Vous êtes sur la page 1sur 2

42 | ASSET MANAGEMENT

How predictive analytics can


optimise asset maintenance
BADRINATH SETLUR explains how to use data to address key strategic objectives.


C
ollecting and distilling digital information Oil change: This involves oil replacement at a xed
and extracting meaning from it holds great schedule because real-time oil quality analysis is not
potential to enhance customer satisfaction, possible.
reduce total cost of ownership, optimise Intercooler cleaning: This involves intercooler cleaning
resources and improve compliance. at a xed schedule because real-time analysis of coolant
Little wonder then that predictive analytics a process and intercooler eciency is not available, even though inlet
of using statistical and data mining techniques to analyse and outlet temperatures are measured.
historic and current data sets, create rules and predictive Bearing replacement: This involves bearing
models, and predict future events is fast becoming a vital replacement at a xed frequency because no scientic
instrument to realise asset life cycle cost reduction and way to determine bearing life left is deployed even
improve the speed and accuracy of decision-making. though lubrication oil temperature, viscosity and bearing
vibrations are measured regularly.
Predictive analytics for assets Major overhaul: This involves a major overhaul of the
Assets are seldom standalone. They exist as a system of assets equipment at a xed frequency, usually more than that of
where they feed o each other. Between asset procurement/ the aforementioned procedures, because the manufacturer
commissioning and decommissioning/salvage lies the has prescribed it, even if all vital parameters look all right.
productive life of an asset. Regular upkeep or maintenance is All this entails avoidable downtimes, unnecessary
needed to maximise this life. There are two perspectives on maintenance expenses, lots of collected yet unutilised data,
how predictive analytics can help optimise asset maintenance: additional Capex (capital expenditure) for back-up equipment
Individual equipment perspective: Typically, and complexity of operations due to the need to accommodate
maintenance frequency is dened based on various multiple downtimes, coordinate with downstream and
parameters such as asset age, asset criticality, operating upstream customers, and so on.
environments, risk of failure and so forth. As a result, more That can cost a lot of money; however, the maintenance
frequent maintenance would mean higher maintenance requirements are real. Operating equipment close to failure
expenses. Often the risk of failure forces over-maintenance thresholds may be riskier than we think. Why then are we
of assets. Over-maintaining can mean replacing lubricants talking about predictive analytics? Primarily, predictive
or bearings that still have life left. The ability to predict a analytics can help to determine failure thresholds, identify
failure can help move maintenance activities closer to the buffer zones and quantify risks in operating equipment
real need for maintenance and reduce over-maintenance. within known buffer zones. Thats also how predictive
Productivity perspective: As can be imagined, a analytics can help optimise equipment maintenance.
system of assets, designed and scheduled for maximum
productivity, can benet immensely if a failure of an event Power of predictive analytics
requiring maintenance is known in advance. Alternative Let us reimagine the same scenario with predictive analytics
schedules or plans can be prepared to ensure maintaining capability. Predictability does not mean the maintenance
productivity levels. software will tell you to replace a certain bearing on a certain
Ironically, in almost all common maintenance scenarios, piece of equipment in a certain process on a certain date at a
the ability to predict failures and move maintenance certain time.
procedures closer to failure exists. It is just that the risk of But it would indicate the probability (for example, there
failure keeps many organisations from taking those steps. is 89 percent chance) of a bearing failing at a given time.
Let us consider a multi-stage centrifugal air compressor. Business rules can be built to instruct a bearing replacement
Some of the typical maintenance procedures carried out today above a certain threshold (for example, when there is a more
on centrifugal equipment include: than 70 percent probability of failure). Evidently, this allows

FM AUGUST | SEPTEMBER 2015 www.fmmagazine.com.au

41959_42-43_asset manage.indd 42 10/07/2015 12:09 pm


risks one is willing to take to be quantied (for example, in time. For example, there is an 88 percent chance that
70 percent is the risk threshold). the product will operate successfully after three years of
Oil change: Oil quality over a period of time can be operation.
plotted to predict with a degree of condence what the oil Probability of failure given time: This indicates the
quality (oil viscosity or decomposition) will be at a given probability that a unit will fail at a particular point in
time. This can then help to answer the question: is there time. This is also known as unreliability. For example,
a signicant dierence in the quality of oil supplied by a 12 percent chance that the unit could potentially fail
dierent vendors? after three years of operation (probability of failure or
Intercooler cleaning: Temperature delta between unreliability) is the same as an 88 percent chance that it
inlet and outlet for coolant water and air is a good way to will operate successfully (reliability).
measure eciency. A lower delta would mean less heat Mean life: This indicates the average time that the units
taken away and lower eciency, assuming everything else in the population are expected to operate before failure.
is intact. This delta over a period of time can be plotted This metric is often referred to as mean time to failure
to determine statistically what the value will be at a (MTTF) or mean time before failure (MTBF).
given point of time and whether or not there is a need for B(X) life: This indicates the estimated time when the
maintenance. This can help gure a trend. probability of failure will reach a specied point (X
Bearing replacement: Predicting bearing life can be percent). For example, if 10 percent of the products are
tricky, as identical bearings tend to display dierent expected to fail by four years of operation, then the B(10)
endurance lives. Reliability of bearings has to be predicted life is four years.
based on the installed population of bearings and
analysing failure data to identify patterns. Predictive analytics and
Major overhaul: Something like this will be dicult to maintenance practices
predict statistically (or there may not be a need). But if Traditionally, maintenance practices are classied as reactive,
individual components such as intercoolers and bearings preventive, predictive and reliability-centred maintenance
are working ne, making a judgement or qualitative (RCM). However, predictive and RCM approaches are the ones
decision on the need for a major overhaul should be that best leverage predictive capabilities.
straightforward. Reactive maintenance: In this, practice equipment is
All this can help widen maintenance intervals, push allowed to run until it breaks. No actions or eorts are
maintenance activities closer to when needed and taken to maintain the equipment.
consequently, optimise asset operations. Preventive maintenance: This involves activities
carried out for the purpose of maintaining equipment in
Its not difficult or expensive satisfactory operating condition by systematic inspection,
Does building predictive analytics capability cost lots of detection and correction of failures either before they occur
time, money and eort? The answer to this big question is or before they develop into major defects.
an emphatic no. There is always a sweet spot where savings Predictive maintenance: In this, practice statistical
from predictive capabilities (reduction in maintenance techniques are used to determine the condition of in-service
expenses, Capex, spares utilisation and so on) outweigh the equipment in order to predict when maintenance should
cost of building those capabilities (software, human resources be performed. This approach promises cost-savings over
etc). Once done, the benets through direct asset-related routine, time-based preventive maintenance, because the
saving (fewer maintenance dollars, fewer spares, longer life tasks are performed only when warranted.
span), and organisational saving (optimal teams, increased RCM: This emphasises the use of predictive maintenance
operational eciencies) are immense. techniques in addition to traditional preventive measures.
When properly implemented, RCM provides companies
KPIs to monitor assets with a tool to achieve lowest asset Net Present Costs (NPC)
KPIs (key performance indicators) are dierent for dierent for a given level of performance and risk.
asset types, operating conditions, criticality and so forth. Due Predictive analytics is the process of moving from hindsight
consideration needs to be given to the role of statistics and to insight. While data and distilling data are the key, equally
the related process changes and change management, among important is the way in which organisations instrument,
others, required to build a predictive maintenance culture. capture, create and use data to address their strategic
Data and statistics will provide quantication, but will need objectives.
to be applied with business insights to derive benets (for
example, the 70 percent risk quantication mentioned earlier). The author, Badrinath Setlur, is assistant vice president
Here are a few commonly used KPIs: Consulting, Manufacturing, Logistics, Energy and Utilities at
Reliability given time: This indicates the probability Cognizant, a provider of information technology, consulting
that a unit will operate successfully at a particular point and business process outsourcing services.

www.fmmagazine.com.au AUGUST | SEPTEMBER 2015 FM

41959_42-43_asset manage.indd 43 10/07/2015 12:09 pm

Vous aimerez peut-être aussi