You are on page 1of 7

150

European Journal of Operational Research 39 (1989) 150-156


North-Holland

Theory and Methodology

The effect of ignoring routes


when locating depots
Said S A L H I a n d G r a h a m K. R A N D
Department of Operational Research, Gillow House, University of Lancaster, Lancaster, LA1 4YX, England

Abstract: In physical distribution the location of depots and vehicle routes are interdependent problems,
but they are usually treated independently. Here, we evaluate the effect of ignoring routeing when locating
depots by using a two stage process (location and routeing). Using data from standard problems, it is
shown that the best solution after the location stage does not necessarily generate the lowest cost solution
after the routeing stage. This feature is found both when the best locations obtained from a variety of
methods are compared, and when a single method is evaluated for different numbers of depots. A sensible
way to determine the best combination of methods used in location and routeing is proposed.
Keywords: Depot location, vehicle routeing, distribution

1. Introduction
In the literature of Operational Research applied to distribution studies there are relatively
few papers concerned with the combination of
routeing and location-allocation (see Or and
Pierskalla [9], Jacobsen and Madsen [4], Nambiar,
Gelders and Van Wassenhove [8], Laporte and
Nobert [5], Madsen [6], and Perl and Daskin [10]).
However, the interdependence between the location-allocation problem and the vehicle routeing
problem has been recognised by both academics
and practitioners as stated by Rand [11]: " M a n y
practitioners are aware of the danger of suboptiraising by separating depot location from vehicle
routeing".
The combined problem has a large variety of
applications. These include the location of a district sales office when the salesman has to visit a
number of customers (shops), the location of a

truck depot which provides pick up and delivery


service for a number of companies, etc.
In location-allocation solution methods, it is
assumed that a tour consists of a visit to a single
customer, but in practice this is seldom the case.
The occurrence of a tour to more than one
customer makes this assumption unrealistic,
though it seems to have been frequently made.
As a simple example of the effect routeing may
have on depot location, consider three customers
which can be served by a single vehicle in a single
tour. These customers form a triangle whose
vertices are A, B and C (see Figure 1).
According to methods based on the minimum
sum of radial distances,the optimal location is the
Steiner point (see Miehle [7]). Let this point be O.
Since all three customers are served on the same
route then the length of such a route is
L = rnin{ OA + A B + B C + OC,
OA + A C + CB + OB,

Received March 1986; revised April 1988


0377-2217/89/$3.50 1989, ElsevierSciencePubfishers B.V. (North-Holland)

OC + A C + A B + OB }.

S. Salhi, G.K. Rand / The effect of ignoring routes when locating depots

This approach is tested on seven standard problems, taken from the literature (see Eilon,
W a t s o n - G a n d y and Christofides [2]). It should be
noted that in problems 1, 2, 3, and 4 a drop time,
(a time per customer, to represent the time to
make a delivery), equivalent to 10 distance units is
used. In the remaining problems a drop time is
not included.

151

F i g u r e 1. Possible l o c a t i o n of a d e p o t in case of a o n e r o u t e 3
customers

On the other hand if the depot location is


chosen at a customer site, the length of the route
would be
L ' = A B + BC + CB
(this is the perimeter of the triangle). The triangular inequality leads to L ' < L.
If the set of customer locations to be visited are
all on a single tour, the optimum location of the
depot is any point on the shortest Hamiltonian
circuit passing through these points.

2. T h e analysis of the problem

The purpose of the analysis is to show whether


the location yielding the lowest cost at the location stage maintains its superiority at the routeing stage. To evaluate this consistency, three approaches have been used.
(i) For one depot, the consistency between
the solutions obtained from different location algorithms, assuming radial distances.
(ii) As (i) but with no restriction on the number of depots.
(iii) Consistency using a particular algorithm.
The location-allocation methods used in the
analysis are:
- R D A (the random destination algorithm, see
Cooper [1]);
- ALA (the alternate location allocation, see
Cooper [1]);
- D R O P (the drop method, see Feldman,
Lehrer and Ray [3]).
For each of a set of problems, RDA, ALA and
D R O P locations are determined, and for each
location routes are generated through a routeing
procedure developed by Salhi and Rand [12].

Case 1: A single depot


When there is only one depot there is no need
to determine accurate values for the unit cost (the
variable cost per mile) and the fixed cost of the
depot. A value for the depot cost of 1000 units
and a mileage cost of one unit per mile were used.
The results are reported in Table 1. For each
problem four depot locations were used. The first
was the one given in the problem statement. The
second, third and fourth were generated by RDA,
A L A and D R O P respectively.
(X, Y) are the coordinates of the depot. DEC~
and DEC 2 are the ranking of the locations by cost
after the location stage and after the routeing
stage respectively. DEV~ and DEV 2 are the %
deviations from the best solution, in terms of total
distance, after the location stage and the routeing
stage respectively. R A D denotes the total radial
distance (one way only, because this is usually
what is minimised in location-allocation problems). R O D is the total routeing distance, including the total drop time, AVER is the average
routeing distance per route (vehicle), and nv is the
number of vehicles (routes).
The total mileage is R O D less the total drop
time, i.e.
total mileage = R O D - ( N C D T ) ,
where N C is the number of customers and D T is
the drop time.
Some comments on the results are required.
For instance, for problem 6 there is not much to
choose between the RDA, ALA and D R O P methods since these give similar costs (distance), which
are lower than the distance derived from the location given in the problem. However, when routeing is introduced, the routes generated by the
location stated in the problem yield the lowest
cost and the best of the previous locations gives a
solution which is 7.7% worse. Similar results are
also found in problems 1, 4, 5, and 7.

S. Salhi, G.K. Rand / The effect of ignoring routes when locating depots

152

Table 1
Inconsistency for one depot problem
No.

145.0
147.7
147.7
148.0

215.0
224.7
224.7
232.0

266.0
271.9
272.4
255.0

RAD

DEV 1
(%)

DEC 1

ROD(nv)

DEV 2
(%)

DEC~

AVER

582.7
562.5
562.5
565.9

3.4
0.0
0.0
0.6

4
1
1
3

585.3(4)
600.8(4)
600.8(4)
613.8(4)

0.0
2.5
2.5
4.8

1
2
2
4

146.2
150.2
150.2
153.5

235.0
250.5
250.3
257.0

995.4
931.1
931.2
942.9

6.4
0.0
0.0
1.2

4
1
2
3

948.7(5)
934.2(5)
934.5(5)
959.8(5)

1.4
0.0
0.0
1.5

3
1
2
4

190.0
187.0
187.0
192.0

162.0
174.9
166.0
175.0

354.0
363.0
360.7
363.0

1268.4
1253.9
1254.8
1253.9

1.2
0.0
0.1
0.0

4
1
3
1

882.9(4)
839.2(4)
854.7(4)
839.2(4)

4.8
0.0
1.5
0.0

4
1
3
1

220.0
210.0
213.7
210.0

292.0
304.7
304.4
304.0

425.0
421.7
421.2
427.0

943.3
851.4
851.8
860.9

9.4
0.0
0.0
1.0

4
1
2
3

810.1(4)
808.6(4)
797.5(4)
803.9(4)

1.6
1.4
0.0
0.7

4
3
1
2

202.5
202.0
199.0
201.0

30.0
35.9
35.4
32.0

40.0
38.9
38.6
39.0

1201.2
1174.1
1174.2
1183.4

2.2
0.0
0.0
0.7

4
1
2
3

554.5(5)
564.6(5)
569.7(5)
530.7(5)

4.5
6.4
7.4
0.0

2
3
4
1

211.0
213.0
215.0
206.0

40.0
39.9
39.6
40.0

40.0
36.7
36.0
37.0

1815.4
1798.1
1798.5
1798.2

0.9
0.0
0.0
0.0

4
1
3
2

884.1(10)
952.6(10)
1032.7(10)
955.2(10)

0.0
7.7
16.8
7.8

1
2
4
3

168.0
175.0
183.0
175.5

35.0
32.4
31.6
37.0

35.0
34.0
33.3
31.0

2494.7
2484.2
2487.9
2525.6

0.4
0.0
0.0
1.9

3
1
2
4

861.1(8)
850.3(8)
847.2(8)
860.4(8)

1.6
0.3
0.0
1.5

4
2
1
3

232.0
231.0
230.0
232.0

In such cases decision makers should shift to


the location given in the problem though this gave
a less a t t r a c t i v e r e s u l t a t t h e l o c a t i o n s t a g e .
In problems 2 and 3 the solution after both
s t a g e s a p p e a r s t o b e c o n s i s t e n t . A l t h o u g h t h e r e is
n o c h a n g e i n t h e r a n k i n g o f t h e l o c a t i o n s it is
w o r t h n o t i n g t h a t t h e r e l a t i v e d e v i a t i o n s a r e sign i f i c a n t . F o r i n s t a n c e , i n p r o b l e m 3, t h e s o l u t i o n ,
obtained by the location given in the problem, has

methods show equal probability of giving the


l o w e s t costs. I n c o n c l u s i o n 5 p r o b l e m s o u t o f 7 a r e
affected by the inconsistency of the solutions.

a c o s t p e n a l t y o f j u s t 1.2% a f t e r t h e l o c a t i o n s t a g e
b u t t h i s p e n a l t y b e c o m e s f o u r t i m e s g r e a t e r (4.8%)
a f t e r t h e r o u t e i n g stage. R D A g e n e r a t e s s i m i l a r
r e s u l t s w h i c h a r e j u s t 0.1% w o r s e a f t e r t h e l o c a t i o n
s t a g e b u t a f t e r r o u t e i n g t h i s i n c r e a s e s t o 1.5%.
F o r t h e l o c a t i o n s t a g e t h e R D A m e t h o d is
superior to the others, but when the routeing
a l g o r i t h m is t a k e n i n t o a c c o u n t all t h e l o c a t i o n

1
2
3
4
5
6
7

Table 2
Parameters used in the problems
No. a

NC

(3) 21
(4) 22
(5) 29
(7) 32
(8) 5o
(9) 75
(10) lOO

DT

D
Vehicle Vehicle cost
capacity
per mile per day
bl
b2

10
10
10
10
0
0
0

6
4.5
4.5
8
8
7
10

0.44
0.35
0.33
0.39
0.38
0.41
0.38

0.127
0.112
0.112
0.130
0.130
0.127
0.141

a Denotes the problem number in [12].

47
44
44
51
51
49
55

150
189
213
201
211
175
231

153

S. Salhi, G.K. Rand / The effect of ignoring routes when locating depots

Table 3
Inconsistency within the methods (RDA, ALA, DROP)
No.

Method

ND

C1

ROD (nv)

DEC 2

RDA
ALA
DROP

1
2
1

347.5
347.0
349.0

RAD
562.4
334.1
565.9

DEC 1
2
1
3

C2
337.9
476.4
339.7

600.8 (4)
533.5 (5)
613.8 (4)

1
3
2

RDA
ALA
DROP

2
1
2

422.9
425.9
419.5

634.3
931.1
627.0

2
3
1

493.1
407.2
489.0

818.9 (5)
934.5 (5)
785.7 (5)

3
1
2

RDA
ALA
DROP

2
2
2

365.3
365.4
365.5

501.0
501.2
501.6

1
2
3

421.5
421.6
422.6

663.3 (4)
663.9 (4)
671.8 (4)

1
2
3

RDA
ALA
DROP

2
2
2

424.1
424.9
424.3

574.6
576.6
575.2

1
3
2

515.3
467.9
513.1

763.1 (5)
789.6 (4)
747.0 (5)

3
1
2

RDA
ALA
DROP

2
1
2

541.3
546.2
546.0

898.2
1174.2
910.5

1
3
2

574.8
432.5
575.8

506.2 (6)
569.7 (5)
513.6 (6)

2
1
3

RDA
ALA
DROP

3
3
3

759.4
761.4
762.1

1159.5
1125.4
1127.1

1
2
3

934.8
983.1
983.2

725.5 (11 )
720.2 (12)
721.4 (12)

3
l
2

RDA
ALA
DROP

4
4
3

849.5
854.1
856.1

1182.8
1194.9
1463.6

1
2
3

1058.2
997.1
901.4

751.3 (10)
708.9 (9)
738.6 (9)

3
2
1

The depot cost is 100 units. ND is the number of depots.


C l and C2 denote the total cost at the location stage and the routeing stage respectively.
C 1= ( D C x N D ) + ( a x RAD).
C2 = ( D C x N D ) + ( b 1 x [ R O D - ( N C x D T ) ] ) + ( b 2 xnv).

Case 2

I n o r d e r to give a r e a s o n a b l e c o m p a r i s o n b e t w e e n costs for r o u t e s a n d for r a d i a l d i s t a n c e s , t h e


u n i t m i l e a g e cost u s e d in the l o c a t i o n a l g o r i t h m
was e x p r e s s e d as a f u n c t i o n o f the u n i t r u n n i n g
c o s t ( b l ) a n d t h e u n i t fixed c o s t ( b 2 ) u s e d in
routeing. The following expression was used:
a = b I + bz/D,

w h e r e D is the e s t i m a t e d d a i l y d i s t a n c e p e r v e h i c l e
( d e t e r m i n e d f r o m T a b l e I as t h e a v e r a g e v a l u e o f
A V E R o v e r the f o u r l o c a t i o n s in e a c h p r o b l e m ) .
T h e u n i t c o s t a n d o t h e r p a r a m e t e r s are s u m m a r i s e d in T a b l e 2.
T h e p e r f o r m a n c e o f t h e t h r e e m e t h o d s u s e d in
l o c a t i o n has b e e n t e s t e d in the f o l l o w i n g w a y . F o r
each problem the cheapest solution among the
t h r e e at t h e l o c a t i o n s t a g e is k n o w n a n d t h e issue
is w h e t h e r this l o c a t i o n m a i n t a i n s its s u p e r i o r i t y
w h e n the l o c a t i o n s are i n p u t in the r o u t e i n g stage.

I n this p a r t i c u l a r c a s e t h e s o l u t i o n s d o n o t s e e m to
b e c o n s i s t e n t at b o t h stages (see T a b l e 3). E v e n
w h e n the n u m b e r o f d e p o t s is t h e same, the c o n s i s t e n c y is n o t m a i n t a i n e d at the r o u t e i n g stage.
F o r i n s t a n c e , in p r o b l e m 4 the c o s t at the l o c a t i o n
s t a g e a p p e a r s to b e m o r e o r less t h e s a m e b u t at
the r o u t e i n g s t a g e a b i g c h a n g e is f o u n d . T h e least
a t t r a c t i v e d e c i s i o n at t h e l o c a t i o n s t a g e gives the
l o w e s t c o s t at t h e r o u t e i n g stage.
When the solutions have different numbers of
d e p o t s , t h e o n e w i t h t h e l o w e s t n u m b e r s e e m s to
b e t h e m o s t p r o m i s i n g . T h i s m a y b e the result of
t h e v a l u e c h o s e n for t h e d e p o t cost. N o t e t h a t
p e r f e c t c o n s i s t e n c y is f o u n d for p r o b l e m 3.

Case 3
T o test t h e c o n s i s t e n c y o f t h e s o l u t i o n s g e n e r a t e d b y a p a r t i c u l a r m e t h o d the u n i t cost f o u n d in
T a b l e 2, t h e o p t i m a l n u m b e r o f d e p o t s d e t e r m i n e d
by that particular method and a fixed depot cost

154

S. Salhi, G.K. Rand / The effect of ignoring routes when locating depots

Table 4
Instability within a given method (RDA)
No.

ND

ROD (nv)

DEC 2

1
2

C1
347.5
347.9

RAD
562.4
336.2

1
2

DEC 1

337.9
476.2

600.8 (4)
531.8 (5)

1
2

1
2
3

425.9
425.0
485.5

931.1
634.3
531.2

2
1
3

407.2
493.1
592.7

934.2 (5)
818.9 (5)
816.4 (5)

1
2
3

1
2
3

513.5
365.3
432.5

1253.9
501.0
401.6

3
1
2

370.0
421.0
619.2

839.2 (4)
663.3 (4)
564.6 (4)

1
2
3

1
2
3

432.0
424.1
501.8

851.4
574.6
517.4

2
1
3

409.1
515.3
620.7

808.6 (4)
763.1 (5)
802.9 (5)

1
2
3

1
2
3

546.2
541.3
577.2

1174.2
898.2
729.6

2
1
3

428.4
574.8
671.7

564.6 (5)
506.2 (6)
483.2 (6)

1
2
3

1
2
3
4

837.2
783.4
759.4
815.7

1798.1
1422.9
1159.5
1013.8

4
2
1
3

710.9
844.5
934.8
1080.6

952.6 (10)
799.1 (11)
725.5 (11)
701.4 (12)

1
2
3
4

1
2
3
4
5

1043.9
893.3
856.2
849.5
912.3

2484.2
1824.6
1463.8
1182.8
1084.9

5
3
2
1
4

659.9
807.9
901.3
1058.2
1096.5

850.3 (8)
784.2 (9)
738.1 (9)
751.3 (10)
704.8 (9)

1
2
3
4
5

of 100 units were used. Here, solutions at the


r o u t e i n g stage were g e n e r a t e d n o t o n l y with the
b e s t solution f o u n d in the l o c a t i o n stage b u t with
m a n y others. F o r i n s t a n c e if m d e n o t e s the o p t i m a l n u m b e r of d e p o t s f o u n d b y a given l o c a t i o n
m e t h o d ( R D A , for instance) the analysis is b a s e d
o n the 1 . . . . . m a n d m + 1 d e p o t p r o b l e m s a n d
solutions g e n e r a t e d at the r o u t e i n g stage f r o m all
these possibilities. T h e solution with the least cost
is selected at the r o u t e i n g stage, regardless of the
result at the l o c a t i o n stage. I n c o n s i s t e n c y is also
discovered here, see T a b l e 4.
Comments

In general the r a n k i n g of l o c a t i o n s after the


l o c a t i o n stage does n o t seem to b e r e p e a t e d after
routeing. This c h a n g e is crucial for strategic decision planning. I n this analysis the last two cases
m a y a p p e a r to b e less c o n v i n c i n g b e c a u s e o f the
w a y the i n p u t is i m p l e m e n t e d , especially w h e n the
d e p o t cost has m o r e i m p a c t o n decisions. N e v ertheless, the results f o u n d in T a b l e 1 (one d e p o t

C2

p r o b l e m ) s h o w that p o o r decisions m a y b e m a d e if
care is n o t t a k e n at the l o c a t i o n stage.
A c c o r d i n g to the results the n u m b e r of vehicles
u s u a l l y increases with the n u m b e r of depots. This
seems s u r p r i s i n g since the total of r a d i a l distances
decreases with the increase of the n u m b e r of depots. I n c r e a s i n g the n u m b e r of d e p o t s leads to
m o r e u n d e r u t i l i s e d vehicles b e c a u s e of the lack of
flexibility. H a v i n g firm b o u n d a r i e s for each d e p o t
m a k e s the s o l u t i o n s s u b o p t i m a l a n d then n o t very
attractive. F o r instance, let us a s s u m e for one
d e p o t that there is o n e u n d e r u t i l i s e d vehicle. F o r 2
d e p o t s there m a y b e one u n d e r u t i l i s e d vehicle for
each d e p o t a n d this m e a n s two u n d e r u t i l i s e d
vehicles in the w h o l e system. Again, when the
n u m b e r o f d e p o t s is small the p o s s i b i l i t y of c o m b i n i n g r o u t e s is g r e a t e r b u t if the n u m b e r of
d e p o t s increases, t h o u g h the c a p a c i t y a n d the time
c o n s t r a i n t s m a y p e r m i t it, c o m b i n i n g routes f r o m
different d e p o t s m a y n o t b e permissible.
F u r t h e r , an i n c r e a s e in the n u m b e r of d e p o t s
does n o t necessarily decrease the total r o u t e i n g
distance, b e c a u s e o f the g e o g r a p h y o f the

S. Salhi, G.K. R a n d / T h e effect of tgnoring routes when locating depots

155

customers and the location of the depots. For


instance in problem 7, see Table 3, R D A with 4
depots generates a longer distance than D R O P
with 3 depots. The same feature is found in Table
4 in problem 4 and problem 7.

and 3 two methods are first equal using this


criterion.
It should be noted that the routeing heuristic
used may have an effect on the choice of the
combination.

3. The choice of the combination

4. Conclusion

Here, an approach for choosing a location


method which generates robust solutions at the
routing stage is proposed. A sensible way to select
the best combination of methods used in location
and routeing is to create a set of problems which
may reflect the real situation and test all possible
combinations for each number of depots. For each
number of depots a table of results can be obtained and the best combination selected. To
choose the combination, criteria such as the total
cost over all problems or the frequency of occurrence of the best two combinations (for instance)
may be useful. For instance, in the one depot
problem (see Table 1) four combinations are used
but here only three will be considered, RDA, A L A
and DROP, all using the refinement routeing process (see Salhi and Rand [12]). The location stated
in the problem is discarded. The results are summarised in Table 5 (these are derived from Table
1). R D A appears to be superior to the others for
both criteria. It yields the lowest total routeing
distance (number of vehicles is found to be the
same for all combinations) and it has the highest
frequency of occurrence for being the best and
second best combination. DEC1 and D E C 2 denote
the ranking of the solutions for the total distance
(TD) and the frequency of occurrence. In the
column 'frequency', ( p , q) denotes that the
method combined with the refinement routeing
process is the best p times out of the seven
problems and q times it is second best (see D E C 2
in Table 1, but ignore the results for the first
location in each problem as this is the location
given in the problem statement). In problems 1

To summarise the results we can see that 5


problems out of 7 are affected by the use of the
location allocation techniques. Here we have
shown, through an example and test problems that
the ranking of locations at the first stage (location
allocation) are not always the same after the
routeing stage.
Such a strategic combined problem may be
worth investigating. The problem which decision
makers may encounter is to evaluate the gain in
changing sites against the constraints in changing.
The analysis may also be of interest for companies
which open new depots or an entire new distribution system. Another point which may strike distribution planners is how to rely on a variable
input such as routes when looking at a strategic
problem such as depot location. A consistent input which includes past data and forecast values
may be worth considering.

Table 5
Overall summary of results
Method

TD

DEC 1

Freq.

DEC 2

RDA
ALA
DROP

5560.3
5637.1
5563.0

1
3
2

(4,2)
(3,1)
(2,2)

1
2
3

Acknowledgment
S. Salhi is very grateful to the Algerian Government for sponsorship.

References
[1] Cooper, L., "Heuristic methods for location-allocation
problems", S I A M Review 6 (1964) 37-53.
[2] Eilon, S., Watson-Gandy, C.D.T., and Christofides, N.,
Distribution management: Mathematical modelling and
practical analysis, Griffin, London, 1971.
[3] Feldman, E., Lehrer, F.A., and Ray, T.L., "Warehouse
location under continuous economies of scale", Management Science 12, (1966) 670-684.
[4] Jacobsen, S.K., and Madsen, O.B.G., "A comparative
study of heuristics for a two level routing-location problem", European Journal of Operational Research 5 (1980)
378-387.
[5] Laporte, G., and Nobert, N., "An exact algorithm for
minimising routing and operating costs in depot location",
European Journal of Operational Research 6 (1981)
224-226.

156

S. Salhi, G.K. Rand / The effect of ignoring routes when locating depots

[6] Madsen, O.B.G., "Methods for solving combined two


level location routing problems of realistic dimensions",
European Journal of Operational Research 12 (1983)
295-301.
[7] Miehle, W., "Link-length minimisation in networks", Operations Research 6 (1958) 232-243.
[8] Nambiar, J.M., Gelders, L.F., and Van Wassenhove, L.M.,
"A large scale location-allocation problem in the natural
rubber industry", European Journal of Operational Research 6 (1981) 183-189.
[9] Or, I., and Pierskalla, W.P., "A transportation location-

allocation model for regional blood banking", AIIE


Transactions 11 (1979) 86-95.
[10] Perl, J., and Daskin, M.S., "A warehouse location-routing
problem", Transportation Research 19B (1985) 381-396.
[11] Rand, G.K., "Methodological choices in depot location
studies", Operational Research Quarterly 27 (1976)
241-249.
[12] Salhi, S., and Rand, G.K., "Improvements to vehicle
routeing heuristics", Journal of the Operational Research
Society 38 (1987) 293-295.

You might also like