News
Final POM2 workshop – 25th March 2014 – Library Campus Arenberg
Innovative solutions for predictive maintenance and performance optimization in industry
Participation free of charge, registration is mandatory
Project info
Project type: IWT-SBO project
Duration: July 1st, 2011 – March 31th, 2014
Project coordinator: FMTC
Consortium: FMTC – CIB (KUL) – ETRO (VUB) – SCD (KUL) – MeBioS (KUL) – DTAI (KUL) – PMA (KUL) – DNI (Lessius)
Subcontractor: Verhaert
IWT project number: 100031
Keywords
Maintenance, Predictive Maintenance (PdM), machine diagnosis, machine prognosis, classification, data mining, industrial process optimization, run-time process optimization
Conference papers
Major results
Methodology for optimization of industrial processes
The methdology developed in POM2 relies on the overall equipment efficiency (OEE) as a measure of the performance of industrial production lines but it takes also the cost aspects into account.It is typically assumed that the higher the OEE the better and companies try to reach an as high as possible OEE for their assets. However, there is an optimum trade-off between the cost for increasing OEE and the profit increase. Our methodology takes this return on investment into account and also shows how it can be achieved.
For more details see the following presentation: POM2 optimization methodology
USE CASE: packaging machine.
The methodology has been successfully applied to several industrial cases, for instance a packaging machine. In this case the machine has to be periodically cleaned otherwise bad products are manufactured. These products will be rejected by the quality control and leads to a decrease in profitability. However cleaning the machine requires a production stop which has also a negative impact on asset efficiency and ultimately on profitability of the production line. Cleaning moments are at present decided by the machine operator and are highly dependent on the experience of the operator. Using our methodology we have determined the optimal moment for performing a clean such as the profit per unit of product is maximized. An 11% increase in profit can be obtained in this way.
(More: MAINTENANCE COST VERSUS PRODUCT QUALITY WCEAM_2011)
USE CASE: steel wire processing machine.
The methodology was further applied to a steel wire processing machine. During production, the machine can reach a critical temperature. When this happens, the normal procedure is to stop the machine and wait until the temperature has sufficiently decreased. We have developed a model on how the machine heats up and use it to adjust the machine parameters such that the critical temperature is not reached. This control procedure has lead to production gains of 17%.
(More: Prognostics_for_optimal_maintenance_COMADEM_2013)
(Dutch version: Productiesnelheid optimaliseren via data van een conditioning monitoringsysteemle)
Predictive maintenance of industrial machinery
FMTC and DTAI has worked on the predictive maintenance of an industrial centrifuge. Due to the process the centrifuge becomes unbalanced and this puts a high stress on the bearings of the centrifuge and leads to excessive vibrations and premature degradation or even failure. Consequently the centrifuge needs to be cleaned regularly. It is preferably if the maintenance moments could be planed in advance such that they do not lead to major production disruptions. By modelling the degradation process the level of vibrations can be predicted up to 14 days in advance which can allow a much better planning of the maintenance actions.
(More info: 6- Predicting RUL – Beau)
Charge optimal control for battery operated systems
Batteries have a limited energy storage capacity therefore optimal utilization of battery-powered equipment is desirable. Our researched has shown that by exploiting existing flexibility in machinery operation the battery lifetime can be extended substantially. A second conclusion was that controllers that take into account the specific way the batteries discharge perform better than controllers that are optimized for energy efficiency.
The use case considered in the project has been that of a robot arm operated on batteries. The battery lifetime has been extended with 50% when using an energy or charge optimal controller with respect to the case of a time optimal controller.
(More info: 9- Charge-optimal control of battery-powered equipment)
(Dutch version: Techniline – Ladingsoptimale aansturing robot)
Run-time optimization of industrial machines
FMTC and KUL-MEBIOS has developed a run-time methodology for optimizing operation of industrial machines. The method is enhancement and refinement of the Evolution Operation method. The main advantage of the method is that processes can be optimized without stopping the normal machine operation, therefore optimization does not causes any production losses. The method searches for the combination of the machine settings that will optimize a predefined cost function. The settings are changed slightly and the cost function is estimated for each combination. The algorithm tries to find the best combination of the settings in the shortest possible time. Applied to the use case of the badminton robot energy gains of 5% in average have been obtained on top of an already energy optimal control.
(More info: 8- EVOP for workshop v10)
(youtube video: https://www.youtube.com/watch?v=OUe5KgczJyc&list=UUaSYdLwkjeviuQJTHEkqmXQ)
Cheap MEMS accelerometers compete with expansive piezoelectric accelerometers for detection of bearings faults / failures
Cost effective monitoring systems are essential for implementation of the predictive maintenance programs through continuous monitoring. We have shown that low cost versions of accelerometers can successfully compete with the high end versions, for monitoring the health of rotating industrial machinery.
(More: Cheap MEMS accelerometers compete with expansive piezoelectric accelerometers, 3- pom_final_workshop_cm_v4)
Bearing condition monitoring using cheap temperature sensors: a cost-effective solution to industry
The biggest hurdle in the widespread adoption of Predictive Maintenance (PdM) programs by the industry is the cost of sensors needed for reliable PdM. In this context, cost-effective Condition Monitoring (CM) systems are needed. We have shown that in certain circumstances they can act as a cost effective way for the online monitoring of industrial machine.
(More: Bearing condition monitoring using cheap temperature sensors)
Continuous condition monitoring: A way robustly predicting catastrophic failures of your machine
The advent of cost effective sensor for condition monitoring opens new possibilities for performing continuous machine monitoring. The major advantage of this technique with respect to periodic inspection is that it can assess the condition of bearings with much higher accuracy. While periodic inspection can lead to false conclusions, permanent monitoring can accurately predict bearing lifetime and avoid catastrophic bearing failures.
(More: Continuous condition monitoring A way robustly predicting catastrophic failures of your machine)