of 37
8/14/2019 Are521 Transportation11.Ppt
1/37
The Transportation Problem
For ARE 521 QUANTITATIVETECHNIQUES
Offered by Prof. G. DSouzaPrepared by Tanya Borisova
8/14/2019 Are521 Transportation11.Ppt
2/37
Transportation Problem This problem involves the shipment of a homogeneous
product from a number of supply locations to a number ofdemand locations.
Problem: given needs at the demand locations, how shouldwe take limited supply at supply locations and move thegoods. Further suppose we wish to minimize cost.
Based on McCarl 2005
8/14/2019 Are521 Transportation11.Ppt
3/37
Basic Concept Objective:Minimize cost
Variables:Quantity of goods shipped from eachsupply point to each demand point
Restrictions:
Non negative shipmentsSupply availability at a supply pointDemand need at a demand pointBased on McCarl 2005
8/14/2019 Are521 Transportation11.Ppt
4/37
Formulating the Problem:Basic notation and the decision variable We denote the supply locations as i
We denote the demand locations asj Let us define our fundamental decision
variable as the set of individual shipmentquantities from each supply location to
each demand location and denote thisvariable as Movesupplyi,demandj
Based on McCarl 2005
8/14/2019 Are521 Transportation11.Ppt
5/37
Formulating the Problem:The objective functionWe want to minimize total shipping cost so we need an expression for
shipping costLet us define a data item per unit cost of shipments from each supplylocation to each demand location as costsupplyi,demandjOur objective then becomes to minimize the sum of the shipment costsover all supplyi, demandj pairsMinimize
supplyi demandj
costsupplyi,demandj Movesupplyi,demandj
hich is the per unit cost of moving from each supply location to eachdemand location times the amount shipped summed over all possibleshipment routesBased on McCarl 2005
8/14/2019 Are521 Transportation11.Ppt
6/37
Formulating the Problem: Constraintsthe sum of outgoing shipments from the supplyi supply point to all possible
destinations (demandj) to not exceed supplysupplyi (resource availabilityconstraints)
Movesupplyi,demandj supplysupplyidemandj
shipments into the demandj demand point should be greater than or equal
to demand at that point. Incoming shipments include shipments from allpossible supply points supplyi to the demandj demand point (minimumrequirement constraints).supplyi
Movesupplyi,demandj demanddemandj
nonnegative shipments Movesupplyi,demandj 0Based on McCarl 2005
8/14/2019 Are521 Transportation11.Ppt
7/37
Based on McCarl 2005
8/14/2019 Are521 Transportation11.Ppt
8/37
Standard FormMin c ij xiji =1j =1mn
= 1, .., ni = 1, .., m
xj =1mnij
ai , bj,
xi =1ij
xij 0
here
i supply locations,j demand locations,cij unit shipping cost, xij shipment volumes,bj demand levels,ai supply limits
8/14/2019 Are521 Transportation11.Ppt
9/37
Exampletrucking company has contracted to supply weekly threesupermarkets located in Modesto, Sonoma, and Fresno,California, with potatoes kept at two warehouses in
Sacramento and Oakland, also in California. The weeklyavailability (supply) of potatoes at the warehouses is 400and 600 tons, respectively. The supermarkets weeklyneeds (demands) are 200, 500, and 300 tones of potatoes,respectively. (the transportation unit costs are given below.)
The objective is to find supply routs that minimizes the totaltransportation cost of the trucking firm. The contractstipulates that all supermarkets must receive at least thequantity of potatoes specified above.
Based on Paris 1991
8/14/2019 Are521 Transportation11.Ppt
10/37
Example (cont.) Two warehouses: Sacramento and Oakland Three supermarkets: Modesto, Sonoma, and
Fresno Quantity availability: Supply availableSacramento 400 tones
Oakland600 tones Demand requiredModesto200Sonoma500Fresno300Based on Paris 1991
8/14/2019 Are521 Transportation11.Ppt
11/37
Example (cont.):Transportation Unit Cost (dollars per ton)
arehouse
ModestoSacramentoOaklandBased on Paris 1991
Supermarkets
Sonoma404050
30Fresno8090
8/14/2019 Are521 Transportation11.Ppt
12/37
Network Representationc11 (cost11)1) Sacramento
400c12 (cost12)
c13 (cost13)x12 (move12)
2. Sonoma500
x22 (move22)x11 (move11) 1. Modesto
200
x21 (move21)c21 (cost21)2) Oakland
600c22 (cost22)c23 (cost23)x13 (move13)
x23 (move23)3. Fresno
300Based on Paris 1991
8/14/2019 Are521 Transportation11.Ppt
13/37
Example:Problem Formulation Objective function:
minimize total cost of shipmentMinTC =
Min cost11move11 + cost12move12+ cost13move13 + cost21move21 +cost22move22 + cost23move23 Demand constraint move11 + move21 demand1 move12 + move22 demand2 move13 + move23 demand3
cost moveiji
ij
moveiij
demand
8/14/2019 Are521 Transportation11.Ppt
14/37
demand j ,for allj
E l
8/14/2019 Are521 Transportation11.Ppt
15/37
Example:Problem Formulation (cont.) Supply constraints move11 + move12 + move13 supply1 move21 + move22 + move23 supply2 moveij
supplyi ,for all i
moveij supplyi ,for all i
E l T bl
8/14/2019 Are521 Transportation11.Ppt
16/37
Example:Tableaumove11(x11)
move12(x12)
move13(x13)
move21(x21)
move22(x22)
move23(x23)
4015080
4013090
min
8/14/2019 Are521 Transportation11.Ppt
17/37
min200500300
-400-60011
-1-1-1-111
-1
1
8/14/2019 Are521 Transportation11.Ppt
18/37
-1
E l Optimal Shipping Pattern
8/14/2019 Are521 Transportation11.Ppt
19/37
Example: Optimal Shipping PatternDestinationModestoSonomaFresnoUnits
ariableUnits
ariableUnits
ariable
Sacramento100move11(x11)
300move13
(x13)Oakland100move21(x21)
500move22(x22)
E l Optimal Solution
8/14/2019 Are521 Transportation11.Ppt
20/37
Example: Optimal Solution Objective value $47,000VariableMove11(x11)
Value1000300100
5000
Reducedcost
020
00
EquationSupply1Supply2Demand1
Demand2
Slack
8/14/2019 Are521 Transportation11.Ppt
21/37
Slack0000
0Shadowprice
Eps040
3080Move12(x12)
Move13(x13)
Move21(x21)Move22(x22)
010Demand3
Move23(x23)
Example: Solution
8/14/2019 Are521 Transportation11.Ppt
22/37
Example: Solution shadow price represents marginal values
of the resources, e.g.
marginal value of additional units at supplylocations are (close to) zero
marginal value of unit decrease in demand atdemand locations = unit transportation cost to
that location reduced cost represents marginal costs offorcing nonbasic variable into the solution
Based on McCarl 2005
Example: Reduced Cost
8/14/2019 Are521 Transportation11.Ppt
23/37
Example: Reduced Cost Assume the trucking company is required to ship 1tone of potatoes from Sacramento to Sonoma
Shipment from Sacramento to Sonoma cost $50 The trucking company also decreases supply from Oakland to Sonoma (saving of $30) To stay within supply limits, the company decreases
supply from Sacramento to Modesto (saving of $40) To meet demand constraints, the company increases
supply from Oakland to Modesto (cost $40) Resulting change in transportation cost = reduced cost = $50 - $30 + $40 - $40 = $20
8/14/2019 Are521 Transportation11.Ppt
24/37
Transportation Model:
8/14/2019 Are521 Transportation11.Ppt
25/37
Transportation Model:Important Extensions Transshipment
Transshipment through intermediatewarehouses is permitted
Spatial equilibrium Quantities supplied and demanded depend on
prices
GAMS IDE
8/14/2019 Are521 Transportation11.Ppt
26/37
GAMS-IDE GAMSIDE GAMS Generalized Algebraic Modeling System IDE Integrated Development Environment, A Windows graphical interface to run GAMS
Using GAMSIDE Open the IDE through the icon Create a project Create or open an existing file of GAMS instructions
GAMSIDE Files
8/14/2019 Are521 Transportation11.Ppt
27/37
GAMSIDE FilesprojectBased on McCarl and Gillig
What is a
8/14/2019 Are521 Transportation11.Ppt
28/37
What is aProject? The project location
determines where allsaved files are placed(to place files elsewhereuse the save as dialogue) The project saves file names and program options
associated with the effort. It is recommended to define a new project every time youwish to change the file storage directory.
GAMSIDE
8/14/2019 Are521 Transportation11.Ppt
29/37
GAMSIDEProject
Define a project name
and location. Put it in adirectory you want to use. All files associatedith this project will be saved in that directory.
In the File name area type in a name for the project fileyou wish to use. Save the file.
File name extension gprstands for GAMS project.
Slide is taken from McCarl and Gillig
8/14/2019 Are521 Transportation11.Ppt
30/37
Opening or Creating
8/14/2019 Are521 Transportation11.Ppt
31/37
a File with GAMSInstructions
You can
Create a new file Open an existing file Open a model library file
File name extension .gms GAMSIDE recognizes codes (and adds color
differentiation) only in the files saved as .gms files.
GAMS Model Library
8/14/2019 Are521 Transportation11.Ppt
32/37
GAMS Model Library collected models to be
used as of examples Standard textbook examples
to illustrate problem formulation / GAMS features models that have been used in policy or sector analysis and
are interesting for both the methods and the data they use
Output and Log files
8/14/2019 Are521 Transportation11.Ppt
33/37
Output and Log files Log file Located in the same directory as the project file File extension - .log
Contain whatever GAMS-IDE shows during the run (content of Process window) Clicking on any of the black-colored line will activate the .lst output file Clicking Reading Solution for model open the .lst file and position
the window at the SOLVE SUMMARY
Clicking on the red-colored line will cause the cursor to jump in
the .gms file to the line with error Output (listing) file Located in the same directory as the project file File name extension - .lst Stores results of the model run
Running GAMS programs
8/14/2019 Are521 Transportation11.Ppt
34/37
g p g click on the run button, or use File menu, or
press the F9 key Note Option Compile performs
initial check of the program
Option Run check the program, and (if there isno error) run optimization and generate output
GAMS Help
8/14/2019 Are521 Transportation11.Ppt
35/37
GAMS Help GAMS Users Guide McCarl Guide
Solver Manual (discussion of various GAMSSolvers) GAMS documents additional materials onGAMS, IDE, Solvers
Useful GAMS Websites
8/14/2019 Are521 Transportation11.Ppt
36/37
Download GAMS student versionhttp://www.gams.com/download/
GAMS development corporationhttp://www.gams.com/ Bruce McCarls web-sitehttp://agecon2.tamu.edu/people/faculty/mccarl-
bruce/
References
8/14/2019 Are521 Transportation11.Ppt
37/37
McCarl, B. Course Materials from GAMS 2 Class.http://www.gams.com/mccarl/useide.pdfMcCarl B. 2005. Basic LP problem formulationshttp://agecon2.tamu.edu/people/faculty/mccarl-bruce/622class/overhead05basiclpformulate.pdf
McCarls and Spreens book athttp://agecon2.tamu.edu/people/faculty/mccarl-bruce/books.htmMcCarl B & D Gillig. Introduction to GAMS IDE.http://agecon2.tamu.edu/people/faculty/mccarl-bruce/641clas/02_641_intro_gamsIDE_.pdfParis 1991. An Economic Interpretation of Linear Programming. Iowa
State University Press / Ames