Sei sulla pagina 1di 8

Optimizing shutdown intervals yields significant savings

Introduction
For many industries planned shutdowns planning assumption. A self-perpetuating
or outages are a fact of life. Whilst such cycle often develops as a result:
shutdowns are often essential to ensure
system safety and/or regulatory “We have a shutdown every (x) year(s)
compliance, from an operational because we have so much work to do…”
perspective they are undesirable because AND/OR
they cause loss of system availability,
“We have to use the opportunity of the
incur significant planning, coordination (annual) shutdowns to do as much work
control overheads and often introduce as possible while we can….”.
system instabilities and recommissioning
problems during the start-up phase. Shutdown Interval Optimisation
at SABIC Innovative Plastics
A lot of good work has been done in the
field of shutdown planning and execution, SABIC Innovative Plastics (SABIC IP) is part
achieving reductions in direct costs and of the Saudi Basic Industries Corporation,
shutdown durations. However, the core and has four polycarbonate production
challenge facing Asset Managers is in facilities in Europe and the USA. Each site
answering the question: had an annual or bi-annual shutdown for
inspections, maintenance and engineering
When (or how often) should I shutdown projects.
in the first place?
In April 2008, following a review of
This is not an easy question to answer. possible processes and available tools, the
The key role of any Asset Manager is to management team decided to partner
optimise the costs, risks and performance with The Woodhouse Partnership Ltd
of owning and operating the assets over (TWPL) in carrying out a Shutdown
their whole life cycle – this includes Optimisation Project across the four sites,
minimising the ‘Total Business Impact’1 all with the first site being Bergen op Zoom,
expenditures, risks and performance ‘lost The Netherlands. The project received
(performance shortfall, quality losses, senior management support with a goal
downtime losses etc). With big money at of significant cost reduction and
stake, many (sometimes conflicting) performance improvement.
regulatory obligations and uncertain risks,
such goals are difficult to identify, let Shutdown Interval Optimisation
alone demonstrate in their attainment. in Bergen op Zoom
Shutdown timing or intervals, in
particular, have often become established The polycarbonate manufacturing facility
with no real strategic thought process at Bergen op Zoom is summarised in
beyond the next single occasion or a cyclic Figure 1, with the primary production
units that were the subject of the study
circled in red:
1
See www.MACROproject.org and
optimisation guidance in PAS 55-2:2008

© The Woodhouse Partnership Ltd SABIC case study Page 1


Bergen Op Zoom: TAR-critical unit groupings
External sales

BPA 2
Phenol
Acetone BPA 1

Salt Chlorine Phosgene Resin Compounding

Brine Rec.

Utilities Power
Nat. Gas (Cogen) External sales
Steam
CO
(Air Liquide)
Pipeline

Figure. 1 Polycarbonate Process Flow: BoZ

Project Methodology
The first issue addressed was to define were first evaluated individually, and then
“what constitutes a shutdown?” For this collectively, using the leading edge APT
project a shutdown was defined as: software tools for optimal timing and
‘bundling’. This methodology is shown in
“A planned event (to execute a pre- Figure 2 below:
defined asset maintenance,
inspection or engineering workscope) The first step in the process is to select,
that reduces plant output by 50% from the entire workload, only those tasks
production for a period of at least 24 which truly drive the requirement to shut
hrs (or equivalent).” down the plant. This was achieved
through a series of screening and filtering
This definition was essential in order to
processes. The selective filters ranged
filter out activities that do not really need
from a simple rule-based screening to
a major outage.
assessment and challenge by
multidisciplinary teams. The results of the
The TWPL Criticality Method was then
screening and filtering process are shown
calibrated to assess the impact and
in Table 1.
potential urgency of each task that
appeared to require a shutdown of the
plant. These were then challenged and
filtered down to a list of the truly
shutdown-dependent activities, which

© The Woodhouse Partnership Ltd SABIC case study Page 2


Shutdown
Optimization
Overall Process

All
possible Select Optimize De-
Analyse Re-optimize Optimized
tasks to Shutdown Shutdown bottleneck
Individual shutdown shutdown
be Critical bundles projects &
Critical Tasks scenarios strategy
considere Tasks of tasks actions
d

Tools used: Screening APT APT Assess value: APT


Filtering & -INSPECTION -SCHEDULE APT-PROJECT -SCHEDULE
Assessment -PROJECT
-MAINTENANCE
-LIFESPAN

Figure 2 Shutdown Optimisation Methodology & Tools

Table 1 Filtering & Screening Results

From the table it will be noted that from a process, so APT2 software tools were
total of 24,260 tasks, 7,875 were listed as employed to model the risk assumptions
‘shutdown dependent’ yet only 86 were and perform extensive “what if?” studies
actually found to be truly shutdown (given the inevitable lack of hard data).
dependent. This result showed the extent
of the shutdown cycle self-perpetuating When analysing optimal intervals we
habit described in the introduction above. needed to be able to put a value to all
direct and indirect costs. This is
Analysis of Individual Tasks particularly difficult when assigning value
The next step was to determine the to the consequences of failure, especially
optimal interval for each individual task.
This can be a complex and difficult
2
See www.decisionsupporttools.com

© The Woodhouse Partnership Ltd SABIC case study Page 3


when dealing with health and safety, A typical shutdown-dependent task is the
environmental or reputational cleaning of the Phenol “vent header”
consequences. This difficulty is pipe. This task is required because fouling
compounded when there is uncertainty of of the pipe reduces process efficiency.
associated assumptions. Great care was Currently the task is carried out every 2
therefore taken by the study facilitators to years. The task was analysed using the
extract tacit knowledge from the multi- APT-MAINTENANCE module. The Total
disciplinary teams participating. Range- Impact results for different task intervals,
estimating techniques were used to set with ‘optimistic’ and ‘pessimistic’
the bounds of possibility for uncertain extremes of uncertainty, are shown in
data. Where this uncertainty caused the Figure 3 below.
optimum to vary considerably, the
quantified financial implications provided The figure shows that, between
the motivation to search for more robust pessimistic extreme and optimistic
and accurate data or additional expertise. extreme, the optimal cleaning interval
varies little (24-30 months) and a 4-year
Following individual task analysis, many of cycle would be much more costly in terms
the tasks revealed optimal intervals equal of performance constraints. Furthermore,
to, or greater than a 4 year ‘ideal target’ the small variation in costs at the optimal
interval. Others, however, showed interval showed there was no financial
optimal intervals which were less than 4 motivation to collect more accurate data.
years – representing ‘bottlenecks’ to the
achievement of longer shutdown Since the optimum interval for this task is
intervals. An example of such a task is less than the desired site interval of 4
detailed below. years, we then evaluated the net cost of
the ‘bottleneck’. This is shown in Figure 4.
Example Task Analysis: Vent Figure 4 shows the contributing factors in
the ‘most likely’ scenario.
Pipe Cleaning
Sensitivity testing to extremes of assumptions

Range for optimum


interval shown to
be 24-30 months

Figure 3 Range Estimates Vent Header Cleaning

© The Woodhouse Partnership Ltd SABIC case study Page 4


In this case, the optimum interval is at 30 optima to share downtime or other
months, and the difference in total impact bundling advantages with other tasks.
incurred by delaying to 48 months, is
approximately €1,200/month or
€14,400/year.
4 yrs = target interval
for other tasks

Net consequences
of delay

APT-SCHEDULE does this in “real time” in


Figure 4 impact of Delaying Vent Header Cleaning
just a few minutes by utilising self-
Optimising Bundles of Tasks learning “genetic algorithm”, normal
The next step was to assess the best way simulation techniques would take around
APT-MAINTENANCE © Decision Support Tools Ltd
of combining tasks with different optimal 4 weeks to ‘solve’ the optimal scheduling
intervals into the ‘least compromise’ of just 10 tasks. There were 32 cost/risk-
groupings and shared timings. To conduct and timing-sensitive tasks carried forward
this extremely complex task the APT- to this ‘optimal bundling and scheduling’
SCHEDULE software tool was employed. stage, the results of which are shown in
This module explores the effects of Figure 5.
moving individual tasks away from their

Activities forcing
short-cycle shutdown
work

Figure 5 APT-SCHEDULE initial results

© The Woodhouse Partnership Ltd SABIC case study Page 5


The results show that there were a production impact of this programme
number of tasks which still required and (and any alternative).
justified a frequent (annual or biannual) So the full impact of the changes was
shutdown. The next step was therefore directly quantifiable – and represented 7-
to “de-bottleneck” or remove the figure annualised net benefits. With the
requirement for these high frequency sensitivity testing already built in, and full
shutdown dependent tasks. Table 2 lists ‘drill-down’ clarity about which activities
some of the 15 de-bottlenecking actions contribute how much to the urgency and
identified for the Bergen op Zoom site; in justification for each shutdown, SABIC
most cases they represented very small senior management were able to commit
investments, such as purchase of a spare to the revised plan very rapidly, and all
equipment item to enable ‘hot swapping’ de-bottlenecking actions were approved
during normal operations. and underway within 2 months of the
study. The Bergen op Zoom site is thus
changing to the new, 4-yearly shutdown

Study Describtion ref. Determining PM Debottleneck Possible sollution Anticipated cost


Failure mode Frequency timing Keuro
needed
CHL-041 GRP- Anolyte Knock-out V14124 De-lamination of GRP 12 Month ASAP Order spare unit € 450
drum
CHL-045 GRP- Suction tank II , V1471 De-lamination of GRP 12 Month ASAP Order spare unit , € 180
caustic service

BPA1-001 Acid removal collumn V510 Reduction in 24 Month re-clad within 30 replace cladding of vessel € 150
wallthickness Month
BPA2-046 Tar vent Pipe - cleaning SW 2150 Fauling lost efficiency 18 month WC at opportunity on- line / 50 % improvement € 60
before next
cleaning window
BPA2-021 Presure controll valve N2 PCV If it is too low we stop 12 month WC at opportunity Spare unit at stock €5
service to Flaker line 1, 21502120/ the system ( line 1
planned inspection 1 flaker), when the flow is
too high we spill N2
BPA2-024 Phenolic Recycle water XC Fauling lost efficiency 6 month WC At opportunity / Spare unit at stock €5
breather, service on 21107103 eliminate root
V21170, planned Cause root
inspection

Table 2 De-bottlenecking actions

Having identified the de-bottlenecking cycle as soon as these actions are


tasks and removed them from the task completed.
list, the APT-SCHEDULE analysis was re-
Next Steps
run. The results of this are shown in
Figure 6 (below). The Bergen op Zoom project proved a
This study revealed that the optimum great success and the process is now
shutdown interval is now 4 years. The being rolled out to the other SABIC IP sites
APT-SCHEDULE outputs included a full with the support of TWPL facilitators.
NPV calculation of the costs, risks and

© The Woodhouse Partnership Ltd SABIC case study Page 6


Figure 6 De-bottlenecked Schedule

Lessons Learned
When left unchallenged, shutdowns and participants to the study. The structured
their intervals become a self-fulfilling range estimating and sensitivity analysis
prophecy. Where this is the case, it is techniques were also major factors in
likely that the organisation’s business obtaining confidence and credibility in the
processes will have evolved in order to analysis results.
support the pattern, concentrating on
efficient delivery of the work rather than The credibility of the APT toolkit also
challenging its need and timing helped: in contrast to ‘decision support’
that concentrates mostly on data analysis,
Challenging the shutdown dependency of
APT was the output of the European
any task is essential, but can only be
MACRO R&D project that focussed on
successfully done by competent
quantification of cost/risk trade-off when
multidisciplinary teams and full cost/risk
data is not available. They also enabled
appraisal, with appropriate decision-
very comprehensive modelling of
support tools and robust methods for
multiple, interacting risks to determine
coping with the inevitable uncertainties
the best compromise position.
and poor information.
As with all change programmes the
participation of all stakeholders was
crucial. In particular, the safety-critical
areas of inspection and Process Safety
Management (PSM) were vital

© The Woodhouse Partnership Ltd SABIC case study Page 7


For more information
Alex Thomson
Principal Consultant
The Woodhouse Partnership Ltd
E: alex.thomson@twpl.com
T: + 44 (0) 1635 298800

Julie Fowler
Marketing Communications
The Woodhouse Partnership Ltd
E: Julie.fowler@twpl.com
T: + 44 (0) 1635 298800

www.sabic.com
www.twpl.com

© The Woodhouse Partnership Ltd SABIC case study Page 8

Potrebbero piacerti anche