Coupling Bolt Failures In Short Shaft Systems For Two-Stroke Engine Installations

Herbert Roeser Dilip Kalyankar Trans Marine Propulsion Systems, Inc. & Nils Juhlin NovaComp Engineering, Inc.

Synopsis The importance of proper pre-loading of coupling bolts cannot be too strongly stated. Most coupling bolt problems relate to inadequate pre-loading of the bolts during installation, or relaxation of the bolt load in service due to rough or inaccurately machined coupling faces. This will eventually result in working of the coupling joint under the applied loads leading to fretting and seizing of the bolts and coupling faces which eventually will lead to fatigue failures of the coupling bolts. This paper looks into the effect of the dynamic response of the system on the bolt design with two examples from our own experience in this industry. One case experienced vibratory torque exceeding the nominal torque of the system and the other case experienced vibratory torque equal to the nominal torque of the system. Since nominal torque is used as design criteria for bolt design according to the classification societys requirement, one of the cases had high vibratory torque exceeding the design criteria of the coupling bolts. Compounding the problem is Poissons effect, on the bolt flange and the surrounding flange material, which severely reduces the interference fit during bolt preloading. This, combined with excessive vibratory torque, generates relative micro motion between the bolt shanks and flange holes as observed in actual failures. The strength of a bolted assembly for resisting dynamic loads depends greatly upon the relative flexibilities of the bolts and of the reacting members. If the bolts could be made very flexible and

the flange assembly very rigid, then, only a very small proportion of the dynamic load would be felt by the coupling bolts until actual separation of the coupling faces occur. Consequently, it is very important in future design of coupling bolts and coupling flanges to provide very rigid flange designs and to give the coupling bolts as much flexibility as possible by under cutting the bolt shanks as much as other considerations will permit. The amount of under cutting that can be permitted will depend on the length of the fitted portion of the bolt, and the magnitude of the interference fit required. It must be understood that if the interference fit is too excessive it can become a problem to achieve proper pre-loading of the bolts during assembly because the stretch of the bolt will tend to be restricted to the portion between the nut and the fitted part of the shank. This may lead to the strain not being distributed along the length of the bolt, but will tend to be confined to the undercut shank and threads between the fitted portion and the abutment face of the nut.

Introduction: Engine manufacturers have all followed more or less the same design parameters to improve their engines, by raising the power output while reducing the fuel consumption. This means that with the improved engine efficiency, the calorific value of the fuel used can be converted as efficiently as possible into power. For the optimum results, the duration of the combustion must be kept as short as possible. By doing so, all the energy is released from the fuel immediately as the piston passes the top dead center position. Due to the much higher combustion pressures, this released energy is converted more efficiently into tractive power. The higher combustion pressures and temperatures at which modern engines

operate, meant a thorough redesign of these engines. This also means that the dynamic response of a system involving these modern engines has changed, particularly the torsional response of the system. One aspect of greater complexity of todays dynamic problems is the absolute increase in the quantity of calculations that are required. In this paper the author will describe two cases with very similar installations. One case where loose shaft flanges were detected in the propeller shaft system, with damaged coupling bolts and a second case, operating without any problems. From our research with many installations we found that the problem discussed in this paper largely depends on the dynamic 2

response of the system at or near the barred speed range, coupling bolt design parameters, and coupling bolts interference fit.

1.0 Torsional response of both systems and their respective barred speed ranges: In regards to two stroke engine installations from a torsional point of view, only the first two modes of vibration are of interest. In order to predict the behavior of any physical system, a model suitable for mathematical analysis is necessary. The requirement for the model is that the model is able to predict the behavior of the system with sufficient accuracy. The development of higher output engines and increased complexity in operation and design of propulsion plants has proven the need for mathematical models capable of predicting behavior at non-resonant conditions, effect of damping, branched systems and vibratory torques and heat losses in elastic couplings and torsional dampers. Respective manufacturer often specifies mass-elastic properties of each component in a shaft system. In addition to the stiffness and inertia values, the damping coefficients and dynamic magnifier are part of a complete mass-elastic system. All these values constitute the mathematical model of the shaft system to be analyzed.

In general, a torsional vibration analysis consists of two parts, first part is natural frequency and mode shape calculations, and the second part is the forced damped calculations. The natural frequency and mode shape calculations will determine the resonant or critical speeds, and the relative deflection curve of each mode of vibration. The vector sum is calculated based on the firing order and deflection curve and indicates the relative significance of each order for a specific mode of vibrations. Fig. 1 & 2 show the mass-elastic systems for the two cases to be analyzed.

Fig. 1 Mass Elastic System - Case 1 3

torque reaches a value at point 1 before it goes through the barred speed range. This point 1 is already above the nominal torque value of the system marked by point 3. The nominal torque value of the system marked by point 3 is used in design of the coupling bolts. When the system comes out of the barred speed range, it reaches a torque value marked by point 2, which is even higher than point 1, and hence the system has still higher vibratory torque than the nominal torque value. This is the system, which has experienced coupling bolt failures. Fig. 4 shows case 2, where the vibratory torque reaches point 1 before entering the barred speed range. This vibratory torque value is equal to the nominal torque value of the system marked by point 3. When the system comes out of the barred speed range, the vibratory torque value reaches point 2, which is very low compared to the nominal torque value of the system at point 3. This system never had any coupling bolt failures. Both these systems in case 1 & 2 have a quick pass through the barred speed range and hence the vibratory torque values inside the barred speed range as shown in the following graphs never materialize. Only the vibratory torque values outside the barred speed range get enough time to materialize.

Fig. 2 Mass Elastic System - Case 2

Based on the calculated results, the stress amplitude in the propeller shafting at the nodal point for the first mode of vibration warrants a barred speed range for case 1 between 38 to 46.5 rpm, and for case 2 between 53 to 65 rpm. This is in accordance to ABS rules since both class of vessels involved are classed by ABS. Following Fig. 3 and Fig. 4 show the vibratory torque transmitted through the system in case 1 and 2 respectively. As can be seen from Fig. 3, in case 1 the vibratory 4

2.0 Coupling Bolt Design Based on ABS Rules: The coupling bolt design depends on the required diameter of the intermediate shaft. The required diameter of the intermediate shaft is calculated from the nominal torque of the shaft system at 100% MCR. As can be seen from the attached calculations, the bolts are designed in accordance to pure shear loading, no other loading is considered in the design calculations of the coupling bolts.

Fig. 3 Vibratory Torque Case 1

SHAFT DESIGN ACCORDING TO ABS According to ABS class rules, the intermediate and the propeller shafts are designed using the following formula

D = 100k 3

H 560 N U + 160

where,
k = 1.0 for shafts with integral forged coupling flange H = engine power at MCR (kW) N = engine speed at MCR (rpm) U = minimum specified tensile strength of the shaft material N /mm 2

Intermediate Shaft for Double Eagle:

D = 100k 3
Fig. 4 Vibratory Torque Case 2 5

H 560 N U + 160

D = 100 1.0 3

7941 560 96 650 + 160

Coupling Bolt for Double Eagle:

D = 385.27 mm

db = 0.65

Intermediate Shaft for Liberty Eagle:

D3 (U + 160) NBU b

db = 0.65

385.27 3 (650 + 160) 10 615 650

D = 100k 3

H 560 N U + 160

db = 69.96 mm

D = 100 1.0 3 9465.9 560 127 780 + 160

D = 354.11 mm

COUPLING BOLT DESIGN ACCORDING TO ABS According to ABS class rules, the coupling bolts are designed using the following formula
D3 (U + 160) db = 0.65 NBU b

Coupling Bolt for Liberty Eagle:

D3 (U + 160) db = 0.65 NBU b

db = 0.65

354.113 (780 + 160) 6 700 780

db = 73.36 mm

where, D = required diameter of the intermediate shaft based on nominal torque (mm) U = minimum specified tensile strength of intermediate shaft material ( N /mm 2 ) N = number of bolts in one coupling B = bolt circle diameter (mm) U b = minimum specified tensile strength of the bolt material in ( N /mm 2 )

Actual dimension for the bolt diameter used = 85 mm

Fig. 5 Coupling Bolt Design Case 1

3.0 Finite Element Analysis: Based on our observations of the several failed coupling bolts in real life, fretting corrosion on the bolt shank indicates micro motion between the bolt shank and the bolt holes in the coupling flanges. A detailed analysis became necessary to study the coupling bolt failures and hence a finite element analysis was conducted on a coupling bolt used to transfer torque between two flanges. The loading consists of an interference fit followed by a preload that is generated from the prescribed bolt torques. Results indicate that the interference fit is severely reduced due the application of the preload. A shaft torque load is also simulated to illustrate the loading cycle that the shank of the coupling bolt experiences.

4.0 Analysis and Results: The finite element model shown in Fig. 7 was developed using MSC Mentat and is based on geometry from the actual drawings. The geometry is representative of the Double Eagle system. The model consists of two shaft sections, the coupling bolt and a ring of preload elements that is used to generate the axial preload in the bolt. In order to quantify the effect of the varying stiffness in the flange, around the bolts, a 3D model was required. The inner ring of elements of the flanges were assigned an Fig. 6 Coupling Bolt Design Case 2 elastic modulus 10 times that of steel in order to simulate the shaft stiffness, thereby 7

enabling the use of fewer elements. The following loads were implemented and are imposed on the model sequentially; 1. 0.005 mm radial interference fit between the coupling bolt shank and the flanges. This is achieved by thermally expanding the bolt. 2. Preload the bolt to 545.5kN axial load (5200 N-m torque). This is achieved by thermally expanding the ring of preload elements. 3. Apply opposing torques to the two flange sections This analysis was intended to qualitatively demonstrate what loads the coupling bolt experiences during assembly and loading. As such, exact material specifications were not required and standard steel properties were assumed for all parts as shown in Table 1. A contact method is utilized which prevents penetration between components, but allows them separate for tensile forces normal to the contacting surfaces. When in contact, the components are permitted to slide relative to one another. A coefficient of friction equal to 0.15 was used in all analyses. Circumferential symmetry conditions were applied to the external vertical flange faces (with circumferential normals) using rigid surfaces such that only radial and axial displacements were permitted. Fig. 8 shows the contact pressure (normal stress) distribution on the coupling bolt shank due to the interference fit and also due to the preload. Note that the contact pressure has reduced to zero on the majority of the shank after 100% of the preload has been applied. The plot shown in Fig. 9 illustrates 8

Fig. 7 Finite element model of coupling bolt/flanges assembly.

Component Material Elastic Poissons Modulus Ratio (GPa) Flange Coupling bolt Preload ring Steel Steel N/A 206.9 206.9 206.9 0.3 0.3 0.3

Table 1. Elastic Properties

how the contact pressure is reduced as a function of preload. The curves represent the normal stress as a function of circumferential distance around the shank. The results indicate that the interference fit will be lost once 60% of preload is applied. The variation in the normal stress relating to the circumferential position on the shank is due to the surrounding stiffness variation of the flanges.

b.) due to preload Fig. 8 Normal stress distribution on coupling bolt

Interference Contact Stress Distribution

A D

A B C
At At At At At 0% of preload 18% of preload 31% of preload 45% of preload 59% of preload

-2 -4 -6 -8 -10 -12 -14 0 50 100 150 200 250 Circum ferential Distance Around Shank (m m )

a.) due to interference fit Fig. 9 Normal stress as a function of circumferential distance around shank for various preloads

The preceding results indicate that minor gapping will exist. When a torque is applied to the shaft, these gaps will allow the transferring loads to be more concentrated at the reduced contact areas. Fig. 10 shows the contact pressure that develops. As this load 9

cycles from negative torque to positive torque, the coupling bolt will react by moving within the flange hole and reversing the contact pressure location. During this process, small sliding will occur between surfaces.

Fig. 12 Associated maximum principal stress distribution at 1100kN.m shaft torque

Fig. 10 Load concentration during applied torque to shaft

Fig. 13 Associated minimum principal stress distribution at 1100kN.m shaft torque

5.0 Findings of the Finite Element Analysis: The interference fit is negated by the Fig. 11 Associated Von Mises stress distribution at 1100kN.m shaft torque 10 application of the specified preload. This is due to Poissons effect which causes the bolt

diameter to contract and the flange hole diameter to dilate. These two opposing motions generate a minor gap between the shank and the flange hole. Axisymmetric models were previously analyzed for radial interferences of 0.0025 and 0.010 mm. Those cases also lost their interference fit during preload. As a result of the minor gapping, small relative motion between the shank and the flange hole will occur during the cyclic loading from vibratory torque. This may lead to damage associated with wear, such as fretting. From the results of the FEA shown in Fig. 11, 12 &13, it can be seen that the stresses are highest in low diameter areas of the bolt. This is an expected outcome of the analysis but the observed coupling bolt failures do not occur in these areas of highest stresses. This further supports the theory that the observed failures are due to stress concentration developed by micro motion. The FEA model was developed for illustrative purposes only and not for any design purpose. Approximations, such as the flat rigid periodic boundary conditions and the artificially stiff inner ring of elements to represent the shaft, were used to reduce model development and analysis time. As a result, these approximations may affect the accuracy of these stress results. Hence, these results should be considered from a qualitative point of view and should not be used as a design guide. 11

6.0 Conclusion: At present, most classification societies base the bolt design on pure shear. From the several examples of coupling bolt failures that we have experienced, the majority of the failure of these bolts does not occur in shear but they occur from fretting corrosion due to micro motion. The future design of the coupling bolts and coupling flanges should include the dynamic response of the system. Considering the nominal torque of the system for the design of coupling bolts is not enough, especially in cases where vibratory torque exceeds the nominal torque of the system. In such cases, considering the peak vibratory torque value outside the barred speed range is essential in reducing the failures due to fretting corrosion. The Poissons effect on the bolt shank and the flange should also be considered in the bolt design by making the bolts as flexible as possible and making the flange as stiff as possible.

References: Practical Solutions of Torsional Vibration by W. Ker Wilson. Volume Three Strength Calcualtions Third Edition. Failure Analysis of Propulsion Shafting Coupling bolts by Yuzhong Song, Hiromi Shiihara, Daisuke Shiraki, and Yuya Nagayama.