Supply Chain Management
Software Solutions versus Policy Design
J Urgen Strohhecker
SIMCON GmbH Business School of Banking and
Friedrich-E bert-Str. 70 Finance (HfB)
D-68723 Schwetzingen Sternstrake 8
GERMANY D-60318 Frankfurt am Main
E-Mail: jstr@gmx.de GERMANY
1 Why Supply Chain Management?
Since the mid-90s Supply Chain Management (SCM) is becoming
more and more popular. The companies rediscover the optimiza-
tion of planning and controlling the production and logistics net-
works as a key factor of success. In the context of a European
Research Project (DELPHI Project No. 26965) a field survey of
644 companies in Europe shows that only 14 % of these compa-
nies haven't heard anything about SCM and more than *%4 intend
to engage in SCM improvements within the next three years.
As interviews carried out all of the interviewed companies use a
software system supporting production planning and control.
However, more than 90 % are not satisfied with it and mention
significant weak points. Hence, most of the companies seem to
intensify their SCM activities because of their disappointment
with the results provided by their MRP or ERP software system.
They assume that they will solve all or most of their problems by
expanding the existing software solutions with components of
SCM software. Software manufacturer see increasing possibilities
of making money and therefore do everything to encourage their
clients in this belief. As studies of the American Manufacturing
Research Institute show the strategy of the software manufac-
turer is obviously very successful. A market volume of 19.3 Bil-
lions of Euro is expected in 2003 for SCM software - an increase
of more than 700 % compared to 1998.
20.
Expected 15
Market Volume
for SCM 10-
software in
Billion Euro 5.
1998 1999 2000 20001 2002 2003
Figure 1: Expected Market Volume for SCM Software!
However, it seems to be doubtful whether the solely software
solution brings the success the companies hope for. Assuming
that the integration of software systems along all elements of the
supply chain will improve the quality and speed of information
1 AMR Research, Inc, 1999
flows, the result should be lower cycle time and cost. SCM, how-
ever, has a much bigger success potential. If the elements along
the supply chain coordinate their order und delivery policies and
stay with these policies, they should outperform pure software
solutions easily.
To confirm or reject this conjecture a model based system
dynamics study is performed. Simulation experiments are con-
ducted to show the effect of different interventions.
2 Supply Chain Reference Modes and Model
Structure
Regardless whether industry is under investigation most supply
chains show a similar dynamic behavior. Typical for supply
chains are2
¢ oscillations in orders and stocks even if the market demand is
almost constant except small random variations.
¢ increasing amplitudes of oscillations along the supply chain.
¢ high levels of excess inventory at the end of product life cycles.
¢ increasing surplus inventory along the supply chain.
1.4
_ Final Products
1.2) . oh
| v
1.0- 1.4
P Intermediate
0.8 fw Nd M\ Goods 42
‘ fn ”. ir \ : aa Pa
14] “1% "y 1.0
1.2 0.8
| Materials
1.0!
0.811 BSA riaiareasieteemnn Bae eEnieninaaeree |
1945 1955 1965 1975 1985 1995
Figure 2: Oscillation and Amplification in the US Economy
2 See Jay W. Forrester: Industrial Dynamics, 1961, p. 28-29. Peter M.
Senge: The Fifth Discipline, 1990, p. 27-39. John D. Sterman: Business
Dynamics, 2000, p. 743-749.
Figure 2 illustrates oscillation and amplification using the exam-
ple of the US economy.
The system dynamics supply chain model used in the following
investigation is based on Forrester’s work presented in his
“Industrial Dynamics”. Compared to the original Forrester model
the following major extensions are made:
A contractor is added to the supply chain. Therefore the supply
chain comprises four stages: retailer, distributor, factory and
contractor. The contractor supplies the factory with parts for
the product. The structure of the contractor module is very
similar to that one of the factory.
The production capacity of both factory and contractor are mod-
eled as variable. That is why policies for increasing and
decreasing production capacity when demand is changing are
added.
Quality management and quality control have become very
important for companies. Therefore a quality control stage is
modeled as extension to the manufacturing process.
Supply Chain Management has a lot to do with cooperation
and trust. For taking this into consideration some equations
are added modeling the development of confidence in the other
supply chain partners.
When the model is simulated with noisy customer orders as test
input a typical supply chain dynamic behavior is the result.
Figure 3 shows oscillating incoming orders with increasing ampli-
tude along the supply chain.
2
- \
2,000
1,500
1,000
500
WT
0
0 50 100 150 200
Time
orders received by retailer 4 4 4 4 1. units/period
orders received by distributor units/period
orders received by factory units/period
orders received by contractor units/period
Figure 3: Supply Chain Reference Dynamics - Incoming Orders in the Base
Case of Noisy Demand
See J ay W. Forrester: Industrial Dynamics, 1961, p. 21-35 and p. 135-186.
4
Figure 4 shows the dynamics of inventories in the base case of
noisy customer demand. As in Figure 3 there are oscillations with
amplification along the supply chain.
8,000
n fy
6,000
4,000
2,000
~ Y
0
0 50 100 150 200
Time (Periode)
inventory at retailer = = 3 4 a units period
inventory at distributor units/period
inventory at factory units/period
inventory of parts at factory units/period
inventory of parts at contractor 5 units/period
Figure 4: Supply Chain Reference Dynamics - Levels of inventory in the Base
Case of Noisy Demand
The life cycle demand test input is generated by a Bass diffusion
structure.4 The orders received by the retailer show the smooth
course similar to a normal distribution. The average of orders
over the whole life cycle is as in the case of noisy demand 1,000
units per period.
The supply chain reference modes in the case of life cycle demand
are shown by Figure 5 and Figure 6. Again oscillations occur, and
the amplitude is increasing along the supply chain. The levels of
inventory in the supply chain are high at the end of the product’s
life. Especially the inventory of goods in the factory is in period
200 almost as high as at the peak around period 100.
The base case simulations show that the system dynamics supply
chain model is able to produce the characteristic dynamic behav-
ior that one can find in real supply chains. Therefore it seems to
be a useful tool for experiments with different Supply Chain
Management strategies.
4 See Frank M. Bass: A New Product Growth Model for Consumer Durables,
in: Management Science, Vol. 15, Nr. 5, J anuary, 1969, p. 215-227. See also
Peter Milling: Diffusionstheorie und Innovationsmanagement, in: Erich Zahn
(Hrsg.): Technologie- und Innovationsmanagement, Berlin, 1986, S. 49-70.
4,000 \\
W\
0 Lie]
| g"'
1,000 af —<
0 50 100 150 200
Time
orders received by retailer + : + : 1 units/period
orders received by distributor units/period
orders received by factory units/period
orders received by contractor
units/period
Figure 5: Supply Chain Reference Dynamics - Incoming Orders in the Base
Case of Life Cycle Demand
sso id f 4
\\
5,000 } wae
0 ansilieed WY! SS
0 50 100 150 200
Time (Periode)
inventory at retailer = units/period
inventory at distributor units/period
inventory at factory units /period
inventory of parts at factory = - 4 4 4— units/period
inventory of parts at contractor s S Sa 5 units/period
Figure 6: Supply Chain Reference Dynamics - Levels of inventory in the Base
Case of Life Cycle Demand
3 Simulation Results of Supply Chain
Management Practices
Most of the companies offering SCM software and consulting use
catchwords in their marketing like quick response systems, ven-
dor management inventory systems, efficient consumer response
systems or continuous replenishment systems.> There are
however some core concepts behind all these slogans. First of all a
very common SCM practice is to speed up information flows and
order processing activities. Electronic data exchange is introduced
and reduces the transportation delay for orders dramatically.
Secondly most SCM systems try to improve the forecasts of
incoming orders. Therefore several trend extrapolation techniques
are usually made available in the software systems. The goal is to
substitute the wide spread estimates of future orders based on
individual experience. In a third step mechanisms are provided
for exchanging and synchronizing the individual forecasts along
the supply chain.
Having the system dynamics supply chain model described above
it is possible to simulate these common SCM practices. Speeding
up information flows and order processing results in shorter delay
times. It is assumed that the order processing delay
* between retailer and wholesaler is reduced to almost one
quarter,
¢ between wholesaler and factory is reduced to 40 % and
« between factory and contractor is reduced to 2/3.
“TT
“TAA
eu VIN \
5 V
10,000
0 20 40 60 80 100 120 140 160 180 200
Time (Periode)
supply chain inventory units
supply chain inventory - base case units
Figure 7: The Effect of Speeding Up Information Flows on Supply Chain
Inventory in the Case of Noisy Demand
The effect of this simulation experiment on the inventory of the
whole supply chain shows Figure 7. It is obvious that the typical
5 Have for example a look at:
http://www.sap.com/solutions/scm/index.htm
http://www.baan.com
http://www.i2.com
oscillations in inventory are not restrained. There is however a
postponement of the curve. This means that faster order pro-
cessing and transportation results in a quicker response to steps
in customer demand. As Figure 8 illustrates new levels of cus-
tomer demand are communicated faster through the supply
chain. This reduces the oscillation intensity of inventories and
therefore the level of costs.
eae AS 3 = 3
EDA \A ei
600 “
0 25 50 75 100
Time
orders received by contractor oS init period
orders received by contractor - base case a units /period
orders received by retailer units/period
Figure 8: Effect of F aster Order Processing on the Orders Received by the
Contractor in the Case of a Step in Customer Demand
That speeding up information flows and order processing is posi-
tive shows another simulation experiment with life cycle demand
as test input. As Figure 9 indicates the peak in supply chain
inventory is reduced by 25 % through shorter delay times in the
information flows. To a significant less degree the high surplus
levels of inventory are decreased at the end of the product life
cyde.
80,000
60,000 lan
- \
2 (VV EN
» Lert PT
0 20 40 60 80 100 120 140 160 180 200
Time (Periode)
Vid
H
(i
I
supply chain inventory = = “= = ~ = = units
supply chain inventory - base case — units
Figure 9: The Effect of Speeding Up Information Flows on Supply Chain
Inventory in the Case of Life Cyde Demand
More effect on inventories has the optimization of forecasts. In
the base run each member of the supply chain uses a simple trend
extrapolation forecast of the future value of incoming orders
based on its past behavior. The trend extrapolation forecast is
modeled using Vensim’s forecast function with an average time of
24 periods and a horizon of 6 periods. The optimization of these
two forecast parameters - separately for each supply chain stage
- leads to a much better performance of inventories compared to
the reference scenario with life cycle demand. Figure 10 confronts
these two scenarios.
80,000 iN
60,000
C PN
40,000 N \ TN
NY
20,000 N N NS
Laced aaa a
lam
0 bor]
0 20 40 60 80 100 120 140 160 180 200
Time (Periode)
supply chain inventory ~ -- -: “| -: -t “: units
supply chain inventory - base case “ - units
Figure 10: The Effect of Optimized Demand F orecast on Supply Chain
Inventory in the Case of Life Cyde Demand
If all the supply chain members cooperate in the field of forecast-
ing the demand and exchange data the effect of surplus inventory
at the end of the product life cyde can be extremely reduced as
Figure 11 indicates. It is true that the level of inventory is higher
in the early stages of the product life cycle, but no expensive sales
activities are necessary to sell the surplus inventory at the end of
the life cyde. Therefore the total supply chain costs will be less
compared to the base case in spite of higher inventory costs.
6 See Ventana Systems, Inc.: Vensim® 4 Reference Manual, 1999, p. 73.
80,000 iN
60,000 \ a
MN
40,000 Fa ee
a
an,
20,000 Ast | ~!
bP Ps
0 Lt |
0 20 40 60 80 100 120 140 160 180 200
Time (Periode)
supply chain inventory units
supply chain inventory - base case units
Figure 11: The Effect of Optimized and Synchronized Demand Forecast on
Supply Chain Inventory in the Case of Life Cycle Demand
As Figure 12 shows, synchronizing and optimizing the demand
forecast in the supply chain has a positive effect on inventory
oscillations in the case of noisy demand too. Therefore it is really
worth wile to intensify cooperation and data exchange along the
supply chain even if the demand takes an almost constant course.
om LL, ALIN
“AAA BS
mV AVIAN
Y
10,000 iy,
0 20 40 60 80 100 120 140 160 180 200
Time (Periode)
supply chain inventory units
supply chain inventory - base case units
Figure 12: The Effect of Optimized and Synchronized Demand Forecast on
Supply Chain Inventory in the Case of Noisy Demand
All the optimization strategies discussed up to now are however
not able to change the general level of inventories in the supply
chain and to speed up the cycle time (see Figure 13 and Figure
14).
10
30,000 il
25,000 \ }
20,000
2 a
LZ
15,000 . J
10,000
0 20 40 60 80 100 120 140 160 180 200
Time (Periode)
supply chain inventory - szenario 3 -: -: -: -: units
supply chain inventory - szenario 2 units
supply chain inventory - szenario 1 units
supply chain inventory - base case 4 units
Figure 13: The Level of Supply Chain Inventory in the Different Scenarios
Discussed (Noisy Customer Demand)
60
50
40
30 y)
7
20
0 20 40 60 80 100 120 140 160 180 200
Time (Periode)
supply chain cycle time - szenario 3
supply chain cycle time - szenario 2
supply chain cycle time - szenario 1
supply chain cycle time - base case ~
Figure 14: The Supply Chain Cycle Time in the Different Scenarios Discussed
(Noisy Customer Demand)
For a better overall performance of the supply chain, especially
for lower levels of inventories and reduced cycle time, changes in
the policies of all supply chain members are necessary. Optimiz-
ing the decision rule determining the level of desired inventory
results in dramatically lower supply chain inventories and
reduces the supply chain cycle time by one third (see Figure 15
and Figure 16).
11
30,000
20,000
x
at
10,000
0 20 40 60 80 100 120 140 160 180 200
Time (Periode)
supply chain inventory - optimized policy Suni
supply chain inventory - optimized software —— tits
supply chain inventory - base case units
Figure 15: The Supply Chain Cycle Time in the Different Scenarios Discussed
(Noisy Customer Demand)
60
50
40 ra\ A
® 7 "
AM
ana NKa
80 100 120 140 160 180 200
Time (Periode)
20 fh
0 20 40 6
oh pry hw H a
supply chain cycle time - optimized policy
supply chain cycle time - optimized software
supply chain cycle time - base case
Figure 16: The Supply Chain Cycle Time in the Different Scenarios Discussed
(Noisy Customer Demand)
4 Conclusions
Using a system dynamics supply chain model based on F orrester’s
work presented in his “Industrial Dynamics”, different strategies
to improve performance of the supply chain are implemented in
the model and simulated. Assuming that the integration of
software systems along all elements of the supply chain will
improve the quality and speed of information flows, the simu-
12
lation shows lower oscillations of inventories and lower surplus
inventories at the end of a product life cyde. SCM, however, has a
much bigger success potential as further simulations show. If the
elements along the supply chain coordinate their order und
delivery policies and stay with these policies, they will outperform
pure software solutions easily.
5 References
Frank M. Bass: A New Product Growth Model for Consumer Dur-
ables, in: Management Science, Vol. 15, Nr. 5, J anuary, 1969, p.
215-227.
J ay W. Forrester: Industrial Dynamics, 1961.
Peter M. Senge: The Fifth Discipline - The Art & Practice of The
Learning Organization, 1990.
John D. Sterman: Business Dynamics - Systems Thinking and
Modeling for a Complex World, 2000.
Peter Milling: Diffusionstheorie und Innovationsmanagement, in:
Erich Zahn (Hrsg.): Technologie: und Innovationsmanagement,
Berlin, 1986, S. 49-70.
Ventana Systems, Inc.: Vensim® 4 Reference Manual, 1999.
13
CC BY-NC-SA 4.0