Loosening is a process of rotation of the bolt in a direction opposite to the tightening. This can cause a loss of preload (tension created in the screw shank upon tightening) which can cause failures in the machinery [9]. For instance, Olmi [10] investigated the crack generated in a geartrain housing of an asphalt milling machine. The experimental simulation of the entire process with the assembly and the possible external loads showed that there was a stress increase due to the load hysteresis. The stress peak was present especially at the curvature where the crack originated. This load hysteresis was made possible due to the vibrational loading by the engine, which resulted in the self-loosening of the bolts. Therefore, he suggested design improvements along with the use of dampers to prevent the vibrational loading generated by the engine. Another preload loss due to dynamic loading and reported in the literature occurred in a pinion shaft for wheel loaders [10]. Croccolo et al. [11] investigated the fracture of the pinion shaft at its threaded end. A ring nut was used to preload the angular roller bearings. It was shown that the preload loss that took place under the nominal duty cycle is around 15 kN, due to which loosening starts occurring. This loosening results in the crack of the weakest portion of the assembly, i.e., the root of a sharp notch next to the threaded portion where the ring nut engages. The suggestion for maintaining the preload given was to use either an adhesive threadlocker or a mechanical locking device. Otherwise, the design of the assembly should be altered to lessen the amount of the loading, which includes using more compliant sleeves or even removing the inner sleeve [11].

Preload loss of threaded fasteners is also affected by the types of materials in use. For instance, Den Otter and Maljaars [12] reported that when stainless-steel bolts are used in aluminium-plated slip-resistant connections, significantly greater preload loss is observed as compared to the carbon steel bolts as specified in ISO 898-1. This preload loss is due to several reasons, among which viscoplastic behaviour, plate contraction and drop in the temperature are mentioned by the authors. The viscoplastic behaviour includes creep and relaxation phenomena. Therefore, it was suggested to use a higher safety factor (1.3–1.5 instead of 1.25 provided in EN 1999-1-1) depending on connection configuration to compensate for the increased preload loss due to material combination in the ultimate limit state [12]. It is worth remarking that a higher factor of safety determines an increase in the weight and cost of the engineering structure [13]. Therefore, if these aspects are crucial, the safety factor should be minimized by taking into account all the available information such as thread friction, rough surface finishing and the actual distance between the bolt axis and the edge of members [14].

Moreover, the initial preload application can also affect the loosening of the bolted connections. Bhattacharya et al. [15] studied the effects of initial preload application on M16 steel bolts that loosening is minimum when the initial preload is greater than or equal to 12 kN, as shown in Table 1. This might be due to the pronounced deformation of the points of contact with greater force values, which enable them to offer more friction [15].

Various theories have been put forward to explain the process of loosening. The first theory on the rotation mechanism in bolts was coined by Sakai [16] in 1978. He reported that loosening occurs when the torque for bolt underhead slip is greater than the loosening torque for thread surface slip and lesser than the tightening torque required by the threaded portion. It was assumed in the theory that the nut is fixed; hence, the condition for loosening was based only on the torques on the bolt surfaces. In the classical theory used by Junker [17] (also known as the “complete slippage theory”), loosening is said to occur when frictional forces on the threaded surfaces are completely overcome. Yamamoto and Kasei [18] proposed a two-stage theory, where the first stage represents slippage on bolt-nut thread surfaces producing torsion on the bolt shank. In the second stage, the rotation of the bearing surface occurs on a moving plate, which results in a release of torsional deformation and produces loosening. In Junker’s theory, the bolt loosens first; however, in Yamamoto’s theory, the nut loosens first [18]. In contrast to the classical theory, Pai and Hess [19] proposed the “local slippage accumulation theory”. According to such a theory, complete slippage is not necessary to initiate loosening. Loosening starts when local areas of slippage accumulate together over the number of vibration cycles. Recently Li et al. [20] extended Sakai’s theory of self-loosening to include free nuts and free bolts under rotary vibrational loading, which agreed well with the experimental evidence. According to their theory, local slippage does not account for loosening, as opposed to the Pai and Hess theory. In Pai and Hess’s theory, the transverse load can introduce eccentricity, which in turn allows the piling up of local slippage. However, Li et al. [20] introduced a theory for rotational loading in which there is no eccentricity and hence does not allow accumulation of local slippage. These theories are summarised in Table 2.

Loss of preload can occur with or without external excitation. This phenomenon of loss of preload has been studied from both standpoints in the past. Bretl and Cook [21] performed a static numerical analysis of the loading distribution over the threads during bolt tightening. Moreover, for the calculation of thread force, Miller et al. [22] established an equivalent spring model for fastening of the bolt and the nut. This approach is relatively simple and immediate. Miller’s model was later improved by Daadbin [23] by adding stiffness and damping, which he claimed to be influential factors for load transmission. Further investigation into the problem was performed by Daadbin and Chow [24], who examined the impact of the loading scheme on the preload loss. However, they neglected the deformation of the thread and modelled only one thread, because of which their model was unrealistic. Nevertheless, they concluded that the preload loss is less for the fine pitch, as it leads to a greater coefficient of friction and duration of applied force [24]. Moreover, an exponential expression for the stiffness of the members of bolted connections was developed using the numerical modelling by Wileman et al. [14]. Additionally, Chen et al. [1] made a significant improvement in Miller’s model by considering a multiple-thread system and adding the threads and bolt shank deformations. They modelled both the static and dynamic loading of the bolts and also determined if the bolt is exceeding the elastic limit by excessive initial preload. They calculated the effects of various aspects including the lead angle, material type, and external excitation parameters using the developed models. They concluded that a larger preload results in the extended service life of a bolted connection, whereas the trend is opposite for the lead angle of the threads and frequency of excitation force. They demonstrated the use of the model in the optimization of the design of a missile tail wing. Using the model, it was shown that preload loss, e.g., was decreased by decreasing the lead angle of the threads [1].

The various kind of loadings on the bolted joints can result in the loosening of the bolted joints. Work has been done to study the loosening process under different loading conditions such as axial, transverse, rotational vibration, impact loading and thermal loadings.

A. Vibrational Loading

B. Impact Loading

C. Thermal Loading