|CAPACITY ASSURANCE STRATEGIES
re Tee ee
USMS
Common Sense
aT
Maintenance
MCAThs rice focuses on this type of arponch in
reohingeome common maintenance Ung oe
‘minennc coe ang Thevoainbetr eee
‘titenocsdchion ling The betes
ton of maintenance nc iproved equipment el
‘biny and incu compeny oe
Silos, io, sos
(Computerised Maintenance Management System),
EAMs (Enerprie Aset Management msems) and CM
have ws that inhibit heir elec; ere are some of
these drawbacks—plus ugetons or ome,
|LRCM peys ite atenion to historic data and is fen
seen as competitive to the main historical dat source
(the CMS) Any objective view, though, sees them
complementary to each other. Howe for example, 0
beter tack the occurence frequency of RCM fare
‘modes tan ona CMMS work order? And how bet 0
start RCM analyisthanby examining flute modes tha,
Ive actully occurred? The relation link
RCM and CMS databases f
2.An RCM realty is that fare modet FMs) are not
comprehensive; thee dependent on Wit sensi:
‘ppled by RCM snaps, Studie show thatthe gap
Detween the FMs predicted and thote actualy observed
‘an be huge. Relig thir tation rears enhancing
"the ROM databace by ein actual maintenance exer
ec ast happens rom the CHS.
Most ure naps pres the ture by looking back
wards (Lathe CMMShisery), Thisapproach however,
"ply omis, for example the RCM concep of poten
th fares (PS). There are two exons forth Fst,
the CMMS doesnot low for sich data to be cole,
‘Second, the tehnicans are not eine 6 ecognne acta
Re Song this poten eal fr minor mada: to
the MMS work order preessand modicamofraningat
‘the proverbial ol fe (ce for he working in dy day
mattenance persion)
4.0M data should prompt intligent decion-making.
“The steps ace involved. ist collect the right data (nd
stop callecng the wrong data). Second, do the right
analysis Third use the dita for making the right dei
son In ou experience, upwards of 70% of the dt as
‘no predicive ait and, of coarse, key dss miing
‘A cited point i reution of the problem i that CM
lata must reate to the fare mode the question is how to
emorsrate this
‘SEPTEMBER 2008
CAPACITY ASSURANCE STRATEGIES Cease)
‘Thesimpl statement above uses a number ofa,
1 Which fates do we predict Refer to the objective of
‘maintenance “edi ale” and we "Cost faite ab
the peimary determing factor, defined
ee
aeease
stearate
Penalty Costs, on Costs, Fines
PAY}
ered bation (inion ed
seg rere rex) ne know tt an
eget tae pay Ise repair cos Sey
etonnentl ot poli brane ee
So nee toe cto ins ede os ipl are
cox report (ig that sey salneanes manages
‘Sold e o ihe ds ach mont
Fig 1 Amsintecance mine sbaldrectve Sipe
‘singl ilrcost rope every month
‘That ype of report daws ou attention to the overall
tof the faites eater than frequency and duration.
‘Bad Actor are redefined as"Bad Cost Actor”
ronune.com (23ay CAPACITY ASSURANCE STRATEGIES
2. How dowe measure resitancet flr! Thi has rest
‘heoreial importance—bul asa praca ise
‘complecand not wel understood. By sabttating perfor
mance” ta pron for"relstance oar the concept
‘Dxome simpler nd eae to underand As iltrated
in Fig therefore,» 1000-GPM pur has "aed if
it pars “ony” 999 gallons per minute (he required
amount being necesary for fedtock supp cooing
‘Purpose, etc). Thus an instance of functional fare
(FB) canbe gui defined and jut a ety recorded
‘on the work order,
4.2 Tho conoptt“portrncs” sour understand
than escent alre”
_2.Asperformance slides down the slippery scope of the PF
‘ate, the point of acteration in the tte of dreds
tone fen clearly apazetin practic, thus suggerting
the PF pla Acne can dene « ec
‘condition value forthe P pont (nthe Fig 2 ea
ite1100 gon per minute) Using ePaper
tion greatly eases data collection and analy: Equally
Important the fact hat the PF acs as warning signal
needing a malstenance response
What if the PF and FF points are not predicable?
‘Or what if theyre denial? Les use eletonie and
slectrieal equipment as an example. The PF and FF
points cleat exs—but are simultaneous, Condition
‘monitoring will not help excep o advise of complete
fare Here, we tespond wih standby uns, pug-out
gin placements nd sir techniques-(Note the
‘Sagar in Fig 3)
Stow does age Bt into the equation? As Nowlan and
Heap pointed outage deel impact fare in only 2
all umber of ser. et intitivey, we el hat ae
24) wanTenance TEcaNOLOGY
3. Conon nostro wil at blp when PE ad FF
nr apredcabie rental
an lnportant factor AS an aleratv lets define ia
"wothing ge” This has wo implications for poy
of failure: bad and stress neptve) and aut orice
Sa te Ln on dca eh
swe deta to operating hous hgh equls tot time
‘minus outof-service time). Thi" equires us to record
“pensions” on the work ble (aba alteratie @ 2
ProranF. 9
6 Hom dower the multe sears ofCM dato the
we nse Proprtonl Hats Moding
ee tcl showing which veal bare
rece ipa on th fate ode ih
te ie pone Techie it no EXAKT,
1 proc devloped by Dr Aten Jasin tthe
‘aie oan Repent we of hs tool a
Shr mot CM dat a le norton he
Incidence of fle and th, ame gore
preicorSck dt doesnt wed colin ya
‘chr wrtng og nd oe coin veri re
frequen nising
7am predicting fale, the predictive ability of Co data
‘st be acre and content. EXAKT achies this
by providing probaly of fallure in «given period
(Completion of» mission, prioe to « mainunance
shutdown, ete), and applying statistical test showing
‘onfdenclevels Relating the thee elements of flare
probability confidence levels and cot of falure proves
Song insght into the “bes” maintenance tac 12
follow Low confidence levels prompt both conservative
sexion (6 preempt the FF point) and enhanced data
‘collection, specially when the cos flares high
‘Sepreween 20aa CAPACITY ASSURANCE STRATEGIES eat
oe
These shortcomings of CM asa trv predictor of file
rrompt the development of a beter apposch. The
‘gueedoutput bimproved eelabityanalsrbt here
tre akeny many eee ella tol on the
‘aah Whats the mising a? Let lt relablity
‘nbase—olleting and holding the many data inputs
randy for the analy
Relablydetabose
Key to he elblty database oe the sources of data Here
is aproposed structure:
1. Historical data—primaiy rom the CMMS—but with
smple modifications tothe work ode to accommodate
themising FR, PFand Suspersiondata Also asexplained
er thie ade a cos eference tothe appropriate BEM
htabase record.
2 Garren status data—primaiy from the CM sensors
“These il ie us (long with PLCs SCADA and others)
the beating nto the current equipment condition.
ta. These willl ws which fares should
expected bated on equipment asessment
‘operating coer asin the RCM database
“p accommodate these data sources, Living RCM
(LRCM) software (a shown in Fig 4) as been devel-
opel I links the dita soures and acts a a data tac
‘op cole, cleansing and toring the dat create «
relly database. This the feedstock for commer
rely tool (sch s EXAKT and other). >
Fi, Ug REM software inks data sore an ac ade
tre cop, collecting clenig ad trig data to crete #
reli databan
Using logic, statistics and
‘well-accepted techniques, we can
improve maintenance decision-making.
ater, weliked the RCM and CMMS databases. These
are lay complementary in prompting a beter snder
nding of fare and reablty. Contrary t common
practic, the best output ofan RCM anal snot sty
tomes in the engnering office, bu an Improved work
order Sars attctoryoutps of «work order ean
improved RCM recorder the work ads new nowt
‘ge oF anew flue mae Looking atthe activity flow in
ig 5, the advantage of inegrating CMMS and RCM are
lie eident:
1. Inspection romp identi of
til fares
» ee rec (ert,
an on-thespot emi reventive action}
22a Neely deed prewntecn
ph ae ete
ion failure, we need to challenge the value of
mw
4 LRCM prompts the tedtnician completing the work
‘order not to use the pial Fault Code (he vale of
thick is highly questionable, and in our experience
‘arly ued), bt rather te acces the lure mode inthe
ROM database and insert in the work order.
“significant” tak alde to our knowledge (eg.
new fire mode or new eee compared tothe RCM
record), «temporary record is erated by LRCM and
awaits validation by the ECM analysis team
(6 Add the RCM record number to record the occurrence
and frequency ofthe REM lure mode—e very vale
ahleanalysisaid
7. Binal an unexpected ocarrence of failure mode in
rita equipment demands several responses therepie
ofthe equipment the repair ofthe REM record AND
the epi ofthe RCM logic AND all the other records
that used the sume log ase of access the RCM dts
‘base om the CMMS thes becomes crcl to creating s
regime of Living Reliab,CAPACITY ASSURANCE STRATEGIES
oes maitonancs improvement actully happen?
The bottom linet Does application of the techniques
described here ad vale —doritimprove deo making?
Here are sever indore .
Ue rep. The cost model optimies the eStbination
‘of preventive work and runt allure, compared to the
‘current mie of maintenance tactics A second modeling
‘option provides the optimum balance of PM and run
{o fade (RT) to achieve the minimum downtime or
(eodel thre) to achieve a given minima evel of rel
bility. Indatral experience shows cot redacsone of,
‘shout 20% to 46 of curent maintenance cost ung
{he customer cost at athe buen Subetential cost
reduction certains achierable
21 the qualy of maintenance improved? Refer to one
‘of RCM fandamentas A hey insight isthe ue of PFs
to prevent Fema easly measurable and sfchacking
[PL hich sb int the nai program would bes
‘emarlable improvement if our vibration sna or i
tnalyis programs could tel ur whether they ae doing
‘hai job propery Or not
26| MANTENANCE TECHNOLOGY
Fi. § The advantages ligating CMM and RCM ae gue evident.
2. Von ming prod neve hi
"lade flowing bese gk
Sorrow pay o
eo eT ees ee
downgncnatca tele ena”
un Risk = Cos of Failure x Probability of Failure
‘Nox, calculate the PM cost (sing paral! loge, but
ferent cot number) ines the probably of doing
‘the PM (whichis 10%), We an then define he"PM.
Rik”
‘PM Risk = PM Cost x 100%
«By comparing the Rn Risk the PM Risk we can
deep Rain
isk Ratio = Ratio of Run Risk to PM Risk
‘Wie can now decide whether «$40,000 investment in 2
PM to avoid the Rn Risk of $200,00 (comprising 22336
probabil of an $800,00 flare) is» good decision —a
‘Rk Ratio of 51. Or, should we spend $360.000 to eli
SEPTEMBER 20085) eri ___CAPACITY ASSURANCE smearzcs A
; The result is better selection of maintenance tactics, improved
| equipment reliability and increased company value.
‘aethe 15% probabiityofariskot1 33 milion falure—a__Adalogialnecatep wecanetabishif tek Ra (atis +
Risk Rati of 1.1 (se Fy 6) tisappzent tat thehigher ce) litte npr ns poi Inadidon, ||
the Rsk Ratio the grever the PM's leverage in reducing by racking change inthe Reka troughtine (Fg iwe ||
| tisk Likewise the highe the PMS ROL, the more value ster the ne cede hutdown o end fhe on,
eed othe company ‘whether th Rk ai rend eens opting Frames
mnie —
ina teponer = |e ae (og) SE 5S i
faeces ae sleneene peal |
efi ik Ue mi cet Tine Oa. fet |
mie
SoS A mo om el] |
vali adtaa te
fenpem.
2K days 1 ame TOS
BE ne
eer
PMI SENSORS
Taltrontush atte
Heer tne
We stand! hehind ALL of our protucts!
SEPTEMBER ate MronuNe.com 127DR caraciry
In other words, if it does net add value, don't do it.
a
ia
‘Shutdown Date en |
fl B
: Voces |
ae
i> busines lope leads to msntenance improvements
tod heed elgg: cot: Pong an cbeie name of ba
new ks ofgarong fundaton for improved decor ang: Eh
Sep ib saighforvard and ean be plemented wih nial cane oan
(ching Sade pos: Moret wh each Fen we pt coe ose i
“hats to done Garb that maimeance says prove of
baie MT
‘Ben Stevens is present of OMDEC Inc, bused in Godfey, ON, Canada
OMDEC (Optimel Maintenance Decions Inc) provider axet management
‘onlin taining and sofware solutions for Cents around tke globe. E-mail
‘ecereacn sce