Assembly Line Balancing
Jaime Joo
MBA 530 – Section 1
Brigham Young University
Outline
• What is Assembly Line Balancing?
• How can Assembly Line Balancing benefit
your operations?
• Classic approach to ALB
• Let’s practice!
• ALB in the real world
• Conclusions
What is Assembly Line Balancing (ALB)?
ALB is the procedure to assign tasks to
workstations so that:
•Precedence relationship is complied with
•No workstation takes more than the cycle
time to complete
•Operational idle time is minimized
How can Assembly Line Balancing benefit
your operations?
A balanced line:
• Promotes one piece flow
• Avoids excessive work load in some stages
(overburden)
• Minimizes wastes (over-processing, inventory,
waiting, rework, transportation, motion)
• Reduces variation
Unbalanced Line
!?

Zzz
Zzz
Work
Station 1
Work
Station 2
Work
Station 3
Work
Station 4
10 sec
40 sec!
20 sec
15 sec
Undesirable waiting
Overproduction!
Generates waste
Balanced Line




Work
Station 1
Work
Station 2
Work
Station 3
Work
Station 4
25 sec
25 sec
20 sec
15 sec
•Promotes one piece flow
•Avoids overburden
•Minimizes wastes
•Reduces variation
Line Balancing prerequisites
Prior to balancing a line we must:
• Determine the required workstation cycle
time (or TAKT time), matching the pace of
the manufacturing process to customer
demand
• Standardize the process
Classic approach to ALB
Also known as SALBP* (Simple Assembly Line
Balancing Problem), the classic approach to ALB
is an heuristic process to optimize assembly lines
simplifying the problem to a basic level of
complexity
*Dubbed SALBP by Becker and Scholl (2004)
Example
The next table shows the tasks performed in a
production line. Our goal is to combine them into
workstations. The assembly line operates 8 hours
per day and the expected customer demand is
1000 units per day. Balance the line and calculate
the efficiency and theoretical minimum number of
workstations.
Example (cont.)
Task
Task Time (sec) Preceding Task
A
13
-
B
11
A
C
15
A
D
20
B
E
12
B
F
13
C
G
13
C
H
18
D, E
I
17
F, G
J
15
H, I
K
9
J
Total Time:
156
Example (cont.)
• Step 1: Draw a precedence diagram according to the given sequential
relationship
20 sec
11 sec
B
D
18 sec
12 sec
H
13 sec
A
E
15 sec
9 sec
13 sec
J
K
15 sec
F
C
13 sec
G
17 sec
I
Example (cont.)
• Step 2: Determine Takt time or Workstation Cycle Time
C=Production time per day / Customer demand (or output per day)
C= 28800 sec (8 hours) / 1000 units = 28.8
• Step 3: Determine the theoretical number of workstations
required
N= Total Task Time / Takt time
N= 156 / 28.8 = 5.42 (~6 workstations)
Example (cont.)
• Step 4: Define your assignment rules. For
this example our primary rule will be
“number of following tasks” and the
secondary rule will be “longest operation
time”
Example (cont.)
• Step 5: Assign tasks to workstations following the assignment rules
and meeting precedence and cycle time requirements
To form Workstation 1:
11 sec
13 sec
B
A
15 sec
WS1: A+C=28 sec
Cycle Time
met!
C
Following tasks: 5
Lot: 15>11!
Following tasks: 5
Example (cont.)
• Forming Workstation 2:
20 sec
11 sec
13 sec
B
D
12 sec
E
A
B+D>Cycle time!
15 sec
13 sec
C
F
LOT:_F&G>E
WS2: Operation time=24 sec (<C)
13 sec
G
Arbitrarily choose F
Example (cont.)
• Following the same criteria we achieve our balancing with
7 workstations
Workstation Task
1
2
3
4
5
6
7
A
C
B
F
D
G
E
H
I
J
K
Remaining
Task Time Unassigned
Time
13
15.8
15
0.8
11
17.8
13
4.8
20
8.8
13
15.8
12
3.8
18
10.8
17
11.8
15
13.8
9
4.8
Feasible
Task with
Task with
Remaining most
LOT
Tasks
followers
B, C
B, C
C
E, F, G
E, F, G
F, G
E
K
-
Example (cont.)
• Step 6: Calculate Efficiency
– Efficiency= Total Task Time / (Actual number of workstations *
Takt Time)
– Efficiency= 156 / (7*28.8) = 77%
How to interpret this efficiency?
Is this the best efficiency achievable?
Let’s Practice
• We have found a new market for our
product. This market is less demanding so
we have decided not to include a particular
feature, specifically the feature added by
task I. As a consequence, task time in F
drops to 5 seconds and task time in G drops
to 8 seconds. Balance the line according to
the other conditions.
Let’s practice (cont.)
Task
Task Time (sec) Preceding Task
A
13
-
B
11
A
C
15
A
D
20
B
E
12
B
F
13 5
C
G
13 8
C
H
18
D, E
I
17
F, G
J (new I)
15
H, F, G
K (new J)
9
I
Total Time:
156 126
Let’s practice (cont.)
• Let’s take some time to solve this new
problem. This time we will calculate
keeping the primary and secondary rules as
in the original problem.
Let’s practice (solution)
• Precedence diagram
20 sec
11 sec
B
D
18 sec
12 sec
H
13 sec
A
E
15 sec
9 sec
5 sec
I*
J*
15 sec
F
C
8 sec
*Previously J & K respectively
G
Let’s practice (solution)
• Takt time
C = 28,800 sec / 1000 units = 28.8
• Theoretical number of workstations
N = 126/28.8 = 4.38 (~5 workstations)
• Primary rule: number of following tasks
• Secondary rule: longest operation time
Let’s practice (solution)
• Following the rules and observing cycle time and precedence
we obtain:
Remaining
Task Time Unassigned
Time
Workstation Task
1
2
3
4
5
6
A
B
C
E
D
H
G
F
I
J
13
11
15
12
20
18
8
5
15
9
15.8
4.8
13.8
1.8
8.8
10.8
20.8
15.8
0.8
19.8
Feasible
Remaining
Tasks
B, C
E, F, G
F
I
-
Task with
Task with
most
LOT
followers
B
E
Let’s practice (solution)
• Efficiency = 126/(6*28.8) = 73%
Challenge: Is this the best efficiency
achievable? Try to solve with LOT as the
primary rule and you will obtain a 5
workstations balance, increasing efficiency
to 87%
ALB in the real world
The simple ALB problem approach is limited by some
constraints:
• Balance on existing and operating lines
–
–
–
Workstations have spatial constraints
Some workstations cannot be eliminated
Need to smooth workload among workstations
• Multiple operators per workstation
– Different paces among operators, different lead times within the
same workstation
ALB in the real world (cont.)
• Operator spatial constraints
– Different workstation imposed working positions
– More than one task to be performed in what should be the space
for one task
• Multiple Products
– Coping with different products, some operations are needed for
some products but not for others
– Some products can introduce “peak times” in some workstations
• Different task times performed in different shifts
– Particularly when introducing new employees or workers with
some degree of incapacity
Conclusion
• Simply Assembly Line Balancing is a valid
method to optimize assembly lines. However,
many variables found in real operating lines
increase the complexity of the problem. More
complex algorithms have been developed to solve
the difficult task of balancing large scale industrial
lines. Some of them are commercially available in
software.
References
• F. Robert Jacobs & Richard B. Chase “Operations and
Supply Management, The Core” McGraw-Hill/Irwin First
Edition
• Emanuel Falkenauer “Line Balancing in the real world”
Optimal Design
• Paul Swift. http://www.beyondlean.com/linebalancing.html