Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Lecture 05
Review of Inventory Models
• STEPS:
Steps to find a practical T:
1. Calculate T*
2. Pick a base time period (day, week, etc.)
3. Find the lowest value of k that satisfies the following relationship:
Power of Two Policy
Steps to find a practical T:
1. Calculate T*
2. Pick a base time period (day, week, etc.)
3. Find the lowest value of k that satisfies:
Example:
Demand: 25.000 units per year Step 1
Ordering cost: 750 $/order Q*=3.131 units/order
Cost of product: 25,5 $/unit T*=0,125 years/order
Holding cost: 15% of unit cost per year TCosts= $11.976 $/year
Step 2.
Base time period: weeks. T*=52(0,125)=6,5 weeks/order
T*/√2=6,5/√2=4,59. √2T*=√2(6,5)=9,19
Step 3.
The value of 2k that is between those bounds is 8 =(23), hence, k=3
Then T*=8 weeks=8/52=0,153
Now we update Q=TD=0,153(25.000)=3846 units and Tcosts= 12.230$/year
Economic Production Quantity: Finite Replenishment
We previously assumed that replenishment quantity arrives at the same time. Now, it
becomes available at a rate of P per unit time rather than all at one time. So P is
the production rate. In this case we assume that it is internally replenished.
So we have now the EOQ with concurrent production and the average inventory level
is now Q 1 − D 2 . This lowers holding costs and leads to higher EPQ.
P
P
P D
P-D
?
Economic Production Quantity: Finite Replenishment
D =(P-D)*Q/P
P P D
=QP/P-DQ/P
P-D
? ?
P-D
? =Q-QD/P
=Q(1-D/P)
Insights:
Since demand is 1680 per day and the production rate is 1000 per
hour and assuming that the plant operates 24 hours per day:
▪ D = 1680(365) = 613.200
▪ P = 1000(24)(365) = 8.760.000
2 AD 2(150)(613.200)
Q* = = 31.449
h(1 − D / P) 0, 20(1 −
613.200
)
8.760.000
Other EOQ models
Inventory
Q Reorder
Lever (R)
Average R=D*LT
Leadtime
Usage (d)
Current Inventory
R Position
Demand
Variability
D = d1 + d 2 + + d L
(forecast error)
▪ Usually we assume that the demand each day has the same
mean and standard deviation. Thus, we can determine:
X −
Z= Nor (0,1)
Z 0.5 Z 0.95 = 1.65
▪ The size of the safety stock (ss) should be large enough to
accommodate the demand during the lead time with a probability of 1-
a, the service level
mean = Ld, Std. Dev. = √L*d
Mean demand
during lead time L D
ss Pr D Ld + ss a
Example
2dA 2(100)(100)
Q= = = 1000
h 0, 02
Pr D Ld + ss a
Z 0,95 = 1, 644
▪ Z0,95 = 1,644
Z1−a =
X −
=
( Ld + ss ) − Ld
=
ss
L d L d
▪ Solving for ss:
ss = Z1−a L d
▪ In our example:
R = d L + Z L d
▪ In our example
R = (100)(2) + 1, 644 ( )
2 25 = 258,15
Types of Service Levels
1. Cycle Service Level (CSL) or Probability of not having a stock out in a
replenishment cycle. This is usually represented by 1- a
2. Fill rate (fr) level.- probability or proportion of demand filled from stock.
It is usually represented by 1- b
CSL is related with whether there will be a stock out, while fill
rate is related with how many unit we will be short in case of a
stock out.
fr = 1 – Expected shortage per replenishment cycle (ESC)/Q
∞
𝐸𝑆𝐶 = න 𝑥 − 𝑅 𝑓 𝑥 𝑑𝑥 R = DL + ss
𝑥=𝑅
For a normal distribution:
−( x − )
2
1 x − DL
f ( x) = e 2 2
dx = L dz Z=
2 L
𝑠𝑠 𝑠𝑠
𝐸𝑆𝐶 = −𝑠𝑠 1 − 𝐹𝑠 + 𝐿 𝑓𝑠
𝐿 𝐿
Example fill rate (fr)
59 59
−59 1 − 𝐹𝑠 + 35,35𝑓𝑠 =0,6968
35,35 35,35
; 0,1,1 )+35,35 ∗ (𝑛𝑜𝑟𝑚𝑑𝑖𝑠𝑡
59 59
-59*(1-normdist ; 0,1,0 =0,6968
35,35 35,35
𝐸𝑆𝐶 0,6968
𝑓𝑟 = 1 − =1- =0,9993
𝑄 1000
En Excel:
Fs()=normdist(x;0;1;1)
fs()=normdist(x;0;1;0)
Demand when lead time is stochastic
Current Inventory
Position
Demand
Variability
(forecast error)
Leadtime
Variability
Safety Stock with service levels (Normal dist.)
R = dL + Zd L
R = d L + Z L d2 + d 2 L2
Example
R = 363,15
Exercise
Z1−a =
X −
=
( Ld + ss ) − Ld
=
ss
=
100
= 0, 4714
L d L d 2 *150
100 100
ESC = −100 1 − Fs + 212,132 f s = 43,86
2 *150 = 212,132 212,132
ESC 43,86
fr = 1 − = 1− = 0,956
Q 1000
EOQ with planned backorders
▪ What happen if now we want to evaluate if it is convenient to order Q
under a certain level of backorders?
▪ So lets consider now:
Inventory
Ce=cost of excess
inventory (h)
Q Q-b
Cs=shortage cost
T2
T1 Time
b b
T
EOQ with planned backorders
▪ What is my total cost function?
D 1 T 1 T
TC (Qb) = A + ce 1 ( Q − b ) + cs 2 ( b )
Q 2 T 2 T
Q Q −b b Ce=cost of excess
= = inventory (h)
T T1 T2 Q-b
Q
T1 Q − b T2 b Cs=shortage cost
= = T2
T Q T Q T1 Time
b b
T
Then we can substitute in TC(Qb):
D ( Q − b )2 b2
TC (Qb) = A + ce + cs
Q 2Q 2Q
EOQ with planned backorders
▪ If we take the derivatives and some math, we will get the following:
2 AD cs + ce cs + ce
Qb* = = Q*
ce cs cs So what is the
inventory policy?
ceQb * cs
b* = = 1 − Qb *
( cs + ce ) ( cs + ce )
Amount
Ordered (Q)
Sales
observed
Demand
Distribution (D)
Newspapers Shortage
leftover (overage)
Single Period Model
Solution:
Lets compute the Critical Ratio:
CR=(20.000-12.000+3.000)/(20.000+3.000-5.000)=0,6111
px − cQ if x Q
Profit ( Q ) =
pQ − cQ if x Q
E Profit ( Q ) = pE x − cQ − pE Units Short
In the first term I am multiplying my price per unit times the expected
demand, but we do not necessarily sell all the units, so we have to
substract the exptected units short. The second term is simply the cost of
the units ordered Q.
Newsvendor: expected profit
▪ If we have the penalty B and salvage s parameters then:
−cQ + px + s (Q − x) if x Q
Profit ( Q ) =
−cQ + pQ − B ( x − Q) if x Q
( )
E Profit ( Q ) = p ( E x − E US ) − cQ + s Q − ( E x − E US ) − B ( E US )
= ( p − s ) ( E x ) − ( c − s ) Q − ( p − x + B ) E US
In the first term is the expected number of units that i will sell, is the
expected total demand minus the units that i will be short. The first term is
the revenue i get. The second term is just the cost of each ítem i order. The
third term is the salvage value of the Q as the amount that i order minus
the units that i sell. The last term has to do with the penalization that i pay
for the expected units with short.
Newsvendor: expected units short
▪ So the expected demand, units sold and units short is (assuming
the continues case, and the analougous for the discrete one):
x− Q−
Z= =Z = = Norminv ( CriticalRatio )
= ( p − s ) ( E x ) − ( c − s ) Q − ( p − x + B ) E US
Hint: Estimate first, the Expected units short for each case and then we
estimate the total profit.
Answer:
Case 1: $238.600.000
Case 2: $176.389.000
Base-Stock Policy S*
Continuous review policy that assumes that demand is variable, random
and continuous. Lead time is constant and deterministic. Planning horizon
is infinite.
This policy is very simple, and what it simply do is that you place an order
everytime that you face demand. If you have a demand of four units, then
you place an order of four units. However, your order will arrive after a
certain leadtime. So, the question here, is what is your inventory position
(IP), that is what we refer as the Base-Stock or S*?
S*
Inventory on Hand
Base-Stock Policy S*
How can I define S* ? I want to set it so that I cover the demand during
the leadtime. Hence, I can use the critical ratio to determine what is the
service level (LoS) I want to offer:
cs
LoS = P DLT S * = CR =
ce + cs
The Base Stock S* policy is computed as: S * = DLT + Z LOS DLT in which
ZLOS is determined as NORMSINV(CR) or in the tables of a standard
normal distribution
S*
Inventory on Hand
Base-Stock Policy S*
Example
Consider the case in which we have a daily demand that follows a Normal
Distribution with mean 100 and standard deviation 15. Lead time is two
days. Excess cost of inventory is $5 per unit per day and the shortage cost
is $25 per unit per day.
Solution: