Reliability prediction for man-machine production lines.

Karaulova, T. ; Pribytkova, M.

1. Introduction

The reliability of the whole mechanism and machine can be assessed using the well-known methods (e.g. Reliability block diagrams, Fault tree analysis, etc.). The application of these methods requires to assess (or more precisely predict) the reliability of individual items. However the most credible approach to prediction of item reliability is utilizing of internationally accepted reliability databases and reliability prediction methods (Vintr, 2007). Process reliability is a method for identifying problems, which have significant cost reduction opportunities for improvements. It started with the question: "Do I have a reliability problem or a production problem?" Sometimes the problems are identified with a root for maintenance improvements. Very often the problems have roots in the operations area (Barringer, 2000). Strict requirements of reliability on modern devices, long-term error-free operation, durability, reparability, storability, make it necessary to determine as a most important technical parameters of industrial production process (Lendvay, 2002).

Reliability theory is the foundation of reliability engineering. Reliability engineering provides the theoretical and practical tools whereby the probability and capability of parts, components, equipment, products and systems to perform their required functions for desired periods of time without failure, in specified environments and with a desired confidence, can be specified and predicted.

The goal of this paper is to determine the root causes of a production system failures and pinpoint potential areas for reliability improvement and identify appropriate actions to mitigate the effects of those failures.

2. Reliability Management

Reliability of a system or component is not achieved accidentally; it is to be built into the system or component. Reliability is an inherent characteristic of the system, similar to the system's capacity or power rating. It needs to be addressed at every stage of the product or system development including design, manufacturing, testing and maintenance phases (Rao, 1992). In the design phase, proper design methods related to the components, materials, processes, and so on, have to be carefully selected. The objectives at this stage are to ensure that well-established design procedures are applied, known materials and processes are used, and the areas of uncertainly are highlighted for further action.

Failure causes:

Design/Construction

* Design error

* Manufacturing error

Operational

* Machine tool error

* Failure to follow procedures

Human Error

* psychology

* experience

* motivation

External

* Environment

* Suppliers error

Reliability of manufacturing processes can be obtained from daily production data when process failure criteria are established. to maintain quality of products, it is required that defects should be discovered in early stage and it is necessary to minimize the loss.

The reliability of production system may be roughly defined as:

R=f(IC)--f(DE)--f(OP)--f(HF)--f(EXT) (1)

where R--total reliability of the production system: f(IC)--intrinsic capability of the system driven by physical laws and technical potential, generally considered as an ideal upper bound; f(DE) -design errors; f(OP)--the effect of a machine tools failures and wrong technology; f HF)--the effect of human factors, and f(EXT) suppliers faults. In many cases, the human factor parameter f (HF) cannot be separated from external and operational factors.

3. Object of Analysis

Present study considers the process of real retractors' production line existing at AS Norma Estonia.

A retractor is the main part of the seatbelt. It fulfils all functions of safety during a crash. The retractor has a locking mechanism that stops the spool from rotating when the car is involved in a collision (Fig. 1-a).

[FIGURE 1 OMITTED]

The whole process of retractors' assembling is long and consists of many consecutive operations. The layout of the line is illustrated in the (Fig. 1-b). All operations of components assembling are manual and all the testing operations are automated. The line is divided into 13 working stations responsible for certain type of process operation. All the stations are jointed into a conveyor. The retractors of this line are manufactured in 4 main variants and from these a total of 170 sub variants can be assembled. Most parts are general for all retractor models but some parts are special for each model. The retractors are designed with special pallets to make it possible to assemble all models in the same pallet.

The aim was to create model of the production process with follow-on analysis of the production line reliability applying various reliability analysis methods. System analysis starts with development of process model using IDEF0 methodology (Figure 2). On the base of IDEF0 model different methods of reliability analysis have been used: Failure Mode and Effects Analysis (FMEA), Fault Tree Analysis (FTA) and Reliability Block Diagram (RBD). All they are interrelated and help to understand the logical structure of failure modes of a system.

[FIGURE 2 OMITTED]

4. Production Process Failure Mode and Effect Analysis (FMEA)

Process FMEAs are performed on the manufacturing processes. They are conducted through the quality planning phase as an aid during production. The possible failure modes in the manufacturing process, limitations

in equipment, tooling, gauges, operator training, or potential sources of error are highlighted and corrective action taken. FMEA is a live document where all manufacturing operations must be calculated, beginning from the individual components to mounting. Fragment of FMEA analysis is introduced in Figure 3.

When the modes of failure and the effects have been determined, it is necessary to decide which of these to focus upon for resolution. It would be inefficient to work on every failure mode and its potential effect, so a method of prioritization will include: severity of the effect (S); probability of the failure mode occurring (O); and probability of failure detection (D). Each index ranges from 1 (lowest risk) to 10 (highest risk), see Figure 4.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

The overall risk of each failure is Risk Priority Number (RPN). The risk priority number provides a qualitative numerical estimate of the design risk. This number is then used to rank order the various concerns and failure modes associated with a given design as previously identified in the FMEA.

RPN = (S) x (O) x (D) (2)

If the RPN value (2) is high as a result of the severity of effect, it will likely require a product redesign. If the RPN value is high due to probability of occurrence, then it will be necessary to investigate the process for control and capability to reduce this value. If the RPN value is high due to the inability to detect defects, it also may be necessary to redesign the part to reduce, or eliminate, the probability of this defect recurring. One-hundred percent inspection may be considered as a stop-gap measure. Process FMEA requires that we take an analytical view of our procedures.

In Figure 5 is introduced result of FMEA analysis for retractors' production line. In account is taken only failures, which PRN value is higher than 100. There are four types of failures, but most of them have as root causes operators errors and machine tools errors. The basic attention in this research will be given to studying of the reasons of emergence of these failures.

As shows FMEA analysis there are four types of reasons of process faults: human faults, machine tools faults, supplier's faults, and design faults.

On the base of minimal cut set analysis (Fig. 6) is well seen that most faults are done by operators (55%), and equipment faults (44%) are on the second place. In this manuscript we will consider reasons of these faults more explicitly.

Analysis of human-machine systems recognizes that both humans and machine elements can fail, and that human errors can have varying effects on a system. In some cases, human errors result from an individual's action during operation, while others are a consequence of system design or manner of use. Some human errors cause system failure or increase the risk of such failure while others merely create delays in reaching objectives. Thus, as with other system elements, the human elements exert a strong influence on the design and ultimate reliability of all humanmachine systems (MIL-HDBK-338, 1998).

5. Human Factors in Production Process

Human factors refer to environmental, organisational and job factors, and human and individual characteristics, which influence behaviour at work. It takes human as an integral part of plant design and procurement from the earliest stages. Figure 7 (Lowe & Kariuki, 2004) shows that an undesired event is as a result of latent conditions and active errors. organisation, management systems and facility design, when inadequate, are the causes of latent conditions. Latent conditions do not immediately affect the functioning of the system but in combination with other factors like active operator error and/or a local trigger (high temperature, high pressure) they could result to a disaster.

[FIGURE 7 OMITTED]

It is commonly accepted by concerned professionals that human factors (HF) can have a significant impact on the safe and reliable operation of manufacturing lines. Reliability considerations usually start with the system description, specification and performance. Product design, operations and maintenance are taken into account together with data on random failures and more systematic types of failures from specific data collection and analysis campaigns. The relationship between reliability of a production process and human must be understood more explicitly. Operator tasks and actions may be described by using structural analysis and reliability analysis methods.

Much of the system development process depends on quantitative measures. Consequently, for human-machine systems, it is necessary to define a set of parameters that includes the human as well as the hardware. Fortunately, it is possible to construct a set of analogues to conventional reliability, maintainability, and availability measures (Siegel et. al., 1977). Two examples follow.

Human Performance Reliability = No. Human Task Success/No. Human Task Attempt (3)

Human Availability = 1 - Unmanned Station Hours/Total Hours (4)

These parameters can be used in simulations and can be used in probability compounding models as well.

Human errors take many forms and are due to many causes. There are types of human errors (Fig. 8) that are not caused specifically by design, although good design practices can reduce the occurrence of these errors (Reason, 1992; Baybutt, 2005).

[FIGURE 8 OMITTED]

5.1 Man-Machine Interface

Classically, human factors often deal with the man-machine interface (Fig. 9). While this model captures many important human factors a more complete model is required to capture all those of importance in process. It is necessary fully analyse the person-process interface and its impact on system operation. Consequently, in order to model human factor in process we must define completely the person-process interface. This requires that we define a person and process in terms meaningful for performing human factors studies.

[FIGURE 9 OMITTED]

Qualitative allocation is simply the selection of which functions are best performed by the human and which are best performed by the machine. Table 1 identifies the functions at which humans and machines excel. In general, the human is better at handling a variety of different information-processing tasks, adapting to new tasks and environments, devising new procedures, and resolving unexpected contingencies. The greatest limitations of the human are the rate of data processing and the amount of immediate retention.

5.2 Human Reliability Improvement

Human error has an important influence on the quality of production processes. Methods of predicting human errors can be applied with success to production processes. This enables the development of measures by which human error probability can be reduced. ordered in terms of the influence on the reduction of the human error probability, they include:

1. ergonomic measures

2. organizational measures

3. Education and training measures

Human reliability improvement can be achieved by considering the following conditions:

* Simplification: There exists a trend to individually over-simplify a complex situation. Therefore a simplification of tasks and an ergonomic improvement of the technical layout can prevent this tendency.

* Design, clearness, and precision: Due to the tendency to ignore redundant information, often the important information hidden in the redundant environment can be neglected. This can be avoided by a corresponding clear design.

* Indication and identification: Often the feedback information is inadequately semantically identified, since the designer assumes that this is known to the operator by his education.

* Arrangement: The ergonomically correct arrangement of indicators and control elements depends on the task. This is a latent error-prone situation in fixed wiredup control panels especially in unusual situations, which can be fundamentally avoided by software supported systems.

* Reduce the possibilities of confusion.

6. Equipment Reliability Maintenance

There are many causes of machine failure, and their properties are different. Some are depending on the age of machines, and some are purely stochastic. Monitoring of machine operation is not necessarily effective. often bad maintainability is only improved by early design changes. There are various types of maintenance, such as time-based maintenance and condition-based maintenance. It is complicated whether to adopt time-based maintenance or condition-based maintenance.

In order to achieve rational total life cycle management, it is strongly desired to realize a systematic planning method of maintenance operations. Reliability-Centered Maintenance (RCM) is one of the well-established systematic methods for selecting applicable and appropriate maintenance operation types. In RCM, failure consequences and their preventive operations are systematically analysed, and feasible maintenance planning is determined. The rough process of RCM is as follows (Kimura et al., 2002):

(1) Target products or systems of maintenance should be clearly identified, and necessary data should be collected.

(2) All the possible failures and their effect on target products or systems are systematically analysed.

(3) Preventive or corrective maintenance operations are considered. Selection of operations is done based on rational calculation of effectiveness of such operations for achieving required maintenance quality, such as reliability, cost, etc.

The above steps are repeated to realize feasible maintenance planning. Step (2) is the core of the RCM process. It is generally very tedious and time-consuming, and its contents are fundamentally the same as failure Mode and Effect Analysis (fMEA). In the following sections, efficient and practical approach to FMEA based on computer-aided technology is explained.

6.1 Equipment Faults Definition

Reliability and design engineers determine current reliability performance by collecting and analyzing data received from a number of sources, including data of internal testing of the system and customer feedback. In situations where data is not available, but reliability performance needs to be determined, preliminary engineering judgments, mathematical predictions, and consensus using the opinions of experts can be used as a first cut at data values.

All equipment exhibits three distinct failure modes through its operational life cycle:

* The first weeks or months after introduction exhibit failures, which arise from manufacturing defects which fail under stress.

* once the equipment is established in operation, it exhibits random failures, which arise in components for a variety of reasons. random failures are poisson distributed.

* As the equipment reaches the end of its useful life, it begins to exhibit wearout failures. Wearout failures are normally (Gaussian) distributed.

[FIGURE 10 OMITTED]

The 'Bathtub Curve' illustrates failure frequency over the product life cycle (Fig. 10). Random failures arise throughout the useful life of equipment. They are exponentially distributed.

[R.sub.S] (t) = [e.sup.-[lambda]t] = exp(-[lambda]t) (5)

The failure rate [lambda] is a measure of how frequently they arise. High stress [right arrow] high failure rates!

MTBF = Mean Time Between Failures

MTBF = 1/[lambda] [lambda] = 1/MTBF (6)

Wear out failures arise at the end of the useful life of equipment. They are normally distributed.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (7)

* The mean wearout time fi is a measure of the average time at which they arise. The standard deviation of mean wearout time o is a measure of their spread in time.

* The spread in wear out failures depends on the quality of the components and the types of loads they are subjected to over their life cycle.

* Wear out arises in mechanical components and connectors due to cyclic mechanical loads, in semiconductor chips due to cyclic thermal loads and junction diffusion effects.

6.2 Equipment Reliability Modeling

Functional block diagrams are used to develop the basic concepts for the equipment and to evaluate their feasibility. A block diagram provides a clear picture of how the equipment functions and can be used to create a reliability model. Reliability Block Diagram (RBD) models are one of the tools that can be used to create the reliability model of equipment. One of the easiest ways to describe the basic ideas used in the creation of RBD models is to create a simple RBD; for a more detailed description of the diagrams look at the sources listed in the references.

A reliability block diagram (RBD) is a drawing and calculation tool used to model complex systems. The goal of an RBD is to produce a series of images representing portions of a system that is to be analyzed. Once the images are configured properly, and data for these images is provided, calculations can be performed in order to calculate the failure rate, MTBF, reliability, and availability of the system.

Some mathematical relationships between the system reliability and the reliabilities of its components are given next. In the following, RS denotes the reliability of the system, and Ri denotes the reliability of the ith component, where i = 1, 2, ..., n and the system has n components. In addition, in the following relationships, it is also assumed that all components work or fail independently of each other.

System reliability accouting for sequential processes:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (8)

The failure rate [h.sub.S] (t) for the system is given by

[h.sub.S](t) = [n.summation over (i=1)] [h.sub.i](t) (9)

where [h.sub.i] (t) is the failure rate of ith component.

If all the components have an exponentially distributed time to failure, we have

[h.sub.S](t) = [n.summation over (i=1)] [[lambda].sub.i] (10)

And MTBF for system is given by

1/[n.summation over (i=1)] [[lambda].sub.i] (11)

And for the parallel processes

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (12)

If the time to failures for all the components is exponentially distributed with MTBF [theta], then the MTBF for the system is given by

[n.summation over (i=1)] [theta]/i (13)

where [theta] = MTBF for every component (Salvendy, 2001).

A system can contain a series, parallel, or combination of series and parallel connections to make up the network.

[FIGURE 11 OMITTED]

For retractors production line RBD scheme is illustrated in Figure 13. Each module from IDEF0 model (A11, A12, ... A37) is defined by a separate block. All blocks are sequentially connected into whole process and only A31 stage is used twice as being parallel stations.

Failure rate for all retractors production line equipment may be calculated as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (14)

The last RBD analysis allows take into consideration temporal characteristics and made the conclusion about maintenance improvement.

7. Conclusions

There are some basic practices that can be applied to improve reliability. These include:

* Simplicity. Simplification of equipment configuration is one of the basic principles of designing-for-reliability. Added parts or features increase the number of failure modes. A common practice in simplification is referred to as component integration (the use of a single component to perform multiple functions).

* Redundancy. Another reliability improvement practice is to include more than one way to accomplish a function by having certain components or subassemblies in parallel, rather than in series. Beyond a certain point, redundancy may be the only cost-effective way to design reliable equipment.

* Proven Components and Methods. To the extent possible, designers should use components and methods that have been shown to work in similar applications. Using proven components can minimize analyses and testing to verify reliability, thus reducing time and costs of demonstrating reliability of the equipment.

* Failure Detection Techniques. Reliability of equipment can be improved by incorporating failure detection methods or self-healing devices such as periodic maintenance schedules, monitoring procedures, automatic sensing and switching devices.

* Ergonomics or Human Factors Engineering. The equipment design must consider human factors aspects such as the person-machine interface, human reliability, and maintainability.

The reliability improvement process is an iterative process of setting goals, then evaluating (predicting), comparing, and improving those goals. central to the reliability improvement process is data collection and analysis; design improvements; and operations and maintenance procedure improvements.

DOI: 10.2507/daaam.scibook.2009.49

8. Acknowledgement

Hereby we would like to thank the Estonian Ministry of Education and Research for targeted financing scheme SF0142684s05 that enabled us to carry out this work.

9. References

Barringer, P. (2000). Process Reliability and Six-Sigma, National Manufacturing Week Conference "Track For Manufacturing Process and Quality", Chicago

Baybutt, P. (2005). Human Factors in Process Safety and Risks Management: Needs for Models, Tools and Techniques

Kimura, F., Hata, T. & Kobayashi, N. (2002). Reliability-Centered Maintenance Planning based on Computer-Aided FMEA, The 35th CIRP-International Seminar on Manufacturing Systems, Seoul, Korea

Lendvay, M. (2004). Dependability Assurance of Industrial Production Processes, Proceedings: Science in Engineering, Economics and Education, Budapest

Lowe, K. & Kariuki, S.G. (2004). Methods for Incorporating Human Factors during design phase, Loss Prevention and Safety Promotion in the Process Industries, Loss Prevention Prague, pp. 5205-5215

MIL-HDBK-338B (1998). Military Handbook, Electronic Reliability Design Handbook

Rao, S. (1992). Reliability Based Design. ISBN 0-07-051192-6 McGraw-Hill, Inc. New York

Reason, J. (1992) Human Error, Cambridge University Press, http://www.cs.bath.ac.uk/~anneb/humanErr.pdf Accessed: 2009-04-15

Salvendy, G. (2001). Handbook of Industrial Engineering: Technology and Operation Management, A Wiley-Interscience, ISBN 0-471-33057-4

Siegel, A., LaSala, K. & Sontz, C. (1977) Human Reliability Prediction System User's Manual, Naval Sea Systems Command

Vintr, M. (2007). Reliability Assessment for Components of Complex Mechanisms and Machines, 12th IFToMM World Congress, France

This Publication has to be referred as: Karaulova, T[atyana] & Pribytkova, M[arina] (2009). Reliability Prediction for Man-Machine Production Lines, Chapter 49 in DAAAM International Scientific Book 2009, pp. 487-500, B. Katalinic (Ed.), Published by DAAAM International, ISBN 978-3-901509-69-8, ISSN 1726-9687, Vienna, Austria

Authors data: Dr. Karaulova, T[atyana]*; M. Sc. Eng. Pribytkova, M[arina] **, * Tallinn University of Technology, Ehitajate tee 5, 19086, Tallinn, EE, **AS Norma, Laki 14, Tallinn, EE, tatjana.karaulova@ttu.ee, marina.pribotkova@autoliv.com

Tab. 1. Human-machine comparative capabilities

HUMAN SUPERIORITY                MACHINE SUPERIORITY

1. Originality (ability       1. Precise, repetitive operations
   to  arrive at new,
   different problem
   solutions)

2. Reprograinming rapidly     2. Reacting with minimum Lag (in
   (as in acquiring new          microseconds, not milliseconds)
   procedures)

3. Recognizing certain        3. Storing and recalling large
   types of impending            amounts of data
   failures quickly (by
   sensing changes in
   mechanical and
   acoustic vibrations)

4. Detecting signals          4. Being sensitive to stimuli
   (as radar scope               (machines sense energy in
   returns) in high-noise        bands beyond human's sensitivity
   environments                  spectrum)

5. Performing and operating   5. Monitoring funcnons (even under
   though task-overloaded        stress conditions)

6. Providing a logical        6. Exerting large amounts of force
   description of events
   (to amplify, clarify,
   negate other data)

7. Reasoning inductively      7. Reasoning deductively (in
   (in diagnosing a general      identifying a specific item as
   condition from specific       belonging to a larger class)
   symptoms)

8. Handling unexpected           --
   occurrences (as in
   evaluating  alternate
   risks and selecting the
   optimal alternate or
   corrective action)

9. Utilizing equipment           --
   beyond its limits as
   necessary (i.e.
   advantageously using
   equipment factors for
   safety)

From An Introduction to the Assurance of Human Performance in Space
Systems, SP-6506, NASA, 1968.

Fig. 5. Classification of the failures in production process according
to RPN grether than 100 and root causes of these failures

Operator fault    240   140   128   128   126   140   108   108
Machine failure   144   128   120   126   240   192   144
Supplier fault    108   126
Wrong design      108

Fig. 6. Classification of the failures in production process according
to minimal CutSet analysis

Operator fault    55,56%
Machine failure   44,44%