Inventory Management
Claudio Arbib
Università dell’Aquila __________
Part I: periodic orders
Contents
1. Introduction
Problem outlook
2. Delivery costs 3. Inventory costs
4. Single resource management with periodic orders (EOQ)
fixed demand
regularly variable demand
5. Drawbacks of EOQ
1. Introduction
• Generally speaking,
producing a good involves the use of resources
available in limited amounts
• A resource is normally regarded as a stock
– in input, to feed production during some time span
– in output, to be delivered to customers
• Managing stocks normally asks for the solution of
decision problems
Main cost sources
• transport
• inventory
• workforce
Decisions
• how much is to be purchased/produced
• when one has to purchase/produce
Data & constraints
• process parameters
• demand
• logistic infrastructure
Problem outlook
• Generally speaking,
producing a good involves the use of resources
available in limited amounts
• A resource is normally regarded as a stock
– in input, to feed production during some time span
– in output, to be delivered to customers
• Managing stocks normally asks for the solution of
decision problems
Example A plant makes use of two resource types.
Let b11 = 10 tons b21 = 9 tons be the amount of resource 1 and 2 available on day 1
Process parameters:
The process consumes
a1 = 2 units of resource 1 a2 = 3 units of resource 2 per unit of finished product.
The daily production capacity of the plant is qmax = 2 tons
Problem outlook
• Let xt be the amount produced on day t
• Then x1 < b11/a1 x1 < b21/a2 0 < x1 < qmax
Example A plant makes use of two resource types.
Let b11 = 10 tons b21 = 9 tons be the amount of resource 1 and 2 available on day 1
Process parameters:
The process consumes
a1 = 2 units of resource 1 a2 = 3 units of resource 2 per unit of finished product.
The daily production capacity of the plant is qmax = 2 tons 5
3 2
• The solution x1* = 2 fully exploits the production capacity of day 1
• At the beginning of day 2 stock levels therefore are
b12 = b11 – a1x1* b22 = b21 – a2x1*
6 3
Problem outlook
• Let xt be the amount produced on day t
• Then x1 < b11/a1 x1 < b21/a2 0 < x1 < qmax
• Iterating the procedure one gets
x2 < b12/a1 x2 < b22/a2 0 < x2 < qmax
• The solution x1* = 2 fully exploits the production capacity of day 1
• At the beginning of day 2 stock levels therefore are
b12 = b11 – a1x1* b22 = b21 – a2x1*
6 3
3 1 2
• The max production level is x2* = 1, which does not
exploits the production capacity of day 2
• To maximize production one then has to replenish with at least r = a2qmax – b22 within t = 1
3 5
3 2
2. Delivery costs
• Summarizing, we were able to focus on the following decisions:
– purchase amount r > 3 – purchase day t < 1
• Delivery costs are normally formed by two terms:
• A fixed cost s0, to be paid even if one moves a single gram of resource
• A variable cost which
increases with the ordered amount r of resource, for example
• To further narrowing on this issue we must examine the cost sources
• Let’s begin with delivery costs
• Delivery costs are normally formed by two terms:
• A fixed cost s0, to be paid even if one moves a single gram of resource
• A variable cost which
increases with the ordered amount r of resource, for example
cost
quantity
0 10 20 30 40 50 60
s0
s0 + s1r
ks0+ s1r for (k – 1)d < r< kd
s0 + s1r for 0 < r < d s0 + s1d + s2(r – d) for r > d, s2 < s1
s0 for any r < D
Delivery costs
ks0 for (k – 1)d < r < kd
Delivery costs
• Also specific aspects of the logistic infrastructure
– supplier multiplicity – link capacity
– …
have an impact on costs
• Heterogeneous and/or still non-critical resources can be loaded onto a single vehicle up to its capacity, so
avoiding fixed cost duplication
Supplier 2 R1, R3 Customer
Supplier 1 R1, R2
Supplier 3 R2
Delivery costs
• Also specific aspects of the logistic infrastructure
– supplier multiplicity – link capacity
– …
have an impact on costs
• Heterogeneous and/or still non-critical resources can be loaded onto a single vehicle up to its capacity, so
avoiding fixed cost duplication
Customer Supplier 1
R1, R2
Supplier 2 R1, R3 Supplier 3
R2
freight consolidation
Delivery costs
• It can be then convenient to synchronize replenishment so as to share transport resources and obtain economy of scale
0 25 50 75 100
mon tue wed thu fri sat
resource 1 resource 2 Inventory levels
0 25 50 75 100
mon tue wed thu fri sat
purchase 1 purchase 2
Replenishment (today for tomorrow) four
deliveries
Delivery costs
• It can be then convenient to synchronize replenishment so as to share transport resources and obtain economy of scale
0 25 50 75 100
mon tue wed thu fri sat
purchase 1 purchase 2
Replenishment (today for tomorrow) four
deliveries
0 25 50 75 100
mon tue wed thu fri sat
resource 1 resource 2 Inventory levels
0 25 50 75 100
mon tue wed thu fri sat
resource 1 resource 2
Delivery costs
• It can be then convenient to synchronize replenishment so as to share transport resources and obtain economy of scale
0 25 50 75 100
mon tue wed thu fri sat
resource 1 resource 2
0 25 50 75 100
mon tue wed thu fri sat
purchase 1 purchase 2 Replenishment (today for tomorrow)
Inventory levels
just three deliveries
3. Inventory costs
• However, holding a material good entails a cost, that can be basically viewed at in terms of lost return
• Think to a bank account: every year, interests give you an extra, whose value, say U, increases with
– the permanence of money on account – the official interest rate st
on every day t of the period T
U =
S
stvtbttÎT
1
|T|
0 25 50 75 100
mon tue wed thu fri sat
resource 1 value (€) Stock level and economic value
g0(t)bt
Inventory cost for one resource unit on day t
Inventory costs
• The less stock is maintained in the system;
the less is the material latency in the system
(also called lead time);
0 25 50 75 100
lun mar mer gio ven sab
Scorta 1 Valore Livello e valore economico della scorta
(€)
0 25 50 75 100
lun mar mer gio ven sab
Scorta 1 Valore Livello e valore economico della scorta
(€)
Production or logistic system
resources products
the less are inventory costs.
• Inventory costs therefore depend on:
– the economic value of each resource
– the average amount hold or hidden in the system within the planning period
Drawbacks
• Usually, the inventory cost behavior is opposite to that delivery costs
0 20 40 60 80 100
mon tue wed thu fri sat
inv entory purchase
high inventory level two deliveries
per week
0 20 40 60 80 100
mon tue wed thu fri sat
inv entory purchase
low inventory level three deliveries per week
0 20 40 60 80 100 120
1 2 3 4 5 6
inventory
replenishment total
Drawbacks
• In a given planning period T
– the inventory cost decreases – the delivery cost increases
with the frequency f = 1/Dt of resource replenishment in the period
minimum cost
Replenishment frequency
cost
4. Periodic orders
• In order to model the problem of minimizing the total cost (inventory + delivery) let us begin with assuming that:
– a single resource is consumed at a fixed absorption ratio a0 up to some limit bs (safety stock)
magazine
Stock level (b0= 10)
safety stock bs
consuming process
(a0 = 2)
− the stock level b(t) in the magazine linearly decreases with time
− the average inventory in the interval [0, DT]
corresponds to the blue area = ò0DT b(t)dt stock level b(t)
DT time t
b(t) = b0 –a
0t
stock level DT
time
safety stock bs
Periodic orders
• In order to model the problem of minimizing the total cost (inventory + delivery) let us begin with assuming that:
– a single resource is consumed at a fixed absorption ratio a0 up to some limit bs (safety stock)
Halving the replenishment period (doubling frequency)
=
Halving variable inventory + Doubling deliveries
Dt 2
− purchase occurs periodically with fixed period Dt at a cost of s0 euros per delivery, independently on the amount delivered
s(Dt) = s0× g(Dt) = g0×
ò
0Tb(t)dtT Dt
(E
conomicO
rderQ
uantity)
• One can then express the
inventory and delivery costs in a given time horizon T as:
g 0(t), s(r)
t, r
s0 g0
• Assume that
– the holding cost g0(t) of a
resource unit in a unit time (e.g.
one day) does not change with time: g0(t) = g0
– the cost s(r) of a single delivery does not depend on the amount r delivered: s(r) = s0
Dt b 0–b s= a 0Dt
bs b(t) = b
0 – a
0t
Area = a0Dt2/2 + bsDt
g0 T (a0Dt2 + 2bsDt) g2Dt0T(a0Dt/2 + bs)
(E
conomicO
rderQ
uantity)
cost
time
• There exists a single optimal solution Dt *, independent on T and bs, that can be computed through the first derivative:
• The total management cost equals
s(Dt) = s0× g(Dt) =
ò
T Dt
• One can then express the inventory and delivery costs in a given time horizon T as:
Dt* = Ö 2s0/a0g0
g0T(a0Dt/2 + bs)
C(Dt) = T( Dt s0 + a0g0Dt + g0bs)
2 a0g0 s0
0 = –
2 Dt2
Variable absorption ratio
• In some cases (e.g. perishable goods, obsolescence etc.) the constant absorption rate assumption can be too simplistic.
• Example. In a food industry the amount of usable product decreases with the time t that has passed from its purchase.
Hence the amount necessary to obtain a unit of finished product increases with t, or in other words, the stock absorption rate
is an increasing function of time: a = a(t)
Call r = b0 the stock of fresh semi-finite material acquired at t0 = 0. After t hours, the stock level will be
b(t) = b0 – ta(t)dt
0
Inventory cost
• In general, the perishing rate of a product has an exponential behavior. Suppose a(t) = a0elt, with a(0) = a0 (initial absorption rate) and l > 0. One then has
Call r = b0 the stock of fresh semi-finite material acquired at t0 = 0. After t hours, the stock level will be
b(t) = b0 – a0 (elt – 1) l
stock level
time
Dt
safety stock bs acquired lot b0
b(t) = b0 – a(t)dt
0 t
Inventory cost
Let Dt be the replenishment period. To avoid useless holding, one has to choose the replenishment b0 so that the residual stock at time Dt equals the safety stock bs
b(t) = b0 – a0 (elt – 1) l
stock level
time
Dt
safety stock bs acquired lot b0
bs = b0 – a0 (elDt – 1) Þ
l b0 = bs + (ea0 lDt – 1) l
b(t) = bs + (ea0 lDt – elt) l
• In general, the perishing rate of a product has an exponential behavior. Suppose a(t) = a0elt, with a(0) = a0 (initial absorption rate) and l > 0. One then has
Inventory cost
The inventory cost depend on the period Dt. In order to find it, integrate b(t) within the interval [0, Dt]
stock level
time
safety stock bs acquired lot b0
= bsDt + a0lDt elDt – la02 (elDt – 1) b(t) = bs + (ea0 lDt – elt)
l
0 Dt
0 Dt
a0
ò
(bs + el lDt)dt – al0ò
elt dt• In general, the perishing rate of a product has an exponential behavior. Suppose a(t) = a0elt, with a(0) = a0 (initial absorption rate) and l > 0. One then has
Dt
Inventory cost
• Calculations give the total area as a function of Dt
= bsDt + a0lDt elDt – la02 (elDt – 1)
T Dt
stock level
time
safety stock bs
Dt
acquired lot b0
This area is repeated T/Dt times in the planning period T Area = T bs + eal0 lDt – la2Dt0 (elDt – 1)
Inventory cost
• Calculations give the total area as a function of Dt
stock level
time
safety stock bs
Dt
acquired lot b0
The inventory cost is proportional to the area and to g0 Area = T bs + eal0 lDt – la2Dt0 (elDt – 1)
g(Dt) = g0T bs + eal0 lDt – la2Dt0 (elDt – 1)
Total cost
The total cost is formed by
• the inventory cost
• the delivery costs (proportional to s0
and to the number of deliveries in the plannig period)
• the cost due to product losses by perhishing (proportional to the product value v0 and to
the amount perished during Dt)
amount (= bs) actually in the magazine after Dt hours
stock level
Dt
amount (= b0 – a0Dt) absorbed by the process at a rate a0 after Dt hours amount (= b0 – a0Dt – bs) perished in Dt hours (with b0 – bs = a0(elDt – 1)/l)
g(Dt) = g0T bs + eal0 lDt – la2Dt0 (elDt – 1)
s(Dt) = s0T/Dt
p(Dt) = v0(b0 – a0Dt – bs)
See slide
(elDt – lDt – 1)
v0a0 l
Total cost
The total cost is formed by
• the inventory cost
• the delivery costs
• the cost due to product losses by perishing C(Dt) =
= g0T bs + eal0 lDt – la2Dt0 (elDt – 1) + s0 +
g0Dt v0la0 (elDt – lDt – 1)
Optimal replenishment period
• Like in the case of constant absorption rate, the optimal replenish ment period Dt* is computed through the first derivative of C(Dt)
C(Dt) =
= g0T bs + eal0 lDt – la2Dt0 (elDt – 1) + s0 +
g0Dt v0la0 (elDt – lDt – 1) 0 = a0 elDt – l21Dt2 – elDt + elDt – s0 +
a0g0Dt2 1
lDt
1
l2Dt2 v0 (elDt – 1)
g0T
1 – + + elDt = –1 + l
2s0 /a0g0 l2Dt2 1
lDt
1 l2Dt2
v0 Tg0 v0
g0T
Optimal replenishment period
Optimal replenishment period
0,00 200,00 400,00 600,00 800,00 1000,00 1200,00
1 2 3 4 5 6 7 8 9 10
l = 0,2 a0 = 5 g0 = 2 s0 = 10.000 v0 = 5.000 T = 1.000
1 – lDt1 + + e1 lDt = 1 + ll22sDt0 /a2 0g0 –
l2Dt2
v0 Tg0 v0
g0T
Dt* ~ 3,467
5. Limits of EOQ
• EOQ does not allow us to
1) express the delivery cost as a function of the amount of replenishment and/or of the distance from supplier
s0
s0 for any r < D
assumed by EOQ
ks0 for (k – 1)d < r < kd
more realistic
cost
amount 0 10 20 30 40 50 60
Limits of EOQ
• EOQ does not allow us to
2) express the impact of resource value oscillation in time on inventory costs
assumed by EOQ
resource value
possible value oscillation
resource value
Limits EOQ
• EOQ does not allow us to
3) take into account fluctuations of the resource absorption rate due, for instance, to demand variation
stock level
time
Dt
assumed by EOQ
stock lelve
time
Dt
possible demand fluctuation
Limits of EOQ
• EOQ does not allow us to
4) optimize the simultaneous management of more resource types
optimal replenishment of resource 1
optimal replenishment of resource 2
10 deliveries
optimal replenishment of resource 2
sub-optimal replenishment of resource 1
6 deliveries
Limits of EOQ
• The first of those four limits:
1) express the delivery cost as a function of the amount of replenishment and/or of the distance from supplier
can sometimes be easily dealt with
In fact, for a fixed absorption rate (b(t) = b0 – a0t), if the delivery cost has the form s(r) = s0 + s1r
cost
quantity 0 10 20 30 40
s0
s0 + s1r
s(Dt) = [s0 + s1(b0 – bs)] =
= [s0 + s1a0Dt] =
= s0 + Ts1a0
T Dt
T Dt
T Dt
its sum during T is
Limits of EOQ
• The first of those four limits:
1) express the delivery cost as a function of the amount of replenishment and/or of the distance from supplier
can sometimes be easily dealt with
The cost term Ts1a0 does not depend on period Dt and does not affect the first derivative of the total cost.
Hence, it does not affect the value Dt* of the optimal replenishment period.
Note that this is not true when the absorption is not constant.
cost
quantity 0 10 20 30 40
s0
s0 + s1r
s(Dt) = sDtT 0 + Ts1a0