Skip to content

Transportation Model

Special case of LP, which deals with shipping of commodities from \(m\) sources to \(n\) nodes

The number of basic vars and independent constraints will be \((m+n-1)\)

It will always be a minimization problem, since transporation model deals with the shipping cost of commodities

To solve, we've to 1. Find initial BFS 2. Find optimal solution

Matrix

\(m \times n\)

For every cell \(v_{ij}\), make sure that

\[ \begin{aligned} v_{ij} &= \text{argmin}(D_i, S_j) \\ \sum v_i &= D_i \\ \sum v_j &= S_j \end{aligned} \]

Balanced Transportation

\[ \sum \text{Supply} = \sum \text{Demand} \]

Find initial BFS

(Basic Feasible Soln)

Any basic feasible solution will have \((m+n-1)\)

Basic vars \(v > 0\)
Non-basic vars \(v=0\)
  • If the problem is balanced, no issues
  • Else add an extra row/column to compensate
  • If no penalties, cell costs = 0
  • If \(\exists\) Penalties, cell costs = penalties

Soln can be find using one of the following methods

Method Steps
North-West Corner Traverse from top-left to bottom-right
Least Cost
Vogelโ€™s Approximation
(VAM)
1. Calculate row-wise penalties \(P_i = \min_2(v_i) - \min(v_i)\)
2. Calculate row-wise penalties \(P_j = \min_2(v_j) - \min(v_j)\)
3. Pick the row/column with max penalty. Cross out the penalty (and hence entire row/column), if the row/column is completely utilized
4. In that row/column, we select the cell with least cost
5. Repeat, excluding the recently-allocated cell

Find optimal solution

Method of multipliers. 2 conditions need to be satisfied

  • Supply limits & demand requirements remain satisfied
  • Shipments through all routes must be \(\ge 0\)

Checking optimality

  1. Set \(u_1 = 0\)

  2. Find \(u_i, u_j\) for cells with \(u_i + v_j = c_{ij}\)

  3. Find BIJ for empty cells: \(\text{BIJ} = u_i + v_j-c_{ij}\)
  4. If BIJ \(\le 0\) for empty cells, solution is optimal (this is minimization problem)
  5. Else, non-optimal

Optimizing

  1. Find the entering var as the empty cell with the most +ve BIJ
  2. Put \(\theta\) there
  3. Make a square/rectangle, with corners as non-empty cells or \(\theta\) cell
  4. Add
  5. \(-\theta\) to corner cells in the same row/col as \(\theta\) cell
  6. \(+\theta\) to other corner cells
  7. \(\theta\) = Max value that \(\theta\) can assume, which is obtained using the cells in the same row/column of \(\theta\) cell
  8. Evaluate all the cells

If 0 appears in non-basic var, and there is no other potential entering var, it implies that optimality is already reached, and future iterations give Alternative Optima.

Degenerate BFS

If in a cell we find zero mentioned, it means that all corresponds to basic var that has assumed \(v=0\). It implies degenerate basic feasible solution.

Maximization Problem

For eg, if we want to maximize the distribution of foods (not worried about money, for eg during a natural disaster).

The values of \(c_{ij}\) will be the +ve output. Since transportation problem is always minimization.

  1. \(c_{ij} = -1 \times c_{ij}\)
  2. Solve as usual
  3. Total output = -1 x total cost
Last Updated: 2024-01-24 ; Contributors: AhmedThahir

Comments