Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 47

Distribution Problems in

Supply Chain
Pick up raw materials/ Deliver to customers
(Travelling Salesman Problem)
-- 12 8 16 9
12 -- 18 11 7
2 8 18 -- 10 8
3 16 11 10 -- 9
9 7 8 9 --
1, F

4 "Given a list of cities and the distances


5 between each pair of cities, what is the
shortest possible route that visits each city
exactly once and returns to the origin city?"
Formulation
Let Xij = 1 if the salesman visits city j immediately after visiting city i. The objective is to minimize

 d
i j
ij X ij

From every city visited, the salesman should go to exactly one city. This is given by

X j
ij 1
.
The salesman reached the present city from exactly one city. This is given by

X i
ij 1

Xij = 0,1.

Plus sub tour elimination constraints


Formulation
Minimize 12X12 + 8X13 + 16X14 + 9X15 + 12X21 + 18X23 + 11X24 + 7X25 + 8X31 + 18X32 + 10X34 +
8X35 + 16X41 + 11X42 + 10X43 + 9X45 + 9X51 + 7X52 + 8X53 + 9X54

Subject to

X12 + X13 + X14 + X15 = 1


X21 + X23 + X24 + X25 = 1
X31 + X32 + X34 + X35 = 1
X41 + X42 + X43 + X45 = 1
X51 + X52 + X53 + X54 = 1
X21 + X31 + X41 + X51 = 1
X12 + X32 + X42 + X52 = 1
X13 + X23 + X43 + X53 = 1
X14 + X24 + X34 + X54 = 1
Solution
The above formulation is that of an assignment problem except that Xjj = 0. The optimum solution is
given by X13 = X24 = X31 = X45 = X52 = 1 with objective function value = 43. This is infeasible to the TSP
because it contains two subtours 1-3-1 and 2-5-4-2.
The optimum solution to the TSP with subtour
X12 + X21 ≤ 1 elimination constraints is given by X15 = X24 = X31 =
2
X13 + X31 ≤ 1 X43 = X52 = 1 with distance = 45.
3
X14 + X41 ≤ 1
X15 + X51 ≤ 1 2
1, F
X23 + X32 ≤ 1 3

4 X24 + X42 ≤ 1
1, F
5 X25 + X52 ≤ 1
X34 + X43 ≤ 1
4
X35 + X53 ≤ 1
5
X45 + X54 ≤ 1
Heuristics
Heuristics are mental shortcuts for solving problems in a quick way that delivers a result that is sufficient
enough to be useful given time constraints

Heuristic technique, is any approach to problem solving or self discovery that employs a practical
method that is not guaranteed to be optimal, perfect, or rational, but is nevertheless sufficient for
reaching an immediate, short-term goal or approximation. Where finding an optimal solution is
impossible or impractical, heuristic methods can be used to speed up the process of finding a
satisfactory solution. Heuristics can be mental shortcuts that ease the cognitive load of making a
decision

Nearest neighbor
Twice around the tree
-- 12 8 16 9
12 -- 18 11 7 Nearest neighbour
8 18 -- 10 8
16 11 10 -- 9 1 – 3 – 5 – 2 – 4 – 1 = 8 + 8 + 7 + 11 + 16 = 50
9 7 8 9 -- 2 – 5 – 3 – 1 – 4 – 2 = 7 + 8 + 8 + 16 + 11 = 50
3 – 5 – 2 – 4 – 1 – 3 = 8 + 7 + 11 + 16 + 8 = 50
4 – 5 – 2 – 1 – 3 – 4 = 9 + 7 + 12 + 8 + 10 = 46
5 – 2 – 4 – 3 – 1 – 5 = 7 + 11 + 10 + 8 + 9 = 45
-- 12 8 16 9 Twice around the tree
12 -- 18 11 7
8 18 -- 10 8 Minimum Spanning Tree
16 11 10 -- 9 7 + 8 + 8 + 9 = 32
9 7 8 9 --
Twice around the tree
1–3–5–2–5–4–5–3-1
2

8 3 Solutions
8 1 – 3 – 5 – 2 – 4 – 1 = 8 + 8 + 7 + 11 + 16 = 50
1, F
7 1 – 5 – 2 – 4 – 3 – 1 = 9 + 7 + 11 + 10 + 8 = 46
1 – 2 – 5 – 4 – 3 – 1 = 12 + 7 + 9 + 10 + 8 = 46
4
5 9
Milk run (Travelling Salesman Path problem)
Milk run is a delivery method used to transport mixed loads from various suppliers to one
customer. It can also be used to describe a transportation route that has many stops. The
term originated in the dairy industry.

Milk run is a delivery approach that involves using one vehicle to pick up goods from various
suppliers and transporting them in a single shipment to one customer.

The Traveling Salesman Problem (TSP) is the challenge of finding the shortest path or
shortest route for a salesperson to take, given a starting point, a number of cities (nodes),
and optionally an ending point.
Path problem between 1 and 4
-- 12 8 16 9 Let Xij = 1 if the salesman visits city j immediately after visiting city i. The first
12 -- 18 11 7 and last cities are 1 and 4 respectively. The objective function is to
8 18 -- 10 8 Minimize 12X12 + 8X13 + 9X15 + 18X23 + 11X24 + 7X25 + 18X32 + 10X34 + 8X35 +
16 11 10 -- 9 7X52 + 8X53 + 9X54
9 7 8 9 -- Subject to
X12 + X13 + X15 = 1
X23 + X24 + X25 = 1
X32 + X34 + X35 = 1
X52 + X53 + X54 = 1
X12 + X32 + X52 = 1
X13 + X23 + X53 = 1
X24 + X34 + X54 = 1
X15 + X25 + X35 = 1
Path problem between 1 and 4
-- 12 8 16 9
12 -- 18 11 7
8 18 -- 10 8 Solver gives the solution X13 = X34 = X25 = X52 = 1 with Z = 32
16 11 10 -- 9
9 7 8 9 -- We add X25 + X52 ≤ 1

We get the solution X13 = X35 = X52 = X24 = 1 with Z = 34

Heuristics

1. Nearest neighbour
2. Spanning tree based
Path problem between 1 and 4 - Heuristic
-- 12 8 16 9
12 -- 18 11 7
8 18 -- 10 8 Nearest neighbour
16 11 10 -- 9 1 – 5 – 2 – 3 – 4 = 9 + 7 + 18 + 10 = 44
9 7 8 9 --

Heuristics

1. Nearest neighbour
2. Spanning tree based
Path problem between 1 and 4 - Heuristic
-- 12 8 16 9 Heuristics
2
12 -- 18 11 7 3
Spanning tree based
8
8 18 -- 10 8
8
16 11 10 -- 9 1, F
7
9 7 8 9 --
4 1 – 3 – 5 – 2 – 4 = 8 + 8 + 7 + 11 = 34
5 9 1 – 3 – 2 – 5 – 4 = 8 + 18 + 7 + 9 = 42

1 and 4 should have


degree 1. 2, 3, 5 have
degree 2
TSP with time windows (Alternate formulation for TSP)
Let tj be time at which the salesman (starts from) or reaches city j. If there are n cities,
we assume that city 1 becomes city n+1 (which is the final destination). The
formulation from Black (1983) is considered for illustration.

Minimize tn+1 – t1 -- 12 8 16 9
Subject to 12 -- 18 11 7
tj – t1 ≥ d1j j = 2,...,n 8 18 -- 10 8
tn+1 – tj ≥ d1j j = 2,...,n 16 11 10 -- 9
9 7 8 9 --
ti  t j  dij for i = 3,...,n; j < I; tj ≥ 0
Formulation t1 t2 t3 t4 t5 t6
0 29 8 18 36 45
Minimize t6 – t1 t3 - t2 +1000d23<= 982
1 1 0 0 0 0
subject to t2 - t3 -1000d23<= -18 d23 d24 d25 d34 d35 d45
t2 - t1 ≥ 12 t4 - t2 +1000d24<= 989
t 3 - t1 ≥ 8 Solution: 1 – 3 – 4 – 2 – 5 – 1
t2 - t4 - 1000d24<= -11
= 8 + 10 + 11 + 7 + 9 = 45
t4 - t1 ≥ 16 t4 - t3 +1000d34<= 990
t 5 - t1 ≥ 9 t3 - t4 - 1000d34<= -10 Constraints representing the time windows.
t6 - t2 ≥ 12 t5 - t2 + 1000d25<= 993 These are t2 ≤ 20, t3 ≤ 30 and t4 ≥ 50.

t 6 - t3 ≥ 8 t2 - t5 - 1000d25<= -7
Solution: 1 – 2 – 3 – 5 – 4 – 1
t6 - t4 ≥ 16 t5 - t3 + 1000d35<= 992 = 12 + 18 + 8 + 9 + 16 = 63
t 6 - t5 ≥ 9 t3 - t5 - 1000d35<= -8
t5 - t4 + 1000d45<= 991 Solution 0, 12, 30, 50, 38, 66

t4 - t5 - 1000d45<= -9
Pick up and delivery problem (Dial a ride problem)
Pickup and delivery problem where the salesman visits some cities to pickup material
and delivers it in other cities. The salesman may have to visit all the pickup points first
and collect all the material and then deliver them to the delivery points.

If the salesman is a taxi driver, every pickup has to be followed by a delivery because
each pickup may be a passenger who has to be dropped at the destination. This
particular version of the pickup and delivery problem is called the “dial-a-ride
problem”. Sometimes, these problems also have associated time windows.
Formulation
Assume that cities 1 and 4 are pickup points and cities 2 3 and 5 are delivery points.
Assume that the demand in each city (2, 3 and 5) is d. Assume that the salesman picks up
2d from city 1 and d from city 4. Find the sequence that minimizes total distance
travelled.
t2 ≥ t1 + d12 The optimum solution is t1 = 0, t2 = 27, t3 = 45, t4
t3 ≥ t1 + d13 = 16, t5 = 34 and t6 = 53 and objective function
t5 ≥ t1 + d15 value = 53
t2 ≥ t4 + d24
1 – 4 – 2 – 5 – 3 – 1 = 16 + 11 + 7 + 8 + 8 = 50
t3 ≥ t4 + d34
t5 ≥ t4 + d45 (Importance of triangle inequality)
TSP with subsets – Orienteering problem,
1. The salesman collects revenue in each city and returns to the starting city when a
certain amount is collected. The problem is to travel minimum distance to collect
the required amount. This problem is called the orienteering problem. The
salesman can also return to a known destination.

2. Another version is to maximize the amount collected subject to a distance


constraint. If the revenue in each city is 1, the problem reduces to visiting
maximum number of cities and returning to the starting point subject to a distance
constraint

3. There are several suppliers. Visiting minimum number of suppliers to collect a


certain fixed quantity
Formulation
Let Yj = 1 if the salesman visits city j. Let be the time at which the salesman visits city j.
Also city n+1 is the same as city 1.
Minimize tn+1 – t1
Subject to
tj – t1 ≥ d1jYj
tn+1 – tj ≥ d1jYj
t j  ti  d ijYY
i j
for all i,j = 2,.., n and i ≠ j.
Yj = 0,1.
DijYij +tj – ti ≤ Mδij
DijYij +ti – tj ≤ M(1-δij)
Yij ≥ Yi + Yj -1
2Yij ≤ Yi + Yj
Yij, δij = 0,1
Illustration
-- 12 8 16 9 10 The revenue in cities 2 to 6 are 100, 70, 80, 60
12 -- 18 11 7 8
8 18 -- 10 8 12 and 90.
16 11 10 -- 9 11
9 7 8 9 -- 9 Solve the orienteering problem where the
10 8 12 11 9 -- salesman has to collect at least Rs 300

1 – 2 – 3 – 4 – 5 – 1 has collection = 310 and distance = 12 + 18 + 10 + 9 + 9 = 58


Formulation t3 - t2 +1000d23 + 18y32≤ 1000
t2 - t3 -1000d23 + 18y32≤ 0
Let Yj = 1if the salesman visits city j. t4 - t2 +1000d24 +11y42≤ 1000
Let tj be the time at which the salesman visits city j. t2 - t4 - 1000d24 + 11y42≤ 0
Also city 7 is the same as city 1. t4 - t3 +1000d34 + 10y43≤ 1000
t3 - t4 - 1000d34 + 10y43≤ 0
The formulation is to Minimize t7 – t1 t5 - t2 + 1000d25 + 7y52≤ 1000
Subject to t2 - t5 - 1000d25 + 7y52≤ 0
subject to t5 - t3 + 1000d35 + 8y53≤ 1000
t2 - t1 ≥ 12 t3 - t5 - 1000d35 + 8y53≤ 0
t3 - t1 ≥ 8 t5 - t4 + 1000d45 + 9y54≤ 1000
t4 - t1 ≥ 16 t4 - t5 - 1000d45 + 9y54≤ 0
t5 - t1 ≥ 9 t6 - t2 + 1000d26 + 8y62≤ 1000
t6 - t1 ≥ 10 t2 - t6 - 1000d26 + 8y62≤ 0
t7 - t2 ≥ 12 t6 - t3 + 1000d36 + 12y63≤ 1000
t7 - t3 ≥ 8 t3 - t6 - 1000d36 + 12y63≤ 0
t7 - t4 ≥ 16 t6 - t4 + 1000d46 + 11y64≤ 1000
t7 - t5 ≥ 9 t - t - 1000d + 11y ≤ 0
Formulation (continued)
y32 - y2 -y3 ≥ -1 2y63 -y6 -y3 ≤ 0 The optimum solution to the binary IP is given by Y2 = Y3
2y32 -y2 -y3 ≤ 0 y54 - y5 -y4 ≥ -1 = Y5 = Y6 = 1, t2 = 16, t3 = 36, t5 = 9, t6 = 24 and t7 = 44. The
y42 - y2 -y4 ≥ -1 2y54 -y5 -y4 ≤ 0 minimum distance travelled is 44. From the tj values we
2y42 -y2 -y4 ≤ 0 y64 - y6 -y4 ≥ -1 observe that the route is 1-5-2-6-3-1 with distance = 44.
y52 - y2 -y5 ≥ -1 2y64 -y6 -y4 ≤ 0 The total amount collected is 320.
2y52 -y2 -y5 ≤ 0 y65 - y6 -y5 ≥ -1
y62 - y2 -y6 ≥ -1 2y65 -y6 -y5 ≤ 0 (The solution Y3 = Y4 = Y5 = Y6 = 1 gives us a total revenue
2y62 -y2 -y6 ≤ 0 of 300. This is not chosen because the distance travelled
y43 - y4 -y3 ≥ -1 is more).
2y43 -y4 -y3 ≤ 0
y53 - y5 -y3 ≥ -1
2y53 -y5 -y3 ≤ 0
100y2 + 70y3 + 80y4 + 60y5 + 90y6 ≥ 300
y63 - y6 -y3 ≥ -1
Yj ≥ 0, Yij = 0,1
Vehicle Routeing – (Supply and demand)
Vehicle routing problem (VRP) is a generic name given to a whole class of problems
concerning the optimal design of routes to be used by a fleet of vehicles to serve a
set of customers. It is a generalization of a TSP. The distance of the cities from the
depot is 12, 8, 16, 9, 10 and 11.
7 6 Capacity = 20
2
8 1 3

-- 18 11 7 8 8
5
18 -- 10 8 12 6 4
D
11 10 -- 9 11 9
7 8 9 -- 9 12 8 6
6 5
8 12 11 9 -- 10
8 6 9 12 10 -- Distance = 12 + 18 + 8 + 7 = 45; 16 + 11 + 10 + 11 = 48
Total = 93
VRP Formulation (Achuthan and Caccetta,1991)
If there are N customers, we have a graph with
8 1
N+2 nodes, where node N+1 is the depot
(source) and node N+2 is also the depot
7 2
(destination). The objective function is
6 Minimize

3
8,D
Z cij X ij
7,D ( i , j ) A
5
4

6
5

8
6
VRP Formulation (Achuthan and Caccetta,1991)
yi  y j   L  1 X ij  L
𝑁

∑ 𝑋 𝑖𝑗 + 𝑋 𝑁 +1 , 𝑗 =1
¿ 𝑖=1
¿ 𝑖≠ 𝑗
yN  2  yN 1  L  1
𝑁

∑ 𝑋 𝑖𝑗 + 𝑋𝑖 , 𝑁 +2=1 ui  u j  WX ij  W  Q j
¿ 𝑗=1
¿ 𝑗≠ 𝑖
u N  2  u N 1  W
𝑁

∑ 𝑋 𝑁 +1 , 𝑗 =𝑉 vi  v j   cij  T  X ij  T
¿ 𝑗=1
¿
vN  2  vN 1  T
𝑁

∑ 𝑋 𝑖, 𝑁 +2 =𝑉
¿ 𝑖=1
¿

X ij  0,1 for all (i, j )  A


Solving for V = 2
Minimize
8X12+11X13+7X14+8X15+8X16+12X18+18X21+10X23+8X24+12X25+6X26+8X28+11X31+10X32+9X34+11X35+9X36+16X38+7X41+8X42
+9X43+9X45+12X46+9X48+8X51+12X52+11X53+9X54+10X56+10X58+8X61+6X62+9X63+12X64+10X65+11X68+12X71+8X72+16X73+9X
74 +10X75+11X76

X21+X31+X41+X51+X61+X71 = 1; X12+X32+X42+X52+X62+X72 = 1 Initial solution is


X14 = X28 = X36 = X48 = X51 = X63 = X71 = X75 = 1 with
X13+X23+X43+X53+X63+X73 = 1; X14+X24+X34+X54+X64+X74 = 1
Distance = 68. This has a subtour 3 – 6 – 3.
X15+X25+X35+X45+X65+X75 = 1; X16+X26+X36+X46+X56+X76 = 1 We add X36 + X63 ≤ 1 and solve.
X12+X13+X14+X15+X16+X18 = 1; X21+X23+X24+X25+X26+X28 = 1
X31+X32+X34+X35+X36+X38 = 1; X41+X42+X43+X45+X46+X48 = 1 Next solution is
X51+X52+X53+X54+X56+X58 = 1; X61+X62+X63+X64+X65+X68 = 1 X16 = X28 = X34 = X48 = X51 = X63 = X72 = X75 = 1 with
X71+X72+X73+X74+X75+X76 = 2; X18+X28+X38+X48+X58+X68 = 2 Distance = 69. This has 2 paths 7 – 5 – 1 – 6 – 3 – 4 - 8
and 7 – 2 – 8 with distance = 69
Solving for V = 1 (TSP)

Minimize
8X12+11X13+7X14+8X15+8X16+12X18+18X21+10X23+8X24+12X25+6X26+8X28+11X31+10X32+9X34+11X35+9X36+16X38+7X41+8X42
+9X43+9X45+12X46+9X48+8X51+12X52+11X53+9X54+10X56+10X58+8X61+6X62+9X63+12X64+10X65+11X68+12X71+8X72+16X73+9X
74 +10X75+11X76

X21+X31+X41+X51+X61+X71 = 1; X12+X32+X42+X52+X62+X72 = 1
X13+X23+X43+X53+X63+X73 = 1; X14+X24+X34+X54+X64+X74 = 1
Initial solution is X15 = X26 = X34 = X41 = X58 =
X15+X25+X35+X45+X65+X75 = 1; X16+X26+X36+X46+X56+X76 = 1 X63 = X72 = 1 with Distance = 57. This has a
X12+X13+X14+X15+X16+X18 = 1; X21+X23+X24+X25+X26+X28 = 1 tour 7 – 2 – 6 – 3 – 4 – 1 – 5 – 8
X31+X32+X34+X35+X36+X38 = 1; X41+X42+X43+X45+X46+X48 = 1
X51+X52+X53+X54+X56+X58 = 1; X61+X62+X63+X64+X65+X68 = 1
X71+X72+X73+X74+X75+X76 = 1; X18+X28+X38+X48+X58+X68 = 1
V = 2 and 3 cities for a vehicle
yi – yj + 4Xij ≤ 3 for i = 1, ...7, j = 1,..., 8; i ≠ j. There are 42 constraints. We also add the
constraint y8 – y7 ≤ 4 instead of y8 – y7 ≤ 7. These ensure that each vehicle goes to exactly
three cities.

The optimum solution to the MILP is given by X14 = X23 = X36 = X48 = X51 = X68 = X72 = X75 = 1
with distance = 72. Since the routes start from the depot (city 7), the two routes are d – 2
– 3 – 6 – d and d – 5 – 1 – 4 – d. The two routes end at the depot (city 8). The individual
distances are 38 and 34 with a total of 72.
With demand and capacities
We consider the case where the capacity of the vehicle is 20 and the demands in the six cities are 8, 7, 6, 5,
6 and 8. We remove the three city limit on the two vehicles by removing the set of constraints. We add the
constraints ui – uj + 20Xij ≤ 20 – dj. We write these constraints the same way as the earlier set. We have 42
constraints. For illustration the first constraint is u1-u2+20X12≤ 13. We also add the constraint u8 – u7 = 20

The optimum solution to the MILP is given by X18 = X24 = X35 = X41 = X58 = X63 = X72 = X76 = 1 with distance = 76.
Since the routes start from the depot (city 7), the two routes are d – 2 – 4 – 6 – d and d – 6 – 3 – 5 – d. The
two routes end at the depot (city 8). The individual distances are 35 and 41 with a total of 76. It is to be
noted that each vehicle visits three vehicles due to the number of vehicles considered and due to the
demand. The best solution with three cities for each vehicle (earlier solution with Z = 72) is infeasible to the
demand constraints. It is also to be noted that if we add the three city restriction (another 43 constraints)
along with the demand constraints, we will get the same solution with Z = 76.
Distance restriction (37)
We add the distance restriction of 37 by adding the constraints vi – vj + (37 + dij)Xij ≤ 37. We write
these constraints the same way as the earlier set. We have 42 constraints. For illustration the first
constraint is v1-v2+55X12≤ 37. We also add the constraint v8 – v7 = 37. We do not add the demand
constraints and the three city restriction constraints.

The optimum solution to the MILP is given by X15 = X23 = X34 = X48 = X58 = X61 = X72 = X76 = 1 with
distance = 73. Since the routes start from the depot (city 7), the two routes are d – 2 – 3 – 4 – d
and d – 6 – 1 – 5 – d. The two routes end at the depot (city 8). The individual distances are 36
and 37 with a total of 73. It is to be noted that each vehicle visits three vehicles due to the
number of vehicles considered and due to the demand. The best solution with three cities for
each vehicle (earlier solution with Z = 72) is infeasible to distance limit of 37 for a vehicle. It is
also to be noted that if we add the three city restriction (another 43 constraints) along with the
demand constraints, we will get the same solution with Z = 73.
Heuristics – Savings based – Clarke Wright

• If a vehicle has to go separately to city i and city j from the origin, the total distance
covered will be 2[d0i + d0j].
• If a vehicle leaves the depot goes to i and then j and returns to the depot, the distance
covered will be (d0i + d0j + dij).
• The saving will be the difference between them. The saving Sij = (d0i + d0j - dij).
Example The distance of the cities from the depot is 12, 8,
16, 9, 10 and 11.
-- 18 11 7 8 8
i-j Saving i-j Saving
18 -- 10 8 12 6 2-6 13
3-6 18
11 10 -- 9 11 9 1-3 17 5-6 11
7 8 9 -- 9 12 3-4 16 4-5 10
8 12 11 9 -- 10 1-6 15 2-4 9
3-5 15 4-6 8
8 6 9 12 10 -- 1-4 14 2-5 6
1-5 14 1 -2 2
2-3 14
Solution for 2 vehicles and 3 cities
3 – 6 = 18, 1 – 3 = 17;
4 – 5 = 10; 2 – 4 = 9 6 trips from depot and back = 66 x 2 = 132
D – 6 – 3 – 1 – D = 11 + 9 + 11 + 12 = 43 Savings = 54
D – 2 – 4 – 5 – D = 8 + 8 + 9 + 10 = 35 Distance = 78
Total = 78
Example (new solution) The distance of the cities from the depot is 12, 8,
16, 9, 10 and 11.
-- 18 11 7 8 8
i-j Saving i-j Saving
18 -- 10 8 12 6 2-6 13
3-6 18
11 10 -- 9 11 9 1-3 17 5-6 11
7 8 9 -- 9 12 3-4 16 4-5 10
8 12 11 9 -- 10 1-6 15 2-4 9
3-5 15 4-6 8
8 6 9 12 10 -- 1-4 14 2-5 6
1-5 14 1 -2 2
2-3 14
Solution for 2 vehicles and 3 cities
3 – 6 = 18, 2 – 3 = 14;
1 – 4 = 14; 1 – 5 = 14 6 trips from depot and back = 66 x 2 = 132
Savings = 60 Savings = 60
Distance = 72
Example (TSP) The distance of the cities from the depot is 12, 8,
16, 9, 10 and 11.
-- 18 11 7 8 8
i-j Saving i-j Saving
18 -- 10 8 12 6 2-6 13
3-6 18
11 10 -- 9 11 9 1-3 17 5-6 11
7 8 9 -- 9 12 3-4 16 4-5 10
8 12 11 9 -- 10 1-6 15 2-4 9
3-5 15 4-6 8
8 6 9 12 10 -- 1-4 14 2-5 6
1-5 14 1 -2 2
Solution for 1 vehicle 2-3 14
3 – 6 = 18; 1 – 3 = 17; 6 trips from depot and back = 66 x 2 = 132
1 – 4 = 14; 5 – 6 = 11; Savings = 66; Distance = 66
2–5=6 Nearest neighbour = D – 2 – 6 – 1 – 4 – 3 – 5 – D
D–4-1–3–6–5–2-D Distance = 8 + 6 + 8 + 7 + 9 + 11 + 10 = 59
9 + 7 + 11 + 9 + 10 + 12 + 8 = 66
Example (capacity) The distance of the cities from the depot is 12, 8, 16,
9, 10 and 11. Demand = 8, 7, 6, 5, 6, 8; capacity = 20
-- 18 11 7 8 8
i-j Saving i-j Saving
18 -- 10 8 12 6 2-6 13
3-6 18
11 10 -- 9 11 9 1-3 17 5-6 11
7 8 9 -- 9 12 3-4 16 4-5 10
8 12 11 9 -- 10 1-6 15 2-4 9
3-5 15 4-6 8
8 6 9 12 10 -- 1-4 14 2-5 6
1-5 14 1 -2 2
Solution for 2 vehicles; capacity = 20 2-3 14
3 – 6 = 18, 3 - 5 = 15;
1 – 4 = 14; 2 – 4 = 9
Savings = 56

Distance = 132 – 56 = 76
Point to point delivery (Bin packing)
Nine point to point deliveries each requiring 32, 26, 23, 17, 16, 14, 10, 10 and 2 minutes are to be made.
Each vehicle is available for 50 minutes. Solve the bin packing problem optimally?

The bin packing problem is about finding minimum number of bins (vehicles) required to make these
deliveries. We start with a heuristic solution:

V1 = 32 + 17
V2 = 26 + 23
V3 = 16 + 14 + 10 + 10
V4 = 2

This heuristic is called “First Fit decreasing heuristic”. This gives a solution with 4 vehicles. We should
check if there is a solution with 3 vehicles.
Formulation (for 3 vehicles)
The formulation is to minimize y1 + y2 +y3
Subject to
x11 + x12 + x13 = 1
x21 + x22 + x23 = 1
x31 + x32 + x33 = 1
x41 + x42 + x43 = 1
x51 + x52 + x53 = 1
x61 + x62 + x63 = 1
x71 + x72 + x73 = 1
x81 + x82 + x83 = 1
x91 + x92 + x93 = 1
32x11 + 26x21 + 23x31 + 17x41 + 16x51 + 14x61 + 10x71 + 10x81 + 2x91 -50y1 <= 0
32x12 + 26x22 + 23x32 + 17x42 + 16x52 + 14x62 + 10x72 + 10x82 + 2x92 - 50y2 <= 0
32x13 + 26x23 + 23x33 + 17x43 + 16x53 + 14x63 + 10x73 + 10x83 + 2x93 - 50y3 <= 0
Yj, Xij = 0,1
Solution (3 vehicles)
Nine point to point deliveries each requiring 32, 26, 23, 17, 16, 14, 10, 10 and 2 minutes are to be made.
Each vehicle is available for 50 minutes. Solve the bin packing problem optimally?

The optimum solution to the formulation is given by Y1 = Y2 = Y3 = 1; X12 = X52 = X92 = 1; X21 = X61 = X71 = 1; X33
= X43 = X83 = 1 with Z = 3. This shows that we need three bins. Items 1, 5 and 9 go to one bin, items 2, 6 and
7 to a bin and items 3, 4 and 8 to the third bin.
Bin packing as a series of knapsack problems
Maximize 32Y1 + 26Y2 + 23Y3 + 17Y4 + 16Y5 + 14Y6 + 10Y7 + 10Y8 + 2Y9
Subject to
32Y1 + 26Y2 + 23Y3 + 17Y4 + 16Y5 + 14Y6 + 10Y7 + 10Y8 + 2Y9 ≤ 50
Yj = 0,1.

Y1 = Y5 = Y9 = 1 with Z = 50.

Y2 = Y6 = Y7 = 1 with Z = 50

Y3 = Y4 = Y8 = 1 with Z = 50
4 vehicles – balance loads - Heuristic
Nine point to point deliveries each requiring 32, 26, 23, 17, 16, 14, 10, 10 and 2 minutes are to be made. Each
vehicle is available for 50 minutes. Solve the bin packing problem optimally for 4 vehicles to balance loads

Item 1 goes to V1 – Load = 32; Item 2 goes to V2; Load = 26; Item 3 goes to V3; Load = 23; Item 4 goes to V4; Load = 17
Item 5 goes to V4; Load = 33; Item 6 goes to V3; Load = 37; Item 7 goes to V2; Load = 36; Item 8 goes to V1; Load = 42
Item 9 goes to V4; Load = 35
4 vehicles – balance loads - Formulation
4 9 9 9

∑ 𝑡 𝑖 𝑋 𝑖𝑗 ≤ 50 ∑ 𝑡 𝑖 𝑋 𝑖𝑗 ≤ 𝑢 ∑ 𝑡 𝑖 𝑋 𝑖𝑗 ≥ 𝑣 Minimize u - v
∑ 𝑋 𝑖𝑗= 1
𝑖 =1 𝑖 =1 𝑖 =1
𝑗 =1

Nine point to point deliveries each requiring 32, 26, 23, 17, 16, 14, 10, 10 and 2 minutes are to be made. Each
vehicle is available for 50 minutes. Solve the bin packing problem optimally for 4 vehicles to balance loads

The optimum solution is X12 = X24 = X33 = X43 = X51 = X64 = X71 = X81 = X92 = 1; u = 40 and v = 34
Vehicle 1 gets 5, 7 and 8; Load = 36; Vehicle 2 gets 1 and 9; Load = 34; Vehicle 3 gets 3 and 4; Load = 40;
Vehicle 4 gets 2 and 6; Load = 40
Case study – Distribution (Inventory vs Transportation)

Nine customers – 32, 26, 23, 17, 16, 14, 10, 10, 2
Demand (day 1) – 12, 8, 3, 4, 10, 10, 7, 8, 9
Day 2 – 10, 9, 3, 8, 10, 9, 8, 2, 9
Inventory cost/item 2, 3, 5, 6, 4, 8, 2, 4, 5
Vehicle capacity = 20
Cost/unit time (distance) = 1

Optimize sum of inventory and transportation


Heuristic solution
• Cannot pick 2 days demand for item 1
• Two days 4 vehicles; pick daily demand = 150 x 2 = 300
• Pick item 1, 2, 3, 4, 5 on both days, Combine 6, 7, 8, 9
Cost = 2 x (32 + 26 + 23 + 17 + 16) + 8 x 10 + 2 x 7 + 4 x 8 + 5 x 9
= 2 x 114 + 80 + 14 + 32 + 45 = 399

Optimum cost?
Formulation
Let Xijk = 1 if item i goes to vehicle j on day k 72 Xijk and 9 Yi variables
Yi = 1 if day 2 demand for item i is picked on day 1
8 + 9 + 9 + 9 + 9 = 44 constraints
Minimize

∑ 𝑑 𝑖 𝑋 𝑖𝑗𝑘 ≤ 𝐵 ∀ 𝑗 , 𝑘
𝑖
❑ ❑

∑ 𝑋 𝑖𝑗 1=1 ∀ 𝑖 ∑ 𝑋 𝑖𝑗 2 ≤ 1 ∀ 𝑖
𝑗 𝑗
❑ ❑
∑ 𝑋 𝑖𝑗 2+ 𝑌 𝑖 =1 ∀ 𝑖 ∑ 𝑏𝑖 1 𝑋 𝑖𝑗 1+ 𝑏𝑖 2 𝑌 𝑖 ≤ 𝐵 ∀ 𝑖
𝑗 𝑗
Solution
Objective function value = 290 (267 + 23)
X121 = X142 = X211 = X232 = X311 = X441 = X412 = X541 = X541 = X641 = X622
= X731 = X732 = X831 = X931 = X912 = Y3 = Y8 = 1

Vehicle Day 1 (Items + Distance) Day 2 (Items + Y3 = Y8 = 1


Distance)
V1 2, 3 26 + 23 4, 9 17 + 2 Item 3 = 3 x 5 = 15
V2 1 32 6 14 Item 8 = 2 x 4 = 8
V3 7, 8, 9 10 + 10 + 2 2, 7 26 + 10 Total = 23
V4 4, 5, 6 17 + 16 + 14 1, 5 32 + 16
Total 150 117
Simple Rule
Nine customers – 32, 26, 23, 17, 16, 14, 10, 10, 2; Cost/km = 1
Demand (day 1) – 12, 8, 3, 4, 10, 10, 7, 8, 9 Vehicle capacity = 20
Day 2 – 10, 9, 3, 8, 10, 9, 8, 2, 9
Inventory cost/item 2, 3, 5, 6, 4, 8, 2, 4, 5

Item 1 – No capacity
Item 2 – Distance = 26; Inventory = 9 x 3 = 27 (not economical)
Item 3 – Distance = 23; Inventory = 3 x 5 = 15 (keep inventory) – Saving = 8
Item 4 – Distance = 17; Inventory = 32
Item 5 – Distance = 16; Inventory = 10 x 4 = 40
Item 6 - Distance = 14; Inventory = 9 x 8 = 72
Item 7 – Distance = 10; Inventory = 8 x 2 = 16
Item 8 – Distance = 10; Inventory = 2 x 4 = 8 (Economical) – Saving = 2
Item 9 – Distance = 2; Inventory = 9 x 5 = 45

Total saving = 10; Total distance (no inventory) = 300; Cost = 290

What happens when vehicle visits multiple customers - TSP


Topics
• TSP
• Path problem
• Time Windows
• Orienteering
• VRP
• Capacity, Distance, Cities,
• Bin packing
• Inventory vs Transportation (multi period)

You might also like