SOLUTIONS TO PROBLEMS 369

14.18 Optimizing a constraint

The ‘simplest’ version of this constraint (with minimum right-hand side coeffi-
cient) is

6x1 + 9x2 − 10x3 + 12x4 + 9x5 − 13x6 + 16x7 + 14x8 ≤ 25.

This is also the equivalent constraint with the minimum sum of absolute
values of the coefficients.

14.19 Distribution 1
The minimum cost distribution pattern is shown in Figure 14.6 (with quantities
in thousands of tons).

Newcastle

C1
50
Liverpool
10 C2
Birmingham

50
35 C3

5
Brighton
C4
55 London 40

20 55
40 C5

C6
Exeter

Figure 14.6

There is an alternative optimal solution in which the 40 000 ton from Brighton
to Exeter comes from Liverpool instead.
This distribution pattern costs £198 500 per month.
Depot capacity is exhausted at Birmingham and Exeter. The value (in reduc-
ing distribution costs) of an extra ton per month capacity in these depots is £0.20
and £0.30, respectively.
370 MODEL BUILDING IN MATHEMATICAL PROGRAMMING
This distribution pattern will remain the same as long as the unit distribution
costs remain within certain ranges. These are given below (for routes which are
to be used):

Route Cost range

Liverpool to C1 −∞ to 1.5
Liverpool to C6 −∞ to 1.2
Brighton to Birmingham −∞ to 0.5
Brighton to London 0.3 to 0.8
Brighton to Exeter −∞ to 0.2
Birmingham to C2 −∞ to 1.2
Birmingham to C4 −∞ to 1.2
Birmingham to C5 0.3 to 0.7
London to C5 0.3 to 0.8
Exeter to C3 0 to 0.5

Depot capacities can be altered within certain limits. For the not fully utilized
depots of Newcastle and London changing capacity within these limits has no
effect on the optimal distribution pattern. For Birmingham and Exeter the effect
on total cost will be £0.2 and £0.3 per ton per month within the limits. Outside
certain limits the prediction of the effect requires resolving the problem. The
limits are:

Depot Capacity range

Birmingham 45 000–105 000 tons
Exeter 40 000–95 000 tons

N.B. All the above effects of changes are only valid if one thing is changed
at a time within the permitted ranges. Clearly, the above solution does not satisfy
the customer preferences for suppliers.
By minimizing the second objective, it is possible to reduce the number of
goods sent by non-preferred suppliers to a customer to a minimum. This was
done and revealed that it is impossible to satisfy all preferences. The best that
could be done resulted in the distribution pattern shown in Figure 14.7, where
customer C5 receives 10 000 tons from his non-preferred depot of London. This
is the minimum cost such distribution pattern. (There are alternative patterns
which also minimize the number of non-preferences but which cost more.) The
minimum cost here is £246 000, showing that the extra cost of satisfying more
customers preferences is £47 500.
 SOLUTIONS TO PROBLEMS 371

Newcastle
10 C1
10
50
Liverpool
C2

Birmingham 35
50 C3

50
Brighton C4
London
30
10
40 C5
40 20

Exeter
C6

Figure 14.7

14.20 Depot location (distribution 2)

The minimum cost solution is to close down the Newcastle depot and open a
depot in Northampton. The Birmingham depot should be expanded. The total
monthly cost (taking account of the saving from closing down Newcastle) result-
ing from these changes and the new distribution pattern is £174 000. Figure 14.8
shows the new distribution pattern (with quantities in thousands of tons).
This solution was obtained in 40 iterations. The continuous optimal solution
was integer. Therefore, no tree search was necessary.

14.21 Agricultural pricing

The optimal prices are:

Milk £303 per ton

Butter £667 per ton
Cheese 1 £900 per ton
Cheese 2 £1085 per ton

The resultant yearly revenue will be £1992 million. It is straightforward to

calculate the yearly demands that will result from these prices. They are:
372 MODEL BUILDING IN MATHEMATICAL PROGRAMMING

Milk 4 781 000 tons

Butter 384 000 tons
Cheese 1 250 000 tons
Cheese 2 57 000 tons

50 C1

Liverpool
10 C2
Birmingham

20
40 C3
70 10
50

Brighton C4
10
London 10

40 C5

C6
25 Exeter 25

Northampton

Figure 14.8

The economic cost of imposing a constraint on the price index can be obtained
from the shadow price on the constraint. For this example, this shadow price in
the optimal solution indicates that each £1 by which the new prices are allowed
to increase the cost of past year’s consumption would result in an increased
revenue of £0.61.

14.22 Efficiency analysis

The efficient garages turn out to be 3 (Basingstoke), 6 (Newbury), 7
(Portsmouth), 8 (Alresford), 9 (Salisbury), 11 (Alton), 15 (Weymouth), 16