Steps to Build a Model Transportation Problem:
• Business Problem: Figure out ___ in order to • Business Problem: Minimizing total transportation cost while meeting the demand
maximize/ minimize ____ while considering • Decision variables: xᵢⱼ = # of units to ship from factory at i to store at j
resource constraints Shadow price of supplier : -ve or zero
• Objective function: Min c₁₁x₁₁ + c₁₂x₁₂ + ... + cₙₘxₙₘ
• Analytical Model: Shadow price of receiver : +ve or zero
• Constraints: (if D<S)
• Decision Variables (& No. of variables) 1) Supply constraints: Qty shipped out should not be more than available (i.e. ≤)
• Objective Function x₁₁ + x₁₂ + ... + x₁ₘ ≤ supply₁, …….., xₘ₁ + xₘ₂ + ... + xₘₘ ≤ supplyₘ
• Constraints 2) Demand constraints: Qty Shipped in should be greater than or equal to Qty Reqd (≥)
*Include Units x₁₁ + x₂₁ + ... + xₙ₁ = demand₁, …….., x₁ₘ + x₂ₘ + ... + xₙₘ = demandₘ
Graphical Model 3) Non-negativity : xᵢⱼ ≥ 0
• Plot the constraint lines & figure out feasible region * If D>S, then Supply constraints: Do not Ship < what is available to meet demand (≥ S1)
• Determine the coordinates of vertices by solving Demand constraint: Give me whatever is available (≤ D1)
simultaneous eqn in 2 variables NOTE: Optimal Solution will not always choose individual least expensive routes, it will
• Plug the vertices in the obj fn & compare the values choose routes such that total cost is minimised. SEE THE OVERALL COST!
• Figure our which vertex max / min the obj fn
Telecom Optics
Assumptions underlying a linear optimization model
• Single objective
• Linearity
• Divisibility: Variable can take fractional values
• Data certainty: Coeff. are not random variables
1. Feasible Region
• Feasible region is defined purely by the constraints;
Obj fund has no role
• Feasible region can be: Bounded, Unbounded,
Empty (for infeasible LP)
• LP can have redundant constraints which can be
omitted w/o affecting feasible solution
2. Optimal Solution
• Always on the vertices or edges i.e. multiple
solutions (but solver gives only corner points)
• Determined by Objective function
3. Binding Constraints
• Resources completely used @optimality
• Line passing through the optimal soln
• corresponding resources are critical resources
The time to complete task B, tB, is a constant (4 days),
• Binding const. are defined w.r.t. optimal sol. tA can take as long as 10 days, but can be shortened
• Check for non-neg cont. also, if x1 = 0 à Binding by assigning additional resources to the task. For
every unit of additional resource assigned to task A,
4. Shadow Price the completion time reduces by half a day. Let xA
• Marginal value of the resource or constraint denote the number of additional resources assigned
• Change in obj. fn w.r.t a small change in availability to task A (clearly xA >=0). Due to budget constraints
we can assign at most 18 additional units to task A
of resource (everything else const.) Which of the following is the correct linear program
• Unit : slope or derivative (∆Obj fn/ ∆constraint RHS) that minimizes the project completion time subject to
• Sh. P. is same b/w a range of resource availability these constraints?
• Sh. P. of Non-Binding constraint = 0
• Diminishing Marginal Val. (Sh P↓ w/ ↑ in RHS)
5. Objective Function Coefficient
• Within a range of obj fn coeff. à same optimal sol
6. Reduced Cost (of a variable)
• = Marginal Benefit – Marginal Cost = MR - MC
• =Obj fn coeff– sumprod(Resource consump., Sh P.)
• Unit : Rate of Change
• Red. Cost = 0; when product ∈ optimal sol
• *** Used when engaging in a new activity
• Indicates how much the Obj fn coeff of the variable
must be improved before the variable is “utilized”
on its own merit (it is neither at max nor min).
• Regardless of whether you are maximizing or
minimizing, a positive reduced cost means
increasing x will increase the objective value and a
negative reduced cost means increasing x will
decrease the objective value. (MR –MC > 0)
Solver Errors:
• Solver could not find a feasible solution: Feasible
region = empty/ LP infeasible (Qty reqd ≥ available)
• Obj. cell values do not converge: Some constraints
are missed à region is very large/ unbounded
Types of Integer Problems
• All integer programs have linear equalities and
inequalities and some or all of the variables are
required to be integer
• If all variables are required to be integer, then it is
usually called a pure integer program
• If all variables are required to be 0 or 1, it is called a
binary integer program, or a 0-1 integer program
• If some variables can be fractional and others are
required to be integers, it is called a mixed linear
integer program (MILP)
Objective Function, Constraints for:
1) Custom Molder
2) Cosmos
3) Transportation problem
4) Frosty
Custom Molder Transportation Problem
• Business Problem: Figure out the quantities of 6-oz & 10-oz
glasses to produce in order to maximize profits while
considering the resource & demand constraints
• Decision variables: x₁ = # of 6oz glasses to produce; x2 = # 10
oz glasses to produce
• Objective function: Max 500x₁ + 450x₂ ($ per 100 cases)
• Constraints:
1) Production (Hrs): 6x₁ + 5x₂ ≤ 60
2) Storage (’00s cu. ft.): 10x₁ + 20x₂ ≤ 150
3) Demand for 6-oz Glasses (‘00s of cases): x₁ ≤ 8
4) Non-negativity : x₁, x₂ ≥ 0
# hrs of Molding-machine cap was inc from 60 to 61 hrs
• How is the feasible region impacted by this change? Increases
• New optimal objective value? Same (Inc is within limits)
• Per unit improvement in profit? Inc by Sh. P*Inc in qty of resp.
constraint= 78.5*1
Profit cont of 6oz glass is inc from $500 to $501 (per 100 cases)
• How is the feasible region impacted by this change? No change
• New optimal solution? No change
• New optimal objective value? Inc by (optimal qty of 6-oz
glass*Inc in Obj fn coeff = 6.43*1
New product : Prod time = 8 Hrs /100 cases; Storage Cap = 1000
cu ft /100 cases; Profit cont is $6.00 per case
• Should you produce? Red Cost = 600 – (8*Sh P of Prod cons +
10*Sh P of Stor Cap cons) Shortest Path Problem Assignment Problem
Astros & Cosmos
Objective function: Max (20*A + 30*C – 1000) Assignment Problem: # of supply locations = # of demand locations.
Constraints: Supply @each supply node = 1 & demand @each demand node = 1.
• A + 2*C ≤ 120 (Picture tube Hrs) Eqn = ∑(Unit Feasible Solution = 1-to-1 mapping
• A + C ≤ 96 (Chassis assembly Hrs) Contribution*xi – • Business Problem: Minimizing cost or time
• A ≤ 70 (Astros circuit boards) Variable Cost* xi ) – • Decision variables: xᵢⱼ = {0,1} ; 0 if task i is not assigned to resource
• C ≤ 50 (Cosmos circuit boards) Fixed Costs j; 1 if task is assigned to resource j and one job can only be worked
• A, C ≥ 0 (Non negativity) on by one person, with its respective costs Cij
We use the –ve sign for demand as a convention
Suppose that workers in the picture tube department are willing
to work overtime for $21/hr. How many hours of overtime, if
any, should they employ? – Sh. P is 15rs i.e maximum price
which should be paid to obtain an additional unit of re- source is
15 but we are demanded 21 so we should not use overtime.
• Business Problem: How many cartons should frosty purchase ?
• Decision variables: xᵢⱼ = # of units to ship from factory at i to store at j
• Objective function:
Southwest Airlines