The effect of joint forces and torques on speed variation in automobile assembly lines

Abstract

BACKGROUND:

In production industries, physical loads have been observed for employees. The impact of these loads has been investigated in automotive industry assembly lines.

OBJECTIVE:

The purpose of this study was to evaluate the effects of different takt times on the forces and torques in body joints during loading.

METHODS:

Data were collected using an integrated system, including a motion capture system as a hardware subsystem and a tool for physics-based human simulation as a software subsystem.

RESULTS:

The results were compared for a human working for 3 seconds and 5 seconds; for 3 seconds, there was a negative impact on the bottom of the torso (L4), top of the torso (T12), left shoulder and upper arm (Lshoulder) joint forces and torques, but there was a reduction in the joint forces and torques on the right shoulder and upper arm (Rshoulder). Furthermore, the results reveal fluctuations and peak values in all joint forces and torques at the initial intervals due to the variation in speed.

CONCLUSIONS:

The initial acceleration is highly correlated with the risk of musculoskeletal disorders, physical loads and ergonomic problems. This study illustrates the importance of providing appropriate processing times for operators.

Keywords

Physical loads motion capture system (MOCAP)real-time ergonomics evaluation

1 Introduction

There are various body positions during an industrial process, such as sitting, bending, crawling, kneeling and moving. The actions are more frequently performed during the manufacturing processes. As the competition of the manufacturing industry is increasing throughout the world, many companies tend to improve product quality, reduce product development cost and shorten time of design-to-assembly for new products [1]. Assembly processes constitute a majority of the cost of a product [2]. Reducing the assembly takt time may be an appropriate solution for reducing costs. For this reason, the operator must move rapidly, as the process time is interconnected with the speed of the operator. Moreover, the different speeds of operator movements cause different loads on the whole body. Therefore, detecting the occupational risk factors, work position standards and following ergonomic interventions are highly recommended [3]. There have been many previous studies to analyse the effect of loads on joint forces and torques during different processes, especially in the automotive industry. The automotive industry is particularly suitable for a study on ergonomics, as it is characterized by sophisticated production systems and high degrees of automation as well as production processes with a high number of manual tasks [4]. Also, work-related fatigue is common among automobile factory employees [5]. Ergonomics problems were to a great extent considered as potential quality risks already in the manufacturing engineering context [6]. On average, 80% of the medium or high ergonomics load tasks showed quality problems [6]. In the last years, direct measurement tools are developed introducing real-time posture data collection using sensors placed on the operators under analysis [7 –9]. These methods require a complex and cost-intensive hardware setup and a lot of effort to analyse and interpret recorded data in real-time [10]. It is also known that motion capture systems to realize task simulation give good results [11].

The goal of this study was to use a measurement tool to quantify operator joint forces and torques during an assembly process and compare the results of working fast and slow. Accordingly, appropriate operating times and speeds may be iteratively found for the operator.

2 Methods

This study was conducted during the rear axle assembly process of the production line of the automotive company TOFAS. First, a MURI analysis was conducted on a selected production line; this is an ergonomic analysis in which a scoring system is used to assess postures and movements during a process, and a table is referenced to indicate 9 defective movements that may occur while a job is performed. The score from the MURI analysis is used to determine the risk level after the posture and motion analysis is performed [12]. The purpose of this analysis was to reveal the riskiest process performed on the production line. According to the MURI results, red items require action to eliminate ergonomic problems (see App. 1 for detailed results). Two critical processes were identified loading the wheel group onto the rear axle and tightening while holding with the other hand.

As a result of the analyses, we focused on the following related processes, separately: wheel group loading, fixing 4 screws (left & right), fixing brake cables, fixing brackets and positioning cables.

In this way, we were certain that the operators were exposed to the physical loads before the real-time experiment. The integrated system was applied to a real case of a selected rear axle group assembly line in the automotive industry; the operators walked to the wheel group line, picked up the rear wheel group, walked to the rear axle fixture and positioned the wheel group on the rear axle (Fig. 1). The operators’ joints were exposed to various physical loads due to gravity and the mass of the rear wheel. The exposure time of loading influences the joint forces and torques. The directions and axes of the loads on the body segments are illustrated in Fig. 2. F is the sum of the muscle force and joint reaction force. T is the torque produced by the muscle force about the joint centre (joint torque) [13]. We analysed the joints and the effects of different process takt times, which are correlated to the operators’ speeds.

Fig. 1.

(a) Walk to the wheel group line; (b) Pick up the rear wheel group; (c) Walk to the rear axle fixture; (d) Position the wheel group on the rear axle.

Fig. 2.

Representation of free body diagram of a segment.

2.1 Participants

A total of seven healthy subjects, four women and three men, participated in the study and provided informed consent. For women and men, respectively, the mean age was 27 and 34 years, the mean weight was 65 and 75 kg, and the mean height was 1.71 and 1.75 m (Table-1). All subjects were right-handed. Each subject performed ten trials, and two different takt times (3–5 seconds) of data were collected for each trial.

Table 1
Subject Information

Gender Age Weight (kg) Height (m)

Woman (n = 4)

1 28 50 1.70

2 21 65 1.74

3 24 75 1.65

4 35 70 1.75

Mean 27 65 1.71

Man (n = 3)

1 47 90 1.80

2 30 67 1.77

3 25 68 1.68

Mean 34 75 1.75

Gender	Age	Weight (kg)	Height (m)
Woman (n = 4)
1	28	50	1.70
2	21	65	1.74
3	24	75	1.65
4	35	70	1.75
Mean	27	65	1.71
Man (n = 3)
1	47	90	1.80
2	30	67	1.77
3	25	68	1.68
Mean	34	75	1.75

2.2 Work sample

There were two important objects used by the operators working in the assembly process. One object was the rear wheel (Fig. 3a). There were 5 types of rear wheel groups, varying in diameter (250 to 270 mm) and weight (9.9–12.2 kg). The 12-kg wheel was used in this study. The weight of the wheel has a direct load on the operators’ joints due to lifting and carrying to another area to assemble. The other object was the rear axle (Fig. 3b). There are also 5 types for this group, having different brackets and minor weight differences.

Fig. 3.

(a) Rear wheel; (b) Rear axle.

2.3 Experimental design

2.3.1 Hardware subsystem for data collection

We used a measurement tool for data collection. The primary data for full-body movements were collected using a motion capture system called Gypsy Gyro-18 (Animazoo, UK), whose name is based on the garment containing potentiometers and gyros. The garment also contained straps, triangle clips and adjustment nuts, which made it fully compatible with the body of the operator. Furthermore, the garment was suitable for all body types and did not restrict the operators’ movements (Fig. 1).

First, we created an existing skeleton structure by marking fixed points on the body with a program called Autocal (Fig. 4a). The operator then put on the Gypsy suit (Fig. 4b). The operator’s movements were recorded as soon as the process began. During data transfer, another program called Cobraview simultaneously converted the actual image of the examined process to a simulated image (Fig. 4c).

Fig. 4.

(a) Joint points (b) Gypsy on the operator (c) Cobraview model images. (d) Screen display of a simulation.

The Gypsy system includes 13 sensors, which detect the joint movements. The unit transmits the data to a personal computer using a connection cable or receiver for real-time processing. To achieve this transfer in the most accurate manner, all connections must be handled with care (App. 2). The system is completed with the introduction of a useful software tool developed for the real-time collection of activities information, such as kind of task, time and methods to execute the activity [10].

2.3.2 Software sub-system for the data processing

First, the full-body biomechanical model was built using a special tool in the hardware system, connecting the collected motion data to the virtual environment where the study was performed.

Secondly, the collected data from Gypsy are converted into numerical values as torque and force. Therefore, we used another program, Digital Biomechanics 1.0, which is physics-based virtual prototyping software was used to develop a simulation of human model (Fig. 4d). Unlike previous human simulators, digital biomechanics uses control systems to regulate behaviour as the simulated humans perform tasks and respond to events in their environment [14]. Digital Biomechanics has 3 modes in which the data are processed step by step; accordingly, we can configure our simulation, run the simulation based on chosen inputs and review the results using graphics and plots. Depending on time, the simulation can be run for the amount of time shown in the run box (Fig. 5). The run time is multiplied by takt time values in each trial so the simulation generated the same number of samples as the multiplication; this is an advantage that supports data reliability. The software subsystem is integrated with the hardware subsystem and follows the conceptual logical scheme shown in Fig. 6–7.

Fig. 5.

Digital Biomechanics Run Box and Command – Line.

Fig. 6.

Logical scheme of the integrated real-time system (Software Sub-System).

Fig. 7.

Logical scheme of the integrated real-time system (Software Sub-System).

3 Results

Full automation with advanced technologies in production processes is desirable, especially in assembly lines. However, there is still a need for humans in production processes, and the physical loads and risks are therefore inevitable. The mean force and torque values for 3 and 5 seconds are shown in Fig. 8 and Fig 9; the plot diagrams illustrate the time-dependent behaviours of the axial joint forces and torques on the bottom of the torso (L4), top of the torso (T12), left shoulder and upper arm (Lshoulder), right shoulder and upper arm (Rshoulder), separately. There were notable peaks when the physical loads began.

Fig. 8.

Force diagrams for 3 and 5 seconds (mean).

Fig. 9.

Torque diagrams for 3 and 5 seconds (mean).

The forces and torques remaining at zero indicated a positive direction of the axial coordinates with peak values in the vertical direction, and vice versa. There curves were different for the joint forces and torques occurring in 3 and 5 seconds. There were noticeable initial peaks in the curves (Fig. 8), as the operator accelerated him/herself during the shortened process time. Hence, depending on the initial acceleration, there were fluctuations in the joint forces and torques at the beginning of the graphs. Table 2 lists the effect of speed on the mean forces and torques for the L4, T12, Lshoulder and Rshoulder body segments for 3 and 5 seconds.

Table 2

Peak values of mean force and torque variables for working in 3 (sec.)-fast and 5 (sec.)-slow (n = 7)

Variable	Joint
	L4	T12	Lshoulder	Rshoulder
Mean Force (N.)
3 (sec.)	3048.76	2737.21	869.77	1978.02
5 (sec.)	2715.59	1880.88	676.90	2610.30
Mean Torque (N.m)
3 (sec.)	951.98	933.17	204.49	238.34
5 (sec.)	447.12	208.36	90.06	340.65

4 Discussions

In production industries, physical loads are observed on employees because of the body movements and activities. These activities expose workers to ergonomic risk factors such as awkward postures, frequent heavy lifting, repetitive motions and hand/arm and whole body vibration [15].

Motion capture is a powerful tool used in a large range of applications towards human movement analysis. Although it is a well-established technique, its main limitation is the lack of dynamic information such as forces and torques during the motion capture [16]. Therefore, the technique that allow full-body data collection are used mostly for the real-time ergonomic assessment based on the most used methodologies, such as OWAS, OCRA, RULA and others. In general, motion capture systems are used to collect data regarding joint angles, body joints’ flexion-abduction positions and body segment postures. The obtained data are processed by specific real-time tools, developed according to one or more ergonomics evaluation methodologies. To date, assessment of outputs from motion capture systems for work- related injuries in operators has been based primarily on surveys, subjective reports, and observational methods. There is a need to develop quantitative techniques to better understand risk factors and develop preventive interventions [17]. Researchers in most cases cannot determine the necessary figures and numerical values. A big difficulty that prevents many numerical and axial approaches at biomechanical analyses is that researchers do not have equipment to evaluate collected motion data. Such equipments require a complex and cost-intensive hardware set-up and a lot of effort to analyze and interpret recorded data in real-time. The original direction of this study is to determine the forces and torques on the body joints which occur on the operator working at the assembly station as 3 axes. Also, in different scenarios, data to be received at a real production station can be quantified using MOCAP and Simulation tool. Detailed numerical analysis is possible with this tool. So far nearly most of the motion capture systems used in scientific studies are focused on specific areas of the human body such as the ankle, knee and hip joints. However, few studies have focused on all body movements while creating axial values and this paper is one of them. The wear of Gypsy connection points, there were many problems in data transfer. Problems had been examined with root cause analysis. When there was a problem with the parts, it was provided and repaired. Numerous trials and observations were made on the healthy operation of the measurement tools. Furthermore, thanks to the permits of the company authorities, it was possible to collect real-time data in the selected production line. Manual material handling (MMH) task is the most common cause of work-related musculoskeletal disorders (MSDs) [18]. In assembly of the heavy vehicles, a large part of the total load on the musculoskeletal system is associated with handling and the operation of powered tools [19]. This study reports estimates for the joint forces and torques in different process times for different speed levels.

In the literature, there are different studies on body joint loads. W.P. Neumann et al. [20] conducted a case-control study for reporting low back pain; a posture and load sampling approach was implemented to measure the physical exposure. Fu Guo et al. [5] found that the fatigue was significantly correlated with work-related factors, especially working environment and monotony. Mikael Forsman et al. [19] aimed to elucidate the load patterns resulting from chosen joint position, tool type, and joint characteristics in a bus chassis assembly plant, where the production system was based largely on manual operations on stationary objects. David M. Andrews et al. [21] provided maximal acceptable peak forces and impulses for females performing rubber hose installations in the automotive manufacturing. Additionally, studies have been conducted in different areas outside the automotive industry. Duarte and Orselli [22] found that the water depth and moving velocity mostly influence the internal musculoskeletal loads. Hutchinson et al. [23] investigated the impact on all joints while rising from a chair and found larger dynamic percentages of force and torque for the fast trials compared to the slow trials. In a similar study using dynamic torque calculations, the results showed that increases in chair-rise speed are accompanied by increases in joint torques [24]. Shankar et al. [25] analysed workers from hand screen-printing (HSP) role in textile industries developing countries. This study results inferred that HSP workers were prone to lower back and shoulder pain followed by knees and ankle feet regions. Bulduk et al. [26] aimed to investigate job satisfaction among aircraft baggage handlers and their exposure to work-related musculoskeletal disorder risk factors.

The aim of this study is to quantify the force and torque values on the operator joints with a measuring tool and to reveal the physiological loads that occur on the worker when compared to the fast work, slow work. Obtained values will allow for the interpretation of joint forces and torques for two different time periods for different speed levels.

Finally, the results of the present study showed different forces and torques when performing a process within two distinct takt time periods. We observed a decrease in the joint forces and torques on the Rshoulder; however, there were increases in the L4, T12, and Lshoulder joint forces and torques for all subjects. When we examined the mean joint forces and torques, the peak values increase and decrease in the same direction for each trial (Figs. 8-9). Therefore, the results also support that subject tendency affects the joints differently during the process, which was seen for Rshoulder values for right-handed operators. There were vastly different effects of operator characteristics on the joint forces and torques (i.e., skeleton structure, body mass, speed); however, the only common aspect between the operators was their right–handedness.

Speed levels are critical for operators’ ergonomics. This paper illustrates the importance of providing appropriate processing takt times for operators. With different trials, the suitability of these times can be explored. Also, as an ergonomic recommendation, robots for handling the object if needed can be used for carrying the rear wheel from the rear wheel group assembly line (RWAL) to rear axle assembly line (RAAL) (Fig. 10). Operators can physically interact with this robots and share common workplace during transfer. Robot can undertake loading while human will perform carrying. This situation will improve the ergonomics and efficiency.

Fig. 10.

Suspension assembly line-flow.

5 Conclusions

Employees working in production industries have to deal with a very large amount of perception and interpretation, making critical decisions and using machines in a fast and accurate manner. In particular, as in the study, if a person is working with an object, there is a force applied by the object on the muscle, and this is defined as the load. Exposure to this force creates an overload at the physical joints and it results in fatigue as causing in decreased performance and function of the body component. This triggers loss of cautiousness in the person, loss of perception and deceleration, difficulty in thinking, reduced desire to be productive, loss of productivity in mental and physical work. If sufficient time is not provided for the work done by the employee, this can turn into a risk factor. In fact, doing a job fast does not mean making it right. The momentum that will emerge when an employee accelerates work creates a favourable environment for the emergence of ergonomic problems.

People feel that the fatigue in their bodies increases when they do the same work fastly in daily life. One of the most important results of the study is that to see this feeling statement verified numerically. Because, the same situation is also examined in this study for operators working in assembly line. Assembly is the most common process in production lines. This paper demonstrates the different loads on selected body joints for an assembly process completed in 3 and 5 seconds. Accelerating a process leads to different loads of joint segments in the body. The results of this study confirm that the speed and dominant hand of the operator affect the joint forces and torques.

Future studies will be focused on load handling in different production lines with various objects. This research method will be implemented in different departments of the automobile company, it will enable the system to be widespread and processes can be done in the correct time to increase productivity. The ideal operator time may be found experimentally, implementing variations to provide stability while inducing harmless loads. The technique used here may be applied to left-handed and disabled individuals working in production lines.

Conflict of interest

None to report.

Footnotes

Acknowledgments

The authors would like to thank the support and funding from TOFAS R&D Center.

Ergonomics assessment matrix on working movements in manuel processes (MURI Anaysis)

Connections

References

Wang

, Ng

, Ong

, Nee

AYC

. Assembly planning and evaluation in an augmented reality environment. International Journal of Production Research. 2013;51(23-24):7388–404.

Boothroyd

, Dewhurst

. Product Design for Assembly. New York: McGraw-Hill; 1989.

Samaei

, Tirgar

, Khanjani

, Mostafaee

, Bagheri Hosseinabadi

. Effect of personal risk factors on the prevalence rate of musculoskeletal disorders among workers of an Iranian rubber factory. WORK: A Journal of Prevention, Assessment & Rehabilitation. 2017;57(4):547–53.

Thun

, Lehr

, Bierwirth

. Feel free to feel comfortable—An empirical analysis of ergonomics in the German automotive industry. International Journal of Production Economics. 2011;133:551–61.

Guo

, Wang

, Ning

. Subjective measures of work-related fatigue in automobile factory employees. WORK: A Journal of Prevention. Assessment & Rehabilitation. 2017; Vol. Preprint, 1–8.

Falck

, Ortengren

, Hogberg

. The impact of poor assembly ergonomics on product quality: A cost-benefit analysis in car manufacturing. Human Factors and Ergonomics in Manufacturing & Service Industries. 2010;20(1):24–41.

Bernmark

, Wiktorin

CCPS

. A triaxial accelerome-ter for measuring arm movements. Applied Ergonomics. 2002;33:541–7.

Freivalds

, Kong

, You

, Park

. A comprehensive risk assessment model for work-related musculoskeletal disorders of the upper extremities. In Proceedings of the Human Factors and Ergonomics Society Annual Meeting. 2000;44(31):5–728. SAGE Publications.

Radwin

, Lin

. An analytical method for characterizing repetitive motion and postural stress using spectral-analysis. Ergonomics. 1993;36:379–89.

10.

Battini

, Persona

, Sgarbossa

. Innovative real-time system to integrate ergonomic evaluations into warehouse design and management. Computers & Industrial Engineering. 2014;77:1–10.

11.

, Tian

, Duffy

. Performing ergonomics analyses through virtual interactive design: Validity and reliability assessment. Human Factors and Ergonomics in Manufacturing & Service Industries. 2015;22(3):256–68.

12.

Ohno

. Toyota Production System: Beyond Large-Scale Production, Productivity Press, ISBN 978-0-915299-14-0, 1998.

13.

Available from: http://www.kwon3d.com/theory/jtorque/jtorque.html#fre

14.

Boston Dynamics. Digital biomechanics 1.1.1 User Guide. 2005.

15.

Everett

. Ergonomic Analysis of Construction Tasks for Risk Factors for Overexertion Injuries. Center for Construction Engineering and Management, University of Michigan. 1997; NTIS Publication No. PB97-207179, Ann Arbor, MI.

16.

Latella

, Kuppuswamy

, Nori

. Force and motion capture system based on distributed micro-accelerometers, gyros, force and tactile sensing. Conference Proceedings Paper-Sensors and Applications. 2015.

17.

Dhyani

, Roll

, Gilbertson

, Orlowski

, Anvari

, Li

, Anthony

, Samir

. A pilot study to precisely quantify forces applied by sonographers while scanning: A step toward reducing ergonomic injury. WORK: A Journal of Prevention, Assessment & Rehabilitation. 2017; Preprint: 1–7.

18.

Ramadan

, Alkahtani

. Development of a device to reduce the risk of injury in handling unstable loads. WORK: A Journal of Prevention, Assessment & Rehabilitation. 2017; Preprint: 1–11.

19.

Forsman

, Cyren

, Möller

, Kadefors

, Mathiassen

SE.

. Activity in five muscles during joint securing using pneumatic nutrunners. International Journal of Industrial Ergonomics. 2001;29:21–32.

20.

Neumann

, Wells

, Norman

, Frank

, Shannon

Kerr

. A posture and load sampling approach to determining low-back pain risk in occupational settings. International Journal of Industrial Ergonomics. 2001;27:65–77.

21.

Andrews

, Potvin

, Calder

, Cort

, Agnew

, Stephens

. Acceptable peak forces and impulses during manual hose insertions in the automobile industry. International Journal of Industrial Ergonomics. 2008;38: 193–201.

22.

Orselli

MIV

, Duarte

. Joint forces and torques when walking in shallow water. Journal of Biomechanics. 2011;44:1170–5.

23.

Hutchinson

, Riley

, Krebs

. A Dynamic analysis of the joint forces and torques during rising from a chair. IEEE Transactions on Rehabilitation Engineering. 1994;2(2):49–56.

24.

Pai

, Rogers

. Speed variation and resultant joint torques during sit-to stand. Archives of Physical Medicine and Rehabilitation. 1991;72:881–5.

25.

Shankar

, Naveen Kumar

, Mohankumar

, Jayaraman

. Prevalence of work-related musculoskeletal injuries among South Indian hand screen-printing workers. WORK: A Journal of Prevention, Assessment & Rehabilitation. 2017; Vol. Preprint, 1–10.

26.

Bulduk

, Bulduk

, Giiler

. Job satisfaction among aircraft baggage handlers and their exposure to risk factors for work-related musculoskeletal disorders: A case study. WORK: A Journal of Preven tion, Assessment & Rehabilitation. 2017;56(2):301–8.