 |
 |
 |
 |
 |
 |
 |
||
|
|
|||||||||
“Do
not follow where the path may lead.
Go instead where there is no path and leave a trail.”
Ralph Waldo Emerson
You have decided to build and implement optimized CBM policies within your physical asset management organization. What new information procedures will you have to set in motion for this initiative to succeed? Generally, maintainers will, henceforth, record the five reliability-centered knowledge elements (see Chapter 1. page13 and Chapter 2. page19) into the CMMS database whenever they close a work order. They will assemble this new data, initially, as "structured free text". Many information technology specialists and maintenance management consultants frown upon the suggestion of free text. Bear with us while we demonstrate that this new approach will liberate physical asset managers from a long standing "reliability information impasse".
Relational databases (upon which all CMMS applications are built) are powerful tools because they enable us to organize important information. We want our information organized in order that we can retrieve and analyze it in a variety of ways. Analysis always precedes the success of any endeavour.
Conventional relational database techniques are ideal for storing (for instance) lists of people, their addresses, phone numbers, and other attributes that may be useful for efficient business analysis and processing. As an example, we could extract a mailing list from a database in order to send a sports car brochure to single men between the ages of 19 and 35 with a minimum annual income of, say, $50,000. Our relatively simple optimizing objective, "to print and mail brochures to a likely group of car buyers", will be satisfied by a database table having specific columns for sex, age, marital status, and revenue. Maintenance and reliability optimizing objectives, on the other hand, will require substantially greater intricacy of information resolution than that illustrated by the ‘sports car’ example.
To improve maintenance policies, we must analyze our work order database. That database contains a complex diversity of fact and judgement. The section entitled The “failure code” problem (page 18) illustrated numerous practical difficulties in selecting and compiling failure modes from conventional drop-down lists in a CMMS. We discovered that most work order records, while they may contain considerable information on the repair tasks that were accomplished, are barren with respect to the details on the prior state of the equipment. Reliability analysis, nevertheless, operates primarily on the latter. Let us proceed, in this chapter, to discover, how to implement an information structure and an optimization process that will:
The foregoing represents a considerable challenge. With adequate purpose and guidance, however, we shall proceed in a step-by-step manner. The investment is incremental, but each advance will return visible benefits both to the maintenance department and to its client organization.
We begin by selecting one or more candidate items that meet these criteria for effective CBM:
Next we add reliability enabling data, to existing work order records in the CMMS. We begin by examining every past work order record related to the item for which we are attempting to build an optimal CBM model. As we study each record, we add to it, reliability data that we will have structured according to the principles outlined here. This process will, undoubtedly, involve research, interviews, and (to some extent) guesswork. Accuracy (what really happened) will have been dimmed, quite often, by the mists of time. Regardless of the inclusion of approximations and assumptions, the structured data resulting from this exercise will, nevertheless, provide the foundation for our initial CBM decision model. More importantly, it will inaugurate, a generalized everyday process for effective long term reliability information management.
Most CMMSs allow you to allocate (or add) new fields to the work order table and to update the work order input form accordingly. If this is not possible (or only partially possible) in your CMMS, then a single text field[3] will work just as well. Alternatively, a mix of reliability information (both in the text field and in dedicated fields) may be used. In the text field we will insert key phrases, suffixed with (for example) " :". Each key phrase shall correspond to a reliability information element.[4] In this form, we extract and, subsequently, we will use the data for reliability analysis and optimal decision making. Add (if possible) the following information elements (or those that are not already included) to the work order data structure. For an item to be analyzed, populate these fields retroactively in the item’s historical work order records. For example:
If an "RCM Reference Number"[5] (information element 9) is entered, then the following fields (10-14) must remain unfilled. (They will be ignored by the EXAKT work order processing program.) If "RCM Reference Number" is null, a record will be be added automatically to the RCM table (see next section) that will contain the values (the reliability-centered knowledge elements) of the fields 10-14 for the item.
Often, a single work order will cover several functions, failures, failure modes, or even several items. In those cases a single work order record may generate (or refer to) several "reliability" records (in the RCM table). Each will contain a unique combination of the fields 1, 10, 11, and 12. The repair person will provide two or more sets of the (16 reliability) information elements on same work order record.[6] For example, assume that the planner decided that it would be expedient, at this time, to remove and replace the motor coupling since it was getting close to the age at which an age based renewal was scheduled later on in the year (knowing that the coupling on Crusher 1017A had failed earlier in the year causing considerable secondary damage). The mechanic, upon removing the coupling observed no damage nor sign of excessive wear. Hence, in a second set of 16 data elements, he would populate element 15 with “S”. The mechanic may indicate in “Additional comments : ”, “coupling fully functional, no damage observed”. “S” will indicate to a future reliability study that the coupling was unfailed at the time of replacement – that is its life cycle was “suspended”. The process by which the EWOP method generates additional work orders is described in greater detail in http://www.omdec.com/articles/p_EXAKTWorkOrderProcessor.html.
One may argue that, this degree of “wordiness” and detail in completing a work order, especially where a general overhaul was performed, is onerous and excessive. One would agree, however, that it is worthwhile to adopt a degree of informational completeness that is proportional to the consequences and probability of the failure of the equipment in question. Recalling the example of the ingot transporter in the section Significant components on page 58, the repair person will devote less time and apply a smaller amount of detail to a less critical equipment, say, one whose functions are duplicated by a backup system. Recall too, that most of the 16 reliability information elements for a failure mode of an item, need be entered manually, only once. Thereafter, in future incidences, the RCM record is merely referenced (using information element 9) in the work order record. (Recall the advantages of the one-to-many integrity constraint illustrated in Figure 2‑2 on page 21). Moreover, in the section Failure codes (ahead in this chapter on page 181) we describe the evolution of accurate failure codes that will eventually reduce the clerical verbosity of this approach, almost to zero.
The EXAKT work order processing (EWOP)[7] program will process all (or specified) work order records. It will extract reliability data (when formatted according to the 16 reliability information elements) and insert them into the Events and RCM tables, structured as follows:
Ident: Crusher 1017B
Date: May 14, 2004
WorkingAge: 17,400,897
Event: EF1885[8]
Work_Order_Number:
349798
RCM_Reference_Number: 1885
Events is the table (along with various CBM database tables) needed reliability analysis and for building an optimal CBM decision model. The EWOP program will populate the Event field (of the Events table) with values, such as B, EF, ES[9] to which it will append the RCM reference number. This will facilitate the data mapping procedure in EXAKT for marginal analysis (see Example 3 Complex Items on page 150).
Item: Crusher 1017B
Function: To crush 130T/day of ore from 100 cm to 30 cm.
Failure: Crushes only 60 T/day
Cause: Multiple trips, and operator stoppages due to increased drive side north bearing resistance and overheating of motor.
Effects: Operator sometimes notices increased temperature on guage in control
room. He adjusts capacity downward to allow bearing to cool. Sometimes he does
not notice and motor trips. This can result in an uptstream blockage that
requires downstream process to stop after 90 minutes. Bearing vibration levels
in the high frequency range usually rise preceding this event.
Consequences: Operational. Lost capacity at this time due to high market demand.
Direct impact on customer service as well as lost revenue and increased
expense.
Date entered or modified: May17, 2004
By: André Lacasse
Date verified:
By:
The RCM table constitutes our reliability-centered knowledge base. The EWOP program inserts new records into the RCM table. Each record in the RCM table will contain a unique combination of the values in Item, Function, Failure, and Cause. That is, these four fields constitute a primary key in the RCM table.
One may ask why two fields (7 and 8) containing the same information are required.
7. Working age
of equipment at time of work order : 17,400,897
metric tons
8. Working age
of equipment when component placed back into service : same
Sometimes part of an item is taken out of service for a minor servicing[10] while the rest of the item continues operating. That is, the other components of the item (or equipment) continue to accumulate working age while one component is in a state of “suspended animation”. An example might be one cylinder of a gas compressor that is taken off line for a certain period of time (weeks or months) while the other cylinders resume operation. In this case the fields 7 and 8 (of the set of 16 reliability information elements referring to that cylinder) would hold different values of working age.[11] 4 would hold the working age of the equipment when the cylinder was taken off line, while 5 would hold the working age of the equipment when the cylinder was placed back on line. The difference between the two will be excluded automatically from counting towards the actual working age of the component[12].
The process not only tracks the life cycles of significant components, but also permits a determination of the relationship between failure modes and condition monitoring data. The granularity at which we choose to identify a component within our maintenance information process will depend on the probability of failure and the gravity of the consequences[13] of the failure mode. If, for example, in the case of a hydrogen compressor, a valve failure stops the process line, likely to result in operational losses, then it will be worthwhile to identify the failure mode as having occurred at a specific valve position on a specific cylinder. Subsequent analysis[14], will reveal behavior patterns that may call for an adjustment of maintenance policy or operation, or a physical modification.
The EWOP program will insert (at least) two records for each workorder into the Events table. One record will correspond to the ending event of the previous lifecycle and the other will correspond to the beginning event of the next lifecycle. Some work orders will not involve a replacement or major repair (with respect to the failure mode in question). In those cases the lifecycle[15] will neither end nor begin, but the component will continue to accumulate working age where it left off at the moment of the work order. The information element. “C” (for continue) in field 15 will indicate this[16] to the EWOP. For example:
indicates a minor repair or adjustment rather than a failure. No lifecycle beginning or ending events will be added by the EWOP.
The information in field 5, regarding a minor repair, may be significant. It may be used by EXAKT, if, in the model building procedure, we judge it to have an impact on monitored data. For example, a re-alignmment of mating shafts would “reset” vibration levels to a lower value, but accumulated damage on the bearing up to that point will not have been reduced. The predictive model, in this way, “knows” about a “minor repair” and will not erroneously interpret the reduction in vibration as a “rejuvenation” of the bearing. This notion (of accounting for the effect of minor repairs on condition monitoring data) was discussed and illustrated in Figure 10‑20 The effect of an oil change on page 141.
For work orders that cover multiple items, functions, failures, or failure modes, the EWOP program will generate additional pairs of records in the Events table, and, if the RCM reference number is not given, will insert additional records into the RCM table.
By using the methods of the section “Implement a reliability-centered knowledge structure” (page 175) we will, in essence, have converted our CMMS into a “reliability database”. Henceforward, we may analyze and exploit its content in a variety of ways.[17]
We note an important relationship between the RCM table and the Events table. A record in the RCM table may relate to one or more records in the Events table. Similarly a record in the Work Order table (of the CMMS) will relate to one or more records in the Events table. However each Events table record will correspond to only one record in the RCM table and only one record in the Work Order table. By enforcing these "integrity constraints" the EWOP program will help ensure that the data analysis required for building a CBM predictive model, proceeds quickly with minimal fuss and error.
Once the Events table has been generated, we may proceed with the data preparation phase. During the data preparation phase of the EXAKT analysis, we will perform an important step, called data mapping. We map each failure mode (currently rendered in the events database in “free text”) to a failure code. Each new failure code, based on careful consideration, in the full context of the five knowledge elements, may, henceforth, populate the evolving drop-down lists of the CMMS. Rather than having eliminated the use of failure codes, for the reasons mentioned in The “failure code” problem (page 18), we have merely deferred their use until after their establishment as accurate descriptive attributes of the failure modes affecting a physical asset in its operating context. Hereafter, we may select from a list of failure codes that relate directly to RCM records.
Figure 12‑1 Mapping of Events to individual components or failure
modes
Figure 12‑1 illustrates an interactive tool used in EXAKT for mapping events in the Events table to a particular failure mode or component upon which we wish to perform reliability analysis. A reliability model (or CBM model) refers to a targeted failure mode or component. In the dialog of Figure 12‑1 we are mapping events that relate to a RCM record in the RCM table to a model that will describe the failure behavior (including the influence of any CBM monitored variables).
7. Working age
of equipment at time of work order :
8. Working age of equipment when component placed back into service :
Working age is the working age of the item, the equipment, or possibly it is a measure of the cumulative throughput of the production line in which the equipment operates. In all cases, the EWOP program (and EXAKT) will keep track of the individual working ages of every significant component. (A discussion of Significant components may be found in Chapter 3. on page 58 and a discussion of Keeping track of system component on page 58). EXAKT accounts for component working age by using the Failure type (15), Date (3-4) and Working Age (7-8) fields. In many manufacturing and process settings a real-time production database maintains a record of the dates production volume, product formats, stoppages, fuel types, or other relevant data. This information may be used to update the working ages, automatically, by relating the date fields in the two databases. The updating of WorkingAge may be accomplished either within the CMMS application or by the EWOP program.
The fields (or reliability information elements),
1. Item :
10. Function lost or compromised or
threatened :
11. Failure :
12. Cause :
uniquely describe a reliability-centered knowledge record. The combination of these four data fields make up the fundamental unit of knowledge regarding the reliability characteristics of any physical asset. We may consider the combination as being the “lowest common denominator” or the fundamental knowledge unit upon which an understanding of all maintenance requirements in an organization may be achieved.
Field 10 describes a function that has been lost, compromised or threatened by the events that led to issuance of the current work order.[18] On page 176 we described the complexity that a single work order may cover more than one item, each having a broad spectrum of functions. For example, a single work order may cover the overhaul of an equipment containing many items. Recall the definition of an item as a component or assemblage of components that is convenient to analyze as a group. In the general case, where no RCM analysis has yet been performed, items will not yet have been formally defined. It is recommended in those cases, that the item be considered as the equipment itself, as it is currently registered in the CMMS. If, in the future, it is found that it is more desirable to define an item within the equipment, at a lower level of "indenture", this can be easily done at that time with minimal impact on the various database tables.
The initial decision model that we build was based on data in the CMMS. We will have modified that data during the analysis and model building phase by using the methods outlined in the section “Implement a reliability-centered knowledge structure” on page 175). The model will often fail one or more of EXAKT’s statistical validation tests. Do not despair. Use the model, but, do not abandon CBM interpretation methods and/or more intrusive inspection methods currently in use. Continue to populate the work order database in the novel way described here. As potential failures of the item are uncovered, update and reverify the model by re-running the EWOP program frequently and rebuilding the model. Verify the model using each of the tools in EXAKT (graphical statistical analysis, Kolmogorov-Smirnov test, and cost comparison analysis). Thanks to the new reliability data procedures that you will have implemented, eventually, with little additional effort, you will generate a “good” (statistically validated) optimal decision predictive model. Proceed similarly to build predictive models for other equipment items and their failure modes where the criteria on page 175 are met.
The RCM table constitutes a living and growing knowledge base[19]. Quality control of the information that populates this table, is an important issue. The QC process will require methods similar to those suggested in Figure 2‑6: Extending the Use Case "Complete the work order Form" (page 25).
In this chapter we learned about some of the practical steps in
implementing optimal CBM policies. We built our methods upon our understanding
of the fundamental reliability-centered knowledge elements introduced in Chapter
1. and Chapter
2. By actually implementing an EXAKT CBM optimized
model, we have set in motion a broader process of building and using a living
reliability-centered knowledge base. In addition to more effective CBM, we
will, without doubt, quickly discover a multitude of visible, spin-off benefits
to other physical asset management initiatives undertaken by our organization.
All physical asset management improvement programs (for example RCM, TPM, RBI,
Six-sigma, and many others) depend on good information. Because the information
in a reliability-centered knowledge base has been rendered in its most
fundamental form, it may be analyzed,
processed, and used to great effect within the structure of any other
physical asset management improvement methodology.
[1] The closer the intrinsic relationship between the monitored data and a deteriorating failure mode the smaller the size of the sample required to build a predictive decision model. The EXAKT statistical tools will indicate a level of confidence in the model that we have built.
[2] Before building the model, we don’t know which way exactly, but initially we suspect that our monitored data does possess predictive content. The purpose of an EXAKT analysis is to discover the “exact” nature of the relationship between monitored data, working age, and the probability of the occurance of failure, particularly of potential failures, with respect to specific failure modes.
[3] This will be a SQL LONGVARCHAR in the database. For example a Memo field in MS Access.
[4] There are 16 “reliability information elements”. These include the 5 “knowledge elements” (10-14) introduced in Chapters 1 and 2.
[5] This is a foreign key from the RCM table.
[6] Hence, the dedicated fields may contain the first set and the text field the subsequent set(s) of reliability data. The EWOP program will extract data from both sources.
[7] A program that processes a report generated by the database (CMMS) application, inserting records into the Events and RCM tables. Multiple sets of the 17 data elements may appear in a single work order record. For the second and third set, null values in elements 1-7 will be populated by the EWOP program by duplicating the values in the preceding set. Hence they need not be filled by the repair technician unless they are to contain different values.
[8] Although this was a potential failure, whose consequences were mitigated by the CBM vibration monitoring task, the bearing was considered to be in a failed state. Reliability analyses such as Pareto, Weibull, or PHM will treat this lifecycle (as referenced in the RCM record) as one “ending with failure”, EF (as opposed to “ending by suspension”, ES).
[9] As they were defined in “Recording Events” page 57. B is the beginning event of a component-failure mode lifecycle, EF is the ending event by failure (including a potential failure), and ES is the renewal of the component for any reason other than its failure. Typically another component will have failed and it was considered expedient to replace this one at this same time. The event symbol will contain a reference to the relavant RCM record. (See http://www.omdec.com/articles/p_EXAKTWorkOrderProcessor.html).
[10] One that does not “zero-time” the component. See http://www.omdec.com/articles/p_KeepingTrackofComponents.html for a discussion of suspended animation.
[11] In this case, the EWOP program, will insert the events BSA (begin suspended animation) and ESA (end suspended animation) into the Events table. EXAKT will, in this way, use the accurate working age of each significant component for modeling and prediction.
[12] The working age methodology of EXAKT resolves the familiar problem of keeping track of the ages of specific components in an item. See Keeping track of system component on page 58
[13] The combined probability and gravity of consequences is known, in reliability circles as “risk”.
[14] Weibull, PHM, Pareto, etc.
[15] Regrading the failure mode or component in question.
[16] Only necessary, if for some reason, the technician decides to create a sub work order to cover this item-function-failure-cause. Otherwise, EXAKT will accumulate working age for this component normally. A work order covering a minor maintenance such as an oil change may be handled in this way.
[17] See Chapter 3. Analyzing data page 33
[18] Sometimes the work order will have been issued by the CMMS as a time based inspection, overhaul, or replacement. The maintainer should take this opportunity to relate this work order to a RCM record using the very same information methods described thus far.
[19] Maintainers, engineers, operators, OEMs will access it often, now that it contains the data required for reliabilty analysis and model building. In effect, the CMMS will have become transformed into an “intellectual asset” and knowledge source at the service of the physical asset management team.
|
