0% found this document useful (0 votes)
150 views17 pagesChapter - Single Facility Location Problem
Industrial engineering
Uploaded by
Safeer khanCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
150 views17 pagesChapter - Single Facility Location Problem
Industrial engineering
Uploaded by
Safeer khanCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 17
* 172 | Single-Facility Location Problems
(or (4,2), (8, 5), (11, 8), and (13, 2). Furthermore, suppose that the cost per
foot traveled is the same between the new machine and each existing machine,
and the number of trips per hour made between the new machine and the
existing machines are 1/6, 1/3, 1/3, and 1/6, respectively. '
From (4.6) and (4.7) it is apparent that the optimum x coordinate value
and optimum y coordinate value are independently determined. It is instruc
tive to consider the number of material-handling trips between the new facili-
ty and each existing facility over a 6-hour period. Thus, we make one trip
to the existing facility located at (4, 2), two trips to the location (8, 5), two
trips to the location (I, 8), and one trip to the location (13, 2). Considering
the x coordinates, we weight each according to the number of trips to give
the sequence 4, 8, 8, 11, 11, 13. The median location is any point in the closed
interval [8, 11]. You might want to compare several points in this interval
against the definition of a median location to satisfy yourself that this is the
optimum x coordinate location. Considering the y coordinate, we have the
sequence 2, 5, 5, 8, 8, 2. Arranging these in increasing order, we have 2, 2,
5, 5, 8, 8. Thus, the median location is any point in the closed interval [5, 5],
or the optimum y coordinate location is y = 5.
For location problems involving a large number of existing facilities
and large values of the weights, there is a more efficient way of determining
the median location. We divide the accumulated weights by 2 and determine
the location such that no more than one half the weights are to the right
of the location and no more than one half are to the left. An example will
serve to illustrate the procedure. A rent-a-car company has five offices locat-
ed in a large city. Customers desiring a cat can pick up and deliver the car
to any of the five offices. The company wishes to locate a maintenance fa-
cility in the city to service the cars. The coordinate locations of the five offices
are (0, 0), (3, 16), (18, 2), (8, 18), and (20, 2). The numbers of cars transported
per day between the new maintenance facility and the offices equal 5, 22,
41, 60, and 34, respectively. What location for the maintenance facility will
minimize the distance cars are transported per day?
The optimum x coordinate location will be determined first. The exist-
ing facilities are ordered according to their x coordinate values. Next the
weights are accumulated. For this example, the total number of cars trans-
Table 4.1. x COORDINATE SOLUTION
v
Existing x-Coordinate Gauuee
Facility Value Weight Weight
1 0 3 5
Z 2 2 27<81
e 60 87> 81
3 18 al re
: 2 4 162
xt =8 Rectilinear-Distance Location Problems | 173
ported per day equals 162. Thus, the median location corresponds to a cu-
mulative weight of 162 ~ 2, or 81. As Table 4.1 shows, the corresponding
x coordinate value for the new facility is 8, since the cumulative weight first
exceeds 81 at x equal to 8.
A similar procedure is employed to obtain the optimum y coordinate
value for the new facility. As shown below, the optimum y coordinate value
is 16. The new maintenance facility should be located at the point (8, 16) to
minimize the distance cars are transported.
An obvious question to be asked at this point is, “What if the point
(8, 16) is not available as a location site?” The point may, for example, be
Table 4.2, y COORDINATE SOLUTION
Existing y-Coordinate Cumulative
Facility Value Weight Weight
1 0 5 5
3,5 2 41 +34 80 < 81
2 16 2 102 > 81
4 18 0 162
y= 16
inaccessible or may coincide with the location of another structure, a river,
or a municipal park. Since such an occurrence is likely for a number of fa-
cility location problems, the question deserves additional consideration.
One approach that may be employed is to construct contour lines of the cost
function. A contour line is a line of constant cost in the plane (since we are
considering planar locations). Thus, locating the new facility at any point
ona given contour line results in the same total cost.
Contour lines provide considerable insight into the shape of the surface
of the total cost function, as well as a useful means of evaluating alternative
locations for the new facility. Contour lines (also called iso-cost lines or
level curves) indicate at a glance the cost penalty associated with the choice
of a nonoptimum location. This is important in practice, since all factors
important in the location decision have not been included in the objective
function. The final decision, in which subjective judgment enters, can be
simplified through the construction of contour lines.
A procedure for constructing contour lines for the rectilinear distance
problem is given next. Figure 4.2 should be consulted in going through the
procedure,
1. Plot the points (a1, b;),-+ +s (Gms), and draw perpendicular lines
(parallel to the x and y axes) through each point.
2. Consider the vertical lines to be numbered 1,2,...,? from left to
right, and the horizontal lines to be numbered 1, 2,..., q from bottom to
top. 174 | Single-Facility Location Problems
Sas °
i ot i
|
2, =
M So | Sy Se
>, —
My So} So | Sop
ic Gs G
mm, My
Figure 4.2. Procedure for constructing contour lines.
3. Calll the x intercept of the jth vertical line c,, and the y intercept of
the ith horizontal line d,. Denote the region delimited by vertical lines j and
Jj + land horizontal lines i and i + 1 by [i, j]. (In order that all regions are
numbered, imagine there is a vertical line numbered 0 to the left of vertical
line 1, a vertical line numbered p + 1 to the right of vertical line p, a horizon-
tal line numbered 0 below horizontal line 1, and a horizontal line g + 1
above horizontal line g.)
4, Let C, and D, be the sums of the weights associated with vertical
line j and horizontal line i, respectively. [For example, if the points (5, 3)
and (5, 10) have the associated weights of 6 and 8, respectively, and the second
vertical line has an x intercept of 5, then C, = 14,] It is convenient to place
the numbers C, at the bottom of the corresponding vertical lines and the
numbers D, at the left of the corresponding horizontal lines.
5. Compute the numbers
N= -¥ = Ew,
Ni=No+2D,
Ma=N,+2D,
M,=My1+2C,= dw, N= N14 +2D,= Sw,
co
and place them as indicated in Figure 4.2, Rectilinear-Distance Location Problems | 175
6. The slope S,, of any contour line passing th ion [if] i
puted as follows: ea
M,
Sy = - =
2]
When N, is zero, the contour line is vertical in region [i,j].
7. To find a point (x*, y*) that minimizes the total cost expression
(4.5), there are four cases to consider:
(a) M,-1 <0, M, > 0, N,-1 <0,N, > 0; take x* =c,, y* =d,.
(b) M,-1 <0, M,=0,N,-1 <0,N,>0; take x* to be any point
between c, and ¢41, )* = d,.
(© M,-1 <0, M,> 0, N,-1 <0,N, =0; take x*=c,; take y* to
be any point between d, and d,s.
(d) M,-, <0, M,=0,N,., <0,N,=0; take x* to be any point
between c, and c,,,, and y* to be any point between d, and d,41.
8. Given the above information, a contour line can be constructed,
‘As an example to illustrate the procedure, consider the problem previ-
ously treated in which the existing facilities have locations of (4, 2), (8, 5),
(11, 8), and (13, 2) with associated weights of 1/6, 1/3, 1/3, and 1/6, respec-
tively. Figure 4.3 shows the slopes in the different regions, while several
contour lines for this problem are plotted in Figure 4.4. Any point (x*, y*)
where x* is between c, = 8 and c, = 11, and y* = d, = 5 is a minimum
cost location.
The justification for all the steps required in constructing the contour
lines, with the exception of step 8, is straightforward, and will now be given.
After going through step 4, the expression (4.5) becomes
fey) = BGlx— + ¥ Diy ail 48)
where ¢, <¢,<---<¢, and dy ] Terminal building
fu =(8,1)
(0, 0)
Figure P4.6
The number of arriving flights per day for each air line is 36, 22, 28, and 18,
for Pokey, Early, Delayed, and Undecided, respectively. If a separate trip is
made from the baggage receiving area to a passenger pickup point for each
arriving flight, and rectilinear distance is assumed, where should the pickup
point be located ? Suppose that the pickup point must be outside and on either
the south or west side of the terminal; where should the pickup point be
located ? Suppose that a conveyorized system is to be installed with separate
conveyor belts joining each airline and the pickup point. If the conveyor
layout follows a rectilinear layout pattern, what is the optimum location for
the pickup point?
Mrs. Black has to pick up several items for her daughter’s party at various
locations. She wants to locate a parking space such that she can get to all the
stores in a minimum amount of time without reparking. Each block is square
—100 feet on a side. Streets running north to south are numbered consecu-
tively. Those running east to west are lettered consecutively. The bakery is
at 6th and E; she must walk half as fast as normal, so that she won't drop the
cake or pastries. At 10th and D is the grocery store. The dress shop is at
12th and G. Mrs. Black picks up her daughter, Kathy, from the beauty parlor
at 10th and G, and they walk twice as fast as normal back to the car, so that
no one will see her with her hair up. It is assumed she must stay on the side-
walks that enclose each block—distance used crossing streets is considered
ible. Determine the location of the parking space that satisfied her objec-
negl
tive.
‘A portable lunch wagon is to be located along the main aisle of an industrial
” plant. The vendor wishes to locate the wagon so that the total walking dis-
tance for his customers is minimized. On a coordinate system the main aisle
runs parallel to the x axis at y = 10. Work stations are located at the follow- 200 | ‘Single-Facility Location Problems
49.
4.10.
441.
ing coordinate points. Travel from the work stations to the main aisle is
rectilinear,
Station Location Station Location
1 (4,4) 6 (12, 12)
2 12) 7 (10, 14)
3 64) 8 (12, 6)
4 (6,12) 9 B14)
5 @,8) 10 (10, 4)
(a) If one customer is located at each station, where should the wagon be
located along the main aisle to minimize curhulative customer walking
distance?
(b) If one customer is located at stations 1, 2, 3, 4, and 5, and three customers
are located at stations 6, 7, 8, 9, and 10, where should the wagon be located
along the main aisle to minimize cumulative customer walking distance?
For the rectilinear-distance problem, it is known that some point c, minimizes
Ai@d). Explain why Djo} Cy; <4 Dh wm and Yj-1 CG; Sh Di wes ie,
explain why the median conditions are necessary conditions for a minimum.
[Property 3 of Section 4.3 and the subsequent discussion establish that if c,
is a point for which the median conditions hold, then ¢, minimizes f,(x);
what is being asked for here is the converse.]
For the rectilinear-distance location problem, explain why, for the median
conditions
any point between c, and cxs1 minimizes f,(x).
Assuming Euclidean distances, what is the optimum location for a single
new facility if P; = (0,9), Ps = (1, 2), Ps = (7, 10), Py = (8, 3), and w=
W, = Ws = Wa?
4.12. Plot the contour line passing through the point (1, 0) for the problem
S(x,y) = 3\x — 1] + 2]1y — 1] + 1ly — 2]
‘What is the value of f(x, y) for this contour line?
(4.13) Suggest an iterative procedure for solving the problem
minimize 3, wie — ai + ( — b)*F, where k > 0
The procedure should agree with the one given in the case for k = }, What
Teasons can you suggest for considering values of k other than 4? Problems | 201
|) Given existing facilities at points P, = (0, 0) and P; = (6,0), with w, = wa
1, plot several contour lines for the (a) rectilinear-distance model, (b)
squared-distance model, and (c) Euclidean-distance model.
4.15. For Problem 4.14, which points will minimize the total cost function for each
of the three cases?
4.16. Suppose that two people go to a shopping center. They have agreed before-
hand that each will visit one store alone and then return to their parked car.
The two stores are at opposite ends of the shopping center, and the car can
be parked at any spot along a straight sidewalk running between the two stores.
Where do you think they would park their car and why? Does your answer
suggest that there may be quantifiable approaches to location problems other
than minimum total cost approaches; if so, what are they?
4.17. Using the contour-line approach for evaluating locations for a new machine,
determine the optimal location based on the following data:
Locations of Existing Machines Possible Locations for New Machines
i % H * my
1 8 a 1 8 10
2 13 2) 2 16 v7
3 13 13 3 2 16
4 18 10 4 16 3
5 2 10 5 16 8
6 8 16
7 18 20
The pallet-load movement per day between the new machine and the ith
existing machine, w,, is given to be
we
100
100
150
100
200
250
100
Rank the alternative locations in order of preference using contour lines.
Vy Rectilinear travel is used.
yas, rhe ABC Auto Parts Company has 11 retail sales stores in the city of Green-
— ville. The ABC Company needs a new warehouse facility to service its retail
stores. The locations of the stores and the expected deliveries per week from
saweenn| 202 |
4.19.
4.20.
Single-Facility Location Problems
the warehouse to each store are
Store Location (miles) Expected Deliveries
1, 0)
(a, 3)
2,0)
(2, —-1)
, -2)
(-1,0)
(1,0)
(-1,3)
(-2, -2)
(-3, -1)
(-3,1)
BS cwrqauaen=
VeaNaUeHeee
Assume that travel within the city of Greenville is rectilinear and that after
each delivery the delivery truck must return to the warehouse.
(a) If there are no restrictions on the warehouse location, where should it
be located?
(b) If the ABC Company is going to open a new retail outlet at the warehouse
location and this outlet must be at least 1 rectilinear mile away from
an existing store, where should it be located?
A plumbing contractor wishes to minimize the cost of pipe for a water waste
disposal system in a building under construction. The locations of all the
facilities that require water waste removal have already been determined.
However, all the individual drain pipes must be connected at some point
along the inside basement wall of the building; ie., y = 0. From the connec-
tion one main drain pipe is used to connect with the sewer system. Assume the
(x, y, 2) coordinates of the connection points within the building are P;
(8, 20, —8), P2 = (12, 10,2), Ps = (25, 35,2), Ps = (40, 20,2), Ps
(25, 35, 12), Ps = (30, 40, 12), P; = (15, 25, 22), and Py, = (25, 35, 22). The
main drain must be connected to the sewer line at Py = (20, —50, —10).
The cost per foot of drain line from the connection point to each facility is
wy = $4, w2 = $5, ws = $3, we = $1.50, ws = $3, we — $1.50, wy = $2,
and w, = $3, respectively. The cost per foot of pipe for the main drain is
$15. Pipe lines will be laid out rectilinearly. Determine the optimum location
for the main drain connection. What is the optimum location if the main
drain connection is constrained to have x <1 and z << —8?
‘As Minister of Defense for the planet Xerxes in the galaxy Euclid, you are
seriously concerned about your four most aggressive rivals and plan to
place one outpost in space fully equipped with an army of laser ships to
be dispatched, as necessary, to attack any of the four rival planets. The posi-
tions in space of the four rivals, relative to an arbitrarily selected origin, are
Ri =(2,2,2), Ra = (12, 6,8), Rs = (10,0, —4), and Ry = ©, 20, —10),
measured in millions of miles. The critical travel time is a function of the Problems | 203
square of the Euclidean distance traveled (due to the repelling forces generated
from the four planets). You have assessed the probability of an attack originat-
ing from each planet and assigned the following values: p; = 0.4, p2 = 0.6,
P31 = 04, and p, + 0.9. (Notice that attacks are not mutually exclusive
events.) Fortunately, due to the magnitude of the diameters of the orbits of
all planets and the outpost (order of 10 million light-years), the entire coordi-
nate system is considered to orbit, such that any changes in the relative posi-
tions of the planets due to orbiting can be ignored. Where should the outpost
be located to minimize the expected distance traveled?
4.21. One new facility is to be located with respect to three existing facilities, and
straight-line movement is used. Annual item movement between the new
facility and each existing facility is given to be w; = 5, w2 = 2, and ws = 10.
The coordinate locations of the existing facilities are Py = (2,6), P2 =
,0), and Ps = @, 3).
(a) Give upper and lower bounds on the straight-line distance traveled per
year. Use (4.44).
(b) Determine the optimum location for the new facility, assuming cost is
proportional to the square of the straight-line distance traveled.
(©) Determine the optimum location for the Euclidean problem.
4.22, Are the following statements true or false? If false, give a reason for the
statement being false.
(a) Let P; = 0,0), P2 = @, 10), Ps = (x,5), and w) = w2 = ws. With
rectilinear distance, the optimum location for the new facility is not
dependent on the value of x.
(b) Under some conditions the rectilinear, Euclidean, and squared Euclidean
problems will yield the same location for a given problem.
(©) The partial derivatives 0 f/x, are always defined for both the Euclidean
and squared Euclidean problem.
(a) The single-facility rectilinear problem is easiest solved using linear
programming.
(© The rectilinear, Euclidean, and squared Euclidean problems can have
multiple optimum solutions.
(Ina single-facility location problem (of the type treated in this chapter)
involving existing facilities at P, = (10, 10), P2 = (12, 15), and P; =
(16, 18), the location (16, 10) can never be an optimum location.
4.23, Let four existing facilities be located at Py (5, 0), Ps = 5, 5),
/and Py = (0,5) with w: = 2, w2 = 1, wa 2. Determine the
optimum location for a single new facility when cost is proportional to (a)
rectilinear distance, (b) squared Euclidean distance, and (¢) Euclidean dis-
tance. Compare the minimum total cost obtained in part (c) with the upper
->bound from (4.44), using the solution obtained in part (a).
Let four existing facilities be located at P, = (0,0), P, = (6,0), Ps =
(6, 16), and Py = (12, 16) with w; = w2 = ws = wy = 2. Determine the
optimum location for a single new facility when cost is proportional to (a)
rectilinear distance, (b) squared Euclidean distance, and (c) Euclidean distance. 204 | Single-Facility Location Problems
4.25.
YY
‘Compare the minimum total cost obtained in part (c) with the upper bound
from (4.44) using the following rectilinear solutions: (6, 0), (6, 8), and (6, 16).
A city wishes to locate a fire station so as to minimize the cost of fire damage.
After some analysis, five major areas have been identified in the city as being
primary risk areas. The feeling is that fire damage increases in proportion to
the square of the time required to answer an alarm, It is also felt that time and
distance are directly proportional. For purposes of the analysis straight-line
distances are assumed, Weighting factors have been obtained such that the
fire loss at area k is equal to w,d2, where w, is the weighting factor and d,
is the Euclidean distance from the fire station to area k. The coordinate loca-
tions of the areas, along with the weight factors, are
+ ate ae we
1 4 12 4
2 8 2 2
5 14 10 2
4 2 3 8
5 14 4 9
What is the optimum location for the fire station?
4.26. Four machines are located at the coordinate points (0, 4), (0, 8), (4, 0), and
421.
4.28.
(8, 0). Thirty per cent of the item movement between the existing machines
and a new machine is between the new machine and each of the existing
machines at (0, 4) and (4, 0). Twenty per cent is between the new machine
and each of the existing machines at (0, 8) and (8, 0). Rectilinear movement is
used. Construct the contour line passing through the positive quadrant and
beginning at the point (0, 12). If the machine locations are interpreted as
docks and the contour line is interpreted as the periphery of the warehouse,
what warehousing interpretation might be given to the percentage weights?
Explain,
Given three existing facilities located at P; = (0, 0), P: = (0, 4), and P; =
(4,2) with w; = wz = ws = 1, determine the upper and lower bounds on
the minimum value of the objective function for the single-facility Euclidean
problem using (4.44).
Let 21,..., 2, be a feasible solution to the dual of the rectilinear problem.
The following conditions will be called complementary slackness conditions:
fey — x) =|z)||e; — x| and (\z,| — C)|x — ¢j| =O forj =1,...,p.
(@) Given a feasible solution to the dual problem z;,...,z, and a point
x such that the complementary slackness conditions hold, prove that x
minimizes f,(x), and z1, . . ., zp isa maximum feasible solution to the dual
problem.
(b) Suppose that x minimizes f,(x) and z,,...,z, is a maximum feasible
solution to the dual problem; prove that the complementary slackness
conditions hold.
(© Given that cy
