Killingsworth, William with J. Chris White and Robert Sholtes, "Designing and Managing the Virtual Enterprise - A Breakthrough Application for System Dynamics Using SimBLOX and scmBLOX", 2011 July 24-2011 July 28

Online content

Fullscreen
Designing and Managing the Virtual Enterprise
A Breakthrough Application for System Dynamics Using SimBLOX and scmBLOX

William R. Killingsworth
Executive Director, MIT Forum for Supply Chain Innovation
Massachusetts Institute of Technology
77 Massachusetts Avenue, Room 1-176
Cambridge, Massachusetts 02139

billk@ mitedu

Director, Office for Enterprise Innovation and Sustainability
University of Alabama in Huntsville
301 Sparkman Drive
Huntsville, Alabama 25899
256.824.4434
william.killingsworth@ uah.edu

J. Chris White
President, ViaSim Solutions
519 IH-30, #247
Rockwall, TX 75087
972.722.9255
icwhite@ viasimsolutions.com

Robert Sholtes
President, SimBLOX
14413 Autumn Crest Road
Boyds, MD 20841
301-219-3990

rsholtes@ SimBLOX.com
Abstract

Almost everything we use today is manufactured by a virtual enterprise composed of
hundreds of companies. These large distributed systems have led to numerous
problems and challenges across multiple industries. The need is great for an analytical
technique to examine the performance of a large-scale virtual enterprise. System
Dynamics has been successfully used to model these large enterprises and assess the
impacts on system behavior of changes in demand and various parameters. These
large-scale enterprise models, however, are complex and time consuming to build and
are difficult to restructure. For enterprise management, the ability to reconfigure the
network of companies in response to external forces is critical, and models of the
enterprise must have similar flexibility and rapid re-configurability. Using System
Dynamics agent models of factories, distribution centers and customers, scmBLOX
uses drag and drop features that enable fast construction of enterprise models and
rapid assessments of alternative enterprise structures. Replacement of a make-to-stock
factory for a make-to-order factory or the addition or elimination of distribution centers
can be quickly evaluated. On-going research is focusing on the interplay between
enterprise structure and performance, the development of additional agent models and
new features for current agent models, and the assessment of optimization strategies
such as push-pull boundaries within the global virtual enterprise.

Designing and Managing the Virtual Enterprise
A Breakthrough Application for System Dynamics Using scmBLOX

Background

Almost everything we use today is manufactured by a virtual enterprise - a large
distributed network of companies that provides parts, components, and sub-assemblies
to a final manufacturer or integrator. It is not unusual for hundreds, even thousands, of
companies to be involved in the production of a product by a large virtual enterprise.
Many of these virtual enterprises span the globe. Consider the Boeing 787 Dreamliner.
The wings are made in J apan, the wing tips in Korea, the landing gear in the UK, the
horizontal tail in Italy, and on and on. The only major part of the plane made in Boeing's
“home” state of Washington is the vertical tail. (Avery, 2007) Automobiles are also
produced by widely distributed virtual enterprises. The Hyundai Genesis, a “Korean”
car, has a transmission built by ZF, a German company, an instrument cluster by
Continental, another German company, and fasteners by TRW, an American company.
(Automotive News)

The virtual enterprise has arisen largely as a result of outsourcing, the underlying
strategy that has led the evolution, if not revolution, away from the highly integrated
vertical firm to a widely distributed network of suppliers. The Ford Rouge plant, located
near Detroit, is a classic example of a highly integrated vertical enterprise. Developed
between 1917 and 1928, the Rouge was an automotive complex that transformed ore
into an assembled automobile. (The Henry Ford) Henry Ford’s idea was to achieve a
continuous, nonstop process from raw material to finished product.

The Rouge plant contained ore docks, steel furnaces, coke ovens, rolling mills, glass
furnaces and plate-glass rollers. Each day, workers smelted more than 1,500 tons of
iron and made 500 tons of glass. Henry Ford's ultimate goal was to achieve total self-
sufficiency by owning, operating and coordinating all the resources needed to produce
complete automobiles. Ford Motor Company owned 700,000 acres of forest, iron mines
and limestone quarries in northern Michigan, Minnesota and Wisconsin. Ford mines
covered thousands of acres of coal-rich land in Kentucky, West Virginia and
Pennsylvania. Ford even purchased and operated a rubber plantation in Brazil.

Today, Ford's strategy seems somewhat hard to believe, and, ironically, both grandiose
and quaint. The modern MBA and the Wall Street analyst would decry the capital tied-
up in low-margin operations, the failure to focus on core competencies and the payment
of high manufacturing wages in an urban American setting. The highly vertically
integrated firm has now become the virtual enterprise.

The motivation typically cited for the rise of the virtual enterprise is the drive for lower
costs, but in reality, the motivations for outsourcing are much broader. These reasons
include:

Reduce and control operating costs

Improve company focus on core competencies

Gain access to world-class capabilities

Free internal resources for other purposes

Gain control of a function that is time-consuming to manage or is out of control
Insufficient resources are available internally

Reduce research and development investment

Share risks with a partner company

Grow sales in new and emerging markets through local participation

As pointed out by Chesbrough and Teece, incentive and responsiveness give the virtual
enterprise its advantage. (Chesbrough and Teece, 1996) “Virtual companies coordinate
much of their business through the marketplace, where free agents come together to
buy and sell one another’s goods and services; thus virtual companies can harness the
power of market forces to develop, manufacture, market distribute, and support their
offerings in ways that fully integrated companies can’t duplicate.” Because of the
multiple motivations and advantages, outsourcing became a rapidly growing, if not
fashionable, phenomenon. As companies began the process of outsourcing, however,
some researchers began to raise warning signs and provide guidance on strategies and
identifying and managing risks. In 1993, writing in the MIT Sloan Management Review,
Stuckey and White described a number of situations in which outsourcing was very risky
to a company. (Stuckey and White, 1993) These risks were related to the design and
production of both the components and system being produced. Figure 1 summarizes
their strategy guidance. The results stress the danger of outsourcing when the situation
involves integral component and system architecture and new or emerging production
processes.
Component
Production

Do Not Outsource
When the Component

Se tek eine Designed Will Be
Design in a New or

Do Not Outsource
When Component Q: When
System HL 0 System Will Be
Assembly itiirac : Hi Produced

_ Ina New or Emerging
‘Components Assembly Process

Figure 1. Outsourcing Related to Product and System
Design and Production

In the Best of HBR in 1996, Chesborough and Teece, analyzed outsourcing from the
perspective of the technology and whether capabilities existed outside or needed to be
developed. They identified considerable risk involved with outsourcing an integrated
technology in which the capability had to be developed. In that case, vertical integration
was much preferred. Their guidance is summarized in Figure 2.

Capabilities  Capabilitics Must
Exist Outside Be Developed

Outsource, Ally With
Autonomous Do Not Technology
Or Stand-Alone Vertically Developer or

Technology Integrate Bring Technology
In-House

Systemic or URN
5 ; y grate,
Integrated Technology Bringing Technology
i ging g)
Technology Provider Development
With Caution In-House

Figure 2. Outsourcing Related to Autonomous or Integrated
Technology and the Presence of Capabilities
In 1999, Charles Fine addressed outsourcing in the book Clockspeed. (Fine, 1999) His
strategy guidance identified the best outsourcing opportunity as a stand-alone or
modular item with dependency for capacity only, not knowledge. The worst outsourcing
situation was a systemic or integral item and being dependent for both knowledge and
capacity.

Dependent for Dependent for
Capacity Only Knowledge & Capacity

A Potential

Autonomous, Best Outsourcing Trap —
Stand-Alone, or Outsourcing Consider Vertical
Modular Item Opportunity Integration
Systemic or Can Live Situa
Integral Item With Should V

Outsourcing

Figure 3. Outsourcing Related to Autonomous or Integrated
Items and the Dependency for Knowledge and Capacity

In general, these strategy guidelines have been widely ignored. Risky, difficult and
innovative components and sub-assemblies are routinely outsourced as are items
requiring new production techniques. As one would expect, this has led inevitably to
numerous problems and challenges across multiple industries. It would seem that all of
the benefits of outsourcing are expected to be realized, and all of the risks are ignored.
As Stuckey and White observe, “In all cases, decisions to integrate or disintegrate
should be analytical rather than fashionable or instinctual.”

Modeling the Virtual Enterprise Using System Dynamics

The obvious question, of course, is how does one analytically examine the likely
performance of a virtual enterprise that may involve many companies. System
Dynamics is certainly an appropriate technique for analyzing the dynamic performance
of a virtual enterprise. These large networks of companies function as complex multi-
tier, multi-channel supply chains. System Dynamics has been used to analyze supply
chains from its very beginning as a modeling and simulation tool for policy analysis.
Forrester’s (1958) groundbreaking article in the Harvard Business Review demonstrated
fundamental supply chain dynamic behavior such as how small changes in retail sales
and promotional activity can lead to large swings in factory production, i.e., the so-called
bullwhip or Forrester effect. Forrester (1961) also included a supply chain model and
demonstrated various modes of behavior. Forrester’s models included factory,
distribution and retail tiers in the supply chain but no suppliers to the factory. More
recently, Sterman (2000) has addressed supply chains with several models and case
studies. Again these are forward looking supply chains from factory to customer with
perhaps a single supplier. Huang and Wang (2007) addressed the bullwhip effect in a
closed loop supply chain using a simple model based on Sterman’s (2000) structure.
Simchi-Levi (2008) and Lee (1997) address bullwhip from an analytical perspective.
Schroeter and Spengler (2005) addressed the strategic management of spare parts in
closed-loop supply chains. Angerhofer (2000) presents a thorough discussion of
system dynamics modeling in supply chain management. Killingsworth, Chavez, and
Martin (2008a and 2008b) addressed an extended supply chain with multiple supply
channels and multiple tiers using System Dynamics. This model incorporated twenty-
seven manufacturing facilities and five distribution centers.

The overall supply chain system modeled by Killingsworth (2008a) is shown in overview
in Figure 4. This supply chain extends from raw material to final customer. Demand
arises from four regions of the world. Each region has an inventory, and these regional
inventories are replenished from a central distribution inventory thus totaling five
distribution centers and inventory control points. Supply of parts comes from three
sources: production of new items, commercial repair of returned items, and the potential
for repair at a government facility. Each type of production requires that a number of
parts be integrated into the major sub-assembly. In general, the repair and overhaul
process requires fewer component parts than new part production.
ee Ordering Process [Demand Info_ as
— H :
Supplier Tiers 1 rN H
3.201 7 ry '
¥ C ae Items in '
Component One ITH E pene —> | Use at |» '
New N Region A} | '
Component Two {OB Items >| T H '
: invento Items in] | '
Component Three {OB ions vs avd ria Mesa
Commercial 08 H
Component F our sn ke Overhaul |——} 1 t
N Items in] “J
Component Five TAC Vv > Inventory,» Use at |p
Depot E Region C| |
Component Six Overhaul : '
0 Items in]!
Component Seven LHI R fp] Inventory, _» | Use at |p
Y D Region D
Component Eight LHP 1

Returns for Overhaul
Figure 4. High Level View of Multi-Tier, Multi-Channel Supply Chain

The component parts are each produced through a three-tier supply chain, i.e., three
companies in each channel. With eight sub-assemblies, this results in twenty four
manufacturing facilities plus the final three integrating facilities. The overall supply
process is managed in a feedback fashion by an ordering process often embedded in
an ERP system. This process determines the recommended buys for new items and
the recommended number of parts to undergo repair and overhaul. The supply chain
control system compares current levels of inventory, including due-ins and due-outs,
with anticipated needs to calculate recommended buys and repairs. Since the
procurement of new items and the repair of returned items lead over time to changes in
inventory, the system truly functions in a feedback control fashion to manage the supply
chain. (Killingsworth, Chavez, and Martin, 2008b)

Figure 5 provides a more detailed view of the flows present in the model. With twenty
seven manufacturing sites and five distribution centers, this represents a difficult and
challenging model to construct. Figure 6 illustrates a traditional System Dynamics
model structure for just one of the eight component supply chains. Each company in the
enterprise requires its own production and inventory models as well as logic for placing
orders to its supplier. The model portrayed in Figures 4, 5 and 6 were developed using
Vensim as the simulation language. These large-scale enterprise models are very
complex and very time consuming to build and test, and once built, they are typically
difficult to restructure. For the modern enterprise, however, the ability to reconfigure the
network of companies in response to external forces is critical, and models of the
enterprise must have similar flexibility and rapid configurability. A key goal in enterprise
management is access to dynamic system dynamics models that enable and support
fast and agile management response.

Special Programs

Component One Supply Chain Suppiy \e- =
amponent’ mrgssceess|| com, Ee
ae ee a cere
Orders 1
T a
1 -_ ONHAND REGION A
Component Two Supply Chan A ! in|
Dee -= i Tig
Orders —
H ane
‘Component Three Supply Chain . I Orders I 1
; t
Tet 1 c.5 | R componens ——! 1 (SH!
Jt fel: Inventory bl Customers
‘Component Four Supply Chain | ew tems a) Ke
Component Orders for Tit| !
Ter Ter Tet Sao ic| eee ryt
Ones | 1 ifs REGIONC
' ories "|
1
— ia laa
"
[Geeta eer fe SHE cog — le |e]!
Orders N i)
tT] 4 =] Commercial Repair yt
‘Component Six Supply Chain Orders for yyy REGIOND
Component |B] Components | 1 fiLb yy pl
== >| ordes "Jol tg
Orders ‘0 1 vale
‘ 8 ' 1] R|
‘Component Seven Supply Chain enponithes H it 1
cempanents, | | comments et |
571 seen | |
Orders: Orders for 7 i}
‘Component Eight Supply Chain 1
be ie he -- ae

Return to Overhaul

Figure 5. More Detailed View of Multi-Tier, Multi-Channel Supply Chain
094041
per sensenuey
a
oo, ethos

‘eu 2708009
ey

Figure 6. Detailed View of Model Structure

a

En \ pune a, \
sna 09 an, \ sab amDge \
so umenuat \ f
ae \
i 00 \
nova nates eno vomauo3uey: \
reno 0g — uiiay maps 09 aque apo
sauna gS K men e009

sais my Ke
eusue4av0D :

amyunaeus

\
03 seo

“* pues 0 seu sa ey aves

eg Buys 2009
MEL

as 109
‘oi lea Budus

FFFFFFFFFFRFEFFE
Modeling the Virtual Enterprise:
A New Approach Using System Dynamics and ScmBLOX

Currently available SD modeling tools (e.g., iThink, Vensim, Powersim) are excellent at
developing many types and levels of models. These current tools, however, can be
limiting when it comes to building large-scale, dynamic System Dynamics models. For
very large enterprise models, today’s tools become cumbersome, difficult to manage,
and prone to error due to large amounts of manual manipulation. While some features,
such as arrays in iThink and subscripts in Vensim, are designed to help “replicate”
model structures, these features only allow exact duplication of the same structure in a
“parallel” process (e.g., Product A and Product B moving through identical processes,
but with different cycle times or inventory levels).

As an example, consider the simple Beer Game model. Even though the model
structure for a warehouse is generic, arrays or subscripts cannot be used to connect the
Wholesaler to the Distributor to the Retailer. Three copies of this generic warehouse
structure must be created and manually connected together on the modeling interface.
Each time this replication occurs in building larger supply chains (e.g., 10 warehouses in
a row), the copy-paste-connect process must be used. For complex warehouse
models, there may be several variables that need to be connected each time. The same
is true for multiple manufacturing facilities in a multi-tier manufacturing process. This
creates many opportunities for human mistakes, for example, making an inappropriate
connection. Moreover, as a model gets larger and larger, this process introduces new
issues related to navigation, visibility, and management. If each individual model for a
warehouse is 3 pages, then connecting 10 of them together creates a 30-page model,
plus any additional model structure needed for aggregate bookkeeping, etc. Moreover,
this large model is now largely “static” with regard to structure. If the user desires to
change the order of the warehouses, add more warehouses, or remove some
warehouses, a good amount of additional technical modeling would be required.
Structural changes cannot be done “on the fly” at the macro-level.

SimBLOX is a technology platform that is designed to allow creation and easy
management of macro-level models, including dynamic structural changes at a high-
level. (SimBLOX) It uses a “building block” format in which simulation models represent
the building blocks of a larger, more complex system. In the case of scmBLOX, a
warehouse model is a building block (represented by an icon) that can now be dragged-
dropped onto the model layout and then easily connected with other building block
models (i.e., icons). (ViaSim Solutions) Model visibility and management are greatly
improved while still keeping the underlying model structure desired. In SimBLOX
terminology, the building block models are called SimBRIX and are represented by
icons as shown in Figure 7.
Simulation “agent” model

E-Je] I r

SimBRIX “icon”

gp

Factory

Figure 7. Agent Model for Factory

The technology platform SimBLOX has been used to develop scmBLOX to enable
modeling of large-scale supply chains and virtual enterprises. Currently, scmBLOX
contains four SimBRIX or agent models:

> MTO Make-to-Order Factory
MTS Make-to-Stock Factory
Bap we-oc Warehouse / Distribution Center

& Customer Customer

Using scmBLOX, almost any number of factories and warehouses can be linked
together to create multi-tier and multi-channel enterprise models. scmBLOX is still
undergoing development and additional agent models are in development as well as
additional features being built into the current agents.

Figure 8 illustrates the primary relationships in the Make-to-Order factory. In this agent
System Dynamics model, fulfillment of customer orders is enabled by production. There
is no inventory of product. If for some reason, production lags behind orders, a backlog
of orders builds-up and influences the level of production. Average orders drive two
factors: production capacity and orders to suppliers for material. For example, if
average orders are on a growing trend, then production capacity would tend to increase
in order to meet demands. Similarly, orders to the supplier would also increase.
Figure 9 illustrates the primary relationships in the Make-to-Stock factory. In this agent
System Dynamics model, fulfillment of customer orders is enabled by inventory. If for
some reason inventory is unable to meet current orders, a backlog of orders builds-up.
Average orders drive three factors in this agent: production capacity, orders to suppliers
for material, and a value for the target inventory. In the Make-to-Stock agent, if
average orders are on a growing trend, then the target inventory would tend to increase
in order to meet demands.
Figure 10 presents a high level view of the model structure for the Warehouse/DC agent
model. The Warehouse/DC is managed to maintain a target inventory based upon
average orders. If, for example, average orders are increasing, then the target inventory
would also increase to enable meeting of demands. If inventory is unable to meet
current orders, a backlog develops that is worked-off over time as the inventory is

replenished.

Figure 10. Overview of Warehouse/DC

Note: Diagram Is Intended to Show Primary Relationships and Does Not Include
All Variables Such as Time Delays, Averaging Periods, otc.

> Inventory ——> Fulfill Orders

| _ to Customer
Recelve Material Backorders \
from Suppliers ~~ Order Rate
From Customers

| VA

Order Rate Target Average
¢_—_—. —
To Suppliers Inventory Order Rate

ScmBLOX is structured for the user with drag and drop functionality. The icons are
simple dragged onto the model layout area and then links are clicked and dragged to
connect the icons. Figure 11 presents a screen shot illustrating the drag and drop

process.
Figure 11. scmBLOX User Interface

As may be seen in Figure 11, the icons for MTS (Make-to-S tock), MTO (Make-to-Oder),
WH/DC (Warehouse/Distribution Center) and Customer can be dragged into the model
field and given specific names. It is important to note that for each agent dragged into
the model structure, there are many input parameters that must be specified. These
input assumptions such as time lags, desired inventory coverage, etc. are on the right
side of the screen for user input.

A very important feature of scmBLOX is that for the manufacturing portion of the virtual
enterprise (as opposed to the distribution part of the enterprise), the structure is driven
and determined, just as in the real world, by the Bill of Material (BOM) of the product.
That is, the components that are manufactured and integrated into sub-assemblies
which are then integrated into the final product not only map into the BOM but the
structure of the virtual enterprise as well. Figure 12 illustrates the user interface for
specifying the BOM. As with the other agents, there are input assumptions that are
specified for each component and sub-assembly.
Figure 12. Bill of Material (BOM)

a a!
Cee

o_o e
=

‘Subeystem 2

build BOM |

aa

Using these tools, a model of a very complex virtual enterprise and associated supply
chain can be easily developed. Figure 13 illustrates such a model. Note that customers
are not only present for the final product but also exist for the components and sub-
assemblies. This is important because in many situations, components and sub-
assemblies are used in multiple products, and developments in one product line can
spell-over and adversely affect other product lines.
Simulation and Analysis Using scmBLOX
Base Case Analysis

Figure 14 presents the Bill Of Material (BOM) for an example product. The product
system is composed of two subsystems. Subsystem 1 is made from two parts: Part 1A
and Part 1B, and Subsystem 2 is also fabricated from two parts: Part 2A and Part 2B.
Figure 15 presents the Base Case structure for the supply chain of the system. As may
be seen in Figure 15, there are two customers: one customer for the complete system
and one customer for subsystem 2. The demands of these customers are independent.
The system is assembled in a Make to Order facility. The system factory receives
subsystem 1 from a Warehouse/Distribution Center and receives subsystem 2 directly
from a Make to Order factory. The MTO factory for subsystem 2 also ships directly to
the Customer for subsystem 2. Subsystem 1 is assembled at a Make to Stock factory.
Two MTS factories supply Part 1A. A single MTS factory supplies Part 1B. The MTO
factory for subsystem 2 receives Part 2A from two MTS supply factories and receives
Part 2B from a single MTS factory.
Part 1A

Subsystem 1

Ee atl
Subsystem 2

Part 2B

Figure 14. Bill of Material for System

Bi og

Bill of Material MTS1- Part 1A

MTS2- Part 1 iS = : “8
—— MTO- System Customer - System

MTS - Part 1B

3P>———

MTS - Part 2B

Customer - Subsystem 2
MTS1 - Part 2A

MTS2 - Part 2A

Figure 15. Base Case Supply Chain
Each factory has a broad array of input parameters that specify its costs and operations.
There are five primary categories of input assumptions for each factory:

Costs - Delivery
Costs -- Production
Costs - Sourcing
Finance
Production

The assumptions are different for the MTS and MTO factories since the MTS factory
must have assumptions regarding inventory and its management. Figure 16 presents
as an example the production assumptions for the MTS factories.

Production

Average Yield (Edit value...)
Days of Finished Goods Inventory Desired | 2

Days of Raw Material Inventory Desired 2

Days to Add Capacity 180 Days

Days to Remove Capacity 180 Days

Firm Forecast (Edit value...)

Forecast Adjustment (Edit value...)

Forecast Averaging 20 Days

Forecast Method MovingAverage

Order Backlog Adjustment 0.05

Order Distribution (Edit value...)

Planned Production Capacity Changes (Edit value...)

Production Capacity Calculation Method Calculated

Push vs. Pull Pull

Raw Material Order Adjustment Time 0.05

Shipping Details (Edit value... a
Specified Production Capacity (Edit value... 7

Specified Target Finished Goods Inventory | (Edit value...
Specified Target Raw Material Inventory (Edit value...
Target Finished Goods Inventory Calculati... | Calculated
Target Raw Material Inventory Calculation... | Calculated

LS: Definition x

Figure 16. Production Input Parameters for MTS
Production

Average Yield (Edit value...)
Days to Add Capacity 180 Days
Days to Remove Capacity 180 Days
Order Distribution (Edit value...)

Planned Production Capacity Changes _ (Edit value.
Production Capacity Calculation Method | Calculated
Shipping Details (Edit value...)
Specified Production Capacity (Edit value...)

m7

Figure 17. Production Input Parameters for MTO

Once the parameters for all of the factories, warehouses and customers have been
entered, scmBLOX enables a wide range of events to be simulated and examined:

Increase or decrease in demand by any customer;
Loss of capacity by any factory;

Increase in delivery or shipping times;

Raw material shortages; and

Change is distribution strategy.

For the Base Case, it is assumed that Customer 1 increases the order rate from 100
systems per day to 125 at day 10 and that this increased demand lasts until day 70 and
then decreases to 80 systems a day for sixty days and then returns to 100 systems per
day. Customer 2 orders a constant 50 subsystems 2 per day throughout the simulation
period. Figure 18 presents the customer orders for the simulations.

Base Case Customer Orders

Multi-Tier 16 March 2011\Customer - System\New Units Ordered
—— Multi-Tier 16 March 201 1\Customer - Subsystem 2\New Units Ordered
140 T ; :

130 + 4
120
110
100
90
80
7o + 4
60 ¢ 4
50
40 + + +

ie} 50 100 150

Time

A. Units

Figure 18. Base Case Customer Orders
Figure 19 presents both the customer orders as well as the customer receipts. As may
be seen, system receipts lag the increase in orders and then exceed the order in order
to reduce backorders. After the system orders decrease, the receipts continue at a
higher level until the backorders are cleared. When the demand increases again, there
is a slight lag in receipts. It is interesting to note the increase in demands for the system
adversely affects the customer for subsystem 2 given that shortages in subsystem 2
reduce deliveries. These types of cross-channel impacts are often unexpectedly
encountered in the real world.

Base Case Customer Orders and Receipts

—— Multi-Tier 16 March 2011\Customer - System\New Units Ordered
—— Multi-Tier 16 March 201 1\Customer - Subsystem 2\New Units Ordered
—— Multi-Tier 16 March 201 1\Customer - System\Total Units Received
—— Multi-Tier 16 March 201 1\Customer - Subsystem 2\Total Units Received
160 T , :

140 4 gh j

120 + | 4

100 4

A. Units

so + J

60 + /\ 4

40 + 4

ie} 50 100 150
Time

Figure 19. Base Case Customer Orders and Receipts

Figure 20 presents the Units Shipped for the Base Case. Key points are that the
Warehouse/DC enables the shipment of subsystem 1 to meet the growth in demand.
However, since subsystem 2 is supplied by a Make to Order factory, shipments lag the
increase in demand. As a result, shipments of the complete system are limited by the
availability of subsystem 2. The shipments of subsystem 1 from the Make to Stock
factory exhibit greater variations because of the corrective ordering process of the
warehouse - this is an indication of bullwhip in the lower tier.

Figure 21 presents the Base Case Inventories. As may be seen, all inventories decline
with the increase in demand, however, inventories of subsystems 2A and 2B are more
or less depleted in an attempt to satisfy demand. This lack of subsystems then restricts
production of the complete system. When demand drops on day 70, inventories begin a
rapid climb due to production in the pipeline but are then brought under control and
begin to be reduced. Finally, when demand increases on day 130, inventories for
subsystem 2A and 2B are once again under pressure.
Base Case Units Shipped

—— Multi-Tier 16 March 2011\WH-DC - Subsystem 1\New Units Shipped
—— Multi-Tier 16 March 2014\MTO - System\New Units Shipped

— Multi-Tier 16 March 2011\MTS - Subsystem 1\New Units Shipped
—— Multi-Tier 16 March 2011\MTO - Subsystem 2\New Units Shipped

250 T
200 + 4
@ 150 4
2
3
< 100 4
50 f |
() ' ,
ie} 50 100 150
Time
Figure 20. Base Case Unit Shipped
Base Case Inventories
— Multi-Tier 16 March 201 1\WH-DC - Subsystem 1\Finished Goods Inventory
—— Multi-Tier 16 March 201 1\MTS - Part 2B\Finished Goods Inventory
—— Multi-Tier 16 March 201 1\MTS2 - Part 2A\Finished Goods Inventory
3500 T ; T

A. Units

Time

Figure 21. Base Case Inventories

Figure 21 presents base case backorders for the system and subsystem. The greatest
increase in backorders occurs at the Make-to-Stock factory for Subsystem 1. This
backlog largely arises due to the incoming orders arriving from the Warehouse/DC in an
attempt to meet its own orders as well as correcting for its drop in inventory. These high
levels of orders tend to create instability at the MTS factory and initiate a bullwhip effect.

Base Case Backorders

—— Multi-Tier 16 March 2011\MTO - System\New Orders Backlog
—— Multi-Tier 16 March 2011\MTO - Subsystem 2\New Orders Backlog
~~ Multi-Tier 16 March 2011\MTS - Subsystem 1\New Orders Backlog
—— Multi-Tier 16 March 2011\WH-DC - Subsystem 1\New Orders Backlog
1600 , r :

A. Units

Time

Figure 21. Base Case Backorders

Alternative Cases of Enterprise Structure
Elimination of the Distribution Center

In an effort to cut costs, many distribution centers have been closed within the virtual
enterprise. This action reduces inventory holding expense, building operational costs
and labor. The downside risks, however, arising from closing a distribution center are
lower levels of customer service and product availability. To examine the impacts of
such a structural change, the Warehouse/DC was eliminated from the enterprise model
shown in Figure 15. The warehouse stored five days of subsystem 1 inventory. Deleting
this agent and establishing the direct shipping link from the MTS subsystem 1 factory to
the MTO system factory required less than a minute of modeling time. The new
enterprise model is shown in Figure 22. Figure 23 presents customer orders and
receipts for the new enterprise assuming the same demand pattern as the base case.
As may be seen, closing the warehouse has no impact on customer service. Thus
costs have been reduced and customer service has not been adversely impacted. The
reason for this outcome is relatively straightforward. The Make-to-Stock factory for
subsystem 1 plus the warehouse for subsystem 1 meant two levels of inventory for
subsystem 1. As a result, subsystem 1 was available for system assembly in both the
base case and the alternative. Assembly of the final system, however, was limited by
the availability of subsystem 2. This simulation makes that constraint very clear.

Faas ogee ,

Bill of Material MTS1 - Part 1A

Pe ——— = ‘ f &
MTS2-Part 1A . 7 e Customer - System

MTS - Subeystem 1 MTO- System

Ig a
MTS - Part 1B
e >)

Sa >) Customer - Subsystem 2
MTS - Part 2B z

F Ps : sy MTO - Subsystem 2
MTS1 - Part 2A

age —

MTS2 - Part 2A

Figure 22. Enterprise with No Distribution Center

Customer Orders and Receipts No DC

—— Multi-tier No DC 16 March 201 1\Customer - System\New Units Ordered
Multi-tier No DC 16 March 201 1\Customer - System\Total Units Received
Multi-tier No DC 16 March 201 1\Customer - Subsystem 2\New Units Ordered
460° Multi-tier No DC 16 March 201 1\Customer - Subsystem 2\Total Units Received

A. Units
88

20 ' ' '
0 50 100 150
Time

Figure 23. Customer Orders and Receipts for Enterprise with No Distribution Center
Elimination of Distribution Center and Addition of Make-to-Stock

One strategy for improving customer service is to produce subsystem 2 in a make-to-
stock facility rather than a make-to-order facility. This would provide an inventory buffer
for subsystem 2 and support final system assembly. This model change is
accomplished by deleting the MTO factory, dragging a MTS icon to that location, and
creating the new flow links. Again, this took less than a minute of modeling time. This
structural change is shown is Figure 24. In the previous two cases, the make-to-stock
factory for subsystem 1 attempted to maintain two days of final inventory. For this
case, that target is increased to five days. A similar target inventory is established for
the new make-to-stock factory for subsystem 2. Figure 25 presents customer orders
and receipts for this case, again assuming the same demand pattern. As may be seen,
the cross impact on the customer for subsystem 2 has been greatly reduced and
service improved. Itis this ability to rapidly assess impacts on performance of structural
sensitivity that makes this approach so valuable to management of the global virtual
enterprise.

fie a
Bill of Material MTS1 - Part 14

ge {aL

—— GP > + a yf P
MTS2- Part 1A BQO — customer System
MTS - Subsystem 1 Gi Sider
MTS- Part 1B
ge ~
MTS- Part 2B
So > >
‘gh MTS - Subsystem 2 — 4
MTS1 - Part 2A Customer - Subsystem 2
a
MTS2-Part 2A

Figure 24. Enterprise with No Distribution Center But with Improved Make-to-Stock
No DC with MTS Customer Orders and Receipts

—— Multi-Tier No DC with MTS 17 March 201 1\Customer - System\New Units Ordered
—— Multi-Tier No DC with MTS 17 March 201 1\Customer - System\Total Units Received
—— Multi-Tier No DC with MTS 17 March 2011\Customer - Subsystem 2\New Units Ordered
Multi-Tier No DC with MTS 17 March 2011\Customer - Subsystem 2\Total Units Received
160 T T T

ge

A. Units
3
8

20 A A A
10) 50 100 150

Time

Figure 25. Customer Orders and Receipts for Enterprise with No Distribution Center and
With Improved Make-to-S tock Management

Conclusions

Almost everything we use today is manufactured by a virtual enterprise - a large
distributed network of companies that provides parts, components, and sub-assemblies
to a final manufacturer or integrator. It is not unusual for hundreds, even thousands, of
companies to be involved in the production of a product by a large virtual enterprise.
The virtual enterprise has arisen largely as a result of outsourcing, the underlying
strategy that has led the evolution, if not revolution, away from the highly integrated
vertical firm to a widely distributed network of suppliers. Risky, difficult and innovative
components and sub-assemblies are routinely outsourced as are items requiring new
production techniques. As one would expect, this has led inevitably to numerous
problems and challenges across multiple industries. It would seem that all of the
benefits of outsourcing are expected to be realized, and all of the risks are ignored. A
great need exists for an analytical technology that enables improved design and
management of the virtual enterprise.

System Dynamics is an appropriate technique for analyzing the dynamic performance of
a virtual enterprise. These large networks of companies function as complex multi-tier,
multi-channel supply chains. System Dynamics has been used to analyze supply
chains from its very beginning as a modeling and simulation tool for policy analysis.
These large-scale enterprise models, however, are very complex and very time
consuming to build and test, and once built, they are typically difficult to restructure. For
the modern enterprise, however, the ability to reconfigure the network of companies in
response to external forces is critical, and models of the enterprise must have similar
flexibility and rapid configurability. A key goal in enterprise management is access to
dynamic system dynamics models that enable and support fast and agile management
response.

SimBLOX is a technology platform that is designed to allow creation and easy
management of macro-level models, including dynamic structural changes at a high-
level. Ituses a “building block” format in which simulation models represent the building
blocks of a larger, more complex system. In the case of scmBLOX, a warehouse model
is a building block (represented by an icon) that can now be dragged-dropped onto the
model layout and then easily connected with other building block models (i.e., icons).
Model visibility and management are greatly improved while still keeping the underlying
model structure desired. The technology platform SimBLOX has been used to develop
scmBLOX to enable modeling of large-scale supply chains and virtual enterprises.
Currently, scmBLOX contains four SimBRIX or agent models: Make-to-Order Factory,
Make-to-S tock Factory, Warehouse/DC, and Customer.

Using scmBLOX, almost any number of factories and warehouses can be linked
together to create multi-tier and multi-channel enterprise models. scmBLOX is still
undergoing development and additional agent models are in development as well as
additional features being built into the current agents.

A very important feature of scmBLOX is that for the manufacturing portion of the virtual
enterprise (as opposed to the distribution part of the enterprise), the structure is driven
by the Bill of Material (BOM) of the product. That is, the components that are
manufactured and integrated into sub-assemblies which are then integrated into the
final product not only map into the BOM but the structure of the virtual enterprise as
well.

An example product BOM was developed and a related virtual enterprise model was
created using system dynamics agents. Simulations were performed to examine the
performance of the virtual enterprise. Rather than conducting parameter sensitivity
analysis as is typically done, scmBLOX allows structural sensitivity assessments to be
rapidly conducted. Addition or elimination of distribution centers can be easily modeled
and performance measured. Revising the manufacturing and outsourcing strategy by
adding factories can also be easily addressed. On-going research is focusing on the
interplay between enterprise structure and performance, the development of additional
agent models and new features for current agent models, and the assessment of
optimization strategies such as push-pull boundaries within the global virtual enterprise.
References

AutomotiveNews, http://www.automotivenewseurope.com/assets/P DF/CA7062082.P DF

Angerhofer, Bernard J ., Marios C. Angelides. “System Dynamics Modeling in Supply
Chain Management Research Review.” Paper presented at the 2000 Winter Simulation
Conference, Orlando, FL, December 10-13, 2000.

Avery, Susan. “Boeing Executive Steven Schaffer is Named Supply Chain Manager of
the Year for the 787 Dreamliner Project.” Purchasing, Oct. 2007.
http://www.purchasing.com/article/CA6489135.html? q=DistributorHechnology+can+rem

ove+costs +from+the+supply+chain+&q=boeing+suppliertcollaboration.

Chesbrough, Henry B. and David J. Teece, Organizing for Innovation: When Is Virtual
Virtuous?, Bestof HBR 1996, Harvard Business Review.

Fine, Charles, Clockspeed: Winning Industry Control in an Age of Temporary
Advantage, Basic Books, 1999.

Forrester, J ay W. Industrial Dynamics. Cambridge: MIT Press, 1961.

Forrester, J ay W. “Industrial Dynamics: A Major Breakthrough for Decision Makers.”
Harvard Business Review (J uly-August 1958).

Huang, Lizhen, Wang. “The Bullwhip Effect in the Closed Loop Supply Chain.” Paper
presented at the 2007 International Conference of the System Dynamics Society and
:50" Anniversary Celebration, Boston, MA, J uly 29 - August 2, 2007.

Killingsworth, William R., Regina K. Chavez, and Nelson T. Martin. ‘The Dynamics of
Multi-Channel Supply Chains for High-Value Government Aviation Parts.” 26"
International Conference of the System Dynamics Society, Athens, Greece, J uly 20-24,
2008a.

Killingsworth, William R., Regina K. Chavez, and Nelson T. Martin. “The Dynamics of
the Government Supply Process for High-Value Spare Parts, 26" International
Conference of the System Dynamics Society, Athens, Greece, J uly 20-24, 2008b.

Lee, HauL., V. Padmanabhan, Whang, Seungjin. “The Bullwhip Effect in Supply
Chains.” Sloan Management Review 3837 (Spring 1997): 93-102.

Shroeter, Marcus, Thomas Spengler. “A System Dynamics Model for Strategic
Management of Spare Parts in Closed-Loop Supply Chains.” Paper presented at the
23" International Conference of the System Dynamics Society, Boston, MA. J uly 12 -
21, 2005.
SimBLOX, www.simblox.com

Simchi-Levi, David, Philip Kaminsky, Edith Simchi-Levi. Designing and Managing the
Supply Chain: Concepts, Strategies and Case Studies. New York: McGraw-Hill Irwin,
2008.

Sterman, J ohn D. “Business Dynamics: System Thinking and Modeling for a Complex
World.” McGraw-Hill Companies, Inc., 2000.

Stuckey, John and David White, When and When Not to Vertically Integrate, Sloan
Management Review, MIT, Spring 1993, Vol. 34, Number 3.

The Henry Ford, http://www.thehenryford.org/rouge/historyofrouge.aspx

ViaSim Solutions, www.viasimsolutions.com

Metadata

Resource Type:
Document
Description:
Almost everything we use today is manufactured by a virtual enterprise composed of hundreds of companies. These large distributed systems have led to numerous problems and challenges across multiple industries. The need is great for an analytical technique to examine the performance of a large-scale virtual enterprise. System Dynamics has been successfully used to model these large enterprises and assess the impacts on system behavior of changes in demand and various parameters. These large-scale enterprise models, however, are complex and time consuming to build and are difficult to restructure. For enterprise management, the ability to reconfigure the network of companies in response to external forces is critical, and models of the enterprise must have similar flexibility and rapid re-configurability. Using System Dynamics agent models of factories, distribution centers and customers, scmBLOX uses drag and drop features that enable fast construction of enterprise models and rapid assessments of alternative enterprise structures. Replacement of a make-to-stock factory for a make-to-order factory or the addition or elimination of distribution centers can be quickly evaluated. On-going research is focusing on the interplay between enterprise structure and performance, the development of additional agent models and new features for current agent models, and the assessment of optimization strategies such as push-pull boundaries within the global virtual enterprise.
Rights:
Date Uploaded:
December 31, 2019

Using these materials

Access:
The archives are open to the public and anyone is welcome to visit and view the collections.
Collection restrictions:
Access to this collection is unrestricted unless otherwide denoted.
Collection terms of access:
https://creativecommons.org/licenses/by/4.0/

Access options

Ask an Archivist

Ask a question or schedule an individualized meeting to discuss archival materials and potential research needs.

Schedule a Visit

Archival materials can be viewed in-person in our reading room. We recommend making an appointment to ensure materials are available when you arrive.