gap between a washer and a locknut for the rear propeller hub (Fig. 8b).

Both above conditions should not have existed in the propulsion section of a torpedo before or after an UNDEX.

Fig. 8

(a) Loose washer, marked ③ in Fig. 2. (b) Gap between the locknut and the washer for the rear propeller hub, marked ④ in Fig. 2

In addition to the above anomalies as the conclusive evidence of an NK submarine torpedo attack on the Cheonan, the SK government showed several different versions of the torpedo propulsion section. For example, the image in Fig. 9a is the torpedo propulsion section on the Daepyeong, the fishing boat that salvaged out of the seabed of the incident site (Fig. 2) [5]. Its shaft had a burn mark at ① (Fig. 9b), which could not be a consequence of an UNDEX. This burn mark is missing in the images of the torpedo propulsion section in the JIG report on the Cheonan sinking (Fig. 9c, d) [1]. The shaft ① and the coupler ② in Fig. 9a and b are completely different from those in Fig. 9c and d. These are also different in the torpedo propulsion section in the display case in Fig. 4. However, the aluminum propellers and the gear box remained the same. JIG did not even acknowledge or explain why the shaft and coupler of the torpedo propulsion section in its report or the one in the display case are different from the original version on the Daepyeong.

Fig. 9

A comparison of torpedo propulsion sections. (a) On the salvage ship (b) Enlarged view of the burn mark ① and the coupler ② in (a). (c) From the JIG report (d) Enlarged view of the shaft ① and the coupler ② in (c)

Lastly, JIG changed the length of the shaft ⑥ between the gearbox and the front propeller. It was 85 mm on the Daepyeong (Figs. 2 and 7a). JIG shortened it to about 20 mm in the display case (Fig. 4b). JIG did not preserve what it called “a smoking gun” of the Cheonan sinking with care. Above all, the profuse corrosion products in Fig. 7, the loose washer between the front and rear propellers in Fig. 8a, and the gap between the washer and the locknut in Fig. 8b indicate that the torpedo propulsion section that JIG presented as “a smoking gun” of an NK’s submarine attack on the Cheonan has nothing to do with the Cheonan sinking.

Main Deck over Empty Turbine Room

The Cheonan was broken up midship where the turbine room was. In Fig. 10, the yellow color indicates the turbine room hull plate that broke away. The blue and red lines in Fig. 10 delineate the two fracture planes, roughly 7.2–7.8 m apart.

Fig. 10

Illustrations of the Cheonan, showing the locations of the fracture planes and the turbine room hull plate (yellow) that broke away

With the turbine room taken out, the diesel engine room in the stern half of the Cheonan began to sink almost immediately. The bow half fell to its starboard, drifted for several hours, and eventually sank. The two halves of the Cheonan, salvaged out of the seabed, are now on display at the Cheonan Memorial Hall (Fig. 11a). The two fracture planes, the bow and stern ends, are shown in Fig. 11b. The stack and demister in Fig. 10 were separated from the deck.

Fig. 11

(a) An overall view of the Cheonan, salvaged and displayed (b) The stern and bow ends of the fracture planes at ② in (a). The deck hangs like a roof over the turbine room that was broken away

Although the turbine room was gone, the main deck remained like a roof over it (Fig. 11b). ② in Figs. 10a and 11b mark the same location where the turbine room hull plate joined the deck. The deck, marked ② and ③ in Fig. 11b, used to be continuous. The deck fractured there, across the openings, and deformed into an inverted V shape. However, very little was lost or blown off the deck itself.

The HMAS Torrens, an Australian Navy destroyer escort, was sunk during a test to validate the combat system of Australian submarines [6]. Figure 12a and b show that a torpedo UNDEX under the Torrens created a water jet column due to a bubble jet effect of an UNDEX [7]. The stern half of the Torrens sank almost immediately (Fig. 12c).

Fig. 12

The HMAS (His Majesty’s Australian Ship) Torrens during an under-keel torpedo explosion test

Figure 13a is by the SK Ministry of Defense [8]. It illustrates the power of the bubble-jet effect of an UNDEX. The bubble-jet (e.g., Fig. 12b) blew off both the superstructure and the deck of the Torrens (Fig. 13b). Torren’s fracture plane has no deck protruding out of it. Then, if Cheonan’s turbine room had been blown away by a torpedo UNDEX, its deck should have also been blown away, just like the Torrens (Fig. 13b).

Fig. 13

(a) Illustration of the bubble jet effect of a torpedo explosion by the SK government (b) The HMAS Torrens after the under-keel torpedo explosion test in Fig. 12, showing the deck blown off

Therefore, the presence of the deck over the empty Cheonan’s turbine room, like a roof (Fig. 11b), is proof that the Cheonan never experienced a torpedo UNDEX. Instead, a collision with a submarine could have taken the Cheonan’s turbine room out, causing it to break up and sink, leaving the deck behind in situ.

Deformation of Turbine Room Hull Plate

As mentioned, the Cheonan’s deck over the turbine room deformed into an inverted V shape, like a roof (Figs. 11b and 14). This may be attributed to an UNDEX, like the Torren’s deck (Fig. 13b). Alternatively, however, the Cheonan’s deck over the turbine room could have deformed into an inverted V because its underside was impacted by the turbine and various objects in the turbine room during a collision with a submarine and its aftermath (Fig. 15).

Fig. 14

The fracture plane of the stern end of the Cheonan. ⑤ and ⑥ mark the same locations as those in Fig. 17

Fig. 15

The underside of the deck of the stern fracture plane in Fig. 14. The arrows point to metal deformation because of impacts by solid objects

The Cheonan’s keel and hull girders were broken at two locations, about 8 m apart, at both the bow and stern ends of the turbine room (Figs. 10a and 11b). This would be unlikely from a torpedo UNDEX because it can break a ship at one location due to hogging and sagging, if at all, not at two locations. In short, an UNDEX may not take out the entire turbine room.

The turbine room hull plate that broke away was salvaged out of the seabed (Fig. 16). The turbine itself was detached from its base (Fig. 16a) and salvaged separately. ① and ② in Figs. 10, 11, and 16a mark the same locations as reference. Figure 17 is an illustration of the turbine room that broke away. ①-④ mark the longitudinal fracture shown in Figs. 10b and 16. The following fracture and deformation characteristics of the turbine room hull plate are consistent with a collision and not with an UNDEX.

• (1) A torpedo UNDEX could not have produced the longitudinal straight fracture, ①-④ for over 7 m, and the transverse fracture ⑦ in Figs. 16, 17 and 18. Instead, a torpedo UNDEX would have produced a rupture in the hull plate rather than straight fractures. According to Lee et al., a close-proximity UNDEX causes target plates to deform or rupture [9]. No target plates formed a single straight fracture, like ①-④, due to an UNDEX.

• (2) A torpedo UNDEX could not have produced the ridge ⑧ in Figs. 16 and 17. An UNDEX can only exert a smooth pressure transition across ⑧, incapable of producing the ridge ⑧.

• (3) The longitudinal T-stiffeners inside the area ⑨ are bent toward starboard. This is also inconsistent with an UNDEX, because it will not produce a force toward the starboard in some localized areas.

• (4) The port hull plate, ①-④ in Figs. 17 and 18a, was bent up almost 90° and pressed tightly against the turbine girder (Fig. 18b). Yet, the adjoining hull plate above it, ⑤-⑥ in Fig. 17, was bent upward only gently, forming a shallow arc. These differences are possible only because of a collision. An UNDEX would have caused rupture at ① in the hull plate or deformed ①-④ and ⑤-⑥ in an equal amount.

• (5) Shock waves from an UNDEX cannot cause the localized metal deformation ⑪ or the cracking ⑫ and ⑬ in Fig. 18b.

• (6) The center-vertical-keel (CVK) and the turbine support girders deformed into an arch (Figs. 18a, 19) and the starboard hull plate, ⑩ in Fig. 17, into a U-shape (Fig. 19). This may have resulted from the uplifting force by a submarine (Fig. 18a) but is improbable from an UNDEX.

Fig. 16

(a) The turbine room hull plate being salvaged out of the seabed (b) The turbine room bottom hull plate in (a), turned 90° counterclockwise. ⑧ marks a ridge and ⑨ where the reinforcing T-bars inside were bent toward the starboard

Fig. 17

Sketch of the turbine room hull plate that broke away. ①-④ mark the same longitudinal fracture as that in Figs. 10b and 16. ⑧ marks the ridge shown in Fig. 16b

Fig. 18

Deformation of the turbine room hull plate. (a) A side view of the turbine room hull plate in Fig. 16b. (b) Enlarged view of the bow end in (a). ⑪, ⑫ and ⑬ mark localized deformation and cracking because of collision rather than shock waves from an explosion

Fig. 19

Turbine room hull plate, starboard, deformed into a U shape. ⑩ marks the same location as that in Fig. 17

Collision as the Cause of the Cheonan Sinking

As discussed above, Cheonan’s metal fracture and deformation characteristics are inconsistent with damage due to an UNDEX. They are, however, consistent with a collision with a solid object. The only solid object that can break up the Cheonan because of a collision in a navigable sea is a submarine.

The Cheonan was sunk about 2 km (1.2 miles) off the coast of Baengnyeong Island, near NK in the Yellow Sea. The SK military maintains several sentry posts there with thermal observation devices (TOD). Figure 20a-d comes from a TOD video that captured the Cheonan sinking [10]. The scenes of the Cheonan before, during, and immediately after the breakup into two halves were not included. The first scene of the Cheonan in this TOD video is Fig. 20a, in which the Cheonan had already been broken apart, most of its stern half was already sunk, and the bow half fell to its starboard. Within one minute, the stern half submerged completely. Some speculate that the SK military released only an edited version, eleven minutes long, from a much longer original file.

Fig. 20

Thermal object device (TOD) images of the sinking of the Cheonan’s stern on March 26, 2010. The white arrows point to an object that moved about. (a), (b) Between the stern and bow (c) Pushed the bow around (d) Moved away against the current flow

Figure 20b shows that a dark object, marked by the white arrow, moved toward the stern first. Then, it moved back toward the fracture plane of the bow half (Fig. 20c), pushed it around clockwise completely against the current flow (Fig. 20d), and departed toward the right.

Figure 21 is an enlarged view of Fig. 20b. The arrows point to what could be a submarine. It had the power to move back and forth and turn the Cheonan’s bow half around against the current flow. Since the Cheonan’s bow was facing the current flow when it fell to its starboard, it could not have turned 180° around without being influenced by an external power, like a submarine.

Fig. 21

Enlarged view of Fig. 20b. The arrows point to a dark image, which could be a submarine

Suppose a submarine was blowing to the surface at night when it happened to collide with the Cheonan’s portside at an angle. The top of the conning tower could have rammed into the Cheonan at ⑦ in Fig. 17. The submarine continued to move at an angle toward Cheonan’s starboard-stern, producing the transverse fracture ⑦ and the longitudinal fracture ①-④ in the Cheonan’s port hull plate.

The turbine support girders and the center-vertical-keel (CVK) formed an arch (Fig. 18a) when the submarine lifted the Cheonan upward in the middle of its turbine room. As the submarine moved upward and toward the Cheonan’s starboard-stern, the entire turbine room could have broken away, leaving the deck behind (Fig. 11b). At the same time, the starboard hull plate ⑩ of the turbine room would have deformed into a U-shape because it was being pushed upward and toward the starboard (Fig. 19). The port hull plate ①-④ would bend up and be pressed against the turbine girder (Fig. 18b) by the conning tower. Thus, the fracture and deformation characteristics of the Cheonan are consistent with a collision with a submarine as much as they are incompatible with an under-keel torpedo explosion.

Kim and Caresta analyzed the spectra of seismic signals recorded at the time of the Cheonan sinking [11]. They concluded it was possible that the Cheonan was sunk because of a collision with a submarine.

Conclusions

• (1) The Cheonan, a South Korean (SK) warship, was sunk while on a routine mission in the Yellow Sea in March 2010. An SK government report concluded that a North Korean (NK) submarine torpedo attack sank the Cheonan. This report presented a torpedo propulsion section that was allegedly salvaged from the seabed of the incident site and insisted that it was the conclusive evidence of the NK’s torpedo attack on the Cheonan. However, a forensic corrosion analysis of this torpedo propulsion section proved it was not involved in the Cheonan sinking.

• (2) Almost an entire turbine room of the Cheonan was broken away. Yet, the deck remained like a roof over an empty space where the turbine room was. If a torpedo underwater explosion (UNDEX) blew away the turbine room, its deck should have been blown away, too. Thus, the presence of the deck over the Cheonan’s turbine room is the evidence that there never was a torpedo attack on the Cheonan.

• (3) The Cheonan was broken up at two places, about 8 m apart, as a result of the turbine room being taken out. This is inconsistent with a torpedo UNDEX because it can break a ship in halves at one place where a maximum bending stress from hogging and sagging exceeds the midship structural strength.

• (4) The metal deformation and fracture characteristics of Cheonan’s turbine room are inconsistent with the damage due to shock waves from an UNDEX. Instead, they are consistent with contact stresses and forces that can develop during a collision with a submarine.