Evaluation of Quality for Flow Drill Screw (FDS) Joining Process Based on Loosening Torque
Article information
Abstract
Abstract
Although the SPR process is widely used for joining of multi-materials, the FDS process is used when one-way joining is required, such as space constraints or closed cross-sectional parts. The quality inspection for the FDS process is basically an appearance inspection or cross-sectional inspection. There is no non-destructive test method that can be used in the actual mass production line, and a method of utilizing loosening torque is being studied. In this study, the factors affecting the loosening torque are analyzed, and an experimental equation that can predict the loosening torque is derived. Since friction force between the screw and the upper plate and sticking on the screw thread resist against loosening torque, the prediction equation for each factor is considered and the least square method is used to obtain the coefficients of the experimental equation. The prediction equation of the loosening torque derived through experiments on 36 material combinations predicts the experimental values well. When a quality requirement value with a given margin is applied, it can be used as a non-destructive test method.
1. Introduction
Environmental and automotive fuel efficiency regulations are prompting the increased use of aluminum materials in automobile body components. Conventionally, steel plate components have been joined and assembled using spot welding. However, the partial application of aluminum parts rendered the conventional spot welding method ineffective for joining dissimilar materials with existing steel plates or aluminum materials. Various dissimilar material joining methods, such as the self piercing rivet (SPR), are employed to join dissimilar materials or same aluminum materials. However, when connecting components with closed cross-sections, such as aluminum extrusion parts, a unidirectional joining method is required. To achieve this, the flow drill screw (FDS) process, as shown in Fig. 1, is adopted. Given the significant impact of this joining process on body stiffness and crash performance, evaluating the quality of the joining is mandatory. Kim et al.1) highlighted four aspects that influence the quality of flow drill screw (FDS), whereas Kang et al.2) examined the quality based on FDS process variables. Lee et al.3) assessed the pull-out strength in relation to the rotation speed and the applied pressure and derived a mathematical regression equation. Furthermore, numerous studies have examined the influence of quality based on the process variables4-6).
Process steps of FDS assembly without pre-hole [EJOT]
Typically, for spot welding, basic non-destructive inspection techniques like ultrasonic inspection or driver test are employed in the field, along with cross-section analysis. However, for joining methods like SPR and FDS, no standardized quality inspection method is available for field application. The objective of this study is to derive a prediction equation for the inspection criterion of joining quality, using the loosening torque as a non-destructive inspection method for the FDS process.
2. Analysis of the FDS Process
2.1 Analysis of the Joining Process
This study investigates a method for indirectly assessing the joining quality of the FDS process by using the loosening torque, which is the inversion of the tightening torque in the FDS process. As shown in Fig. 1 below, the FDS process is a joining method that involves high-speed rotation of a screw to penetrate the material being joined. At the same time, this process applies a bonding force to the material by means of high-temperature frictional heat transmitted from the screw thread. Fig. 2 displays the torque exerted on the screw in each stage. As shown in Fig. 2, the torque consistently increases during the initial penetration stage, followed by a temporary decrease after penetration. However, as the screw head makes contact with the material, the torque experiences a rapid increase, ultimately completing the joining process after rotating up to the maximum tightening torque applied to the equipment. In general, the loosening torque is directly influenced by the tightening torque, making it a significant determining factor. If the FDS joining quality is high, the screw will remain securely fastened when subjected to reverse torque, up to a critical point, enabling non-destructive testing.
Torque during FDS assembly [EJOT]
2.2 Analysis of the Cross Section
The cross-section of the actual FDS joint is shown in Fig. 3. Screws, which are applied depending on the chosen material combination, are primarily classified into two varieties, M4 and M5, based on the diameter of the threads. Typically, the combination of materials to be joined is used when it is difficult for the joining equipment to reach the bottom portion of the material or when the lower part has a closed cross-sectional shape that interferes with the direct joining point. It is mostly applied to two or three layers, with the lowest layer frequently consisting of aluminum extrusion or cast material, and the upper layer being aluminum or steel plate. If the topmost material is an ultra-highstrength steel plate that presents a challenge for a screw to penetrate, it is occasionally joined by creating a pre-hole.
Sections of FDS assembly
In Fig. 3(a) (b), the cross-sections obtained from destructive inspection and CT scan are displayed simultaneously, categorized by the material of the top plate. The variables influencing the loosening torque can be categorized into two types: the friction between the upper screw surface and the top plate material, which generates resistance prior to rotation while applying the loosening torque, as depicted in Fig. 3(c), and the joining force between the threads and the material resulting from high temperature. This resistance is anticipated to be significantly influenced by the characteristics of the target material, including its type, thickness, and contact area. Furthermore, unlike the frictional force experienced by conventional bolts, the screw thread connection remains in a sticking condition as a result of the high-temperature rotational frictional heat. Consequently, it is intricately linked to the tightening torque, which directly impacts the frictional heat.
2.3 Relational Formula
The process of loosening will begin when the two resistance factors indicated earlier become the same as the torque exerted in the loosening direction. In this study, such critical torque is referred to as the loosening torque (Tloosening). Although it is not possible to calculate this loosening torque theoretically, an empirical formula based on fundamental tests is developed for use in practical non-destructive testing. The formula for loosening torque is obtained by separating it into two main components as shown below.
2.3.1 Friction Force of the Screw Head
As shown in Fig. 3(a)(b), when the upper layer of material is penetrated, it bends upward as a result of the rotational force and moves upward into the head space. In this manner, the material is introduced to fill up the vacant area within the head until the rotation is fully completed, ultimately making contact with the head surface, as shown in the head part of Fig. 3(c). Hence, the frictional resistance that affects the loosening torque can be calculated using Eq 1) provided below.
In other words, the friction coefficient (μ), contact area (A), and axial compressive force (P) can be assumed with first-order functions. The axial compressive force can be substituted by the tightening torque (τ), which exhibits a linear proportionate relationship7). Furthermore, it may be associated with the elastic coefficient of the material in contact. However, as this coefficient is a constant value determined for each material and rivet, taking into account the friction coefficient and contact area, the ultimate relational formula can be represented as in Eq (1) by incorporating the constant coefficient k1. The frictional behavior of the upper plate caused by the screw may vary depending on the steel plate and aluminum utilized. As a result, the relational formula in Eq (1) will be calculated separately for the steel plate (k1_ST) and aluminum (k1_AL).
2.3.2 Joint Force at the Screw Thread
Similarly, as shown in the screw section in Fig. 3(c), the joint force of the screw section is determined by the joint force between the material and the screw thread, as well as the joint area (A). Once the joint force is lost, a certain amount of frictional force will exert frictional forces similar to those of a screw. The loosening torque of these bolts typically exhibits a linear relationship with the axial force (P). Similar to Eq (1), the axial force can be substituted with the tightening torque7). The joint force is also related to the contact area and has been quantified using the thickness tAL and tST, which are directly proportional to the constant diameter (D), for each material. The components can be simplified in this manner because all proportional quantities, including the friction coefficient, joint characteristics for each material, and screw thread diameter, are constants and function as coefficients of the tightening torque. Furthermore, although the contact with the upper plate is characterized by elasticity, the screw thread section involves plastic deformation. Therefore, the yield strength (YP) of each material was also taken into account. Finally, it can be structured as a function of coefficients k2_AL and k2_ST, with YP differentiated by material, as shown in Eq (2). Given that the stacking order of materials is irrelevant, it can be determined by summing the aluminum materials that make contact with the screw thread and the steel plate materials separately.
3. Loosening Torque
3.1 Empirical Formula for Loosening Torque
The equations obtained for the two sections differentiated in the preceding chapter two can be structured as shown in Eq (3) below.
That is, coefficients k1, k2_AL, and k2_ST were used to represent all the constant physical qualities, form, or area-related values that depend on the kind of material. The yield strength, which exhibited variations even within the same steel plate or aluminum material, was applied separately. Here, the value of k1 is determined by the steel plate and aluminum components. It should be selectively applied based on whether the upper plate is steel plate (k1_ST) or aluminum (k1_AL). Deriving these four coefficients theoretically or analytically is challenging. Therefore, this study measures the actual loosening torque through experiments using various combinations and uses the least square method to calculate the difference between the experimental value and the expected value of the four coefficients, as shown in Eq (4).
To verify the precision of the least square method of Eq (4), we performed experiments on 18 different combinations (N) for each screw type. The experimental conditions and results for the M4 screw are presented in Table 1, while the experimental conditions and results for the M5 screw are summarized in Table 2. The tightening torque is a standardized process parameter. In this study, three values of 6 Nmm, 7 Nmm, and 8 Nmm were used. However, in situations where normal jointing was impossible based on the combination, experiments were carried out using the adjusted tightening torque. The material-specific yield strengths were as follows: A365.0 (100 MPa), A6063S (175 MPa), A6N01S (205 MPa), SGACUD (167 MPa), SGARC340 (186 MPa), SGARC440 (265 MPa), SGAFC-590DP (325 MPa), and SGAFC780DP (400 MPa).
Experimental conditions and results for M4 screw
Experimental conditions and results for M5 screw
The values of the coefficients to minimize the sum of squared errors, obtained by substituting the expected Loosening torque from Eq (3) into Eq (4), are k1_AL, k1_ST, k2_AL, and k2_ST, which may be found in Table 3. These coefficients were substituted into Eq (3) to utilize as the final predictive formula.
Coefficients for predictive equation derived from experiments
Fig. 4 presents a graph that compares the expected values, which are obtained by reflecting the determined coefficients, with the experimental values.
Regression results of prediction equation for loosening torque using experimental data
3.2 Analysis of Results
As shown in Fig. 4, the majority of the specimens exhibit a strong adherence to the trend, with minimal deviation between the expected and experimental values. The M4 screw had a Root Mean Square Error (RMSE) magnitude of 0.711 and an R2 (R squared) value of 0.778. The M5 screw exhibited a somewhat high absolute value of the loosening torque, resulting in an RMSE value of 1.151 and an R2 value of 0.719. R2 ranges from 0 and 1; a higher value indicates a stronger expressivity of the chosen independent variable on the dependent variable. While there is no absolute standard, the fact that both types of screws have values more than 0.5 indicates that this study has identified significant influencing factors that can accurately predict the loosening torque for the FDS joining process. Furthermore, the relational formula between these factors was well established. The following recommendations might be regarded as supplementary improvement strategies for R2.
A representative improvement is the observed deviation in the loosening torque due to experimental factors in certain combinations. In order to minimize experimental deviation, this study focused on specimens that had no gap between the screw head and the upper plate. However, the evaluation results of the nine specimens under the same conditions showed some relatively low values with significant deviations. The average value of five or more specimens, excluding these, was used instead. Given that this phenomenon is more likely to occur in combinations involving aluminum castings, it is crucial to minimize the levels of quality deviation in aluminum castings. Furthermore, it is plausible that some factors incorporated in the predictive formula are yet absent. This study derived a predictive formula that incorporates the yield strength (YP) of each material. However, it is expected that the accuracy of the formula will be enhanced by taking into account measurable process conditions, such as the screw rotation speed (rpm).
Likewise, the predictive formula for the Loosening torque obtained in this study should be interpreted as the highest torque that may be observed when the joint is in an ideal condition. It should not be automatically considered as an aberrant joint condition solely based on its lower value. For instance, because the threads of the actual screw are only formed below a specific location, the assumptions made during the formulation of the predictive formula contain some errors. Nevertheless, when the safety margin is taken into account in the predictive equation for assessing the integrity of the joint using the FDS process, it is deemed to be a commendable benchmark for quality control.
4. Conclusion
This study presents a summary of research findings on inspection methods other than destructive inspection, such as cross-sectional analysis or tensile evaluation, for assessing the quality of the FDS process. The joint section influencing the loosening torque was largely categorized into two groups. The predictive equation for the loosening torque was derived based on the frictional force exerted on the surface of the screw head and the fixation caused by the rotation of the screw. The fundamental equation was obtained by analyzing the simplified physical behaviors of the joint force. The coefficient value that minimizes errors in the predictive equation developed in this study was identified by measuring the actual loosening torque under different experimental settings.
To summarize, the predictive equation developed and validated in this study has an R2 value of 0.7 or above. If it is checked to determine if it does not deviate from the torque that reflects the margin as per the management requirement level, it can serve as a non-destructive inspection method. The inspection method that utilizes the Loosening torque mentioned in this study is insufficient for capturing the entire quality of the FDS process. It should be conducted alongside the appearance check, and if the loosening torque falls below the suitable threshold, an investigation of the underlying reason through a formal destructive test is also required.