Dysfunctioning-Based Approach for Indicator Identification to Support Proactive Maintenance

Proceedings of the 20th World The International Federation of Congress Automatic Control Proceedings of the 20th World Congress The International Federation of Congress Automatic Control Proceedings of the 20th9-14, World Toulouse, France, July 2017 The International Federation of Automatic Control Available online at www.sciencedirect.com Toulouse, France,Federation July 9-14, 2017 The International of Automatic Control Toulouse, France, July 9-14, 2017 Toulouse, France, July 9-14, 2017

ScienceDirect

IFAC PapersOnLine 50-1 (2017) 13704–13709 INDUSTRIAL SYSTEM FUNCTIONING/DYSFUNCTIONING-BASED INDUSTRIAL SYSTEM FUNCTIONING/DYSFUNCTIONING-BASED INDUSTRIAL SYSTEM FUNCTIONING/DYSFUNCTIONING-BASED APPROACH FOR INDICATOR IDENTIFICATION TO SUPPORT INDUSTRIAL SYSTEM FUNCTIONING/DYSFUNCTIONING-BASED APPROACH FOR INDICATOR IDENTIFICATION APPROACH FORPROACTIVE INDICATORMAINTENANCE IDENTIFICATION TO TO SUPPORT SUPPORT APPROACH FORPROACTIVE INDICATORMAINTENANCE IDENTIFICATION TO SUPPORT PROACTIVE MAINTENANCE PROACTIVE MAINTENANCE , A. VOISIN11, S. DEEB11, E. ROMAGNE22, B. IUNG11, F. LORANGE22 T. LALOIX1,2 1,2

T. LALOIX LALOIX1,2,, A. T. A. VOISIN VOISIN11,, S. S. DEEB DEEB111,, E. E. ROMAGNE ROMAGNE222,, B. B. IUNG IUNG111,, F. F. LORANGE LORANGE222 1,2 1,2, A. VOISIN1 , S.Boulevard DEEB , E. , B. IUNG F. LORANGE LALOIX Université deT.Lorraine, CRAN UMR 7039, desROMAGNE aiguillettes, B.P. 70239 ,F-54506 Vandœuvre-lès-Nancy Université 7039, Boulevard Université de de Lorraine, Lorraine, CRAN CRAN UMR UMR 7039, [email protected]) Boulevard des des aiguillettes, aiguillettes, B.P. B.P. 70239 70239 F-54506 F-54506 Vandœuvre-lès-Nancy Vandœuvre-lès-Nancy (e-mail: (e-mail: [email protected]) Université de Lorraine, CRAN2 UMR 7039, Boulevard des105, aiguillettes, B.P. 70239 F-54506 Vandœuvre-lès-Nancy (e-mail: [email protected]) Usine Renault Cléon, BP 76410 Cléon, France 2 2 Usine Renault Cléon, (e-mail: [email protected]) Usine Renault Cléon, BP BP 105, 105, 76410 76410 Cléon, Cléon, France France (e-mail: [email protected]) 2 2 Usine(e-mail: [email protected]) Renault Cléon, BP 105, 76410 Cléon, France (e-mail: [email protected]) (e-mail: [email protected]) Abstract: Failure Mode, Effects and Criticality Analysis (FMECA) is a well-known method used in Abstract: Failure Failure Mode, Mode, Effects Effects and and Criticality Criticality Analysis Analysis (FMECA) is is a well-known well-known method method used in in Abstract: safety analysis, reliability analysis, risk assessment and (FMECA) maintenance aobjectives. Moreover, used in the safety analysis, analysis, reliability analysis, risk assessment and (FMECA) maintenance objectives. Moreover, in the the Abstract: Failure Mode, Effects and Criticality Analysis is a well-known method used in safety reliability analysis, risk assessment and maintenance objectives. Moreover, in manufacturing domain and particularly in the context of the deployment of the factory of the future, its manufacturing domain and particularly particularly in the the context of ofand the maintenance deployment of of the factory factory of of the the future, future, its safety isanalysis, reliability analysis, risk assessment objectives. in the manufacturing domain and in the deployment the its usage not entirely satisfactory in relation tocontext the indicator to be observed. Thus, theMoreover, paper presents a usage is not entirely satisfactory in relation to the indicator to be observed. Thus, the paper presents a manufacturing domain and particularly in the context of the deployment of the factory of the future, its usagemethodology is not entirely satisfactory in relation to the observed. Thus, the paper presents new based on a coupled approach of indicator FMECA to andbeHazard Operability analysis (HAZOP)a new methodology based on aa coupled coupled approach of indicator FMECA to andbeHazard Hazard Operability analysis (HAZOP)a usagemethodology isaim notisentirely satisfactory relation to the observed. Thus, the paper presents new based on approach FMECA and Operability analysis (HAZOP) which to contribute to theindeployment ofof proactive maintenance strategies by clearly identify which aim is to contribute to the deployment of proactive maintenance strategies by clearly identify new methodology based on a coupled approach of FMECA and Hazard Operability analysis (HAZOP) which aim is to contribute to the isdeployment of formalization proactive maintenance by clearly pertinent indicator. This approach based on the of conceptsstrategies of knowledge whichidentify permit pertinent indicator. This approach approach is isdeployment based on on the the formalization of concepts conceptsstrategies of knowledge knowledge whichidentify permit which aim isthe to contribute of formalization proactive maintenance by methodology clearly pertinent indicator. This based of of which permit to constitute first pillarstoofthe proactive maintenance approach. Applicability of this is to constitute constitute the the first first pillars of proactive proactive maintenance approach. Applicability of this this methodology methodology is pertinent This approach is based maintenance on the formalization of Applicability concepts of knowledge which permit to pillars of approach. of is illustratedindicator. on a machining center sub-system. illustrated on the machining center sub-system. to constitute first pillars of proactive maintenance approach. Applicability of this methodology is illustrated on aa machining center sub-system. Keywords: functional analysis, dysfunctional analysis, indicator monitoring, maintenance © 2017, IFAC Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. illustrated on a(International machining center sub-system. Keywords: functional analysis, dysfunctional analysis, indicator monitoring, maintenance Keywords: functional analysis, dysfunctional analysis, indicator monitoring, maintenance Keywords: functional analysis, dysfunctional analysis, indicator monitoring, maintenance    degradation or failure mode, it allows several indicators to be 1. INTRODUCTION degradation or mode, several indicators to  1. INTRODUCTION degradation or failure failure mode, it it allows allows severalby indicators to be bea considered for its monitoring. Moreover, providing 1. INTRODUCTION considered for its monitoring. Moreover, by providing or failure mode, it allows several indicators to beaa Since early 2000, proactive maintenance, in Europe, and degradation 1. INTRODUCTION considered for its monitoring. Moreover, by providing structured approach, it will broaden the use of such strategies Since early 2000, proactive maintenance, in Europe, and structured approach, it broaden the use strategies Since earlyand 2000, maintenance, and considered prognostic healthproactive management (PHM) in in the Europe, US has been for its monitoring. Moreover, providing it will will the use of ofbysuch such strategies in helpingapproach, engineering to broaden develop monitoring solution ona prognostic and health management (PHM) in the US has been Since early 2000, proactive maintenance, in Europe, and structured in helping engineering to develop monitoring solution on structured approach, it will broaden the use of such strategies prognostic (PHM) in the maintenance US has been developed and in health order management to go beyond classical engineering to develop monitoring solution on bothhelping new system and existing ones. Such indicator structure developed in order to go beyond classical prognostic (PHM) in the maintenance US has been in both new system and existing ones. Such indicator structure developed in health order to classical maintenance strategies. and Both rely management on go the beyond use of monitoring parameters in helping engineering to develop monitoring solution on both new system and existing ones. Such indicator structure can be viewed as a health check-up of the system (Abichou et strategies. Both rely on the use of monitoring parameters developed inanticipation order go beyond classical maintenance can be viewed as aaand health check-up of the system (Abichou et both new system existing ones. Such indicator structure strategies. Both rely to on of thethe usefailure of monitoring parameters allowing the of the system. Health al. can be viewed as health check-up of the system (Abichou et 2015). It can be used as i) an alarm for the product/system allowing the anticipation of the failure of the system. Health strategies. Both rely on the use of monitoring parameters al. 2015). It can be used as i) an alarm for the product/system can be viewed as a health check-up of the system (Abichou et allowing the anticipation of the failure of the system. Health monitoring, and prognostics processes support such al. 2015). It can be used as i) an alarm for the product/system user, ii) an appropriate diagnostics tool for failure monitoring, and prognostics processes support such allowing the They anticipation ofthe thedefinition failure ofand the management system. user, ii) an appropriate diagnostics tool for failure 2015). It can be used as i) an alarm for the product/system monitoring, and require prognostics processes support Health such anticipation. of al. user, ii) anandappropriate diagnostics tool for failure mechanism, iii) an input for the product models to anticipation. require the and of monitoring, and failure prognostics processes support mechanism, and iii) an input for the product models to ii)lifeantime appropriate diagnostics tool for failure anticipation. They They require the definition definition and management management of degradation and mode set of indicator relatedsuch to user, mechanism, and iii) an input for the product models to enhance in form of continuous collected data-base degradation and failure set of related to anticipation. or They require mode theasdefinition and management enhance life time in form of continuous collected data-base degradation and failure mode of indicator indicator related of to (Catelani components (sub-)systems aset whole. On the management mechanism, and iii) an input for the product models to enhance life time in form of continuous collected data-base et al. 2015). components or (sub-)systems as whole. On management degradation and failure mode of indicator related to (Catelani et al. 2015). enhance life time in form of continuous collected data-base components (sub-)systems as aaset whole. On the the management side one findor some proposal starting mainly from MIMOSA (Catelani et al. 2015). side one some proposal starting mainly from MIMOSA components (sub-)systems as a whole. On the management demonstrate the interest of such combined approach, side one find find proposal starting from MIMOSA initiative andorsome ISO 13374 standard andmainly further enhanced, for To (Catelani et al. 2015). To demonstrate the of such combined approach, initiative and ISO 13374 standard and further enhanced, for side one find some proposal starting mainly from MIMOSA To demonstrate the ainterest interest combined approach, section 2 will present state ofofthesuch art on the combination of initiative and(Callan ISO 13374 and further enhanced, for instance, by et al. standard 2006). When considering indicator section 2 will present a state of the art on the combination of To demonstrate the interest of such combined approach, instance, by (Callan et al. 2006). When considering indicator initiative and ISO 13374 standard and further enhanced, for section 2 will present a state of the art on the combination of FMECA and HAZOP tools and its limits regarding our instance, by (Callan et al. 2006). When considering indicator design, monitored parameters are usually given in ad hoc FMECA and HAZOP tools and its limits regarding our section 2 will present a state of the art on the combination of design, monitored parameters are usually given in ad hoc instance, by (Callan et al. 2006). When considering indicator FMECA and HAZOP tools and its limits regarding our problematic. Then Section 3 will describe the proposed design, monitored parameters are usually given in ad hoc solution. Only few works focus on the design of the problematic. Then Section 3 will describe the proposed solution. Only few works focus on the design of the FMECA and HAZOP tools and its limits regarding our design, monitored parameters are usually given in ad hoc Then to Section 3 will describe the proposed combined approach face these limits and section 4 its use solution. works Most focusofon thearedesign parametersOnly to be few monitored. them relatedofto the problematic. combined approach to face these limits and section 4 its problematic. Then Section 3 will describe the proposed parameters to monitored. Most them related to solution. Only focusof theare combined to face these limits and section 4 will its use use a case approach study. Finally conclusion and perspective be parameters to be be monitored. Most ofon them aredesign relatedof to the the on use of metrics infew orderworks to determine if some already define on aa case study. Finally conclusion and perspective will be combined approach to face these limits and section 4 its use use of metrics in order to determine if some already define parameters to be monitored. Most of them are related to the on case study. Finally conclusion and perspective will be use of metrics in order for to determine alreadypurpose define sketched. parameters are suitable monitoringiforsome prognostic sketched. on a case study. Finally conclusion and perspective will be parameters are suitable for monitoring or prognostic purpose use of metrics in order to determine if some already define sketched. parameters are suitable for monitoring or root prognostic purpose (Coble & Hines 2009). When considering cause analysis, 2. PAST WORK COMBINING FMECA and HAZOP sketched. (Coble & 2009). When considering cause analysis, parameters are suitable for monitoring or root prognostic purpose (Coble & Hines Hines When considering root (Medina-Oliva et2009). al. 2012) conclude that no cause uniqueanalysis, tool is 2. 2. PAST PAST WORK WORK COMBINING COMBINING FMECA FMECA and and HAZOP HAZOP (Medina-Oliva et al. 2012) conclude that no unique tool (Coble & Hines 2009). When considering root cause analysis, techniques, and decision support based on (Medina-Oliva et al.the 2012) conclude that causal no unique tool is is Numerous 2. PAST WORK COMBINING FMECA andsystems HAZOP able to characterize knowledge about relationship Numerous techniques, and decision support systems based able to characterize the knowledge about causal relationship (Medina-Oliva et al. 2012) conclude that no unique tool is Numerous techniques, and decision support systems based on on the integration of FMECA & HAZOP were proposed in able to characterize the knowledge about causal relationship between degradation. They suggest to combine several tools the integration of FMECA & HAZOP were proposed in Numerous techniques, and decision support systems based on between degradation. They suggest to combine several tools able to characterize the knowledge about causal relationship the integration of FMECA & HAZOP were proposed in different works with the objectives of reliability analysis, between degradation. They suggest to combine several tools in order to capture the several aspects of the required different works with the objectives of reliability analysis, the integration of FMECA & HAZOP were proposed in in order to capture the several aspects of the required between degradation. They suggest to combine several tools worksrisk with the objectives of reliability analysis, safety analysis, assessment and maintenance objective. in order toNevertheless, capture the when several aspects parameters of the required knowledge. considering to be different safety analysis, risk assessment and maintenance objective. different works with the objectives of reliability analysis, knowledge. Nevertheless, when considering parameters to be in order to capture the several aspects of the required knowledge. considering parameters to be safety analysis, risk assessment and maintenance objective. monitored, itNevertheless, requires the when definition of sensors. Such sensor support reliability analysis,and (Medina-Oliva et al. 2010) safety analysis, risk assessment maintenance objective. monitored, it requires the definition of Such knowledge. considering parameters to be To monitored, itNevertheless, requires thetowhen definition of sensors. sensors. Such sensor sensor To support reliability analysis, (Medina-Oliva et cannot always be related the root causes. To support reliability analysis, (Medina-Oliva et al. al. 2010) 2010) proposed a methodology to develop a tool for assessing the cannot be the monitored, it requires theto of sensors. Such sensor proposed cannot always always be related related todefinition the root root causes. causes. a methodology to develop a tool for assessing the To support reliability analysis, (Medina-Oliva et al. 2010) proposed a methodology to develop a tool for assessing the dependability and reliability of an industrial system. The idea Hence, always this paper aims to propose a combined approach to dependability and reliability of an industrial system. The idea cannot be related to the root causes. proposed a methodology to develop a tool for assessing the Hence, this paper aims to propose a combined approach to of an industrial system. The idea is to formalizeand thereliability interactions between an industrial system Hence, paper aims propose atocombined approachThe to dependability ease thethisdefinition of toindicators be monitored. is to formalize the interactions between an industrial system dependability and reliability of an industrial system. The idea ease the definition of indicators to be monitored. The Hence, this paper aims to propose a combined approach to the interactions between an industrial andto formalize the support system (maintenance system) system using ease the approach definitionis of indicators to beandmonitored. The is proposed based on FMECA HAZOP tools. andto formalize the system (maintenance system) using is interactions between industrial system proposed approach based on HAZOP tools. ease definitionis of indicators to beand The the support support system system) using processing andthedata models(maintenance such as an SADT (Structured proposed approach is of based on FMECA FMECA andmonitored. HAZOP The the combination both approaches enables tools. the and processing and data models such as SADT (Structured and the support system (maintenance system) using The combination of both approaches enables the proposed approach is based on FMECA and HAZOP tools. dataTechnique) models such as SADT models (Structured Analysis andand Design and qualitative such The combination of relationship both approaches the processing identification of causal betweenenables root cause, Analysis and Design Technique) and qualitative models such processing and data models such as SADT (Structured identification of causal relationship between root cause, The combination of both approaches enables the Analysis and Design Technique) and qualitative models such identification of causal relationship roota cause, degradation, failure and flow deviation.between Hence, for single as FMECA, HAZOP. First, they used FMECA to model as FMECA, HAZOP. First, they used FMECA to model Analysis and Design Technique) and qualitative models such degradation, failure and flow deviation. Hence, for a single identification of causal relationship between root cause, as FMECA, HAZOP. First, they used FMECA to model degradation, failure and flow deviation. Hence, for a single as FMECA, HAZOP. First, they used FMECA to model degradation, failure and flow deviation. Hence, for a single 1 1 1 11

Copyright © 2017, 2017 IFAC 14268 2405-8963 © IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Copyright © 2017 IFAC 14268 Copyright 2017 responsibility IFAC 14268 Peer review©under of International Federation of Automatic Control. Copyright © 2017 IFAC 14268 10.1016/j.ifacol.2017.08.2544

Proceedings of the 20th IFAC World Congress Toulouse, France, July 9-14, 2017 T. Laloix et al. / IFAC PapersOnLine 50-1 (2017) 13704–13709

failure modes of the functions, failure modes of the components, failure consequences and the criticality of the failure. Second, HAZOP has been used to model flow deviation, causes of flow deviation and failure consequences (impact on the flow). Related to the reliability of robotics, (Maier et al. 1995) described a methodology to identify the derivation of the reliability and make safety analysis by applying methods and technique such as HAZOP, functional analysis, FMECA and FTA (Fault Tree Analysis). HAZOP allow them to focus on specific aspect that can have catastrophic consequences. Then, the identification of particular top events, identified by FMECA, together with the creation of functional system reliability models, enabled to carry out a fault tree analysis and to calculate their probabilities. This proposition is limited to a simple sequential application of methods lacking their combination and do not consider monitoring parameters. In safety domain, (Casamirra et al. 2009) presented a safety analysis of a high-pressure storage equipment in hydrogen gas refueling station by integrating FMEA, HAZOP and FTA techniques. The work is intended to assess the refueling station design taken into consideration its safety level, at least from the occurrence frequency point of view. It constitutes a basis for further refined studies which considers the consequences aspects, allowing the assessment of the plant risk. Subsequently, (Daramola et al. 2011) presents a conceptual semantic case-based framework for safety analysis, which facilitates the reuse of HAZOP and FMEA experiences. The aim of the framework is to provide credible tool support for safety analysis processes in order to facilitate early identification of potential system hazards, and enable the interoperable reuse of knowledge for both HAZOP and FMEA activities. Another safety analysis was performed to determine possible accidental events in the storage system used in the liquefied natural gas regasification plant using the integrated application of FMECA and HAZOP methods. The proposed integrating analysis (FHIA) has been designed as a tool for the development of specific criteria for reliability and risk data organization. It aims to provide more recommendations than those typically provided by the application of a single methodology. FHIA provides also an exhaustive list of events or combinations of events that affect the same or different TEs (Top Events). This allows to focus on the critical points of a hazard before making a quantitative assessment of the occurrence probability (Giardina & Morale 2015). These scientific contributions are mainly focused on safety analysis and do not address the topic of indicators definition. In risk assessment domain, (Mechhoud et al. 2016) examined the implementation of a new automated tool dedicated to risk analysis, and assessment of the consequences of the accident scenarios by combining the HAZOP and FMEA methods. The combination enables to localize the problem and its cause in every component. In this method, the authors involve creating two interlinked evaluation models. The first model is evaluated by a criticality matrix extracted from the HAZOP and FMECA analysis. The second is evaluated using the accident scenarios model extracted from the distance effect. To identify the risk of failure levels, (Reitz et al. 2013)

13705

proposes a dysfunctional analysis approach based on the joint application of both methods FMEA and HAZOP. The FMEA method focuses on the failure modes that could affect every function, but does not study the malfunctions caused by possible deviations. To address this shortcoming, the author proposed to complete the FMEA approach by involving the HAZOP method. Motorola developed a hybrid HAZOP and FMEA technique for risk assessment approach. This technique separates the risk factors related to human safety, the environment, facility and product damage and business interruption. It provides a systematic method to thoroughly review failure modes and the effects of failures and deviations on the overall system. As these deviations are identified, the HAZOP nodes and the deviation are logged on the FMEA’s worksheet. HAZOP deviations are noted on the FMEA worksheet as potential failure modes. Each of these deviations are reviewed to determine the consequences and logged onto the worksheet as potential Effects failure. The HAZOP causes are logged also as potential cause mechanisms (Trammell & Davis 2001). Hence it constitutes a combined FMECA/HAZOP approach. While some coupled approach of risk domain consider causal relationship, they do not define monitoring parameters. Finally, and related to maintenance objective, (Gabbar et al. 2003) present a system design and mechanism of improved RCM (reliability-centered maintenance) process that is integrated with a CMMS (computerized maintenance management system). In this work, the failure model is developed using HAZOP and FMECA techniques. HAZOP was used to report the possible deviations, causes, and consequences for equipment, while FMECA defines the different failures with the different levels of details along with the criticality of each failure. The combination of these failure and deviation assessment techniques facilitates the analysis of the root cause. Similarly, for the production of large motors, and in particular to identify predominant failure modes, (Edwin & Syam 2010) proposed to apply simultaneously some methods such as FPD (Failure Progression Diagrams), FMEA and HAZOP to the motor sub-systems. He focuses on advantages of simultaneous failure analysis using these three methods. These proposals focused on classical maintenance needs but do not consider the monitoring needs. (Cocheteux et al. 2009) proposed a coupled FMECA-HAZOP approach for proactive maintenance in order to integrate information for prognostic purpose within an extended FMECA worksheet. The parameters to be prognosticated are described as well as the behavior and the influence factors. However, these parameters are prescribed while we are looking for a way to define them. Finally, this review of combined FMECA and HAZOP approach used in reliability analysis, safety analysis, risk assessment and maintenance domain, highlights some lacks in regard to proactive maintenance strategy. Indeed, these works do not consider the definition of monitoring parameters with causal relationship consideration. To face these limits, this paper proposes a functioning/dysfunctioning-based approach for indicator identification.

14269

Proceedings of the 20th IFAC World Congress 13706 T. Laloix et al. / IFAC PapersOnLine 50-1 (2017) 13704–13709 Toulouse, France, July 9-14, 2017

3. INTEGRATED FMECA-HAZOP FOR MONITORING SPECIFICATION The proposed approach is based on some original works related to functional and dysfunctional analysis found in literature and adapted to face the deployment/development of proactive maintenance strategy in manufacturing domain within an integrated framework. This methodology is based on combining FMECA and HAZOP analysis. As a starting point for dysfunctional analysis, functional knowledge on the system is required (Medina et al., 2010; Cocheteux et al. 2009). Thanks to this knowledge, degradation and failure mode and flow deviation are assessed as well as the causality relationship and failure propagation (Wang et al. 2014). Furthermore, monitoring indicators are identified for each dysfunctional mechanism as (Catelani et al. 2015) do in the case of failure mode. The proposed approach uses these different steps to formulate a coherent methodology (Fig. 1). First, functional knowledge A – Functional analysis

of the system (Fig. 1. A) is identified. It gives knowledge about hierarchical topo-functional structure of the system from the main functions down to the components supporting sub functions. Then, dysfunctional analysis is performed (Fig. 1. B) for identifying the degradation and failure modes (FMECA) and the corresponding flow deviation (HAZOP). Inter related causal relationship are identified in order to consider degradation to failure propagation. A quantification phase of the causality relationship between root causes and failure modes is performed in order to weight a cause on the failure mode against each other. Finally, monitoring indicators are proposed. Dysfunctional analysis phase contribute to the creation of nested multi-level indicators. Thus, proposed methodology is based on the pillars established by (Cocheteux et al. 2009) regarding causal relationship identification, on which we propose to associate on each degradation or deviation a monitoring parameter and to quantify causal relationships, both methods merged into a single sheet as in (Trammell & Davis 2001).

B - Dysfunctional analysis FMECA

Identify subfunctions

Identify sub-function support mechanism Identify Input flow (IF) and output flow (OF) properties of the function

Identify degradation and failure modes Identify flow property deviation

Determine causal relationship at current level of abstraction

Determine multilevel* causal relationship

Determine degradation indicator for maintenance decision making

Identify monitoring parameters

Quantify causal relationship

HAZOP

* From other sub-systems of the same level to the upper sub-system level

Fig. 1. System knowledge formalization methodology 3.1 Functional analysis The functional analysis allows identifying the functional structure of the system. It formalize the interaction between the system function down to each of the sub-systems until the component level (elementary functions). The system functional modeling is based on the principle of decomposition of activity into sub-activities until elementary activities supported by technological mechanism (component). This latter represents the resource used to perform the activity. Regarding manufacturing context, these activities can be grouped by type of process, e.g. part transformation (machining process, stamping process), parts assembly (welding process, bonding process). Activity consumes and produces flows representing (i) items which are transformed by the activity, (ii) energy, resources and activity support, (iii) knowledge allowing how to do activity, (iv) triggers related to the activity (figure 2); (Medina-Oliva et al. 2010), (Santarek & Buseif 1998). These flows are characterized by a quantity of object per unit of time which are defined by spatial, temporal and morphologic attributes (Medina-Oliva et al. 2013). These latter can be more generally considered as functional properties, e.g. torque, speed, flow rate, etc. Functions are linked through the chain of input/output flows. The input/output flow chain will be used in the dysfunctional analysis to propagate the effect of

degradation. Such formalization can be supported by a method such SADT coupled with notions of system theory rules (e.g. flow concept). 3.2 Dysfunctional analysis Input flows (Definition of the properties of flows, definition of flow states and flow causal relationships in the HAZOP)

Flow to trigger the function (informational) Ouput flows (Definition of the properties of flows, definition of flow states and flow causal relationships in the HAZOP)

FUNCTION Causes of mal-functioning (Deviation on the input flow (HAZOP) or deterioration of the support (FMECA))

Support of the function (mecanism representing the support of the function, its failure mode and its effects on input or ouput flows in FMECA)

Consequences of malfunctioning (Deviation on the output flow or on another function input flow (HAZOP) and effect on the support mecanism of the function or of another function (FMECA))

Fig. 2. Knowledge formalization, adapted from (MedinaOliva et al. 2012) The aim of the dysfunctional analysis is to identify the degraded and failure modes of the component and of the flows as well as their causes and consequences on the behavior of the sub-systems and finally on the system performances. Transition between normal to abnormal state of a system can be analyzed in terms of the degradation mechanism of the support the function (figure 2). This degradation is then both spread to the rest of the system

14270

Proceedings of the 20th IFAC World Congress Toulouse, France, July 9-14, 2017 T. Laloix et al. / IFAC PapersOnLine 50-1 (2017) 13704–13709

through the flow exchanging between processes and the propagation of the degradation mechanism (Medina-Oliva et al. 2010). Hence, the impact of component degradation on flow deviation and on downstream functions has to be explicited. Such aim is achieved by combining FMECA and HAZOP in an integrated analysis extending previous work done by (Casamirra et al. 2009), (Maier et al. 1995), (Edwin & Syam 2010). Degradation modes of the mechanisms supporting the functions and failure consequences are identified through the use of FMECA. Flow deviations, their causes and consequences are identified through the use of HAZOP aspect. The first step consists in the identifications of the degradation and failure mode for every bottom level function/component. A classical FMECA approach, including degradation is conducted. In the study, one can include catalectic failure mode, but, since in proactive maintenance, the anticipation ability is based on the progressive propagation of degradation, they will not be considered for monitoring purposes. Considered failure mode comes from degradation modes and corresponds to their final state. The second step is to complete FMECA analysis with the impact of degradation modes on the output flows of the function using HAZOP approach. This lead to add to FMECA worksheet some column related to the flows properties and their deviations. In this paper, the considered properties are defined on more specific flow of energy or material and are only quantitative. The set of generic deviations can be limited to {NO, LESS, MORE} (Cocheteux et al. 2009).

cause another one within the same function/component (e.g. bearing wear leads to bearing vibrations) or between function/component (e.g. bearing vibrations lead to shaft vibrations). (iii) corresponds also to the propagation of failure mode between the abstraction levels. Indeed, the degradation of a bearing, cause the degradation of the motor then the degradation of the performance characterized by an increase of power consumption until the breakdown of the production system. (iv) corresponds to the effect of a degradation mode on the output flow of the function/component. The wear of a bearing cause a torque loss of the output rotation. (iv) corresponds to the impact of a flow deviation on a degradation mode. For instance, voltage surges causes motor aging. 3.4 Dysfunctional causal relationship quantification Each degradation and deviation leading to a failure mode does not have the same weight. The quantification of such a weight of causal relationship has to be evaluated to estimate causes criticality. We thus propose to adapt the standard criticality calculus of FMECA methodology, by considering occurrence (F), dynamic and probability that failure mode occurs once appearance of the cause (G), and level of detectability (D) (Table 2). It produces a general criticality indicator (risk priority number – RPN) for low level interactions: RPN = F*D*G. Table 2. Causality relationship quantification Value

Table 1. Flow properties standard deviation

1 2

3.3 Dysfunctional causal relationship

(i)

is well known and constitute the starting point of failure or degradation mode (Medina et al., 2012).

(ii) corresponds either to chain of degradation mode at the same level of abstraction. A degradation mode can

Frequency criteria Probability that the cause occurs Cause of failure is practically non-existent (1 fault during an average lifespan) Cause of failure appears rarely (less than 1 fault / year )

3

Cause of failure appears occasionally (1 fault / year)

4

Cause of failure appears frequently (1 fault / 3 month)

Value

Dysfunctional causal relationship is a basic notion of the proposed approach. Dysfunctional causal relationship is the chain of cause-effect relation occurring. Causal relationship between sub-systems of the same abstraction level corresponds to link of degradation modes and the relation to the support degradation to the flow deviation. Regarding particularly the causal relationship with upper abstraction level, the degradation modes leading to the upper effect will be considered as the root causes of failure mode at this abstraction level. Indeed, the drift of the sub-system behavior leads to the failure mode impacting the system performance. The vertical and horizontal causal relationship are between: (i) root cause and degradation mode, (ii) degradation mode and degradation mode, (iii) degradation mode and flow deviation, (iv) flow deviation and degradation mode. This chain starts from the root cause goes through the lowest component level up to the system level.

13707

Value 1 2 3 4

Non-detection criteria Probability that the cause will not be detected (2) Total detection of the initial cause or the failure The failure mode or cause are partially detected Low detection of the failure mode or cause No detection of the failure mode or cause

(2) Early w arning sign

Gravity criteria Gravity of the causes leading to failures (1)

1

Minor impact: no notable impact of the cause on the failure

2

Medium impact: low probability of failure with rapid dynamic OR high probability with low dynamic of the failure

3

Major impact: high probability and rapid dynamic of failure

4

Catastrophic impact : immediate failure

5

Safety/quality : failure leading to non conformity or dysfunction for the final customer

(1) Impact on the dynamic and probability that the failure mode occurs from apparition of the cause

3.5 Degradation mode monitoring Causality relationship knowledge guides the monitoring strategy. The monitoring of a particular failure mode can be mainly related to: (a) the degradation mechanism itself, (b) the causes of the degradation and (c) the effects of the degradation mode. Hence, sensing strategy could be focused on: usage degradation for (b), physical mechanism for (a) and flow property measurement for (b) and (c). Moreover, degradation related parameters can be merged to focus on a specific failure mode identification. Monitoring parameters can be both focused on the system performance related to the function and on the degradation related to the component. System performance is characterized by the

14271

Proceedings of the 20th IFAC World Congress 13708 T. Laloix et al. / IFAC PapersOnLine 50-1 (2017) 13704–13709 Toulouse, France, July 9-14, 2017

effectiveness (ratio of results to objectives) and the efficiency (ratio of results to engaged resources). Among monitoring parameters, we identify physic characteristics of the mechanism supporting the function (e.g. temperature, vibrations, acoustic emission), system performance (via output flow properties) and resulting effects on upstream or downstream degradation mechanism or function properties (e.g. torque rise, output reduction, vibrations). Health indicator concept represents the link between the monitoring parameters and the degradation mode assessment. Finality of this approach is to estimate the performance of the system through its both aspects (i.e. performances of the system itself and performance of the product/service its delivering), by means of the aggregation of these indicators (Bouthaina et al., 2015; Laloix et al. 2016). This knowledge is the key of the decision making process. These considerations are not discussed in this paper. 3.6 Worksheet Table 3. FMECA/HAZOP analysis worksheet Element Function

Properties

Failure mode / Deviation

Causes

Local effet

Sub-system effect

D F G RPN

Monitoring parameters

Classical FMECA analysis is extended by HAZOP analysis characteristics: properties and deviation items (Table 3). Horizontal causal relationship is identified through the deviation properties, failure mode, causes and local effect concepts rather than vertical one is highlighted by sub-system effect concept. Then, related monitoring parameters can be associated. 4. APPLICATION CASE A proof of concept (POC) of the proposed methodology has been applied in RENAULT Cléon factory. The application case is a machining center GROB G520. It manufactures engine cylinder block (Fig. 3). POC deals with the electro spindle sub-system in order to identify degradation relationships, monitoring parameters and assess root causes criticality.

Rotate cutting tool function (A3) represents the third subfunction of the machining center. This function is supported by the electro spindle rotating system (Fig. 4). Rotate cutting tool is split into 3 sub-functions which decompose the electro spindle rotating function: Provide rotational movement (A31), Guide rotational movement (A32) and Transmit rotational movement (A33). Clamping information

Rotation speed order

Clamped cutting tool Rotation speed information Rotate cutting tool

Electrical power

Cutting tool rotation

Electrospindle rotating system

Fig. 4. Rotate cutting tool function of the machining center Each of these elementary function is respectively supported by motor, bearings and clamping unit. Regarding especially the Guide rotational movement function, the input flow is rotational movement, characterized by torque and speed properties. The output flow is represented by guided rotational movement, characterized by spatial positioning precision and guiding resistance properties. 4.2 Dysfunctional analysis Thanks to functional analysis, the dysfunctional analysis of mechanism supporting the function is performed (Table 4). Failure modes of the component are identified, as well as the output flow properties. Then, causes and local effects and effects on the sub-system are completed. The causes are linked with the component state and the deviation of the input flows while flow deviations are related to the qualitative or quantitative deviation of a property compared to its nominal value. Table 4. Extract of the FMECA/HAZOP analysis Element

Function

Properties

Failure mode / Deviation

Causes

Local effet

Sub-system Monitoring effect parameter

Lack of lubricant (seal degradation) Bearing vibrations

Bearings

Guide rotational movement

Bearing shock Bearing wear Shaft vibrations

LESS spatial LESS spatial positionning positionning precision of precision the tool

Vibration parameters

LESS spatial LESS spatial positionning positionning precision of precision the tool

Vibration parameters

MORE rotational speed Bearing vibrations

Fig. 3. GROB machining center / Renault engine cylinder block

LESS

Rolling resistance

MORE

Bearing shock Shaft vibrations Unbalanced shaft

4.1. Functional analysis As a major element of machining center, electro spindle system plays central role in the machining process. Its finality is to transmit a required rotation speed to the cutting tool, precisely in space and ensure a required energy to the cutting zone for metal removal (Abele et al. 2010). Electro spindle purpose represents the function: to rotate cutting tool. This function is composed of elementary sub-systems.

Spatial positionning precision

Bearing wear

MORE engine

MORE rolling Engine torque resistance of

Degradation of bearings (bearing vibrations), can lead to a deviation of an output flow (LESS spatial positioning precision) of the elementary function, leading to a deviation of a downstream function (function A33 – Transmit rotational movement), but also impact upstream function (function A31 – Provide rotational movement: MORE engine torque). It impacts also upper abstraction level of the system (function A3 – Rotate cutting tool) by deviation of one of its output flow property (spatial precision of cutting to rotation)

14272

Proceedings of the 20th IFAC World Congress Toulouse, France, July 9-14, 2017 T. Laloix et al. / IFAC PapersOnLine 50-1 (2017) 13704–13709

and finally impacts the machine performance level (LESS product performance). Criticality quantification shows that bearing vibrations is the major cause of the spatial positioning imprecision of the spindle and thus the tool on the workpiece during machining process (Table 5). This phenomenon can lead to a surface roughness or a deviation of required dimension precision of the machine part. It also establishes the set of significant root causes in the occurrence of bearing vibrations are bearing shock, bearing wear and shaft vibrations. The two first ones are intrinsic bearing phenomenon whereas shaft vibrations is provided by downstream function component. A mean to prevent such phenomena is to orient the monitoring toward the shaft root causes which leads to its failure mode “shaft vibrations”. Some kind of different monitoring strategies may lead to develop a proactive approach. Regarding especially bearing application case, two types of monitoring parameters can be identified, (i) failure mode oriented i.e. bearing vibration monitoring, (ii) consequence/effect oriented i.e. engine torque monitoring. Combination of both monitoring can allow a more precise estimation of degradation state of the component. Quantification of the impact of root causes of a failure mode is another element of decision of the monitoring. Table 5. Extract of the Criticality evaluation Properties

Failure mode

Bearing vibrations

Spatial positionning precision

LESS

Rolling resistance

MORE

Causes

D

F

G

RPN

Lack of lubricant (seal degradation)

4

1

3

12

Bearing shock

2

2

4

16

Bearing wear

2

2

4

16

Shaft vibrations

2

2

4

16

MORE rotational speed

4

1

2

8

Bearing vibrations

2

3

3

18

Bearing shock

2

2

3

12

Shaft vibrations

2

2

3

12

Unbalanced shaft

2

2

3

12

Bearing wear

2

2

2

8

6. CONCLUSIONS This paper proposes a methodology based on functional and dysfunctional analysis which lead to identify the most relevant parameters to be monitored by means of degradation causality relationship identification and quantification on a single worksheet. It aims at introducing concepts of system knowledge in the context of performance assessment of manufacturing system. Finally, a proof of concept is realized to prove the applicability of the methodology. As further work, first can consist in defining a formal framework of these concepts of knowledge by means of meta-models (Daramola et al. 2011), secondly in evaluating the cause consequences at system abstraction level in term of system performance in regards to system finality (Khorshidi et al. 2016). REFERENCES Abele, E., Altintas, Y. & Brecher, C., 2010. Machine tool spindle units. CIRP Annals - Manufacturing Technology, 59(2), pp.781–802.

13709

Abichou, B., Voisin, A. & Iung, B., 2015. Choquet integral capacity calculus for health index estimation of multi-level industrial systems. IMA Journal of Management Mathematics, 26(2), pp.205–224. Callan, R., Larder, B., Sandiford, J. & IEEE, 2006. An integrated approach to the development of an intelligent prognostic health management system. 2006 IEEE Aerospace Conference, Vols 1-9, pp.3578–3589. Casamirra, M., Castiglia, F., Giardina, M. & Lombardo, C., 2009. Safety studies of a hydrogen refuelling station: Determination of the occurrence frequency of the accidental scenarios. International Journal of Hydrogen Energy, 34(14), pp.5846–5854. Catelani, M., Ciani, L., Cristaldi, L., Faifer, M., Lazzaroni, M. & Khalil, M., 2015. Toward a new definition of FMECA approach. Conference Record IEEE Instrumentation and Measurement Technology Conference, 2015– July, pp.981–986. Coble, J. & Hines, J.W., 2009. Identifying optimal prognostic parameters from data: a genetic algorithms approach. Proceedings of tlpppkhe Annual Conference of the Prognostics and Health Management Society, pp.1–11. Cocheteux, P., Voisin, A., Levrat, E. & Iung, B., 2009. Methodology for assessing system performance loss within a proactive maintenance framework. IFAC Proceedings Volumes (IFAC-PapersOnline), 13(PART 1), pp.1889–1894. Daramola, O., Stålhane, T., Moser, T. & Biffl, S., 2011. A Conceptual Framework for Semantic Case-based Safety Analysis. 16th IEEE International Conference on Emerging Technologies and Factory Automation, pp.1–8. Edwin, V.K.D. & Syam, S.A., 2010. Reliability assurance of large electric motors in a process plant. 2010 2nd International Conference on Reliability, Safety and Hazard, ICRESH-2010: Risk-Based Technology and Physics-of-Failure Methods, pp.642–647. Gabbar, H.A., Yamashita, H., Suzuki, K. & Shimada, Y., 2003. Computeraided RCM-based plant maintenance management system. Robotics and Computer-Integrated Manufacturing, 19(5), pp.449–458. Giardina, M. & Morale, M., 2015. Safety study of an LNG regasification plant using an FMECA and HAZOP integrated methodology. Journal of Loss Prevention in the Process Industries, 35, pp.35–45. Khorshidi, H.A., Gunawan, I. & Ibrahim, M.Y., 2016. Data-Driven System Reliability and Failure Behavior Modeling Using FMECA. IEEE Transactions on Industrial Informatics, 12(3), pp.1253–1260. Kumar, S., Dolev, E. & Pecht, M., 2010. Parameter selection for health monitoring of electronic products. Microelectronics Reliability, 50(2), pp.161–168. Maier, T., Volta, G. & Wilikens, M., 1995. Reliability of robotics: an overview with identification of specific aspects related to remote handling in fusion machines. Fusion Engineering and Design, 29(C), pp.286–297. Mechhoud, E.-A., Rouainia, M. & Rodriguez, M., 2016. A new tool for risk analysis and assessment in petrochemical plants. Alexandria Engineering Journal, 55(3), pp.2919–2931. Medina-Oliva, G., Iung, B., Barberá, L., Viveros, P. & Ruin, T., 2012. Root cause analysis to identify physical causes. 11th International Probabilistic Safety Assessment and Management Conference and The Annual European Safety and Reliability Conference, PSAM11 - ESREL 2012. Medina-Oliva, G., Weber, P. & Iung, B., 2013. PRM-based patterns for knowledge formalisation of industrial systems to support maintenance strategies assessment. Reliability Engineering and System Safety, 116, pp.38–56. Medina-Oliva, G., Weber, P., Levrat, E. & Iung, B., 2010. Use of probabilistic relational model (PRM) for dependability analysis of complex systems. IFAC Proceedings Volumes (IFAC-PapersOnline), 9(PART 1), pp.501–506. Laloix, T., Iung, B., Voinsin, A. & Romagne, E., 2016. Towards the control of product quality from the process deviation monitoring: Overview and investigation in automotive sector. 3rd IFAC Workshop on Advanced Maintenance Engineering, Service and Technology, IFAC A-MEST 2016 Reitz, A., Levrat, E. & Pétin, J., 2013. Quantitative risk analysis in radiotherapy using Bayesian networks. Annual Conference of the European Safety and Reliability Conference, ESREL 2013, Aug, p.pp.520. Santarek, K. & Buseif, I.M., 1998. Modelling and design of flexible manufacturing systems using SADT and Petri nets tools. Journal of Materials Processing Technology, 76(1–3), pp.212–218. Trammell, S.R. & Davis, B.J., 2001. Using a modified Hazop/FMEA methodology for assessing system risk. Proceedings 2nd International Workshop on Engineering Management for Applied Technology. EMAT 2001, pp.47–53. Wang, Y., Deng, C., Wu, J., Wang, Y. & Xiong, Y., 2014. A corrective maintenance scheme for engineering equipment. Engineering Failure Analysis, 36, pp.269–283.

14273