A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

CHAPTER 4

RELIABILITY OF PROCESS MODELS

1. INTRODUCTION
2. THE REDUCTION APPROACH
3. THE MINIMUM PATH AND MINIMUM CUT SET METHODS
3.1. The Path Tree Algorithm
3.1.1. Film Deposition Case Study
3.2 . Evaluation of Minimum Cut Sets
3.3 . The Cut Set - Activity Incidence Matrix
3.4. The Path Set - Activity Incidence Matrix
4. SUMMARY
REFERENCES
QUESTIONS
PROBLEMS

1. INTRODUCTION
Reliability analysis of an IDEF3 model identifies critical activities in the model, improves
its performance, and decreases operating cost of the process. Such an IDEF3 model can be
complex as the model may include a large number of activities. This complexity warrants
applications of the reliability evaluation techniques.
In the recent years, attempts have been made to apply the IDEF methodology for
analysis of system reliability and project risk assessment. Ang and Gray (1993) examined the
adequacy of IDEF methodology and suggested a number of modifications and enhancements in
order to improve IDEF descriptive power for project risk assessment. Kusiak and Larson
(1994) integrated techniques for analysis of system reliability with an IDEF model. Kusiak and
Zakarian (1996a) developed a fault tree based methodology for reliability and risk assessment
of parent activities in IDEF3 models. This chapter is based on Kusiak and Zakarian (1996) and
it extends the system reliability evaluation techniques, such as, the reduction method, and
minimum path and cut sets methods to reliability evaluation of IDEF3 models.
The relationship between activities in IDEF3 is modeled with three types of links:
precedence, object flow, and relational. The precedence and object flow links expresses the
simple temporal precedence between activities. The relational links highlight the existence of a
relationship between activities. The logic of branching within a process is modeled using an
AND (&), OR (O), and exclusive OR (X) junction boxes. To represent the reduction approach
for reliability evaluation a serial and parallel system modeled with IDEF3 graphical syntax is
discussed next. Assume the states of activities are statistically independent.
1) Parallel system modeled with an AND logical junction
The activities following an AND junction are performed in parallel. Therefore, the
system fails if any of the parallel activities fails (see Figure 1).

1
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

R1
1

& R2 &

Figure 1. Parallel system with an AND logical junction

2) Serial system modeled with precedence, object flow, or relational links

For the activities arranged in a series, the system fails if any of the serial activities fails
(see Figure 2(a), 2(b), and 2(c)).

(a) (b) (c)

R1 R2 R1 R2 R1 R2
1 2 1 2 1 2

Figure 2. Serial system: (a) precedence link, (b) object flow link,
and (c) relational link

The system reliability of parallel activities following an AND junction, and the system
reliability of serial activities connected by precedence, object flow, or relational links is
determined from (1)
n
Rs = ∏ R j (1)
j=1
where: Rs is the reliability of the system and Rj is the reliability of activity j.
The system reliability for each of the four cases in Figure 2 and 3 is
Rs = R1 R2
3) Parallel system modeled with an exclusive OR logical junction
In an exclusive OR junction box only one of the several parallel activities is carried out
(see Figure 3). Each activity has a certain probability of occurrence.

R1
1

X R2 X

Figure 3. Parallel system an exclusive OR logical connector

The system reliability of parallel activities in Figure 3 is determined from (2)

n
R s = ∑ P jR j (2)
j=1
where: Rs is the system reliability, Pj is an occurrence probability of activity j,

2
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

n
and ∑ P j = 1 . For the system in Figure 4, Rs = P1R1 + P2R2, and P1 + P2 = 1.
j=1
3) Parallel system modeled with an OR junction box
In an OR junction box the activities are arranged in parallel and system functions if k out
of n parallel activities in the system function. The value of k depends from the underlying
process. To illustrate the reduction approach and minimum path and cut sets methods, assume
that for an OR junction, the system functions if 1 out n parallel activities functions (see Figure 4).
R1
1

O R2 O

Figure 4. Parallel system modeled with an OR junction box

The system reliability for the parallel activities in Figure 4 is determined from (3)

n 
R s = P ∑ x j ≥ k (3)
 j=1 
1 if activity j functions
where, xj = 
0 if activity j fails
For the process in Figure 4 the reliability is calculate as follows:
Rs = P(x1 = 1, x2 = 1) + P(x1 = 1, x2 = 0) + P(x1 = 0, x2 = 1) = R1 R2 + R1(1 - R2) + R2(1 -
R1) = R1 + R2 - R1 R2
where: P(xj = 1) = Rj is the probability that activity j functions, and P(xj = 0) = (1 - Rj) is the
probability that activity j fails.
In the next section, three different methods for reliability evaluation are discussed:
• Reduction method
• Minimum path set method
• Minimum cut set method

2. THE REDUCTION APPROACH

The basic idea behind the reduction approach applied to an IDEF3 model is to reduce
its size by combining appropriate parallel AND, OR, and exclusive OR branches, and the
activities connected by precedence, object flow, or relational links until a single equivalent
activity is obtained. This equivalent activity represents the reliability of the original IDEF3 model.
To illustrate the reduction approach, consider the IDEF3 model of the packaging process
presented in Figure 5.

3
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

R5 = 0.95
5 Activity Number and Name:
R3 = 0.97
1 - Define system requirements
3 O R6 = 0.97 O 2 - Develop weight, size, and shape
6 specifications
R1 = 0.99 R2 = 0.98 R9 = 0.98 3 - Perform detailed design of subsystem A
1 2 & R7 = 0.95 & 9 4 - Perform detailed design of subsystem B
7 5 - Design packaging module 1A
R4 = 0.99 6 - Design packaging module 2A
4 X R8 = 0.96 X 7 - Design packaging module 1B
8 - Design packaging module 2B
8 9 - Prototype build

Figure 5. Packaging design process represented with an IDEF3 model

In the first reduction, activities 5 and 6 are combined in a single activity 56. Since
activities 5 and 6 immediately follow an OR junction box, the reliability of the equivalent single
activity 56 is determined as R56 = R5 + R6 - R5R6 = 0.9985
In the second reduction, one can combine activities 7 and 8 in an equivalent activity 78.
The reliability of the equivalent activity 78 is R78 = P7R7 + P8R8 = 0.957
where: P7 = 0.3 and P8 = 0.7 is the occurrence probability for activity 7 and 8, respectively,
and P7 + P8 = 1. Following the above logic, the reliability of the IDEF3 model in Figure 5 is Rs
= R1R2R3R4R56R78R9 = 0.8725.

3. THE MINIMUM PATH AND MINIMUM CUT SET METHODS

The reduction approach presented in the previous section can be used if the number of
activities in an IDEF3 model is relatively small. When a process model is large, more efficient
methods based on the notion of a path set and a cut set are used.
A minimum path set of an IDEF3 model is the minimum set of activities whose
functioning ensures the functioning of the model. Consider the IDEF3 process model in Figure
5. There are four minimum path sets {1, 2, 3, 4, 5, 7, 9}, {1, 2, 3, 4, 5, 8, 9}, {1, 2, 3, 4, 6, 7,
9}, {1, 2, 3, 4, 6, 8, 9} exist in this system. One can see in Figure 5, that system will function if
all the activities of at least one minimum path set are functioning. Therefore, the structure
function of the system in Figure 5 is defined by (4) (Barlow and Proschan 1981).
∅ (x) = max{min{x1, x2, x3, x4, x5, x7, x9}, min{x1, x2, x3, x4, x5, x8, x9},
min{x1, x2, x3, x4, x6, x7, x9}, min{x1, x2, x3, x4, x6, x8, x9}} (4)

1 if the IDEF 3 model functions

where, ∅ = 
0 if the IDEF 3 model fails

1 if activity j functions
xj = 
0 if activity j fails
The structure function ∅ (x) for path sets of the model in Figure 5 is represented in
Figure 6.

4
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

1 2 3 4 5 7 9

1 2 3 4 5 8 9

1 2 3 4 6 7 9

1 2 3 4 6 8 9

Figure 6. The representation of structure function ∅ (x) for path sets

A minimum cut set of an IDEF3 model is the minimum set of activities, whose failure
leads to the failure of the model. Therefore, the model fails if at least one minimum cut set is not
functioning. There are seven minimum cut sets in the IDEF3 process model in Figure 5, namely
{1}, {2}, {3}, {4}, {5, 6}, {7, 8}, {9}. Hence, the structure function ∅ (x) of this IDEF3
model is defined in (5).
∅ (x) = min{max{x1}, max{x2}, max{x3}, max{x4} max {x5, x6},
max{x7, x8}, max{x9}} (5)
The representation of ∅ (x) for cut sets of the model in Figure 5 is shown in Figure 7.

5 7
1 2 3 4 9
6 8

Figure 7. The representation of structure function ∅ (x) for cut sets

The minimum path and minimum cut representations developed above provide means
for systematically computing the reliability of the system Rs simply by taking the expectation of
structure function (Barlow and Proschan 1981), that is
Rs = E(∅ (x)) (6)

Assuming in equation (4) the probability of occurrence of paths {1, 2, 3, 4, 5, 7, 9} and {1, 2,
3, 4, 6, 7, 9} equals P7, and the probability of occurrence of paths {1, 2, 3, 4, 5, 8, 9} and {1,
2, 3, 4, 6, 8, 9} is P8, an upper bound on the reliability of IDEF3 process model can be
obtained by setting P7 = P8 = 1. The latter is equivalent to replacing an exclusive OR junction
with an OR junction box. Furthermore, one may see that in determining the cut sets of the
IDEF3 model in Figure 5 the exclusive OR junction box is treated as an OR junction.
Replacing it with an AND junction leads to a lower bound of the reliability of the IDEF3
process model. In this case the minimum cut sets are: {1}, {2}, {3}, {4}, {5, 6}, {7}, {8},
{9}.

3.1. The Path Tree Algorithm

In this section, a path tree algorithm for deriving the minimum path and cut sets of an
IDEF3 model is presented.

5
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

The reliability network of an IDEF3 process model corresponds to the original model
except that the junction boxes are replaced with edges representing the corresponding logical
boxes. Table 1 presents the elementary IDEF3 graphical components and the corresponding
reliability network.
The two reliability networks for an AND junction presented in Table 1 are equivalent
from the reliability point of view, however, they are different from the process flow prospective.
Using the representation from Table 1, one can obtain the reliability network corresponding to
the IDEF3 process model in Figure 5 (see Figure 8).

Table 1. IDEF3 graphical components and corresponding reliability network

IDEF3 Graphical Components Equivalent Reliability Network

2 1 2 3

1 & & OR

3 1 3 2

2 2

1 O O 1

3 3

2 2

1 X X 1

3 3

1 2 1 2

The terminology used by an algorithm to determine path sets is presented.

Terminology:
Start point Node in the network that has only outgoing arrows (O point).
End point Node in the network that has only incoming arrows (D point).
Minimum path A path from the start point O to the end point D where no node is
traversed more than ones.
Minimum cut set A set with minimum number of activities whose failure ensures the
failure of the end point D.
Tree A representation that shows the relationship among all activities in
the model from the start point O to the end point D.

6
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

5 7
O D
1 2 3 4 9

6 8

Figure 8. Reliability network of the IDEF3 model in Figure 5

The algorithm to determine the path sets in an IDEF3 model is presented next.

Algorithm 1
Step 1: Begin the start (origin) point O at the top of the tree (level k = 1).
Step 2: Place the activities connected to the start point O at the next level (level k = 2).
Step 3: Obtain level k + 1 activities connected to the activities at level k.
Step 4: If there is no more activities at the level k + 1, set k = k + 1 and repeat step 3.
Step 5: Repeat steps 3 and 4 until each activity at the last level is the end point D.

To illustrate Algorithm 1, consider the reliability network of the IDEF3 process model in
Figure 5 shown in Figure 8.
In step 1 of the algorithm, the start point 1 is placed on the top of the tree (level 1).
There is only one activity 2 connected to the start point 1. Therefore, this activity is placed at the
next level (level 2). Following the steps of the algorithm, the tree in Figure 9 is obtained. It is
seen in Figure 9 that each lowest level activity corresponds to the end point D = 9.
From the tree in Figure 9 one can easily obtain the minimum pat sets of the IDEF3
model, i.e., {1, 2, 3, 4, 5, 7, 9}, {1, 2, 3, 4, 5, 8, 9}, {1, 2, 3, 4, 6, 7, 9}, {1, 2, 3, 4, 6, 8, 9}.
Those are the paths which connect the start point O = 1 to end point D = 9 in the tree.

1 Level 1

2 Level 2

3 Level 3

5 Level 4
6

4 4 Level 5

7 8 7 8 Level 6

9 9 9 9 Level 7
Figure 9. Tree structure for the network in Figure 8

7
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

Algorithm 1 presented above identifies the minimum path sets where no cycles exist in
an IDEF3 model. An extension of Algorithm 1 for generation of minimum path sets of an
IDEF3 model with cycles is presented next.

Algorithm 2
Step 1: Start with the start point O at the top of the tree (level k = 1).
Step 2: Place the activities connected to the start point O at the next level (level k = 2).
Step 3: Obtain level k +1 activities connected to activities at level k.
Step 4: If any activity obtained at level k + 1 is already included in a path, disregard this
activity.
Step 5: If an activity obtained at level k + 1 is the same as any activity obtained at the higher
level of any other path, fathom the path.
Step 6: If there is no more activities at the level k + 1, set k = k + 1 and repeat steps 3 - 5
until each activity at the last level is fathomed or corresponds to the end point D.

To illustrate Algorithm 2, consider the IDEF3 process model in Figure 10 and the
corresponding reliability network in Figure 11.

R11
11
R5
5
R3 R10
3 O R6 O 10 X
6
R1 R2 R9
1 2 & R7 & 9
7
R4
4 X R8 X
8

Figure 10. IDEF3 process model

11

5 7
O D
1 2 3 10 4 9

6 8

Figure 11. Reliability network of the IDEF3 model in Figure 10

Following the steps of Algorithm 2, the tree in Figure 12 is obtained. One can see from
Figure 12 that each lowest level activity is either fathomed or corresponds to the end point D.

8
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

From tree in Figure 12 one can obtain the minimum pat sets of the IDEF3 model, i.e., {1, 2, 3,
5, 10, 4, 7, 9}, {1, 2, 3, 5, 10, 4, 8 9}, {1, 2, 3, 6, 10, 4, 7, 9}, {1, 2, 3, 6, 10, 4, 8, 9}, {1,
2, 3, 5, 10, 11, 4, 7, 9}, {1, 2, 3, 5, 10, 11, 4, 8, 9},{1, 2, 3, 5, 10, 11, 6, 4, 7, 9}, {1, 2, 3,
5, 10, 11, 6, 4, 8, 9}, {1, 2, 3, 6, 10, 11, 4, 7, 9}, {1, 2, 3, 6, 10, 11, 4, 8, 9}, {1, 2, 3, 6,
10, 11, 5, 4, 7, 9}, {1, 2, 3, 6, 10, 11, 5, 4, 8, 9}. Those are the paths that connect the start
point O to end point D in the tree by expanding the fathomed activities whenever it is
appropriate.
1 Level 1

2 Level 2

3 Level 3

5 6 Level 4

10 10 Level 5

4 11 11 4 Level 6

7 8 5 6 5 6 7 8 Level 7

9 9 10 10 9 9 Level 8
Fathom
Disregard 4 Disregard Level 9
4
Figure 12. Tree structure for the network in Figure 11

Note that only the first repeating sequence, e.g., (5, 10) of the path set {1, 2, 3, 5, 10, 11, 4, 7, 9},
has been shown in the tree in Figure 12.

3.1.1. Film Deposition Case Study

Figure 13 shows an IDEF3 model of the film deposition process (Nguyen and Bachner 1987).
Knowing the reliability of the activities involved in the process in Figure 13, the reduction approach
presented earlier in this chapter determines the reliability of the process as Rs = R1R234567R8 = 0.884,
where R234567 = (P2R2R4) + (P3R3R567), R567 = R5 + R6R7 - R5R6R7, and P2 = 0.45, P3 = 0.55 is the
occurrence probability for activity 2 and 3, respectively, P2 + P3 = 1. This result could not be obtained
with the traditional reliability formulas due to logical connectors involved in the process model. In the
system in Figure 2, a circuit is formed through the execution of a set of appropriate activities. For
example, one such set may include activities {1, 2, 4, 8} (MOD film technology). Other possible
subsets of activities guaranteeing the formation of a circuit are the sets {1, 3, 5, 8} and {1, 3, 6, 7, 8}
(thin film technology). For a small size model, similar to the one presented in Figure 13, these sets could
be possibly obtained by the inspection of the model. However, when the number of activities and logical
junctions in the model is large, which is the case of full-scale industrial processes, formal approaches
have to be used. Moreover, these sets are successfully used in evaluating the reliability of the system as
well as in developing technique for tracking the functional state of a process under, given the state of
activities. For example, assuming that in the process in Figure 13 the sputtering and electroplating
operations have failed, Algorithm 1 presented in this section determines the only set of activities {1, 3, 6,

9
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

7, 8} that guarantees formation of a circuit with the corresponding process reliability Rs = R1R3R6R7R8
= 0.824.

R2 = 0.98 R4 = 0.95
Activity Number and Name:
2 4 1 - Evaluate a circuit
2 - Perform sputtering
R1 = 0.98 R8 = 0.96 3 - Perform screen printing
R5 = 0.95
1 X X 8 4 - Perform pattern plating
5 5 - Obtain final layer by electroplating
R3 = 0.95 6 - Obtain final layer by adding
R6 = 0.97
3 O O a fritless gold
6
7 - Perform subtractive etching
& 8 - Form a circuit
R7 = 0.95 &
7
Figure 13. IDEF3 model of the film deposition process

3.2. Evaluation of Minimum Cut Sets

The earlier presented Algorithm 1 and Algorithm 2 generate minimum path sets of an
IDEF3 model. The algorithm presented in this section determines the minimum cut sets provides
that minimum path sets have been determined. In the literature, the minimum cut sets by
combining nodes that break the minimum path sets (Rai and Aggarwal 1978, Rosenthal 1979).
The algorithm described next follows the same logic.

Algorithm 3
Step 0: Initialize the current solution set S1 = {Ø}, and two working sets S2 = {Ø} and S3
= {Ø}.
Step 1: Place activities associated with the start point O and end point D in set S1.
Step 2: Select a minimum path set. If any activity from the minimum path set selected occurs in
the remaining minimum path sets, add this activity to set S1.
Step 3: Place the activities of the minimum path set that are not in set S1 in set S2.
Step 4: Place all the activities of the model that are not in set S1 in set S3, and set the number of
activities in a q-tuplet, q = 2.
Step 5: Find all possible q-tuplets of the activities in S2 and S3. Each q-tuplet is a cut set if
each minimum path includes one of its activities. Add the cut sets obtained to the set S1.
If the maximum number of activities in the cut sets generated q < total number n of
minimum path sets in the model, then repeat step 5 by creating (q + 1)-tuplets from S2
and S3; otherwise, stop.

To illustrate Algorithm 3, consider the IDEF3 reliability network in Figure 8. The minimum path
sets obtained from the path tree algorithm are: {1, 2, 3, 4, 5,7, 9}, {1, 2, 3, 4, 5, 8, 9}, {1, 2,
3, 4, 6, 7, 9}, {1, 2, 3, 4, 6, 8, 9}.

Step 0: Initialize: S1 = {Ø}, S2 = {Ø}, S3 = {Ø}.

Step 1: Activities 1 and 9 are added to set S1. Therefore S1 = {1, 9}.
Step 2: The path {1, 2, 3, 4, 5, 7, 9} is obtained.

10
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

Activities 2, 3, and 4 appear in each minimum path set.

Therefore, S1 = {1, 2, 3, 4, 9}
Step 3: Activities 5 and 7 are included in S2 = {5, 7}.
Step 4: The activities of the model that are not included in the set S1 are: 6 and 8.
Therefore S3 = {6, 8}.
Step 5: The possible 2-tuplets are: (5, 6), (5, 8), (7, 6), (7, 8).
Since minimum path set includes one element of the 2-tuplets (5, 6) and (7, 8),
therefore S1 = {1, 2, 3, 4, 9, (5, 6), (7, 8)}
The maximum number of activities in the cut sets generated q = 2 < total number of
minimum path sets n = 4. Create 3-tuplets from activities in S2 and S3. The 3-tuplets
are: (5, 6, 8), (7, 6, 8). Note that the 3-tulets (5, 7, 6), (5, 7, 8) are not listed as they
include (5, 7) that belong to the same minimum path set. Since q = 3 < n. Create 4-
tuplets from the activities in S2 and S3. The 4-tuplet is (5, 6, 7, 8). S1 = {1, 2, 3, 4, 9,
56, 78} are the minimum cut sets of the IDEF3 model. One may see that combinations
(5, 6, 8), (7, 6, 8), (7, 5, 6), (5, 7, 8), and (5, 6, 7, 8) are not considered, as they
include the tuplets (5, 6) and (7, 8).

3.3. The Cut Set - Activity Incidence Matrix

For the cut sets of an IDEF3 model, a cut set - activity incidence matrix [a ij ] can be
constructed. The cut set - activity incidence matrix includes Rj entries, where Rj is the reliability
of activity j in IDEF3 model and indicates that the activity j is included in the minimum cut set i.
The cut set - activity incidence matrix allows one to check the functional state of the model
under the current state of activities. As soon as the states of the activities are determined, one
can obtain the reduced cut set - activity incidence matrix by removing the columns
corresponding to all activities that have failed. If every row in the reduced matrix contains at
least one positive entry Rj, then the IDEF3 model functions.
Activity
1 2 3 4 5 6 7 8 9
1 R1 C
2 R2 u
3 R3 t
[aij ] = 4 R4
5 R 5 R6 s
6 R 7 R8 e
7 R9 t

Figure 14. Cut set - activity incidence matrix of IDEF3 model in Figure 5

To illustrate the above concept, consider the cut set - activity incidence matrix in Figure
14. Assume the activities 6 and 7 are not functioning. One can transform matrix [ a ij] into [ a1ij]
by removing the columns corresponding to non-functioning activities 6 and 7 (see Figure 15).
Each row of the matrix [ a1ij] contains at least one positive entry Rj, therefore one may conclude
that under the current state of the IDEF3 activities, the process is operational.

11
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

Activity
1 2 3 4 5 8 9
1 R1 C
2 R2 u
3 R3 t
[a1ij ] = 4 R4
5 R5 s
6 R8 e
7 R9 t

Figure 15. Transformed matrix [ a1ij]

3.4. The Path Set - Activity Incidence Matrix

Using the concept presented in the previous section, one can construct a path set -
activity [Pij] incidence matrix for an IDEF3 model. The [Pij] matrix consists of entries Rj, where
Rj is the reliability of activity j and indicates that this activity is included in the minimum path set i.
The path set - activity incidence matrix allow one not only to check the functional state of the
model, but also to calculate the reliability of IDEF3 model under the current state of activities.
The algorithm for calculating reliability for an IDEF3 model from [Pij] matrix is presented next.

Algorithm 4
Step 1: Select the columns associated with activities that have failed in the model and draw
a vertical lines vj through each of these columns.
Step 2: For each Rj crossed by the vertical line vj, draw a horizontal line hi.
Step 3: Transform the incidence matrix [ Pij] into [ P1ij] by removing rows and columns
corresponding to all the vertical and horizontal lines drawn in Step 1 and 2.
Step 4: If [ P1ij] is empty, then stop. Model does not function.
Otherwise, calculate r i = ∏ R j for all j of matrix [P 1ij] .
j

Step 5: Calculate the reliability of the new IDEF3 model as R s = 1 - ∏ (1 - r i ) , where i ranges
i
over all minimum path sets in matrix [ P1ij] .
To illustrate Algorithm 4, consider the path set - activity incidence matrix of IDEF3
model in Figure 16. Assume the activities 5 and 7 are not functioning.
Activity
1 2 3 4 5 6 7 8 9
1 R1 R2 R3 R4 R 5 R7 R9
2 R1 R2 R3 R4 R 5 R8 R9 Path
[Pij ] = set
3 R1 R2 R3 R4 R6 R7 R9
4 R1 R2 R3 R4 R6 R8 R9

Figure 16. Path set - activity incidence matrix of the IDEF3 model in Figure 5

Steps of Algorithm 4
Step 1: Columns 5 and 7 of matrix [Pij] are selected and vertical lines v5 and v7 are
drawn.

12
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

Step 2: Three horizontal lines h1, h2, and h3 are drawn. The results of Step 1 and 2 are
presented in matrix (9).
Activity
1 2 3 4 5 6 7 8 9
1 R1 R2 R3 R4 R5 R7 R9 h1
2 R1 R2 R3 R4 R5 R8 R9 h2 (9)
[Pij ] =
3 R1 R2 R3 R4 R6 R7 R9 h3
4 R1 R2 R3 R4 R6 R8 R9
v5 v7
Step 3: Matrix (9) is transformed into matrix (10).
1
1 2 3 4 6 8 9
[ Pij] = 4 R 1 R2 R3 R 4 R6 R8 R9 (10)

Step 4: Since matrix (10) is not empty, therefore

r i = ∏ R j = R1R 2 R3 R4 R6R 8R9
j

Step 5: The reliability of the new IDEF3 model is

R s = 1 − ∏ (1- ri ) = R1 R 2 R3 R 4 R 6 R8 R9 = 0.85
i

4. SUMMARY
Reliability analysis of IDEF3 models is of interest to practitioners and researchers for
several reasons. It produces critical activities of the process, improves its performance, and
decreases downtime and operating cost of the process. This paper extends the system reliability
evaluation techniques, i.e., the system reduction approach and minimum path and cut sets
method for reliability evaluation of IDEF3 models. Representation of IDEF3 models as
reliability graphs, generation of the minimum path and cut sets of IDEF3 models with a path tree
algorithm, and reliability analysis of IDEF3 models are the issues discussed in this chapter. An
algorithm for computing reliability of an IDEF3 model from a path set - activity incidence matrix
is also presented.

REFERENCES
1. Ang, C. H. and Gray, R. (1993), IDEF0 modeling for project risk assessment, Computers
in Industry, 22, 31 - 45.
2. Arunkumar, S. and Lee, S. H. (1979), Enumeration of all minimal cutsets for node pair in a
graph, IEEE Transactions on Reliability, 28, 51 - 55.
3. Barlow, R. and Proschan, F. (Eds) (1981), Statistical theory of reliability and life
testing: probability models, McArdle Press, Inc., Silver - Spring, MD.
4. Kusiak, A., Larson, T. N., Wang, J. (1994), Reengineering of design and manufacturing
processes, Computers and Industrial Engineering, 26, 521 - 536.
5. Kusiak, A. and Larson, N. (1994), System reliability and risk assessment: a quantitative
extension of IDEF methodologies, Stanford University: AAAI Spring Symposium, 88 -
93.
6. Kusiak, A. and Zakarian, A. (1996), Reliability Evaluation of Process Models, IEEE
Transactions on Components, Packaging, and Manufacturing Technology - Part A,
19, 268-275.
7. Kusiak, A. and Zakarian, A. (1996a), Risk Assessment of Process Models. Journal of
Computers and Industrial Engineering, 30, 599-610.

13
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

8. Menzel, C., Mayer, R. J., Edwards, D. D. (1994), IDEF3 process descriptions and their
semantics, in Intelligent systems in design and manufacturing, (C. H. Dagli and A.
Kusiak, Eds), ASME Press, New York, 172 - 212.
9. Nguyen, P. H. and Bachner, F. J. (1987), A New Metallization Technology for Advanced
Interconnects on Substrates. IEEE Transactions on Components, Hybrids, and
Manufacturing Technology, 12, 571 - 576.
10. Rai, S. and Aggarwal, K. K. (1978), An efficient method for reliability evaluation of a
general network, IEEE Transactions on Reliability, 27, 206 - 211.
11. Rosenthal, A. (1979), Approach to comparing cut - set enumeration algorithms, IEEE
Transactions on Reliability, 28, 62 - 65.
12. Ross, D. (1985), Applications and extensions of SADT, Computer, 18, 25 - 34.
13. O'Sullivan, D. (1994), Manufacturing Systems Redesign: Creating the Integrated
Manufacturing Environment. Prentice Hall, Englewood Cliffs, New Jersey.

QUESTIONS
1. Why traditional methods can not be used for reliability evaluation of IDEF3 process
models?
2. Name the formulas for evaluation of reliability of process models.
3. What are the three main methods for the reliability evaluation?
4. Which reliability methods can be easily computerized and why?
5. What are the advantages and disadvantages of the reduction approach?
6. When a path set method should be used?
7. When the cut set method should be used?
8. Which method generates a lower bound of the process model reliability?
9. Which method generates an upper bound of the process model reliability?
10. Which algorithm can be used to compute reliability of a process model with cycles?

PROBLEMS
1. List all minimum path sets and cut sets of the IDEF3 model in Figure A1.
7

2 5 & &

1 O O 4 & 8 & 9

3 6

Figure A1. Process model

2. Consider the product realization process represented with the IDEF3 model in Figure A2. Determine the
minimum path sets and cut sets. Calculate the overall reliability of the process model with the reduction approach
and using the minimum path sets. The probabilities associated with the exclusive OR junction are P2 = .6 and P3 =
.4.

14
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

Activity Reliability
1 .95
2 .98
3 .90
4 .90
5 .96
6 .92
7 .94
8 .96

4 10
2 4
2

1 14
X X
1 8
3 15
5 6 11
3
O 5
O
7
16
8 12
& &
6

9 13
7

Figure A2. Process model

3. For the product development process represented with the model in Figure A3, determine the minimum path
and cut sets. Provide interpretation of the reliability obtained with the two methods. Assume that the reliability of
each activity Ri = 0.95, i = 1, ..., 11.

R9
9
R5
5
R3 R10
3 & R6 & 10 X
6
R1 R2 R 11
1 2 X R7 X 11
7
R4
4 O R8 O
8

Figure A3. Process model

4. Find the overall reliability of the manufacturing process represented with the model in Figure A4. Use the
reduction and minimal path methods. The probabilities associated with the exclusive OR junction are P23 = .8 and
P24 = .2.

15
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

R5 = 0.96
5
R3 = 0.98
3 O R6 = 0.97 O
6 R10 = 0.98
R1 = 0.93 R2 = 0.99 R9 = 0.98
1 2 X R7 = 0.97 X 9 10
7
R4 = 0.99
4 & R8 = 0.96 &
8

Figure A4. Process model

4. For the process model in Figure A5 and the data in Tables A1 and A2 determine:
(a) The model reliability using the reduction method,
(b) The minimum and maximum value of the process reliability,
(c) What action would you take to increase the value of the minimum reliability path by 5% ?

P1
O 2 O

1 X
X 7

P2
4 & 5 &

Figure A5. An IDEF3 process model

Table A1. Reliability data Table A2. Probability data

Activity Reliability Probability Value

1 0.85 P1 0.3
2 0.90 P2 0.7
3 0.90
4 0.95
5 0.90
6 0.88
7 0.94

6. Consider the product development process represented with the IDEF3 model in Figure A6. The risk of
performing activities on time is measured with reliability.

16
A. Kusiak, Engineering Design: Products, Processes, and Systems, Academic Press, San Diego, CA, 1999.

The reliability of each activity is as follows:

Activity Reliability
1 .95
2 .95
3 .90
4 .95
5 .96
6 .94
7 .96
7 .97

The probabilities associated with the exclusive OR junction are P2 = .2 and P3 = .8

2 4

X X
1 8
5

& &
3
6
O O

Figure A6. Process model

(a) Compute the minimum min cut set and maximum min path set.
(b) Compute the overall reliability of the model with the reduction method.

17