This action might not be possible to undo. Are you sure you want to continue?

# Chap 8.

**The Transportation and Assignment Problem
**

by Dr. Peitsang Wu Department of Industrial Engineering and Management I-Shou University

**The Transportation Problem
**

m resources, n destination

si number of nit supplied by source i dj number of unit required by destination j cij transportation cost per unit shipped from source i to destination j

**The Transportation Problem
**

Objective minimize total transportation cost xij number of product shipped from source i to destination j

cmn emand d1 d2 d3 . . dn . .. . . . n .The Transportation Problem ost per unit istributed estination 1 1 2 / / / / m c11 c21 / / / / cm1 2 c12 c22 1 3 c13 c23 1 .. .. . . . c1n c2 n / / 1 1 / / upply s1 s2 / / / / sm source cm 2 cm 3 . . . . .

S.Properties of Transportation Problem Feasible Solutions Property : m i n j m n ij § s ! § d ! §§ x i !1 j !1 i !1 j !1 If the total supply { total demand. Integer solutions property: Not only si and dj must be integer values. it mean either si or dj represent a bound rather than an exact requirement. But also all the B. . have integer values.F. in this case introduce ³dummy source´ or ³dummy destination´ as the slack variable.

11 1. . The problem can be formulated as the general L.10 1.Example: Production Scheduling Month Scheduled Installations Unit cost Max Unit cost of Production of storage production 1 2 3 4 10 15 25 20 25 35 30 10 1.015 0.015 0.08 1.015 0.13 0.P. model.015 Let xj be the number of engines to be produced in month j.

4) xij: number of engines produced in month i for installation in month j cij: cost associated with each unit of xij .Formulate as Transportation Problem si: production of jet engines in month i (i =1«.4) dj: installation of jet engines in month j (j =1«.

.Formulate as Transportation Problem Cost per unit distributed Destination 1 2 3 4 1 2 source 3 4 Demand ? ? ? Supply ? ? ? Supply is not fixed quantities.

Formulate as Transportation Problem In fact x11 + x12 + x13 + x14 e 25 x21 + x22 + x23 + x24 e 35 x31 + x32 + x33 + x34 e 30 x41 + x42 + x43 + x44 e 10 And notice that § Supply § demand { 0 i.e. total supply {total demand) + + + = = supply + + + = = demand total supply .total demand = .

115 M 25 1.110 1.110 1.080 1.095 1.Solution Use Big-M method and introduce a new dummy destination d5 Cost per unit distributed Destination 1 1 source 2 3 4 Demand M M M 10 2 3 4 5 0 0 0 0 30 25 35 30 10 1.125 1.130 20 Supply .140 M M 15 1.125 1.100 1.

Example: Distribution of Water Resources Cost per Acre Foot Los Berdoo Devils Colombo River Sacron River Calorie River Min needed requested 16 14 19 30 50 13 13 20 70 70 San Go 22 19 23 0 30 Holly glass 17 15 10 g Supply 50 60 50 .

Example: Distribution of Water Resources Observations: upper bound for Holly glass: i. Demand: bounded variables not constant. now pexcess demand. the demand is fixed. .e. Holly glass can get as much as units. regard less the minimum need.

1 Colombo Sacron Calorie dummy demand 16 14 19 0 50 2 13 13 20 0 70 3 22 19 23 0 30 4 17 15 M 0 60 Supply 50 60 50 50 .

After considering the minimum needs 1 2 (min) (extra) 1 2 3 4 demand 16 14 19 M 30 16 14 19 0 20 3 4 5 17 15 M M 10 6 17 15 M 0 50 Supply 50 60 50 50 13 13 20 M 70 22 19 23 0 30 .

Transportation Simplex Method estination 1 1 2 source / / 2 c12 c22 3 4 5 n c1n c21 Supply ui s1 s2 / / c11 c21 1 1 1 1 1 / / m emand vj cm1 d1 cm2 d2 d3 d4 d5 cmn dn sm Z .

cij cij-ui-vj .Transportation Simplex Method For each cell If xij is a basic variable. cij xij If xij is not a basic variable.

.S.Procedure Step 1: Initialization Step 2: Optimality Test Step 3: Iteration 3 Determine the entering basic variable Determine the leaving basic variable Determine the new B.F.

e. problem. we¶ll have exact one basic variable for each constraint. . any one of the constraints are satisfied.P. the number of basic variables is m+n-1. this set of m+n equations with one extra (or redundant) equation that can be deleted without changing the feasible region i. For transportation problem with m sources and n destinations. The reason is that the functional constraints are equality constraints.Initialization In general L.

Make that allocation large enough to exactly use up the remaining supply in its row or the remaining demand in its column (which ever is smaller) . 2.S.Procedure for Constructing the Initial B. 1. select the next basic variable (allocation) according to some criterion.From the rows and columns still under consideration. .F.

Eliminate that row or column from further consideration. . i.If only one row or column remains under consideration.e. (If the row and the column have the sane remaining supply and demand.S. Otherwise return to step 1. 3.Procedure for Constructing the Initial B. a circled allocation with zero) 4. then the procedure is completed by selecting every remaining variable associated with that row or column to be basic with the only feasible allocation. select arbitrary row as the one to be eliminated.F. The column will be use later to provide a degenerate basic variable.

. Russell¶s approximation method. Vogel¶s approximation method.Alternative Criteria for Initialization Northwest corner rule.

move one column to the right) if source i has any supply remaining. j (that is. Thereafter. Otherwise. then next select xi . if xij was the last variable selected. j+1 (that is. move one row down) . . select next xi+1 .Northwest Corner Rule Begin by selecting x11 (the northwest corner).

Example D e stin a tio n 1 1 2 so u rc e 3 4 (D ) D em and vj 2 1 1 1 0 1 1 2 M 20 3 2 1 2 0 70 4 1 1 M 0 30 5 S u p p ly u i 1 1 1 M 30 50 60 50 50 60 = .

select the variable having the smallest remaining unit cost. breaks arbitrarily) .Vogel¶s Approximation Method For each row and column remaining under consideration. In that row or column has the largest difference. which is defined as the arithmetic difference between the smallest and the next-to-the-smallest unit cost cij still remaining in that row or column. calculate its difference. (If tie.

Example estination 1 1 16 Source 2 14 3 19 4 emand di erence 2 16 14 19 0 20 70 3 13 13 20 4 22 19 23 0 30 0 5 17 15 Supply di erence 50 60 50 50 x 60 Select eliminate column .

Example Destination 1 1 16 Source 2 14 3 19 4 M Demand difference 30 2 16 14 19 0 20 3 13 13 20 M 70 4 5 17 15 M 0 Supply difference 50 60 50 20 60 Select x = eliminate row .

Example estination 1 1 Source 2 3 emand di erence 16 14 19 30 2 16 14 19 20 3 13 13 20 70 5 17 15 40 Supply di erence 50 60 50 Select x eliminate ro .

Example 1 Source 2 3 Demand difference 14 19 30 2 14 19 20 3 13 20 70 5 15 M 40 Supply difference 60 50 Select x = eliminate column .

use or ourth iteration in degeneracy) .Example 1 Source 2 3 emand di erence 14 19 30 2 14 19 20 3 13 20 20 Supply di erence 20 50 Select x eliminate ro (not column 3.

Example 1 Source 3 19 2 19 3 20 Supply 50 Select x31 x32 x33 z emand 30 20 0 .

determine its v j which is the largest unit cost cij still in that column. . For each destination column j remaining under consideration. which is the largest unit cost cij still remaining in that row. For each xij not previously selected in these rows and columns. Select the variable having the largest negative value of (ij (Tie breaks arbitrarily). determine its.Russell¶s Approximation Method For each source row i remaining under consideration. calculate ( ij ! cij ui v j .

Example Iteration 1 2 3 4 5 6 u1 u 2 u3 u4 v1 v 2 v3 v4 v5 ( ij llocation x = x = x = x = x = x = x = x = Z= .

Destination 1 1 16 Source 2 14 3 19 4 Demand M 30 2 16 14 19 0 20 3 13 13 20 M 70 4 22 19 23 0 30 5 17 15 M 0 60 Supply 50 60 50 50 ui vj .

Vogel¶s approximation : popular. But it pays no attention to cij. Russell¶s approximation : excellent criterion. the solution will be far from optimal. easy to implement by hand. quick implement is computer (but not hand) better solution than the other two methods.Comparison of Three Methods Northwest corner : quick and easy. .

j ) such that xij is non-basic. .Optimality Test A basic feasible solution is optimal if and only if cij ui v j u 0 for every ( i .

..Note Since cij ± ui ± vj = 0 if xij is a basic variable. and to assign it to be zero. cij = ui + vj There are m+n-1 basic variables. A convenient choice is to select ui that has the largest number of allocations in its row (tie broken arbitrarily). m+n-1 equations. but m+n unknowns.

Example

estination 1 1 2

Source

2 16 14 19 13 13 20

3

4 22 19 23 17 15

5

Supply 50 60 50 50 Z=

ui

16 14 19

40 30

10

30 0

3 4 emand vj

20

0 0

30

0 +4 70 30

-22

+3 30 20

50

60

An Iteration

STEP 1:

± Select the one with the largest negative value of cij - ui - vj to be the entering basic variable.

STEP 2:

± Increase the entering basic variable from 0 sets off a chain reaction of compensating changes in other basic variables. ± In order to satisfy the supply and demand constraints. The first basic variable to be decreased to 0 then becomes the leaving basic variable.

An Iteration

STEP 3:

± The new B.F.S. is identified by adding value of the leaving basic variable to the allocation for each recipient cell and subtracting this same amount from the allocation for each donor cell.

Example (STEP 2) 3 1 2 emand 4 17 +4 5 10 1 3 1 3 + 22 40 30 - Supply 50 60 19 +1 15 -2 + 70 30 60 Step 3: (Z=10(15-17+13-13)=10(-2)=10(c25-u2-v5) .

± and then solving the set of equations cij = ui + vj for each ( i . ± If cij ui vj u 0 for every ( i . setting it ui = 0 . Optimality Test: ± Derives ui and vj by selecting the row having the largest number of allocations. ± Otherwise. . by any of the three methods.S. j ) such that xij is non-basic. then the current solution is optimal. go to optimality test.F.Summary Initialization: ± Construct an initial B. go to an iteration.p STOP. j ) such that xij is basic.

select the basic variable having the smallest value.vj STEP 2: ± Determine the leaving basic variable : Identify the chain reaction require to retain feasibility when the entering basic variable is increased.Iteration STEP 1: ± Determine the entering basic variable : Select the non basic variable xij having the largest negative value of cij . .ui . From the donor cells.

F.S. Return to Optimality test.Iteration STEP 3: ± Determine the new B. . : Add the value of the leaving basic variable to the allocation for each recipient cell. Subtract this value from the allocation for each donor cell.

Transportation Simplex Tableaux Iteration 0 1 2 Source 1 16 14 2 16 14 estination 3 4 13 22 40 13 20 5 17 15 Supply ui 50 60 50 10 30 19 30 19 23 0 3 4( ) emand vj 0 19 0 20 30 0 50 60 50 Z= 30 20 70 30 .

Transportation Simplex Tableaux Iteration 1 1 2 Source estination 1 16 14 16 14 2 13 3 22 4 17 15 5 Supply 50 60 50 ui 50 13 19 30 19 20 19 20 23 10 30 0 0 30 3 4( ) emand vj 0 20 0 50 60 50 Z= 30 20 70 .

Transportation Simplex Tableaux Iteration 2 1 2 Source Destination 1 16 14 19 16 14 19 2 13 13 3 4 22 19 17 15 5 Supply ui 50 50 20 40 23 M 60 50 50 = 3 4(D) Demand vj 20 30 M 30 0 20 M 20 70 0 0 0 30 30 20 60 .

331 Table 8.23 ) Iteration 3 1 2 Source 3 4(D) Demand vj 30 19 19 16 14 1 2 16 14 +2 20 13 Destination 3 22 4 17 15 5 Supply 50 ui 50 13 19 20 23 40 60 50 30 0 20 0 0 0 30 20 70 30 20 60 50 Z= .Complete set of transportation simplex tableaux (P330.

Increasing a single supply by and a single demand by .Sensitivity Analysis Changing the objective function coefficient of a non basic variable. Changing the objective function coefficient of a basic variable.

vj µs remain unchanged. Thus the current basis will still be feasible.Changing Coefficient of a Nonbasic Variable The change of coefficient of a non basic variable xij will leave the r. In row 0.h. only the coefficient of xij changed. T 1 c B B is not changed.s. of the optimal tableau unchanged. ui . thus as long as the coefficient of xij in the optimal row 0 is non negative the current basis remains optimal. .

Assure the Optimal Tableau Destination 1 1 Source 2 3 Demand vj 8 +2 9 12 6 2 10 3 9 4 +7 7 +2 5 30 2 Supply 35 50 40 ui 0 3 3 10 13 +3 9 16 25 5 45 14 +5 45 6 10 20 6 +3 30 10 Z = 1.020 .

. for what values of will the current basis remains optimal? d c11 u1 v1 ! if c¶11u 6 the current basis remains optimal.Ex: Change c11 from 8 to 8+ Change c11 from 8 to 8+ .

The coefficient of each non basic variable in row 0 may change.Changing Coefficient of a Basic Variable Since we are changing c T. we must find the new uiµs. as long as all non basic variables are non negative. c T B 1will B change too. To determine whether the current basis remains optimal. . and vj µs and use these values to price out all non basic variables. The current basis remains optimal.

020 .Previous Optimal Table Destination 1 1 Source 2 3 Demand vj 8 +2 9 12 6 2 10 3 9 4 +7 7 +2 5 +3 30 10 Supply 35 50 40 ui 0 3 3 10 13 +3 9 16 25 5 45 14 +5 45 6 10 20 6 30 30 2 Z = 1.

v3 ! . v1 ! . v2 ! u2 ! . 10 ( u1 v3 ! 0 d let u1 ! 0 u2 v1 ! .Ex:Change c13 from 10 to 10+ d d d c13 u1 v3 ! 0 i. u1 v3 ! u3 v2 ! . v4 ! . u3 ! u1 v2 ! u3 v4 ! .e. u2 v3 ! .

Ex:Change c13 from 10 to 10+ Price out each non basic variable d c11 : d c14 : d c22 : d c24 : d c31 : d c33 : ¾ ± ± ± ± ± ¿ ± ± ± ± ± À .

Increasing both si and dj by This change maintains a balance transportation problem.g. increase source 1 and destination 2 by 1 unit. Since both si and dj increase . this will cancel out the supply demand balance. New ±value = old -value + ui + vj e. New cost = 1020 + 1*0 + 1*(6) = 1026 . Now.

increase xij by (ii) If xij is a non basic variable in the optimal solution. . find the loop involving xij and some of the basic variables. Increase the value of this odd cell by and go around the loop. Find an odd cell in the loop that is in row i.Increasing both si and dj by We may also find new values of the decision variables as follows: (i) If xij is a basic variable in the optimal solution. alternatively increasing and then decreasing current basic variables in the loop by .

d2 by 2 D estination 1 8 1 9 Source 2 14 3 D em and vj 45 6 12 6 2 10 3 9 4 Supply u i 35 7 0 10 13 25 5 45 9 16 50 5 3 10 20 6 30 10 30 30 2 40 3 .Ex: Increase s1.

d2 by 2 x12 is a basic variable in the optimal solution. Now .Ex: Increase s1.

Destination 1 2 3 4 Supply ui 1 Source 2 3 Demand vj 45 6 10 46 10 20 6 26 4 30 30 10 35 50 40 0 3 3 30 Z=1020+u1+v1 = 2 . d1 by 1 x11 is a non basic variable in the optimal solution.Ex: Increase s1.

The Assignment Problem Assumptions: The number of assignees and the number of tasks are the same. There is a cost cij associated with assignee i performing task j ( i . . j = 1«.n ) The objective is to determine how all n assignments should be made in order to minimize the total cost.. Each task is to be performed by exactly one assignee. (denote by n) Each assignee is to be assigned to exactly on task.

Example ocation n=4 1 2 3 4 achine n = 3 2 15 .13 20 3 5 7 10 6 1 13 16 12 11 ocation n=4 1 2 3 4 1 13 16 12 11 achine 2 15 12 20 3 5 7 10 6 n=4 4 0 0 0 0 (D) .

Z n ! x n § § j ! 1 i ! 1 ij c ij x ij § j ! 1 ! 1 ! 1 i.e. .t . xij ! ¯ 0 ° if not. i. j § i ! 1 x ij x ij u 0 . ( actually xij n_a ) 1 ® if assignee i performs task j.The Mathematical Model n n Min s .

The Mathematical Model Assignment problem is a special case of transportation problem: Number of sources m = number of destination n Every supply si = 1 Every demand dj = 1 .

So it automatically satisfies binary restriction. the constraints prevent any variable from being greater than 1. More over.The Two Properties Feasible solution property Integer solution property holds here. . and the non-negativity constraints prevent values less than 0.

/ / / / 1 Demand 1 . . c 22 . 1 1 1 1 Source / / / / n / / c n 2 . . .Formulate as a Transportation Problem Destination 1 2. c nn 1 . .. n Supply c1n c2 n / / 1 1 1 2 c11 c 21 / / / / c n1 1 c12 . . . . . .. . . ..

Our example 1 1 2 3 4(D) Demand 13 15 5 0 1 2 16 M 7 0 1 3 12 13 10 0 1 4 11 20 6 0 1 Supply 1 1 1 1 .

etc). m=n). waste tine. we¶ll have 2n-1 basic variables. other algorithms do better than transportation simplex method in terms of computer implementation ( tine. with the special structure of assignment problem (si = dj = 1.e. memory.Disadvantage of using Transportation Simplex Method Using transportation simplex method. Transportation simplex method solves the general transportation-type problem. . these are n-1 degenerate basic variables (which equal to 0). but exactly n of these xij = 1. i.

. ± For this new matrix. (called the reduced cost matrix) by subtracting from each cost the minimum cost in its column.Hungarian Method: STEP 1: ± Begin by finding minimum element in each row of the n x n cost matrix. find the minimum cost in each column. ± Construct a new matrix. ± Construct a new matrix by subtracting from each cost the minimum cost in its row.

Hungarian Method: STEP 2: ± Draw the minimum number of lines (horizontal and/or vertical) that are needed to cover all the zeros in the reduced cost matrix. an optimal solution is available among the cover zeros in the matrix. ± If n lines are required to cover all the zeros. . GOTO STEP 3. ± If fewer than n lines are needed to cover all the zeros.

.Hungarian Method: STEP 3: ± Find the smallest nonzero element (call its value k) in the reduced cost matrix that is not covered by lines draw n in STEP2. ± Return to STEP 2. ± Now subtract k from each uncovered element of the reduced cost matrix and add k to each element of the reduced cost matrix that is covered by two lines.

Example row min 14 2 7 2 5 12 8 4 8 6 3 6 7 5 9 10 Column min .

Example 9 0 3 0 1 4 6 p Smallest Uncovered:1 0 10 4 4 0 5 2 0 4 .

Transshipment Problem A supply node is a node that can send goods to another node but cannot receive goods from any other node. . A demand node is a node that can receive goods from other nodes but cannot send goods to any other node. A transshipment node is a node that can both receive goods from other nodes and send goods to other nodes.

add a dummy demand to balance the problem.Solution Procedure Step 1: If necessary. Step 2: Construct a transportation tableau as follow: ± A row consists of supply node and transshipment node. s =total available supply. ± The transshipment node will have a supply equal to (node¶s original supply)+s. and a demand equal to (node¶s original demand)+s. ± A column consists of demand node and transshipment node. .

Example .