Integrated optimization of maintenance interventions and spare parts requirements for a partially observable multi-component system

Research output: Contribution to conferencePoster

Abstract

We consider a multi-component system in which a condition parameter (e.g., vibration or temperature) is monitored. The outcome of monitoring indicates whether the system is functioning properly, defective, or has failed. However, the condition signal does not reveal which component in the system is defective or has failed. Maintenance is performed by a service provider who has specialized knowledge about the system. The maintenance service provider needs to infer the exact state of the system from the current condition signal and the past data, in order to decide when to visit the customer for maintenance and which spare parts to take along. We model this problem as a partially observable Markov decision process and propose a grid-based solution method to numerically obtain the optimal policy. We analyze the value of having better sensors in the system by considering a case where the maintenance service provider can fully observe the deterioration level of each component. The analysis results show that the positive impact of having full information on the components’ deterioration levels increases as the return cost for the components is getting higher. On the other hand, the positive impact of having full information on the components’ deterioration levels decreases when the ratio of the preventive and corrective maintenance costs is close to either 1 or 0. Additionally, we compare the optimal policy with the corrective and preventive maintenance policies in which the maintenance service provider brings all the components to the customer. The comparison results indicate that the positive impact of employing the optimal policy improves when the return costs for the components increases.
Original languageEnglish
Publication statusPublished - 27 Nov 2018
Event5th Data Science Summit (DSSE 2018) - Muziekgebouw Frits Philips, Eindhoven, Netherlands
Duration: 27 Nov 201827 Nov 2018
Conference number: 5

Other

Other5th Data Science Summit (DSSE 2018)
Abbreviated titleDSSE 2018
CountryNetherlands
CityEindhoven
Period27/11/1827/11/18

Fingerprint

Deterioration
Costs
Preventive maintenance
Monitoring
Sensors
Temperature

Keywords

  • POMDP
  • Condition Based Maintenance
  • Spare Part Optimization

Cite this

@conference{9e460b7bae9e4359b54c425175e090c0,
title = "Integrated optimization of maintenance interventions and spare parts requirements for a partially observable multi-component system",
abstract = "We consider a multi-component system in which a condition parameter (e.g., vibration or temperature) is monitored. The outcome of monitoring indicates whether the system is functioning properly, defective, or has failed. However, the condition signal does not reveal which component in the system is defective or has failed. Maintenance is performed by a service provider who has specialized knowledge about the system. The maintenance service provider needs to infer the exact state of the system from the current condition signal and the past data, in order to decide when to visit the customer for maintenance and which spare parts to take along. We model this problem as a partially observable Markov decision process and propose a grid-based solution method to numerically obtain the optimal policy. We analyze the value of having better sensors in the system by considering a case where the maintenance service provider can fully observe the deterioration level of each component. The analysis results show that the positive impact of having full information on the components’ deterioration levels increases as the return cost for the components is getting higher. On the other hand, the positive impact of having full information on the components’ deterioration levels decreases when the ratio of the preventive and corrective maintenance costs is close to either 1 or 0. Additionally, we compare the optimal policy with the corrective and preventive maintenance policies in which the maintenance service provider brings all the components to the customer. The comparison results indicate that the positive impact of employing the optimal policy improves when the return costs for the components increases.",
keywords = "POMDP, Condition Based Maintenance, Spare Part Optimization",
author = "O. Karabag and {Eruguz - {\cC}olak}, A.S. and R.J.I. Basten",
year = "2018",
month = "11",
day = "27",
language = "English",
note = "5th Data Science Summit (DSSE 2018), DSSE 2018 ; Conference date: 27-11-2018 Through 27-11-2018",

}

Integrated optimization of maintenance interventions and spare parts requirements for a partially observable multi-component system. / Karabag, O.; Eruguz - Çolak, A.S.; Basten, R.J.I.

2018. Poster session presented at 5th Data Science Summit (DSSE 2018), Eindhoven, Netherlands.

Research output: Contribution to conferencePoster

TY - CONF

T1 - Integrated optimization of maintenance interventions and spare parts requirements for a partially observable multi-component system

AU - Karabag, O.

AU - Eruguz - Çolak, A.S.

AU - Basten, R.J.I.

PY - 2018/11/27

Y1 - 2018/11/27

N2 - We consider a multi-component system in which a condition parameter (e.g., vibration or temperature) is monitored. The outcome of monitoring indicates whether the system is functioning properly, defective, or has failed. However, the condition signal does not reveal which component in the system is defective or has failed. Maintenance is performed by a service provider who has specialized knowledge about the system. The maintenance service provider needs to infer the exact state of the system from the current condition signal and the past data, in order to decide when to visit the customer for maintenance and which spare parts to take along. We model this problem as a partially observable Markov decision process and propose a grid-based solution method to numerically obtain the optimal policy. We analyze the value of having better sensors in the system by considering a case where the maintenance service provider can fully observe the deterioration level of each component. The analysis results show that the positive impact of having full information on the components’ deterioration levels increases as the return cost for the components is getting higher. On the other hand, the positive impact of having full information on the components’ deterioration levels decreases when the ratio of the preventive and corrective maintenance costs is close to either 1 or 0. Additionally, we compare the optimal policy with the corrective and preventive maintenance policies in which the maintenance service provider brings all the components to the customer. The comparison results indicate that the positive impact of employing the optimal policy improves when the return costs for the components increases.

AB - We consider a multi-component system in which a condition parameter (e.g., vibration or temperature) is monitored. The outcome of monitoring indicates whether the system is functioning properly, defective, or has failed. However, the condition signal does not reveal which component in the system is defective or has failed. Maintenance is performed by a service provider who has specialized knowledge about the system. The maintenance service provider needs to infer the exact state of the system from the current condition signal and the past data, in order to decide when to visit the customer for maintenance and which spare parts to take along. We model this problem as a partially observable Markov decision process and propose a grid-based solution method to numerically obtain the optimal policy. We analyze the value of having better sensors in the system by considering a case where the maintenance service provider can fully observe the deterioration level of each component. The analysis results show that the positive impact of having full information on the components’ deterioration levels increases as the return cost for the components is getting higher. On the other hand, the positive impact of having full information on the components’ deterioration levels decreases when the ratio of the preventive and corrective maintenance costs is close to either 1 or 0. Additionally, we compare the optimal policy with the corrective and preventive maintenance policies in which the maintenance service provider brings all the components to the customer. The comparison results indicate that the positive impact of employing the optimal policy improves when the return costs for the components increases.

KW - POMDP

KW - Condition Based Maintenance

KW - Spare Part Optimization

M3 - Poster

ER -