Reliability Analysis of Electronic Systems using

Markov Models

István Matijevics
Polytechnical Engineering College, Subotica, Serbia and Montenegro,
[email protected]

Zoltán Jeges
Polytechnical Engineering College, Subotica, Serbia and Montenegro,
[email protected]

Abstract: Programmable Electronic Systems are tools for safety protection applications in
industrial processes. These electronic solutions have special circuits and architectures.
Markov Models can expressively represent the operation of a programmable electronic
system as various system components fail and/or are repaired. This paper describes one
method and shows examples of the reliability analysis of control system. In model are
multiple failure rates as a function of failure state, common cause failures, on-line
diagnostic capability of a programmable electronic system, multiple failure modes, and
different repair rates as a function of failure state.

Keywords: Reliability analysis; Programmable electronic controls; Markov models

1 Introduction [2]
In process industry nowadays there are a great number of PES (Programmable
Electronic Systems) system applications. These systems are very important for the
management of risk. These systems consist of sensors, computers
(microcontrollers) and actuators. The unwanted failure events damage the
environment and cause loss of production and investments in equipment.
New international standards (IEC61508 [3] and ISA-S84.01 [4,5]) are required
especially for high safety applications and quantification of the achieved safety.
The following main objectives are necessary in the teaching on reliability in PES
(Programmable Electronic Systems) [1]:
• Reliability specifications-oriented design,
• Re-design after analyzing field data,
• Reliability analysis of an existing design,
• Failure analysis of components, circuits or systems,
• Maintainability analysis of an existing design and
Understand and apply reliability standard.

2 Most Used Analysis Techniques [2]

Some reliability analysis techniques are graphically represented in Fig. 1. These
techniques are grouped into:
• quantitative – the interval between the resulting numbers and the ratio of the
resulting numbers has a meaning - and
• qualitative – the resulting numbers are only used for distinction or rank
ordering.

Figure 1
Most used analysis techniques

• Analysis by experts: based on previous experience in similar applications.

• FMEA (failure mode and effect analysis and derivatives): bottom-up analysis
of a system, by examining all component failures and determining the effects
of these failures on the entire system.
• Parts count analysis or component count analysis: is an analysis technique to
calculate the failure rate of a system when the failure rates of its components
are known.
• RBD (reliability block diagrams): a model of the behavior of a system by
showing graphically the condition for a successful operation.
• Hybrid techniques: combinations of reliability block diagrams and Markov
analysis results for redundant configurations.
• FTA (fault tree analysis): top-down method, how basic events may lead to a
certain top-event.
• Markov analysis: the safety of a system is analyzed by representing the
system by beans of the different states and transitions between these states.

3 Programmable Electronic Systems Architectures

(PES) [5]
Traditional automatic protection systems used in industrial processes mechanical
relays. A PES offers advantages for these safety protection applications including
fast response times, digital communications capability and extensive on-line
diagnostics to detect electronic component failures.
The on-line self-diagnostic capability of the system is a critical variable. Good
diagnostics improve both safety and availability. Two types of diagnostics are
used in a PES:
• reference diagdostics and
• comparison diagnostics.

3.1 Failure Modes and Effects Analysis

An FMEA (Failure Mode Effect Analysis) is a bottom up technique that used
qualitatively, quantitatively or as a combination of both and is very effective in
identifying critical component failures in a PES. An FMEDA (Failure Mode
Effect and Diagnostic Analysis) is an FMEA variation. It combines standard
FMEA techniques with extensions to identify online diagnostic techniques. It is a
technique recommended to generate failure rates for each important category:
• safe detected,
• safe undetected,
• dangerous detected and
• dangerous undetected
in the safety models.
Fig. 2 shows an input circuit from PES, Fig. 3 shows FMEDA done on the input
circuit.

Figure 2
PES input circuits

FMEDA Failures/billion Safe dange

hours rous
compo Mode Effect criticality FIT Safe Dang. Det.
nent 0 0
R1 short Loose filter Safe 0.13 0.125 0 0 0.50 0
open Logic 0 Safe 0.50 0.50 0 1 0 0
C1 short Logic 0 Safe 2 2 0 0 0 0
open Loose filter Safe 0.50 0.50 0 0 0 0
R2 short Overvoltage Dang. 0.13 0 0.13 0 0.50 0
open Logic 0 Safe 0.50 0.50 0 1 0 0
R3 short Logic 0 Safe 0.13 0.125 0 0 0 0
open overvoltage Dang. 0.50 0 0.50 0 0 0
D1 short Logic 0 Safe 2 2 0 0 0 0
open Blow out Dang.
circuit 5 0 5 0 0 0
D2 short Logic 1 Dang. 2 0 2 0 0 0
open Blow out Dang.
circuit 5 0 5 0 0 0
OC Led No light Safe
dim 28 28 0 0 0 0
Tran. Logic 1 Dange.
short 19 0 19 0 0 0
Tran. Logic 0 Safe
open 5 5 0 0 0 0
R4 short Logic 0 Safe 0.13 0.125 0 0 0 0
 FMEDA Failures/billion Safe dange
hours rous
open Logic 1 Dang. 0.50 0 0.50 0 1 0
38.88 0.025
71 32 7
Total Safe Dang.
Failure rates

Figure 3
FMEDA for PES input circuit

4 Markov Models for Reliability Analysis of PES [3]

Markov model (failure state diagram) is good tool in reliability analysis of PES,
because the method is flexible and gives a realistic model. The method can include
the following:
• common cause failures,
• multiple failures,
• different repair times and
• variable failure rates.
Markov model is a state diagram model with circles and arrows. The circles
represent the component states (working or failed), the arrows stand for the
direction of transitions between the states (failure or repair), so the arrows are
directed arcs. The failure or repair rates are presented by the arrows with numeric
values. A simple Markov model (one repairable component) is presented on Fig.
4.

Figure 4
Markov model of repairable component
The component is in state 1, if it is successful, or in state 2, if it failed. The model
can move from state 1 to state 2 at a rate of λ12 (the failure rate), or from state 2
to state 1 at µ 21 (the repair rate).

5 Common Cause Failures [4]

Common cause failures are simultaneous outages of many components, caused by
a single traumatic event (Fig. 5).

Figure 5
State-time diagram including common cause failures

The stochastic model for common cause failures will be derived from the state
space of two stochastically independent components (Fig. 4). λ1 and λ2 are the
outage rates of components 1 and 2, while µ1 and µ2 denote their repair rates.
λ and µ are generally known as transition rates.
 Figure 6
State-space of a system with two stochastically independent components

In the state space in Fig. 7, containing the possibility of the occurrence of common
mode failures of two components, there is a direct transition from state 1 to 4,
determined by the common cause outage rate. This rate is determined by the mean
time T ( Bc ) between two successive common cause outages:
1
λc = (1)
T ( Bc )
 Figure 7
State-space of the system with two components including common cause outages

The rateλc (which will be further assumed as being equal for both components) is
dependent on the system, in contrast to the component-specific rates λ1 and λ2 , so
it is system-specific.
According to Fig. 7, to determine the transition rates from state 1 to state 2 and 3,
respectively, the outage rates λ1 and λ2 must be reduced by λc . The reason for
this is that the outage rates λ1 and λ2 represent all the outages of the separately
studied components. However, inside the system, some of them are single outages,
while the rest are common cause outages. So, the sum of the transition rates for
transitions starting from state 1 is equal to λ1 + λ2 , just as in case of independent
outages.

6 Limited Repair Capacities

The number of repair teams is not unlimited. Only one repair team will be
considered here, therefore in double outages the repair of the component which
failed second must be delayed until the completion of repair of the component
which failed first (Fig. 8).
 Figure 8
State-time diagram including repair postponability

The possibility for repair postponability results in an additional outage state A *

in the state pace of the component in Fig. 9. Starting from the operating state B,
according to whether another component has failed first or not, there are two
possible transitions, to states A and A * respectively. Figure 10 shows the state
space of a system consisting of two components when employing only one repair
team.

Figure 9
State-space of a single component including repair postponability
 Figure 10
State-space for a system with only one repair team

7 PES Example with 4 Logic Parts [3]

There is a PES example in [3] in control electronics. The architecture of the
system is given in Fig. 11, the Markov model for the system in Fig. 12 and Fig.
13.

Figure 11
Two channel logic architectures
There are detected and undetected failures:

To properly account for common cause failures, each failure rate should be
partitioned into normal and common cause. This result in eight failure rates for
each physical set of channels in PES:

Figure 12
Markov model of PES (calculate-calculate mode)
 Figure 13
Markov model of PES (calculate-verify mode)

References
[1] Jorge Marcos, Luis Molinelli, Santiago Fernandez-Gomez, “Software-
Aided Reliability Education”, ASEE/IEEE Frontiers in Education
Conference, TIC-18, October 10-13, 2001 Reno
[2] J. L. Roovroye, E. G. van den Bliek, “Comparing safety analysis
techniques”, Reliability Engineering and System Safety”, 75 (2002) 289-
294
[3] Julia V. Bukowski, Wiliam M. Goble, “Using Markov models for safety
analysis of programmable electronic systems”, Elsevier, Isa Transactions
34, 1995 pp. 193-198
[4] István Matijevics, Lajos Józsa, “An Expert-system-assisted Reliability
Analysis of Electric Power Networks, Engng Applic. Artif. Intell. Vol. 8,
No. 4, pp. 449-460, 1995
[5] William M. Goble, Julia V. Bukowski, A. C. Brombacher, “How
diagnostic coverage improves safety in programmable electronic systems”,
Elsevier, Isa Transactions 36, pp. 345-350, 1998