An application of fuzzy cognitive mapping in optimization of inventory functi...
Complexity Science & Adaptive Supply Networks
1. Complexity Science and Adaptive
Supply Networks:
One Answer to the Challenge
of Sea Enterprise
Major Kelly G. Dobson
Commandant of the Marine Corps National
Fellow
IBM Business Consulting Services
2.
3. Complexity Science 1
Complexity Science and Adaptive Supply Networks:
One answer to the Challenge of Sea Enterprise
Major Kelly G. Dobson
Commandant of the Marine Corps’ National Fellow
IBM Business Consulting Services
Supply Chain and Operations Solutions
May 21, 2003
4. Complexity Science 2
Abstract
The Challenge of Sea Enterprise calls upon the naval services to draw on
the lessons of the business world to make activities such as operating the supply
chain cheaper, more efficient, easier to use, and less manpower intensive.
Additionally, with an eye to future war fighting strategies, the naval services must
transition to anticipatory, more flexible logistics which leverage information and
provide needed support where and when it is most needed. Building upon both
evolutionary and revolutionary examples from the business world, the naval
services have the opportunity to leverage Agent-Based Modeling, aided by
dynamic tracking technologies, into a truly anticipatory, responsive, and adaptive
supply network. This network could not only answer the challenge of Sea
Enterprise, but also would adapt well to form the nucleus of the future joint supply
network.
5. Complexity Science 3
Complexity Science and Adaptive Supply Networks:
One answer to the Challenge of Sea Enterprise
The Challenge of Sea Enterprise
“Among the critical challenges that we face today are finding and
allocating resources to recapitalize the Navy.” (40) These are the opening words
Admiral Clark used to describe Sea Enterprise, an essential element of Sea
Power 21. He went on to say that, “we will make our Navy’s business processes
more efficient to achieve enhanced warfighting effectiveness in the most cost-
effective manner.” (40) Admiral Clark then sums up the means and the goals of
Sea Enterprise: “Drawing on lessons from the business revolution, Sea
Enterprise will reduce overhead, streamline processes, substitute technology for
manpower, and create incentives for positive change.” (40)
One response to the challenge of Sea Enterprise might be simply to
capitalize on the lessons learned from previous experience and incrementally
improve current Navy business processes. However, as recent events suggest,
Sea Enterprise must also promote the development of solutions capable of
supporting emerging military tactics. President Bush, aboard the USS Abraham
Lincoln, noted that: “Operation Iraqi Freedom was carried out with a combination
6. Complexity Science 4
of precision and speed … Marines and soldiers charged to Baghdad across 350
miles of hostile ground in one of the swiftest advances of heavy arms in history”
When discussing the capabilities required to support concepts such as
Sea Basing and other emerging strategies, Vice Admiral Moore, Deputy CNO for
Fleet Readiness and Logistics, and Lieutenant General Hanlon, Commanding
General, Marine Corps Combat Development Command, asserted that the naval
services’ future logistics enterprise must: “leverage information to achieve
efficiencies and provide support at the time and place of greatest impact.” (82)
They went on to say that naval service logistics must “shift toward anticipatory,
responsive logistics.” (82)
Focusing specifically on the supply chain, the challenge of Sea Enterprise
then becomes three fold: First, given the recent glimpse at the future, how does
the supply chain need to change in order to support a broader spectrum of
conflict? Second, what are the lessons from the business revolution? Finally, how
are these lessons applicable to the naval services’ supply chain in order to
reduce overhead and improve effectiveness? Accepting Admiral Moore’s and
General Hanlon’s description as a starting point for the future characteristics of
the naval services’ supply chain, the next question becomes are there relevant
lessons from the business revolution?
7. Complexity Science 5
Key Lessons from the Business Revolution
Complexity Science
One of the applicable lessons from the business revolution is complexity
science. Complexity science is not a new field of study, but a new approach for
studying complex, adaptive systems. Adaptive systems consist of numerous,
varied, simultaneously interacting parts, called agents. The goal of complexity
science is to uncover the underlying principles and emergent behavior of
complex systems, often
invisible using
Birds Flocking
The basic flocking model consists traditional approaches.
of three simple steering behaviors
which describe how an individual
boid maneuvers based on the
positions and velocities its nearby Separation The difference
flockmates:
between traditional
Separation: steer to avoid
crowding local flock mates
methods of analysis
Alignment: steer towards the
average heading of local flock and complexity science
mates
Alignment
involves a shift in focus
Cohesion: steer to move toward
the average position of local flock
mates and methodology.
Reynolds - Boids Traditional methods
Cohesion
rely on cause-and-
effect analysis: by
knowing all the factors that affect a situation, one can predict the outcome of the
situation. Conversely, complexity science holds that behavior is often
unpredictable and analyzing the factors of a situation may not gain the requisite
8. Complexity Science 6
insight. As an example, complexity scientists discuss the steering behaviors of
birds: each individual bird maintains separation, alignment, and cohesion with the
other birds in the flock. (Sidebar) Given these three factors particular to each bird
in the flock, it is unlikely one would predict that the group of birds flock, but that is
what they do as emergent behavior from their steering behavior interactions.
Agent-Based Modeling (ABM)
To capitalize on the insight offered by complexity science, scientists and
corporations have developed Agent-Based Modeling (ABM) which uses
collections of autonomous decision-making entities called agents. Each agent in
the simulation assesses the current situation and makes decisions based upon
its set of rules. The rules themselves are not the essential product of the
simulation; rather the benefit comes from the interactions between agents and
the emergent behavior these interactions produce.
But to glimpse at emergent behavior requires numerous iterations – many
times the number required for traditional simulations – and until fairly recently,
there was insufficient computing power to make these multiple simulation runs in
a cost effective manner. However, because of recent capabilities and product
improvements, analysts can run the simulations hundreds or thousands of times
to develop a distribution of emergent behavior while incurring only nominal costs.
By comparing this behavior to historical data, the analysts validate the accuracy
of the model. Once validated, the model provides something that most traditional
approaches cannot: the ability to model changes to the system, such as
9. Complexity Science 7
obstacles or bottlenecks, and predict how the real system agents would adapt to
these changes. This ability changes ABM from a purely analytical tool to a
predictive tool. ABM offers the potential to accurately model not only the main
elements of the naval services’ supply chain, but all the interactions and
“workarounds” that become such an integral part of the dynamic system. This
ability to extract useful information from agent interactions led Procter and
Gamble (P&G) to use ABM tools in an effort to reduce supply chain inventory.
P&G Case Study: Evolutionary business rules
In 1998, P&G had already achieved a 50% reduction in their inventory,
and was looking for an additional 25% reduction in an effort to control costs.
P&G’s desire to cut inventory seemed to run counter to their need to keep
products such as Tide and Comet on the store shelf. Using ABM, P&G found that
a “seemingly logical policy sending out only full trucks actually created
disruptions along the supply chain … [resulting in] supermarket shelves that were
empty of its key products.” (Bylinsky, 5) Supply chain agents within P&G’s ABM
recognized this self-induced obstruction and correctly modeled a new,
evolutionary approach: “letting some trucks travel with partial loads and making
delivery times more flexible.” (Bylinsky, 5) Not only did the proposed solution
meet predicted results, it exceeded them. After implementing the ABM
modifications, “Procter & Gamble Co. saves $300 million annually on an
investment of less than 1% of that amount” (Anthes, 1)
10. Complexity Science 8
While the return on investment of the P&G example is impressive, similar
results might have been attainable by traditional methods and are evolutionary in
nature. On the other hand, what Air Liquide did with AMB was truly revolutionary.
Air Liquide Case Study: Revolutionary business rules
Air Liquide is a Houston-based industrial gas firm, which supplies “liquid
oxygen, nitrogen, and other gases to
10,000 customers from more than
Radio Frequency Identification (RFID)
300 sources through 30 depots,
This tag, approximately the size of a
shirt button, is:
using 200 trucks and 200 trailers.”
a “smart object” implementation for
item/object tagging that enables end-to-end (Mucha) The scope and complexity
asset awareness. At its core, RFID uses tags,
or transponders that have the ability to store
information that can be transmitted wirelessly in of Air Liquide’s supply chain was
an automated fashion to specialized RFID
readers, or interrogators. This stored daunting with “3 trillion daily
information may be written and rewritten to an
embedded chip in the RFID tag. When affixed to
various objects, tags can be read when they combinations among all its
detect a radio frequency signal from a reader
over a range of distances and do not require
line-of-sight orientation. The reader then sends constituent parts; it took 22 full-time
the tag information over the enterprise network
to back-end systems for processing. (Levine, 3) logistics analysts nearly half a day to
Conceptually, the logistics supply
chain could tag everything from pallets, boxes, generate a delivery schedule that
even down to individual items if their size or
importance demanded. This would provide the would get every product to its
dynamic tracking visibility that so many other
programs seek, but with a much higher degree
of granularity in that each tag is able to know destination on time.” (Mucha) Using
the contents of its attached container. Also, the
cargo would now ‘know’ its destination, required
delivery date, and associated cargo, which in ABM, the truck “agents” were not
turn would allow en route synchronization and
adaptive rerouting when tied with the proper only programmed to find the shortest
ABM system.
routes, but to remember those routes
and compare them with other routes
11. Complexity Science 9
found, optimizing short-cuts and compiling new routes from sections of previously
optimized routes. Most importantly, because of the power of ABM, “just one Air
Liquide analyst is needed to create daily shipping and production schedules
across its numbingly complex supply chain in about two hours.” (Mucha) With the
proven cost savings and overhead reduction of P&G’s efforts and the manpower
reduction and adaptive supply chain of Air Liquide, ABM offers some potentially
revolutionary supply chain management lessons.
Real Time Modeling
The business examples demonstrated ABM’s ability to be both
evolutionary and revolutionary with its approaches to greater supply chain
effectiveness. But even in the Air Liquide example, the information optimized had
some time delay inherent to it – the analyst based the schedule on the known
conditions at a certain point the day prior. While the analyst was able to very
rapidly respond to a bottleneck or an obstacle such as an interstate shutdown,
the information he worked with was not the most current due to this time delay.
What if it were possible to remove that time delay? While a powerful tool in its
own right, one can greatly enhance ABM’s power by supplying the model with
real time data from the actual supply chain. Several technologies, including RFID
(sidebar) offer the potential for dynamic tracking. With the advent of low-cost
computing capacity, Agent-Based Modeling, and dynamic tracking technology,
the naval services have the potential to develop a real time adaptive supply
system.
12. Complexity Science 10
An Illustration of Military Applications
As an example of ABM’s potential when supplied with dynamic tracking
information from the naval services’ supply chain, let us look at a critical node in
an existing supply chain: Sigonella, Italy. Currently, when ships deploy to the
Mediterranean, each group typically leaves an expeditor at Sigonella to rescue
frustrated cargo and ensure that all the cargo destined for the target group
actually makes it to that group. Expeditors rely on ship-to-shore communications
for priorities and a shore-based information system to know what cargo is
inbound or is lost en route for what ever reason. Additionally, the expeditor
maintains a list of priority cargo that takes precedence over other, lower priority
cargo. While the expeditor can be highly effective, he represents a manpower
intensive workaround to a supply chain problem. Additionally, the work of one
expeditor may well prove counter to the work of another, adding greater
inefficiency to the system.
Contrast the expeditor system with an ABM supply chain leveraging
dynamic tracking. In this system, each piece of cargo becomes its own expeditor.
Using RFID as an example, each tag retains knowledge of its host’s contents, its
destination, its required delivery date, and even associated cargo necessary for
this cargo to be useful for the end user. Since this data is stored on the RFID tag,
and not part of a remote system located at Sigonella, the loss of the facility or a
system at the facility does not destroy the required destination of the cargo.
Additionally, by capturing all dynamic tracking data via remote interrogation and
feeding it real time to the ABM, the system constantly learns and optimizes itself,
13. Complexity Science 11
even allowing cargo synchronization with partner cargo en route, relieving
manpower requirements on the end user. This capability alone might make ABM
worth the cost of investment, but this new system really shows its strength when
something goes wrong.
Imagine that some terrorist faction detonates a bomb at Sigonella,
effectively shutting down the node and putting all the expeditors out of action. For
a traditional supply chain to react to this situation, news of the bombing must first
make its way back up the supply chain to the managers, potentially taking on the
order of minutes or as long as days. With the knowledge of the lost node, the
supply chain managers must determine alternate routes and enact those routes.
Then, still in a reactive mode, they must assess the impact that changing to
alternate routes has had on other nodes and adjust accordingly, potentially
routing too much cargo through ports with insufficient capacity. This further
congests the supply chain and potentially leads to individual supply chain
managers developing solutions that create even more congestion.
Now, take that same scenario, but this time using ABM to manage the
supply chain. Because of the dispersed nature of the ABM and the visibility
provided by dynamic tracking, the system could potentially recognize that there is
a problem with the Sigonella node before anyone even finds out that a bomb
went off. Recognizing the impact to cargo in the system, ABM considers the time
sensitive nature of shipments and automatically reroutes critical shipments.
Simultaneously, ABM down-grades the priority of items in the supply chain that
14. Complexity Science 12
depend on other items unavoidably delayed. Finally, ABM, leveraging its
predictive nature and emergent behavior analysis capabilities, anticipates the
impact of routing changes on the entire system, preemptively eliminating the
potential bottlenecks. If cargo is somehow isolated from the master ABM
network, it still retains all of its destination information. Similar to mission specific
orders and commander’s intent, the cargo assesses the situation at the next
node and continues toward its intended objective.
Turning Supply Chains into Supply Networks
Lieutenant General Van Riper, USMC retired, spoke at a conference titled
Preserving National Security in a Complex World in September of 1999. During
his comments – A General Perspective on Complexity – General Van Riper
reminded his listeners, “if you do not cast your net widely and look at places that
traditionally Marines wouldn’t look, you are not going to find the right answers …”
(Van Riper, 179) Using complexity science and Agent-Based Modeling to
manage the naval services’ supply chain would definitely be a wide cast of the
net. However, the question remains: while the potential of cheaper, more
efficient, simpler, and less time consuming alternatives appear successful in the
business world, is it too great a hope to believe that they could produce the same
results for the naval services?
P&G was so impressed with the transformation of their supply chain, they
renamed it a supply network. According to Larry Kellam, P&G’s director of supply
network, “Chain connotes something that is sequential, that requires handing off
15. Complexity Science 13
information in sequence … we believe it has to operate like a network …” where
all the parts are dynamically interacting. The payoff from successfully applying
this new way of thinking about logistics – the challenge of Sea Enterprise – holds
tremendous potential in both cost and effectiveness. By recognizing this potential
to transform how the military thinks about supply, the naval services have the
opportunity to lead the transition to the supply networks needed to properly
support tomorrow’s warfighting requirements. And this technique would adapt
well to cut across the bounds of the traditional service specific supply lines to
form the nucleus of a joint supply network.
16. Complexity Science 14
References
Allan, T. (Consultant). (2003). The Adaptive, Automated Supply Chain. [Microsoft
PowerPoint Presentation]. Tampa Bay, FL: IBM.
Anthes, G. H. (2003, January 27). Agents of Change. ComputerWorld. Retrieved
May 8, 2003 from the World Wide Web:
www.computerworld.com/softwaretopics/erp/story/0,10801,77855,00.html
Bergonzi, C. (2001, September). Thriving in the econosphere. Continental,
75-79.
BiosGroup Complexity Science Overview and Toolkit. (2002). BiosGroup, Inc. 4.
Bush, G. W. (2003, May 1) [Full text of speech aboard the USS Abraham
Lincoln]. Washington Post on the Web. Retrieved from the www:
http://www.washingtonpost.com/wp-dyn/articles/A2627-2003May1.html
Bylinsky, G. (2000, November 27). Look who’s doing R&D. Fortune Industrial
Management & Technology. [Excerpt] 5.
Clark, V. (2002, October). Sea Power 21. Proceedings, 33-41.
Giordano, A. A. (2003, April). Make the supply chain combat ready. Proceedings,
40-42.
17. Complexity Science 15
James, G. E. (1996). Chaos Theory: The essentials for military applications.
Newport, RI: Naval War College Press.
Levine, R. (Director of Emerging Business Technologies). (2003, February 24).
Smart Chip & Automated Technology Solutions from IBM. [Microsoft
PowerPoint Presentation]. Chicago, IL: IBM.
Magruder, C. B. (1991, May 1). Recurring Logistic Problems As I Have Observed
Them. Washington, DC: U.S. Government Printing Office.
Marine Aviation Weapons and Tactics Squadron One. (2000, May 5). A Marine
Expeditionary Brigade in 2010: An analysis of operational potential and
logistical capabilities.
Moore, C. W., Hanlon, Jr. E. (2003, January). Sea Basing: Operational
Independence for a New Century. Proceedings. 80-85.
Mucha, T. (2002, November). The wisdom of the anthill. Business 2.0. Retrieved
May 14, 2003 from the World Wide Web:
http://www.business2.com/articles/mag/print/0,1643,44528,00.html
Reynolds, C. (1995, June 29). Boids (Flocks, Herds, and Schools: a Distributed
Behavior Model). Retrieved May 13, 2003 from the World Wide Web:
http://www.red3d.com/cwr/boids/
Roston, E. (2001, May). Nature’s bottom line. Time Bonus Section: Your
Business, pp. Y9, Y10.
18. Complexity Science 16
The making of a futurist: an interview with Simon Ellis. (2003, January 1).
[Interview with Simon Ellis]. Supply Chain Management Review. Retrieved
May 8, 2003 from the World Wide Web:
http://www.manufacturing.net/scm/index.asp?
layout=article&articleid=CA276608&text=agent
Van Riper, P. (1999, September 13). A general perspective on complexity.
[Address to Preserving National Security in a Complex World Conference,
Cambridge, MA. September 12-14, 1999]. Conference Summary
brochure.
Waldrop, M. M. (1992). Complexity: The emerging science at the edge of order
and chaos. New York: Touchstone.