Maintaining asset’s desired availability can be a daunting task but not anymore (Part 1): A reliability approach
List of Contents Summary
3
Introduction
4
What is Reliability Centred Maintenance (RCM)?
5
Reliability Centred Maintenance Model Deliverables
6
RCM Process
8
Reliability Engineering: Failure Model
10
Flexing PDM Inspection Frequency
13
Performance based Partnership
17
Virtual Tools – Performance and Data Analysis
20
Prognostic and diagnostic tools
21
Conclusion
22
References
23
PAGE 2 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Summary This paper addresses some of the key disadvantages associated with conventional calendar based maintenance, where an average client spends an extensive amount of money on asset maintenance and in return the existing calendar based maintenance model often offers poor visibility on asset’s operational status, availability and savings. With very few options facility owners often seek to reduce their spending on maintenance but without the appropriate performance indicators clients may face a potential risk of losing visibility on their life critical assets operating conditions which can often lead to unscheduled downtime, poor availability and surge in maintenance cost. Uptime Plus proposes a maintenance model which is an adaptation of both Reliability Centred Maintenance and Performance based partnership model [1],[3],[5] aimed to overcome the major disadvantages of conventional calendar based maintenance and provide a cost effective solution to improve asset’s reliability, efficiency and most importantly identifies Key Performance Indicators (KPIs). With the aid of condition and performance monitoring tools clients can now have a broader spectrum of their asset’s operational status consequently mitigating any potential failures at an early stage where the cost of intervention is less.
PAGE 3 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Introduction Infrastructure maintenance industry has seen a tremendous growth over the span of 10 years, where the introduction of new technologies have paved a way for clients to increase their life critical asset’s reliability and efficiency, but the cost of maintaining a critical asset at a desired availability is still high due to hours and the labour required to perform the necessary maintenance tasks and quite often clients and service providers lose visibility of assets operational status and other key parameters (reliability, probability of failures, availability, failure rate and Mean Time Failures). Considering the recent economical climate, clients are often forced to resort to more conservative policies that would enable them to reduce their spending on maintenance and label some of the risks caused due to lack of maintenance as “Accepted Risk” but in reality it has an indirect impact on the engineering resilience of the facility. In addition to that, identifying spares for the critical assets is a task that challenges both parties (client/ service provider), where a huge sum of money is spent on buying those spares without any information on asset’s operational status, In most cases the acquired spares are either kept on site or at a safe house, quite prepared for the impending failures. It seems logical and can be deemed as “Common Sense”, the questions that needs to be answered is “Does the asset really need a spare at this point in time?” and “what happens if that asset or some of its components becomes obsolete?” Oversized inventory can lead to insufficient use of the capital and can cause serious impact on savings.
Clients can also have a tough time in making decision regarding when to replace an asset as in most cases they don’t have any site/field information that tells them how well their critical asset is currently performing, and usually in the facility maintenance industry asset replacement is carried out on “better safe than sorry” principle, which again sounds appropriate but the policy is not suitable for all assets. The primary objective for any maintenance policy is to let the user know that the deployed policies and procedures have actually helped an asset to reach or extend its documented life expectancy and maintain the desired availability throughout its service life. Unscheduled downtime is the single parameter that every individual in the industry tries to avoid, every firm who offers critical infrastructures services have their own proactive measures to mitigate the failures but only a handful of them acknowledges the facts that unscheduled downtimes are inevitable and reliability studies have shown over 20% of asset failures are age related. It is imperative to clearly understand the deployed policies & procedures should focus on finding the optimum inspection interval. The traditional approach towards this issue is performing preventive inspections/tasks at predefined frequencies (i.e. Monthly, six monthly, yearly), now that sounds like the right option but what if the actual maintenance activity is itself the root cause of the asset failure? In FM these failures are known as “Maintenance Induced Failures*”. [See footnote]
*Maintenance Induced Failure: Type of failure occurs when a maintenance technician performs an intrusive inspection or service on equipment and induces or causes a failure.
PAGE 4 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
What is Reliability Centred Maintenance (RCM)? The term “Reliability Centred Maintenance” has various definitions and the most suitable one is defined as “The process used to determine the maintenance requirements of any physical asset in its operating context” [1], In simple words, it means the maintenance tasks are performed only when its required by identifying failure modes for the particular asset and collating ages to failure data to determine Predictive (PdM) and Preventive (PM) inspection intervals. It is a method that identifies applicable and effective maintenance tasks required to maintain the inherent reliability of an asset with minimum cost.
cost is called “Start up Cost” which is caused during hardware acquisition process i.e. buying the tools and setting up training required for the engineers to implement and maintain the desired RCM standards. It has been proven and acknowledged by the reliability engineering community, the Return on Investment (ROI) from RCM is on average between 25%-30% [1]
This methodology has been widely acknowledged in the process industries, hospitals and by the aviation manufacturers where reliability and business continuity is the life line for these industries and any unscheduled downtime can cause serious financial or health & safety impact. In an ideal world, RCM should be a key process in facility maintenance (FM) industry but unfortunately FMs perspective towards RCM is not very optimistic as it has been classified as a “complex” process with too many “variables” involved and the most common response for not leaning towards this methodology is the myths that surrounds around RCM.
Surprisingly NO, over 80% of asset failures are not due to age therefore performing conventional calendar based maintenance, replacements or overhauls do not increase asset’s reliability, In addition to that performing calendar based maintenance might increase the risk of maintenance induced failures which are often hidden. Failure rate of an asset subjected to RCM is far less compared to an asset that undergoes conventional calendar based maintenance because sometimes “Less is more”.
Myth #1: “Operational cost is HIGH during RCM implementation” The answer is YES, but it is a one off cost, in RCM world this surge in the operational
Myth #2:“Sounds good in theory but in reality performing less maintenance in a critical assets poses threat to asset’s operational efficiency and it is vulnerable to failures”
Myth #3: “Replacing PPM tasks with predictive inspections has a negative impact on the maintenance model” Performing predictive tasks does not replace the original preventive tasks; it is an efficient decisive method that allows engineers to identify when to perform the specified intrusive maintenance.
PAGE 5 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Reliability Centred Maintenance: Deliverables Maintenance models are basically devised to improve the operational efficiency of assets, reduce downtime and enable facility managers to allocate their resources more efficiently by providing clear visibility on asset’s KPIs, i.e. how well they are performing at any given state?, Does conventional calendar based maintenance model deliver these aspects?
acts as a catalyst during asset deterioration process and in most cases makes assets prone to premature failures. The proposed maintenance model is a fusion of performance based partnership model [refer section 3] and reliability centred maintenance, where the latter is used to identify the failure modes, key performance indicators and reliability parameters via ages to failure data and the former is used to identify the minimum maintenance conditions and the PM and Pdm tasks intervals to meet its specified performance level. By combining the two maintenance models, some of the major disadvantages of performance based partnership approach such as loss of flexibility and the ability to deal with changes is mitigated as the model is devised to evolve constantly based on its performance.
Sadly no, all it does is carry out tasks at a regular interval i.e. constantly intervening with the asset, and creates an optimistic view that “failures are reduced or eliminated because maintenance was carried out before any failures could occur”, the statement above is not aimed to dismiss preventive maintenance strategy (PM) and say “It is all wrong”. PM is the most essential aspect in asset maintenance and its full efficiency can be only achieved, if it’s utilised in part. Constantly intervening with an asset
Ages to failure Data
Predicitve Inspection PPM FMEA
Reliability parameters
RCM
Savings
Performance
Reduced Downtime
Availability
Figure 1: Graphical representation of RCM inputs and service outputs.
PAGE 6 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
The model allows the user to identify Key Performance Indicators (KPIs) on critical assets by identifying the possible causes that could affect the specified KPIs. For example, consider an 11KV transformer required to be kept at an availability of 99.98%, the first step in identifying the KPI is to perform FMEA analysis [8] and identify the possible failure modes that can occur in that transformer based on its current operational context. Once failures modes are identified, select appropriate maintenance tasks (PdM, PM) to be performed at appropriate intervals. In this case, KPIs for an 11KV transformer will be the secondary voltage (tolerance of Âą5%), cooling, winding temperature and Insulation. Based on these parameters a minimum operating condition can be devised which provides a clear objective for the engineers that would allow them to address some of the key questions:
How the asset should be maintained? The level of maintenance required? What are Key Performance Indicators (KPIs) to be monitored? What are the potential causes that could affect the specified KPIs?
-
The model is aimed to - Improve the performance of critical assets - Increase asset availability and reliability - Reduce asset downtime - Increase cost savings - Optimise asset replacement strategy - Identify hidden failures and monitor current use of time and resources
Asset No: TX2123355 100 Performance Scale
Threshold limit-70
80 60
2010 2011
40
2012 20 0 Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
2010
50
50
50
80
80
80
90
90
90
90
90
90
2011
70
70
70
90
90
90
70
70
70
90
90
60
2012
90
90
90
80
80
80
60
60
60
60
60
50
Month
Figure 2: Typical Performance monitoring chart
PAGE 7 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
RCM Process A. Asset classification Asset classification is the most essential process in implementing RCM, where it needs to be thorough and assets should be classified in any one of the two types. The whole idea behind this process is to use the Type
Criteria
Critical
1-1 1-2
Non Critical
2-1 2-2 2-3 2-4
time and resource efficiently and clearly identify level of maintenance tasks required based on asset’s criticality (Business and Functional).
Important to business function and continuity where the user can’t afford for unscheduled downtimes. Asset’s failure can induced failures to other critical asset connected to it. Does not pose serious threat to business continuity and does not incur financial loss. The user can afford for unscheduled downtime. Assets with random failures* User can afford run to fail.*
Table 1: Asset Classification Criteria B.
Failure Modes and Effects Analysis (FMEA)
Failure Modes and effects analysis (FMEA) is a form of reliability study that identifies possible failure modes in an asset which in turn enables the engineers to decide the appropriate maintenance tasks that can be of predictive or preventive in nature that would enable them to mitigate possible failures modes. - Failure Modes - (what could go wrong?) - Cause (what could cause those failure modes?)
- Effects - (what is the consequence?) Once these elements are identified, each failure mode will be rated from [1 – 10] for their Severity, Likelihood of Occurrence and Likelihood of Detection based on asset history & available condition monitoring tools, then the Risk Priority Number (RPN) can be calculated based on eq (1), The RPN provides a clear indication on failure modes that are critical and has high probability of occurrence based on the scale mentioned in table [2]
RPN = Severity x Occurrence x Detection ………………………………… Eq (1) Scale
Status
RPN = 0- 25 RPN = 26- 125 RPN>125 Table 2: Generic RPN scale *- Assets with random failure patterns, where it is no longer is cost effective to maintain it can also be classified as rogue asset
PAGE 8 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
C. Failure Pots The term “Failure Pots” refers to ages to failure data, which is crucial for calculating the failure distribution and reliability parameters. The process of collating ages to failures data is a continuous process and accurate predictions can be obtained if it’s constantly updated and returned to the RCM facilitator on a monthly basis.
Every failure has a pattern; in order to identify the failure pattern the engineers should have the visibility of “when the asset failed?”, “what caused the failure?” and “number of occurrences?” The possible cause of failures can be identified by performing FMEA analysis
Why do we need it? Asset
PUMP#23
Function
Pumps cold water to the chillers 2 & 3
Failure modes Number of Failures
F1 2
F2 0
F3 1
F4 1
F5 3
F6 0
Time (hours)
25000
33000
37000
37500
23900
50000
Table 3: Typical Ages to failure data The above table contains sample ages to failure data of a pump, where F1 to F6 represents the failure modes as mentioned in the table [4] and time (hours) indicates the .
time those failures were detected or occurred, and the specified failures modes are not limited since FMEA is meant to be a continuous process
Failure Modes
Causes
F1 F2 F3 F4
Damaged Impeller Bearings Cavitations / Clogged suction pipe Excessive Loads, Overheating, Lubricant failure, corrosion
F5
Excessive vibration
F6
Age
Table 4: Typical Failure Modes Failure Distribution 60000 50000
Hours
40000 No.of.failures
30000
Failure Modes
20000 10000 0 2
1
1
1
3
1
Number of Failures
Figure 2: Typical Failure distribution
PAGE 9 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Reliability Engineering: Failure Model Calculating reliability parameters and predicting asset failures [2] is often deemed as “A time consuming” process but in reality it is not that hard once the process for acquiring ages to failure (i.e. raw data) has been laid out. The two key parameters required to perform failure predictions are the number of failures and the time it was detected or occurred. The process is aimed to shed some light on following the questions: -
-
When an asset is going to fail? What is time interval between two successive failures for a particular asset? What is the reliability of the asset? How much time do I have to perform the remedial actions?
Failure Rate (λ): A failure rate is the ratio between number of failures occurred and the time at which they were detected, it is usually denoted in failures per year, it is crucial to understand what is a failure* and what are the assumptions?, [see footnote] λ=R/ T…….. Eq (2) Where, R – Number of failures T – Sample time or Operational time it was detected
Mean Time between Failures (MTBF) It is one of the most misunderstood variables in RCM, as it is often confused with assets life expectancy. It is defined as the mean or average time between two
successive failures. The simplest method to calculate MTBF is mentioned below,
MTBF=1/λ…Eq (3) Where, λ – Failure rate, refer [eq (2)] Failures Average time between two successive failures FA
MTBF
FB Time
Figure 3: Showing Interval between two successive failures FA and FB respectively *
The term “Failure” is defined as the termination of the ability of the asset as whole to perform its required
function, termination of the ability of any individual component to perform its required function but not the termination oft he ability of the asset as whole to perform.
PAGE 10 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Reliability [R (T)] It is the measure of resistance to failure of an asset and it is directly proportional to MTBF or failure rate –t/MTBF
R(t)=e
Annual Failure Rate (AFR) This parameter calculates failure rate for a group of assets that are operational 24 x 7 and has same functional objective, it is denoted by
…Eq(4)
Probability of Failure [F (t)] It is a measure that indicates the unreliability of an asset based on its failure data, the output basically denotes whether the likelihood of failures will increase or decrease at any given time.
AFR = Failures in the sample period x (52 weeks/ Number of weeks in the sample period) Number of Units in the population
F (t) = 1- R (t)….Eq (5)
Figure 4: Snap shot of reliability parameters dialog box from the Predictive Maintenance Management (PMM) tool
PAGE 11 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Mean Time to Repair (MTTR) It is the measure of the average time required to repair a failed asset or its components and it is usually expressed in hours
Availability (A) It is generally defined as the degree to which an asset and its component are in operable and committable state at any point in time when it is needed.
MTTR = Total Downtime in hours .Eq (6) Number of Breakdowns
A = MTBF/ MTBF+MTTR ….Eq (7)
2.1 Maintenance Review As mentioned in the earlier section, RCM is an optimum mix of predictive, preventive and reactive maintenance, it is vital that this principle is reflected on the maintenance planner by performing a maintenance review and identify the most appropriate method of maintenance activity required for an asset. -
-
-
Acquire the list of maintenance tasks. Identify and exempt the tasks that are required to satisfy health and safety legislations which can only be performed by intrusive maintenance from the review. Factor FMEA results Identify the tasks that can be subjected to predictive Inspections. Identify the tasks that can be subjected to preventive Maintenance Identify the tasks that can be subjected to visual Inspections While amending or assigning the frequencies for PdM inspections, the type of the condition monitoring test and objective of the original PM task
-
should be considered, for example – If a maintenance task aimed to verify whether there are any excessive vibrations in a pump on a yearly basis, a non intrusive vibration analysis is preferred and recommended to be performed on a six monthly basis, as the cause of excessive vibration in any rotary asset can grow rapidly thus increasing inspection frequency will enable engineers to keep track on the operational status of the asset and perform remedial works before it the exceeds specified tolerance level. It is important that the assigned frequency should be feasible and cost effective; the entire maintenance review should be performed by the RCM team since most of the decisions are made based on engineers/managers field experience.
PAGE 12 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Flexing PdM Inspections Intervals Maintenance models often don’t allow the user to change the inspections interval as it could have perceived impact on the cost, availability or resilience of the facility. The proposed RCM model allows “Flexing” the
PdM inspection interval based on the following parameters: - Asset’s Failure Pattern - Age - Operational life
Failure pattern of an asset is dependent on the failure rate, and there isn’t a definitive pattern for all assets, it varies based on the load, environmental condition, temperature,
design, shipping, and installation. But most assets follow a failure pattern called “Bathtub-Curve”.
Failure Rate
Phase 3: Wear Out Period
Phase 1: Infant Mortality
Phase 2: Operating Life
Time
The curve itself is classified into three phases and it is dependent upon “shape parameter” denoted by the symbol β (Beta),
If β < 0, it is classified as Phase 1 (Infant Mortality) or asset prone to early failures.
If β =1, it is classified as Phase 2 (Operational life) which indicates the asset entered into its operational life or useful life
decide the appropriate maintenance tasks, in most cases the PdM inspection frequency will be increased in order to monitor the status of the asset, where it gives sufficient time for the engineers to organise the remedial actions. The analysis works well for assets that are operational 24 x 7, but for critical back up assets (for example a standby generators) the probability of failure can be calculated by
and prone to random failure and finally if β > 1, it is classified as Phase 3 (Wear out Period) which indicates engineers that the asset has high probability of failing and suitable replacement or remedial actions is required. The ages to failure data is again crucial to calculate these parameters, based on the β value engineers can schedule or
β
Qn = 1 – exp [(n-1) * τ-γ] / n * exp [-[nt- γ / n] ………..Eq (8) Where, Qn= Probability of Failure over the entire interval n; η = Characteristic Life Parameter; β = Shape Parameter; γ = Location Parameter; τ = Inspection Interval n = Number of times the component operated in it s life.
PAGE 13 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
CDF= 0.632
Weibull Analysis 1 0.9 0.8 0.7
CDF= 0.632
Beta
0.6 0.5 0.4 0.3
Beta = 1.71
0.2
Scale parameter = 2200hrs
0.1 0 0
500
1000
1500
2000
2500
3000
3500
Hours
Figure 6: Typical Weibull Probability Plot (based on the sample data)
P-F Curve It is commonly defined as “A visual representation of the behaviour of an asset as it approaches failure”; The P-F Curve is plotted against two parameters asset condition and time. Once a failure has been identified (Via PdM or Visual Inspection) it is labelled as Point ‘P” called Potential Failure, Usually the potential failures become visible at around 70% of asset’s operational life, and the interval between potential failure and functional failure is called as “P-F
which means the asset or its components had shown an early sign of deterioration and it can lead to the Catastrophic or Functional Failure point ‘F’ where an asset can no longer be in operation or can no longer perform its specified function Interval”. The general rule is, during the P-F interval the asset must be inspected at least once; the inspection can be predictive, intrusive/visual.
PAGE 14 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
2009 2010
POTENTIAL FAILURE Symptom 1 Detected
2011
2012 B
Increase the inspection interval to 6 monthly to exactly predict Point ‘F’ and also it is more feasible and cost effective approach
Functional IS TASK INTERVAL PRACTICAL? = Yes
PF Interval C I PdM Inspection Interval (I) = 6 Monthly
OPERATING AGE/Time
Figure 7: Typical P-F Curve The Inspection interval (I) can be calculated by, I = PF/n…Eq (9) [see footnote] Where, PF = Duration or Interval between Potential Failure and Functional Failure
n = Number of inspection carried out during PF Interval
*Example: If PF Interval = 8 years, Minimum number of inspections carried during the PF interval is 2 I =PF/N = 8/2 = 4 or 4 monthly
Based on the asset’s failure and deterioration pattern, the predictive inspection frequency is varied. Intrusive maintenance is performed only when the asset operational condition is in amber to red transition period, i.e. the optimum point of intervention and the maintenance interval
is tuned accordingly so that engineers do not lose the visibility of the source. It is a manual process and it’ll be usually be carried out by a PPM manager based on the reliability and the field information provided by the RCM facilitator
The failure pattern illustrated below is a typical bath tub curve, but it is very unlikely that all the assets will follow this pattern as there are six different types of failure pattern. As discussed in the earlier section,
based on the ages to failure data asset specific failure pattern can be identified and can be used during the flexing process. [See note]
PAGE 15 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Asset Condition
Wear Out Period
Infant Mortality Operating Life
Asset Condition Curve
PdM Inspection Interval
Time Figure 8: Predictive Inspection interval (PdM) flexing * Asset’s MTBF is not factored during PdM inspection interval flexing, as in most cases manufacturers MTBF and asset operational MTBF will not be the same
PAGE 16 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Performance Based Partnership – Business Model Performance based partnership / Performance based maintenance [4],[5],[6] is a maintenance approach where Uptime Plus acts as an engineering consultant and takes full responsibility in maintaining the condition of all the assets in the facility within the agreed budget. In this approach the performance standards are agreed instead of maintenance techniques consequently shifting the risk from client to Uptime Plus prior to contract agreement. The performance requirements for this approach can be divided into qualitative and quantitative requirements where the former implies that the client needs are expressed in the form goals and objectives which are usually derived from the functional and performance requirements, the latter implies the standard verification methods (audits). The performance of a critical infrastructure can be determined by its asset’s condition and deterioration rate where this approach predominantly focuses on condition based maintenance or monitoring tools which would enable the engineering consultant to have the full spectrum of an asset’s operational information. Performance requirements are not just technical, the performance of service delivery (e.g. Response time) is also accounted. The flow chart depicted in figure 9 is a visual representation of the performance based partnership approach and the objective of this model is to improve the quality & reliability of the assets, make cost savings and provide budget certainty
and development of a long term relationship. In the initial stage, client will liaise with a group of maintenance contractors where in this stage Uptime Plus will act as an engineering consultant and contribute to the planning process in which the maintenance intervals are predetermined, and proposes bespoke maintenance strategy within the constraints of performance requirements. Key Performance Deliverables: - Improve asset’s reliability and quality - Aid client to achieve direct cost savings - Reduce risks associated with compliance and legislation. - Provide clear visibility of asset’s operational status - Manage and monitor the performance of life critical assets - Being innovative in developing new maintenance strategies. An engineering consultant will take the responsibility for providing evidence of business related financial risks associated with various maintenance scenarios. For example, when the consultant reports a defect or deterioration on a critical asset, e.g. an 11KV transformer, client will be supplied with the following information - Type of fault - Source of fault - Ages to failure date (In visual form) - Deterioration rate - Time to fail - Remedial and replacement strategy
PAGE 17 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
execution phase, completion of each task will be reported back to the client where all the tasks and its execution frequency will be monitored and assessed by the client. In the assessment period, the audit results of both parties will be compared and evaluated whether to confirm the Service Level Agreements (SLA) and the specified performance requirements which were agreed during the specification phase are met.
During the performance specification phase, both the client and consultant would liaise with one another and decide the performance threshold level for all critical assets, i.e. they will decide the required availability for the assets and the minimum level of maintenance conditions. Business continuity and the impact caused due to failures are the decisive parameters for this process, i.e. “How important is that asset to my business?” and “what will be the impact on business, if it fails?” and this task can be time consuming and the role of an engineering consultant is to help the client to conclude a general agreement during this process by providing an universal table of critical assets, desired availability, key performance indicators (asset specific) and inspection interval to monitor the performance.
Asset Performance Indicator API = [Conditional assessment of the asset x 10] …eq [10] Where the rating interprets [see foot note], >80 API – Asset at good or high service level 70 > API < 80 – Asset at marginal condition 60 > API < 70 – Asset at deteriorating condition API < 60 – Asset at poor or critical condition
Key performance indicators and the level of maintenance required to satisfy the specifications will be derived by Uptime Plus facilitators. After the agreement, a bespoke maintenance model will be devised and sent to the client for final approval. In the
Asset Performance Indicator
Asset Number Interval
First quarter
122675
Description
Second Quarter
Packaged Chiller Unit
Supports
Third Quarter
Ist Floor COMMS room
Fourth Quarter
2010
50
50
50
80
80
80
90
90
90
90
90
90
2011
70
70
70
90
90
90
70
70
70
90
90
90
2012
90
90
90
80
80
80
60
60
60
60
60
60
Table 6: API table * - Asset performance indicator less than 70 requires preventive /intrusive maintenance
PAGE 18 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Long term agreement process: Client Budgeting maintenance project
Specify Maintenance requirements
Uptime Plus
Collate Project Information Conclude General Contract Specify provisional performance criteria
Determine starting Point
Project assessment
Condition assessment
Devise maintenance scenarios and performance criteria
Supervise the process
Devise activity plan
Conclude Partnership Agreement
Devise project plan and PPM
Conclude performance and maintenance interval
Execution of work
Assessment of completed tasks
Assessment of performance indicators
Devise maintenance plan
Completion of work
Agree performance guarantees Evaluate the partnership
Periodic performance audit
Adjust maintenance scenarios and activity plan
Figure 9: Performance based Partnership flow chart
PAGE 19 OF 23
M AI N T A I NI N G AS S E T â&#x20AC;&#x2122; S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Virtual Tools – Performance and Data Analysis Performance monitoring tool is the most essential aspect in the model, where all the ages to failure data, current operational status, asset information and their hierarchy are stored. The tool allows the user to edit or add an asset and provide a visual representation of asset’s operational status and keep track on the existing PPM planner, remedial actions. In return the tool enables the user to acquire the valuable historic data that would allow the RCM facilitator to determine asset’s failure pattern consequently results in devising bespoke maintenance strategy
Deriving bespoke asset replacement and critical spares strategy is possible which is usually based on the failure rate. Low MTBF doesn’t necessarily means that the asset should be replaced because asset replacement is entirely age related not on MTBF, having the historic information of critical assets enables the user to distinguish between MTBF and asset life expectancy, resulting in optimising the existing asset replacement model.
Figure 10: Snapshot of the PMM database and asset life expectancy window
PAGE 20 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Prognostic and Diagnostic tools A quintessential aspect in both RCM and performance based partnership approach is the condition assessment. Visual inspection is often the primary method to access the operational status of the asset, but the amount information that can be extracted via this method is limited as there are constraints that limit its efficiency (e.g. human errors, spurious alarms). In some cases, detecting failures can often challenge even the most experienced engineers since some of the early signs of deterioration are hard to detect or almost impossible during visual inspections. With the rapid growth of sensors and signal processing technology [9] engineers can now have a much broader spectrum of their assetâ&#x20AC;&#x2122;s operational status and allows them to detect early deterioration signs and even some of the hidden failures. Thermal Imaging [7] (Thermography) is one of the efficient non-intrusive methods to detect any thermal anomalies on electrical assets, and works on the principles of joules heating effect, these heat signatures increase when the current in a particular conductor increases (overloaded) and it can be easily be detected by Infra-red scanning. It is suitable for detecting over-loads and loose connections in fuses, switch gears, transformers and bus bars. In HVAC, thermal imaging is used to detect refrigerant leaks, leaking pressure gauges where the method can be used to replace the quarterly intrusive leak detection checks on chillers. In rotary assets, it is ideal for locating the root cause of overheating. It is suitable for identifying overheated bearings
or rollers, misalignment of shaft, pulley or coupling and lubrication failure Deterioration in fuel tanks, oil filled transformers and pipe works can be identified via fluid sampling that basically detects any fluid contamination and provides indication on the level of deterioration. Partial Discharge (PD) is an electrical discharge that does not completely bridge the space between two conductors. The discharge may be in a gas filled void, in a solid insulating material, in a gas bubble, in a liquid insulator. When partial discharge occurs in a gas, it is usually known as corona. Partial discharge is accepted as a standard protocol test for high voltage assets by power sectors. In addition to that partial discharge detectors are equipped with ultrasonic sensors where they are used to detect arcing and corona in HV/MV switchgears and transformers. Vibration analysis is an efficient nondestructive testing tool for the buildingâ&#x20AC;&#x2122;s rotary assets, basically the tool analyses the vibration signature of high speed rotary equipments such motors, pumps which has a on board diagnosis tool with the clever algorithm that can prioritise repair recommendations. The vibration analyser is equipped with tri-axial accelerometer and a two- point laser tachometer (speed measurement) for precise vibration sampling to identify bearings looseness, misalignment, unbalance, gear problems and bent shaft.
PAGE 21 OF 23
M AI N T A I NI N G AS S E T â&#x20AC;&#x2122; S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
Conclusion The proposed model provides complete transparency over critical asset’s KPIs and with aid of reliability predictions supported by performance & condition monitoring tools, asset failures are detected at an early stage where the cost of intervention is minimum, consequently enabling facility owners to achieve substantial cost savings and enables them to maximise asset’s service life and in certain case extends asset’s life expectancy. In addition to that asset’s failure pattern is determined in order to identify the desired maintenance frequency consequently resulting in a dynamic maintenance planner which is mapped against asset’s failure pattern, operational status and age. This approach increase asset’s reliability, availability and maintains downtime well below the threshold level. Overall the reliability and performance based approach for asset maintenance is an effective replacement to the conventional calendar based maintenance.
About Authors: Andrew Dutton CEM Director, Integral UK, 1290, Aztec West Almondsbury, Bristol BS32 4SG Email: Andrew.Duttton@integral.co.uk
Laxmi Vajravel Critical Infrastructure Manager 1290, Aztec West Almondsbury, Bristol BS32 4SG Email: Laxmi.Vajravel@integral.co.uk
PAGE 22 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H
References [1] Introduction to Reliability-Centred Maintenance by John Moubray, ISBN-10, 0750602309, ISBN-13 9780750602303 [2] Paul Barringer, P.E, Predict Failures: Crow-AMSAA 101 and Weibull 101, Barringer & Associates, Inc, Proceedings of IMEC 2004 International Mechanical Engineering Conference December 5-8, 2004, Kuwait, Published by Kuwait Society of Engineers. [3] Smith & Hinchcliffe, RCM--Gateway to World Class Maintenance, 1st Edition, 2003, Butterworth-Heinemann, ISBN: 9780080474137 [4] Ad Straub, Performance based Partnership forms for Maintenance by Dutch housing Associations by, 2005, OTB Research Institute for Housing, Urban and Mobility Studies, Delft University of Technology [5] Igal M. Shohet, & Ad Straub, Performance-Based-Maintenance: A Comparative Study between the Netherland and Israel, 2010 EFMC (European Facility Maintenance Conference) [6] Ad Straub, The Maintenance Contractor as Services’ Innovator In Performance-Based Partnerships, TU Delft OTB Research Institute for Housing, Urban and Mobility Studies ,The Netherlands. [7] Business Focused Maintenance, Samples and Schedules by Jo Harris and Paddy Hastings, 2004, BSRIA 70174 December 2004 ISBN 0 86022 604 2 Printed by Multiplex Medway Ltd. [8] V. Narayan, Effective Maintenance Management – Risk and Reliability Strategies for Optimizing Performance, April 2004, Industrial Press Inc., ISBN 0-8311-3178-0 [9] Andrew K.S. Jardine, Daming Lin, Dragan Banjevic, A review on machinery diagnostic and prognostics implementing condition based maintenance, Mechanical Systems and Signal Processing, Volume 20, Issue 7, p. 1483-1510 [10] Alan Pride, CMRP, Reliability Centred Maintenance: http://www.wbdg.org/resources/rcm.php, last Updated on 06-07-2012.
PAGE 23 OF 23
M AI N T A I NI N G AS S E T ’ S DE S I RE D A V AI L A BI L I T Y C A N BE A D AU N T I NG T AS K BU T N O T A NY M O RE ( P A R T 1) : A RE L I A BI L I TY AP P R O A C H