aratere shaft failure investigation final report interisland line

Aratere Shaft Failure Investigation

Final Report

Interisland Line

15 October 2015

Revision: 0

Reference: 239422

Project 239422 File 1 Plus 2 final draft report.docx 15 October 2015 Revision 0

Document control record

Document prepared by:

Aurecon New Zealand Limited

Level 1, 102 Customhouse Quay Wellington 6011

PO Box 1591 Wellington 6140 New Zealand

T F E W

+64 4 472 9589 +64 4 472 9922 [email protected] aurecongroup.com

A person using Aurecon documents or data accepts the risk of:

a) Using the documents or data in electronic form without requesting and checking them for accuracy against the original hard copy version.

b) Using the documents or data for any purpose not agreed to in writing by Aurecon.

Document control

Report title Final Report

Document ID Project number 239422

File path P:\239422\3 Project Delivery\3 Reports\1 Plus 2 final draft report.docx

Client Interisland Line Client contact

Rev

Date Revision details/status Prepared by

Author Verifier Approver

0 15 October 2015 Final Report S Jenkins S Jenkins P Wells S Jenkins

Current revision 0

Approval

Author signature Approver signature

Name Name

Title Title

Project 239422 File 1 Plus 2 final draft report.docx 15 October 2015 Revision 0 Page i

Contents 1 Executive Summary 1

2 Introduction 3

3 Background Information 4

4 Initial Observations 7

5 Investigation in Singapore 14

6 Investigation Plan 17

6.1 Consistency with physical evidence 18

6.2 Detailed analysis 19

6.2.1 External event 19

6.2.2 Power and motor system defect 19

6.3 Shaft system defect 20

6.3.1 Shaft design 20

6.3.2 Shaft material specification 20

6.3.3 Metallurgical defect 20

6.3.4 Alignment 20

6.3.5 Torsional vibration 20

6.3.6 Whirling vibration 21

6.4 Propeller system defect 21

6.4.1 Propeller design 21

6.4.2 Manufacture 22

6.4.3 Propeller fitting 27

6.4.4 Propeller performance 28

7 Discussion and conclusion 31

Appendices Appendix A

Aretere Starboard Propeller Shaft Failure Investigation by Quest Integrity Group

Appendix B

Metallurgical Investigation by Lloyd’s Register Marine

Appendix C

Stbd Propeller Damage Repair Proposal by Stone Marine Shipcare Ltd

Appendix D

Fatigue Crack Growth Analyses of Propeller Shaft by Det Norske Veritas

Appendix E

Failure Analysis and Condition Assessment of Starboard and Port Propeller Shafts by Matcor

Project 239422 File 1 Plus 2 final draft report.docx 15 October 2015 Revision 0 Page ii

Appendix F

Propeller Reports by Recon

Appendix G

Alignment and Measurement Results by Det Norske Veritas

Appendix H

Condition Assessment of Starboard and Port Rudder Shafts by Matcor

Appendix I

Advise on Propeller Shaft Replacement by Lloyd’s Register EMEA

Appendix J

Propeller Improvement Study by Wartsila

Appendix K

Surface Etch Inspection of Propellers by Quest Integrity Group

Appendix L

A Comparison of the Scanned Geometry with Design Geometry by Marin

Appendix M

Analysis of Propeller Forces and Cavitation by Marin

Project 239422 File 1 Plus 2 final draft report.docx 15 October 2015 Revision 0 Page 1

1 Executive Summary The final outcome of the detailed investigations was that the most robust approach to determining the cause of failure of the starboard shaft of the Aratere was to consider the basic and undeniable physical evidence that is available.

The first fact is that the failure was a uni axial fatigue failure, requiring a fluctuating force acting in a constant direction across the shaft. The position of the fracture just inside the propeller hub requires that force to be capable of generating a couple on the propeller itself.

The other factor that needs to be considered is the presence of fretting between the shaft and the propeller hub. Fretting is capable of reducing the ability of steel to resist fatigue failure arising from fluctuating forces, but is not by itself capable of generating a shaft failure. The fluctuating force still has to be present, and still has to be large enough to drive the fracture through the shaft material.

It is generally agreed that a shaft which complies with Class rules is capable of resisting the normal fluctuating forces generated by a propeller as the blades pass through the wake field.

In a well formed propeller, with all four blades performing equally, the fluctuating force will not be uni axial as each blade will generate a similar force differential as it passes through the wake field giving four equally spaced fluctuating forces that, if they were capable of generating a bending failure, would generate more than one single origination point.

The uni axial fatigue fracture on the starboard shaft shows a single originating point close to the C blade.

To determine whether fretting had contributed to the origination of the fatigue failure metallurgists were instructed to examine and compare the port and starboard shafts, and particularly to examine the area of the starboard shaft where the uni axial fatigue crack originated.

The conclusions of that examination were that the fretting damage to the surface of the port shaft was worse than the fretting damage to the surface of the starboard shaft, and that in the area of the origination of the uni axial fatigue crack there was no evidence of fretting damage.

It would be reasonable to conclude that if fretting was the primary root cause of the failure it would have occurred on the port shaft where the fretting damage was worst.

It is therefore necessary to find another factor to explain why the fracture occurred on the starboard shaft and was distinctly uni axial.

The physical evidence clearly shows that there is no cavitation damage to the suction surfaces of the port propeller. The physical evidence clearly shows that there is cavitation damage to the suction surfaces of the starboard propeller and that this cavitation damage is most severe on the C blade of the starboard propeller.

In addition cavitation damage to the paint coating on the starboard rudder was significant and consistent with cavitation streaming from the starboard propeller and collapsing on the rudder surface. The position of this paint damage was consistent with the cavitation being generated by a blade or blades passing through the wake field. There was no such damage present on the port rudder.

Underwater dive surveys showed that this cavitation damage was present from the first underwater dive survey following the installation of the new high efficiency propellers.

This physical evidence proves that there are significant differences between the port and the starboard propeller systems.


The physical form of the two propellers was recorded digitally using high definition laser scanning providing an accuracy of plus or -2 mm. The shape of the port and starboard propellers were then compared digitally using the machined forward face of the propeller hub as a reference plane.

The differences of form between the two propellers were then assessed by the differences shown between the digital models of two propellers when they were aligned rotationally.

If the two propellers were identical all differences should have fallen within the tolerance zone. The comparison showed there were significant and random differences between the two propellers with the differences being most pronounced between the two C blades. By comparison the differences between the two D blades were significantly less and largely within the tolerances zones.

Considering the physical evidence available, and by comparison between the port and starboard propeller and shaft systems we get the following summary of facts:

1. Fretting on the port shaft was worse than fretting on the starboard shaft.

2. The naturally fluctuating forces of the port propeller were not able to initiate or drive a fatigue failure on the port shaft despite the higher level of fretting present.

3. There was no fretting damage present on the surface of the starboard shaft where the fracture originated.

4. There was clear evidence of abnormal performance of the starboard propeller by way of cavitation damage to the suction surfaces of the propeller and paint erosion on the rudder.

5. The fluctuating forces generated by the starboard propeller initiated and drove a uni axial fatigue failure which originated close to the C blade.

6. The surface damage from cavitation was most pronounced on the suction face of the C blade of the starboard propeller.

7. By comparison between the scanned shapes of the port and starboard propellers the C blade of the starboard propeller was most significantly different from other blades on the starboard propeller and from matching blades on the port propeller.

8. It is known that cavitation affects the capability of a propeller blade to generate thrust.

9. One non-performing blade on a propeller would generate a uniaxial force which fluctuated once per rotation in a consistent transverse direction across the shaft.

10. That fluctuating force would generate a couple on the propeller that would act to maximum effect at the plane where the fracture occurred on the starboard shaft.

On the basis of the physical evidence it is reasonable to conclude that a malformed C blade on the starboard propeller was the primary cause of the failure. If this blade had been well formed and the propeller had performed symmetrically the uni axial driving force required to initiate and drive the fracture would not have been present.


2 Introduction On 5 November 2013 on a sailing from Picton to Wellington the starboard shaft of the Aratere failed shortly after the ship left the Tory Channel. Once tests had established that the propeller had been lost the ship proceeded on one shaft to Wellington harbour where it berthed successfully and unloaded and was then shifted alongside.

An incident investigation team was immediately set up and arrangements made to survey any damage and record the fracture face on the tail shaft. The underwater survey revealed no hull damage and provided good high-resolution photographs of the fracture face which was protected from corrosion by the cathodic protection systems on the ship. The fracture face was subsequently protected by a grease filled cap, which was removed once the vessel was docked for repair to allow metallurgical examination.

The starboard propeller was found in 120 m of water approximately two nautical miles from Tory Channel it was recovered on 10 December and returned to Wellington where it was stored at the Transport Accident Investigation Commission (TAIC) warehouse.

The propeller and the stub of the shaft which was still retained in the propeller hub were examined by the investigating team, by Quest Integrity and Lloyds Technical Department metallurgically, and by Stone Marine to assess the feasibility of repair of the superficial damage to the propeller blades.

It was noted at this time on the recovered starboard propeller that there was a small bend at the tip of the C blade, and that the suction faces all blades showed varying degrees of surface cavitation damage, with the C blade showing the most severe damage. A review of recent underwater surveys showed that the bent tip was not present in the 2012 survey, but was noted as present but not requiring any remedial action in the 2013 survey.

A repair plan was made and the Aratere sailed on one shaft to Singapore where the vessel was repaired in dry dock. The repairs undertaken included replacing both tail shafts, realignment of both shafts, replacement of both rudder stocks, and refitting the original propellers as the investigation team considered that the cause of the failure was likely to be related to the high efficiency propellers fitted in 2011 although at that stage no definite root cause had been established.

The opportunity to carry out other work on the vessel while it was in dock was used to perform maintenance to the stabilisers, propulsion machinery, and vehicle decks which was unrelated to the failure of the starboard shaft.

Following repairs the vessel was placed back in service and has operated without problems since.


3 Background Information The outline specification of the Aratere is:

Ship type:

Passenger/ RORO cargo ferry

Service speed:

19.5 knots

Gross tonnes:

17,816

Deadweight:

5,464 tonnes

Number of propellers:

Two, 3.95 m diameter, four blade, fixed pitch inward rotating (Currently fitted)

Total kW:

2 x 5,200 kW at 160 rpm

Length B.P. (m):

183.5 (as modified 2011)

Arrangement of shaft, propeller and rudder. The exposed shaft end is where the starboard propeller was mounted before it was lost.


The port side arrangement is similar. In this picture the new high efficiency propeller is in place and the rudder is inclined towards the camera.

This close up shows the area between the aft end of the stern tube and the hub of the propeller. This photo is the port side, the starboard side was similar.


The fracture in the starboard shaft occurred just inside the propeller hub.

This sectional drawing shows the shaft end and the approximate position of the fracture.


4 Initial Observations The high-resolution underwater photographs showed a distinctive pattern on the fracture face which clearly indicated that the failure was a uni axial fatigue failure. This type of failure is caused by a fluctuating force that increases and decreases stress on one side of the shaft and which generates a fatigue fracture with a single origination point that progresses across the shaft from the side where the force is being applied and results in the final overload failure occurring on the opposite side from the fluctuating force.

This fracture is distinctive and cannot be generated by any other loading pattern.

It was noted that the fracture face was approximately 20 mm inside the propeller hub, and that a certain amount of damage to the propeller hub could clearly be attributed to relative movement between the two halves of the shaft as the failure progressed.


Fretting marks on the shaft stub Fretting marks and failure damage on the propeller hub

When the shaft stub was removed from the propeller hub there were marks on both the shaft and the bore of the propeller which indicated that there may have been fretting occurring at the shaft to propeller hub interface. Fretting is a form of surface damage which occurs when there are a very small relative movements between two surfaces in very close contact. Fretting is known to reduce the ability of steel shafts to resist fatigue loading, but while it can facilitate the initiation of a fatigue crack, the full development of the crack into a fracture still requires a significant fluctuating force capable of driving the fracture through the body of the shaft.1

Detailed measurements were undertaken of both components to eliminate the possibility that the propeller was off centre, and within the tolerances of the measuring equipment it was confirmed that the hub and shaft were both constructed in accordance with the original drawings.

The fracture face on the shaft hub was heavily corroded which limited the amount of information that could be determined from detailed microscopic metallurgical examination, however the examination did establish that there was no discernible metallurgical defect in the shaft stub in the area of the origin of the fatigue crack, and that the steel shaft material, based on hardness measurements, appeared to be within the specification for shaft strength.

1 Mechanics of fretting fatigue crack formation Szolwinski and Farris Wear 198 (1996) 93-107.

Fretting marks

Fretting marks

Damage to hub when propeller broke away from main shaft


Micrograph of the origin on propeller stub

The Stone Marine examination established that the damage and the bent tip were repairable but noted that there was evidence that cavitation had been originating at defects on the leading edge and depressions on the propeller surface. Measurements also showed that there were some significant depressions on the propeller blades which were unexpected. Stone Marine also expressed the opinion that the bent tip was typical of normal operational damage and was unlikely to have any significant effect on propeller performance and did not represent a threat to the integrity of the propulsion system.

Cavitation erosion caused by small indentations and poor edge form to the leading edge.


Depression in forward or bow (suction) surface of blade ‘C’

Examination showed that the cavitation damage on the starboard propeller was most severe on the C blade.

Surface replicas of the surface damage to each blade in order A, B, C, D


C Bow D Bow

Cavitation patterns were observed on the suction (Bow) side of the starboard propeller varying in depth and area. C blade was worst.


The bend on the end of the C blade did not show signs of metal to metal impact and was thought to be service damage.

To record the propeller shape for future analysis the propeller was scanned using laser digital technology. Analysis of these scans showed possibly significant differences in shape between blades, and particularly showed that the C blade, the blade with the most severe cavitation damage, appeared to be the most significantly different blade in terms of propeller form.

The quality assurance documents from the propeller manufacturer recovered and examined and where possible compared to the measurements taken. Analysis of the documents themselves showed some inconsistencies which were noted for further investigation.

The bent tip of the C blade was measured and Quest Integrity carried out an elastic/plastic analysis of the bend to determine the load that the creation of this bend this would place on the shaft. The estimate of the instantaneous stress at 80 MPA1 was not sufficient to fracture the shaft and was considered unlikely to have played a part in the initiation of the fatigue failure.

These inspections are recorded in the reports from Quest Integrity, Lloyds Register, and Stone Marine. (Appendices A, B and C)

At this time the maintenance records of the vessel recovered and reviewed, but no indication of any problems could be found.

The vibration records were also recovered and reviewed although the nearest monitoring point was the aft bearing of the main gearbox which was originally considered to be so remote as to the unlikely to produce any useful information. However the analysis did show that the blade pass frequency (4xRPM) could be determined within the vibration spectra confirming that some useful data was available. Although it was not conclusive, there were no obvious signs of any shaft vibration above 3 Hz. Recordings were not taken below 3 Hz (180 RPM) for reasons relating to the sensitivity and discrimination of the recording apparatus.

1 The quest calculations in Appendix A has been corrected for a dimensional error that underestimated the stresses.


Some rough order calculations were carried out in an attempt to establish stress levels in the shaft at the plane where the fracture occurred but were inconclusive and required a large number of assumptions which the investigation team considered rendered the results of indicative value only.

To ensure that the vessel could sail safely to Singapore on one shaft DNV undertook a fracture analysis calculation which used a combination of material properties and potential defect size to provide a theoretical baseline against which the remaining shaft could be assessed. This indicated that a defect of 40 mm at a stress level of 110 MPA would be unacceptable. 110 MPA was the stress level assessed by Lloyds’ technical experts as being the likely operating stress level on the shaft. (Appendix D)

The port shaft was then tested from the rear face of the shaft with ultrasonics following a validation exercise on a dummy section of shaft with calibrated defects to prove the technique. The ultrasonic tests showed no defects in the port shaft.


5 Investigation in Singapore In the dry dock in Singapore the fracture face on the starboard shaft was uncovered and examined metallurgically by an independent local organisation (Matcor), and by the investigation team and Lloyds’ metallurgist, both of whom had seen the failure face on the shaft stub.

The fracture face on the tail shaft after removal of the protective cap.

It was also noted that there was an unusual pattern of paint removal on the starboard rudder which was consistent with cavitation damage. This paint damage was not present on the port rudder.

Outboard side leading edge to right Inboard side Leading edge to left

(The arrow shows the approximate centreline of the tail shaft and propeller)

Following in place examination the tail shaft was removed and a small section of shaft which included the fracture face was cut off and taken to the Matcor metallurgical laboratory for detailed examination.

The port propeller was examined in place and then removed, and the end of the tail shaft subject to magnetic particle inspection and detailed metallurgical inspection in place to see if there was any sign of distress or incipient failure, and to examine in detail the fretting damage which was also present


under the port propeller hub. The surface of the propeller was closely examined for any evidence of cavitation damage. None was found.

An independent metallurgical laboratory in Singapore Matcor was commissioned to carry out a detailed metallurgical examination of the surface of the starboard tail shaft in the area of the failure, and to conduct a full metallurgical investigation of the origin area of the fracture. They were also commissioned to carry out a detailed examination of the surface of the port tail shaft from the aft stern tube bearing to the aft end of the taper which secured the propeller.

Their report on the work they did is included as Appendix E and reached the following conclusions:

1. There was no metallurgical defect at the origin of the fatigue failure.

2. There was no surface damage from fretting at the origin of the fatigue failure.

3. Fretting damage on the port shaft was more severe than on the starboard shaft.

4. There was no sign of cracking or incipient failure on the port tail shaft.

5. In their opinion the fracture was caused by a significant uniaxial fluctuating bending forces.

The two high efficiency propellers were taken to Recon Propeller Engineering in Singapore where they were measured in accordance with ISO 484 so that the anomalies in the manufacturer’s quality assurance documents could be checked against an independent set of measurements.

This report is included as Appendix F.

While in Singapore the alignment of the shafts was thoroughly checked and the bearings examined for signs of vibration damage. While the alignment was found to be less than satisfactory there was no damage to the bearings which could be attributed to vibration although there was a small area of fatigue failure on both aft stern tube bearings which was consistent with normal loading. There was no wiping of the bearing material, and no unusual wear patterns when assessed against ISO 7416 – 1 2008

Starboard aft bearing Port aft bearing

The full alignment report is included as Appendix G.

Assessment of actual vibration of the shaft system was not possible as the missing starboard propeller meant the original conditions could not be reproduced. However the expert undertaking the alignment was asked for his opinion on the likelihood that vibration was a contributory cause and stated in his view it was unlikely and that stress levels on the shafting that could be attributed to vibration would be less than five MPA. Vibration was monitored during sea trials following the fitting of the original propellers and completion of shaft alignment and no vibration was detected.

Given that bearing position and condition is a significant element in the onset of vibration it was considered unlikely that vibration had been a problem with the original drive configuration.


Late in the repair process it was discovered that the rudder stocks were cracked and that in particular the starboard rudder stock had growing fatigue fractures on the port and starboard side is indicating that some force had been bending the rudder stock from side to side.

Matcor also examined these fractures and determined that there were vestiges of fatigue failure on the fracture surface. This is consistent with the expected loading that fractured the shaft.

This report is included as Appendix H.

The high efficiency propellers were shipped back to New Zealand on the returning Aratere and both propellers were scanned at the same time, in the same conditions, using the same equipment to higher accuracy than the original scan of the starboard propeller to allow digital comparison between the port and starboard propellers and also to provide a digital model for analysis and hydrodynamic modelling.


6 Investigation Plan As the root cause was not immediately obvious an investigation plan was developed to allow for a comprehensive review of all aspects of the starboard propulsion system. The underlying strategy was to eliminate through tests or physical evidence as many potential causes as possible and to narrow down the remaining causes into areas where additional test and investigations could be used to provide further useful information.

A key strategy in the investigation was the comparison of the port and starboard propulsion systems since they were in essence identical when constructed, yet the port system showed no signs of distress or incipient failure even under detailed metallurgical examination during the dry docking in Singapore. From this the investigation team were able to include or eliminate factors by comparison between the two systems. If something was the same on both systems and it had not initiated a failure on the port shaft it was assessed as being unlikely to be a root cause of the failure. If a significant difference existed between the two systems this difference was assessed as requiring further detailed examination as a possible root cause.

To provide some structure the system was analysed and divided into four primary systems based on operational elements of the propulsion system. These were the propeller, the shaft, the power and motor system, and an external event. Observations were accumulated under each heading and potential causes evaluated with a view to confirming or eliminating their possible contribution.

The report will follow this structure dealing with the areas that were able to be eliminated before dealing in detail with the areas where there were contributory factors or root cause.


6.1 Consistency with physical evidence In reviewing each of the potential causes it is important to consider the information provided by the physical evidence.

1. The fracture face was undeniably a uni axial fatigue failure.

2. The forces on the shaft had been sufficient to drive the fracture through the shaft.

3. The unusual paint damage on the starboard rudder required explanation.

4. Surface cavitation damage was present on the starboard propeller but not the port propeller.

5. Surface cavitation was worst on the C blade of the starboard propeller.

6. The fatigue fracture started close to the root of the C blade and progressed away from the C blade through the shaft.

7. The plane of fracture was about 20 mm inside the propeller hub.

These facts require us to find a fluctuating force capable of generating a bending couple on the propeller that will act at the plane of the fracture rather than at the point where the tail shaft emerges from the aft stern tube bearing, the point at which the maximum stresses from bending forces are expected to occur.


6.2 Detailed analysis

6.2.1 External event

There was always a possibility that the fracture had been initiated by some external event, but the nature of a fatigue fracture is that it occurs over time, so the single event does not remove the need for a uniaxial fluctuating force. It was also considered that an external event would leave some physical evidence of impact on the propeller and no evidence apart from the bent tip of propeller blade C, and some incidental impact marks on other propeller blades was found.

The analysis carried out by Quest to determine the force required to bend the tip of the C blade required forces that generated less than 80 MPA in surface stresses, and these were, by themselves, not sufficient to initiate a fracture.

In addition there were no reports in the ship’s log of any significant impact incidents.

6.2.2 Power and motor system defect

The nature of forces in the drive system allow the defects to be considered in three areas. The torque or twisting forces in the shaft that turn the propeller, the thrust in the shaft which pushes the ship through the water, and some instability in the electrically controlled drive motor system.

Torque The first important fact is that the motor and drive system of the ship had not changed specification since the original build so the possibility of an overload in the shaft from the system is remote.

A typical torque failure

In addition a torque failure produces a characteristic fracture which runs at 45° to the main axis of the shaft, and is completely different from the uniaxial fatigue failure observed on the starboard shaft. On this basis a failure related to torque can be positively ruled out.

Thrust The new high efficiency propellers produce 7% more thrust than the original propellers fitted to the ship. This is well within the design safety factors, and if it had been a problem we would expect to see evidence of this on both systems as they are identical. In addition a uniaxial fatigue failure requires tensile stresses while the thrust of the propellers only generates compressive stresses in the shaft. On this basis a failure related to thrust can be positively ruled out.

Drive instability or a power surge There is no record of any drive instability during the entire service life of the ship. In addition should this have been the cause of the failure the nature of the fracture would have been significantly different, either being a characteristic torque failure, or a fatigue failure with multiple points of origin.


Our conclusion is that the failure does not have a root cause in the power or motor system.

6.3 Shaft system defect

6.3.1 Shaft design

As part of the repair the shaft design was reviewed by Lloyds who required no change to the original design. In addition the port shaft, which showed no signs of distress or failure, was identical to the starboard shaft. On that basis a design fault of the shaft can be eliminated as a root cause.

6.3.2 Shaft material specification

The shaft material was tested and met the required specification in the design document.

6.3.3 Metallurgical defect

As noted it is often the case that a small metallurgical defect is found at the origin of a fatigue failure. The origin area of the fracture face was examined by three independent metallurgists all of whom could not find any defect under microscopic examination. It is reasonable to conclude from this that there was no metallurgical defect present.

The possibility was raised by the investigation team that there was an existing crack in the starboard shaft which predated the fitting of the high efficiency propellers. This cannot be completely ruled out as the investigating team were unable to find any record of crack testing carried out on the shaft between the removal of the original propeller and the fitting of the high efficiency propeller. However, the investigation showed the following:

there was no visual indication of a crack recorded at any point

no crack was seen by dockyard or Interisland Line staff present when the propellers were replaced

no cracks were found on the port shaft using magnetic particle testing after the propeller was removed in Singapore

there was no indication of an historical crack found during metallurgical investigations of the fracture surface.

Although this possibility cannot be definitely ruled out the investigating team consider that the presence of a pre-existing crack providing an origination point on the starboard shaft was unlikely.

6.3.4 Alignment

Although the alignment of the shaft was less than ideal at the time of the failure both shafts were in a similar condition and the port shaft did not fail. During the course of the alignment carried out during repairs in Singapore the expert from DNV was asked to assess whether alignment could have had some effect that could have led to the shaft failure. His professional opinion was that it would not.

In addition consideration of the effects of misalignment, and the constraints imposed by the bearings in the stern tube where the shaft is held in by a forward and aft bearing and two intermediate bearings make it extremely unlikely that misalignment could have created a uniaxial force at the location of the fatigue fracture.

6.3.5 Torsional vibration

It is a known phenomenon in shafting systems that under certain circumstances torsional vibration can arise. This is where instead of turning smoothly and only one direction, a small forward and reverse vibration is set up in the shaft creating fluctuating torsional forces.

As discussed previously torsional failures have a distinctive characteristic and are aligned at 45° to the axis of the shaft. The uniaxial nature of the fatigue failure rules out torsional vibration as a root cause.


6.3.6 Whirling vibration

Whirling vibration is where rotational forces acting on a shafting system initiate a vibration pattern similar to skipping rope. This type of vibration is a function of the stiffness of the shaft and the condition of the bearings and can usually be detected by examining the wear pattern of the bearing lining materials. There was no evidence of whirling seen in the bearings of either shaft.

Whirling vibration would create symmetric forces on the shaft which would result in at least two fracture origination points which is not consistent with the evidence of the fracture surface.

6.4 Propeller system defect Considering the previous analysis, and also the fact that clearly a significant force was required to fracture a 352 mm (13.8 inch) diameter shaft, the propeller system was likely to have some influence in the failure process. Not only were many of the other potential causes ruled out, but the propeller is a large mechanical element generating forces capable of pushing the ship through the water and if there was any problem in the propeller system it has the potential to generate effects that could have significant consequences.

To assist in the analysis of the propeller system this was divided into four sub areas; the design of the propeller, the manufacturing process of the propeller, the fitting of the propeller, and the performance of the propeller in service.

6.4.1 Propeller design

The main information on the design of the new high efficiency propellers is contained in report titled Wartsila Propeller Improvements which is Appendix J to this report. This report presents to the Interisland Line the Wartsila recommendations for a change in propeller design which will improve fuel efficiency as well as providing some additional thrust that will assist in keeping to timetables.

Interisland Line accepted the recommendation of Wartsila and commissioned them to design and build two new propellers which would be fitted to the existing tail shafts. They were to be fitted at the same time as the ship was lengthened.

The new propellers designed by Wartsila were significantly lighter than the original propellers and analysis showed there was a possibility that the shaft/propeller combination may vibrate in service. Two options were proposed to eliminate this potential issue, one being modifying the bearings in the tail shaft, a major and difficult job. The other was extending the rear bonnet of the propeller and adding weight behind the main propeller to recreate the original propeller system characteristics which had operated successfully for many years. This change was assessed as having a minimal effect on the stresses in the shaft. Interisland Line chose the option to extend the propeller bonnet and add the extra weight.

The new propellers had a slightly higher power density in kilowatts per square metre of blade area than ships of similar design and service. The Wartsila propeller improvement report notes that the new high efficiency propellers would be slightly closer to cavitating in service a diagram included in the report shows that as designed the propellers were within accepted service parameters.


We understand that the propellers were designed using digital techniques which calculated the geometry of the propeller to a high level of accuracy – much less than 1 mm. They were specified to be built to the ISO 484/1, the standard for propellers of diameter greater than 2.5 m which has a base tolerance band of plus 2mm minus 1.5 mm. The effect of the wide tolerance relative to the design accuracy is not known.

6.4.2 Manufacture

The primary record of the manufacture the propeller was the quality assurance document provided by Wartsila and reviewed and signed off by DNV. There were a number of issues in this document which raised concerns about the design and certification process. The most obvious of these was that the pitch of the propeller, a fundamental design property, had various values indicating that either the quality assurance process was flawed or that the personnel completing the document did not understand what they were recording.

The pitch was reported at various times as:

pitch stamped on propeller hub and nominated design pitch 4,571.0

average propeller pitch measured by Recon Propeller engineering 4,591.0

calculated average design pitch from average blade design pitches 4,599.8

total mean pitch reported in Wartsila QA document 4,719.8

mean pitch calculated by DNV after reviewing QA document 4,600

Based on the results reported in the Wartsila QA document the propeller does not pass the ISO 484 requirement for pitch accuracy. However, because the reported numbers varied widely, Interisland Line had the propeller measured independently and these measurements showed that the pitch was within the ISO 484/1 tolerance range. This anomaly appears to be because the factory report records the mean pitch as the measured pitch at 0.7R and does not calculate it in accordance with ISO 484/1.

There is no explanation as to why there is a difference between the calculated average design pitch and the pitch stamped on the propeller hub which is the nominated design pitch.

It was noted by a number of experts that the thickness of the propeller blades varied quite significantly although measurements showed that these fell just inside the tolerance band as allowed by ISO 484/1. These observations led to the decision to digitally scan both propellers and carry out a shape comparison which is reported on later in this report.


Surface trails on the starboard C blade similar to the markings left on the surface after weld repairs

It was also observed that there were surface markings on the starboard propeller which resembled welds. These were particularly intense on the C blade of the starboard propeller. Because the International Standard requires a record to be kept of all welds, weld maps and welding procedures these were requested from the Class Surveyors DNV who advised that they had no such records and they were not aware that any welding had taken place on either of the propellers.


To determine whether the surface markings were a casting defect or evidence of extensive weld repair both propellers were subject to surface etching once they were returned to New Zealand. This etching showed that the extensive surface markings were a casting quality defect known as seaming, the cause of which is not understood, but which is generally accepted as not having a significant effect on casting strength. The surface etch did reveal that there were unrecorded welds on both propellers in the B zone which should have been mapped and reported by the manufacturer to DNV.

The report on this work is Appendix K.

No explanation was provided by the manufacturer for the presence of these welds or for the lack of any records.

Once the two propellers were returned to New Zealand they were digitally scanned at the same time using the same equipment in the same environment with digital control measures to allow the

The arrow points to a significant weld repair revealed by surface etching


accuracy of the scan to be assessed as plus or -2 mm for the surfaces. Because both propellers are inward rotating one propeller was then digitally reflected so that the two digital images could be placed together and any differences in shape highlighted.

The propellers were aligned using the machined front face of the hub, and rotated until the A, B, C and D blades were in matching positions and the difference between the two blade surfaces could be assessed. This was mapped using a plus or -2 mm acceptable zone and any differences greater than this were colour-coded. This figure show the suction face which is where cavitation occurs.

It was clear from this comparison that there were significant differences between the A B and C blades of the port and starboard propellers, while the D blades fell largely within the base tolerance zone.

C to C

B to B

A to A D to D

Metres


This figure compares the pressure faces of the two propellers.

The C blades were significantly different, and the C blade of the starboard propeller was also displaced rotationally around its main axis.

D to D C to C

A to A

B to B


Comparison between the mapped differences in shape (colours) and the observed cavitation (inside line) showed a close correlation in location when the two images were overlaid.

6.4.3 Propeller fitting

Because of the regularity of shaft failures where propellers were secured to the tail shaft by a keyway recent shipbuilding practice is to secure the propeller and transfer the driving torque by means of an interference fit between the tapered end of the tail shaft and the tapered bore of the propeller hub.

This interference fit is defined by the distance that the propeller is forced up the tapered end of the tail shaft. This is a controlled procedure generally monitored by the class surveyor and documented in the shipyard records. The design of the interference is intended to hold the propeller firmly on the shaft without movement. There is significant pressure at the interface between the shaft and the propeller hub.

When two metal surfaces are in intimate contact a process called fretting can occur. This is generated by very small relative movements between the two surfaces. It is known that fretting can reduce the ability of steel to resist the initiation of fatigue cracking, resulting in failures at lower fluctuating stresses than would be normally expected. If the interference fit is inadequate fretting can occur. However, fretting can only promote the initiation of the crack, it cannot drive the crack through the shaft. An external fluctuating stress must exist that is great enough to do this.

Although the measurement records of the fitting of the propeller were provided by the shipyard, no other documentation relating to the fitting could be found. The documentation available did show that the required interference fit was achieved although a small anomaly in the push up data suggested that there was some plastic yielding of the propeller hub. This was not considered to be relevant to the failure process as it occurred on both propellers.

Line marking the extent of cavitation on C blade


The metallurgical evidence referred to above confirmed that there had been fretting between the shaft and the propeller hub on both the port and starboard shafts. It also confirmed that there was no fretting damage at the site on the origin of the uniaxial fatigue failure.

6.4.4 Propeller performance

A key factor in the performance of a propeller is a phenomenon known as cavitation. A propeller generates the thrust that pushes the ship through the water in two ways. By the back face of the propeller that pushes on the water as the propeller turns, and by the suction on the front face of the propeller as it sucks the water in front of the propeller towards it. While the pushing force is generally stable, the suction force depends on the water sticking to the propeller face. If the suction becomes too strong the water in front of the propeller cavitates and this force is significantly reduced. This is a situation that propeller designers can control by design and is to be avoided. It is measured by the cavitation number. Cavitation occurs when the number is less than -2.

The design performance of propellers is often checked prior to manufacture by using hydrodynamic modelling techniques. These same techniques were used to assess the performance of the propellers using the scanned shape of the propellers rather than the design criteria. The hydrodynamic modellers were also commissioned to determine, if possible, the effect of the bent tip on the C blade to see whether this was affecting the performance of the starboard propeller in some way.

Marin, a leading hydrodynamic modelling organisation based in the Netherlands undertook modelling. The brief was to determine whether there was a discernible difference between the forces generated by the port and starboard propellers, including modelling the starboard propeller with and without the bent tip. They were also asked to establish the propensity of the propellers to cavitate in an attempt to explain the cavitation pattern on the starboard propeller. They were also requested to assess the shape of the propellers against industry-standard profiles as the actual design profiles were never provided by the manufacturer.

The comparison between the propellers by modelling proved to be somewhat inconclusive. Marin concluded that any difference between the two propellers in terms of forces generated, with or without


the bend on the tip of the C blade of the starboard propeller fell within the uncertainty band of plus or minus 5% associated with their calculations. More accurate calculations were not possible.

They also reported that they were unable to model cavitation that corresponded to the observed damage on the propellers but noted that many of the shape anomalies in the scan had to be smoothed out to allow the calculation methodology to work. Note here that there was physical evidence that cavitation was occurring at small anomalies, hence the smoothing is a valid reason for the failure of the model to reproduce the cavitation.

Marin also noted that their modelling was only intended to assess cavitation originating at the leading edge, and would not detect sheet cavitation originating at surface defects. They were, however, able to provide maps of the propensity of the propellers to cavitate through the calculation of a standard measure called the cavitation number. This showed that according to their calculations the scanned shape of the propellers operated at a cavitation number much closer to the critical level than the number proposed in the propeller improvement report.

Figures 12 and 15 of Appendix M shows the propeller operating at a cavitation number of around -1 not the +1.9 in the Wartsila report (see page 22 of this report).

These results supported cavitation as a significant factor for consideration, but meant that the physical evidence became the most reliable indicator of any performance problems with the propellers.

Marin were provided with the photographs of the cavitation damage on the starboard rudder and asked whether they had ever seen cavitation damage of the type observed on the starboard rudder. They advised they had not, but that having discussed the photos widely amongst the design experts in the organisation they agreed with the Interisland Line investigation team that this damage was almost certainly caused by cavitation streaming off the starboard propeller.

The Interisland Line investigation team were able to show from underwater dive surveys that the paint damage to the starboard rudder was present after one year of service and prior to the appearance of the bent tip on the starboard propeller C blade, thus eliminating any effect that the bent tip may have had on the relative performance of the propellers.

The Marin analysis of the shape of the propellers also showed that there were some significant differences shown in the attached extract from the report which is appended to this document.


Analysis by Marin of the variability of the camber line of the starboard propeller against a design standard. The green line is blade C starboard propeller.

The Marin modelling did show conclusively that the sum of the forces acting on a propeller varies as each blade passes through the wake field which is the area of slower water flow close to the hull. Based on the position of the paint damage to the starboard rudder was clear that the cavitation streaming from the propeller was originating in the wake field.

Given that the C blade of the starboard propeller had the most severe surface cavitation damage the physical evidence points strongly to this blade cavitating as it passed through the wake field, streaming cavitation back towards the rudder, and almost certainly resulting in the force generated by the suction side of that blade significantly reduced. With the other blades performing normally this imbalance of forces would in turn create a couple on the propeller which would fluctuate and produce maximum stresses in the shaft at the forward end of the propeller hub where the fatigue fracture occurred.


7 Discussion and conclusion The investigation team consisted of Kiwirail staff led by an independent consulting engineer, but was empowered to commission such independent experts as were required to clarify technical issues and assist in the determination of the cause of the loss of the starboard propeller of the Aratere. Experts in metallurgy, propeller engineering, shaft dynamics, alignment and vibration, and experts from the international class certification organisations were all involved, as was local expertise in machining, non-destructive testing, and underwater inspection.

These organisations were provided with broad briefs to use their expertise to answer questions, and were encouraged to contribute observations from their experience which could assist the investigation team in reaching a conclusion. The reports from these organisations have been summarised in this overview report and are appended to it for readers who require more technical detail.

The investigation team set out a plan that would allow analysis of all possible credible failure paths and commissioned independent testing with this could contribute value to the investigation process. Some of the failure paths lead rapidly to technical conclusions which ruled them out as credible causes and no further investigation in those areas was carried out.

The availability of a similar drive system on the port side of the vessel provided a valuable benchmark to assess the significance of observed differences and similarities, allowing more weight to be given to the differences as potential contributors to the failure.

In some areas, particularly in the hydrodynamic modelling and theoretical stress analysis areas the number of assumptions that had to be made to allow technical processes to be used led the team to give less weight to the outcome of those analyses.

Historical evidence allowing a timeline to be established where the team could see the sequence in which some of the physical evidence appeared in the record provided valuable information as to when that evidence appeared and allowed certain issues such as the bend on the tip of the C blade of the starboard propeller to be discounted as causative of cavitation evidence as the damage preceded the appearance of the bent tip.

In the end the team concluded that the most reliable information was the physical evidence, particularly the metallurgical investigation and the conclusions reached and the physical evidence of cavitation on and around the starboard propeller. The team also came to the conclusion that the shape differences measured on the starboard propeller when compared to the port propeller were significant and consistent with the physical evidence of the fatigue fracture and the cavitation damage.

Considering the physical evidence available, and by comparison between the port and starboard propeller and shaft systems we get the following summary of facts:

1. Fretting on the port shaft was worse than fretting on the starboard shaft.

2. The naturally fluctuating forces of the port propeller were not able to initiate or drive a fatigue failure on the port shaft despite the higher level of fretting present.

3. There was no fretting damage present on the surface of the starboard shaft where the fracture originated.

4. There was clear evidence of abnormal performance of the starboard propeller by way of cavitation damage to the suction surfaces of the propeller and paint erosion on the rudder.

5. The fluctuating forces generated by the starboard propeller initiated and drove a uni axial fatigue failure which originated close to the C blade.

6. The surface damage from cavitation was most pronounced on the suction face of the C blade of the starboard propeller.


7. By comparison between the scanned shapes of the port and starboard propellers the C blade of the starboard propeller was most significantly different from other blades on the starboard propeller and from matching blades on the port propeller.

8. It is known that cavitation affects the capability of a propeller blade to generate thrust.

9. One non-performing blade on a propeller would generate a uniaxial force which fluctuated once per rotation in a consistent transverse direction across the shaft.

10. That fluctuating force would generate a couple on the propeller that would act to maximum effect at the plane where the fracture occurred on the starboard shaft.

On the basis of the physical evidence it is reasonable to conclude that a malformed C blade on the starboard propeller was the primary cause of the failure. If this blade had been well formed and the propeller had performed symmetrically the uni axial driving force required to initiate and drive the fracture would not have been present.

Eur Ing Stephen Jenkins BE (Mech) MSc CPEng FIPENZ FIMechE

Technical Director Aurecon New Zealand Limited

Technical leader KiwiRail Investigation Team

aratere shaft failure investigation final report interisland line

Documents