Linear Model for Optimization

What is a linear relationship?

Relationship between two or more variable of
the type:
TotalCost=Quantity *Price+ FixedCost
Revenue=Demand*SellingPrice
Profit=Revenue-TotalCost
Distance =Speed * Time
Overtime=1.5*HourlySalary+ 100
In all cases variables are to the power 1, then
only it is linear relation ship


What is a linear relationship?
E.g. In TotalCost=Quantity *Price+ FixedCost if I
want to see impact of Quantity over Total Cost,
and if Price=Rs 10, and FixedCost=100
it would always be 10 times the Quantity add
100 to it irrespective of what value quantity
assume
Linear relationship is formed through three basic
operations
Proportionality, additivity, divisibility

Components of Linear Models
Objective in problem solving, is the criteria by
which all decisions are evaluated. A single
quantifiable objective must be specified by
decision maker(DM).
For example objective might relate to Profit, Cost
or Market-share, but only to one of these.
As we are dealing with optimization objective
would always be either to MAXIMIZE or
MINIMIZE
Maximization problems often involve Profit,
Revenue, Market-share or ROI
Minimization problems involve Cost, Time or
Distance
Once objective is specified it becomes measure of
effectiveness against which alternative solutions
are judged. For example optimal solution will be
the one that yield highest profit
DM can control the value of objective through
choices in the levels of decision variables.

Decision variables represent choices available
to DM, usually w.r.t. amount/quantity of either
an input to a process or an output from a
process. For example, a problem may involve
maximizing profit obtained from sale of three
products, each with its own profit (profit/Unit).
Assuming all what is produced can be sold, you
need to decide how much quantity of each
product to manufacture, that give you greatest
profit Q1,Q2,Q3 become your decision
variable
Ability of DM to select values of decision variables is
subject to some limitations, we call them CONSTRAINTS,
can come from variety of sources e.g. limited availability
of raw material, labor time, machine time, work space,
storage space, capacity to produce.
Also technological, legal, or other standards e.g if you
are producing food item standard may be every Kg
should contain certain amount of minerals, salts,
calories, etc. So that when you are mixing the
ingredients, need to satisfy either ( >= min standard) or
(<= max standards)
There may be other limits imposed based on, forecast,
customer order(max demand cant produce more than
this)
Only solutions that satisfy all specified CONSTRAINTS
are acceptable these are called feasible solutions

The optimum solution is that feasible solution
that yields best value of the objective(e.g
maximum profit or minimum cost).
Many times there may be more than one
optimal solution

Formulating the Model
Identify the inputs
Identify the Objective
Represent Objective in the form of a set of mathematical
function- called Objective function
The units of all coefficients multiplied by the variables
must be same (e.g all $, all Rs, all hours) e.g.
Cost=4 * Qty shipped from plant 1 + 6 * Qty shipped from
plant 2 + 8 * Qty shipped from plant 3
Unit of coefficients 4,6,8 (transportation costs from plant
1,2,3 respectively to warehouse in this case) should be
the same either all in $, or in Rs. or in
Constraints
Four elements:
RHS quantity that specify the limit for that
constraint it must be a constant, not a variable
An algebraic sign that indicate whether limit is
upper bound <=, a lower bound >=, or an equality =
that must be met exactly.
Decision variable to which constraint applies.
The impact that one unit of each decision variable
will have on RHS quantity of the
Case Study
A toy manufacturer makes three versions of a Benten
Gadget. Version 1 requires 10 minutes each for
fabrication and packaging and 2 lbs. of plastic, Version
2 requires 12 minutes for fabrication and packaging
and 3 lbs. of plastic. Version 3 requires 15 minutes for
fabrication and packaging 4 lbs. of plastic. There are 8
hours of fabrication and packaging time available, 200
lbs. of plastic available for next production cycle. Unit
profits are Rs 100, from V1, Rs 500 from V2 and Rs 600
from V3. A minimum of 10 units must be made to fill
previous order. Formulate the model that will
determine optimal production quantities for profit
maximization.
Model Formulation Process
Identify objective Maximize Profit
Identify decision Variables
Number of V1 to produce (lets call it X1)
Number of V2 to produce (lets call it X2)
Number of V2 to produce (lets call it X3)
Identify Constraints by name
Fabrication & packaging time
Quantity of plastic
Quantity of V1 (minimum 10)
Quantity of V2 (minimum 10)
Quantity of V3 (minimum 10)
Write Objective function
Maximize 100*X1+500*X2+600*X3 (represent total profit)
(Note 100,500,600 are profit from each unit of V1,V2 &V3
all represented in Rs.)


Model Formulation Process
Write Constraints in algebraic form
Fabrication & Packaging time
10 * X1 +12 * X2 + 15 * X3 <= 480 mins


Note 8 hrs converted to 480
minutes To make all units in mins
Fabrication & packaging time for
X3 Numbers of Version V13
Fabrication & packaging time for
X3 Numbers of Version V13
Fabrication & packaging time for
X3 Numbers of Version V13
Model Formulation Process
Write next Constraints in algebraic form
Quantity of Plastic
2 * X1 + 3 * X2 + 4 * X3 <= 200


Total Plastic available.
All units in lbs.
Plastic used for
X3 Numbers of Version V13
Plastic used for
X3 Numbers of Version V13
Plastic used for
X3 Numbers of Version V13
Model Formulation Process
Write next Constraints in algebraic form
Minimum Quantity (10) of each version to be
produced
X1 >= 10
X2 >= 10
X3 >= 10

Objective, Decision variables & Constraints
Summarized
Objective function
Maximize 100*X1+500*X2+600*X3 (represent total
profit)
Decision Variables

Constraints
10 * X1 +12 * X2 + 15 * X3 <= 480 mins
2 * X1 + 3 * X2 + 4 * X3 <= 200 lbs.
X1 >= 10
X2 >= 10
X3 >= 10







X1 X2 X3
Let us see what is known at this point?
Three version of toy
Plastic quantity
Fabricn+Packg time
Profit from each Ver
Decision Variables
Objective Function
Constraints(time)
Constraints(Plastic)
Constraint min No
10X
1
+

12X
2
+ 15X
3
<= 480
V1 V2 V3
X
1
X
2
X
3

100 500 600
100X
1
+ 500X
2
+ 600X
3

10 12 15
2 3 4
2X
1
+

3X
2
+ 4X
3
<= 200
10 10 10
>= >= >=
X
1
X
2
X
3


SUMPRODUCT
of Profit &DV

SUMPRODUCT
of Time &DV
SUMPRODUCT
of FabPacg
&DV

Quantity of
Each Version
should be at
least 10
Model on Excel Sheet
Excel Model
We should
substitute
X1,X2,X3 with
some values say
15,15,15
c

This will
automatically
change to
15,15,15 see
formula sheet
Excel Model with Formula
Excel Model with Formula
Excel Solver Dialogue Box
Excel Solver Solution
Go through this presentation at
least two times, then do it yourself.
Did it help you to understand
Linear Optimization models?
Let me know the feed back
through the mail.