Adaptive Supply Chains PDF
Adaptive Supply Chains PDF
Adaptive Supply Chains PDF
UiS Brage
http://brage.bibsys.no/uis/
INTRODUCTION
Stages to
Adaptivity
Adaptive
Stage
Collaborative
Stage
Integrated
Stage
Quality of Timeliness of Depth of
Information
information information information
Density
Visibility
Management
Velocity
Management
Variability
Management
Process
Density
FIGURE 2. HP’s Layered Approach for Real Time Supply Chain Solutions
Initialization Module:
Initial values from current distribution
chain are assigned to parameters
Strategic Module:
Resolve the strategic decisions
Tactical Module:
Resolve the tactical decisions
Y
First iteration ?
N
N
Convergence?
Termination:
The new distribution chain
and location of wholesalers and retailers). After this, from the
output of the strategic module, tactical decisions can be made
(e.g., optimal transportation schedules and routes, inventory control
parameters) in the tactical module.
After determining the tactical parameters, the parameter values are
input into the strategic module again, starting another re-optimization
cycle. Successive design results are compared at the end of each
iteration cycle. When there is no remarkable difference between the
successive iteration results (iteration results converge), the
distribution chain is established.
When market conditions change, there may be changes in the
values of some of the parameters such as demand, transportation
costs, etc. First, for these new values, the tactical module is invoked
(operation related parameters are re-calculated) to adapt to the new
market situation. If the resulting operations-related parameter values
differ significantly from the older values, then the nucleus enterprise
may re-start the iteration cycles until the iteration result converges: at
this point, a new distribution chain should be found that is adaptable
to the new market.
Optimal Design of the Modules
In an earlier paper, Ma and Davidrajuh (2005) presented
complete mathematical details for all these modules (the current
paper is concerned with how an adaptive supply chain can be
developed by realizing these mathematical details). Rather than
repeating these details, a summary is given in the following
subsections.
Initialization Module
Knowledge from past and present
Collaborations (Initial conditions)
Strategic Module
Iterations
Tactical Module
Transportation Inventory
Module Control
(From Module
wholesaler to (Wholesalers,
retailers) distribution
center)
Transportation Inventory
Changing
Module Control
Market
(From Module
condition
distribution (At retailers)
center to
wholesalers)
Submodule for inventory control at retailers. According to
Tijams (1994), there are mainly two types of inventory control
models: periodic review (R, S) model and continuous review (s, Q)
model. For the (R, S) model, an order is placed every R unit time to
raise the inventory level to S. For an (s, Q) model, an order Q is placed
when on hand inven-tory is less than or equal to s. Ma and
Davidrajuh (2005) proposes the use of an (s, Q) model for inventory
control policy.
Submodule for transportation from wholesaler to retailers.
After determining optimal (s, Q) for each retailer, transportation
between wholesalers and retailers can be planned. In practice, the
retailer order quantity is normally small, so it is possible for a vehicle
to serve several retailers in one journey. In such situations, the
following questions are raised:
1. How to cluster retailers?
2. How to determine routes for
vehicles?
Vehicle routing algorithms can answer these questions. At
present, there are mainly three types of routing algorithms:
heuristics (Tijms, 1994; Christofides, 1985; Viswanathan and Mathur,
1997; Chao, 2002), mathematical programming (Popken, 1994;
Wendy et al., 1999), and genetic algorithms (Gabbert et al., 1991;
Chen and Gen, 1996; Chen et al., 1998). Ma and Davidrajuh (2005)
propose genetic algorithms for routing, for ease of implementation.
Submodule for inventory control at wholesalers and at
distribution center. A wholesaler supplies several retailers; adding
together random demands at retailers can generate the random
demand process at the wholesaler. For the distribution center, the
demand parameter determination process is the same one used for
wholesalers.
Submodule for transportation from distribution center to
wholesalers. Normally, the amount of product demanded by a
wholesaler is large. Hence, a vehicle can only serve one wholesaler
in its journey. Thus, there is no routing problem in this transportation
model (if there is a routing problem, then the mathematical model used
for the submodule “Transportation model from a wholesaler to its
retailers” can be used). The unit product delivery cost from
distribution center to a wholesaler can be calculated by summing
together the unit loading cost at the distribution center, the unit
transportation cost from distribution center to wholesaler, and the
unit unloading cost at the wholesaler.
From Design to Development
This subsection describes how the proposed distribution chain can
be developed as a business application, using Service Component
Architecture (SCA). SCA is based on Web services (WWW, 2005).
Beisiegel et al. (2005) and Beatty et al. (2003) explain its concepts.
Subsystem StrategicDecisions
StrategicBusinessProcess
Distribution
Chain Initialization HandleIterativeProcess
Design
Subsystem TacticalDecisions
InventoryWholeSDist
InventoryRetailer
TacticalBusinessProcess
TransportWholeSDist
TransportDistRetail
enterprise data and market data through Service Data Objects (Beatty et
al., 2003); for brevity, these details are not shown.
The second subsystem is the IterativeProcess subsystem that is
responsible for executing the iterative process. The module
component HandleIterativeProcess of this subsystem realizes the
iterative process by starting the iterative process for the first time,
getting the services of the necessary subsystems (StrategicDecisions
and TacticalDecisions) during successive iterations, and finally,
terminates the iterative process when it converges.
The third subsystem is the StrategicDecisions subsystem that will
compute the strategic design values. This subsystem has a component
module called “StrategicBusinessProcess,” which is the realization of
the business logic represented by the strategic module in Figure 5.
The fourth subsystem is the TacticalDecisions subsystem, which is
responsible for computing tactical values. For this purpose, the
subsystem gets the services of four other module components:
Inventory WholeSDist (meaning inventory control at wholesalers and
distribution centers), InventoryRetailers (inventory control at retailers),
Transport-WholeSDist (transport scheduling from wholesalers to
distribution centers), and TransportDistRetail (transport scheduling
from distribution centers to retailers).
According to Figure 6, the subsystem TacticalDecisions gets the ser-
vices of the four other modules locally. However, an alternative design
may place these four module components as remote service references
to TacticalDecisions.
Advantages of the New Iterative-Based Modular System
The main benefits of the proposed distribution chain are discussed in
the following section. The benefits are grouped under the three basic
criteria of adaptive supply chains: visibility, velocity, and variability.
Visibility of key data by all collaborating enterprise is increased
because:
REFERENCES
J. Beatty, S. Brodsky, M. Nally, and R. Patel, Next-Generation Data Programming:
Service Data Objects: A Joint Whitepaper with IBM and BEA, BEA Systems, Inc.
and International Business Machines Corp, November 2003.
M. Beisiegel, H. Blohm, D. Booz, J.-J. Dubray, M. Edwards, B. Flood, B. Ge, O.
Hurley, D. Kearns, M. Lehmann, J. Marino, M. Nally, G. Pavlik, M. Rowley, A.
Sakala, C. Sharp, and K. Tam, SCA Service Component Architectue: Assembly
Model Specification, SCA Version 0.9, November 2005.
A. C. Caputo, F. Cucchiella, L. Fratocchi, P. M. Pelagagge, and F. Scacchia, “Analysis
and evaluation of e-supply chain performances”, Industrial Management & Data
Systems, Vol. 104 No. 7, pp. 546-57, 2004.
M. I. Chao, “A tabu search method for the truck and trailer routing problem”, Comput-
ers & Operations Research, Vol. 29, pp. 33-51, 2002.
R. Chen, and M. Gen, “Fuzzy vehicle routing and scheduling problem using genetic al-
gorithms”, Genetic Algorithms and Soft Computing, Springer, Berlin, pp. 683-709,
1996.
X. Chen, W. Wan, and X. Xu, “Modeling rolling batch planning as vehicle routing
problem with time windows”, Computers & Operations Research, Vol. 25 No. 12,
pp. 1127-1136, 1998.
P. Childerhouse, R. Hermiz, R. Mason-Jones, A. Popp, and D. R. Towill, “Information
flow in automotive supply chains–identifying and learning to overcome barriers to
change”, Industrial Management & Data Systems, Vol. 103 No. 7, pp. 491-502, 2003.
N. Christofides, Vehicle routing: The traveling salesman problem. John Wiley & Sons
Ltd, New York, NY, 1985, pp. 431-448.
R. Davidrajuh, “Realizing a new e-commerce tool for formation of a virtual enter-
prise”, Industrial Management & Data Systems, Vol. 103 No. 6, pp. 434-45, 2003.
P. Gabbert, D. Brown, and C. Huntley, “A system for learning routes and schedules
with genetic algorithms”, Proceedings of 4th International Conference on Genetic
Algorithms, Atlanta, GA, 1991.
Hewlett-Packard (2004). HP Real Time Supply Chain Solution. Hewlett-Packard De-
velopment Company.
i2 (2004) i2: The Foundation for Supply Chain Optimization. i2 Inc.
W. Kozaczynski, “Composite nature of components”, 1999 International Workshop on
Component-Based Software Engineering, http://www.sei.cmu.edu/cbs/icse99/
papers, 1999.
H. Ma, and R. Davidrajuh (2005). An iterative approach for distribution chain design in
agile virtual environment, Industrial Management and Data Systems, Vol. 105 No.
6, pp. 815-834.
H. Ma, and Z. Deng, “Architecture of methodology for distribution chain design”, Pro-
ceedings of International Conference on e-Business, Beijing, 2002.
D. A. Popken, “An algorithm for multi-attribute, multi-commodity flow problem with
freight consolidation and inventory costs”, Operational Research, Vol. 42 No. 2,
pp. 274-286, 1994.
Z. Rahman, “Use of internet in supply chain management: A study of Indian compa-
nies”, Industrial Management & Data Systems, Vol. 104 No. 1, pp. 31-41, 2004.
SAP (2002). Adaptive Supply Chain Networks. SAP White Paper.
A. Silberschatz, G. Gagne, P. Galvin (2005). Operating Systems Concepts. 7 ed., John
Wiley & Sons.
H. C. Tijms, Stochastic Models: An Algorithmic Approach. John Wiley & Sons, New
York, NY, 1994, pp. 51-71.
S. Viswanathan, and K. Mathur, “Integrating routing and inventory decisions in
one-warehouse multi-retailer multi-product distribution systems”, Management
Science, Vol. 43 No. 3, pp. 294-312, 1997.
A. J. A. Wang and K. Qian, Component-Oriented Programming. Wiley-Interscience:
John Wiley & Sons, 2005.
W. Q. Wendy, H. B. James, and P. Iyogun, “An integrated inventory-transportation
system with modified periodic policy for multiple products”, European Journal of
Operational Research, Vol. 115 No. 1, pp. 254-269, 1999.
W. Wu, C. Chiag, Y. Wu, and H. Tu, “The influencing factors of commitment and busi-
ness integration on supply chain management”, Industrial Management & Data
Systems, Vol. 104 No. 4, pp. 322-333, 2004.
World Trade (2004) “Adaptive Supply Chain” Mar 2004. Vol. 17 No. 3; pg. 41, 1 pgs.
World Wide Web Consortium (2005). Web Services. Retrieved on October 15, 2005
from http://www.w3.org.