Little's Law describes the relationship between inventory levels, flow rates, and flow times in production systems. It states that inventory equals the flow rate times the flow time. The document also defines concepts like capacity, utilization, cycle time, batching, and economic order quantity, which are important for analyzing production systems.
Little's Law describes the relationship between inventory levels, flow rates, and flow times in production systems. It states that inventory equals the flow rate times the flow time. The document also defines concepts like capacity, utilization, cycle time, batching, and economic order quantity, which are important for analyzing production systems.
Little's Law describes the relationship between inventory levels, flow rates, and flow times in production systems. It states that inventory equals the flow rate times the flow time. The document also defines concepts like capacity, utilization, cycle time, batching, and economic order quantity, which are important for analyzing production systems.
Little's Law describes the relationship between inventory levels, flow rates, and flow times in production systems. It states that inventory equals the flow rate times the flow time. The document also defines concepts like capacity, utilization, cycle time, batching, and economic order quantity, which are important for analyzing production systems.
Inventory
(I)
=
Flow
Rate
(R)
x
Flow
Time
(T)
Annual
inventory
holding
cost
%/Yearly
turns
=
I
in
(units),
R
in
(units/time),
T
in
(time)
Annual
inventory
holding
cost
as
a
%
of
COGS
R
=
COGS
NOT
sales
Holding
cost/time
period:
h
*
I
(I=avg
inventory)
Days
of
supply
=
T
=
I/R
=
(Annual
I/COGS)
*
365
E.g.
if
it
costs
$12
to
hold
an
item
for
one
year,
then
h
=
$1/month
and
if
average
inventory,
I
=
100
units,
Turns
then
holding
cost
per
month
is
$1
X
100
=
$100
Turns
=
1/Flow
Time
=
R/I
Holding
cost/unit:
h
*
cost
of
a
unit
*
T
Make
sure
if
you
want
weekly
turns,
flow
time
is
in
weeks
E.g.
Cost
of
holding
inventory
of
electronic
Capacity
and
Flow
Rate
components
is
40%
a
year
so
if
a
component
costs
Capacity
=
(#
units
produced
by
one
machine
or
$120
it
costs
04*120
=
$48
to
hold
it
one
year
Utilization
and
Implied
Utilization
worker/time
taken
to
produce
units)
*
(#
of
machines
or
workers
if
applicable)
Note:
These
can
be
calculated
at
the
level
of
a
resource
of
the
o Should
only
add
capacity
when
the
process
is
process
supply
constrained
(otherwise
no
effect
on
flow
Utilization
=
flow
rate/capacity
rate)
Implied
utilization
=
demand/capacity
Bottleneck
=
Resource
that
constrains
the
entire
Resource
with
highest
IU
is
bottleneck
process
(lowest
capacity)
=
capacity
of
process
o Should
only
add
capacity
to
bottleneck
Cycle
Time
and
Producing
Nth
Unit
Flow
rate
=
min
{Demand,
Capacity
(i.e.
Cycle
time
=
1/R
=
longest
processing
time
(among
bottleneck)}
the
resources)
Time
between
production
of
units
Supply
constrained:
Process
capacity
<
When
system
starts
empty,
time
to
produce
Nth
unit
demand
=
Sum
of
the
processing
times
+
(N-1)
*
Cycle
time
Demand
constrained:
Demand
<
process
When
system
starts
full,
time
to
produce
Nth
unit
=
capacity
Minutes
of
work
N
*
Cycle
time
Blocking
and
Starving
For
demand:
Units/hr
*
PT
(min/unit)
Blocking:
Resource
prevented
from
completing
its
For
capacity:
60min/hr
*
#
of
machines
work
Use
when
processing
time
varies
for
different
Starving:
Resource
could
work
b ut
it
lacks
inputs
to
types
of
units
in
the
process
work
on
E.g.
Demand
is
14
applications/hour.
Each
Blocking
and
starving
at
bottleneck
reduces
R
application
requires
15
minutes
of
work.
Demand
on
the
resource
each
hour
is
14
*
15
=
EOQ
model
210
minutes
of
work.
Capacity
will
be
60min/hr
*
#
of
workers
Q=amount
ordered,
K
=
fixed
cost/order,
h=
annual
%
x
$
value
of
good
you
are
buying,
Batching
&
Setups
Avg
inventory
(AI)
holding
and
order
costs:
B
=
batch
size,
p
=
processing
time,
R
=
flow
rate,
ST
=
setup
time,
C
=
capacity
() = +
Capacity
=
#
units/time
to
produce
B
=
B/(ST
+
B
= /
x
p)
Average
inventory
holding
cost
per
unit
time=
h
x
Batch
size
can
b e
component
pairs
(add
setup
Q
/
2
times
and
processing
times)
Setup
cost/ordering
cost
per
unit
time
=
K*R/Q
Utilization
=
R
x
p
Q/R
=
time
between
orders
Batch
size
=
(C
x
ST)/(1
C
x
p)
Setup
and
inventory
holding
costs
per
unit
Average
I
=
Maximum
I
=
x
B
x
(1
R
x
p)
()/
Maximum
inventory
=
Production
time
x
Rate
of
If
Q
=
Q*
=EOQ
then
you
can
use
increase
=
B
x
(1
R
x
p)
Key
takeaway:
Product
variety
added
to
a
process
( )
=!
with
setup
times
increases
inventory
(i.e.
long
setup
times
are
not
compatible
w ith
large
product
variety)
Amount
that
minimizes
average
(setup
and
Larger
batch
=
more
inventory
=
more
utilization
=
holding)
cost
per
unit
time
is
the
EOQ
(Q*)
which
more
capacity
If
theres
no
inventory
buffer,
p
=
slowest
processing
time.
May
is
given
by
the
formula
have
to
add
p
of
m achine
to
the
setup
time
if
it's
not
the
first
in
the
process
(first
unit
takes
longer
to
produce)
= !
Evaluating
Quantity
Discount
Evaluating
Decisions
1.
Evaluate
Q*
that
minimizes
ordering
and
holding
1.
Maximax
costs
For
each
action
evaluate
the
maximum
payoff
that
2.
If
Q*
is
greater
than
the
quantity
discount
threshold,
could
occur.
order
Q*
Choose
the
action
that
yields
the
maximum
3.
Otherwise:
maximum
payoff.
a) Evaluate
purchasing
+
ordering
+
holding
costs
2.
Maximin
with
Q*
For
each
action
evaluate
the
minimum
payoff
that
b) Evaluate
purchasing
+
ordering
+
holding
costs
could
occur
with
minimum
threshold
for
quantity
discount
Choose
the
action
that
yields
the
maximum
c) Choose
whichever
quantity
minimizes
total
minimum
payoff
costs
3.
Expected
value
Evaluate
the
expected
value
at
each
node
(payoff
x
Queuing
probability)
Stable
when
IU
<
1
Choose
the
action
with
the
highest
expected
68%
Utilization
means
at
any
given
moment:
value
o there
is
a
68%
chance
a
server
is
busy
The
expected
value
of
a
decision
tree
is
the
serving
a
customer
(34%
chance
the
server
expected
value
of
the
first
node
is
idle)
o on
average
68%
of
the
servers
are
b usy
(and
EVPI
and
EVSI
34%
are
idle)
EVPI:
Expected
value
of
decision
with
PI
Original
a=interarrival
time
(time/cust),
m=#
of
servers,
decision
tree
cva
=
sa/a,
cvp
=
sp/p
Draw
tree
starting
with
outcome
and
then
with
Utilization
(U)
=
p/(m*a)
action
To
find
number
of
machines
to
meet
a
utilization
EVSI:
Expected
value
of
decision
with
SI
Original
target
set
p/(m*a)
<
target
(as
a
decimal)
decision
tree
Average
time
in
queue
(Tq):
Draw
tree
starting
with
decision
to
do
test/not
!(!)! + o Test
signal
decision
market
outcome
=
o Dont
test
=
original
decision
tree
Higher
U
=
Higher
Tq
(exponential
increase)
If
EVSI
=
0
subsequent
actions
dont
depend
on
it
Average
time
in
system/Flow
time:
T
=
Tq
+
p
EVSI
=
EVPI
if
the
test
is
perfectly
diagnostic
Average
inventory
in
queue:
Iq
=
Tq/a
Probability
Types
and
Equations
Ideal
Test
Average
inventory
in
process:
Ip
=
p/a
Prior
probability
(a.k.a
marginal
1.
Quick
to
probability):
e.g.
Pr(A)
implement
Inventory
in
system:
I
=
Iq
+
Ip
Joint
probability:
e.g.
Pr(A
and
B)
2.
Cheap
Capacity:
Ip
=
m
*
(1/p)
Posterior
probability
(a.k.a
conditional
3.
Diagnostic
Demand:
D
=
m
*
(1/a)
probability):
e.g.
Pr(A|B)
(high
To
get
a
new
interarrival
times
sum
demands
(CANT
Joint
=
Posterior
x
Prior
P(Pass|Good)
sum
interarrival
times)
Pr(A)
=
Pr(A|B)
x
Pr(B)
+
Pr(A|B )
x
Pr(B )
c c and
P(Fail|Bad)
P(A|B)
=
P(A
and
B)/P(B)
Pooling
Variability
Greatly
reduced
Tq
or
same
Tq
with
less
servers
Variability
will
decrease
R,
increase
I,
and
increase
T
Can
never
have
an
idle
server
and
w aiting
customer
at
Variability
can
reduce
capacity
considerably
without
the
same
time
in
a
pooled
system
sufficient
buffers
(blocked/starved)
Solutions:
1.
Add
buffers
or
2.
Get
rid
of
variability
Quality
Dont
send
rework
through
the
bottleneck
(reduce
Shortest
Processing
Time
(shortest
job
first)
capacity)
Average
=Total
w aiting
time/
#
of
jobs
Dont
feed
the
bottleneck
a
poor
quality
product
Shorter
average
waiting
time
Harder
to
d etect
quality
problems
in
a
large
batch
Same
total
p,
total
amount
of
work,
but
Tq
less
Go
for
quality
at
the
source
Gross
M argin
NO
CALCULATIONS
IN
HEAD!
CONSISTENT
Gross
margin
=
(Revenue
cost)/Revenue
*
100
UNITS!
FINISH
QUESTION
(DONT
STOP
With
slower/lower
inventory
turns
you
need
a
higher
HALFWAY).
YOU
GOT
THIS
J
gross
margin