Introduction
Bolted joints are important parts of mechanical structures, and their reliability is essential due to the potential consequences of failures.
Due to their wide use in industry, bolts and screws have become one of the most standardised parts in mechanical engineering. In spite of being exhaustively informed by many international standards, bolted joints are still causing problems to the industry today, which results in many catastrophic failures and significant losses. In spite of their apparently simple and obvious design, bolted joints are often not properly understood.
This article deals with a general approach to better understanding bolted joints.
Failure modes
There are many failures modes typical to bolted joints. These include: screw thread failure, plastic deformation, insufficient axial tension during installation, static failure, fatigue fracture, delayed fracture, stress corrosion cracking, separation of joined parts, slip between joined parts, loosening of the bolt, and decrease of axial tension [1].
In many flange cases, the design of a proper bolted joint is not as simple as calculating the necessary bolt tension. Infinitely ridged joints behave differently from real joints where the faces can compressand deflectdue to compliance within the structure. In such instances, bending under the head or at the threads can result in bolt failure. As a result, Finite Element modelling and bolt axial/bending stress measurements are now playing a greater than ever role in understanding bolted joints. Depending on the mounting arrangement, a bolted joint can be modelled as an encastré support with the highest bending at the threads, or a pin-pin support, where the internal threads are not constrained and are free to tilt - leading to the highest bending moments under the head.
In several cases experienced by industry today, bolts with a clearance between the shank and the hole can develop strong bending resonance due to inherent low damping in the first mode of vibration. These failure modes are especially important in gearboxes, where a small excitation from Transmission Error (TE), can excite strong resonance in the first bending mode of vibration, leading to rapid fatigue failures.
Understanding bolted joints
To correctly designbolted joints, one must understand that for all intents and purposes, bolts should be treated asextension springs. To achieve correct bolt preload, the bolt must be stretched by a well-defined amount, often taking the bolt close to its yield point. The amount of bolt preload can be calculated from Hooke’s law: knowing bolt stiffness k, the achieved clamping force P can be calculated as:
P = k x (1)
wherex is bolt extension.
From the above equation, it can be seen that when a load is applied which tends to separate the flanges, the bolt will not see any increase in tension (provided that the flanges are infinitely rigid) until the separating load exceeds the bolt preload, and the joint starts to separate. As the flanges are actually compliant, an applied load will result in a deflection of the flanges themselves, and a small change in bolt tension may occur.
It is commonly thought that if a bolt loses its tensionor fails, it should be tightened with a higher tension, or replaced with a bolt of larger diameter. This can, however, result in no improvement to the bolted joint. In fact, in most practical applications, it is better to reduce the diameter of bolts, counter-intuitive as this might seem!
To reduce the risk of losing tension, industrial bolts which secure critical components should be highly elongated. It is an incorrect design approach to use short, large diameter bolts, where the designed tension is achieved with a very small amount of axial elongation. With such a design of bolted joints, a very small amount of thread settling, fretting under the head or joint shrinkage will result in loss of tension and hence failure of the bolted joint. It is better to use several lower diameter bolts, with the same equivalent load carrying capacity than one thick bolt.
The second important design aspect to consider is that even the most accurately controlled tightening torque will not necessarily result in the intended bolt preload. In many practical applications, bolt tension can significantly vary, in spite of achieving the recommended and accurate tightening torque.
There are several influences onthe relationship between tightening torque and bolt tension, including: a varying coefficient of friction in the thread, surface coating, friction between head and clamping face, and manufacturing tolerances. These factors can combine to result in significantdiscrepancies in bolt tension,as summarised in Figure 1.
Figure 1: Graph showing difference in preload as a result of coefficient of friction inaccuracies[1]
A number of independent tests performed in industry confirmthat using a torque control bolt tensioning method may lead to significant problems where the desired tension is compromised. For example, case study 1[2] performed on 36 bolts in an HP steam heat exchanger predicted that bolts should be torqued to 4,380 Nm to reach the desired preload. Upon installation and tightening of the bolts to this desired value, tension measurements were conducted. The results showed that for the bolts to reach the desired preload, the necessary torque was not constant, and in fact ranged from 2,259 - 5,874 Nm (52 - 134%). Case study 2[2] considered similar tests conducted on 72 bolts on a Texaco Tartan pedestal crane, and showed that the required torque varied from 1,700 - 9,000 ft-lbs (61 – 321%), in spite of the predicted torque to achieve correct tension being 2,800 ft-lbs.
There is a general agreement that the best way to ensure high integrity of bolted joints is to measure bolt tension instead of bolt tightening torque. Historically, direct measurements of bolt tension required a specialised and often expensive setup.
Ultrasonic sensors can be used for bolt tension measurements. Such sensors, however, suffer from one major problem: they require expensive setup as well as precision reference faces.
Other solutions (Rotabolt, Valley Forge, etc.) involve direct bolt extension measurements using a suitably inserted gauge pin. As the tension in the bolt increases, which results in overall bolt extension, the gauge pin length remains fixed and retracts relative to the top surface of the bolt. As a result, when the preload drops below a certain level,the change in the relative gauge pin height is detected and indicated by a moving washer or other means.
Although these methods allow reliable tension control, they require manual access to the bolts during inspection (often in remote or offshore locations).
An interesting recent development uses a strain-measuring sensor to indicate bolt tension. The measuring head is fitted directly on the bolt head (see Figure 2). Each bolt wirelessly interfaces with a stationary receiver, allowing truly remote monitoring of bolt tension. Due to its wireless capability, the system can also be used in rotating machinery applications. Data automation behind the Rotabolt-TD system notifies key stakeholders when a loss of tension is detected. Reports are automatically generated and instantaneously deployed by SMS/email, allowing continuous and reliable monitoring of bolts used in mission critical applications.
Figure 2:Rotabolt-TD; wireless, remote bolt preload monitoring system.
Conclusions
Proper understanding of bolted joints is fundamental in creating reliable, long-term engineering solutions. Torque control methods must be used with caution as they may result in significant variations in resulting bolt tension. Newly emerging technologies are likely to play an important role in assisting with reliable bolt installation and remote, long-term bolt tension monitoring in mission critical applications.
References
- T. Sakai. Bolted Joint Engineering – p 19 & 51 Fundamentals and Applications. Beuth 2008.
- B. Deeley. Assured Reliability & Lowest Cost of Ownership through Measurement of Wind Turbine Bolted Joints. 2011.
For further information contact:
Transmission Dynamics