ANALYSIS OF MAINTENANCE DATA IN TERMS OF POSSIBLE APPLICATION OF SPC. Ing. Marek LEIDOLF, MBA, prof. Ing. Darja NOSKIEVIČOVÁ, CSc.

Size: px
Start display at page:

Download "ANALYSIS OF MAINTENANCE DATA IN TERMS OF POSSIBLE APPLICATION OF SPC. Ing. Marek LEIDOLF, MBA, prof. Ing. Darja NOSKIEVIČOVÁ, CSc."

Transcription

1 ANALYSIS OF MAINTENANCE DATA IN TERMS OF POSSIBLE APPLICATION OF SPC Ing. Marek LEIDOLF, MBA, prof. Ing. Darja NOSKIEVIČOVÁ, CSc. VŠB-Technical University of Ostrava, 17. Listopadu 15/2172, Ostrava, Czech Republic, EU, Abstract Together with increasing competition on the market modern companies tries as much as possible to focus on improving production efficiency, reducing costs and losses and improving quality of final products. From these requirements implies a very close link between quality management systems and maintenance. The purpose of this article is to describe data that are used for maintenance of industrial machines and equipment, and analyze whether these data are suitable for application of SPC. It depicts the most used maintenance systems in the industrial field, what kinds of data are gathered from those systems and how these data are processed. Furthermore, it outlines the potential of integration of maintenance systems and SPC that should lead to faster and more efficient reaction to the equipment state shift. Key words: Maintenance, Statistical Process Control, Diagnostics, Quality 1. STATISTICAL PROCES CONTROL The history of using statistical methods for quality improvement is very long. Walter A. Shewhart developed statistical control chart concept, nowadays considered as the beginning of statistical quality control. During World War II statistical methods were used to increase quality of ammunition. In the 1950s, designed experiments (DOE) for product and process improvement were first introduced in US. Japanese progress and lead in this methodology has been the driving force behind the development of DOE into the western world during the 70 s and the 80 s [1]. A control chart is one of the primary and basic techniques of statistical process control. This chart plots the measurements results of quality characteristics in samples taken from the process versus time. It has a center line and two limit lines. The center line (CL) represents where the process should fall, if there are no unusual sources of variability. The limit lines represent control limits (UCL, LCL). If the process is observed beyond these limits, it is considered to be out-of-control The ultimate goal is to identify and separate random changes from changes due to assignable causes. Systematic use of the control chart is an excellent way to reduce variability. The typical control chart is illustrated on Fig. 1. Fig. 1 Control chart [1] The control chart is a fundamental tool used in SPC and it can be applied in many areas of maintenance to improve the quality, such as: equipment availability; equipment downtime rate; number of breakdowns; equipment quality rate.

2 2. MAINTENANCE SYSTEMS OVERVIEW Maintenance systems have been developing since the late 60 s quite rapidly. This follows either from development of new, more complex machines and also from higher requirements on machine reliability, production efficiency, product quality and safety. 2.1 Corrective maintenance Corrective maintenance was in the past the most used system for taking care about equipment. This system does not use any systematic approach; no actions are taken until the equipment failures, as illustrated on Fig. 2 The maintenance cost are very low, on the other hand, the repair cost may be significant. There is now information about the current equipment state, how quick it deteriorates, so the maintenance personal is unable to plan maintenance actions, prepare spare parts, manpower etc. It also does not take into account the root cause of the failure; it just put the machine back to the operational state. That s why this system is no longer used in modern production factories. R(t) probability of operational state t 0, t 1, t 2 time of failure t 1, t 2, t 3 time of completion of repair, time of operational state 2.2 Preventive maintenance Fig. 2 Corrective maintenance [7] Based on the needs described above the corrective maintenance system has developed into the preventive maintenance system. This system is based on the periodical preventive checks (Fig. 3) done by operators or maintenance staff. This preventive plan can be based on the equipment manufacturer recommendation, inhouse staff experience, or on the technical audits done by some external company. The preventive plan is prepared for all machines and divided by periods. The most common are weekly, monthly, half-yearly and yearly check lists or checklists divided by number of equipment run hours, i.e. 2000hrs, 4000hrs, etc. These check lists consist of instruction about what to check, what to clean, what to lubricate, what to replace, etc. They allow scheduling the maintenance activities, preparing spare parts and manpower to realize the inspections. But, there are still some disadvantages of this system. Preventive maintenance still does not reflect the actual state of the equipment. As written above, the system is based on some primary evaluation of the equipment lifetime, but without proper measurements and diagnostics it is unable to reach the optimum. That means that preventive maintenance may be quite costly in some cases, i.e. due to replacement of expensive spare part that is still at its 80% of lifetime.

3 R(t) probability of operational state t 0, t 1, t 2 time of failure t 1, t 2, t 3 time of completion of repair, time of operational state 2.3 Predictive maintenance Fig. 3 Preventive maintenance [7] The first maintenance system that really reflects the actual state of the equipment is predictive or conditionbased maintenance system (CBM). Basically, the equipment is repaired when its required, not according to the preventive plan. To know the exact condition of the equipment, monitoring and checks has to be done. The key idea of this is to get the correct information at the right time. [5] Monitoring and checks come together in a term diagnostics. There are several types of diagnostics methods and in practice it is recommended to use more of them at one time, to obtain the most accurate results. Types of diagnostics according to the type of the analyzed parameters are as follows: Tribo-diagnostics (analysis of lubricants) - it fulfills two main tasks: Monitoring the condition of the lubricant - a lubricant degradation can occur for various reasons (oxidation, penetrating of water or other substances, etc.). Analysis of impurities and wear particles (ferrography) on the base of the material and shape of particles present in the lubricant, an assessment about the place where the machine is damaged is carried on. Thermo-diagnostics (measurements of temperature, thermal imaging) Using local or surface temperature measurements, sites with different temperature can be determined and the cause of the elevated temperature can be deduced (excessive friction, high electrical resistance, etc.). Thermo-diagnostics is widespread in inspections of electrical switch-gears, high voltage lines, hot water pipes, in the steel industry (brick lining of furnaces and chimneys), etc. Examples of the thermo-diagnostics can be seen on Fig. 4. Fig. 4 Thermo diagnostics examples [8]

4 2.3.1 Ultrasonic diagnostics Based on the physical fact that dry friction generates ultrasound. It is also produced when the flow occurs - the leakages due to the leaks and friction in seals, etc. In addition, electrical discharges produce ultrasound as well and therefore this method and instruments based on it are also used by specialists in the field of electrical equipment Vibration diagnostics Vibration signal involves information about the cause of vibration and through its analysis using different methods an emerging or developing fault can be detected. For rotating machines, this is usually the method that covers most possible faults. Basic example on Fig. 5 demonstrates how different cause of vibration is shown in the spectrum. Cause of the vibration can then be distinguished i.e. by using Fast Fourier Transform. Fig. 5 Spectrum of vibrations [5] 3. USE OF SPC FOR ANALYSING PREDICTIVE MAINTENANCE DATA From above mentioned it is obvious that it is possible to collect a lot of different data that describe the machine operational state. The second step would be to analyze the data and reflect these analyses into the preventive maintenance plan. SPC tools and methods should be very useful for these analyses. As an example to show the possibility, data from the vibration diagnostic were chosen, as this method is the most used within all other methods. Using the same method, any other data from different measurements can be analyzed. (temperature, oil viscosity, ultrasonic signals, etc.) as it can be seen in Table 1. Tab. 1 Example of predictive maintenance data suitable for SPC Type of diagnostics Methods Measuring equipment Data Attribute/ Continous data Evaluation method Tribodiagnostics Equipment wear condition monitoring Lubricant degradation monitoring Laboratory methods Amount of wear particles Viscosity Ph Water activity A SPC for attributes Thermodiagnostics Temperature analysis Laser thermometer; Thermo-camera Temperature C SPC for continuous data

5 Ultrasonic diagnostics Signal analysis Ultrasonic detectors Frequency analyzers; Lever analyzers Intensity Pressure Power C SPC for continuous data Vibration diagnostics Signal analysis Multichannel analyzer Vibration transducers Displacement of vibration Velocity Acceleration A/C SPC for attribute or continuous data Basic tool, Shewhart control chart, is the ideal for data analyzing. An overall vibration of roller bearing is chosen as an example, because of its simplicity and common use. Analyzed data should vary randomly around an established mean. Since we are interested in the identification of an earlier deviation of the monitored parameter, therefore only the data related to the normal working stage should be used to establish the control limits. For simplicity, we assume that the observed data over the first stage is normally distributed. To set-up control limits, collected data should be analyzed, looking for data representing normal operational state, or it is also possible to use values recommended by the manufacturer. After having the data, mean and standard deviation are estimated. While having that, collected data are plotted into the control chart, as shown on Fig. 6. This graph also illustrates warning and action limits, which are the indicators for maintenance. After vibrations reach warning limit, physical inspection should be performed, eventually bearings should be lubricated. If the vibrations level reach action limit, it is time for preventive maintenance and the bearing should be replaced. While using this analysis, we can immediately react to the equipment shift, calculate failure probability. While using different tools and analysis, like adaptive moving average or range chart, we can more effectively react on the small process shifts. Fig. 6 Bearing vibration data control chart [3] 4. CONCLUSION The main goal of this article was describing what sort of data are collected from the different types of diagnostic measurements, and if these data are suitable to be analyzed using SPC tools, such as control charts. As has been proven, data from vibrodiagnostics are suitable and usable for Statistical process control. Further research should be conducted to in order to find the best model of control chart that would help us to detect even very small process shifts. Also finding the correlation between data from different measurements methods would be a challenge. ACKNOWLEDGEMENT This paper was elaborated in the frame of the specific research project SP2013/49, which has been solved at the Faculty of Metallurgy and Materials Engineering, VŠB-TU Ostrava with the support of Ministry of Education, Youth and Sports, Czech Republic.

6 REFERENCES [1] MONTGOMERY, D. C., Introduction to Statistical Quality Control. 6th ed. New York: Wiley, 2012, 768 p. [2] PANAGIOTIDOU, S., TAGARAS, G. Statistical Process Control and Condition-Based Maintenance: A Meaningful Relationship through Data Sharing. Production and Operations Management, 2010, Vol. 19, No. 2, pp [3] WANG, W., ZHANG, W. Early Defect identification: Application of Statistical Process Control Methods, Journal of Quality in Maintenance Engineering, 2008, Vol. 14, No.3, pp [4] DELOUX, E., CASTANIER, B., BÉRENGUER, C. Predictive maintenance policy for a gradually deteriorating system subject to stress. Reliability Engineering and System Safety, 2009, Vol. 94, pp [5] BILOŠOVÁ, A., BILOŠ, J. Vibration diagnostics. 1.st ed. Ostrava: VŠB-TU Ostrava, p. [6] FAMFULÍK, J. Teorie údržby. 1st ed. Ostrava: VŠB-TU Ostrava, [7] CASSADY, C.R., BOWDEN, R.O., LIEW, L., POHL, E.A. Combining Preventive Maintenance and Statistical Proces Control: A Preliminary Investigation. IIE Transactions, Vol. 32, 2000, pp [8]