Management Decision Support Simulations
for Technology Investment Planning
Thomas Klaue
Visiting Scholar Lecturer
Sloan School of Management, System Dynamics Industrieseminar der Universitat
Massachusetts Institute of Technology Mannheim, Schlo8
Cambridge, MA 02139 D - 6800 Mannheim, W. Germany
Abstract
Today's investment decisions in the production industry require - as
this industry becomes more and more integrated by information sy-
stems - a careful long-range planning. Investment projects have to
be seen within the network of their environment, and their interde-
pendant impacts can be assessed in a systematic investigation, as
part of a Technology Strategy. Furthermore a Systems approach
helps to clarify the complex process of Technology Innovations.
Systems thinking to support the definition of Technology
Strategies
Over the last years some enterprises gained competitive advantages
by shifting successfully to new technologies. Others carried the bur-
den of significant investment expenses, leading to high fixed costs,
without any economic advantages, weakening the company's compe-
titive strength.
However, uncertainties about the economic future will never be
cleared totally,’ but a systems approach will at least give an insight
into the interrelations of the underlying structures, which will re-
spond to the changes provocated by the strategy decision, and help to
understand the system, i.e. the enterprises past, actual and future
behavior. So, a systems thinking based assessment of the decisions
impacts already in the planning stage will help to reduce the finan-
cial risks as far as possible.
SHORTFLEX - a System Dynamics project to assess the
value adding potential of different Technology Strategies
In 1986 the "Verband Deutscher Maschinen- und Anlagenbauer"
(VDMA) and the "Industrieseminar der Universitat Mannheim" (ISM)
agreed to set a project on how to illustrate the economic efficiency
System Dynamics '90 555
of computerized technologies in the manufacturing industry.2 The
intention was to create a method to assess the value of higher pro-
duction flexibilities, reduced inventories and shorter job execution
times.
For that purpose, a System Dynamics model was designed, which
attempted to translate the - so far mostly blue-sky - conjectures
about the value-adding potential of technology innovations into more
practical business numbers.
The basic requirements defined as the model's task are shown in
Fig.1
_—_ - — )
CHARACTERISTIC OF TECHNOLOGY STRATEGY
DECISIONS RESULT IN TASKS FOR ASSESSMENT
METHOD
non-dynamic assessment ‘i
often not sufficient to —_— > pramnd ens |
prove economic efficiency
multidymensional impacts > complex assess:
within the organization ment required
nonlinear relations > non-linearity of
between variables system
important. impacts:
integration of
of hardly measurable -_> “soft variables”
parameters.
holistic assessment of > faedback-loop
entire system: required
Structure
Fig.1: The basic requirement to develop a Technology Strategy - a problem for
System Dynamics
The model should provide a vehicle to generate a technology strategy
that would be successful under possible "uncertainties", which
might range from shifts in consumer demographics to labour unions
enforcing shorter workweeks and governmental measures, i.e. change
in taxation rates. It should propose ways for a company to meet it's
long-range goal and increase it's competitive advantage under which
556 System Dynamics '90
ever conditions turn out to prevail, and finally determine an econo-
mic advantageous and technical possible solution, paying attention
to organizational aspects as well.
The other main issue was to clarify the feedback-relationships
"under the waterline" within the system, for example those between
production schedule and inventory, which are often underestimated.
The attempt to adjust the production to fluctuating order-rates
leads usually to internal amplifications of such external oscilla-
tions within the organization, and results in production schedules
rapidly changing between overtime and shortwork weeks. This avoids
an adequate long-range planning and makes an economic allocation of
financial ressources more difficult.
This long-term effect, that a firm's demand for capital fluctuates
far more than the demand for the goods produced with that capital
stock was first treated formally by Frisch and Samuelson and is
known as the multiplier-accelerator theory of investment, which
has already been investigated and discussed by John Sterman in
several papers.3
In Fig.2 the basic structure of the DYNAMO-model shows that it's
design of different sections, representing the
- Technology section, the
- Market section, the
- Labour section, the
- Materials handling section, the
- Production section, the
- Cost-calculation section, and finally the
- Financial section,
which includes the financial restrictions about the models invest-
ment decisions, such as liquidity has to be guaranteed at all times.
These sections, everyone delineating a model within the model, are
driven by internal feedback-loops representing the innerdepart-
mental decisions and linked together with external loops to the main
model, describing the interdepartmental decision making process.
The model's behavior was validated by empirical studies in the pro-
duction industry, mostly in plants of the machine-tool and automobil
System Dynamics '90 557
=
SECTIONS OF THE SIMULATION-MODEL
| VALUE
ADDING
MATERIAL
HANDL. $C.
. Fig.2: The structure of the DYNAMO-model SHORTFLEX
industry. The presentation of the model convinced many people by
the capability of the method. But despite the performance of the
model, it was considered more as an academic solution, since it
could only be handled by a System Dynamics and DYNAMO-experi-
enced person - a requirement, which so far is not met by too many
managers at all.
STRATECH - a participatory simulation to illustrate the
economic impact of manufacturing innovations on produc-
tion, inventory, capacity and profit
In consequence, the next step was to make the model's behavior
easier to interpret and it's application easier to handle for the user,
with the intention to serve as a meaningful basis for business
decisions.
558
System Dynamics '90
So, based on the original DYNAMO-model, a STELLA-version was de-
signed on a level of higher aggregation, focusing mainly on the
- . materials flow (including inventories), the
- capacity (capital stock) and the
: order flow.
The structure of the STELLA-model is shown in Fig.3, the arrows de-
scribe the information flow to generate the systems decisions,
while the broken lines represent the decisions, which later will be
required by the participant. For the reference-calculation, these
decisions were made by the computer as well.
NE —
MODEL-SECTIONS AND DECISION IMPACTS
|. ORDER FLOW
SYSTEMS OECISIONS
—-——— PLAYERS DECISIONS
nen
Fig.3: The structure of the more aggregated STELLA-version STRATECH
The simulation examined three different technology strategies under
the same external economic conditions:
= Every strategy had to guarantee the permanent supply of all
it's Orders.
- The Desired Production is determined by the anticipated,
smoothed demand in history, taking into account the systems
delays, i.e. the Job Execution Time.
System Dynamics '90 559
- Desired Inventory is calculated on a "savety basis" to ensure
permanent ability to deliver, dependant on the flexibility of
the manufacturing system.
: Capacity Adjustment requires in advance-planning, since it
is characterized by significant delays.
- The Orders are represented by a fluctuating graph, including
an increase in basic demand in period 20.
The three technology strategies are characterized as:
Strategy 1: "low tech"-strategy, characterized by inflexibility
in manufacturing, long Job Execution Times, low
Fixed Costs per capacity unit and relativly high
Variable Costs.
Strategy 2: "middie tech"-strategy, including medium flexibility,
higher Fixed costs per unit and lower Variable
costs than strategy 1.
Strategy 3: "high tech"-strategy, characterized by high flexibility in
manufacturing, small economic batch sizes and short
Job Execution Times, but requiring high investment
expenses. This leads to high Fixed costs per capacity
unit, and relatively low Variable costs, due to
automatisation of the manufacturing process.
To ensure equal chances for every strategy, the basic calculations
will generate exactly the same economic results - the Cummu-
lative Net Profit - if the order-rate would be stable at it's
average level. So, any different results between the strategies are
due to external fluctuations of Orders and their impacts due to
internal amplifications by the system.
The most important results of the simulation are illustrated in Fig.4
through Fig.8. Fig.4 illustrates, how the fluctuations in Orders are
amplified by the system, dependant on the length of the Job Execu-
tion Time. The delayed response - representing the systems infle-
xibility - leads to accelerated amplifications down the supplyline,
which for Strategy 1 result in consequences even in the Capacity
section. Though the maximum order-rate is always below the initial
capacity, the system tends to capacity investment activities.
560
System Dynamics '90
1 capacity
7500.00
$625.00
3780.00
1875.00
0.0
capacity
7500.00
5625.00
‘3750.00
4
3} 1875.00
4
00
1 capacity
7500.00
5625.00
3780.00
1878.00
0.0
pune so AOE
2 prod_start
S orders
4 prod_out
af
Strat.l
ao 28.00 50.00 75.00 100.00
Time
2 prog_start Borders 4 prod_out
Strat.2
0.0 25.00 $0.00 75.00 100.00
Time
2 prog_start 3 orders 4 prod_out
Strat.3
pe PP POS
00 25.00
50.00
Time
Fig.4: Orders, materials flow and capacity
75.00
100.00
In Strategy 2, the effects of fluctuating Orders on the capacity
planning are already smoothed very well. However, the flexible
Strategy 3, which assumes that the Job Exection Time is as short
as the Desired Delivery Delay, illustrates the equity of Orders,
System Dynamics ‘90 561
Production Start and Production Output. In this scenario, ca-
pacity adjustments are not necessary at all. This, in fact helps long-
range planning to be more reliable. Plotted in terms of average Ca-
pacity Utilization, Fig.5 shows the smoothing effects of higher
flexibility in manufacturing.
1 cap_util
1 1,00 ——_" cr
1 0.750 r
1
1
1 0.500 ’ b
) 0.280 r
Strat.l
' ae 0.0 25.00 50.00 78.00 100.00
Time
1 cap_utl
1 4.00 e
a a! \ ;
‘ 0.750 laN f .
1
1 0.500 a
Strat.2
1 0.250 r
1 00
0.0 25.00 50.00 75.00 100.00
Time
1 cap_unl
1 1.00 4
\ A LN
KS i.
1 0.750 /~ ; is
1
1 0.500
1 0.250 Strat.3
1 0.0 '
oo 25.00 50.00 75.00 100.00
Time
Fig.5: Capacity utilization
562
System Dynamics '90
Fig.6 gives an insight, how the delayed response of Production
Start to changing Orders leads to long-range amplifications in the
manufacturing system. This effect will be reduced, dependant on the
decreasing length of the systems time delay - the Job Execution
Time.
1 orders
Y} 1500.00
4} se2s.00
Y} ars20
XY} s975.00
yy a0
1 orders
4} 1200.00
8} se2s.00
Y} srs0.00
Y}sers00
Yo
1 ordere
3} 7500.00
3} 5825.00
43) 3750.00
3} 1875.00
Yo 00
2 prod_start
i 1
yf. i as i 1 b
‘e
Strat.l
a0 25.00 30.00 78.00 109.
Time
2 prog_start
aad
Strat.2 r
00
Cry 28.00 50.00 75.00 109.
Time
2 prod_start
po fn
Strat.3 t
00
a0 25.00 50.00 75.00 109.00
Time
Fig.6: Orders and production starts response of system
System Dynamics '90
The main task - to serve all the Orders in time - is met in all sce-
narios as shown in Fig.7. Backlog represents the Orders due in the
calculation period, which are served by Sales. Except for some very
little "Noise", both graphs are equal. Hence, the ways to reach this
goal are different, depending on the actual strategy.
1 backlog
25000.00
sone
12800.00
| 6250.00
| 00
1 backlog
2500.00
18750.00
1250.00
i 6250.00
00
1 backlog
2500.00
18750.00
12500.00
4
3} 8280.00
Bs
0.0
2 units_in_prog B inventory 4 sales
Strat. rN
. Ly,
00 25:00 $0.00 78.00 100,00
Time
2 unwts_in_prog S inventory 4 sae
t
Strat.2 r
I~. _ wai
be FF I el 8 a 3 et
00 25.00 50.00 78.00 100.00
Time
2 units a_prog 3 inventory 4 sales
Strat.3
Liz. 134124 ZO 24 2124 Lew
ee ee eee ee eerie
00 28.00 $0.00 . 75.00 oe t00.00
Time
Fig.7: Process of serving orders with different strategies
System Dynamics '90
The inflexible Strategy 1 is shipping the Sales basicly out of stock,
which coveres all the expectable fluctuations until the production
catches up. On the other hand, Units in Progres overshooting de-
mand are stored, a strategy which leads to high level average In-
ventories. Strategy 2 is able to reduce Inventories significantly,
while the flexible Strategy 3 is almost able to abondon it's Inven-
tories at all. Units in Progres equals the Orders to be served as
well as the Sales, illustrating the fact that Sales are directly
shipped off the line. 4 cumner_protit
Fig.8: Economic results
2 tot_costs® 3 revenues
3000000.
} S00900.00 2, .
J, 22800000 f
}} 375000.00 r
2
4, 1s00000.0 |
4} 250000.00 r
Strat.1
4, 780000 00 oN
3} 128000.00 S* /~7
ra Qo
00 Lot.
3) 00
a0 28.00 50.00 75.00 100.00
Time
1 cum_-et_protit 2 tot_costs 3 revenue
4, 300c000 0
500c00 00 r
Py AN
1, 2280000 9
} 375000 00 Vv,
4, 'sco00e 0
250000.00 _—
Strat.2 . a”
1, 7s0000 00 ——
125000 00 wa L
¥
s 00 1 1
a} 99 Go ‘ 25.00 50.00 78:00 100.00
Time
1 cum_rat_protit 2 tot_costs 3 revenues
3000000 0
00000 00 c
ry a
}, 22800000
375000 90 2 7 St
> 1
1, 1so0cea 0 a
3} 250000 20 1
~—
Strat.3
4, 750090 90
3) ‘25000 20 s
1
—
4 20 1
90
a0 25.00 $0.00 75.00 109.00
System Dynamics ‘90 565
Finally, Fig.8 provides a general view about the economic efficiency
of the different strategies. None of them is really unsuccessful, but
the advantage of at least the partly flexible Strategy 2 is signifi-
cant. The fact, that the economic result in Strategy 1 tends to fluc-
tuate between losses and profits, while the other strategies seem to
- provide more stable results, may as well contribute to a more solid
basis for long-range planning as the stabilization of the Capacity
Utilization (Fig.5).
MICROWORLD-interface connects STRATECH to the user
This STELLA-model was linked to a MICROWORLD¢ interface, which
is a handy tool to provide a meaningful connection between an
unexperienced - as far as System Dynamics and STELLA is concerned
- user and the model. By cutting off the decision making loops within
the model (Fig.3), the STELLA/MICROWORLD-model is transformed
into a participatory simulation game.
To run the game, the participant has to make three basic decisions,
one on each level of
- The production planning schedule prod_start, representing
the short-range, operational decision.
- The capacity acquisition ordered_cap as a middle-range 2
decision, and the
- “flexibility” of his manufacturing process, described by the
length of job_exe_time, as the long-range, strategic
decision, which will determine his burden of fixed costs.
"@ File ait Game Options Modet
Oveisteas ——————-——>
ye =——
-— recs” ate ji
Reperts
STRATECH me OP
20 poriieipaiery elmalation a
Graphs & Tables
ise a
ores
ordertiow
sl
. @6Grapns OTabdles
Fig.9: Microworld spreadsheet
566 System Dynamics '90
In addition to the external decisions required by the player, the
spreadsheet shown in Fig.9 provides the player with the information
he needs to make his decisions as well as it illustrates the conse-
quences of these decisions.
The information Reports (Fig.10) and Graphs & Tables (Fig.11)
focus on the description of the relationships between manufacturing
flexibility and the oscillation in desired production, actual
production and inventories. They emphasize the crucial impacts of
delays in the system and between investment decisions and their
actual economic effects for the firm.
@ File Edit Game Options Model
OPERATIONS
week:100
orcare
freacatere
prod_out
inventory
@Graphs OTabies
r 7
& File Edit Game Options Model
ACCOUNTING
week:100
sae
‘avenues
tot.costs
cum_net_profit
a
@6raphs OTapies
Fig.10: Reports of MICROWORLD to the participant
System Dynamics '90 567
" @ File Edit Game Options Model .
matfiow Decisions: 2
axe.tine
o wesw reason 3
nrg ;
o ronet predstart sie
P Oraphs & Tables
exp plan
cash few
order flaw
7|| @Grapns OTabies
@ File Edit Geme Options Model
cap_pian
D erdered_eap
© eapaetty
Fig.11: Graphs &Tables show consequences of paticipants decisions
So, this -participatory simulation assesses the
- technological,
: organizational and
- economical
implications of a certain Technolgy Strategy. It may provide - in a
more sophisticated version, which is still in the process of being
developed - a learning laboratory to get a feeling how to manage the
innovation process in the computerized enterprise, and to handle
difficult markets successfully by anticipating their fluctuations
through the application of the appropriate technology.
System Dynamics '90
Gaffney, R.: Systems Thinking in Business: An Interview with Peter
Senge, in: ReVision, Vol. 7, Num. 2, P. 56-63.
Klaue, Th.: Kosten und Nutzen der industriellen Flexibilitét, Baden-
Baden 1990.
Sterman, J.D.: Misperceptions of Feedback in Dynamic Decision Making,
pres. at the 1986 Judgment/Decision Making society meeting, New
Orleans, LA.
Diehl, E.W.: MICROWORLD user's manual, Cambridge, MA. 1989.
