Sensitivity Analysis: Lindo Input & Results
Sensitivity Analysis: Lindo Input & Results
Analysis
Sensitivity
Analysis
is
used
to
see
how
the
optimal
solution
is
affected
by
the
objective
function
coefficients
and
to
see
how
the
optimal
value
is
affected
by
the
right-‐hand
side
values.
Using
LINDO,
you
can
figure
out
how
to
integrate
different
variables
to
the
problem
without
affecting
the
objective
function.
The
Wilson
Problem’s
objective
function
was
to
maximize
7X1+10X2
and
was
subject
to
X1
<
500,
X2
<
500,
X1
+
2X2
<
960,
and
5X1
+
6X2
<
3600.
The
optimal
value
for
the
Wilson
problem
was
$5,520.
Subject to
X1 < 500
X2 < 500
END
(Optimizer Status)
(Status: Optimal)
(Iterations: 2)
(Objective: 5520)
NO. ITERATIONS= 2
1) 5520.000
NO. ITERATIONS= 2
Now
to
construct
a
dual
problem,
you
would
need
to
formulate
a
minimization
linear
model.
For
the
Wilson
problem
we
will
use
the
decision
variables
U1,
U2,
U3,
and
U4,
since
there
are
4
constraints
in
the
original
problem.
The
dual
model
for
the
Wilson
problem
would
be:
Minimize 3600U1+960U2+500U3+500U4
Subject to
5U1+1U2+U3>=7
6U1+2U2+U4>=10
Uj>=0
END
As
you
can
see
we
have
switched
the
objective
function
to
the
constraints
and
the
constraints
to
the
object
function.
This
would
be
the
dual
problem
for
the
Wilson
Problem.
After
we
have
figured
out
our
program
model,
we
can
use
LINDO
in
order
to
solve
the
problem.
LINDO INPUT
Minimize 3600U1+960U2+500U3+500U4
Subject to
5U1+1U2+U3>=7
6U1+2U2+U4>=10
Uj>=0
END
LINDO RESULTS
1)
5520.000
NO. ITERATIONS= 3
As
you
can
see
the
objective
coefficients
from
the
original
problem
have
become
the
right
hand
side
ranges
in
the
dual
problem.
The
results
from
the
minimization
problem
have
shown
that
the
optimal
value
is
the
same,
$5,520.
The
solutions
being
U1=1,
U2=2,
U3=0,
U4=0,
were
found
at
Step
3.
The
value
column
comes
from
the
Right
Hand
Side
Ranges.
The
slack
or
surplus
columns
result
from
remaining
or
excess
resources.
The
cost
coefficient
can
increase
or
decrease
without
affecting
the
current
optimal
solution
of
5520.
That
is
what
either
output
is
called
combined
report:
Final
report
to
the
primal
and
the
dual
problems.
Definitions
of
LINDO
Terms
Related
to
Original
and
Dual
Wilson
Problem
LP
Optimum:
Tells
which
step
the
optimal
solution
was
found
at.
The
original
Wilson
problem
was
found
at
step
2
and
the
dual
problem
was
found
at
step
3.
Objective
Function
Value:
The
value
to
which
the
decision
variables
were
used
to
optimize
the
objective
function.
The
objective
function
value
for
the
Wilson
problem
and
dual
problem
was
$5,520,
it’s
the
same
as
we
expect.
Variable:
In
the
original
Wilson
problem
the
decision
variables
were
X1
and
X2,
but
in
the
dual
problem
they
were
U1,
U2,
U3
and
U4.
These
variables
represent
the
objectives
being
produced.
Value:
The
values
of
the
decision
variables.
In
the
original
problem,
X1=360
and
X2=300,
and
in
the
dual
problem
U1=1,
U2=2,
U3=0,
and
U4=0.
When
plugged
in
to
the
objective
function
you
will
find
the
optimal
value
of
the
objective
function.
Reduced
Cost:
These
values
indicate
the
amount
the
coefficient
value
in
the
objective
function
must
increase
before
it
has
a
positive
impact
on
the
optimal
solution.
In
both
problems,
the
values
is
0,
because
solution
to
both
problem
are
positive
numbers.
Slack or Surplus: Represents the lack of or excess of resources from production.
Shadow
Price:
The
amount
of
change
to
the
right
hand
side
of
a
constraint
and
is
also
the
solution
to
the
dual
problem.
For
example,
the
dual
prices
in
the
original
Wilson
problem
were
0,
0,
2,
1
and
are
the
solution
for
the
dual
problem
being
U1=1,
U2=2,
U3=0,
and
U4=0.
The
dual
prices
in
the
dual
problem
were
360
and
300,
which
are
the
solutions
for
the
optimal
value
for
X1
and
X2.
1) For
the
first
what
if
analysis,
I
decided
to
delete
a
constraint.
The
constraint
I
deleted
was
the
amount
of
cowhide
allowed
for
production
of
baseballs
and
softballs.
This
constraint
is
binding
which
means
it
will
change
the
optimal
solution.
The
deletion
of
this
constraint
yielded
a
surplus
of
270
for
softballs.
It
takes
less
time
to
produce
baseballs,
so
with
no
limit
to
the
amount
of
cowhide
for
production,
Wilson
would
utilize
production
of
baseballs.
The
optimal
solution
of
the
deleted
constraint
would
maximize
profit
at
$5800,
with
production
of
500
baseballs,
and
230
softballs.
The
LINDO
results
were:
Subject to
X1 < 500
X2 < 500
END
1) 5800.000
VARIABLE
VALUE
REDUCED
COST
NO. ITERATIONS= 1
2)
For
the
second
what
if
analysis,
I
decided
to
add
a
variable.
I
decided
to
add
the
production
of
a
cricket
ball.
Cricket
balls
are
similar
to
baseballs
with
a
core
and
the
use
of
hide
for
the
outer
cover.
Since
there
will
be
no
change
to
the
amount
of
cowhide
or
time
available
for
production,
there
will
probably
be
no
change
to
the
objective
function.
The
profit
per
cricket
ball
will
be
$5
and
the
maximum
amount
of
cricket
balls
produced
daily
will
be
300.
It
will
take
3
sq.
ft.
of
hide
per
dozen
cricket
balls
and
take
90
seconds
to
produce
each
dozen.
After
solving
the
new
variable
in
LINDO,
the
optimal
solution
did
not
change
and
showed
that
the
production
of
the
cricket
ball
is
not
necessary.
That
is,
it
is
not
profitable
to
produce
the
new
product.
One
can
see
that
by
considering
what
goes
into
the
new
product
1.5($1)
+
3($2)
=
$7.5
which
more
that
net
profit
of
5.
Therefore
do
not
product.
The
following
report
confirms
it.
The
objective
coefficient
ranges
remained
the
same
except
for
the
allowable
decrease
of
softballs.
LINDO
results:
Subject to
X1 < 500
X2 < 500
X3 < 300
END
1)
5520.000
NO. ITERATIONS= 2
3) For
the
next
what
if
analysis,
I
decided
to
add
to
the
amount
of
time
allowed
for
production.
I
increases
the
amount
of
time
allowed
for
production
from
960
minutes
to
1500
minutes
per
day.
Since
Wilson
manufacturing
makes
more
profit
from
softballs,
you
would
think
that
the
amount
of
softballs
produced
would
increase
with
them
having
more
time
to
produce
them.
With
all
other
constraints
kept
the
same,
I
plugged
the
new
problem
into
LINDO.
The
amount
of
softballs
produced
increased
from
300
dozen
to
500
dozen
per
day,
and
the
amount
of
baseballs
produced
decreased
from
360
dozen
to
120
dozen
per
day.
The
surplus
of
baseballs
was
380
dozen
for
both
quantity
and
time.
With
the
amount
of
production
time
increased,
Wilson’s
profits
were
increased
to
$5,840.
LINDO
results:
Subject to
X1 < 500
X2 < 500
END
1) 5840.000
NO. ITERATIONS= 2
RIGHTHAND
SIDE
RANGES
4) The
last
what
if
analysis
I
decided
to
analyze
was
the
deletion
of
a
variable.
I
decided
to
delete
the
production
of
baseballs.
The
profit
should
decrease
because
the
production
of
both
baseballs
and
softballs
with
the
current
constraints
are
necessary
to
maximize
profit.
With
the
deletion
of
baseballs,
Wilson’s
profit
from
softball
production
was
$5,000
with
the
production
of
500
dozen
softballs
per
day.
The
surplus
of
time
was
460
minutes
and
the
surplus
of
cowhide
was
600
square
feet.
If
you
were
to
delete
softballs
instead
of
baseballs,
the
profit
from
baseballs
would
be
$3,500,
with
a
surplus
of
460
minutes
and
1100
square
feet
of
cowhide.
So
it
would
be
more
profitable
to
produce
only
softballs
than
only
baseballs,
with
fewer
surpluses
of
materials.
But
it
would
still
be
better
to
produce
both
baseballs
and
softballs
with
the
original
constraints
provided.
LINDO
Results
for
both
scenario’s:
a. Maximize
10X
Subject to
X< 500
X < 960
6X < 3600
END
1) 5000.000
X 500.000000 0.000000
NO. ITERATIONS= 1
b. Maximize 7X
Subject to
X< 500
X < 960
5X < 3600
END
1) 3500.000
X 500.000000 0.000000
NO. ITERATIONS= 0