Reliability, Risk and Safety – Ale, Papazoglou & Zio (eds)

© 2010 Taylor & Francis Group, London, ISBN 978-0-415-60427-7

“RAM + L analysis: A case study in Brazilian refinery”

E. Calixto, Leonardo Carnerio & Adilson Correa Alves

Petrobras, Rio de Janeiro, Manaus, Brazil

ABSTRACT: The main objective of RAM analysis (Reliability, Availability and Maintainability) is
assessing equipment or system performance throughout critical equipment improvement in order to
achieve an availability target. To carry out RAM analysis it is necessary to define the equipment failure
modes which have the highest impact on system availability. The analysis is carried out using historical
failure data and repair time and simulation using a reliability diagram model model. Despite widespread
applicability of this methodology on large complex systems it is vitally important that logistics issues must
to be considered. There are two different approaches, the first one focuses on reliability issues and the
second one on logistic. At this time in Brazil there is no methodology which considers these two issues,
logistic and reliability in only one Methodology, in order to assess huge logistic system regarding reliabil-
ity issues of subsystems and equipments into logistic systems. In fact complex systems logistic analysis do
not take into account reliability issues and the other way rounds.
The RAM + L analysis methodology takes into account logistic and reliability issues in order to have
a more representative result to support improved decisions. The case study consists of a complex system
comprising refineries plants (Vacuum and Atmospheric Distillation Plant, Thermal Cracking Plant, Acid
Water Plant, Cracking Catalytic Plant, Reforming Catalytic Plant, Fractioning Plant, DEA, Nafta and
Diesel Hydrodesulphurization Plant) and Tanks will be carried out to assess advantages, drawbacks and
to compare RAM analysis with the results obtained using the RAM + L analysis.

Keywords: RAM Analysis, RAM + L Analysis, logistic, availability

1 INTRODUCTION A 1 2

The availability analysis of a system is essential to

1
verify the possible enhancements to be carried out
on critical sub-systems and equipment in order to B
maintain the system availability goal. Availability
is influenced by reliability and by maintainability, 2
making it essential to carry out an evaluation of fail-
ure and repair occurrences in order to identify the
most critical equipments with respect availability. Figure 1. Reliability block diagram.
In order to carry out RAM analysis is necessary In case A, reliability is represented as:
to specify the system’s borders and define scope.
R(t) = R(t)1 × R(t)2
This will require an evaluation of sub-systems,
equipment and components which failures repre- In case B, reliability will be:
sent environmental impacts, damages to personal
safety and physical damage, loss of production R(t) = 1 − ((1−R(t)1) × (1−R(t)2))
and system halt. The System Configuration will
comprise a set of blocks linked in series and in will impact the equipment’s availability which may
parallel as illustrated in Figure 1, shown below. be understood as the probability of certain equip-
The reliability of a system is the probability of ment being available for a length of time.
a system working without failure for a specified In general, the analysis of a system must take
period of time. Another important concept of reliability and maintainability into consideration,
availability analysis of a system is the maintain- since these two factors affect the availability of
ability, what means the probability of equipment a given system. It is possible to identify critical
being repaired in a specific period of time. That equipment in terms of maintenance and failure

8
through this analysis, so that managers can take After critical and sensitivity analysis the
the best decision to optimize availability with the conclusion will summarize critical equipments,
lowest cost. logistic resources and further vulnerabilities sys-
The principle objectives of a technical system are tem as well as improvements actions which must
to ensure the realization of continuous operation take in place in order to achieve high complex
process of its components. However, a population system performance.
of units (aircraft engine components, computer
modules, means of transport, etc.) that randomly
fail but are completely repairable requires and 2 Data procurement
effective maintenance infrastructure and logistic
system, that will be available when required. As A huge challenge to Brazilian Oil and Gas Industry
result, reliability and effectiveness of technical is to get good data to perform RAM analysis.
system, being worked in changeable environment, In order to ensure the reliable level of such data,
cannot be analyzed in isolation, without taking maintenance professionals with knowledge of the
into account the numerous links with its logistic these systems took part in this stage and a semi-
support system (Werbinska, 2007). quantitative analysis of failure and repair data is
This case study in a refinery was analyzed with carried out in some cases.
omitting all external factors to the unit, such as A equipment analysis into the causes of system
steam, cooling water and other influences it was downtimes requires failure modes were performed
assumed that all external resources are 100% avail- and identified along time, it means all equipment
able. Failure and repair data of the unit was used failure modes responsible for most of the impacts
in this stage. in respective sub-systems. The failure data equip-
Moreover, System analysis will be carried ment is treated statically in order to define the best
out taking into account reliability availability, PFD (probability density function) which bests fit
maintainability and a complex logistic assessment the historical failure data and is necessary to have
which comprises such plants and regards further software’s support such analysis (weilbull 7 ++ reli-
logistics issues using RAM + L (Reliability, Main- asoft). As instance, is shows an example in Table 1
tainability, Availability + Logistic) methodology. below a Thermal cracking Furnace failures modes
Such methodology comprise phases like scope with each PFD and repair time.
definition, System RAM analysis, Logistic Statistical analysis was performed for more than
assumptions, RAM + L Analysis, critical analysis 200 equipment to enable direct simulation (Monte
sensitivity analysis, and conclusion. Carlo) to represent operation time in 3 years. The
Scope phase has main objective to define analysis coke formation is the most critical event in refiner-
boundaries in order to focus on defined systems ies plants. Coke formations is considered the most
and logistic resource avoiding delays caused for critical failure mode in the RBD modeling (Reli-
scope changes or increment. ability Block Diagram Block), but is considered
System RAM Analysis is required in order a process failure. The Figure 2 below summaries
to consider critical systems and their equipment RAM + L methodology.
which may take a high influence in whole Complex
system availability. In case of not required such
analysis is being performed a logistic analysis only. 3 SYSTEM MODELING
Logistic assumptions is required in order to
model logistic issues like tank levels, flow priori- To perform the availability results in Monte Carlo
ties in process, ships load or other modal capacity, simulation, it is necessary to set up a RBD model.
delays and stock levels.
After performed system RAM analysis and to Table 1. Failures and repair data.
get logistic assumption is possible to comprise all Failure time (years) Repair time (hours)

information and model complex system regarding T AG Failure mode

Variables (PDF) Variables (PDF)

both issues.
After simulation critical analysis are required Coke formation Normal
µ

4,95
ρ

2,66
Normal µ

420
ρ

60
and usually reliability, availability and utilization β η γ Normal µ ρ
will indicate which are the critical systems and F -01 A
Incrustation Weibull
0,51 1,05 4,05 420 60

logistic resources. Exponencial

λ γ Normal µ ρ

In some case, some situation which may increase Others failures

Bi p
0,28 3,22 420 60

system vulnerability was not took into consideration µ ρ Normal µ ρ

and it’s must to considered in model. The stock level, Coke formation Normal
5,23 2,55 420 60

energy supply and facilities unavailability are some F -01 B

Others failures
Exponencial λ γ Normal µ ρ

examples which may decrease system availability. Bi p

0,29 4,07 420 60

9
 1 – Scope Definition 3.2 Atmospheric and Vacuum Distillation
Plant(U-10)
2 – Data Analysis
The main objective of Vacuum Distillation plant is
3 – System RAM Analysis to change out heavy oil portion in light oil product.
Based in general assumptions, The RDB (Reliabil-
4 – Logistics Requirements Assessments
ity Diagram Block) of Vacuum Distillation Plant
5 – RAM+L Analysis comprise five blocks in series which represent
Feed, Desalter, Heating, Furnace, Atmospheric
5 – Critical Analysis and improvements Distillation and vacuum Distillation. That means
actions
if one of block fail down the whole RDB will be
6 – Sensitivity Analysis unavailable. Each subsystem represented for RDB
comprises several equipments with each PDF
7 – Conclusion
based in failures modes. The main assumptions to
make up RDB are:
Figure 2. RAM + L methodology.
• It’s not being regarded facilities availability and
other supply influence in U-10 availability;
Although the system is complex it was decided to • The equipment failures modes are based in fail-
use RBD (Reliability Diagram Block) methodol- ure historical data of own Plant from 2000 to
ogy. In order to perform Monte Carlo Simulation, 2010;
it is necessary to be familiar with the production • Subsystem unavailability represents system fail
flow data which influence losses in productivity. down;
Consequently, some statements and definitions • The average availability target is 98% in 5 years;
regarding process limitations. • Total Production per day is 5.600 m3/day.
In Figure 4 below does shown RDB which com-
3.1 Atmospheric Distillation Plant (U-11) prise three main diagram blocks.
Different to Atmospheric Distillation Plant
Based on general process assumptions, The RDB
(U-10), Vacuum Distillation have a light oil feed-
(Reliability Diagram Block) of the Atmospheric
ing most of time and that condition preserve
Distillation Plant comprise five blocks in series
equipment. In addition, right equipment project
which represent Feed, Desalter, Heating, Furnace,
specification and correct maintenance policy along
Atmospheric Distillation and LPG Treatment.
time permit system restore their life in each 5 years,
This means if one block fails down the whole Sys-
as doing so, System Availability most be high as it
tem will be unavailable.
will be discussed in item 4.
Each Subsystem represented in the RDB, com-
prises several equipments with each PDF (prob-
ability density function). The assumptions to 3.3 Thermal Cracking Plant (U-211)
perform RDB model are:
The main objective of Thermal Cracking is convert
• It’s not being regarded facilities availability and heavy feed from Atmospheric Distillation (U-11)
other supply influence in U-10 availability; into diesel product.
• Subsystem unavailability represents system fail
downtime;
• The average availability target is 97,0% over
3 years; 1- 2- 3- 4- 5-
• Total Production per day is 1.500 m3/day. Feed Desalter Heat Atmospheric LPG
Subsystem Subsystem Subsystem Distillation Treatment
In Figure 3 below does shown RDB which com- Subsystem Subsystem
prise five main diagram blocks.
Despite having a heavy oil feeding most of Figure 3. Atmospheric Distillation RDB.
time, right equipment project specification and Source: Author, 2010.
correct maintenance policy along time permit sys-
tem restore their life in each 5 years, as doing so,
System Availability must be high as it will be dis-
cussed in item 4. In fact, most of equipments are as 1- 2- 3- 4- 5-
Feed Desalter Heat Atmospheric Vaccum
good as new mainly the static ones. Most dynam- Subsystem Subsystem Subsystem Subsystem Destillation
ics equipments as pumps have redundancy and it
permit high performance even thought equipment Figure 4. Distillation RDB.
reliability is not so high. Source: Author, 2010.

10
 Based in general assumptions, The RDB • Total Production per day is 2500 m3/day.
(Reliability Diagram Block) of Thermal Cracking
Plant comprises five blocks in series which rep- In Figure 6 below does shown RDB which
resent Feed and Preheater, Thermal Cracking, comprise eight main diagram blocks.
Fractioning, compression and Stabilization. That
means if one of block fail down the whole RDB 3.5 Nafta Hydrodesulphurization Plant (U-12)
will be unavailable. Each sub system represented The main objective of Nafta Hydrodesulphuriza-
for RDB comprises several equipments with their tion is to separate sulphur component from nafta
PDFs based in failures modes. The main assump- feed from Atmospheric and Vacuum Distillation
tions to make up RDB are: (U-10), Atmospheric Distillation (U-11) and Ther-
• It’s not being regarded facilities availability and mal cracking Plant (U-211).
other supply influence in U-211 availability; Based in general assumptions, The RDB (Relia-
• The equipment failure modes are based in fail- bility Diagram Block) of Nafta Hydrodesulphuri-
ure historical data of similar Unit Plant from zation Plant comprise four blocks in series which
other refinery; represent Feed, Reaction, H2 make up, H2 recycle,
• Subsystem unavailability represents system fail Diesel Fractioning, Drying Subsystem and
down; Cleaning Water . That means if any one of block
• The availability target is 97,0% in 3 years; fail down the whole system is down. RDB (Reli-
• Total Production per day is 1.500 m3/day. ability Diagram Block) will be unavailable. Each
sub system represented in the RDB comprises eight
In Figure 5 below does shown RDB which equipments with their PDF (Probability Density
comprise five e main diagram blocks. Function) based in failures modes.
• It’s not being regarded facilities availability and
3.4 Diesel Hydrodesulphurization Plant (U-13) other supply influence in U-12 availability;
The main objective of the Diesel Hydrodesulphuri- • The equipment failure modes are based in fail-
zation is to separate out sulphur component from ure historical data of similar Plant from other
diesel which come from the Atmospheric and Vac- refinery;
uum Distillation (U-10), Atmospheric Distillation • Subsystem unavailability represents system fail
(U-11) and Thermal cracking Plant (U-211). Based down;
on the general assumptions, The RDB (Reliability • The average availability target is 98% in 3 years;
Diagram Block) of the Diesel Hydrodesulphuriza- • Total Production per day is 2500 m3/day.
tion Plant comprise eight blocks in series which In Figure 7 below does shown RDB which com-
represent Feed, Reaction, H2 make up, H2 recycle, prise eight main diagram blocks.
Diesel Fractioning, Drying Subsystem and Clean- One of the most of important process condition
ing Water. That means if one of block fails down is that H2 make up compressors (A/B) in Diesel
the whole RDB system will be unavailable. Each
sub system represented in RDB comprises eight
equipments with their PDF (Probability Density
Function) based on the historical failures modes. 1.0 - 3.0 - 4.0 - 5.0 -
The main assumptions for this System RDB are: Feed Reaction H2 H2
Subsystem Subsystem Make Recycle
• It’s not being regarded facilities availability and up
Subsystem
Subsystem

other supply influence in U-13 availability;

• The equipment failure modes are based in fail-
ure historical data of similar Plant from other 6.0 - 7.0 - 8.0 -
Fractionning Drying Cleanning
refinery; Subsystem Subsystem Water
• Subsystem unavailability represents system fail Subsystem

down;
• The average availability target is 98% in 3 years; Figure 6. Diesel Hydrodesulphurization RDB.

1- 2- 3- 4- 5- 1.0 - 2.0 - 3.0 - 5.0 -

Feed Thermal Fractioning Compression Stabilization
and Cracking Subsystem Subsystem Subsystem Feed Reaction Fractioning H2
Preheater Subsystem Subsystem Subsystem Subsystem Recycle
Figure 5. Thermal Cracking Plant RDB. Figure 7. Nafta Hydrodesulphurization RDB.

11
Hydrodesulphurization (U-13) supply H2 to heat exchangers and towers in series. That means
both plants U-12 and U-13. In doing so, in case if one of block fail down the whole RDB will be
of unavailability in H2 make up compressors both unavailable. In this case, like other subsystems and
Plants will be unavailability. systems the pumps are in parallels configuration.
It means that is necessary both pumps fail down to
shut down DEA Plant. The main assumptions to
3.6 Acid Gas Treatment (DEA—U-23)
make up RDB are:
The main objective of Acid Gas Treatment plant is
• It’s not being regarded facilities availability and
to separate sulphur component from gas produced
other supply influence in U-26 availability;
in Nafta and Diesel Hydrodesulphurization. Based
• The equipment failure modes are based in
in general assumptions, The RDB (Reliability Dia-
failure historical data of similar Plant from
gram Block) of DEA Plant comprises many types
other refinery;
of equipment like vase, pumps, heat exchangers
• Subsystem unavailability represents system fail
and towers in series. That means if one of equip-
down;
ment fail down the whole System will be unavaila-
• The availability target is at least 98% in 3 years;
ble. In this case, like other subsystems and systems
the pumps are in parallels configuration. It means In Figure 9, Acid water RDB subsystem is
that is necessary both pumps fail down to shut represented comprising vases, pumps and towers.
down DEA Plant. The main assumptions to make One of the most important assumptions in Acid
up RDB are: Water Plant is that in case of unavailability in such
Plants, others plants are unavailable as Vacuum
• It’s not being regarded facilities availability and
and Atmospheric Distillation Plant (U-10), Atmos-
other supply influence in U-23 availability;
pheric Distillation Plant (U-11), Thermal Cracking
• The equipment failure modes are based in fail-
Plant (U-211), Nafta and Diesel Desulphurization
ure historical data of similar Plant from other
(U-2312/U-2313) and Catalytic Cracking Plant.
refinery;
Despite high Acid Water availability, there’s no sig-
• Subsystem unavailability represents system fail
nificant impact in refinery regarding Acid Water
down;
Plants.
• The availability target is at least 98% in 3 years;
In Figure 8, DEA RDB subsystem is represented
3.8 Catalytic Cracking Plant (U-21)
comprising vases, pumps and towers.
The main objective of Catalytic Cracking Plant is
convert heavy feed from Atmospheric and vacuum
3.7 Acid Water (U-26)
Distillation (U-10) into light oil product.
The main objective of Acid water Treatment plant Based in general assumptions, The RDB (Reli-
is to separate out sulphur component from gas ability Diagram Block) of Catalytic Cracking
produced in Nafta and Diesel Hydrodesulphuri- Plant comprises five blocks in series which rep-
zation. Based in general assumptions, The RDB resent Preheating Feed, Conversion Subsystem,
(Reliability Diagram Block) of DEA Plant com-
prises many types of equipment like vase, pumps,

1.0 - 1.1 - 1.2 - 1.3 -

TQ-01 B-01-02 P-01 P-06
1.0 - 1.1 1.2 - 1.3 - AB
T-01 -V-03 P-01 A P-01 B

1.4 - 1.5 - 1.6 - 1.7 -

1.4 - 1.5 - 1.6 - 1.7 - P-04 T-01 V-01 P-05
P-05 P-03 T-02 V-04

1.8 - 1.9 - 1.10 -

1.8 - 1.9 - 1.10 -
B-02 B-03 P-02 P-03 B-03 V-02
AB AB AB

Figure 8. DEA Plant RDB. Figure 9. Acid Water Plant RDB.

12
Cold Area, DEA, Caustic Cleaning. That means 3.10 Fractioning Plant (U-20)
if one of block fail down the whole RDB will be
The main objective of Fractioning Plant is turn out
unavailable. Each subsystem represented for RDB
nafta from Nafta Hydrotreatment Plant (U-13)
comprises several equipments with their PDFs
into heavy and light nafta product.
based in failures modes. The main assumptions to
Based in general assumptions, The RDB (Reli-
make up RDB are:
ability Diagram Block) of Fractioning Plant
• It’s not being regarded facilities availability and comprises eight blocks in series which represent
other supply influence in U-21 availability; towers, pumps, vase and heat exchanger. That
• The equipment failure modes are based in fail- means if one of block fail down the whole RDB
ure historical data of own Unit Plant; will be unavailable. Each subsystem represented
• Subsystem unavailability represents system fail for RDB comprises several equipments with their
down; PDFs based in failures modes. The main assump-
• The availability target is 98,0% in 3 years; tions to make up RDB are:
• Total Production per day is 55 m3/day.
• It’s not being regarded facilities availability and
In Figure 10 below does shown RDB which other supply influence in U-20 availability;
comprise five e main diagram blocks. • The equipment failure modes are based in fail-
The most critical equipment in such plant is ures data from similar equipments of other
compressor in terms of number of increasing fail- refinery;
ures despite due to K/N (2/3) configuration, it not • Subsystem unavailability represents system fail
cause high unavailability impact in whole system. down;
• The availability target is 98,0% in 3 years;
• Total Production per day is 1500 m3/day.
3.9 Reforming Catalytic Cracking Plant (U-22)
In Figure 12 below does shown RDB which
The main objective of Reforming Catalytic Crack-
comprise five e main diagram blocks.
ing Plant is convert heavy nafta from fractioning
Plant (U-20) into reforming nafta product. 3.11 Logistic resources
Based in general assumptions, The RDB (Reli-
ability Diagram Block) of reforming Catalytic Logistic management is that part of supply chain
Cracking Plant comprises five blocks in series process that plans, implements and control the effi-
which represent Reaction, Recontact, Debutan- cient, effective flow and storage of goods, services
izer, Purification and Regeneration. That means and related information from the point of origin to
if one of block fail down the whole RDB will be the point of consumption in order to meet custom-
unavailable. Each subsystem represented for RDB er’s requirement (Ballou 2004). The logistic resource
comprises several equipments with their PDFs like tanks, pipelines and ships have main objective
based in failures modes. The main assumptions to to make products, equipment and raw material flow
make up RDB are: easier along process in order to maximize profits.
• It’s not being regarded facilities availability and
other supply influence in U-22 availability;
• The equipment failure modes are based in relia-
1- 2- 3- 4- 5-
bility requirement and failures data from similar Reaction Recontact Debutanizer Purification Regeneration
equipments; Subsystem Subsystem Subsystem Subsystem Subsystem
• Subsystem unavailability represents system fail
down; Figure 11. Reforming Catalytic Cracking Plant RDB.
• The availability target is 98,0% in 3 years;
• Total Production per day is 800 m3/day.
In Figure 11 below does shown RDB which
1.0 - 1.1 - 1.2 - 1.3 -
comprise five e main diagram blocks.
T-01 P-01 V-01 B-01
A/B

1.0 - 2.0 - 3.0 - 4.0 - 5.0 - 1.4 - 1.5 - 1.6 - 1.7 -

Preheating Conversion Cold DEA Caustic P-02 P-03 P-04 B-02
Feed Subsystem Area Cleaning A/B

Figure 10. Catalytic Cracking Plant RDB. Figure 12. Fractioning Plant RDB.

13
 The logistics recourses configuration mostly represent the three tanks G-401/402/405 one by
is applied to systems regarding its dependence one even thought there was not flow to feed U-11.
and related demands and supply of products. In In this example, there’s no high impact in final
general, in Logistic model assessments is not take result therefore such tank model in RDB meth-
into account the equipment reliability which would odology do not matter to final result but in Acid
take high influence in profits results. Even if that Water Subsystem, if logistic representation be car-
assumption is limited in such Logistic analysis, the ried out, only acid gas that feed such plants will
other way round happen too, in other words, logis- be into account and it not represents plants shut
tic assumption in many cases is not taking into down (U-10, U-11, U-12, U-13 and U-21) cause
account in RAM analysis. That is the main point in case of Acid Water (U-26) shutdown, so many
that will be discuss below in a case study that will plants shutdown to.
comprise Plants and logistic resources together Other good example is U-12, which furthermore
having a Complex Refinery System. than their equipment failures, Acid Water Plant
The main logistic resources in refinery plant case shutdown, H2 make up compressor of U-13 shut-
study are tanks which provide oil to distillation down and PSA (H2 purification ) of U-22 shut-
plants. Such tanks reduce system unavailability down make such plant unavailable.
whenever pumps or other equipment which supply If logistic methodology was carried out, probably
oil to tank shut down. In Figure 13, a good exam- that assumptions would not take into account, cause
ple of logistic mixed with RDB Methodology. logistic focus in product flow. As doing so, such
In first case, both Distillation plants are feed assumption must be represented in order to regards
for tanks. The U-10 is fed by G-01 and G-404. such impact. In Figure 14, such outside U-12 impacts
Both tanks are available and only one of them are represented in RDB condition block.
are enough to supply U-10, being G-404 an active Regarding such assumptions we get into
redundancy. Into such tanks there are equipments conclude that is not possible to model a complex
and their failures. The RDB model Tank failures system without consider logistic and reliability
(internal and external corrosion) in series with two issues. A refinery model example that is consid-
pumps parallel Block being one of them a passive ered complex system with ten plants and tanks
redundancy. will be carried out in order to show RAM + L
In second case, U-11 is fed from G-401/402/405 application.
or G-02 which supply U-11 and U-10 as an active
redundancy. The G-401/402/405 represents k/n
(1/3) configuration RDB that means at least one 4 SYSTEMS SIMULATION
of three must be available to not shut down U-11.
Into tanks are represented tank’s failures and The Simulation (Monte Carlo) has the main
pumps failures. Usually, logistic model probably objective to confirm the system availability

Figure 13. Tank feed distillation plants. Figure 14. Outside U-12 impacts.

14
results in order to determine critical equipment n
or logistics resources (Tanks) in terms of avail- ∑ pti
ability and utilization to give support to improve- EP (t ) = i =1
ment decisions. That RDB (Reliability Diagram n

Block) methodology makes up systems configu- ∑ Pti

i =1
ration regarding each equipment and failures
modes and taking into account failure mode
n
repair time.
For each system mentioned above is performed ∑ pri × ti
i =1
a simulation and after the whole system will be EP (t ) = n
assessed based in RDB (Reliability Diagram
Block) methodology.
∑ Pri × Ti
i =1
In order to run simulation some software like
MAROS (Maintainability, Availability, Reliability, pr1 × t1 + pr2 × t2 + … + prn × tn
and Operability Simulator-DNV) and BLOCK- EP (t ) =
SIM (Reliasoft) are performed and final results are Pr1 × T1 + Pr2 × T2 + … + Prn × Tn
compared among then in order to check result.
Nevertheless, when Systems characteristics are pr1 = pr2 = pr3 = … = prn
not representing completely, it’s possible to simu-
late the effect of equipment failures in System Pr1 = Pr2 = Pr3 = … = Prn
availability. Accord with simulation methodology
it’s possible to represent system life cycle along
time and take into account system down time. pr1 × (t1 + t2 + … + tn )
EP (t ) =
The systems simulations one by one were P r1 × (T1 + T2 + … + Tn )
performed showing the main result. The availabil-
ity and efficiency are approximately the same in
pri = Pri
case 1 and different in case 2. The cases are:
• The Case 1 regards that all equipments (in pr1 × (t1 + t2 + … + tn )
series) shutdowns cause 100% unavailable of EP (t ) =
one specific system capacity production;. pr1 × (T1 + T2 + … + Tn )
• In Case 2, partial of Plant Capacity production
is loss when equipments (in series) shutdowns. (t1 + t2 + … + tn )
EP (t ) =
(T1 + T2 + … + Tn )
The equation below shows case 1, that availability
and efficiency are the same along time. In this case
n
production is always in two condition along time,
0% when equipments shutdown or 100% when ∑ ti
i =1
system working property. D(t) is availability, EP(t) EP (t ) = n
= D(t )
is efficiency, t is time that system is working, T is
nominal time, p is real production and P is nominal
∑Ti
i =1
production.
The equation below shows case 1, in this case the The equation below shows case 2, and in this
system is either up or down or available or unavail- case production depends on loss that equipment
able 100% of total capacity. The case one represent cause in system range from zero to 100%. As the
most of equipments in refineries plants like towers, same, D(t) is availability, EP(t) is efficiency, t is
vases furnace and even pumps (active and passive). time that system is working, T is nominal time, p
Whenever such equipments shutdown, cause 100% is real production and P is nominal production.
of loss production in refineries plants. Such condition happen for instance when some
D(t) is availability, EP(t) is efficiency, t is time heat exchanger shutdown. In some cases is possible
that system is working, T is nominal time, p is real to produce but it’s necessary to reduce production
production and P is nominal production. while heat exchanger is being repaired.
n n n
∑ ti ∑ pri × ti ∑ p′ri × ti
i =1 i =1 i =1
D(t ) = n
EP (t ) = n
+ n
∑Ti ∑ Pri × ti ∑ P ′ri × ti
i =1 i =1 i =1

15
 pr1 × (t1 + … + tn −1 ) plants, in case of plant or tanks shut down, other
EP (t ) = products are produced even though such shutdown
Pr1 × (T1 + … + Tn −1 )
p′ r1 × (t ′1 + … + t ′ n ) occur. That is RAM + L approach, which consider
+ reliability and logistic to make up complex model
P ′ r1 × (T ′1 + … + T ′ n ) that is different from RDB (reliability Diagram
Model) approach which consider all plants in series
pri = Pri and tanks in parallels. In next analysis, improve-
ments actions will be used to compare RAM + L
p′ ri = P ′ ri results with RAM Methodology results.
pr1 × (t1 + … + tn −1 )
EP (t ) = 5 Critical analysis
pr1 × (T1 + … + Tn −1 )
and improvment actions
p′ r1 × (t ′1 + … + t ′ n )
+
p′ r1 × (T ′1 + … + T ′ n ) Regarding system results, the CCR, CTB, Nafta
HDT and Diesel HDT are the most critical plants.
(t1 + t2 + … + tn ) (t ′1 + t ′ 2 + … + t ′ n ) Therefore improvements are to be carried out on
EP (t ) = + systems to eliminate failures or reduce the conse-
(T1 + T2 + … + Tn ) (T ′1 + T ′ 2 + … + T ′ n )
quences therefore improving system efficiency and
n n consequently Complex System efficiency.
∑ ti ∑ t′i On CCR the most critical equipment are reac-
tors due to linkage failure modes therefore the sys-
i =1 i =1
EP (t ) = n
+ n
= D(t ) + D′(t ) tem improvement action is:
∑Ti ∑T ′i • To propose procedures when plant will buid up
i =1 i =1
to avoid linkage in such equipments;
Looking at Table 2 we conclude that the most On CTB plant the most critical equipment is the
critical systems are CTB, CCR, Nafta HDT and furnace due to coke formation therefore the system
Diesel HDT because lowest efficiency value than improvement action are:
defined target. • To reduce decoke time time spalling on line proce-
Regarding RDB methodology, refinery availabil- dure will be carried out in order to reduce time to
ity will be lower than the lowest system availability decoke furnace and reduce unavailability time;
because the systems are in series. That the same On Nafta and Diesel HDT the most critical
regarding efficiency. In fact, that a very conservative_ equipment are reactors due to linkage failure modes
assumption and it can be used to represent complex therefore the system improvement action is:
Systems which comprise all systems and it means
that in case of shutdown in any System the whole • To propose procedures when plant will build up
Complex System will shut down. In this case, refin- to avoid linkage in such equipments;
ery efficiency is lower than 95,77% over the 3 years These improvement actions will results in effi-
period. The results will be improved if improvement ciency improvements as shown in Table 3 below.
are implemented on each critical system. By the After all systems improvement is necessary to
other way round, regarding logistic resources and make up the macro system regarding all plant in
model Complex System which comprise tanks and series based in RDB (Reliability Diagram Block)
Methodology). As doing so, the macro system
Table 2. System efficiency. availability is 93,89% in 3 years and it configura-
System Efficiency target Efficiency result tion is shows in Figure 15 below.
That result represents that refinery will produce
UDA 98,0% 100% 93,89% of total production (3 years) in their higher
UDV 98,0% 100% capacity. In fact, such conservative approach
UFCC 98,0% 100% required a RAM + L Methodology configuration
AA 98,0% 100% that will be carried out on next item regarding
DEA 98,0% 100% logistic issues (tank) and reliability.
CTB 98,0% 95,74%
CCR 98,0% 97,44%
6 RAM + L SIMULATION
Fractioning 98,0% 99%
Nafta HDT 98,0% 95,77%
Diesel HDT 98,0% 97,64%
The RAM + L methodology consider logistic
resources in Complex System efficiency and make up

16
Table 3. System efficiency improvement.

System Efficiency target Efficiency result

[A] [A]
G-01 G-401/402/405
UDA 98,0% 100% U-10
U-11
UDV 98,0% 100%
UFCC 98,0% 100% [S:1] [S:1]
G-01
AA 98,0% 100% G-02

DEA 98,0% 100% Tanks Tanks

CTB 98,0% 98,53%
CCR 98,0% 98,26%
Fractioning 98,0% 99%
U-20 U-21 U-23 U-56 U-12 U-22 U-13 U-211
Nafta HDT 98,0% 99,05%
Diesel HDT 98,0% 98,56%
Figure 15. Macro system RDB configuration.

Figure 16. Actual refinery (RAM + L).

Figure 17. Actual refinery (RAM + L).

17
model which consider both logistic and equipment logistic issues partially solution for Complex System
failure. The final results will show the total efficiency will achieve high utilization in logistic resources
in all products regarding the relation between demand with low efficiency due lower availability plants.
and supply among equipments and systems. Some softwares like MAROS and BlockSIM
The whole system will be represented actual and softwares focus on reliability issues. By the other
future configuration as shows Figure 16 and 17 in side, there are others software focuses more in
page 14 below. The actual refinery configuration logistic like ARENA and TARO software focus on
comprises seven tanks and three plants (U-11, logistic issues. The best solution is develop some
U-10 and U-21). The future configuration con- software which comprise reliability and logistic
sider more seven Plants (U-56, U-23, U-12, U-13, issues like TARO and MAROS, improve logis-
U22, U-20, U-211). tics softwares like ARENA to consider reliability
In such configuration, the U-56 is in series with issues or improve reliability softwares like Block-
U-10, U-11, U-211 and U-21, it means that in case sim to consider logistic issues.
of unavailabity in such plant the other plants will In case study mentioned above the logistic issues
shut down. were simple to be represented but if it would have
The second important condition is that PSA in been considered ships and other logistic resources
U-22 supply H2 to U-12 and U13. It means that it would be harder to be modeling for such soft-
in case of PSA unavailability, U-12 and U-13 will ware. The most important aspect is consider logis-
shut down. tic and reliability issues when Complex System is
The third important condition is make up com- being assessed in order to have a more reliable opti-
pressor in U-13 supply H2 to reactor in this plant mizations and improvements.
(U-13) and to U-12 reactor. In case of compressor
unavailability both plants will shutdown, therefore
such compressor is in series with two plants (U-12
and U-13). References
The final Complex System efficiency is 100%
Ballou, R.H. 2004. Business logistic/supply chain
in 3 years for all products in actual configuration management. New jersey: Pearson Education Inc.
(Tanks, U-10, U-11 and U-21). Billinton, R. and Allan, R.N. “Reliability Evaluation of
The final Complex System efficiency in future Engineering Systems: Concepts and Techniques”, 1st
will vary from 99,14% to 99,86% of total produc- Edition, Plenum Press, New York, 1983.
tion in 3 years. The result is different of RDB Calixto, Eduardo; Schimitt, William. Cenpes II project
methodology that is not consider logistic resources RAM analysis. ESREL 2006, Estoril.
(tanks and pumps) neither all products. Calixto, Eduardo. “The enhancement availability
methodology: a refinery case study”, ESREL 2006,
Estoril.
Calixto, Eduardo. “Sensitivity analysis in critical equip-
7 CONCLUSIONS ments: The distillation plant study case in the Brazil-
ian oil and gas industry”. ESREL 2007, Stavanger.
The RAM analysis methodology comprises logistic Calixto, Eduardo. “Integrated preliminary hazard
issue into RAM analysis being more robust assess- analysis methodology regarding environment, safety
ment of Complex system like refineries. and social issues: The platform risk analysis study”.
In order to perform such analysis is required to ESREL 2007, Stavanger.
obtain so many information about equipment fail- Calixto, Eduardo. “The safety integrity level as hazop risk
ures and to define the logical dependency of sys- consistence. the Brazilian risk analysis case study”.
tems, equipments and logistic resources. ESREL 2007, Stavanger.
Calixto, Eduardo. “The non-linear optimization method-
Although is more realistic analysis commonly ology model: the refinery plant availability optimiza-
in Brazil is rare implemented such assessment due tion case study”. ESREL 2007, Stavanger.
lack of information or integrated system vision. Calixto, Eduardo. “Dynamic equipments life cycle
In general two groups work with different issues analysis”. 5° International Reliability Symposium SIC
as logistic and reliability trying to optimize Com- 2007, Brazil.
plex Systems and equipments by itself without an Calixto, Eduardo. “Using Network Methodology to
integrate perception. Define Emergency Response Team Location: The
In general, there are many software focus on Brazilian Refinery Case Study.” International Journal
system reliability or system logistic giving more of Emergency Management, V6. Inderscience Pub-
lisher, 2009.
importance in one or other aspect without take into Calixto, Eduardo. “Environmental Reliability as a
account both issues. In case of more focus on reli- Requirement for Defining Environmental Impact
ability issues, partially solution for Complex Sys- Limits in Critical Areas.” ESREL 2008, Valencia.
tem will achieve high availability plant but lower Glasstone, S., Laidler, K.J, and Eyring, H.E. The Theory
efficiency due logistic problems. In case of focus on of Rate Processes, McGraw Hill, NY, 1941.

18
Groebel, David, Mettas, Adamantios and Sun, Feng-Bin, ReliaSoft Corporation, Life Data Analysis Reference,
Determination and Interpretation of Activation ReliaSoft Publishing, Tucson, AZ, 2000. Also por-
Energy Using Accelerated Test Data, 47th Reliability tions are published on-line at www.Weibull.com
and Maintainability Symposium. Schmitt, W.F. “Confiabilidade de Sistemas de Dis-
Kececioglu, Dimitri, and Sun, Feng-Bin, Environmental tribuição: Metodologias Cronológica e Analítica”,
Stress Screening—Its Quantification, Optimization Dissertação de Mestrado, UNIFEI, Out./2002.
and Management, Prentice Hall PTR, New Jersey, Werbisnka. S. 2007. Interaction between logistic and
1995. operational system and availability model. Esrel
Kececioglu, Dimitri, and Sun, Feng-Bin, Burn-In 2007. Taylor & Francis Group, London, ISBN
Testing—Its Quantification and Optimization, Pren- 978-0-415-44786-7.
tice Hall PTR, New Jersey, 1997.
Mettas, Adamantios, Modeling & Analysis for Multiple
Stress-Type Accelerated Life Data, 46th Reliability
and Maintainability Symposium.

19