Introduction to RAM

What is RAM?
RAM refers to Reliability, Availability and Maintainability. Reliability is the
probability of survival after the unit/system operates for a certain period of
time (e.g. a unit has a 95% probability of survival after 8000 hours). Reliability
defines the failure frequency and determines the uptime patterns.
Maintainability describes how soon the unit/system can be repaired, which
determines the downtime patterns. Availability is the percentage of uptime
over the time horizon, and is determined by reliability and maintainability.

Why choose RAM Analysis?

RAM has a direct impact on profit through lost production and maintenance
costs. The main objectives of RAM are to increase system productivity, increase
the overall profit as well as reduce the total life cycle cost — which includes lost
production cost, maintenance cost, operating cost, etc.

Some significant figures are:

As much as $100,000 per hour typical production losses can be sustained
in base chemical plants due to non-availability
In oil refineries, production losses are millions of dollars per year for every
1% of non-availability
In oil refineries, the maintenance staff are up to 30% of total manpower
Maintenance spending is one of the largest parts of the operational budget
after energy and raw material costs
Each year over $300 billion is spent on plant maintenance and operations
by US industry, and it is estimated that approximately 80% of this is spent
on systems and people to correct the unplanned failure of machines
(Engineering Maintenance, 2002)
The operation and maintenance budget request of the US Department of
Defence for the fiscal year 1997 was on the order of $79 billion

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
RAM analysis is essential to the system profitability. Even small improvements
in the effectiveness of RAM schemes make large additions to the bottom line.

Role of RAM Analysis

Fig.1 illustrates the interactions and applications of RAM analysis. For an
existing process, maintenance data are usually recorded in the CMMS
(Computerised Maintenance Management System). These data can be analysed
through qualitative and quantitative approaches, as shown in Fig.2.

Failure Analysis Improve

Basic Analysis – Equipment (Qualitative & Equipment
Quantitative) Design

RBD or Fault
Basic Analysis – System Tree

Maintenance Stocking Policy Process

Applications Strategy Optimisation Synthesis or
Optimisation Retrofit

Operation Stage Design Stage

Fig.1 RAM Analysis Cycle

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
 Failure Analysis

Qualitative Analysis Quantitative Analysis

Failure Mode and Distribution

Effect Analysis Parameters
(FMEA) Regression

Fig. 2 Failure Analysis

Failure mode and distribution parameters can be obtained for each unit in the
system. Reliability Block Diagrams (RBD) or Fault Trees (FT) can be used to
represent the logic relationships between component failures and system
failures, and provide the basis for a RAM study. With the failure distribution
data input into an RBD/FT, engineers will be able to understand the RAM
performances of the current system and carry on further developments and
optimisations. In fact, there is a direct relationship between RBD and FT, but
most engineers find RBD easier to use, as it can be more easily related to a
process flowsheet. The approach to RAM analysis adopted by Process
Integration Limited exploits RBDs.

In the operation stage, RAM analysis can be implemented to optimise

maintenance strategy as well as stocking policy for spare parts. The
maintenance strategy will then affect the failure records in the CMMS, which in
turn help engineers revise the maintenance strategy through time. New failure
records will be saved in the CMMS during the operating stage.

In the design stage, RAM analysis can be integrated into the design of the
system configuration, which will ensure the optimum design with balanced
RAM performance and total investment. Moreover, the qualitative and
quantitative analysis of process unit failures can help designers to modify the
structure of a specific process unit to improve the process design.

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
RAM analysis run throughout both operating and design phases to enable the
process to achieve high profitability.

Benefits of RAM Analysis

The following benefits can be obtained from RAM analysis.

Decision making
What maintenance policy should be applied
Investment decisions on maintenance
Resource utilisation
Inspection intervals
Optimum spare part purchasing
Appropriate maintenance scheduling
Understanding the financial implications of maintenance
Decision making based on modelling
Cost management
Managing the cost related to unavailability
Cost of maintenance
Etc.
Integration with other business activities
All projects on a site have an effect on process RAM
RAM needs to involve the whole organization
Meeting the business demand
Reduce outages caused by breakdowns
Reduce the loss of revenues caused by unavailability
Etc.

RAM Analysis Tools

There are several tools to conduct RAM analysis in different stages and for
different purposes, as shown in Figure 3.

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
FMEA (Failure Mode and Effects Analysis) or FMECA (Failure Mode, Effects and
Criticality Analysis) can be used to identify equipment failure modes and their
system effects. FMEA/FMECA provides the basis to system failure analysis. On
system level, Event Tree (ET), Fault Tree (FT) and Reliability Block Diagram
(RBD) can be used to represent the logic relationships between individual
equipment failure modes and system failure modes.

An Event Tree uses binary branching to identify the events leading to system
failures or successes. In a Fault Tree, system failures can be expressed in terms
of combinations of component failure modes, as illustrated in Figure 4a. A Fault
Tree is part of Event Tree and can be transformed from an Event Tree. An RBD,
as illustrated in Figure 4b, is a logical inverse of a Fault Tree. As shown in Figure
4b, blocks are used to represent component failures or failure modes for a
certain component. The approach to RAM analysis adopted by Process
Integration Limited exploits RBDs because most engineers find the RBD easier
to use, as it can be more easily related to a process flowsheet.

With distribution parameters input into an RBD (or ET or FTA), system RAM
performance will be obtained by simulations.

Understand System Failures Hierarchy

Event Tree  Using binary

branching
 Supplementary to FTA

Failure Mode, Effect and Fault Tree  Events leading to

Criticality Analysis Analysis (FTA) system failures
(FMECA or FMEA)  Part of Event Tree

 Provide basis to Event

Tree, FTA, RBD, etc
Reliability Block  Related to Success
Diagram (RBD) Tree
 Logic inverse of FTA

Fig.3 Different RAM Analysis Tools

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
 a) Fault Tree b) Reliability Block Diagram
Fig.4 Fault Tree and Reliability Block Diagram Examples

Deterministic or Probabilistic Modelling

A RAM study is conducted through a probabilistic approach, which is
fundamentally different from a deterministic approach. The difference is
illustrated by an example. As shown in Table.1, 41 failures are recorded in 20
years. Table.2 lists other necessary data.

Table.1 Failure Records

Failure Record
No Time No Time No Time
1 02/05/1990 15 03/04/1996 29 28/01/2002
2 29/05/1990 16 20/10/1996 30 20/09/2002
3 07/07/1990 17 19/09/1997 31 23/05/2003
4 19/11/1990 18 02/09/1998 32 31/01/2004
5 07/04/1991 19 07/02/1999 33 19/10/2004
6 04/09/1991 20 22/07/1999 34 08/07/2005
7 04/02/1992 21 11/01/2000 35 08/04/2006
8 11/07/1992 22 11/04/2000 36 09/01/2007
9 28/01/1993 23 14/07/2000 37 04/11/2007
10 03/09/1993 24 22/10/2000 38 15/09/2008
11 13/04/1994 25 11/05/2001 39 07/01/2009
12 30/11/1994 26 04/07/2001 40 10/05/2009
13 03/04/1995 27 30/08/2001 41 21/05/2010
14 26/09/1995 28 12/11/2001

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
 Table.2 Maintenance and Cost Data
Mean Time To Repair (days) 5
Mean Time of PM (days) 1
CM Cost (K$ per action) 10
PM Cost (K$ per action) 2
Lost Production Cost
240
(K$/day)

Maintenance costs in Table.2 are associated with the penalties of consequent

damages to other units.

a) Deterministic Approach
According to the data in Table.1, Mean Time Between Failure (MTBF) = 7142
day/ 40 = 178 day. Using a deterministic approach, it is assumed the component
will fail every 178 days, as shown in Fig.5.

Up

Time
178 day 178 day 178 day

Fig.5 Failure Analysis using Deterministic Approach

To reduce unexpected failures, two preventive overhaul schemes are

considered:
1. Preventive overhaul at MTBF – 178 days
2. Using rule of thumb: preventive overhaul at 80% of MTBF – 145 days

Based on the economic data, total annual costs with different maintenance
policies are listed in Table.3. The cost comprises corrective maintenance cost,
preventive maintenance cost and lost production cost. As a result, preventive

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
overhaul at MTBF reduces the cost by 19%. If rule of thumb is applied and
overhaul at 80% of MTBF, the total cost is reduced by 23%.

But it is dangerous to use the rule of thumb for every unit, since each unit has
different failure distribution patterns as well as different maintenance cost,
associated penalty cost, etc. It is not possible to find a “golden rule” which can
be applied for all of the process units. Costs could even be increased with a
wrong rule.

Table.3 Total Cost of Different Policy

Maintenance Policy Annual Cost (K$) Cost Reduced (%) Availability
No Preventive Overhaul 2421 Base 0.9726
Preventive Overhaul at MTBF 1960 19.04 0.9778
Overhaul at 80% MTBF 1848 23.67 0.9790

b) Probabilistic Approach
Using a probabilistic approach, the failures are represented by a probability
distribution. The Reliability curve and Probability Density Function curve are
shown in Fig.6. From this:
The probability that the component will fail in 60 days is 10%.
The probability that the component will fail in 120 days is 32%.
The probability that the component will fail in 180 days is 57%.
The probability that the component will fail in 300 days is 90%.

The component will fail most likely around 140 days, although MTBF is 178 days.
This indicates that the component is more likely to fail after 140 days.

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
 Reliability Probability Density Function

1 0.005

0.9 0.0045

0.8 0.004

0.7 0.0035

0.6 0.003

Probability
Reliability

0.5 0.0025

0.4 0.002

0.3 0.0015

0.2 0.001

0.1 0.0005

0 0
0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800
Tim e(day) Tim e (days)

a) Reliability vs Time b) Probability Density Function

Fig.6 Reliability and Probability Density Function Curve

The frequency of preventive maintenance can be optimised based on economic

data. The results of such an optimization are listed in Table.4. Optimum
preventive interval is 105 days.

Table.4 Total Cost of Different Policy

Maintenance Policy Annual Cost (K$) Cost Reduced (%) Availability
No Preventive Overhaul 2421 Base 0.9726
Preventive Overhaul at
1960 19.04 0.9778
MTBF
Overhaul at 80% MTBF 1848 23.67 0.9790
Overhaul at Optimal
1767 27.01 0.9800
Interval

Cost optimisation using a probabilistic approach saves more money compared

with the deterministic approach. Inappropriate preventive overhaul scheme can
be avoided.

Redundancy Optimisation
Using standby components (sometimes referred to as spare or redundant
components) is a common way to increase system availability and profit in the
design stage, or even in retrofit. In addition to the inclusion of standby
equipment, the standby equipment can be exploited in different ways. Instead

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
of having one item of equipment on line and vulnerable to breakdown, there
may be two, with one on-line and one off-line. But these two items of
equipment can be sized and operated in many ways:
2 x 100% one on-line, one off-line switched off
2 x 100% one on-line, one off-line idling
2 x 50% both on-line, with system capacity reduced to 50% if one fails
2 x 75% both on-line operating at 2/3 capacity when both operating, but
with system capacity 75% if one fails
Etc., etc.
Since different units have different availability features and capital costs,
engineers usually face difficulties to identify the optimum process structure –
which unit should have redundancy, how many standby items are required,
what is their capacity and what is their operating policy? Many factors should
be considered when making decisions, e.g. availability target, investment
budget, etc. Engineers need a systematic way to make decisions. Combining
reliability analysis with optimization allows redundancy to be optimised with
different objectives and constraints.

Fig.7 Process of Acetic Acid Production

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
Using Fig.7 as an example, standby pumps are required for the acetic acid
production process. Pump data are listed in Table 5. Fig.8 and Fig.9 show two
different designs. Fig.8 is to minimise the total capital cost with the availability
target as 0.97. Pump D and Pump J are 2x50%, Pump E and G do not have
standby, all the others are 2x100%. Fig.9 maximises system usage within a
capital budget of 3.6x104 K$. Pump C does not have standby; all other pumps
are 2x100%.

A mathematical tool can help engineers make the best decisions on a sound
basis.
Table.5 Basic Information of Pumps
MTBF MTBF
Capacity MTTR Capital Cost
Pump ID (operation) (standby) Availability
(%) (days) (K$)
(days) (days)
Pump A 100% 300 4000 4 0.9868 320
Pump B 100% 300 4000 5 0.9836 628
Pump C 100% 200 2000 10 0.9524 8000
Pump D1 100% 220 4000 10 0.9565 4660
Pump D2 75% 220 4000 8 0.9649 4200
Pump D3 50% 250 4000 8 0.9690 3600
Pump E 100% 330 4000 4 0.9880 1116
Pump F 100% 280 4000 5 0.9825 176
Pump G 100% 300 4000 4 0.9868 1316
Pump H 100% 290 4000 4 0.9864 676
Pump I 100% 300 4000 4 0.9868 820
Pump J1 100% 310 4000 4 0.9873 504
Pump J2 75% 330 4000 4 0.9880 450
Pump J3 50% 350 4000 4 0.9887 390

Pump A Pump B Pump C Pump D3 Pump E Pump F

Pump A Pump B Pump C Pump D3 Pump F

Pump H Pump I Pump J3 Operating I

Pump G

Pump H Pump I Pump J3 Spare

Fig.8 Minimise Capital Cost with Availability Target

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
 Pump A Pump B Pump C Pump D1 Pump E Pump F

Pump A Pump B Pump D1 Pump E Pump F

Pump H Pump I Pump J1 Operating I

Pump G

Pump G Pump H Pump I Pump J1 Spare

Fig.9 Maximise Usage within Capital Cost Budget

Preventive Maintenance Interval and Inspection Interval Optimisation

As shown in Fig.10, the Preventive Maintenance (PM) interval or Inspection
interval need to be optimised to balance the maintenance cost and downtime
penalty cost. Too frequent PM will interrupt the process and increase the cost
of lost production. Too few PM will lead to unnecessary equipment failure and
excessive maintenance cost, in addition to the cost of lost production. The
interval of PM needs to be optimized to give the lowest total cost.

Another approach is to monitor the condition of equipment in order to

determine the maintenance frequency. Many techniques are available, e.g.
vibration monitoring. Such monitoring can be carried out continuously for
critical equipment through on-line instrumentation, or off-line through
inspection. If off-line condition monitoring is used, the frequency of inspection
needs to be determined. Too frequent inspection brings unnecessary costs. Too
infrequent inspection leads to unnecessary breakdowns. If components are
maintained according to their condition, extra cost needs to be invested for the
monitoring instruments. A systematic way is required to evaluate whether the
investment is worthwhile or not, and help engineers to decide what policy
should be used, real-time monitoring or inspection monitoring. The best
inspection interval can be found by mathematical calculations with balanced
maintenance cost, inspection cost, etc.

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
 100

Cost vs PM interval Monitoring Cost (K$/yr)

2200 Penalty Cost (K$/yr)
2100 90 Maintenance Cost (K$/yr)
2000
1900
Cost (K$)

1800 80
1700
1600
1500 70
1400
1300
1200 60
50 100 150 200 No CBM Inspection/6 Inspection/4 Inspection/2 Inspection/1 Real-time
w eeks w eeks w eeks w eek
PM interval (Days)

Fig.10 Cost vs PM Interval/ Inspection Interval

Spare Parts Optimisation

Spare parts need to be stocked in the warehouse in order that units can be
repaired in time if they fail unexpectedly, or for planned maintenance. The
holding cost and depreciation cost can be very high if too many parts are
stocked. But if not enough parts are stocked, delivery delay time could increase
system downtime and lead to extra penalty cost. Stocking policy should be
determined according to the failure patterns of the parts. Mathematic analysis
is necessary to find the best stocking policy and balance the stocking cost
(including depreciation cost) and downtime penalty cost, as shown in Fig.11.

4.5

3.5
Cost (K$/year)

Downtime Penalty Cost

3 (K$/year)
Holding Cost (K$/year)
2.5

2 Purchase Cost (K$/year)

1.5

0.5

0
5/3 4/2 3/1 2/0
Max/Min Stocking No

Fig.11 Total Cost vs. Different Stocking Policies

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com
Methodology
Functions Methodology
Data Regression Rank Regression, Maximum Likelihood
Estimation
RAM Simulation Monte Carlo Simulation
Redundancy Optimisation Markov Analysis with optimization
PM Optimisation Monte Carlo Simulation with optimisation
CBM Optimisation Monte Carlo Simulation with optimisation
Spare Parts Optimisation Monte Carlo Simulation with optimisation

Process Asset Integration & Management (Trading as ProAIM) Ltd

One Central Park, Northampton Road, Manchester, M40 5BP, UK
T. +44 (0)161 918 6790
E. [email protected] www.proaimltd.com