This document summarizes several new and upcoming ISO standards related to condition monitoring and maintenance, including standards on vibration monitoring, oil analysis, thermography, and training/accreditation. It discusses key aspects of the standards, such as guidelines for selecting monitoring techniques based on failure modes and symptoms, and approaches to diagnosis and prognosis outlined in the standards. The standards are designed to provide a common framework for condition monitoring and reduce business risks.
Descripción original:
Título original
An Introduction to ISO Standard Methodology for Condition Monitoring
This document summarizes several new and upcoming ISO standards related to condition monitoring and maintenance, including standards on vibration monitoring, oil analysis, thermography, and training/accreditation. It discusses key aspects of the standards, such as guidelines for selecting monitoring techniques based on failure modes and symptoms, and approaches to diagnosis and prognosis outlined in the standards. The standards are designed to provide a common framework for condition monitoring and reduce business risks.
This document summarizes several new and upcoming ISO standards related to condition monitoring and maintenance, including standards on vibration monitoring, oil analysis, thermography, and training/accreditation. It discusses key aspects of the standards, such as guidelines for selecting monitoring techniques based on failure modes and symptoms, and approaches to diagnosis and prognosis outlined in the standards. The standards are designed to provide a common framework for condition monitoring and reduce business risks.
This paper also outlines developments in the fields of Training and
Accreditation including new and upcoming standards as well as AINDT
direction. Several of the new standards released and under development discussed in this paper are: ISO 17359, Condition monitoring and diagnostics of machines General guidelines ISO 13373, Mechanical vibration and shock - Vibration condition monitoring of machines ISO 13379, Data interpretation and diagnostic techniques which use information and data related to the condition of a machine. ISO 13381, Condition monitoring and diagnostics of machines Prognostics ISO 18436 Part II, Accreditation of organizations and training and certification of personnel - Part II: General requirements for training and certification Vibration Analysis ISO 18436 Sub Parts under development for Oil Analysis and Thermal Imaging Standards are an important part of conducting business and can even establish the legal framework for establishing the validity of goods and services. Standards themselves are a codified form of information that enable the diffusion of technology in a readily assimilated form and can also serve to spread knowledge of requirements for market acceptability and validity. Given that they often contain explicit and detailed technical information they can reduce market uncertainty and business risk. The use of standards form a vital part of the six critical success factors for generating innovative business performance being: Access to knowledge Capacity to absorb and apply new knowledge A sound science and engineering base An intelligent, educated and demanding customer base A competitive business environment, and Access to and willingness to use financial resources Standardisation in Australia has been an ongoing process for many years with varying success in uptake by industry. In the condition monitoring arena there are a new series of standards, both published and under development, focussing on the varying technologies employed, program design and management, diagnostics and prognostics and finally to training and accreditation. These standards are written for the average user and are designed to give the user, both asset owner and service provider, a common platform of understanding and knowledge coupled with training and accreditation guidelines designed to ensure consistent information transfer and knowledge application. All of these standards are designed with business risk minimisation in mind to ensure that both the technologies employed and personnel employing them are utilised and managed in a way that results in a service or product deliverable that is optimised from a cost, efficiency, and risk perspective. 1. General Process Map A general conceptual approach to the overall condition monitoring process is given below in Figure 1 (ISO/CD 13379). An overview of the guideline process from ISO 17359 is shown below in Figure 2. INTRODUCTION This keynote address is intended to brief attendees on the new and upcoming ISO Standards for Condition Monitoring. The paper describes the processes for system design and implementation as well as new concepts in the fields of diagnostics and prognostics. The processes outlined also have direct FMEA links to RCM programs and can be used to implement CM programs directly linked to RCM analyses. An Introduction to ISO Standard Methodology for Condition Monitoring BY LEITH HITCHCOCK, PALL CORPORATION 345 RELIABILITY WORLD 2005 CONFERENCE PROCEEDINGS Figure 1. Conceptual Approach. Figure 2. ISO 17359 General Approach Diagram Extract. Selecting the Right Condition Monitoring Technique (Failure Mode Symptoms Analysis (FMSA)) ISO 13379 FMSA Process The aim of this process is to select monitoring technologies and strategies that maximise the confidence level in the diagnosis and prognosis of any given failure mode. This methodology is designed to assist with the selection of monitoring techniques that will provide the greatest sensitivity to detection and rate of change of a given symptom. Where the confidence in a techniques sensitivity and resulting diagnosis/prognosis accuracy is questionable then the use of additional techniques for further correlation should be recommended. This process is essentially a modification of a FMECA process with a focus on the symptoms produced by each identified failure mode and the subsequent selection of the most appropriate detection and monitoring techniques and strategies. This tool should be used in conjunction with an existing FMECA/RCM analysis that has already identified and ranked possible failure modes. The essential elements of the FMSA process are: listing the components involved listing the possible failure modes for each component listing the effects of each failure mode listing the causes of each failure mode listing the symptoms produced by each failure mode ranking each symptom/monitoring technique combination by a detection rating, severity rating, diagnosis confidence rating and prognosis confidence rating resulting in an overall Monitoring Priority Number (MPN) rating. listing the most appropriate monitoring technique listing the estimated frequency of monitoring listing the most appropriate correlation techniques listing the frequency of monitoring for the correlation techniques When considering monitoring strategies, the following form can also be used: A Failure Mode produces Symptoms, which are best detectable by a Primary Monitoring Technique resulting in a high diagnosis and prognosis confidence when monitored at a given Monitoring Frequency. Increased diagnosis and prognosis confidence can be gained by using Correlation Techniques when monitored at a given Monitoring Frequency. Probability of Detection rating The probability of detection is rated from one to five and is designed to reflect the overall detectability of a failure mode irrespective of the following accuracy of diagnosis or prognosis. This rating is designed to highlight failure modes that: produce symptoms that are detectable but unrepeatable; or, produce symptoms that are undetectable; or, produce symptoms that are not measurable in practice; or, produce symptoms that may be masked by other failure mode symptoms. This is estimated on a scale of 1 - 5, where: 1 means "There is a REMOTE PROBABILITY that this failure mode will be detected" 5 means "It is CERTAIN that this failure mode will be detected" Severity of Failure rating This ranking should reflect any previous FMECA analysis and is designed to rank individual failure modes by risk. This is estimated on a scale of 1 - 4, where: 1 means "Any event which could cause degradation of system performance function(s) resulting in negligible damage to either system or its environment; and no damage to life or limb" 4 means "Any event which could potentially cause the loss of primary system function(s) resulting in significant damage to the system or its environment, and or cause the loss of life or limb " Diagnosis Confidence rating The predicted accuracy of the diagnosis is also rated from one to five. This rating is designed to identify failure modes with: detectable but unrepeatable symptoms; or, unknown symptoms; or, symptoms that are not distinguishable from other failure mode symptoms. This is estimated on a scale of 1 - 5, where: 1 means "There is a REMOTE PROBABILITY of this failure modes diagnosis being accurate" 5 means "It is CERTAIN that this failure modes diagnosis will be accurate Prognosis Confidence rating The predicted accuracy of the prognosis is also rated from one to five. This rating is designed to identify failure modes with: detectable but unrepeatable symptoms; or, symptoms that are not sensitive to changes in degradation; or, unknown failure rates; or, symptoms that are not distinguishable from other failure mode symptoms. This is estimated on a scale of 1 - 5, where: 1 means "There is a REMOTE PROBABILITY of this failure modes prognosis being accurate 5 means "It is CERTAIN that this failure modes prognosis will be accurate" The frequency of monitoring also contributes to the determination of the accuracy of expected prognosis i.e. the greater the frequency of monitoring used the higher the confidence in the expected failure rate and prognosis. Monitoring Priority Number (MPN) The ranking is the multiplication of the four preceding rankings and results in an overall rating of each failure mode. A high MPN value indicates that the nominated technique is the most suitable for the detection, diagnosis, and prognosis of the associated failure mode. It should be noted that a low MPN value does not imply that monitoring is not necessary but rather that a low confidence level for detection, analysis, and prognosis can be expected with the nominated monitoring technique and frequency. The least favourable case is a failure mode with high severity, low detectability, low diagnosis confidence, and low prognosis confidence. The most favourable case is a failure mode with low severity, easily detectable, with known failure modes and associated patterns and therefore high diagnosis and prognosis confidence levels. Continuous re-assessment should be carried out when experience with a new installation has been gained or when a modification has been carried out. 346 RELIABILITY WORLD 2005 CONFERENCE PROCEEDINGS Diagnosis and Prognosis Principles - ISO 13379 and ISO/CD 13381 Diagnosis process is generally triggered by anomaly detection. This detection is carried out by making comparison between the present descriptors of a machine, and reference values (generally called baseline values or data) chosen from experience, from the specifications of the manufacturer, from commissioning tests or computed from statistical data (e.g. long term average). If confidence in the diagnosis and/or prognosis is low, then further verification may be required. If the confidence is high it may be possible to initiate maintenance or corrective action immediately. Diagnosis and prognosis influence factors Influence factors are parameters that effect the deterioration rate of a failure mode such as temperature, viscosity, clearance, load, speed, etc. Each influence factor can be considered a symptom of an existing failure mode and in Figure 3 is represented by the solid lines that connect existing failure mode trends. Influence factors also have effects on the progression and initiation of other either existing or future faults (Figure 3). Improving Diagnosis and/or Prognosis Confidence In order to increase the confidence in the diagnosis/prognosis it may be necessary to carry out an iterative verification process as outlined in Figure 4. This process requires additional analysis of data using either the same technology or a different one or both in order to gain further diagnostic or prognostic information. Setting Alert, Alarm and Trip (Shutdown) Limits The failure definition set point for a parameter/descriptor is the final value that it reaches at the point in time when the item fails. This value is normally determined historically from failure history. The trip set point, however, is the parameter/descriptor value at which the machine is shut down and is normally less than its failure set point. This value is normally determined from standards, manufacturers guidelines and experience. This is the value normally used to define the failed condition, however, this value is not normally reflective of the fully failed condition due to its lower set point required to prevent consequential damage or catastrophic failure. Alert and alarm limits are normally set at a value less than the trip set point. For vibration condition monitoring, information on alarm criteria is contained in ISO/DIS 13373, ISO 10816 and ISO 7919. Usually this value is determined based on the maintenance lead time required, however, such alert values should be cognisant of confidence level of prognosis, future production requirements, spare parts delivery lead times, maintenance planning lead time required, scope of work required to rectify faults, and trend extrapolation and projection. Trending, Extrapolation and Projection The basic difference between trend extrapolation and trend projection is that projection requires the estimation of future data followed by curve fitting whereas extrapolation curve fits only to existing data (Figure 5). Most current curve fitting is extrapolative in nature in that a curve is extrapolated using existing data points This process requires that the behaviour of a set of parameters is understood for a given failure mode set and given conditions. Trend projection requires mathematical equations expressing the rate of change of a variable that describes the deterioration of a given failure mode under given conditions. Simultaneous Data Display Simultaneous Data Display is the simultaneous display of all data within the one system. This concept is paramount to prognostics in that the relationship between parameters can be observed not just the parameters themselves. This is particularly important for different yet possibly interdependent parameters such as bearing temperature and oil viscosity (Figure 6). 347 RELIABILITY WORLD 2005 CONFERENCE PROCEEDINGS Figure 4. Expanded Flow Chart for Diagnosis and Prognosis Confidence Level. Figure 3. Causal Tree Relationships for Diagnosis and Prognosis. One principle of Simultaneous Data Display is that the technique must trend both parameters (unfiltered/unprocessed variables) and descriptors (filtered/processed data) simultaneously. The use of narrow band filters allows spectrums to be divided into discrete elements of which the band amplitude can then be used for simultaneous data display trending. The failure definition set point for each narrow band is the assigned maximum allowable amplitude for each band. This allows, for example, each narrow band amplitude to be plotted against other vibration descriptors, oil analysis results, process parameters and performance values in order to identify and establish relationships between each of them. The difficulty with Simultaneous Data Display presentation is that each variable will have a different unit of measure. This is compounded if the variable can attain the same value more than once in the life of the component (Figure 6). Simultaneous Data Display trending and alarming is also made difficult when the value of the variable in the failed condition is zero e.g. flow or pressure. One key difference between standard simultaneous data display for monitoring and simultaneous data display for prognosis is that for prognosis a common severity axis must be used. For simplicity this can be set to percentage of life usage where 0% life used occurs when the machine has not been operated and 100% life used occurs when the machine is in the failed condition. At this stage data that may approach zero when the machine is in the failed condition, such as flow or pressure, must be inverted to reflect the % Life Usage relationship. It is important when doing simultaneous data display prognosis that for each parameter and/or descriptor being used the following is understood: the start value representing 100% asset life or new condition; the end value representing 0% asset life or failed condition; and how the parameter and/or descriptor behaviour reflects the failure mode development and associated reduction in asset life. Failure Mode Initiation Criteria For future failure modes these influence factors must first be described as initiation criteria in that the same parameter can be both an influence factor for an existing mode and an initiation criteria for a future mode. This introduces the concept of initiation criteria data sets where the root cause of a failure mode can be described in terms of a set of different parameter values that either directly of indirectly measures its occurrence (Figure 7). Direct measurement can take such forms as valve position whereas indirect measures a symptom of a change such as temperature. Prognosis of Failure Mode Initiation Using simultaneous data display techniques the evidence of actions, conditions, and their inter-dependence can be readily observed. Failure mode initiation criteria, expressed as a set of values for all monitored parameters, can therefore be used to trigger alarms indicating that a failure mode has been initiated once this set of values have been achieved or exceeded (Figure 8). Prognosis of failure mode initiation can, therefore, be achieved given a known initiation criteria set and trend projection techniques. The accuracy of the prognosis will depend greatly on whether or not trend extrapolation or projection is used (Figure 8). This method requires a thorough understanding of failure mode initiation criteria, which can be historically and/or statistically generated over time, 348 RELIABILITY WORLD 2005 CONFERENCE PROCEEDINGS Figure 7. Failure Mode Initiation Criteria. Figure 8 - Initiation Prognosis - Projection vs. Extrapolation Figure 5. Extrapolation vs. Projection. Figure 6. Simultaneous data display. sufficient simultaneous data display analysis of a wide set of monitored parameters, thorough understanding of the relationships between and inter- dependence of failure modes, known operating conditions at the time of failure mode initiation, and the behaviour of parameters under varying conditions and influences. Condition Monitoring Program Review The CM process is an on-going process, and techniques which may not have been available, or at the time considered too costly, or too complicated, or unfeasible in some other way, may on review become feasible. Similarly the effectiveness of techniques currently being undertaken in the program should be assessed, and any techniques considered no longer necessary removed. Similarly, warning and alarm values may need revision, due to changes in the machine such as progressive wear, ageing, modification, operation or duty cycle changes. Measured values and baselines may also change due to maintenance work, including component change, adjustment or duty change. In certain cases the baseline may need to re-established following such changes. When maintenance actions have been carried out it is useful to inspect components to confirm that the initial diagnosis or prognosis was correct. Whatever the intent of a Condition Monitoring program its design, implementation, management and review should be carried out in accordance with the latest ISO standards in order to maximise the programs effectiveness and efficiency as well as reduce the programs risk and cost. Recommendations It is highly recommended that asset owners, service providers, and asset insurers become fully conversant with the knowledge and requirements outlined in the above standards in order to minimise the business risk associated with their activities. Asset owners who outsource their condition monitoring should use such standards as the basis for evaluating, selecting and managing service providers and contracts. Asset owners with internal programs should use such standards to audit, validate and manage internal programs to ensure they are efficient, effective and are minimum risk. Service providers should use such standards to design, refine and manage their programs and minimise their business risk. Insurers should use such standards to audit and validate internal and external programs at customer facilities to ensure premium discounts or rebates and claim payouts are warranted. It is foreseen that in the future these standards will form the basis for contracts, litigation and prosecution. It is not unrealistic to expect a future contract for condition monitoring services to read: The condition monitoring program with be designed in accordance with the requirements of ISO 17359, and where applicable ISO 13373 with the program designed by a person certified to a minimum of the second highest level for each technology category applied. The selection of appropriate techniques shall be in accordance with ISO 13379. The vibration analysis component of the program shall be compliant with ISO 13373, ISO 10816, ISO 10817, ISO 7919, ISO 5348, and any other ISO standards relevant to the technique and application. The thermal imaging component of the program shall be compliant with ISO 18434. The lubricant analysis component of the program shall be compliant with ISO 14830. The performance monitoring based component shall be compliant with ISO 13380. Diagnostic processes shall be compliant with ISO 13379 and ISO 13373. Prognostics and all reporting requirements shall be in accordance with ISO 13381. Personnel shall be trained and certified to ISO 18436 with the program managed by a person certified to a minimum of the second highest level for each technology category applied. The program and its compliance to standards will be audited and validated using the above stated standards. The stated standards, and their normative references, shall form the basis of any compliance-based litigation and/or prosecution. This will become the norm particularly when such certification programs become fully mature and if the consequences of a missed failure include injury or death. It is not also unrealistic to see a future where one of the key business drivers for standards implementation and compliance is the insurance sector. As a minimum it should be recognised that standards represent the condensed and refined knowledge of a large group of international experts and as such contain the distilled knowledge of global experts. If for no other reason than education and learning, and a desire to improve, then such standards are a wealth of knowledge for practitioners. By their very nature the use of such standards can result in the standardised implementation of a highly effective, highly efficient, low risk, optimised condition monitoring program that consistently produces repeatable results in a cost effective manner. Surely this alone is sufficient incentive to use them! Reference 1. Framework for the Future, Business Standards, October/November 2003. 349 RELIABILITY WORLD 2005 CONFERENCE PROCEEDINGS