Flow Line Balancing Problem &
Extensions/ Enhancements
A Simple Flow Line
Assume three activities with times of 13
min/unit, 11 min/unit and 8 min/unit, each
staffed by one worker.
Demand is 125 scooters per week, and
the process operates 35 hours per week.
Components
Activity 1
Activity 2
Activity 3
13 min.
11 min.
8 min.
(Problem Scenario from Cachon p.56-58)
Finished Product
Table 4.1 Basic Calculations Related to Idle Time
Worker 1
Worker 2
Worker 3
Activity time
13 min/unit
11 min/unit
8 min/unit
Capacity
1/13 unit/minutes
= 4.61 units/hr
1/11 units/minutes
= 5.45 units/hr
1/8 unit/minutes
= 7.5 units/hr
Process capacity
Minimum {4.61 units/h, 5.45 units/h, 7.5 units/h} = 4.61 units/h
Flow rate
Demand = 125 units/week = 3.57 units/hr
Flow rate = Minimum {demand, process capacity} = 3.57 units/hr
Cycle time
1/3.57 hours/unit = 16.8 minutes/unit
Idle time
16.8 minutes/unit
- 13 minutes/unit
= 3.8 minutes/unit
16.8 minutes/unit
- 11 minutes/unit
= 5.8 minutes/unit
16.8 minutes/unit
- 8 minutes/unit
= 8.8 minutes/unit
Utilization
3.57 / 4.61 = 77%
3.57 / 5.45 = 65.5%
3.57 / 7.5 = 47.6%
Average Labor
Utilization
= 1/3 x (77.4% + 65.5% + 47.6%) = 63.5%
Or = 32 / (32 + 18.4) = 63.5%
{Total = 18.4
min/unit}
p.62
Increasing Capacity by Line Balancing
Comparing the utilization levels in Table 4.1 reveals a strong imbalance
between workers. Imbalances within a process cause micro-level
internal supply-demand mis-matches.
Line balancing attempts to reduce such imbalances through:
Increasing the capacity of the process by reallocating either workers
from underutilized resources to the bottleneck or work from the
bottleneck to the underutilized resources.
Increasing the efficiency of the process by better (analytical) balancing
p.63
De-bottlenecking through
Manual Line Balancing
Scenario: Increased Demand
Demand now increases to 200 units/ week i.e. cycle time
35 hours/ 200 units = 10.5 mins.
Bottleneck now at Worker 1, who is utilized 100% and
produces once every 13 minutes i.e. 161.5 units/ week
(35*60/13)
Worker 2 utilization 11/13 = 84.6%
Worker 3 utilization 8/13 = 61.5%
Overall utilization (13+11+8)/3*13 = 82%
p.64
A Simple Flow Line
Assume three activities with times of 13
min/unit, 11 min/unit and 8 min/unit, each
staffed by one worker.
Demand is 125 scooters per week, and
the process operates 35 hours per week.
Components
Activity 1
Activity 2
Activity 3
13 min.
11 min.
8 min.
(Problem Scenario from Cachon p.56-58)
Finished Product
Table 4.2 Detailed Station Task Times
Total Task time: 1890 sec.
p. 65
Revised calculations
Note that the original estimated times are now replaced
by actual detailed times.
Worker 1 time 792 seconds, i.e. 159.1 units/ week
(35*60*60/792)
Worker 2 utilization 648/792 = 81.8%
Worker 3 utilization 450/792 = 56.8%
Overall utilization (792+648+450)/3*792 = 79.5%
Balancing through Reallocation - 1
Shift last (118 sec.) task from Worker 1 to Worker 2 who
is now overloaded =>
Shift last two (56 & 75 sec.) tasks from Worker 2 to
Worker 3
Worker 1 still bottleneck: time 674 sec. i.e. 186.9 units/
week (35*60*60/674)
Worker 2 utilization 635/674 = 94.2%
Worker 3 utilization 581/674 = 86.2%
Overall utilization (674+635+581)/3*674 ~ 93.4%
p. 66
Table 4.2 Detailed Station Task Times
Total Task time: 1890 sec.
p. 65
Balancing through Reallocation - 2
Shift 2nd last (51 sec.) task from Worker 1 to Worker 2
who is now overloaded =>
Shift 3rd last (84 sec.) tasks from Worker 2 to Worker 3
Worker 3 now bottleneck: time 665 sec. i.e. 189.5 units/
week (35*60*60/665)
Worker 1 utilization 623/665 = 93.7%
Worker 2 utilization 602/665 = 90.5%
Overall utilization (623+602+665)/3*665 ~ 94.7%
See Fig. 4.5 for detailed allocations
p. 6667
Managing Further Demand
Increase
Increased Demand
Demand now 700 units/ week i.e. cycle time 35
hours/ 700 units = 180 sec.
Have to explore new options
p. 66
Parallel Identical Lines
Fig. 4.6
p. 67-68
Total capacity: 4*189.5 (as before) = 758
Same utilization levels
Independence vs. Pooling
Alternatively: use 3 parallel lines to get 3*189.5 = 568.5
output, and use overtime for the remaining 131.5
Station Paralleling (Identical)
Continuing from final solution of 189.5/ week line
Workers required at Step 1: 623/180 = 3.46 ~ 4 (86.5%)
Workers required at Step 2: 602/180 = 3.34 ~ 4 (83.5%)
Workers required at Step 2: 665/180 = 3.69 ~ 4 (92.2%)
Pooling vs. Dedication
Flow & Handling
Fig. 4.6
p. 67-68
Detailed Balancing with Task
Specialization: Revised Calculations
Bottleneck Worker 2: time 180 seconds, i.e. 700 units/
week (35*60*60/180)
Lowest utilization at Worker 3: 115/180 = 63.9%
Overall utilization 1890/12*180 = 87.5%
Fig. 4.6/ p. 68
Detailed
Station
Reallocation
Total Task time: 1890
sec. as before
Table
4.3/ p.
70
Reduced Specialization: Parallel
Work Cells
One worker performs all tasks
Number of workers required to meet demand = 1890/ 180
= 10.5 ~ 11 (saving of one worker due to task pooling)
Fig. 4.8/
p.71
Summary of Line Configurations
Analytical Line Balancing
Balancing Terminology Review
RM
Stn. 1
Stn. 2
Stn. 3
FG
Task/ Element: Small,
distinct subset of work,
taking time tk
Station: Place where
task(s) are performed
Cycle Time (c): Time
available at a station
for performing the
tasks assigned
Station Time (STj):
Sum of task times
assigned to a station
(STj < c)
Balancing Concepts Revisited
To get total work accomplished on line:
k tk < N c
Bounds on c:
tmax < c < k tk
Measures of line balancing efficiency:
Balance Delay
d = (N c - k tk) / N c = 1 - k tk / N c
Smoothness Index
SI = j (STmax - ST j )2
where ST is the total of task times assigned to a Station
Balancing Approaches
1. Given c, Minimize N
(approach used in OM courses, often heuristic based)
2. Given N, Minimize c
Both seek to minimize N c
(i.e. total time-capacity available on the line)
Can use algorithms developed for (1) to solve (2):
Iteratively solve (1) for ?
beginning from lower bound ?
until solution results in ?
Balancing Approaches
1. Given c, Minimize N
(approach used in OM courses, often heuristic based)
2. Given N, Minimize c
Both seek to minimize N c
(i.e. total time-capacity available on the line)
Can use algorithms developed for (1) to solve (2):
Iteratively solve (1) for progressively higher cycle times,
beginning from lower bound (clb = ? ),
until solution results in number of stations within N
