System Management
Chapter 65
65.1 Maintenance
65.2 Support
65.3 Obsolescence
65.4 Replacement
65.5 Upgrades
65.6 Comments
Having commissioned a control system, the em- up in spares which are doing nothing. Use of pre-
phasis shifts to its operation. This chapter there- ferred suppliers for purchasing purposes has the
fore covers system maintenance and support, and very desirable effect of reducing the number of
concludes by consideration of aspects of upgrade types of instrument for which spares have to be
and replacement. held. Standardising on certain models of instru-
mentation also reduces the stockholding.
Whereas most instruments themselves require
65.1 Maintenance little in the way of maintenance, or none, their
This essentially concerns the field instrumentation process interfaces often require extensive mainte-
and the hardware of the control system.Modern in- nance.Particularly notoriousare impulse lines and
strumentation is mostly electronic: self diagnostics sampling systems.As discussed in Chapters 11 and
and remote calibration, either by means of fieldbus 12, for example, there is much scope for distortion
or otherwise, are becoming the norm. With three of signals due to unwanted bubbles of condensate
obvious exceptions,the amount of maintenance re- and/or air in the impulse lines, blockages due to
quired is fairly minimal. Those exceptions are in- deposition of solids in sample lines, failure of air
struments used in protection systems, for which or nitrogen supplies for dip legs, and so on. These
proof testing at specified intervals is mandatory, issues can only be addressed from a knowledge of
analytical measurements and control valves. which instruments are problematic and instigating
From a reliability point of view, if an instru- preventive maintenance at appropriate intervals.
ment survives its first few hours of operation suc- Analytical instrumentation is well known for
cessfully then its subsequent operation is likely to requiring disproportionate amounts of routine
be only subject to random failure of its component maintenance. Many in-line measurements involve
parts. The key issue then is having available suf- direct contact of an electrochemical sensor with
ficient spares to enable replacement in the event the process fluid. There is clearly scope for the
of failure. It is normal practice for companies to sensor becoming dirty or contaminated. Often the
maintain a stock of spares. Given the number of only feasible solution is to use disposable sensors
different types and/or size of instrument, careful and to replace them regularly. Analytical measure-
thought needs to be given to the spares inventory. ments usually involve use of a reference electrode:
Otherwise a substantial amount of capital is tied the standard solutions contained therein degrade
526 65 System Management
with time and have to be refreshed on a regular the necessary skills, and keeping a sufficient stock
basis. Also, the transmitters associated with ana- of spares such as I/O cards. System diagnostics en-
lytical measurements are prone to drift, hystere- able hardware faults to be readily identified and,
sis, etc., and are in constant need of recalibration, in general, they can be quickly resolved by swap-
although modern transmitters are more robust. ping cards, boards, etc. from stock. It is crucial that
Sampling systems for off-line analysers, such as the faulty ones are returned to the supplier and re-
chromatographs, require even more maintenance: placed with new to ensure the stock of spares is
they usually have filters requiring regular replace- functional.
ment, carrier gases or liquids which need replen-
ishing, and so on.
Control valves, with their moving parts and
pneumatic positioners,require some form planned
65.2 Support
maintenance.The most common problem is loss of The final act of acceptance should be the hand-
accuracy in the range and zero calibration due to ing over by the supplier to the end-user of a com-
wear, partly of the actuator but mostly of the po- plete and up to date set of documentation, or else
sitioner. Continual movement of the stem causes of those parts that haven’t been handed over al-
the packing to lose its tightness resulting in leak- ready. The end-user must keep them up to date so
age.Also, due to vibration and shock, other devices an effective change control policy must be insti-
such as limit switches and alarm contacts work tuted addressing the changes both to the applica-
loose. The plug and seat are subject to erosion un- tion software itself and to the related documenta-
der extreme flow conditions and, in time, will lose tion. In the interests of continuity and consistency,
their characteristic and shut off capability. For all it makes sense to use the same policy and modifi-
these reasons, control valves require regular and cation control form as used during commissioning
routine maintenance. The advent of smart valves and described in Chapter 64.
will not obviate the need for maintenance. During the life of a control system there will in-
As far as the hardware of the control system evitably be many changes made for process reasons
itself is concerned, it is normal practice to have such as improvements to operability and safety, ex-
a maintenance contract with the system supplier. tensions to the plant, addition of new products, etc.
There are different types of maintenance contract There will also be changes made for control rea-
according to the level of support required. The two sons: enhancements to the control strategies,intro-
key issues are callout time and provision of spares. duction of optimisation functions, etc. It is impor-
The callout time is essentially the supplier’s guar- tant that the plant management should recognise
anteed speed of response, i.e. the maximum time the value of on-going enhancement and accept the
it takes the supplier to get an engineer onto site to need for deploying control engineering resource to
repair faults. The callout time required depends on support these changes.
the financial impact of a disruption to production. That resource essentially consists of one or
The shorter the callout time the more expensive more engineers,who have relevant applications ex-
the maintenance contract. pertise and experience of using the system, whose
Short callout times can be avoided if the end- time is allocated wholly or in part to system en-
user’s engineers are capable of doing some hard- hancement. Obviously that resource includes the
ware maintenance themselves. This typically in- necessary development environment and test tools
volves them attending a few basic hardware main- too.
tenance courses provided by the supplier to gain
65.3 Obsolescence 527
65.3 Obsolescence plier for existing systems and, usually, a repairs
only service is offered.
A key long-term decision is whether to allow the • A supplier will typically commit to supporting a
control system to become obsolete or to attempt to
system for up to about ten years from the end of
upgrade it incrementally over the years. Central to
the active sales phase. The system is then said to
this is an understanding of the product cycle of a
be obsolescent, as opposed to obsolete which is
control system. This is depicted in Figure 65.1:
what it becomes at the end of the limited phase.
• Generally, as one system enters its classic phase
• Once a system has been developed and released of life, the supplier releases the next generation
for sale it enters an active sales phase. During of system which enters its active sales phase.This
this period, both for commercial purposes and is known as system evolution.
the need to maintain a competitive edge,the sup-
plier is likely to make various upgrades to the Whereas the production cycle of a system is some
hardware and enhancements to the system soft- twenty years,the service life is more likely to be 10–
ware. 12 years.That is because the hardware becomes un-
• At some stage, typically ten years after the sys- reliable, maintenance costs start to rise and the po-
tem was first released,users are given six months tential for loss of production increases. Also, there
notice that the system is to be withdrawn from is the difficulty, as a system approaches the end of
production,whereupon it enters its classic phase its service life, of finding the resource to support
of life. The classic phase lasts up to some seven it. Nobody wants to work with old hardware plat-
years during which there is minimal investment forms using historic languages and obsolescent op-
in the system by the supplier. erating systems.
• Towards the end of the classic phase, users are Given that the average life of a process plant
given a “last chance to buy” before the system is typically 30 years or more, it can be anticipated
enters its limited phase of life. During this phase that any plant’s control system will be replaced at
it is not possible to buy new parts from the sup- least twice in the life of a plant. With the excep-
Phase
Limited
Removed from
active sales Obsolescent Obsolete
Classic
Releas ed
for sale
Last buy
Active
Time (years)
0 5 10 15 20
Fig. 65.1 Control system product cycle
528 65 System Management
tion of the field instrumentation, which tends to tomatic conversion of code from ancient to mod-
be replaced as and when necessary, such replace- ern. The advent of the IEC 61131 and IEC 61512
ments are generally complete: new I/O and other standards, with their formalised physical and pro-
cards, racks and frames, processors, operator sta- cedural models, should lead to automatic transla-
tions and peripherals. tion becoming more effective and widespread.
Another key decision is whether the replace-
ment process should be cold or hot. Cold implies
that the whole system is replaced at once, typically
65.4 Replacement during a plant shut-down, whereas hot implies re-
Replacement is the lazy option and, in the short placement on a loop by loop basis whilst the plant
term, is clearly the cheapest option. This may is still running. This clearly depends on how criti-
be a conscious decision or happen by default. In cal the process is from both production and safety
essence, it means doing nothing, allowing the sys- points of view.
tem to drift into obsolescence, and then to replac-
ing the whole system lock stock and barrel at the
end of its effective service life.
Replacement can be considered to be a repeat
65.5 Upgrades
of the system life cycle, starting with an URS, pro- The alternative to replacement is a policy of con-
gressing through a DFS,and so on.The same rigour tinuous,incremental upgrading.This requires con-
should be applied.However,provided that the orig- siderable effort on an on-going basis. It is certainly
inal DFS has been kept up to date, there is an ex- more expensive in the short term and may even be
isting documentary basis for articulating the re- so in the long term, but does have the very obvi-
quirements of the new system. The effort involved ous attraction of not having to find the capital to
must be substantially less. replace the whole system all at once. Also, because
Particular care should be taken to ensure that the system is up to date, finding the resource for its
the new system software can support the function- maintenance is not as problematic.
ality of the existing system.In practice,the replace- Throughout the active sales phase, the supplier
ment system is likely to offer more but different will enhance and redesign the system’s hardware
functionality than the old. This should be carefully and system software and issue new releases. These
considered and taken advantage of as appropriate. will:
A crucial consideration is the portability of the
• Involve processor upgrades
application software. Many person-years of effort
• Provide additional functionality which may not
will have been invested in developing and refining be required, but could be taken advantage of
it.In many respects the application software encap-
• Require modifications to existing application
sulates all the existing knowledge about the oper-
software
ability and safety, if not the viability, of the plant.
It is essential that some means exists to transfer These releases provide the end user with a migra-
this from the old system to the new. As far as the tion path for an existing control system to evolve
database and configurable software is concerned, towards the next generation. That is the essence of
transfer should be relatively straightforward pro- the upgrade strategy as depicted in Figure 65.2.
vided the function blocks offer the same or similar The upgrade path is truly incremental. It is not
functionality. necessary, or indeed desirable, to realise every new
However, procedural software is a different release because some are only minor and can be
matter. Suppliers must provide software tools bundled into more significant upgrades. Each up-
which enable upgrade routes. The conversion may grade needs to be planned and treated as a project
not be complete but should realise at least 80% au- in its own right for which the original, butupdated,
65.6 Comments 529
Upgrade path which, if not addressed, will result in partial obso-
lescence. That is worse than either of the replace-
Ideal ment or upgrade strategies as the inevitable out-
come is replacement but some of the costs of up-
grading have been incurred too.
Actual
65.6 Comments
Active Classic Limited As stated, the replacement strategy is the easy op-
tion and cheapest in the short term at least. It is
Time (years)
undoubtedly the preferred option of the accoun-
Fig. 65.2 Migration path for system upgrades tants. The upgrade strategy is the preferred option
of the suppliers as it reduces the number of previ-
DFS documents are the basis. There may be a some ous releases the supplier has to maintain. Equally,
re-design work but the project would normally it is in the end users interest to avoid obsolescence.
consist largely of the usual stages of development, There are clearly pros and cons. However, if an up-
testing, commissioning and acceptance. For appli- grade strategy is to be adopted, a policy decision
cation software, the same issues about portability is needed at the outset so that the cost of upgrad-
apply as were discussed under replacement. ing can be budgeted for. Also, there is a potential
Inspection of Figure 65.2 reveals two upgrade trap with the upgrade strategy: it can lead to an
paths: the ideal and the actual.The ideal implies re- up to date version of an old system and, inevitably,
alisation of significant upgrades in a timely man- to replacement. Confidence, by the end user, that
ner. The actual depicts the outcome of a lack of the migration path does indeed enable evolution
commitment: too little and too late. A gap emerges to the next generation is therefore essential.
