Ph. D. Project
Title:
Methodology for developing the remaining potential of manufacturing systems from (prognostic-decision support) coupling for its exploitation in
production
Dates:
2020/02/25 - 2024/05/31
Student:
Supervisor(s):
Description:
As part of the development of its predictive maintenance activities, Renault's Maintenance Engineering Department follows a PHM (Prognostics
and Health Management) approach. The purpose of such an approach is centred on the manufacturing system and the product it produces (impact
of the machine state on the quality of the product). Such approach is structured in 5 key steps: (i) the formalization of knowledge of the means of
production for the PHM, (ii) the elaboration of their health check-up, (iii) prognosis, (iv) decision support and (v) capitalisation. The first 2 steps
have been addressed in a first thesis work and this new thesis work will address the next two steps related to prognosis and decision support.
Prognosis is used to assess future health status, which supports proactive decision-making. The coupling between these two processes brings out
the notion of remaining potential as the residual capacity of the system to accomplish its mission(s).
Beyond the Renault context, the thesis is based on an abductive research protocol, based on the generalization of the problems attacked in order to
be applicable to all industrial systems (and not only to solve the case of RENAULT production machines). This work is part of the IFAC, IEEE,
CIRP and PHM society communities (in which CRAN and RENAULT are strongly involved) through, for example, the work of (Voisin et al.,
2010), (Nguyen, Do and Grall, 2015) and (Rozas et al., 2018). However, these studies only address the prognosis in its coupling with decision
support through a vision centred on the components and rarely the system as a whole. Thus the prognosis provides a RUL (Remaining Useful
Life) which is exploited by the decision support in a sequential and linear way. The RUL is attached to "the amount of time remaining during
which the component or system can perform the function defined by its design specifications" (Zio, 2013). This vision, although important and
necessary, shows its limits in operation, because the management of the missions of an equipment depends on its ability to produce in conformity,
within a given time frame and on a defined time scale. Thus, for the same RUL, two machines will not necessarily have the same capacity. This
vision is therefore not sufficient in relation to the Renault context since it does not address product quality aspects, process characteristics, etc. The
scientific added value of the thesis is thus to develop contributions to take into consideration a holistic vision of the production system (Colledani
et al., 2014; Voisin et al., 2018) and by adapting on the one hand the concept of potential still mainly used in aeronautics to the manufacturing
sector and its specificities and on the other hand by formalizing the coupling between prognosis and decision support.
Consequently, the objective of the thesis is to propose a methodology for developing the remaining potential use of a machining means (and more
generally of production in the broad sense) for decision-making at the most appropriate level considering not only the state of the means, but also
the product it produces. This methodology is intended to be generic (portability to all the resources of the group's plants).
The development of this prognostic system raises three major challenges in relation to PHM solutions:
- Identify the relevant indicators to be predicted that support the vision of remaining potential and are necessary to support potentially multi-trade
decision-making (e. g. driving, maintenance).
- Formalize the needs of decision support in relation to business needs. Thus, since several indicators are available with their future evolution, the
use of several decision criteria for a given job must be studied both in the combination of the criteria and in the use of their levels, which may
involve different scenarios, i. e. different maintenance or management actions.
- Define the most relevant prognostic approaches with regard to health indicators and able to support coupling with decision support by integrating
into their calculation the parameters of the various decision support scenarios. Propose a method for selecting the most appropriate algorithms.
The thesis therefore aims to address these three challenges in order to continue the development of a generic PHM engineering process, initiated
by T. Laloix, by addressing prognostic and decision support processes. Within the framework of the thesis, the application case will be of the
machine tool type and located at the Cleon plant (Normandy) on a line producing engine cylinder casings. Two use cases, one based on
maintenance and the other based on production, will be defined from there to demonstrate the added value of this approach.
You will be attached to the Maintenance Department of Renault Production Engineering, located at the Renault Technocentre in Guyancourt.
Travel will be planned to the production sites (particularly the Cleon plant).
References :
Colledani, M. et al. (2014) 'Design and management of manufacturing systems for production quality', CIRP Annals - Manufacturing Technology,
63(2), pp. 773-796.
Nguyen, K. A., Do, P. and Grall, A. (2015) 'Multi-level predictive maintenance for multi-component systems', Reliability Engineering and
System Safety. Elsevier, 144, pp. 83-94.
Rozas, H. et al. (2018) 'An Approach to Prognosis-Decision-Making for Route Calculation of an Electric Vehicle Considering Stochastic Traffic
Information', PHM Society European Conference, 4(1), pp. 1-9.
Voisin, A. et al. (2010) 'Generic prognosis model for proactive maintenance decision support: Application to pre-industrial e-maintenance test
bed', Journal of Intelligent Manufacturing, 21(2), pp. 177-193.
Voisin, A. et al. (2018) 'Predictive Maintenance and part quality control from joint product-process-machine requirements: application to a
machine tool', Procedia Manufacturing, 16, pp. 147-154.
Zio, E. (2013) Diagnostics and Prognostics of Engineering Systems, Diagnostics and Prognostics of Engineering Systems: Methods and
Techniques.
and Health Management) approach. The purpose of such an approach is centred on the manufacturing system and the product it produces (impact
of the machine state on the quality of the product). Such approach is structured in 5 key steps: (i) the formalization of knowledge of the means of
production for the PHM, (ii) the elaboration of their health check-up, (iii) prognosis, (iv) decision support and (v) capitalisation. The first 2 steps
have been addressed in a first thesis work and this new thesis work will address the next two steps related to prognosis and decision support.
Prognosis is used to assess future health status, which supports proactive decision-making. The coupling between these two processes brings out
the notion of remaining potential as the residual capacity of the system to accomplish its mission(s).
Beyond the Renault context, the thesis is based on an abductive research protocol, based on the generalization of the problems attacked in order to
be applicable to all industrial systems (and not only to solve the case of RENAULT production machines). This work is part of the IFAC, IEEE,
CIRP and PHM society communities (in which CRAN and RENAULT are strongly involved) through, for example, the work of (Voisin et al.,
2010), (Nguyen, Do and Grall, 2015) and (Rozas et al., 2018). However, these studies only address the prognosis in its coupling with decision
support through a vision centred on the components and rarely the system as a whole. Thus the prognosis provides a RUL (Remaining Useful
Life) which is exploited by the decision support in a sequential and linear way. The RUL is attached to "the amount of time remaining during
which the component or system can perform the function defined by its design specifications" (Zio, 2013). This vision, although important and
necessary, shows its limits in operation, because the management of the missions of an equipment depends on its ability to produce in conformity,
within a given time frame and on a defined time scale. Thus, for the same RUL, two machines will not necessarily have the same capacity. This
vision is therefore not sufficient in relation to the Renault context since it does not address product quality aspects, process characteristics, etc. The
scientific added value of the thesis is thus to develop contributions to take into consideration a holistic vision of the production system (Colledani
et al., 2014; Voisin et al., 2018) and by adapting on the one hand the concept of potential still mainly used in aeronautics to the manufacturing
sector and its specificities and on the other hand by formalizing the coupling between prognosis and decision support.
Consequently, the objective of the thesis is to propose a methodology for developing the remaining potential use of a machining means (and more
generally of production in the broad sense) for decision-making at the most appropriate level considering not only the state of the means, but also
the product it produces. This methodology is intended to be generic (portability to all the resources of the group's plants).
The development of this prognostic system raises three major challenges in relation to PHM solutions:
- Identify the relevant indicators to be predicted that support the vision of remaining potential and are necessary to support potentially multi-trade
decision-making (e. g. driving, maintenance).
- Formalize the needs of decision support in relation to business needs. Thus, since several indicators are available with their future evolution, the
use of several decision criteria for a given job must be studied both in the combination of the criteria and in the use of their levels, which may
involve different scenarios, i. e. different maintenance or management actions.
- Define the most relevant prognostic approaches with regard to health indicators and able to support coupling with decision support by integrating
into their calculation the parameters of the various decision support scenarios. Propose a method for selecting the most appropriate algorithms.
The thesis therefore aims to address these three challenges in order to continue the development of a generic PHM engineering process, initiated
by T. Laloix, by addressing prognostic and decision support processes. Within the framework of the thesis, the application case will be of the
machine tool type and located at the Cleon plant (Normandy) on a line producing engine cylinder casings. Two use cases, one based on
maintenance and the other based on production, will be defined from there to demonstrate the added value of this approach.
You will be attached to the Maintenance Department of Renault Production Engineering, located at the Renault Technocentre in Guyancourt.
Travel will be planned to the production sites (particularly the Cleon plant).
References :
Colledani, M. et al. (2014) 'Design and management of manufacturing systems for production quality', CIRP Annals - Manufacturing Technology,
63(2), pp. 773-796.
Nguyen, K. A., Do, P. and Grall, A. (2015) 'Multi-level predictive maintenance for multi-component systems', Reliability Engineering and
System Safety. Elsevier, 144, pp. 83-94.
Rozas, H. et al. (2018) 'An Approach to Prognosis-Decision-Making for Route Calculation of an Electric Vehicle Considering Stochastic Traffic
Information', PHM Society European Conference, 4(1), pp. 1-9.
Voisin, A. et al. (2010) 'Generic prognosis model for proactive maintenance decision support: Application to pre-industrial e-maintenance test
bed', Journal of Intelligent Manufacturing, 21(2), pp. 177-193.
Voisin, A. et al. (2018) 'Predictive Maintenance and part quality control from joint product-process-machine requirements: application to a
machine tool', Procedia Manufacturing, 16, pp. 147-154.
Zio, E. (2013) Diagnostics and Prognostics of Engineering Systems, Diagnostics and Prognostics of Engineering Systems: Methods and
Techniques.
Keywords:
PHM, Prognostics, Decision-making, Residual Performance Lifetime, Predictive Maintenance
Department(s):
| Modeling and Control of Industrial Systems |
