Sie sind auf Seite 1von 28

INTRODUCING BUFFER INVENTORIES IN

THE RBD ANALYSIS OF PROCESS


PRODUCTION SYSTEMS
1
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

CONTENT
Introduction
Objective
Throughput analysis of process production lines by means of
RBDs
Proposition of RBD models for buffered production lines
Validation of proposed method.
Conclusion
2
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

INTRODUCTION
Modeling a system by using RBDs is used to make reliability and availability
analysis.
Presence of buffer inventory plays a role on the propagation of the effect of
failure along the entire systems.
Buffer inventory level should provide proper isolation time before material
starvation and blocking of production.
The buffer inventory level between two production stages.
3
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

OBJECTIVE
To develop RBDs capability to analyze material buffering
To improve RBDs potentials toward including analysis of material starvation and
blocking of production.

4
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

THROUGHPUT ANALYSIS OF PROCESS PRODUCTION LINES BY
MEANS OF RBDS
RBDs allows to reach a multileveled hierarchical model by using a simple procedure.
RBDs can deal with complex production structures.
5
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

PROPOSITION OF RBD MODELS FOR BUFFERED
PRODUCTION LINES
Three different RBD models are proposed for analyzing the behavior of
buffered production lines

First RBD model does not consider the presence of buffers
Second model is based on the assumption that buffers perfectly isolate
the production stages of the line.
Third model considers a non perfect isolation of buffers.

6
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Model 1.
The subsystems are kept as two coupled components so that when a failure occurs
at a subsystem, the other also stops.
Series logic of production system
7
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

RBD model of the production line.
8
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Availability and throughput analysis
9
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Model 2.
The buffer perfectly isolates the two subsystems.
The buffer inventory is high enough to avoid a propagation of the failure
downward, when the failure occurs upwards.
The buffer capacity is high enough to avoid propagation upward, when the
failure occurs downward.
Buffered line with perfect isolation
10
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Availability and throughput analysis
where:
K = 1, 2(1 = upward and 2 = downward subsystem);
A
k =
subsystem availability;
Th
k
nom
= nominal throughput of the subsystem k;
Th
best
= throughput of the line (best case).
11
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Model 3.
Its distinctive feature is the non perfect isolation attitude of the buffer.
Time losses can be propagated along the line in conjunction with blocking or
starvation effects.
RBD logics can be used to identify the state of the subsystems with respect to the
production flow degradation
(i) PFD corresponds to a total reduction of production flow, in case the failure occurs
in a subsystem behaving as a series reliability structure;
(ii) PFD corresponds to a partial reduction of production flow, in case the failure
occurs in a MSS reliability structure (the impact factor allows to determine the
reduction rate);
(iii) PFD corresponds to a no reduction of production flow, in case the failure occurs
in a standby or, in general, in a parallel structure .
12
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Modeling the dynamics of the buffer inventory level
13
W
1,k
no PFD happens after failure
W
2,k
when a partial PFD happens after failure
W
3,k
- when a total PFD happens after the failure
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Analytical solution of the model
Identification of the bottleneck and the non-bottleneck subsystems.

State space analysis of the non-bottleneck subsystem.

Joint availability analysis of non-bottleneck subsystem and the buffer
inventory.

Availability and throughput analysis of the line.
14
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Probability of PFDs
Considering the degraded states w
i,1
of the non-bottleneck subsystem (i.e.,
subsystem upward, so k is fixed to 1).
where:


0,i,1
is the failure rate leading to a transition from the nominal state (indexed
with 0) of the non-bottleneck subsystem (indexed with 1)to any ith degraded
state having a partial or total PFD;

I is total number of degraded states having a partial or total PFD.
15
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Probability that PFDs cause a time loss.
where:
IL
b,t1
is the buffer inventory level at time t
1
;
t
1
is the time when a machine failure occurs;
t
2,i
is the time when the buffer becomes empty;
t
2,i
t
1
is the isolation time toward material starvation.
M(t) is the maintainability of the machine under repair 16
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Expected time loss per PFD
Where ;
17
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Joint availability
The expected time loss per failure of the non-bottleneck can be calculated by,
Total number of failures at the non-bottleneck can be,
where T is the time when the line is open for production.
Equivalent subsystem availability, expressing the joint
availability provided by the non-bottleneck subsystem and
the buffer inventory is
18
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Availability and throughput analysis
where:
A
k =
subsystem availability;
Th
k
nom
= nominal throughput of the subsystem k;
Th
real
= throughput of the line
19
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

VALIDATION OF THE PROPOSED METHOD
The case of a production line taken out from the mining industry is carried on in order
to demonstrate proposed method.
20
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Design of simulation experiments
21
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Three data sets were considered sufficient for the validation along a range of
operating conditions, adequately ample to cover real situations as well as to tackle
border cases.
Input data sets
A first data set was selected in order to test the models under conditions resembling
the real industrial case (data set 1).

a condition of quasi perfect isolation (data set 2)

a condition of quasi perfect coupling between the upward and downward
subsystems (data set 3).

22
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

TH
1
nom
=1.1 units/h;
TH
2
nom
=1.0 units/h.
BC = 4.8 units;
Il
min =
0.5xBC = 2.4 units.
Data set 1
Data set 2
Data set 3
TH
1
nom
=1.1 units/h;
TH
2
nom
=1.0 units/h.
TH
1
nom
=1.1 units/h;
TH
2
nom
=1.0 units/h.
BC = 15 units;
Il
min =
0.5xBC = 7.5 units.
BC = 1units;
Il
min =
0.5xBC = 0.5units.
23
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Performance analysis
Model 3.
Model 1.

A
worst
= 91.90%
Th
worst
= 0.92 unit/h

Model 2.

A
best
= 96.93%
Th
best
= 0.97 unitlh
24
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

State space analysis.
25
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

CONCLUSION
The main result of this paper is the demonstration of a new modeling approach to
introduce buffer inventories in the RBD analysis of a production line with complex
reliability structures.

The method is based on
o Integration of the RBD availability analysis with a state space analysis of the
PFDs and related buffer dynamics.
o The analytical modeling of the joint availability resulting from the combined
action of the non-bottleneck subsystem and the buffer.

APPLICATION

When BC is low model 1 should be used.
Greater the size of BC model 2 and model 3 are advisable.
The quasi perfect isolation by model 2 can achieved only in specific conditions.
Model 1 &3 are used where machines have repair time exceeding isolation time
guaranteed by the buffer.
26
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

REFERENCES
Marco Macchi Introducing buffer inventories in the RBD analysis of process
production systems Reliability Engineering and System Safety 104 (2012) 8495

Barabady J, Kumar U. Reliability analysis of mining equipment: a case study of a
crushing plant at Jajarm Bauxite Mine in Iran. Reliability Engineering & System Safety
2008;93:64753.
27
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

28
f
b
.
c
o
m
/
s
u
l
f
i
c
k
e
r
.
a
l
i
.
i
s
m
a
i
l

Das könnte Ihnen auch gefallen