Dynamic
Simulation and Supply chain Management
White Paper
August 2002
__________________________________________________________
Abstract
This paper examines how dynamic computer simulation can be
applied within the field of supply chain management to support supply chain
planning, diagnose problems and evaluate possible supply chain management
solutions, optimize supply chain operations, and mitigate risk factors. The
paper also provides a methodology for building supply chain applications and
identifies the critical criteria for selecting appropriate supply chain
management software. Finally, an example of a supply chain planning model using
GoldSim simulation software is described.
Contents
Introduction .. .1
Typical Applications of Supply Chain Planning Models . 3
Supply Chain Application Design and Development: The Top Down Approach ...4
Selecting Supply Chain Management Software ..6
Example Supply chain Model ..7
Conclusions .10
About the GoldSim Technology Group .. 10
Introduction
Companies face an increasingly challenging marketplace with a growing field of competitors, higher customer expectations, and complex supplier relationships. Increased competition means that companies face the dual challenge of cutting costs while being more responsive to the market. The need to cut costs is driving companies to outsource business operations, minimize inventories, divest underutilized capital equipment and facilities, and in general run as close to the edge as possible. The need to be more responsive to the market drives companies to expand their product lines and increase options, minimize the time to bring new products to market, and quickly modify product delivery rates to match changes in demand.
As competition and complexity has increased, supply chain management and supply chain planning has emerged as an increasingly important issue for companies. The challenge of supply chain management is to identify and implement strategies that minimize costs while maximizing flexibility in an increasingly competitive and complex market. This paper examines how dynamic simulation tools can be used to better understand supply chain dynamics, diagnose problems and evaluate possible supply chain management solutions, conduct supply chain optimization, and mitigate risk factors.
Supply chain dynamics. The term supply chain
generally encompasses the web of interconnected relationships between the sales
channel, distribution, warehousing, manufacturing, transportation, and
suppliers (see Figure 1). Each component of the supply chain is connected to
other
parts of the supply chain by the flow of materials in one direction, the flow
of orders and money in the other direction, and the flow of information in both
directions. Changes in any one of these components usually creates waves of
influence that propagate throughout the supply chain. These waves of influence
are reflected in prices (both for raw materials, labor, parts, and finished
product), flow of materials and product (within a single facility or between
facilities within the supply chain), and inventories (of parts, labor capacity,
and finished product). How these influences propagate through the system
determines the dynamics of the supply chain.
It has been observed in many cases that supply chain dynamics demonstrate cyclical fluctuations and instability. These fluctuations are typically a result of information and communication delays (e.g., orders may be based on inventory data that is several week old) and inertia (e.g., once an order is placed it may be weeks or months before the production rate can be changed). These fluctuations can result in many undesirable and/or costly inefficiencies, including stock outs, obsolete inventories, unfulfilled customer demand, or idled factories. The objective of supply chain simulation is to understand the dynamics of the system and ultimately to identify and evaluate strategies to minimize inefficiencies in the system.
What is dynamic simulation? In this context, the term simulation is defined as the process of creating a computer model (i.e., a representation) of an existing or proposed system (in this case, a supply chain) in order to identify and understand the factors that control the system. Any system that can be quantitatively described using equations and/or rules can be simulated. In a dynamic simulation, the system changes and evolves with time and the objective in modeling such a system is to understand the way in which it is likely to evolve, predict the future behavior of the system, and determine how to influence that future behavior.
Transactional
versus analytical information technology (IT). To understand how
dynamic simulation fits within the realm of corporate IT, it is important to
differentiate between transactional IT and analytical IT (Figure 2). As
discussed by Shapiro[1]
, transactional IT is concerned with the acquisition,
processing, and communication of information regarding the past and present and
is primarily used at an operational level. Most of the enterprise resource
planning (ERP) software and systems implemented over the past 10 to 20 years
fall into the category of transactional IT. In contrast, analytical IT is
concerned with forecasting, decision analysis, and solving problems. Analytical
IT can be divided into analysis software (such as data mining and statistical
packages) and strategy software (such as optimization and dynamic simulation
software). To maximize the benefits of these different types of software, they
should be integrated such that the analytical IT applications utilize the
information provided by the transactional IT applications.
Supply Chain Integration:
Large companies typically have very complex supply chains with
tens or hundreds of distributors, factories, warehouses, transporters, and
suppliers. Although ERP software has greatly increased the amount of current
information, having timely access to information doesnt always result in an
integrated supply chain management system, nor does it tell us much about the
future. It is virtually impossible for the human mind to fully comprehend and
predict the dynamics of a complex supply chain system. Dynamic simulation
software, such as GoldSim, provides the means to incorporate all the data and
dynamics regarding a complex supply chain into a computer model that can be
used to gain a better understanding of supply chain behavior and improve supply
chain integration.
If the future were predictable, it would be relatively straight-forward to design a supply chain that was optimized for that particular future. In reality, however, the future is uncertain, and a well-designed supply chain must be flexible and fully capable of adapting to a wide range of potential futures. Therefore, any simulation of the supply chain must account for all the possible futures by providing a means of incorporating uncertainty into the analysis.
Typical Applications of Supply chain Models
The process of building a supply chain application provides valuable insights and understanding regarding the behavior and characteristics of the supply chain. Beyond this expanded knowledge, however, most models are developed to address particular issues. Types of issues that can be addressed using supply chain simulation generally fall into the following categories:
- Optimization: Optimization usually involves finding the optimal operational guidelines that either maximize or minimize a particular result, such as minimizing costs and/or risks and maximizing profits. Examples of operating conditions that could be optimized include factors such as inventory levels, investment in maintenance, or geographic distribution of warehousing facilities.
-
Decision Analysis: Decision analysis typically involves the
quantitative evaluation and comparison of two or more alternatives. For example, the decision to build a new
production facility could be evaluated by simulating how the supply chain
would be impacted by the additional facility. Alternatively, the analysis
might be focused on comparison of ten different locations for a new
production facility.
- Diagnostic Evaluation: Diagnostic evaluation is typically conducted when the cause of a particular problem is unknown. Supply chain simulation can provide insight into the cause of the problems and facilitate development and evaluation of various supply chain management solutions. For example, a recurring inventory stock-out problem could be investigated using a supply chain model.
- Risk Management: As discussed earlier, supply chain dynamics can be severely impacted by unanticipated disruptive events. Examples include labor strikes, destruction of key facilities due to fire or floods, a major customer or supplier going out of business, political upheaval in a country that provides a key raw material, and so on. Many corporations are attempting to determine how to prepare themselves for such events. Supply chain simulation can provide an important role in helping companies design redundant systems or mitigation plans to minimize the impacts of disruptive events.
-
Project Planning:
- Changes to a portion of the supply chain can result in major disruptions and short-term or even long-term inefficiencies. Examples include outsourcing a major manufacturing process, building a new plant nearer to your major customers, or switching a major supplier. In contrast to decision analysis, which is focused on whether to implement a project, project planning is focused on implementation of the project in a manner that minimizes cost, stays on schedule, and minimizes potential risks.
Supply Chain Application Design and Development: The Top Down Approach
Successful development of a supply chain application (model)
is similar to any IT project in that the greatest benefits are received when
development is planned and executed by knowledgeable technical experts
following a well-practiced methodology. Most successful methodologies include
the steps shown in Figure 3. This general methodology illustrates an iterative
top-down process, rather than a straight-line process.
The primary advantage of the top-down approach is that
greater detail is only added in areas that govern performance of the system
relevant to the issue at hand. In this manner, important results and return on
investment are provided as quickly as possible. The supply chain application
can be expanded in additional areas as the focus shifts to other issues.
Development of the supply chain application typically involves the following steps:
1.
Define
Objectives: Defining the objectives is a critical first step in any
model development process. With regards to supply chain modeling, it is
important to identify how the model will be used. Will the analysis be used to
optimize existing operations and/or to evaluate the addition of new facilities,
suppliers, and channels? Will the analysis focus on dynamics between facilities
or also be used to address operations within facilities? These and many other
questions must be addressed to determine the scope and breadth of the
simulation model. Clearly, defining objectives must involve the project
sponsor, decision makers, and the investigators building the model.
2. Develop the Conceptual Model: Building a conceptual model of the supply chain is probably the most important (and time-consuming!) part of the entire exercise. This phase typically involves input and feedback from many people within the organization and thoughtful investigation of how the different elements of the system relate. Most people find that the exchange of information and ideas while formulating the conceptual model results in valuable insights and better understanding of the supply chain system prior to even building a simulation model. In addition, the conceptual model-building phase provides a critical opportunity to get buy-in and support from a broad range of constituents within the organization (e.g., operational managers, supply chain experts, senior management).
Although some simulation software packages (such as GoldSim) provide a graphical user interface that can be used to document the conceptual model, a whiteboard is the most common design tool for this phase. The conceptual model should identify all the important components of the model (production facilities, suppliers, distributors, inventories, etc.) the critical relationships between these components (rendered as influence diagrams), the flow of materials and information, and the critical input data and metrics that govern system dynamics and drive decision-making.
- Build the Computer Model: For complex systems, it is preferable to begin building the computer model during development of the conceptual model. The computer model allows the investigators to test components of the conceptual model and demonstrate relationships, thereby providing valuable insight and perspective during development of the conceptual model. By using a top-down hierarchical model such as GoldSim, the modelers are able to develop the overall structure of the model while leaving the details to be developed at a later stage. As the conceptual model is expanded and filled out, additional details are added to the computer model until a complete working model is developed.
- Verify and Calibrate: Verification and calibration of the model should begin shortly after starting model development. As components of the computer model are completed, they should be tested to ensure that the model results are consistent with expected results and/or actual experience. Once the computer model is complete, model results should be compared with historical results to verify that the model captures the behavior of the supply chain. Typically, the verification process will require calibration of some model parameters to achieve a close match with actual results. Verification should be conducted any time the model is significantly modified or updated.
- Data Integration: For complex supply chain models with a large quantity of input data, it is much more efficient to have the computer model retrieve the appropriate data from databases containing the latest information. In some situations, this information can be obtained directly from corporate ERP systems. In other situations, however, it may be necessary to conduct post-processing of the data before the information is suitable for use in the supply chain model. The ERP vender or a third-party data analysis software vender may be able to provide appropriate data analysis tools.
Development of a supply chain model requires a team approach that typically includes the following roles:
- Sponsor: Usually a senior-level manager who is focused on improving operations and bottom-line results. This person should have the authority and the knowledge to marshal the proper resources throughout the organization and direct the model development activities effectively.
- Operational Managers: These managers are usually responsible for some part of the supply chain related to production and manufacturing, transportation, warehousing, purchasing, inventory, marketing and sales, information systems, financial management, and human resources. These are typically the people who know how things work, what processes are in place, and how decisions are made in the real system. Their participation is critical to ensuring that all the important components and decision rules that make up the existing supply chain are identified and incorporated into the conceptual model.
- Supply chain Experts: These participants may either be internal line managers or external consultants with expertise in supply chain management. They typically operate at a corporate level, and focus on coordinating and optimizing overall operations across the supply chain. It is critical that these participants have the expertise to ask the right questions of the operational managers, formulate the conceptual model, and review the computer model of the supply chain to ensure that it accurately represents the conceptual model.
- Modelers: These are the people that translate the conceptual models into the mathematical expressions and relationships that make up the computer model. The modelers should be familiar with general computer modeling concepts (e.g., mass balance, unit consistency, Monte Carlo analysis, probability distributions, etc.). It is critical that the modelers be involved in the conceptual model development to ensure that the operational managers and supply chain experts are providing adequate information regarding relationships between critical components of the supply chain, documented or undocumented operational procedures, and feedback mechanisms.
Selecting Supply Chain Management Software
Numerous venders and consultants claim to provide packaged or custom simulation software suitable for supply chain management applications. Although many of these supply chain management software packages are suitable for simulating simple supply chain systems, some of these packages lack critical features necessary to address real-world problems. When selecting supply chain management software, key features to look for include the following:
· Capability to explicitly incorporate variability and uncertainty into the analysis: Attempting to simulate the future behavior of a supply chain system is complicated by the presence of significant variability and uncertainty (e.g., seasonal changes in demand, variations in production rates, fluctuations in the price of raw materials). Simulation software must have the ability to represent uncertainty regarding input data and system dynamics, the ability to conduct simulations that address the full range of uncertainty, and the ability to present the results in terms of the range and likelihood of different outputs (i.e., probability distributions). This ability is critical for supply chain modeling because many of the important governing parameters and processes are highly uncertain and/or variable.
· Capability to explicitly represent discrete events: Supply chain dynamics can be significantly influenced by random discrete events, such as a labor strike, a warehouse fire, or a supplier going out of business. Simulation software must have the capability to represent random discrete events in a manner that accounts for the likelihood of occurrence, the severity of the event, and the full range of consequences. This is critical for supply chain simulation since discrete (usually disruptive) events, such as labor strikes, supplier bankruptcies, equipment failure, and power outages, can play a critical role in the flexibility, robustness, and overall performance of the supply chain. Moreover, it is not possible to evaluate mitigation plans or alternative strategies without the ability to represent random discrete events.
· Top-down hierarchical model structure: Depending on the depth of analysis, large supply chain planning models can be very complex with thousands of interrelated components. It is infeasible to comprehend and work with such models if they are viewed as a single layer. Complex models must be built using simulation software that allows the investigators to construct hierarchical multi-layer models that represent greater detail at lower levels in the model structure. In this manner, investigators can build, explore, and explain highly-complex models without losing sight of the overall model structure and high-level relationships.
· Capability to dynamically link to external data repositories: Supply chain simulation should be based on current information regarding inventories, order rates, production rates, etc. For large models with large amounts of input data, it can be labor-intensive and burdensome to enter data by hand each time investigators want to update the model. Therefore, it is important that the simulation software have the ability to link to ERP and other database systems that represent the most recent information.
The GoldSim suite of simulation software meets all the requirements outlined above. An example of a supply chain simulation conducted using GoldSim is presented next.
Example Supply chain Model
The example presented here was created using the GoldSim simulation software package (www.goldsim.com). It represents a portion of the supply chain for an automotive manufacturing company. The scope of the model was limited to the OEM division and a single tier 1 supplier, the drive-train division. The objective of the analysis was to assess the impact of reducing several key information delays within the system.
The screen in Figure 4 illustrates the conceptual model for the OEM division. The drive train division has a similar conceptual model. The model simulates the following processes:
· Four different dealership order trends (constant, decreasing, cyclical, 10% increase after one year)
· Stochastic (random) variability in the dealership order rate
· Production scheduling given constraints in plant capacity, materials and labor
· Manufacturing productivity
· Delivery management and logistics
· Materials management (purchasing) with information delays
· Demand forecasting
· The potential for a labor strike (simulated as a discrete event)
Quantities
that are tracked within the model include: parts inventories, backlog,
work-in-progress, finished product, and product-in-transit.
In this particular example, OEM production scheduling is based directly on dealership orders. As a result, the finished product and powertrain production rates go up and down in response to changes in the dealership order rate. Figure 5a shows results for the period from 250 days to 350 days assuming that several key information delays are seven days. Over this period, dealership order rates vary between 170 and 230 units per day, while the OEM production rate varies from 80 to 230 units per day and the powertrain production rate varies from 50 to 270 units per day. This behavior is consistent with typical supply chain dynamics, in that the fluctuations in orders and production increase in magnitude as they propagate down the supply chain. Note that production rates tend to track changes in dealership order rates with a lag of approximately 20 days.
Figure 5b demonstrates how the powertrain production rate is
affected if several key information delays are reduced from seven days to one
day. First, changes in the dealership order rate tend to be reflected in the
powertrain production rate much sooner, with a reduction in the delay from 20
days to less than 10 days. Second, the range in production rates is
significantly reduced, with a maximum production rate of about 200 units per
day (compared with 270 units per day) and a minimum production rate of 80 units
per day (compared with 50 units per day). Although financial factors are not
included in this demonstration model, it is clear that significant cost-savings
would result due to the more stable
production rates.
Conclusions
Supply chain management and supply chain planning have emerged as an increasingly important issue for companies. The challenge of supply chain management is to identify and implement strategies that minimize costs while maximizing flexibility. This paper examines how dynamic simulation tools can be used to better understand supply chain dynamics, diagnose problems and evaluate possible supply chain management solutions, conduct better supply chain planning, optimize operations, and mitigate risk factors.
Supply chain simulation models can be used to address a broad range of
problems and issues. Most of these applications fall into one of the following categories:
- Optimization
- Decision Analysis
- Diagnostic Evaluation
- Project Planning
-
Risk Management
When selecting supply chain management software, key features to look for include the following:
· Capability to explicitly incorporate variability and uncertainty into the analysis
· Capability to explicitly represent discrete events
· Top-down hierarchical model structure
· Capability to dynamically link to external data repositories
________________________________________________________________
About the GoldSim Technology Group
The GoldSim Technology Group is dedicated to delivering
software and services to help people understand complex systems and make better
decisions. Our flagship GoldSim simulation software package is based on
technology developed over the past 10 years servicing such clients as the U.S.
Department of Energy, Bechtel, Booz Allen & Hamilton, IT Corporation, Japan
Nuclear Cycle Corporation, Phelps Dodge, Shell Global Solutions,
Petrobras (Brazil)
and Taisei Corporation (Japan). The GoldSim simulation software package is a generalized simulator suitable for modeling any type of real-world system and has been used to solve problems related to strategic planning, portfolio analysis, risk assessment, program planning, supply chain management, environmental systems, and engineered systems. The GoldSim Technology Group is a division of Golder Associates, Inc., an international consulting firm with over 2,700 employees. For additional information, please contact:
J. Scott Kindred
Director of Marketing
GoldSim Technology Group
Phone: 425-883-0777
Email: skindred@goldsim.com
© 2002 Golder Associates Inc. All rights reserved.
The information contained in
this document represents the current view of Golder Associates Inc. on the
issues discussed as of the date of publication. Because Golder Associates Inc.
must respond to changing market conditions, it should not be interpreted to be
a commitment on the part of Golder Associates Inc., and Golder Associates Inc.
cannot guarantee the accuracy of any information presented in this publication.
Information in this document is subject to change without notice. This white
paper is for informational purposes only. Golder
Associates Inc. MAKES NO WARRANTIES, EXPRESS OR IMPLIED, IN THIS
SUMMARY.
Complying with all applicable
copyright laws is the responsibility of the user. Without limiting the rights
under copyright, no part of this document may be reproduced, stored in, or
introduced into a retrieval system, or transmitted in any form, by any means
(electronic, mechanical, photocopying, recording, or otherwise), or for any
purpose, without the express written permission of Golder Associates Inc.
Golder Associates Inc. may
have patents, patent applications, trademarks, copyrights, or other
intellectual property rights covering subject matter in this document. Except
as expressly provided in any written license agreement from Golder Associates
Inc., the furnishing of this document does not give you the license to these
patents, trademarks, copyrights, or other intellectual property.
Golder, Golder Associates,
and GoldSim are either registered trademarks or trademarks of Golder Associates
Inc. in the United States and/or other countries. The names of actual companies
and products mentioned herein may be the trademarks of their respective owners.
Golder Associates Inc.,
18300 NE Union Hill Road, Suite 200, Redmond, WA 98052, USA.