Defect and Failure Analysis of the Bow Visor Structure of the M. V. "Estonia"
by the Laboratory for Materials Technology and Welding
Techniques of the University of the Armed Forces, Hamburg
As a further consequence of the finding that the Atlantic lock would only be exposed to excessive loads if the visor hinges had partly or completely failed, the Laboratory for Materials Technology and Welding Techniques of the University of the Armed Forces, Hamburg was instructed to examine the hinge remains which had meanwhile been cut off the visor at Hangö and transported to the Royal Technical Institute (KTH), Stockholm.
The investigation was subsequently extended to the connecting and supporting structural elements between the visor and hull, i.e. in addition to the visor hinges also the visor actuator mountings on D-deck, the side lock locations between visor and hull, the Atlantic lock location between visor and hull and the visor stempost area were examined. The investigations also included a visual examination of the parts cut off the MARE BALTICUM ex DIANA II. The results of the difficult and long lasting examinations, partly carried out at KTH, Stockholm, shall be summarised as follows:
(1) The visor stempost
The welds between stempost and shell plating of the visor had failed already during normal service by a spectrum of cyclic loading providing tensile stress at the crack initiation points together with long term corrosive action of seawater after opening of the welds. As a consequence the stempost had not only been separated from the shell plating during normal service but the stempost being a vital part in its load-carrying function was in itself found to be split up into four parts by crack development well before the casualty. The drawing on the next page shows the cracks F1 - F4 of which F1 had penetrated the thickness of the stempost to about 40% and blue paint was found inside the crack. It could be verified that these traces of blue paint originated from penetration of the outside paint into the crack opening at the time the crack had penetrated the stempost already to the extent of about 40% of its thickness.
Note: Reference is made to Section 12.5 where the application of ice paint INERTA is explained in January 1994, when the ESTONIA had been in the shipyard for the last time, thus at that time the stempost was already cracked by ca. 40%.
The investigation of fracture F1 - F4 ascertained that all the transverse fracture had started and penetrated from the outside by corrosion assisted fatigue cracking. In addition, fractures F2 and F3 reveal major bending displacements of their flanks, in particular fracture F3 reveals a specific bending angle of about 70° combined with severe bending deformation following the original crack development.
(2) The Atlantic lock
The inspection of the original lug remains recovered from the wreck of the ESTONIA at KTH, Stockholm revealed that the failure of the Atlantic lock took place by rupture of the 3 lugs welded to the forepeak deck of the ferry, viz. at first to their circular welds to the housing/bushing, subsequently in the remaining cross sections by overload shear at about 40° to the mounting plane on forepeak deck, whereby also bending distortion prior to final cracking of the assembly was found.
The visor mating lug was found to be deformed by bending of about 15°-20° to starboard whereby apparently a crack had opened in the weld between the visor bottom plating and a vertical support stiffener.
The bore of this mating lug had an original diameter of 85 mm, however, which was measured to be 91 x 105 mm with an oval deformation towards forward indicating the main force direction to have been aft.
Note: This is a confirmation of the structural analysis of the visor by the FE method - see Enclosure 34.2.430 - according to which the force direction at the Atlantic lock was found to be aft as long as the visor hinges were intact.
Another downward bending movement of about 8°-10° was found which indicated forward tilting of the visor bottom prior to the failure of the lock.
Note: This is only possible if the hinges had failed already.
It was further found that the position and shape of the visor's mating lug did not fully comply with the available drawings, thus changes had been made during the life of the vessel.
Further examinations revealed that about 90% of the length of welds between the lugs on the forepeak deck and the bushing/housing had been pre-cracked by fatigue. The crack propagation together with the corrosion showed a similar picture as found on the visor stempost.
In summary of the above it can be concluded that the welding seams between the lugs on the forepeak deck and the bushing/housing as well as the welding seams between when the visor lug and the visor bottom have been considerable pre-cracked by fatigue.
Deformation indicate a sideways and downwards movement of the visor during final failure of the Atlantic lock.
(3) The side locks
The starboard lug has been torn off the visor bulkhead in such a way that the crack started at the upper edge of the weld between the lug and the visor plating, then propagation in a downward direction, being inclined to starboard by about 20° against the vertical, leaving a steel tongue at an angle of more than 90° to the bulkhead surface before final separation of the lug from the visor plating.
The port lug location indicates a more vertical down tearing movement during crack propagation without leaving a steel tongue before failure. No detailed investigation of the character of the fractures have been performed nor were the welding seam remains examined.
(4) Actuator mountings on B-deck
The starboard actuator was torn off from the supporting B-deck structure, whereby the welds failed first and the vertical stiffeners below the B-deck plating subsequently, leaving strongly deformed steel tongues behind. Inspection of the dismounted port actuator revealed a repair weld in way of the vertical stiffener at a location with possible high stress during the opening procedure. Investigation revealed the fatigue cracked area to cover 50-70% of the total load-carrying area of this weld.
(5) Actuator hinges at the visor arms
At a distance of 1300 mm forward of the visor's main hinges the actuator hinges (fastening) were fitted underneath the visor arms and permitted the actuators' sideways movements up to an angle of 15°. No specific structural failures could be found by visual inspection. Severe scratch marks were found with complete removal of the paint at an angle of 8-10° against the horizontal. At the linear forward edges of the lug plates deeper surface damage had been caused as compared to the curved aft edges. It can be concluded that these scratch marks are a result of the cutting process of the lug plates through the forecastle deck plating. The scratch mark inclination of 8°-10° against the visor arm would coincide with a forward rotation of the visor around the Atlantic lock while the lug plates were cutting through the deck structure and the side locks had failed already.
(6) The visor hinges
The visor's hinge system design consisted of the starboard and port side hinge arms fabricated from structural platings, with two attached lug plates on each side, carrying support bushings with bronze liners in the respective bore holes. The support bushings were welded to the lug plates with circular fillet welds. The visor hinge lugs were mated by deck hinge fittings, represented by lugs welded to the forecastle deck. The mechanical connection was one bolt on each side. The video image below shows the starboard hinge arm shortly be- fore the accident.
The following picture shows the broken port visor lug plates with the frac-tures A1, A2, A5 and A6.
The outer port side shows a vertical to slightly horizontal bending to port side at the fracture at the lower part of the remaining bore. The inner port side lug and both starboard lugs - see picture below - did not exhibit significant bending deformations. The picture below shows the broken starboard visor lug plates with fractures A3, A4, A7, A8:
The failure of the visor hinges occurred by fractures A1-A8 of the visor lug plates together with the fillet welds to the support bushings. The bolts had been ripped out of the hinges, together with the bushings.
The four visor lug plates revealed various amounts of macroscopic deformation and necking at the fracture locations.
The fractures A1-A4 of the upper locations of the remaining bores were indicating considerable less necking as compared to the lower fractures A5-A8. In particular, the upper fractures A3,A4 of the starboard lugs exhibited only very small fracture edge contractions, yielding substantially identical cross sections of the fractures and the plates as taken from the design drawing. Fracture A4, shows delaminations originating from longitudinally to rolling direction orientated inclusions.
The respective upper fractures of the port side lugs A1, A2 showed visible necking with some reduction of local cross sections. the lower fracture of the outer port side lug A5, was considerable necked representing a typical shear failure.
The lower fractures A6 of both the inner port side and the outer starboard lug A8 were heavily deformed by bending and hammering. As a particular feature, the inner starboard upper fracture A3, was accompanied by numerous secondary cracks. Some of them even penetrated the lug plate in radial direction without significant macroscopic deformation.
The visual investigation of the failed fillet welds between the hinge lugs and the support bushings revealed different cracking modes at the upper and lower fractures of the hinge plates. Close to the upper hinge lug plate fractures, A1-A4, the fillet welds had failed mostly by cracks parallel to either the hinge or the bushing surface.
The cracks propagated also very often below the respective base material surfaces. At the lower fracture areas of the hinge plates A5-A8, the fillet welds seemed to have failed more by a shear fracture of the weld metal, revealing thicknesses between 5 and 7 mm. However, there were also hammering effects on the weld crack surfaces in this area which was more subjected to compression than to tensile stress.
One bushing had been recovered from the wreck while the remaining visor hinge lug plates had been removed from the visor for investigation at the Institut för Hållfasthetslära of KTH, Stockholm.
The fractures and cracks of the upper locations A1-A4 of both the starboard and port side hinge lug plates exhibit ratchet marks and fatigue bench marks which demonstrate the start of fatigue cracking from flame cutting marks in the inner bore of the lugs. This applies also to the lower fracture of the inner starboard lug, A8.
For further identification of the cracking mechanism, a section was removed from the inner starboard upper fracture A3 area containing the above mentioned secondary cracks. As from the metallographic investigation, fig. 46, the cracks developed stepwise and showed side branching with corrosion products being typical for corrosion supported cracking.
The investigation of one of the secondary cracks (A3) revealed numerous smaller secondary cracks also starting from the inner asymmetrically applied deep flame cuts. The formation of respective ratchet marks can be clearly demonstrated at the crack initiation points, which show successive striation marks at various locations of the secondary cracks being created by stepwise crack propagation in radial direction of the visor hinges.
The fatigue cracks of the visor hinge plates did however also initiate from the outer surface of the hinge plates. The inner bore holes of the hinge plates were originally drilled by machining, however, the surfaces visible outside of these areas were noted as having been destroyed by inadequate application of flame cutting.
The fracture of the recovered hinge bushing was fitting to the fracture of the inner starboard hinge lug. It exhibits a gap of about 4-6 mm between lug bore and support bushing. This gap is almost entirely filled with corrosion products, presumably magnetite, which is also found in thick layers at a location where both hinge late and fillet weld failure are represented together with some excess weld metal, see arrow on the picture below which may have provided a local notch effect.
In another cross section, also severe corrosion inside the gap is identified and the surface of the hinge bore and the bushing exhibit a considerable misalignment, which, during welding of the left fillet weld, was overcome by a buffering layer. The misalignment corresponds to asymmetric burning marks, for example of the inner starboard hinge.
For closer investigation only a small part of an adhering fillet weld of the support bushing was available. It included both a full crack of the hinge plate material and a secondary crack close to the weld metal and showed also the flame cut surface of the hinge plate bore hole. While the full crack revealed plastic shear at its flanks, the secondary crack had started from the heat affected flame cut surface and propagated close to the heat affected zone of the weld in radial direction.
Based on the evidence provided by the above investigation, the following conclusions can be drawn:
- The fillet welds between the hinge plates and the bushings seem to have partially failed a considerable time before the casualty due to corrosion supported fatigue.
- The partial failure of these fillet welds provided access of seawater to the gap between the hinge plates and the bushings and caused reduction of load carrying capacity of the hinges.
- Both effects facilitated the start and propagation of extensive corrosion fatigue cracking of the hinge plates at another location which occurred also a considerable time before the casualty as demonstrated by the corroded appearance of the fracture of the inner starboard hinge.
- It has to be assumed that the inner starboard hinge may have failed first with only a small macroscopic plastic deformation at the edges of the fracture and without leaving bending of the rest of the hinge plate.
- As the port inner hinge plate was found to be bent to port and plastic necking at the fracture was noted, it has to be assumed that the port hinges failed after the starboard hinges.
The above explained investigation of the various parts of the ESTONIA visor system has provided sufficient evidence that the visor hinges, in particular, the starboard hinge lugs, were severely precracked, before the casualty. There is also evidence that the fillet welds between the visor hinge lugs and the support bushings failed due to corrosion assisted fatigue.
In order to determine the required strength assessments, a Finite Element Calculation of the stresses in the above welds was carried out which considered four different directions of loads on the hinge lugs and established the location of the maximum fracture inducing radial stresses in the fillet welds for each direction and for various applied loads. The thickness of the fillet welds were taken as 7 mm.
For example, the normal visor opening procedure created a vertical down force of maximum 375 kN on each hinge lug based on nominal visor weight and respective leverages and resulted in maximum radial stress in the fillet weld found at the location as marked by the circle in the drawing below together with the stress distribution in radial direction. Accordingly the opening procedure would thus create a local radial weld joint stress of 75 N/mm².
It is thus obvious that, during opening of the visor, the location of maximum radial fillet weld stress is found in the same region where fatigue cracking of those fillet welds initiated. However, also the other assumed directions of loads on the visor hinges would provide maximum radial stress locations. In particular, a similar stress range as for the opening procedure would result from an assumed aftwards force on the hinges. Such an aftwards load could have created the conditions for crack initiation and failure of the hinge plates, following the failure of the fillet.
From the present investigation the following conclusions are drawn:
- The ESTONIA bow visor system contained considerable amounts of fatigue cracking which was created by dynamic loads during normal service. The visor hinges showed the largest amount of pre-damage, together with the mountings of the hydraulic actuators.
- The failure of the bow visor hinges was initiated by failure of the fillet welds which were vital supporting joints between the visor hinge plates and the supporting bushings.
- The failure of the hinge fillet welds at locations of high stress was initiated by corrosion fatigue a considerable time before the casualty.
- The dynamically loaded hinge plates failed by fatigue cracking and subsequent plastic overloading of the remaining weakened cross sections. The fatigue crack initiation at the hinge plates was favourised by corrosion, reduced load-carrying capacity and burning marks.
- In addition to the visor hinges and the actuator mountings, there were various other pre-damaged locations, in particular, the stempost and the bottom lock area, which may have contributed to overloading of the hinge locations.
The complete Report is attached as Enclosure 30.417.