MRP lot sizing

Lot-Sizing in MRP

◼ Lot-size is the quantity

ordered/produced at one time
◼ Large lots are preferred because:
 Changeovers cost less and capacity
increases
 Annual cost of purchase orders
decreases
 Price breaks and transportation
economies of scale can be utilized
Lot-Sizing in MRP

◼ Small lots are preferred because of:

 Lower inventory carrying cost
 Reduced risk of obsolescence
 Shorter cycle time to produce customer
order
Lot-Sizing in MRP

◼ Economic order quantity (EOQ)

 Does not consider quantity discounts
 Does not always provide the most
economical approach with lumpy
demands
◼ Lot-for-lot (LFL)
 Accommodates lumpy demand
◼ Period order quantity (POQ)
Lot-Sizing in MRP
The best lot-sizing method,
resulting in least cost …

Depends on cost
and demand patterns
Lot-for-lot (L4L)
◼ The simplest lot sizing scheme for MRP
systems is lot-for-lot (abbreviated L4L).
 This means that requirements are met on a
period by period basis as they arise in the
explosion calculus.
◼ LFL is only for convenience and ease of use,
rather than optimal;
◼ However, more cost effective lot sizing plans
are possible.
 These would require knowledge of the cost of
setting up for production and the cost of holding
each item.
...

◼ The problem of finding the best or near

optimal production plan is described as
 Having known the time-varying demand
and costs of setup and holding, what
production quantities will minimize the
total holding and setup cost over planning
horizon?
 Statement of the Lot Sizing
Problem
◼ Assume:
 There is a known set of requirements (r1, r2,... rn)
over an n period planning horizon.
 Both the set up cost, K, and the holding cost, h, are
given.
 The objective is to determine production quantities
(y1, y2, . . ., yn) to meet the requirements at minimum
cost.
 The feasibility condition to assure there are no
stockouts in any period is: j j

 y  r
i =1
i
i =1
i for 1  j  n.
Methods
1 n
◼ One could apply the EOQ formula by defining  =  ri
n i =1
but there are better methods.

◼ Property of the optimal solution: every optimal solution orders

exact requirements: that is,
y1 = r1 or y1 = r1 + r2 ,. . ., or y1 = r1 + r2 + ... + rn
 MRP-the Explosion Calculus

Indented BOM
Fig 7-5 Product Structure Diagram for Harmon Trumpet
Level 0 Level 1 Level 2
1 Trumpet
1 Bell assembly
1 Valve assembly
3 Slide assemblies
3 Valves
 MRP for the valve casing
assembly (LT=4 weeks)

Week 4 5 6 78 9 10 11 12 13 14 15 16 17
Gross requirement (GR) 42 42 32 12 26 112 45 14 76 38
Net requirement (NR) 42 42 32 12 26 112 45 14 76 38
Planned order release (LFL) 42 42 32 12 26 112 45 14 76 38
 MRP- Alternative Lot-sizing
Schemes
◼ Lot for Lot
 Three parameters are required: the average demand , the
holding cost h, and setup cost K;
 Consider the valve casting assembly: K=$132; the holding cost
is $0.6 per unit per week;
◼ If the holding cost is charged against the inventory each week, the
total holding cost incurred from week 8 through 17 is 0;
◼ Since there is one setup each week, the total setup cost incurred
over 10 weeks planning horizon is 10132=$1320 ;
 The cost can be reduced largely by producing larger amounts
less often;
◼ EOQ
 As the “first cut”, EOQ formula is used to determine an
alternative production policy;
 The total of the time-phased requirements over week 8 through
17 is 439, the average is 43.9/week ; 2K  2 132  43.9
Q= = = 139.
 Using ==43.9, h=0.6, and K=132, h 0.6
 MRP- Alternative Lot-
sizing Schemes
◼ If we schedule the production in lot size 139, while ensuring that the net
requirements are satisfied, then resulting MRP is as follows

Week 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Net requirement (NR) 42 42 32 12 26 112 45 14 76 38
Planned deliveries 139 0 0 0 139 0 139 0 0 139
Planned order release (EOQ)139 0 0 0 139 0 139 0 0 139
Ending Inventory 97 55 23 11 124 12 106 92 16 117

◼ Ending Inv.=Beginning Inv.(Ending Inv. In the last period)+Planned

deliveries-Net requirement
◼ Cost of Using EOQ Lot sizing
 During week 8-17, there are 4 setups, the total setup cost is $132*4= $528;
 The cumulative ending inventories are 97+55+…=653, the total holding cost is
6530.6=$391.80;
 The total cost is 391.80+528=$919.80<$1320 for LFL.
MRP- Alternative Lot-sizing
Schemes

◼ The Silver-Meal Heuristic (Named for

Harlan Meal and Edward Meal)
 A forward method that requires determining the
average cost per period as a function of the
number of periods the current order to span and
stopping the computation when this function first
increase.
 Define C(T) as the average holding cost and
setup cost per period if the current order spans
the next T periods;
MRP- Alternative Lot-
sizing Schemes
◼ Let (r1, r2, …, rn) be the requirements over the n-
period horizon;
 Consider period 1.
 If we produce just enough in period 1 to meet the demand
in period 1, then only the order cost K is incurred. Hence,
◼ C(1)=K/1;
 If we produce just enough in period 1 to meet the demand
in both periods 1 and 2, then r2 is held for one period.
Hence,
◼ C(2)=(K+hr2)/2;
 Similarly, C(3)=(K+hr2+2hr3)/3;
 In general, C(j)=((K+hr2+2hr3+…+(j-1) hrj)/j;
 Once C(j)>C(j-1), stop and set y1=r1+r2+…+rj-1 and start
the process at period j.
 MRP- Alternative Lot-sizing
Schemes:
r=(18, 30, 42, 5, 20 ); h=2; K=80
▪Starting in period1 70<80
C(1)=K/1=80;
102.67>70
C(2)= (K+hr2)/2 =(80+(2)(30))/2=70;
C(3)=(K+hr2+2hr3)/3=(80+(2)(30)+(2)(2)(42))/3=102.67;
Stop at period 2, y1=r1+r2=18+30=48
▪Starting in period 3
45<80
C(1)=K/1=80;
56.67>45
C(2)= (K+hr4)/2 =(80+(2)(5))/2=45;
C(3)=(K+hr4+2hr5)/3=(80+(2)(5)+(2)(2)(20))/3=56.67;
Stop at period 4, y3=r3+r4=42+5=47
 MRP- Alternative Lot-sizing
Schemes: (Cont.)
y5=r5=20
y=(48, 0, 47, 0, 20)
Total Cost= $80*3+30*$2+5*$2= $310
➢Example

r=(10,40,30); h=1; K=50

y=(50,0,30)

Notes: Silver-Meal heuristic does not guarantee optimal solution.

Methods (continued)
◼ Least unit cost (LUC) heuristic.
 One minimizes the average cost per unit of demand (as
opposed to the average cost per period as is done in the
Silver Meal heuristic.)
 Heuristics is similar to Silver-Meal Heuristic method
except that instead of dividing the cost over j periods by
the the periods, j, but by the total number of units
demanded from 1 through j period, r1+r2+…+rj;
 C(1)=K/r1;
 C(2)=(K+hr2)/(r1+r2);
 ...
 The average cost per unit of demand over j periods is
given by:

C ( j ) = ( K + hr2 + 2hr3 + ... + ( j − 1)hrj ) /(r1 + r2 + ... + rj ).

 MRP- Alternative Lot-sizing
Schemes:
r=(18,30,42,5,20 ); h=2; K=80
▪Starting in period1 2.92<4.44
C(1)= K/r1 =80/18=4.44;
3.42>2.92
C(2)= (K+hr2)/(r1+r2) =(80+(2)(30))/(18+30)=2.92;
C(3)=(K+hr2+2hr3)/(r1+r2 +r3)=(80+(2)(30)+(2)(2)(42))/ (18+30+42) =3.42;

Stop at period 2: y1=r1+r2=18+30=48

▪Starting in period3
1.92>1.90
C(1)= K/r3 =80/42=1.90;
C(2)= (K+hr4)/(r3+r4) =(80+(2)(5))/(42+5)=1.92;
Stop at period 3: y3=r3=42
 MRP- Alternative Lot-sizing
Schemes:
r=(18, 30, 42, 5, 20 ); h=2; K=80
▪Starting in period4
4.8<16
C(1)= K/r4 =80/5=16;
C(2)= (K+hr5)/(r4+r5) =(80+(2)(20))/(5+20)=4.8;

y4=r4+r5=5+20=25
y=(48, 0, 42, 25, 0)
Total cost= $80*3+30*$2+20*$2= 340

Silver-Meal (48,0,47,0,20) $310

LUC (48,0,42,25,0) $340

Methods (continued)
◼ Part period balancing (PPB)
 Although the silver meal technique
seems to give better results in grater
number of cases, PPB seems to be more
popular in practice.
 Here one chooses the order horizon to
most closely balance the total holding
cost with the set-up cost.
 The exact matching is rare in an integer
number.
 MRP- Lot-sizing:

r=(18, 30, 42, 5, 20 ); h=2; K=80

▪Starting in period1 228 exceeds the setup cost of 80.
Order Total 80 is closer to 60 than to 228
Horizon holding
cost
1 0
2 60 y1=r1+r2=18+30=48
3 228

60 = 30*2
228= 30*2+42*2*2
 MRP- Lot-sizing:
r=(18, 30, 42, 5, 20 ); h=2; K=80
▪Starting in period3
Order Total 90 is close to 80 than is 10
Horizon holding
cost
1 0
2 10 y3=r3+r4 +r5 =42+5+20=67
3 90
Total cost= $80*2+ 30*$2+5*$2+20*2*$2= $310

Silver-Meal (48,0,47,0,20) $310

LUC (48,0,42,25,0) $340

PPB (48,0,67,0,0) $310

Methods (concluded)

◼ Experimental evidence seems to favor the

Silver Meal Heuristic among the four
discussed as the most cost efficient.
◼ Optimal lot sizes can be found by using
dynamic programming.
◼ A heuristic method for lot sizing subject to
capacity constraints is discussed in this
section.
 The Wagner-Whitin Model:
Dynamic Lot Sizing Notation
t a period (e.g., day, week, month); we will consider t = 1, … ,T,
where T represents the planning horizon.

Dt demand in period t (in units)

ct unit production cost (in dollars per unit), not counting setup or
inventory costs in period t

At fixed or setup cost (in dollars) to place an order in period t

ht holding cost (in dollars) to carry a unit of inventory from period t

to period t +1

Qt the unknown size of the order or lot size in period t

decision variables
Wagner-Whitin Example
Data
t 1 2 3 4 5 6 7 8 9 10
Dt 20 50 10 50 50 10 20 40 20 30
ct 10 10 10 10 10 10 10 10 10 10
At 100 100 100 100 100 100 100 100 100 100
ht 1 1 1 1 1 1 1 1 1 1

 Lot-for-Lot Solution
t 1 2 3 4 5 6 7 8 9 10 Total
Dt 20 50 10 50 50 10 20 40 20 30 300
Qt 20 50 10 50 50 10 20 40 20 30 300
It 0 0 0 0 0 0 0 0 0 0 0
Setup cost 100 100 100 100 100 100 100 100 100 100 1000
Holding cost 0 0 0 0 0 0 0 0 0 0 0
Total cost 100 100 100 100 100 100 100 100 100 100 1000

Since production cost c is constant, it can be ignored.

 Wagner-Whitin Example (cont.)
t 1 2 3 4 5 6 7 8 9 10
Dt 20 50 10 50 50 10 20 40 20 30
ct 10 10 10 10 10 10 10 10 10 10
Data At 100 100 100 100 100 100 100 100 100 100
ht 1 1 1 1 1 1 1 1 1 1

Fixed Order Quantity Solution

t 1 2 3 4 5 6 7 8 9 10 Total
Dt 20 50 10 50 50 10 20 40 20 30 300
Qt 100 0 0 100 0 0 100 0 0 0 300
It 80 30 20 70 20 10 90 50 30 0 0
Setup cost 100 0 0 100 0 0 100 0 0 0 300
Holding cost 80 30 20 70 20 10 90 50 30 0 400
Total cost 180 30 20 170 20 10 190 50 30 0 700
 A key observation
Wagner-Whitin Property
If we produce items in t (incur a setup cost) for use to satisfy
demand in t+1, then it cannot possibly be economical to
produce in t+1 (incur another setup cost) .
Either it is cheaper to produce all of period t+1’s demand in
period t, or all of it in t+1; it is never cheaper to produce some
in each.

Under an optimal lot-sizing policy

(1) either the inventory carried to period t+1
from a previous period will be zero (there
is a production in t+1)
(2) or the production quantity in period t+1
will be zero (there is no production in t+1)
Basic Idea of Wagner-Whitin
Algorithm

By WW Property, either Qt=0 or Qt=D1+…+Dk

for some k.

If jk* = last period of production in a k period

problem,
then we will produce exactly Dk+…DT in
period jk*. Why?

We can then consider periods 1, … , jk*-1 as if

they are an independent jk*-1 period problem.
 t 1 2 3 4 5 6 7 8 9 10

Wagner-Whitin Dt
ct
At
20 50 10 50 50 10 20 40 20 30
10 10 10 10 10 10 10 10 10 10
100 100 100 100 100 100 100 100 100 100

Example ht 1 1 1 1 1 1 1 1 1 1

◼ Step 1: Obviously, just satisfy D1 (note we are

neglecting production cost, since it is fixed).
Z1* = A1 = 100
j1* = 1

◼ Step 2: Two choices, either j2* = 1 or j2* = 2.

 A1 + h1D2 , produce in 1
Z = min  *
*

Z1 + A2 , produce in 2
2

100 + 1(50) = 150

= min 
 100 + 100 = 200
= 150
j2* = 1
Wagner-Whitin Example (cont.)
◼ Step3: Three choices, j3* = 1, 2, 3.
 A1 + h1 D2 + (h1 + h2 ) D3 , produce in 1

Z 3* = min Z1* + A2 + h2 D3 , produce in 2
 Z*2 + A3 , produce in 3
100 + 1(50) + (1 + 1)10 = 170

= min 100 + 100 + (1)10 = 210
150 + 100 = 250
= 170

j3* = 1
Wagner-Whitin Example (cont.)
◼ Step 4: Four choices, j4* = 1, 2, 3, 4.
 A1 + h1 D2 + (h1 + h2 ) D3 + (h1 + h2 + h3 ) D4 , produce in 1
 Z* + A + h D + ( h + h ) D ,
 1 produce in 2
Z 4 = min  *
* 2 2 3 2 3 4

 Z 2 + A3 + h3 D4 , produce in 3
 Z*3 + A4 , produce in 4
100 + 1(50) + (1 + 1)10 + (1 + 1 + 1)50 = 320
100 + 100 + (1)10 + (1 + 1)50 = 310

= min 
150 + 100 + (1)50 = 300
170 + 100 = 270
= 270

j4* = 4
Planning Horizon Property
◼ In the Example:
Given fact: we produce in period 4 for period 4 of
a 4 period problem.
Question: will we produce in period 3 for period 5
in a 5 period problem?
Answer: We would never produce in period 3 for
period 5 in a 5 period problem.

If jt*=t, then the last period in which

production occurs in an optimal t+1 period
policy must be in the set t, t+1,… (this
means that it CANNOT be t-1, t-2……)
Wagner-Whitin Example (cont.)
◼ Step 5: Only two choices, j5* = 4, 5.

Z 3* + A4 + h4 D5 , produce in 4
Z = min  *
*

 Z 4 + A5 ,
5
produce in 5
170 + 100 + 1(50) = 320
= min 
270 + 100 = 370
= 320

j5* = 4
◼ Step 6: Three choices, j6* = 4, 5, 6.
And so on.
Wagner-Whitin Example
Solution
Last Period Planning Horizon (t)
with Production 1 2 3 4 5 6 7 8 9 10
1 100 150 170 320
2 200 210 310
3 250 300
4 270 320 340 400 560
5 370 380 420 540
6 420 440 520
7 440 480 520 610
8 500 520 580
9 580 610
10 620
Zt 100 150 170 270 320 340 400 480 520 580
jt 1 1 1 4 4 4 4 7 7 or 8 8
Produce in period 1 Produce in period 4 Produce in period 8
for 1, 2, 3 (20 + 50 + for 4, 5, 6, 7 (50 + 50 + for 8, 9, 10 (40 + 20 +
10 = 80 units) 10 + 20 = 130 units) 30 = 90 units
 Wagner-Whitin Example
Solution (cont.)
◼ Optimal Policy:
 Produce in period 8 for 8, 9, 10 (40 + 20 + 30 = 90 units)
 Produce in period 4 for 4, 5, 6, 7 (50 + 50 + 10 + 20 =
130 units)
 Produce in period 1 for 1, 2, 3 (20 + 50 + 10 = 80 units)

t 1 2 3 4 5 6 7 8 9 10 Total
Dt 20 50 10 50 50 10 20 40 20 30 300
Qt 80 0 0 130 0 0 0 90 0 0 300
It 60 10 0 80 30 20 0 50 30 0 0
Setup cost 100 0 0 100 0 0 0 100 0 0 300
Holding cost 60 10 0 80 30 20 0 50 30 0 280
Total cost 160 10 0 180 30 20 0 150 30 0 580
WW example