gained from reading

grace0802

567265.pdf

Home >Reading homework help >gained from reading

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

BUSML 4382: Logistics Analytics Keely Croxton

Ohio State University Spring 2016

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

BUSML 4382: Logistics Analytics BUSML 4382

Ohio State University

THIS PRINT COURSEPACK AND ITS ELECTRONIC COUNTERPART (IF ANY) ARE INTENDED SOLELY FOR THE PERSONAL USE OF PURCHASER. ALL OTHER USE IS STRICTLY PROHIBITED.

ISBN 1323097554

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

XanEdu™ publications may contain copyrighted materials of XanEdu, Inc. and/or its licensors. The original copyright holders retain sole ownership of their materials. Copyright permissions from third parties have been granted for materials for this publication only. Further reproduction and distribution of the materials contained herein is prohibited.

WARNING: COPYRIGHT INFRINGEMENT IS AGAINST THE LAW AND WILL RESULT IN PROSECUTION TO THE FULLEST EXTENT OF THE LAW. THIS COURSE PACK CANNOT BE RESOLD, COPIED OR OTHERWISE REPRODUCED.

XanEdu Publishing, Inc. does not exert editorial control over materials that are included in this course pack. The user hereby releases XanEdu Publishing, Inc. from any and all liability for any claims or damages, which result from any use or exposure to the materials of this course pack.

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

− In print copy only − Online only

Unmarked items are available both in print and online

BUSML 4382: Logistics Analytics

Table of Contents “Chapter 1: The Value of Supply Chain Network Design” by Watson, Michael; Lewis, Sara; Cacioppi, Peter; Jayaraman, Jay

“Chapter 12: The Art of Modeling” by Watson, Michael; Lewis, Sara; Cacioppi, Peter; Jayaraman, Jay

“Chapter 13: Data Aggregation in Network Design” by Watson, Michael; Lewis, Sara; Cacioppi, Peter; Jayaraman, Jay

Bibliography 67

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016 ii

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

1 THE VALUE OF SUPPLY CHAIN

NETWORK DESIGN

What Is Supply Chain Network Design and Why Is It Important? A firm's supply chain allows it to move product from the source to the final point of consumption. Leading firms around the world, from large retailers to high-tech elec- tronics manufacturers, have learned to use their supply chain as a strategic weapon. A supply chain is defined by the suppliers, plants, warehouses, and flows of products from each product's origin to the final customer. The number and locations of these facilities is a critical factor in the success of any supply chain. In fact, some experts suggest that 800/0 of the costs of the supply chain are locked in with the location of the facilities and the determination of optimal flows of product between them. (This is similar to the notion from manufacturing that you lock in 800/0 of the cost to make a product with its design.) The most successful companies recognize this and place significant emphasis on strategic planning by determining the best facility locations and product flows. The discipline used to determine the optimal location and size of facilities and the flow through the facilities is called supply chain network design.

This book covers the discipline of supply chain network design. Sometimes it is referred to as network modeling because you need to build a mathematical model of the supply chain. This model is then solved using optimization techniques and then analyzed to pick the best solution. Specifically, we will focus on modeling the supply chain to deter- mine the optimal location of facilities (warehouses, plants, lines within the plants, and suppliers) and the best flow of products through this facility network structure.

Here are four examples to illustrate the value of supply chain network design~

• 1

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Example #1

Often, we hear about firms acquiring or merging with another firm in the same indus- try to reduce the overall costs to operate both firms. That is, they justify the new com- bined company by determining that they can deliver the same or more products to the market at an overall lower cost. In firms that make or ship a lot of products, a large por- tion of the savings comes from the merger of the two supply chains. In such mergers, the savings often come from closing redundant plant and warehouse locations, opening new plants and warehouses, or deciding to use existing facilities to make or distribute different mixes of product. We have heard firms claim resultant supply chain savings from $40 million to $350 million over a period of a couple of years. With these kinds of savings, you can only imagine the pressure placed on the supply chain team to deter- mine the new optimal supply chain structure after an acquisition or merger is announced.

Example #2

Often, a large firm will find that its supply chain no longer serves its business needs. In situations like this, the firm will have to transform its supply chain. It may have to close many facilities, open many new ones, and use facilities in a completely different way. For example, a retail firm may have to redesign their supply chain to serve their stores as well as their new online customer base in a more integrated approach. Or a large retail- er may find that some of their product lines have grown significantly and the retailer needs new warehouses to manage this growth. If done right, this type of supply chain transformation can help reduce logistics and inventory costs, better respond to different competitive landscapes, and increase sales and profitability. We have even seen firms highlight this work in their annual reports, therefore showing the importance of this analysis to the firm as a whole.

Example #3

In the spring of 2011, we were working on a project for a global chemical company to help develop their long-term plan for their supply chain. This study was analyzing where they should locate new plants to serve a global customer base. The long-term project suddenly became extremely short-term when the CEO called the project team to inform them that within six hours they were closing their plant in Egypt due to politi- cal unrest. He also indicated there was no timeline for reopening the plant. The CEO immediately needed to know which of the existing plants should produce the products that were currently being manufactured in Egypt and how customer demand was going to be impacted. The team was quickly able to deliver the answers and minimize this sup- ply chain disruption. As seen in this example, supply chain network design models can also be a great tool for identifying risks and creating contingency plans in both the short and long term .

• 2 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Example #4

As consumer behavior and buying patterns change, firms often want to bring their product to the market through different channels. For example, we worked with a con- sumer products company that wanted to analyze different channels such as selling through big-box retailers, selling through smaller retailers, selling direct online, and sell- ing through distributors. This firm wanted to analyze different ways to bring their prod- uct to market and understand what the supply chain would need to look like for each of these cases. That is, they wanted to determine the optimal number and location of plants and warehouses. This would then be a key piece of information to help them determine their overall strategy.

Of course, the details of these studies can be a bit more complicated. As an example, take a minute to think through the possible supply chain for a tablet computer and compare that to the supply chain you envision for a candy bar.

The tablet supply chain faces specific challenges surrounding a time-sensitive delivery of the device for a very demanding high-tech customer market. The tablet maker must also determine how to best balance its partnerships with many contract manufacturers worldwide while still ensuring the highest quality end products. Finally, this supply chain must deal with the high costs for insurance, transport, and storage of these high- priced finished goods.

Conversely, when we shift our thoughts to the supply chain related to the candy bar, we must consider an entirely different set of challenges and objectives that the candy maker must face: government regulations that mandate different requirements for all stages of production paired with a strict shelf life of each unit produced. In addition, raw mate- rial costs, as well as costs tied to temperature control during transit and storage, add up. Major swings in demand due to seasonality or promotions also add the need for flexi- bility within their supply chain.

Despite their differences, both the tablet maker and candy bar maker must determine the best number and location of their suppliers, plants, and warehouses and how to best flow product through the facilities. And building a model using optimization is still the best way for both of them to determine their network design.

As the previous examples highlight, many different types of firms could benefit from network design and many factors go into the good design of the supply chain. Along with ever-growing complexity, the need to truly understand how all these requirements affect a company's costs and performance is now a requirement. Using all these variables to prove out the optimal design configuration commonly saves companies millions of dollars each year.

As you would expect, a network design project can answer many types of questions such as these:

• Chapter 1 The Value of Supply Chain Network Design 3

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

• How many warehouses should we have, where should they be, how large should they be, what products will they distribute and how will we serve our different types of customers?

• How many plants or manufacturing sites should we have, where should they be, how large should they be, how many production lines should we have and what products should they make, and which warehouses should they service?

• Which products should we make internally and which should we source from outside firms?

• If we source from outside firms, which suppliers should we use?

• What is the trade-off between the number of facilities and overall costs?

• What is the trade-off between the number of facilities and the service level? How much does it cost to improve the service level?

• What is the impact of changes in demand, labor cost, and commodity pricing on the network?

• When should we make product to best manage and plan for seasonality in the business?

• How do we ensure the proper capacity and flexibility within the network? To meet demand growth, do we need to expand our existing plants or build new plants? When do we need to add this capacity?

• How can we reduce the overall supply chain costs?

Being able to answer these questions in the optimal manner is important to the overall efficiency and effectiveness of any firm. Companies that have not evaluated their supply chain in several years or those that have a new supply chain through acquisitions can expect to reduce long-term transportation, warehousing, and other supply chain costs from 5% to 150/0. Many of these firms also see an improvement in their service level and ability to meet the strategic direction of their company.

Although firms are happy to find 50/0 to 150/0 reduction in cost, it does highlight that your supply chain might have already missed out on significant savings you may have realized had you done the study a year ago (or two or more years ago). Some firms have realized this and now run this type of analysis on a more frequent basis (say, quarterly). This allows them to readjust their supply chain over time and keep their supply chain continually running in an optimal state while preventing costs from drifting upward.

The frequency of these studies depends on several factors. Historically, it has been cus- tomary to complete these analyses once every several years per business unit, because it was usual for business demographics and characteristics to change over this period of time. For some industries such as high-tech, the frequency was even higher because there may be higher volatility in customer demand, thereby requiring periodic

• 4 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

reevaluation of the network. Any major events, such as mergers, acquisitions, or divesti- tures, should also trigger a network reevaluation study. As noted before, the savings from the optimization of the revised network typically represent a significant part of the savings that justify the merger or divestiture. A current trend we are seeing, however, is to do these studies even more frequently. Business demographics and characteristics are changing faster. In addition, the growth of the global supply chain is driving firms to cycle through studies as they go from region to region around the world. Also, firms are running the same models more frequently to stay on top of changes in their business by adjusting the supply chain. Some firms update these models several times throughout the year.

Determining the right supply chain design involves a lot of quantitative data as well as some nonquantitative considerations. We will discuss this in the rest of this chapter, as well as how we use mathematical optimization to sort through this quantitative data.

Quantitative Data: Why Does Geography Matter? It should be clear by now that the supply chain network design problem is just as much about geography as it is about business strategy. The two cannot be separated.

Take these supply chain considerations for example:

• If you have a plant in the interior of China and some of your customers are in New York, you need to physically get the product out of China, across the ocean, and to New York.

• If you make wood products (like paper or boards), you can locate plants either close to the raw materials (forest areas) or close to your customers (usually located a significant distance away from the large forest areas).

• If you have a warehouse in Indianapolis, you are close to your customers in Chicago, but far away from customers in Miami. If the warehouse is in Atlanta, you are closer to Miami, but farther from Chicago.

• If you make a critical product only in Miami, a hurricane may shut down your operation, causing a loss of revenue.

As the examples highlight, decisions about the location of your facilities impact many aspects of your business and require you to make trade-offs. Specifically, geography drives the following:

• Transportation Cost-You need to move product from its original source to its final destination. The location of your facilities determines the distance you need to move product, which directly impacts the amount you spend on trans- portation. But, also, the location of your facilities determines your access to

• Chapter 1 The Value of Supply Chain Network Design 5

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

transportation infrastructure such as highways, airports, railheads, and ports. Finally, because of supply and demand, different locations may have different transportation rates.

• Service Level-Where you locate relative to your customers impacts the time it takes to get product to your customers. For some products, you can negate great distances by using overnight air freight. But this usually comes at a premium cost.

• Risk-The number and location of your facilities impacts risk. If you have just one location for a critical activity, there is always the risk that a fire, flood, some other natural disaster, a strike, or legal issues will shut down your operation. There is also political risk to consider. Your facility could get confiscated or shut down for political reasons, or the borders may shut down, isolating your facility.

• Local Labor, Skills, Materials, and Utilities-The location of your facilities also determines what you pay for labor, your ability to find the needed skills, the cost of locally procured materials (which is often directly related to the local labor costs), and the cost of your utilities.

• Taxes-Your facilities may be directly taxed depending on where they are locat- ed and the type of operations being performed. In addition, you also need to consider the tax implication of shipping product to and from your locations. In some industries, taxes are more expensive than transportation costs.

• Carbon Emissions-Locating facilities to minimize the distance traveled or the transportation costs often has the side benefit of reducing carbon emissions. In addition, if your facilities consume a lot of electricity, you can reduce your emissions by locating near low-emission power plants.

As the list highlights, geography matters. What makes this challenging is that the geog- raphy often pushes the solution in different directions at the same time. For example, it would be desirable to have a facility close to all the demand. However, demand is typi- cally where people live. And it is usually very expensive, if not impossible, to locate a plant or warehouse in the middle of a major metropolitan area. So the desire to be close to customers pushes locations close to cities. The desire for cheap land and labor (and welcoming neighbors) pushes the best locations further from the city center. In global supply chains these decisions become even more extreme. In some cases it may make sense to service demand from a location on an entirely different continent.

In addition to geography, the next two sections will discuss the importance of ware- houses and multiple plants to your supply chain as well.

• 6 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Quantitative Data: Why Have Warehouses? In this book, a warehouse represents a facility where firms store product or a location where product simply passes through from one vehicle to another. It can be called a dis- tribution center, a mixing center, a cross dock, a plant-attached warehouse, a forward warehouse, a hub or central warehouse, a spoke or regional warehouse, or a host of other terms.

To understand how to optimally locate warehouses, it is important to discuss why ware- houses exist. Wouldn't it be much cheaper for companies to load the product only once, at the manufacturing location, and ship it directly to the customer? Stopping at a ware- house adds loading, unloading, and storage costs, not to mention the cost for two legs to transportation (one leg from the plant to the warehouse and one leg from the ware- house to the customer). In cases where you can ship directly from the plant, it is usual- ly good to do so. Therefore, it is important to ask questions to see whether you can avoid warehouses altogether. But in most cases warehouses are needed in a supply chain for the following reasons:

• Consolidation of Products-Often, you will need to deliver a mix of different products to your customers and these products may come from various sources. A warehouse serves the useful function of bringing these products together so that you can then make a single shipment to a customer. This will be cheaper than having the products ship to the customers directly from each individual source of supply.

• Buffer Lead Time-In many cases, you will need to ship to your customers with lead time that is shorter than that which can be offered by shipping direct- ly from the plant or supplier location. For example, you may promise to ship products to your customers the next day but your plants or suppliers may have a lead time of several weeks before they are able to make the product available to the customer. In this case, the warehouse holds product at a location closer to the customers in order to provide the next day transport promised each time an order is placed.

• Service Levels-Where you store the product and its proximity to the market where it will be consumed is also a measure of the service level the company can provide. The need to be close to customers can create the need for multiple warehouses. Overall cost versus service level is one of the most classic trade-offs in supply chain network design.

• Production Lot Sizes-Setting up and starting the production of a single prod- uct or group of similar products on a line can have a significant fixed cost asso- ciated with it. Therefore, production plans attempt to maximize the number of units of product made during each run. (This production amount is called a lot

• Chapter 1 The Value of Supply Chain Network Design 7

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

size.) Understandably, these lot sizes normally do not match the exact demand from the market at the time. This requires the extra units to be "stored" in warehouses until future demand requires them. Production lot sizes versus inventory storage costs is also a common supply chain design trade-off.

• Inventory Pre-Build-Some industries see huge spikes in the supply of raw materials (seasonal food harvests) or in the demand of finished goods (holiday retail shopping). In the case of raw material supply spikes, some firms must store these abundant raw materials until the time they will be needed for steady monthly production cycles. Other firms must immediately use these raw mate- rials to produce finished goods that are not yet demanded. These additional fin- ished goods must then be stored until demand in future time periods requires them. In the case of demand spikes, companies find themselves with insufficient production capacity to fulfill all orders during peak periods of demand. As a result, they must use their additional capacity during off-peak time periods to make finished-good units to be stored awaiting their use to fulfill the upcoming spikes in demand. The use of costly overtime production versus inventory stor- age costs is another common supply chain design trade-off.

• Transportation Mode Trade-offs-Having warehouses often allows you to take advantage of economies of scale in transportation. A warehouse can help reduce costs by allowing the shipment of products a long distance with an efficient (and lower cost) mode of transportation and then facilitating the changeover to a less efficient (and usually more expensive) mode of transportation for a shorter trip to the final destination (as opposed to shipping the entire distance on the less efficient mode).

It is also important to match up the preceding list of reasons for warehouses with the types of warehouses in the supply chain. A supply chain may have many types of ware- houses to meet many different needs. Here are some common types of warehouses:

•

• Distribution Center-Typically refers to a warehouse where product is stored and from which customer orders are fulfilled. This is the most common and traditional definition of a warehouse. When a customer places an order, the dis- tribution center will pick the items from their inventory and ship them to the customers. These types of facilities are also called mixing centers because they «mix') products from many locations so that your customers can place and receive an order from a single location. If a manufacturing company does not have this type of warehouse in the supply chain, customers may have to place several orders or receive several shipments from different locations depending on where each product they want is made.

• Cross-Dock-Usually refers to a warehouse that is simply a meeting place for products to move from inbound trucks to outbound trucks. The term simply

8 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

means that products pass (or cross) from one loading dock (for inbound trucks) to another loading dock (for the outbound trucks). For example, in the case of a produce retailer with 50 stores, they may have a full truck of fresh peaches arriving at the inbound docks from a single supplier. The peaches are then removed from the truck and some are placed in each of the 50 waiting trucks on the outbound side, according to the relevant store demand. This hap- pens for peaches as well as a host of other produce items. Basically full trucks arrive from a single supplier on the inbound side of the facility, and then trans- ferred to multiple trucks on the outbound side of the facility resulting in fully loaded truckloads with a mix of product from each of the suppliers quickly sent on their way. The best-run cross-dock systems have all the inbound trucks arriving at approximately the same time so that product stays at the cross-dock for only a short period of time.

• Plant-Attached Warehouse-Refers to a warehouse that is attached to a manu- facturing plant. Almost all plants have some sort of product storage as part of their operations. For some, it may simply be a small space at the end of the line where product is staged prior to being loaded onto a truck for shipment. In other cases, the warehouse can act as a storage point for product made at the plant or for products made at other plants. In this case, this warehouse acts like a distribution center co-located with the manufacturing facility. A major benefit of a plant-attached warehouse is the reduction of transportation costs because a product does not have to be shipped to another location immediately after it comes off the end of the line. When you have plant -attached warehouses, some- times the standalone warehouses are called forward warehouses, meaning they are placed "forward" or out closer to customers .

• Hub Warehouse or Central Warehouse-Refers to a warehouse that consoli- dates products to be shipped to other warehouses in the system before moving on to customers. Different from cross-docks, the products are normally stored in these locations for longer periods of time before being used to fulfill demand. The other warehouses in the network are then typically called spokes or regional warehouses.

In practice, you will find many different names for warehouses. These names are most likely just different terms for what is described in the preceding list. In addition to the types of facilities, there are also needs for different temperature classes (frozen, refriger- ated, or ambient), different levels of safety (hazardous or nonhazardous), and different levels of ownership (company owned, company leased, or the company uses a third- party facility).

As an interesting side note and to further illustrate the wide range of warehouse types we have experienced, we have even seen caves used as warehouses. Caves have the nice advantage of maintaining the same (relatively low) temperature and have prebuilt roofs .

• Chapter 1 The Value of Supply Chain Network Design 9

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

If you can get trucks into and out of them and have room to store products, caves also make great warehouses. Kansas City is probably the best-known location with cave warehouses.

Quantitative Data: Why Have Multiple Plants? Similar to our use of the word "warehouse;' we are also going to use the term "plant" to broadly refer to locations where product is made or where it comes from. So a plant could be a manufacturing plant that produces raw material, components, or finished goods, or just does assembly. A plant could be owned by the firm, it could be a supplier, or it could be a third-party plant that makes products on behalf of the firm. These third- party plants are often called co-packers, co-manufacturers, or toll manufacturers (a term used for third-party manufacturers common in the chemical industry).

A plant can also contain multiple production lines. So, often we are determining not only the location of plants, but also the number and location of the production lines.

For companies that make products, the number and location of plants are important. For retailers and wholesalers however, the location of the suppliers is often beyond their scope of control.

Even for a firm that has only one product, the location of the plant can impact trans- portation costs and the ability to service customers. In some cases, the location of the plant is primarily driven by the need to have skills in the right place or the need to be next to the corporate headquarters. However, in most cases, there are choices for plant locations. An even more interesting choice is to determine whether you should locate multiple plants to make the same product-even when a single plant could easily han- dle all the demand.

When you have choices for where to put your plants or the option to have several plants, you must consider some of the same questions we did when locating warehouses. For example, factors that would drive you to have multiple plants making the same product include:

•

• Service Levels-If you need your plants to be close to customers, this will drive the need for multiple plants making the same product. This becomes especially important if your business does not use warehouses. In this case, your plants face the customer and their location can drive service levels.

• Transportation Costs-For producers of heavy or bulky products that easily fill up truckload capacities, you will want to be as close to your markets as possible. This may also drive the need for multiple warehouses.

• Economies of Scale-As a counterbalance to the benefits of transportation, you also want to factor in the economies of scale within production. As mentioned

10 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

previously, the more you make of a given product at a single location, the lower the production cost per unit. This is driven by a reduction in production line setup time and costs and the benefit of being able to create a more focused manufacturing process. So while it may be ideal to have many producti9n loca- tions to minimize transportation cost, economies of scale in manufacturing suggest that fewer plants will be better.

• Taxes-In a global supply chain, it is often important to consider the tax implications of producing and distributing product from multiple or different locations.

• Steps in the Production Process-In a production process with multiple steps, you may need to decide where you should do various activities. For example, it can often be a good strategy to make product in bulk at a low-cost plant, ship it in bulk to another plant closer to the market to complete the conversion to a finished good.

As with the warehouses, a plant can also represent many types of facilities. A plant may represent any of the following:

• A Manufacturing or Assembly Site-This is a site that is owned by the firm that makes products. The products coming out of the plant could be raw mate- rials, semifinished goods, or finished goods.

• A Supplier-This is a location that is not owned by the firm but supplies prod- uct to the firm. The products could again be raw material, semi finished goods, or finished goods. In the majority of cases, you have no control over where your suppliers are located. You may, however, be able to pick which supplier you will purchase from.

• Third-Party Manufacturing Site-This is a location similar to supplier plants, but these sites make product on behalf of the firm and are therefore treated more like the firms' own plants than a raw material supplier. Many firms use third-party manufacturing sites because manufacturing may not be their core competency. Their competitive advantage comes from all other activities includ- ing the design, marketing, and/or final sale of the product. These third-party firms, typically called contract manufacturers, are widely used in the electronics industry. Other third-party firms, termed co-manufacturers, co-packers, or toll manufacturers, are used to simply supplement the capacity of the firm itself. These are quite common with consumer packaged goods like food and bever- ages or chemical companies.

As with warehouses, in practice, you will find many variations in terminology for plants and different types of manufacturing sites and suppliers serving roles similar to those mentioned previously.

• Chapter 1 The Value of Supply Chain Network Design 11

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Solving the Quantitative Aspects of the Problem Using Optimization Because of the supply chain complexities and rich set of quantitative data we have dis- cussed, mathematical optimization technology is the best way to sort through the various options, balance the trade-offs, determine the best locations for facilities, and support better decision making. The mathematical optimization relies on linear and integer programming.

One common misconception we see is that managers sometimes confuse having good data and a good reporting system with having an optimal supply chain. That is, they think that their investment in good data and reporting systems should equip them with the ability to complete network design analysis. (These systems are often called business intel- ligence systems.) But in reality, if your warehouses or factories are located in the wrong places, these reporting systems will not correct the situation or suggest new locations.

So although these systems do, in fact, generate good reports and allow managers to gain good insights, they do not lead to better designs for the supply chain. That is, they are not built to construct models of your supply chains to which you can apply mathemat- ical optimization. At best, they may allow you to evaluate a small handful of alternatives, but in doing so, you have to define all the details of each of the alternatives. Optimization is a complementary, not competitive, technology that allows you to actu- ally determine the best locations for your facilities. And you can let optimization do the heavy number crunching to determine the details of the alternatives (where to locate, what is made where, how product flows, which customer is served by which warehouse). And, in many cases, if the optimization is set up correctly, it will uncover ideas that you never thought about.

Of course, because these problems are of great strategic value to an organization and touch on many aspects of the business, there will be non quantitative aspects you must consider. These non quantifiable aspects are important and are discussed later in this chapter and throughout the book.

With all that said, solving the quantitative aspects of this problem with mathematical optimization is the key to coming up with the best answers. So, let's start there.

To formulate a logical supply chain network model, you need to think about the follow- ing four elements:

• Objective

• Constraints

• Decisions

• Data

• 12 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

The objective is the goal of the optimization and the criteria we'll use to compare differ- ent solutions. For example, the most common objective in strategic network design is to minimize cost. If our objective is to minimize cost, we can now compare two solutions and judge which one is better based on the cost. When the mathematical optimization engine is running, it searches for the solution with the lowest cost. So with this common example you can see that an optimization problem needs to have a quantifiable objec- tive. It is important to point this out because we have encountered many situations in which supply chain managers say their objective is to "optimize their supply chain." The appropriate response is to ask what exactly they want to optimize, or what criteria they will use to determine which of two solutions is better. For example, is the key criteria transportation cost, is it service, is it facilities costs, or something else? Later, we'll dis- cuss optimizing multiple objectives (as long as you can quantify them). And we will show methods for analyzing nonquantifiable factors such as risk and robustness, because these are also very important. But for now, we'll start with one quantifiable objective. When we understand this one, we'll be able to understand more detailed analysis later .

. The constraints define the rules of a legitimate solution. For example, if you want to just minimize cost, it is probably best to not make any products, not ship anything, or have no facilities. A cost of zero is indeed the minimum; however, that's clearly not realistic. So there are some logical constraints we must include such as the fact that you want to meet all the demand. There are also constraints that specify which products may be made where, how much production capacity is available, how close your warehouses must be to customers, and a variety of other details. In this step, you also want to be careful not to specify so many constraints that you prohibit the optimization from find- ing new and creative strategies.

The decisions (sometimes called decision variables) define what you allow the optimiza- tion to choose from. So in the optimization of the physical supply chain, the main deci- sions include how much product moves from one location to another, how many sites are picked, where those sites are, and what product is made in which location. But, cer- tainly, there are other decisions as well. The allowable decisions cannot be separated from the constraints. For example, if you have existing warehouses, you mayor may not be able to close some of these sites.

Finally, you must consider the data you have available to you. There may be factors you would like to consider in the optimization, but you do not have the data to support. In this case, you still need to figure out ways to make a good decision. This could include running multiple scenarios, considering approximate data, or adjusting the data you have. We will further discuss a variety of these techniques later in this text.

After you have thought about your problem in terms of the objective, constraints, deci- sions, and data, there are ways to translate this into a series of equations and then solve the problem using linear and integer programming techniques. This is also sometimes

• Chapter 1 The Value of Supply Chain Network Design 13

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

referred to as mixed integer programming (MIP)-the "mix" refers to a mix of linear and integer variables. We will cover what this means in more detail later in the book, and will also provide information on how these problems work. There are whole cours- es on MIP however, so we will cover the topic only as it relates to supply chain network design problems in this text.

One way to think about a MIP is to think of it as a series of steps that are influenced by the objective, the constraints, and the decision variables. That is, during the steps, the objective steers the solution to more favorable costs and avoids less favorable costs, the constraints set the rules and can prevent it from doing more of what it wants or can force it to do something that is not favorable to the objective, and the decision variables tell it what it is allowed to change. A nice benefit of MIPs is that they solve for all the decisions and consider all the constraints simultaneously. That is, this approach allows you to come up with the overall best solution for a given problem.

Data Precision Versus Significance: What Is the Right Level in Modeling? The discussion about quantitative data brings us to a discussion about data accuracy.

The lessons we learned about significant digits in high -school chemistry will serve us well when doing network design studies.

We learned that measurements were only as accurate as the equipment that took the measurement, and we had to report it as such. And when we combined different meas- urements together in some equation, our answer could be reported only in terms of the measurement with the least amount of accuracy. This accuracy was expressed in the number of significant digits. For example, if we took one measurement and it was 3.2 units and added that to another measurement of 4.1578 units and added them togeth- er, our calculator would give us 7.3578 units. However, we have to report it as 7.4 units because the final answer can have at most two significant digits. To write it otherwise would give the answer a level of accuracy that just wasn't true.

We often assume that more precise data is always better. However, as our lessons in sig- nificant digits taught us, we can be only as precise as our measurements allow. Keeping this concept in mind will serve you well when you are collecting data for your network design models as well.

The adage about bad data leading to bad results, "garbage in equals garbage out," can sometimes just confuse. This adage does not mean that data needs to be precisely meas- ured to a certain number of significant digits. It just means that the data has to be good enough for the decisions we are making .

• 14 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Other problems can result when the precision and detail of the data actually get in the way of making good decisions. The cartoon in Figure 1.1 highlights this point very well.

A 250 kilogram object that looks like a piano is approaching your head at 9.8 meters per second squared.

Figure 1.1 Precision Cartoon

In the cartoon, the extra precision on the left actually makes things worse for our poor analyst (who is about to be hit by a piano). The analyst has to spend too much time try- ing to understand the data and misses the opportunity to take the much-needed action of getting out of the way. In network modeling, our time horizon for making decisions is much longer, but the data is also much more complex. We have seen projects in which the extremely detailed analysis of data causes the project team to miss their opportuni- ty to impact the supply chain in a positive way.

Our goal for collecting data for a network design model is to define the data needed with the right level of significance to make the relevant decisions. Our ultimate respon- sibility is to report the results with the right level of significance for the organization to make decisions. Our goal is not to ensure that every piece of data is significant to ten decimal places. Therefore it is really a waste of time and often a risk to our project's suc- cess when we report data with more significance than is warranted.

Also keep in mind that when we are running a network design project, we are making decisions about the location of our plants and warehouses to support the business in the

• Chapter 1 The Value of Supply Chain Network Design 15

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

future. It is pointless to build a new warehouse to support last year's business. Although you may start your analysis with historical data as a baseline (we'll cover baseline in more detail in Chapter 8, "Baselines and Optimal Baselines"), you will eventually want to consider how this data might be different when considering it in a future state. Two specific elements with future uncertainty, for example, include:

• Demand Data-There will be uncertainty in future demand dependent on overall economic conditions, moves by competitors, success of your marketing programs, and so on.

• Transportation Costs-There can be a lot of variance in future prices of trans- portation dependent on the ever-fluctuating world market price of oil.

Based on the previous, we know our instruments for measuring future demand and transportation cost data are likely going to be inaccurate. Therefore we should consider using fewer significant digits when representing them in the model. If we get push-back to include more significant digits from other stakeholders in the project, we should remind ourselves that if we could accurately predict future demand or the future price of oil, we would benefit more by taking this knowledge to the financial markets.

In Chapter 4, "Alternative Service Levels and Sensitivity Analysis," we will expand our discussion of this topic to include the use of sensitivity analysis and multiple scenarios to help make our best network design decisions despite this problem of uncertainty. That is, just because we don't know the future demand or the future price of oil, doesn't mean we can't still come up with good solutions.

Often times, the problems we mentioned previously tempt teams into giving up on net- work design until they get better and more precise data. People get nervous about mak- ing decisions without enough significant digits or enough precision. There are two problems with this approach, though:

•

• You are fooling yourself that you are not making decisions. If you don't do any- thing with your supply chain, you are making the decision that there is nothing you can do to improve the supply chain right now. And if you make decisions without any formal data collection and modeling, you are using the resource that is the most imprecise and has the fewest significant digits of all: your intuition.

• You are missing a chance to better understand your supply chain and under- stand what data you should be collecting. For example, a good practice that we have found is to make initial assumptions, run scenarios, and test those assumptions by varying the data by +/-100/0, then +/-200/0, and seeing what happens. These runs give you insight into your supply chain and show you the value of the data. For example, if the results are not sensitive to a particular data element, you do not have to spend much time refining that data element as you continue to build out your model.

16 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Of course, when we are looking at the results and output data, we also want to keep our lessons in significant digits in mind. If our significant digits are valid only to the nearest million dollars, then a reported savings of $500,000 should not be considered.

The other concept to keep in mind when analyzing the output comes from introducto- ry statistics. When you first learned about hypothesis testing, you were shown tech- niques for determining whether two different samples were statistically different from each other. For example, if you ran two marketing campaigns in two different markets and one market came back with average sales of $15,000 per store and the other with $17,500 per store, you could not immediately conclude that the second marketing cam- paign was better. In general, the higher the variability of the data in the sample size, the harder it is to claim that there is a statistical difference between the two samples.

Although we cannot apply this direct statistical test to our network design models, we can apply the concept. Because we already know that we have a lot of underlying vari- ability in our data (such as future demand, future costs, and so on), we know that we need to show a fairly large savings from the current state to be comfortable with making the decision to implement the recommended change.

Typically, we look for savings of more than 50/0 to 100/0 compared to the current situation before we recommend a change. That is, when relocating facilities, if you find savings that reduce costs by 1 % to 20/0 or change services levels by 20/0 to 30/0, we would consider this not statistically different from the current state. When the savings are 150/0 to 250/0 how- ever, you can be confident that you have found a statistically different solution.

Of course, you should be careful with this rule. You don't need to completely disregard the value of relatively small improvements either. The key is to prevent the magnitude of the suggested change from swamping the size of the expected benefit. For example, you might study a $300 million supply chain that is already well rationalized, and only find $250,000 in savings. However, this 0.0840/0 savings might be well worth the bother, if it could be realized by reassigning just two customers. In such a situation, you might think of the $250,000 as being proportionally quite large when compared to the total landed cost of serving these two wayward demand points.

As you go forward in this book and with projects for your company, these are important lessons that will serve you well.

Nonquantifiable Data: What Other Factors Need to Be Considered? Our discussion up to this point has focused on the quantifiable aspects of the network design solution. However, there are other factors that you want to consider when mak- ing a final decision. Some of these factors do not lend themselves to being quantified and being considered directly within the optimization runs. This does not mean that

• Chapter 1 The Value of Supply Chain Network Design 17

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

our optimization runs are meaningless, nor does it mean that we should just ignore these factors.

It is important to run the optimization with as much quantifiable information as need- ed. You will want to run a variety of scenarios with different input data. This will then give you a range of potential solutions. From this range, you will then want to apply the consideration of the nonquantifiable aspects. For example, if closing existing facilities and opening new ones in new locations increases your political risk, you want to know whether that new configuration saves you $500,000 or $50 million. You will then be able to judge whether the extra risk is worth it.

We will discuss how you run different scenarios and create a range of possible solutions in future chapters. Some of the nonquantifiable factors you might then want to consid- er include the following:

•

• Firm's Strategy-Your firm may value cost more than service or vice versa. For example, your firm may have a strategy of servicing the top customers at any expense or be committed to a local manufacturing strategy.

• Risk-For global supply chains, you need to worry about placing sites in politi- cally unstable locations, port closures, and the added risk of extra distance between origins and destinations. There is also a risk when you have just a sin- gle location to make a given product or you have a supply chain that is current- ly at capacity and is not equipped to handle any unexpected extra demand.

• Disruption Cost-Firms realize that changes could cause significant disrup- tion, leading to other costs like attrition, lost productivity, and unmet demand.

• Willingness to Change-Some firms may be more willing to change than oth- ers. This can impact the range of solutions you might want to implement. For example, for a firm that is not willing to change, they may be happy to give up savings in exchange for a minimal number of changes.

• Public Relations and Branding-This is especially important for firms with a highly visible brand. If one of these firms opens or closes a new facility (espe- ciallya manufacturing location), it can often make the news. These firms need to consider the public reaction and the impact on their brand.

• Competitors-A firm's supply chain can be impacted by the competition. Sometimes it makes sense to be exactly where the competition is, and in other cases it makes sense to be where they are not.

• Union versus Non-Union-Some firms have strong policies on union affilia- tion or union contracts and want the locations chosen to reflect that.

• Tax Rebates-Although taxes can be modeled directly as a cost (as product crosses borders or tax jurisdictions), there can also be rebates for locating a

18 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

facility in a particular location. This can be hard to quantify during the analysis, but it can be used for negotiating with the local tax authorities.

• Relationships with Trucking Companies, Warehousing Companies, and Other Supply Chain Partners-You may have supply chain partners (like trucking or warehouse providers) that you will not be able to work with in new locations. There is some value to keeping your existing partner relationships. You will need to consider who your new partners will be in the new configuration, or what you will need to do to get these new partners.

There are many more such factors. What is important to remember is that we are not just pushing the "run" button and coming up with the right answer. We want to run a variety of scenarios and then apply other criteria when finalizing the decision. When we finalize this decision, we can realize exactly how the quantifiable factors (cost and serv- ice) are impacted. This can help lead to discussions based on facts, data, and trade-offs (rather than gut-feel, intuition, and emotion).

Nonquantifiable Data: What Are the Organizational Challenges? A supply chain study must span many different areas of an organization: sales, opera- tions, logistics, finance, and IT.

The first challenge, aside from merely gathering all these people into the same room at the same time, is to understand and start to balance the different objectives that each group may have. As you can imagine, each of these groups operates with its own specif- ic goals and these may directly conflict with each other. There are many examples of the various groups' goals, so here are just a few related to what we'll cover in this book:

• Sales Team-Place product as close to customers as possible (create many ware- houses). Have small frequent shipments to customers (many small shipments or more frequent production runs at the plants).

• Operations Team (Production)-Produce large quantities of one product dur- ing each run in order to reduce machine downtime and changeover costs (cre- ates a need for a lot of warehouse storage). Produce product in one location to maximize economies of scale.

• Operations Team (Warehousing)-Quickly move inventory through the ware- houses (minimize storage costs). Minimize warehousing locations to reduce fixed and management costs.

• Logistics Team (Transportation)-Have large shipments on less costly modes of transportation (ocean, rail, or truckload).

• Chapter 1 The Value of Supply Chain Network Design 19

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

• Finance Team-Have the least amount of money tied up in capital (low levels of inventory and operations requiring the least investment in warehousing and production locations). Incur the lowest costs tied to logistics (transportation, warehousing) .

Understanding the different objectives of the different groups is important to any suc- cessful project.

The second challenge you have is collecting and validating data from all these different parts of the organization. The sales group must produce the appropriate historical demand data as well as dependable forecasts of sales in the future. The operations group will be needed to explain the costs, capabilities, and capacities of all the production and storage assets, as well as any related overhead and labor costs. The logistics group is also needed to provide not only current transportation rates but estimates of rates for new potential lanes resulting from a reorganized network. The finance department is depended on for comparing the output costs from the model to the costs within their . financial statements for the same span of time. Doing this provides a validated starting point for the model and a baseline to which we can compare all future model scenarios and output. This data may lie in different systems as well, which only adds to your chal- lenges and often requires IT help to sort out.

Data challenges also come when you are attempting to estimate data for the new poten- tiallocations and product flow paths. (Even though this data may be difficult to collect, this is often the whole point of a study-to consider new alternatives.) Transportation rates for new lanes, potential site costs, capacity, and capabilities, as well as the cost to shut down existing sites, must all be researched and calculated for consideration when we ask the model to make the best decision.

The third and final challenge comes after the modeling is done and you have come to the final decision. The final step of actually implementing the results can be a major challenge in and of itself. People in any company become very comfortable with a cer- tain way of doing things. As a result, it is not always easy to get them to see the "big pic- ture" and the value these changes will bring. Proper involvement from all of the previously mentioned teams within an organization throughout the entire project can assist with this, however, as each team understands the rules and constraints that they ensured were adhered to within the recommended solution. There are many great resources for you to learn about how to implement change like this in an organization, but this topic is beyond the scope of this book.

Making changes to a supply chain may also cause a temporary state of disruption. A supply chain cannot just stop at a moment in time and take on a new structure. It is often important to implement changes over a period of time to minimize the downtime and inconvenience that switching over operations may cause .

• 20 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Despite all of this, however, the more network design projects that companies complete, the better they get at addressing these challenges. And, their future models and recom- mendations improve as a result. Based on the savings we see from firms that develop this capability, we can say with surety that it is well worth it!

Where Are We Going with the Book? This book is organized to give you a set of building blocks to tackle any type of supply chain. We will start with very simple problems and build from there. This field is sur- prisingly deep however, and this book cannot cover every type of model to address every type of business that can benefit from modeling. We have found that advances in opti- mization technology and computing power have opened up new opportunities for answering different questions and using this technology for more and more decisions and we expect this to continue at an even faster pace in the years to come.

In fact, in just the past ten years, we have seen a dramatic increase in the complexity of models that people want to run and the frequency with which they want to run these models. This is a very positive trend as firms derive more and more value from these types of models.

To successfully deploy the more complex models and run them on a more frequent basis, we think it is critical that you understand the basic building blocks.

You will gain a lot of depth and insight from even the simplest problems. Then, later, as you tackle larger and more complex problems, the foundations you learned from the simple problems will continue to serve you well.

End-of-Chapter Questions 1. ABC Bottling Company's sales have been expanding rapidly. Their single plant,

which ships directly to customers, is now out of capacity. What factors should they consider when they decide whether to expand the existing plant or build another one? If they build another plant, what factors should they consider when they locate this plant?

2. A producer of dog food is trying to decide whether they should change the number and locations of their warehouses to better meet projected demand over the next three years. They do a study and determine that their transporta- tion and warehousing costs will be $51 million if they stick with their current structure. They have determined that if they close two warehouses and open two new warehouses, their costs will drop to $50.5 million. Assume that all other costs stay the same. Should they make the change?

• Chapter 1 The Value of Supply Chain Network Design 21

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

•

3. You need to set up a mathematical optimization model. Assume you are model- ing a supply chain for a business with ten warehouses and 1,000 customers. If you set up the model to minimize cost, set the decision variables to decide which warehouse should serve which customers, and set up no constraints, why would you expect the minimal cost to come back as $O?

4. You are helping a firm determine their future transportation costs between their plant in Dallas and their warehouse in Atlanta. Your best estimate, with the data you have, is that the cost will be between $1.70 and $1.80 per mile. You decide to use $1.75 as your cost because it is the mid point. If you are asked to spend more time seeing whether the number should be closer to $1.70 or $1.80, what would be your argument against further refining this number?

5. A small medical supply company in Australia has just developed a never before seen product with major pre-release orders from around the globe already. This company will need more production capacity to support their forecasted sales for this new blockbuster product. If they simply expand their plant in Australia, they estimate that their production, transportation, and warehousing costs will be approximately $450 million (ADD). After a careful network design study, they have found two solutions that people in the company generally like.

a. Solution #1: Estimated cost of $375 million with a new large plant in China to supplement their existing plant in Australia.

b. Solution #2: Estimated cost of $385 million with three neW smaller plants in China, Brazil, and Italy to supplement their plant in Australia. These plants would service their local regions.

(Assume the costs listed here include all the costs that are relevant.) What would be the best reasons for picking solution #1? For picking solution #2? Why is it important for this firm to consider other nonquantifiable factors when determining their best course of action for expansion?

22 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

12 THE ART OF MODELING

._----------------_._-_.- -----_._-----_ .. _--

As you've seen throughout this book, network design is about making a decision about your supply chain. When you are in the midst of analyzing data, are talking to different people in the organization, and are worried about the deadline, the most common trap you can fall into is thinking that the goal of the project is to exactly replicate the current situation. This thinking leads to too much effort in collecting data, too much effort in building a perfect model, and, in the end, failing to create a model that can be used to make a good business decision. In other words, if you try to exactly replicate reality, your model will be just as complicated and just as expensive as reality.

Instead, you need to keep in mind that the objective of the model is to make a good decision for the supply chain. This is why modeling is more of an art than a science. You've seen the science of solving these problems throughout this book. The art of these problems is setting up the models so that they can give the business the information it needs to make a decision and take action.

It should be noted that the definition of a good decision almost certainly includes some timeline. Businesses move fast. A great decision delivered too late is much worse than a good decision delivered in time to act. Keep in mind the adage attributed to Voltaire: "The perfect is the enemy of the good."

Of course, generic adages don't help directly with network design. In this chapter, we will explore the following specific areas that will help you build better models:

1. Start by understanding the supply chain.

2. Separate the trivial from the important.

3. Start with small models and iterate.

4. Run a lot of scenarios-don't be afraid to experiment.

S. Don't be afraid of including things in the model that don't exist in the actual supply chain.

• 217

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

6. Models are not a substitute for due diligence and decision making.

7. Optimization will do anything to save a penny.

The next two chapters cover the important topics of data aggregation and running a project.

Understanding the Supply Chain Before starting your network design project and data collection, it is important to com- plete a scoping session to ensure that all key stakeholders agree on the exact questions being answered, availability of data, aggregation schemes, and output expectations. This discussion and the decisions that come from it will shape the rest of the project from start to finish.

This session should include representatives from Production, Warehousing, Transportation, Finance, IT, and any associated Project Analysts. Ensuring participation from all associated areas from the start ensures significantly less pushback on solution implementation after you are well into the project.

A good early step is to draw a diagram of the physical structure of the supply chain on a white board to understand the current flow. Drawing this diagram helps the team to better understand the full scope of the supply chain. Different members of the team may have thought about only their portion of the supply chain. Seeing the full supply chain in one diagram is a great way to understand it and a great way to get various people in the organization to agree that it is correct.

For example, a retailer may draw the diagram shown in Figure 12.1 on the white board to show the flow of their product from the vendors to their vendor's warehouses (called DCs-distribution centers) and then their own DCs and on to the stores.

~e-+ Vendors Vendor DC

-+~ Retail ~StOreS

Regional DC --+

Retail ~ Central DC ~ ~

~StOreS

Figure 12. 1 Schematic of a Supply Chain for a Typical Retailer

Note that we have not drawn every store, but the general groups and categories of stores. In this supply chain, the retailer is looking back to its vendor locations .

• 218 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

With this diagram in place, the retailer can start to ask questions about whether they really want to include the vendors and vendor locations. They can ask whether they have missed any important flows in the supply chain. For example, does the vendor ship product directly to the stores, and is that important? If these retailers sell online, some- one on the team will quickly realize that this diagram has omitted the online customers. Also, as the team draws the diagram, you might realize that there are different types of stores that require different rules. For example, smaller stores in the city center may receive deliveries from an intermediate depot. Or, if the retailer is a grocer, you may need to include different warehouses for frozen, refrigerated, or ambient warehouse types.

A consumer packaged goods (CPG) manufacturer (or any industrial manufacturing firm) would have a different diagram. For example, a typical CPG company's supply chain may look like the schematic shown in Figure 12.2.

Raw ~ Finished

L MaterialS ~ ~dS

..•• i G. . ~ Mfg. Plants ~ ~ Retailer DC . ' semi-~ , • Finished

Raw Material Goods C ~ Vendors ~ onsumer

R .", Products DCs

Mat:~als "~.' .•. <.G .... ~ "'d ~ ~ Goods ~

Mfg. Plants 82C Channel

Figure 12.2 Schematic of Supply Chains for a Consumer Products Company

In this case, there is more focus on modeling the manufacturing process. The location of the raw material vendors can playa role in what product is made where and where plants are located. There may be multiple manufacturing steps that are performed at different locations. Also, note that the end customers include both the retailer's ware- houses (referred to as the Retailer DC-distribution center) and the B2C (Business to Consumer-or online) Channel. Many manufacturing firms are starting to sell their product directly online. This creates the need for a different type of customer. For exam- ple, they may ship in full truckloads to the retailer's DCs, but may ship small-parcel for the online business. Also, in this case, all product goes to a Consumer Products distri- bution center (DC) before moving to the customers. The team should also ask them- selves whether the plants can ship directly to the customers and bypass an additional distribution center.

An industrial manufacturing company, like a chemical company, will have the same general structure except their customers may be other manufacturing companies.

• Chapter 12 The Art of Modeling 219

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

A pharmaceutical supply chain may have yet another structure and it may have plants located around the world. The diagram shown in Figure 12.3 depicts a typical pharma- ceutical supply chain.

Ingredient! Raw Material

Vendors

API Mfg. FG/Bulk Mfg.

FG DCs Affiliates! Sales Channels

Figure 12.3 Schematic of Supply Chain for a Pharmaceutical Company

A typical pharmaceutical supply chain consists of a series of steps performed at differ- ent plants located around the world. The supply chain starts with sourcing of ingredi- ents and raw materials from specialized vendors, shipping products into an Active Pharmaceutical Ingredient (API) manufacturing location. The API is the chemical for- mulation that makes the drug work. The API is then shipped in bulk to a finished goods (FG) plant that turns the API into pills, tablets, and liquids that are then packaged in the right doses. The finished goods are stocked in a central warehouse (called FG DC in this case) that ships to their regional affiliates and various sales channels.

With the previous examples, you can see how supply chain structures can be very dif- ferent depending on different industries. They can also capture the unique characteris- tics of the company.

As you are creating the diagram, you want to ask as many probing questions as possible to make sure you understand the full modeling requirements. We have provided a set of questions, in the text that follows, as a good guide to help get you started. A typical approach that has worked well in the past is to get the group talking by asking everyone to describe their expectations for the project in general. Often many of the questions will be answered during this initial discussion. The provided questions can then be used to help further the discussion and to finalize official scope documentation. Though not all questions need be answered, the more information gathered the better. The output of this discussion also directly guides your next step, which is to begin the data collection process.

The following list of questions will help you define the objective, constraints, and deci- sion variables. These questions are just a list to get you started. You should make sure you ask plenty of follow-up questions and make sure you capture many details. Later, you will need to sort through the answers and separate the important from the trivial (more on this later in the chapter). But, for now, getting detail is important.

•

1. What is the objective of your network optimization study? For example, are you trying to:

220 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

a. Reduce transportation costs? Reduce total costs? Reduce costs after a merger?

b. Reduce the delivery time to customers?

c. Better use your capacity or add new capacity to cope with growth or a new channel?

2. What are the key decisions you want to make? What specific output should be available after the completion of this project?

3. Which portions of your network may change and which are fixed for the pur- poses of this study?

4. How much of your supply chain do you need to include in the study? Do you need to include both manufacturing and warehousing in the study?

5. What is the desired project start date? (Completion date?) What is driving the completion date? If the timeline is short, you may not be able to create a detailed model.

As part of the study, you also need to understand the data that must be collected. Often, understanding the data will help you ask better questions from the list just presented. That is, these questions should not necessarily be sequential, but should be iterative. To understand the data, it is often a good practice to focus your questions on the physical supply chain and the different types of products, customers, warehouses, plants, and transportation. Here is a starter set of questions:

1. How many different products or SKUs do you manufacture or distribute?

a. How do you currently categorize and group these products?

b. Where do these products come from? What are your options?

c. Will you need to consider the key raw materials that go into the finished goods? How do you categorize your raw materials?

2. How many warehouses do you have? Where are they located? Where do they get product? Who do they ship to? How are they different from each other (are some for all products, some for frozen products, some for online customers, and so on)? How do you measure warehouse capacity? Do you want to consider new warehouses? What are the fixed and variable costs of the warehouses?

3. How many plants and suppliers do you have? Where are they located? What do they make? Where do they send their products? How do you measure their capacity? Can the same product be made in multiple locations? What are the fixed and variable costs of the plants?

• Chapter 12 The Art of Modeling 221

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

4. How many customers do you have? Where are they located? Is there wide varia- tion in the demand of each customer? How can they receive product? How do you group your customers? Do you have different channels? Do the customers have different service requirements (need to be within 100 miles of a ware- house, need to receive product in three days)?

5. How do you ship product throughout your supply chain? Where do you use rail, truckload, less-than-truckload, parcel? What are your rules for when you use different modes?

Now that you have a working diagram of the supply chain and the details behind that diagram, it is time to start thinking about the model. By now, you know quite a lot (too much, maybe) and it is time to determine what to model.

Separating the Important from the Trivial Often, the difference between a model that leads to decisions that save the firm millions of dollars anq a modeling project that fails is the ability to separate the important from the trivial. If a modeling project tries to get down to every last detail, not leave out a sin- gle exception that happens in practice, or include every minute decision, most often it gets bogged down to the point of failure.

When we phrase it like this, it looks easy to avoid this pitfall. However, in practice, it is not so easy. There will be strong forces in play that will push you into more and more detail. For example, there will be a strong feeling that if the model does not represent the supply chain, we cannot trust the results and we cannot get others to believe our analy- sis. If we simply give into these forces and feelings, and try to model ~verything, our models will fail. Instead, we need to understand that these forces exist and work to address them and still build a practical model.

The art of modeling requires that our models capture the important aspects of a prob- lem, that they don't get bogged down in the trivial, and that we and others can trust the results.

There is no magic to separating the important from the trivial, and you may not even be able to do this prior to digging into the problem. Although the list is not comprehensive, here are some tips to help you develop better intuition for separating the important from the trivial:

•

• Realize that there are different types of decisions and aspects to the problem-Just realizing that there may be important and trivial decisions is a good first step. This allows you to realize that you can separate out different decisions and then you can decide which ones to include or exclude later .

222 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

• Remember the 80/20 Rule-This common rule of thumb helps you realize that 800/0 of your sales come from 200/0 of your customers, 200/0 of your products account for 80% of the business, 200/0 of the supply chain changes will lead to 800/0 of the savings, and so on. It is not a hard-and-fast rule, but helps remind the team that if you are spending most of your time trying to understand an aspect of the business that accounts for 1 % of the supply chain costs, you may not be focused on the right things.

• Exceptions are captured in an optimal baseline-Almost every supply chain has many different exceptions and special cases. These can be very difficult, if not impossible, to capture. One thing to keep in mind is that we take care of some of this problem with the optimal baseline. In the optimal baseline, we are coming up with a solution to the supply chain that assumes that there are no exceptions. Then, we compare our optimization runs to the optimal baseline. It is not that we assume that exceptions won't happen; it is that we know they will, and that is why we compare back to the optimal baseline. See Chapter 8, "Baselines and Optimal Baselines," for more details.

• Address the areas the model does not cover-Even if we have removed the trivial aspects from the model, it may still be in your best interest to make sure you cover these aspects. Covering it does not mean modeling it. It could mean adding in the impacts after the model runs, or it could mean just discussing this aspect during the presentation of the results.

Start with Small Models and Iterate We have found that starting with small models and incrementally creating ever- more-complex models is an effective way to get to a final model and final decision.

The outline of this book provided you with different building blocks you can use to start simple and add complexity. You saw how you could start with the simple center of grav- ity model to determine the best supply chain considering just service. Then, you could add complexity by adding one element at a time like

• Capacity

• Transportation costs

• Facility costs

• Multiple echelons

• Chapter 12 The Art of Modeling 223

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

The benefits of this approach are these:

• It allows you to test your data and model incrementally.

• It enables you to better explain your final results if you know the impact that each new element has on the model.

• It allows you to separate the important from the trivial by being able to see the effect of different elements of the model. If something has no impact, you might consider leaving it out.

• You may find that you come up with your answer before you have to build the full model. That is, as you iterate and build more detailed models, you may find that you have a model that answers your questions with less complexity than you were planning on adding.

Run a Lot of Scenarios-Don't Be Afraid to Experiment The benefit of creating a model of your supply chain is that you can test different ideas, different data sets, and different strategies. It is important to note the alternatives to building a model before making a decision:

• Just implement the change and see what happens. This has the advantage of being very real-you know what the results are. Of course, this comes with the big downside that if you get it wrong, the firm could be financially wrecked.

• Use intuition and judgment. This has the advantage of being quick, but the big downside is that it does not take advantage of data nor does it allow you to sort through the myriad of possible solutions.

Many firms rely on some version of these alternatives. It should be no surprise that the resulting supply chains leave a lot of money on the table or don't support the overall business strategy.

With a model you have the ability to do much better, but only if you make use of one of the big advantages of a model-the ability to run a lot of scenarios. Often these scenar- ios are called "what-if" analysis. You can test what would happen if something were changed. And that "something" is up to your creativity.

After you build a model, it is relatively easy to run different scenarios. Besides the debugging benefits we discuss later in this chapter, running scenarios allows you to gain a deeper understanding of your supply chain and better prepare for future possibilities.

Most supply chains are relatively complex with many different costs and parameters. By running the multiple scenarios, you can start to see how changes to one part of the sup- ply chain impact another part, you can see how the variables interact, and you can get a

• 224 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

sense of the important variables. You may also get other, unexpected ideas on how to improve your supply chain. One company we worked with wanted only the lowest cost network. However, when they found a solution that was just $3 million more expensive than the lowest cost solution but doubled the number of customers they could deliver to within one-day, they opted to implement the latter solution.

Also, by running different scenarios, you can prepare better for the future. When we are building a structure for our future supply chain, we have to make our best guesses about demand and costs in the future. So it is always best to test what would happen if demand went up or down, if it varied by region or part of the world, or if we lost or gained a new customer. It is also good to analyze what happens if costs change. For example, what would happen if the price of oil were to drop or rise significantly?

Part of the art of modeling and coming up with good answers is to not only test these varieties, but then come up with better answers. One way to think about a better answer is to come up with a robust solution. A robust solution is one that is reasonably good across a wide range of changes to the data.

For example, we worked with a firm that had designed an optimal supply chain that reduced transportation and warehousing costs by $4 million with no changes to service level. They were ready to implement this solution. However, when we analyzed the solu- tion, we realized that the savings were being driven by one very large customer. If this customer was lost or its demand shrank, the new network would have been $500,000 more expensive than the existing structure. However, by coming up with alternatives, they found a solution that would still save $3.2 million if the new customer stayed and $2.8 million if they lost the new customer. In this case, the $800,000 of savings that they gave up was good insurance and made for a more robust solution.

Don't Be Afraid of Including Things in the Model That Don't Exist in the Actual Supply Chain When you realize that the underlying mathematical optimization engine solves a prob- lem in a systematically ruthless way, you can take advantage of this fact to help you make better decisions and make more efficient models.

For example,

• One way to trick the optimization into maximizing the number of customers within 400 miles of a warehouse is to set up a transportation rate structure so that all customers within 400 miles cost $1 per mile to service, and those out- side of 400 miles cost $10 per mile to service. So if one customer is 399 miles away from the warehouse, each truckload costs $399. If it is 401 miles away, the cost is $4,010. If you don't put any more costs into the model, the optimization

• Chapter 12 The Art of Modeling 225

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

engine will work very hard to get as much demand as possible within 400 miles. The costs of $1 and $10 are purely to guide the optimization to make the deci- sion you want it to make .

• In our discussion of transportation costs, we mentioned that some transporta- tion rates could be quite flat. That is, it may cost the same amount whether the load is going 10 miles versus 200 miles or 400 miles versus 800 miles. This can happen when you ship small-parcel, or ship with a third party with an extensive network in place and the marginal costs to them are just the extra touches-the vehicles are already moving. In this case, the customers may not be assigned to the closest warehouses. One quick trick is to add a small surcharge to all such shipments of something like $0.01 per mile. This cost will be trivial in the over- all scheme of the model, but it can be enough to break the tie when the rates are the same between two warehouses and assign the customer to the closest facility. No such cost exists in the real world and you need to make sure that it gets sub- tracted later (unless it gets rounded out when you report on significant digits).

• For some models focused on the U.S. with significant imports from Asia, you may model the supply location as the port of entry. For example, you model the supplier or plant in Long Beach, California. There is no such plant there, but it helps simplify the modeling without impacting the decision.

This list could go on and on. The key idea is that if you understand the underlying opti- mization engine, you can set up structures to get the model to help make the key trade- offs you need to make.

Models Are Not a Substitute for Due Diligence and Decision Making All through this book, we have been talking about the value and importance of network modeling, so this point may sound like minimizing the value of modeling. However, it is important to remember and remind others about the concept of modeling. A model by definition is meant to be a representation of a real entity, be it a facility or equipment or a supply chain. No matter how detailed or precise (again going by our previous ref- erence to precision in modeling) a model is, it can never represent reality entirely. All network models will have some degree of assumptions and simplified rules. As a result, it is important to use the outputs of the model along with other factors and analysis to make decisions. Network modeling is designed to provide decision support and is not a substitute for required due diligence and decision making. It is important to remember that the results are based on best available data and estimates at the time and may like- ly change in the future .

• 226 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

We worked on a network study for a company where the final results recommended closing a warehouse in New Jersey and opening a new warehouse in Memphis. The existing cost with the New Jersey warehouse was $10 million and the modeled baseline cost was $9.5 million; and this was good enough for decision making. When the model ran, the Memphis solution in the model came out to $7 million with the baseline demand. After the recommendations were accepted and the implementation was started, the Finance team revised their logistics budget for the following year and set it to $7 million.

What is wrong with this? The $7 million cost estimate was based on assumptions and estimates such as historical demand and nonnegotiated transportation rates. The model was designed to answer questions on the distribution network strategy-the scenario costs are meant to help quantify how much better the Memphis solution is compared to, say, a Chicago solution with a total cost of $10 million. This model was not meant to serve as a budgeting tool, and is not app'ropriate to use as a substitute for due diligence and analysis that goes with a budgeting process.

Optimization Will Do Anything to Save a Penny When you set up a network design model and hit the run button on the optimization, the optimization is going to work as hard as possible to find the absolute best solution. If the optimization can find a solution that is one penny cheaper, it will do so.

A good modeler will keep this in mind.

This can impact how you set up the model. For example, you may want to artificially inflate the cost of opening a new plant to ensure that if the model does open the plant, it is a good decision (see the earlier section on including things in the model that don't exist in reality).

This can impact how you analyze results. Specifically, this points back to the need for multiple scenarios (or what-ifs) and due diligence to test whether the model did, in fact, make a decision to save a penny. We have seen that executives have good intuition with regard to this. In our experience, if we show the executives the best four-warehouse loca- tions, they will immediately ask what the cost is of the best three and the best five. Also, if one of your locations is Birmingham, Alabama, they will ask about the cost if it is in Atlanta. They are, in effect, trying to figure out whether the model is recommending one solution over another to save a penny.

• Chapter 12 The Art of Modeling 227

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Debugging Models Earlier in the book, we saw how we can set up the model in ways that prevent a successful run. In the chapter on service levels, we saw how we set up a constraint that said all cus- tomers had to be within 800km of a warehouse and that we could have only three ware- houses. There was no way for both of these things to be true at the same time. Supply chains are complex and the models you build will reflect that complexity. When you set up these models, you want to make sure they are correct. Sometimes the problems are obvi- ous, but smaller, less obvious problems with your model may also lead you astray.

When a good software developer writes a program, they work hard to get the logic cor- rect from the start, they double- and triple-check the logic, and then they thoroughly test the program with many data sets. There is no shame in finding a bug during the development process. Programs are complicated, with many different interactions. This is why the testing of programs is just as important as writing them. In fact, it is impor- tant to note that programs are tested during the writing phase. You do not wait until you have a program to test; you should test it throughout development.

Modeling your supply chain is a lot like writing a software program. Of course, com- mercial-grade network design software is already debugged. But when you set up your supply chain within these programs, you need to make sure it is doing what you want it to do. To ensure that it is working properly, we suggest that you follow the steps a pro- grammer goes through.

To start, you should make sure you understand the key elements of the optimization model you have set up. These elements are explained in Chapter 1, "The Value of Supply Chain Network Design:' but are worth reviewing:

• Objective-What are you optimizing and how do you compare one solution with another?

• Constraints-What are the business restrictions you want to model?

• Decisions-What are you asking the model to do and what choices are you giving it?

• Data-What data will you load into the model and is the data what you expect it to be?

Taking a page from good software development, you should thoroughly test your model. Our experience has shown that people are hesitant to test until they have a com- plete model. We suggest you test throughout the process. In this case, testing a model consists of running a scenario. The scenario need not be complete. By testing and run- ning various scenarios, you can test the logic of your model and the quality of the data. And, as a side benefit, you may also learn a few things about your supply chain that are worth investigating .

• 228 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

What are you looking for when you test? When you run a scenario, you are answering a few basic questions:

1. Did the model come back with a feasible solution? That is, did the optimization problem return a solution or did it come back with a message that the model is infeasible (commercial-grade packages can provide some messages as to what the problem may be).

2. Is the result doing something that makes sense? That is, did it pick locations that make sense, does the flow of product make sense? Does the solution follow the business rules we set up?

3. Do the output values look correct? That is, do the costs make sense?

Even if these three questions can be answered in the affirmative, you still want to test other possibilities. Remember, these models are complicated and there could be hun- dreds of millions of different possible solutions. You want to make sure that bad logic or bad data is not preventing other better solutions from surfacing. Running a lot of sce- narios and forcing different solutions allows you to look for hidden problems that do not immediately surface. If during these tests you come up with a negative answer to one of the preceding questions, you need to dig deeper into your model.

It is also worthwhile, in the early stages of model development, to develop a feel for the nature of optimization. Remember, at the heart of a network design tool is a purely mathematical engine that will inexorably seek the optimal solution. Model analysis and troubleshooting sometimes resembles reverse electronic psychoanalysis, as the flesh- and-blood modeler attempts to deduce the motivations that drive her digital advisor to make such odd decisions. It is better to encounter these conundrums with the simpler and faster models that arise early in the planning process. Thus, the modeler should dedicate some time to playful experimentation throughout the development of industrial-sized models, in order to familiarize herself with the "personality" of the opti- mization problem she is building.

Fixing Infeasible Models If you are at the point where you cannot positively answer the first question, you obvi- ously do not get a chance to evaluate the second two questions. So when the model does not run, what do you do?

First, a quick note about infeasible problems: As you have seen throughout this book, these problems are solved with mathematical optimization. If you simply write a math- ematical optimization program, and feed the model data and constraints that conflict with each other, then the model will simply return an answer to you which states that the problem is infeasible. It would be up to you to try to determine what went wrong .

• Chapter 12 The Art of Modeling 229

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Fortunately, commercial-grade network design software can help you pinpoint some problems. First, you know that the software itself has been tested, so you need to just focus on your model and data. Commercial-grade network design software has the abil- ity to pick up some types of problems automatically. For example, if you load a demand of 100 and a total supply of 90, you will get a message that demand is greater than sup- ply. There are a lot of cases like this. However, not everything can be pinpointed with- out deeper analysis.

You should be aware that not all contradictions can be found in such an easy way. In the example we mentioned at the start of this chapter, every customer was within 800km of at least one warehouse and most customers were within 800km of several warehouses. So you need to run the optimization to sort through the combinations to see whether there is some choice that can meet both restrictions. In more complex models, you may have many such potential contradictions and need the model to sort the combinations.

So for these more complex models, you may still need to do some debugging to get to a feasible model.

One technique we have found to be helpful is to run the model in a mode that allows you to find the least infeasible model by relaxing certain constraints. Some commercial- grade software will have this feature since it requires a difficult reformulation of the problem. This then allows you to determine how close to feasible you are. For example, are you able to meet 99% of the demand or 5O/o?

Moreover, a commercial-grade tool will include a «bottleneck" report that helps you identify the blocking constraints that are limiting your model's ability to meet demand. Thus, if a partial solution is generated that meets 850/0 of demand, but whose only fully utilized facility is the Cleveland warehouse, then you are at least oriented toward a worthwhile starting point in the debugging process. This particular infeasibility might result from the modeler's underestimating the capacity of the warehouse in question, or it might result from some more indirect problem, such as a shortage of shipping choic- es for Midwestern customers. But it would seem reasonable to conclude that the pro- duction capacity and shipping lanes associated with the Mexican production facility are likely not to blame.

Another approach is to create some high-cost dummy facilities and dummy transporta- tion costs that are set up in a way that they will surely be feasible, but come at a high cost. If these facilities or transportation lanes are used, it can help point to potential problems.

Usually, infeasible models are the result of constraints that are too tight, constraints that contradict each other, or complex models with missing links. For example, when we define that we want three warehouses and every customer has to be within 800km, we may have constrained the problem so much that there are no possible solutions .

• 230 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Constraints that contradict each other usually occur when you combine constraints that force a minimum at some point and a maximum somewhere else. For example, you may specify that a plant make at least a minimum of 100 units of a certain item. But because of other capacity restrictions (maybe not at the same location), it may not be possible to make 100 units at that location. The more complex the model, the harder these contra- dictions are to spot. Our experience has shown that examining the minimum con- straints for problems is a good place to start.

Missing links in the supply chain are easy to spot in simple models, and commercial- grade software can pick this up. The problem occurs in complex models with multiple echelons. In these cases there may be links between all the pairs of sites you are interest- ed in. But, based on the way you set up the model, there are certain products that need to get to certain destinations, but cannot do so because of the way you have set up the different flows.

After you have a model that runs and produces answers, you need to address the next two questions (Do the results make sense? and Does the output look correct?).

Fixing Feasible Models When a model is feasible but is producing results that do not make sense, you need to investigate. This is true even when the results don't make sense just a little bit. We have seen many models that have returned very counterintuitive results that turned out to be true. But to convince people they are true, you need to understand why the model gave the answers it did.

We worked with a company that had experienced rapid growth in Asia and needed a production plant in the region. Demand was growing fast in all parts of Asia, but China was by far the largest market. At the start of the project, the team thought that a plant in China would be the best solution. However, when they ran the network design model, the optimization runs kept suggesting a plant in Vietnam. The company was convinced that there was a mistake with either the model or the data. We had to investigate what was causing these results. It turned out that Vietnam did have higher transportation and duty costs, but because of the low cost of manufacturing and the lower fIxed costs, it was indeed the best solution from a cost perspective.

Not all stories turn out like that. Sometimes the strange results just mean that the model is not set up correctly.

This book cannot list all the problems you might encounter. However, keep in mind that the complex model you are debugging is nothing more than a combination of the dif- ferent building blocks we talked about in this book. If you can understand each of the individual elements, it makes the more complex model less intimidating. We have found

• Chapter 12 The Art of Modeling 231

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

that a systematic way to look for problems is to focus on the main elements of an opti- mization problem. Each of these elements tends to lead to different types of problems. Let's take each one in turn:

Fixing the Objective of the Optimization Some problems with models are caused because the objective of the model is not set up like you want or you have not captured some important business aspect of the problem. Remember, the objective function is what guides the mathematical optimization engine and allows the engine to compare one solution to another. The optimization engine does just what you tell it to do, and nothing more.

So if you have an objective to minimize cost, the mathematical optimization engine will return a solution with the lowest cost. If you are looking at the output of this model it may look as though there are "better" solutions that will serve customers better. That is, a customer in Chicago may be getting their product from the warehouse in Dallas and not from the nearby warehouse in Indianapolis. However, you must keep in mind that the optimization engine did not ever consider service in its calculation.

If your objective function includes information about only transportation cost, do not expect the warehouse costs or plant costs to be considered. In fact, a solution based sole- lyon transportation costs may turn out to make bad decisions on the warehouses or plants.

Also, keep in mind that the optimization does not have a mechanism to break ties. To the optimization, a tie is a tie, and whichever solution it finds first, it will return. This can often come up when you have relatively flat transportation costs. That is, it may cost the same to ship to a customer that is 100 miles away as it does to ship to one 500 miles away. So when the mathematical engine chooses to skip the warehouse that is 100 miles away and use the one that is 500 miles away, that is perfectly valid.

Fixing the objective is not always as straightforward as you might think. Sometimes, you can simply add a missing cost element to the objective and the model works fine. Other times, you may not want to add the missing cost, but simply add a constraint to the model to encourage more reasonable solutions. We discussed this earlier in the book when we excluded warehouse costs but put in constraints to limit the number of ware- houses it picked.

The reason that fixing the objective is not straightforward is that you sometimes have objectives that cannot simply be added together. For example, you may want to mini- mize cost and minimize the weighted average distance to customers. We cannot simply combine these. In the section on multi-objective optimization, we talked about the abil- ity to automatically run different objectives at the same time to develop a trade-off curve between the two .

• 232 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

You will find that many real-world problems you will encounter involve multiple objec- tives. Building a trade-off curve between these objectives is a good way to fix problems related to the objective.

Fixing the Constraints of the Optimization When setting up the constraints, you need to strike the right balance between giving the model enough constraints to capture the reality of the business and restricting the model too much so that it is very limited in what results it can return.

In the case where there are not enough constraints, the model tends to return results with low costs but unrealistic business results. That is, the model is doing things that are just not possible. In this case, you need to tighten up the model by adding constraints and restrict- ing things that cannot happen in practice. There is a side benefit to not having enough constraints: Sometimes it can allow you to see possible solutions that you didn't think were possible. On more than one occasion, we have seen results from solutions that the client did not think were possible become possible when they saw the potential savings.

In the case where you have overrestricted the model, the results you are getting do not seem to be significantly different from the current state. That is, the model does not have enough freedom to find different solutions. One way this can happen is if you assume that the way things are done now is a constraint and not just a business practice. To avoid this problem, when you are building your model, you want to make sure that current practices are not necessarily listed as a constraint, and you want to make sure that constraints are real constraints. A real constraint is a real physical limit or a legal or contractual issue, or some rule that the company really cannot violate. If you vet your constraints carefully, you can avoid this problem.

Fixing the Decisions of the Optimization The decisions or choices you allow the model to make are similar to constraints. If you do not give the model many decisions to make, you are limiting potentially different results and are likely seeing very similar answers to the current state of your supply chain. Often it is important to allow the model a variety of decisions to see which ones it takes.

We worked with a client who was looking to open or close warehouses. After many dif- ferent runs, they found savings ranging from $300,000 to $500,000. Then, they allowed the model to make decisions on which plants to open or close. The savings went up to $7 million. They were not ready to make those kinds of decisions, but the model told them about the opportunity. After about 18 months, they changed the locations of their plants based on the model results.

Sometimes giving the model «too many" decisions can be a good thing that leads to bet- ter results.

• Chapter 12 The Art of Modeling 233

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Fixing the Data of the Optimization When debugging a model, we cannot forget about the data. Even though this is the most obvious and is where a lot of problems are found, you should not ignore the other ele- ments of an optimization problem.

There are two main problems with the data:

1. It is wrong.

2. It is not accurate enough (go back to our discussion on significant digits).

When we say the data is wrong, we are thinking of possibilities like the following:

1. You have typos-you put in 1,000,000 when you meant 100.

2. You put in 100 when the correct value is 500.

3. When you moved data around, you accidentally assigned demand for Product #1 to Product #2.

4. Because of the way data was pulled, you are missing data for an entire division, product family, or type of customer.

While we can't provide a magic bullet for avoiding the tedious work of checking the data carefully and doing normal data cleaning work (double-checking, looking for high and low values, looking at the top 200/0, sorting the data by different criteria-customer, product, region, and so on), we have found that running a baseline and a lot of test sce- narios is actually a good way to find problems with the data.

In Chapter 8, we talked about using the current state as a way to validate the model and data. So if the costs match up reasonably well, that is a good way to ensure that you are not missing data or have not misplaced decimal places.

By running the model a lot and forcing certain decisions, you can also look for problems with the data that does not get used in the baseline. That is, when you are optimizing, you may be using plants, warehouses, and transportation moves that are not used in the baseline. By running scenarios, you can see how this data behaves.

Fixing inaccurate data mayor may not be an issue. Before you fix inaccurate data, you should review the discussion on significant digits in Chapter 1. In many cases, it may turn out that inaccurate data is actually as accurate as it needs to be. In other cases, it is important to run scenarios in which you change the data in question by +/-100/0, +/- 200/0, and so on, to see how a data element actually changes the decisions. If the model does not change much as you change the data, this tells you not to focus on this data. If the model does change, you can either refine your model or make sure that your final results are robust even if this data does change .

• 234 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Lessons Learned for the Art of Modeling In this chapter, we introduced the art of modeling and debugging. Although you use optimization to solve these problems, a lot of thought goes into how you set up these models. This is the art of modeling. This chapter is a good starting point for learning the skills of setting up a good model and properly debugging it. But it takes practice to become a good modeler. If you combine what you learned in the first part of this book-what the finished models look like and how they behave-with the lessons from this chapter, you will be off to a fast start.

End-of-Chapter Questions

1. If you are creating a model and one of the key considerations is which port of entry you should use, why would you model the port like you would model a warehouse? What costs would be relevant for picking which port to use?

2. You are creating a model for a manufacturing company with four major plants and ten warehouses. They want to reduce costs by determining the optimal number and location of the plants and warehouses. During the initial set of meetings, one of the team members is very concerned about modeling the "samples" that the firm sends out. As part of the marketing of the product, the firm sends out quite a bit of sample product. This product is pulled from the plants or warehouses and shipped to customers via small-parcel. It accounts.for about 100/0 of the demand records (and about 0.50/0 of the demand volume) and about 30/0 of the logistics costs. Should you include this part of the business in the model?

3. Open the model found in the file Debugging Example I. zip on the book Web site. In this model, the customer wants to minimize the average distance to cus- tomers. However, when the model runs, one warehouse serves the majority of the customers even though there are other warehouses that could be used and are closer to the customers. Why is this happening? How would you fix this model to remedy the situation?

4. Open the model found in the file Debugging Example I I. zip on the book Web site. This model will not solve. Determine what is preventing the model from solving and how you would fix the problem.

5. Open the model found in the file Debugging Example I I I. zip on the book Web site. This model solves, but is returning a cost that is about 100 times high- er than what you expect the result to be. Determine what the problem is and suggest how you would remedy the situation.

• Chapter 12 The Art of Modeling 235

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

•

6. You are starting a network modeling project. When you ask the project manager what the objective of the study is, you get the answer, "We want to optimize the entire end-to-end supply chain." Why doesn't this answer help you? What addi- tional information do you need?

7. Why shouldn't your model be an exact one-to-one replica of the supply chain?

8. Using the diagram of the retailer from this chapter, draw the diagram of a retailer in which the urban stores are served only from regional warehouses, the rural stores are served from either the central warehouse or regional warehous- es, and the online customers are served from a standalone warehouse that receives product directly from the vendors .

236 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

13 DATA AGGREGATION IN

NETWORK DESIGN

In this chapter, we will focus our attention on one of the most interesting, yet challeng- ing parts of network design-data aggregation (we'll refer to it as just "aggregation"). Aggregation is putting the data of your supply chain into logical groups for the purpos- es of modeling. For example, instead of modeling every single item that moves through the supply chain, our model may contain only a small handful of product groups. When we aggregate, we typically aggregate the following:

• Products (from many different individual products to a small number of prod- uct families)

• Customers (from many different actual delivery locations to several hundred aggregate points representing all the customers in a geographic area with other similar traits)

• Plants, warehouses (that are a different building but on the same campus), and vendors (from hundreds of vendors to a small group of important vendors or vendors in the same geographic area)

• Time periods (modeling a year instead of every single day)

• Cost types (from many different line items to a single cost figure)

One of the reasons that aggregation is interesting is that it is both natural in how we think about problems and scary because we worry that we aren't being accurate.

Throughout this book, we have used aggregation strategies without calling them out. Al's example grouped all types of products into one weight based product group; the small- parcel example looked at shipments to about 200 customer points representing 10,000 or more actual shipment points; and so on. In each of these cases, we thought of the models as good representations of the supply chains. They seemed natural. Also, when we diagram a supply chain on a white board, we think of elements of our supply chain

• 237

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

in groups. For example, we can draw flows for products that travel in full pallets and flows for products that travel in bulk containers. Again, all very natural.

However, when it comes to actually aggregating the data to make decisions, we lose our nerve, forget what is natural, and get scared thinking that if we don't directly load all 10,000 products and 45,000 shipment points into the model, it won't be accurate.

We need to get over our fear and learn the techniques of aggregation. Only firms that sell a very small number of products or ship to a small number of endpoints can avoid it. And even these firms may want to aggregate.

First, there are some technical reasons for why you want to aggregate:

• The optimization models we've covered in this book can be difficult to solve. If you simply loaded every product, every vendor, and every delivery point, you will likely find yourself with an optimization model that does not fit into the memory of the largest computer and has no chance of solving.

• If you are running a model for future decisions (and especially those more than several months or a year out), your forecasts at individual products or sites are not close to accurate.

On top of the fact that you technically can't avoid aggregation (which, sadly, won't be enough to convince some people), there are some very practical reasons you want to aggregate:

• Cost and time involved in obtaining, processing, validating, and analyzing data at a very detailed level is prohibitive.

• It can be impossible to understand the big-picture model and the influence and interaction between the key data elements if you are working at a detailed level.

• When the final decisions are made, the executives are going to go back to the natural way of thinking about the supply chain and think about the model in aggregate terms.

Because you cannot avoid aggregation, don't fear it. When aggregation is performed well, models will closely represent the real world without sacrificing accuracy, but at the same time provide manageable and understandable outputs that lead to sound decisions.

Aggregation is also interesting because there is an art to it. The lessons we learned in Chapter 12, "The Art of Modeling," also apply to aggregation. So you will need to use your creativity to solve problems. But there are also some best practices and science we can apply to make our aggregation decisions better. We will cover this later in the chapter .

• 238 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Finally, this is interesting because there are many ways to approach aggregation and they may be different in each model or project. And the same business may use different aggregation strategies to answer different types of questions.

It is important to start thinking about aggregation early in the project-as early as the first discussions about the model. Following are some key questions to ask and consid- er to help drive the right strategies:

• What exactly are we looking to solve for, and what portions of the supply chain will we realistically be able to change? By answering this question, you can get a feel for which data elements are more or less important. If a data element is less important, it is a good candidate for aggressive aggregation. If you are answer- ing questions about plant locations relative to warehouses that won't change, it might make sense to roll all your customers to the warehouse that serves them.

• What data is available? If there are areas where you don't have good data, you might as well aggregate.

• Are there detailed constraints to model? In general, the more detail you want to model in the constraints, the less aggressive you can be in aggregation. If you want to build constraints for just your top customers, you have to split your customers in a geographic area into at least two groups (one for the top and one for the others).

• How are the transportation rates applied? If we are looking at simple trans- portation structures, it may make no sense to add extra details. For example, if we are looking at minimizing the delivery cost of products from our warehouse to customers and all products travel in full pallets, then we might not need any detail on the products.

In the next few sections, we will introduce aggregation strategies for each of the ele- ments of the model as well as discuss each of these strategies in more detail.

Aggregation of Customers Customers are one of the most commonly aggregated elements in network design mod- els. As a reminder, we are using the term "customer" to refer to ship-to locations, so if you ship product to a firm with ten locations, we are talking about the ten locations. A typical supply chain will include hundreds or thousands of customers to which prod- ucts are distributed and where they finally get consumed. These sites may even change from year to year as the customer base shifts. Customer aggregation usually comes down to geographic proximity and types of customers.

• Chapter 13 Data Aggregation in Network Design 239

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

It is pretty clear that customers that are physically close to each other should be grouped together. That is, if you have 25 customers in Denver, you might consider grouping these customers together. Then, if you have 3 distinct types of customers, you may split the Denver customer group into 3. You've gone from 25 points to 3 and still have the essence of your problem.

The most common strategy for grouping customers by geography is to use some form of the ZIP or postal code, or a metro area, or regional area.

For U.S .-based analyses, there are too many five-digit ZIP Codes, so it is most common to aggregate customers by three-digit ZIP Codes. (i.e., the first three digits of the stan- dard five-digit ZIP Code). To pick the point to plot on the map, it is most common to use the largest single customer or five-digit ZIP Code. In general, there tends to be more three-digit ZIP Codes in largely populated areas and fewer in sparsely populated areas. The maps in Figure 13.1 show examples from Illinois and Montana.

Aggregation for State of Illinois Aggregation for State of Montana

Figure 13.1 Example Showing Aggregation at Five- and Three-Digit ZIP Level for Illinois and Montana

In the U.S., most transportation rate structures-especially small-parcel and LTL, are defined or estimated at the three-digit ZIP Code level. So, in one sense, you lose no accuracy by grouping by three-digit ZIP Codes.

In Europe, two-digit postal codes can be used like three-digit ZIP Codes in the U.S . Then, in all regions of the world, cities and regions can be good geographic zones. When doing worldwide studies, you may even aggregate all the customers in one country to a single point. Or if you have a model that is mostly focused on the U.S., you may aggre- gate your non-U.S. customers to the port of exit. The actual method chosen largely depends on the quality of data and how well it ties to the transportation rate structure .

• 240 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

In some countries, aggregation to the district or county level may make sense if aggre- gation by city does not yield a significant reduction in the number of customers. Figure 13.2 shows the difference in aggregation at city versus district level for locations in India.

Figure 13.2 Example Showing Aggregation of Locations in India at City/Town and District Levels

For all models, it is very common to remove outlier and low-demand locations with insignificant demand that will always be difficult to service. For example, Alaska, Hawaii, and Puerto Rico in the u.s. are either modeled at the nearest continental u.s. port or excluded from the model. Or simply removing the low-volume demand points within your data may also save you time and effort.

In addition to geographic strategy, customers can be classified and aggregated using other key categories. You can think of this as stacking several points on top of the geo- graphic aggregation. For example, if you aggregate by three-digit ZIP Code, you may have two points at each three-digit ZIP Code representing the two types of customers you want to model. Some examples of categories include these:

• Required service levels-For example, next-day customers and three-day cus- tomers. This provides the ability to identify and model different service levels and see the impact on the network strategy .

• Shipping methods-For example, LTL, TL, Rail. This allows the user to easily apply transportation rates and business rules by classifying customers that are served by LTL shipments versus TL shipments. See Figure 13.3 as an example .

• Chapter 13 Data Aggregation in Network Design 241

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Customer 1 Norman OK 243,535 45,464 4,354 83% 15% 1%

Customer 2 Ft. Lauderdale FL 75,645 4,643,546 476,464 1% 89% 9%

Customer 343 Des Moines IA 3,235 ,325 5,262 ,352 3,253 ,252 28% 45% 28%

Customer 445 Carson City NV 43,224 24,242 252,552 14% 8% 79%

Customer 134 Bridgetown CT 76,433 63 ,536 767,465 8% 7% 85%

Figure 13.3 Example Showing Breakdown of Volume by Customer by Mode

• Type of delivery location-For example, a store or an online customer. This provides the ability to track and apply specific business rules based on the type of end consumption point for the products.

You can be creative in coming up with customer categories. Just be sure you don't cre- ate so many categories that you have more customers than in your original data set.

Also, don't forget about the art of modeling and coming up with creative solutions. For example, there is nothing that says that when you decide to aggregate customers you have to aggregate all customers. If you have 20 to 30 customers that account for a large portion of your demand, you may simply model those customers as they are and aggre- gate all the others.

Validating the Customer Aggregation Strategy- National Example We may need to validate the accuracy of our aggregation strategy before others in the organization will accept it. We'll do two tests, one at a national level and one more detailed for a region. You may need to come up with other ways to validate your aggre- gation strategy.

This sample model has the following data elements:

• One product sourced from one plant.

• Our objective is to minimize transportation costs only.

• Consider two sets of potential warehouses:

1. 26 potential warehouses-major u.s. distribution points. 2. 60 potential warehouses-major u.s. cities/metropolitan statistical areas .

• 242 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

• For customer data we will use U.S. population data obtained from the U.S. Census Bureau Web site (http://factfinder2.census.gov/faces/nav/jsf/pageslindex.xhtml). We will model this data at two levels of aggregation with this model:

1. Top 16,000 five-digit ZIP Codes with a population of at least 3,000 residents. We will exclude ZIP Codes in Alaska, Hawaii, and Puerto Rico for this example.

2. ZIP Codes aggregated to three-digit ZIP Code level. This yields a total of 876 three-digit ZIP Codes. For the model, we will pick the largest five-digit ZIP within each three-digit ZIP for geocoding.

• The total demand is the same in both cases.

For the first model, we will load data for 16,000 customers and 60 potential warehous- es. For the aggregated model, we will load the model with the 876 three-digit ZIP Codes and the top 26 cities as potential warehouse locations.

The charts in Figure 13.4 give us a graphical view of the inputs for the two customer aggregation strategies. We can see the significant difference in density of customers between the two models. However, the real question is whether this increased level of detail offers any significant benefits in terms of the quality of outputs and associated decision making.

Total Customers: 15,927 Total Potential Warehouses: 60

Total Customers: 876 Total Potential Warehouses: 26

Figure 13.4 Comparison of Input Data-16,OOO Versus 800 Customers

For purposes of simplicity, we are using a truckload rate of $2/mile and 40,000 units per load.

• Chapter 13 Data Aggregation in Network Design 243

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

To test the models, we will run a scenario on each model to pick the best five warehouse locations from the given list of potential sites. The results of the two models are shown in Figure 13.5.

Total Cost: $4,311 ,730 Total Cost: $4,296,088

Figure 13.5 Comparison of Output Results

The charts themselves give insight into how the models compare. Both models picked the same five warehouses: Newark, New Jersey; Atlanta, Georgia; Dallas, Texas; Chicago, Illinois; and Los Angeles, California. This shows that with the right aggregation strategy, we can come up with the same answer as one without aggregation. The model with three-digit ZIP Codes is much easier to run and analyze compared to the un aggregated model.

Now when we look at the actual costs, we see that the aggregated model varies by only 0.363 0/0 from the model without aggregation. As noted in the previous example, it is likely that the forecasted demand has more error than the difference in these solutions.

By using three-digit ZIP Code level aggregation, our customer file went from 16,000 to 800 without loss of accuracy. This model was further simplified by reducing the number of potential warehouses, without any difference in the final solution recommendation.

Validating Customer Aggregation-Regional Example In the previous example, we analyzed and validated the impact of customer aggregation on a broad, national level. In this example, we will focus on a specific market (Chicago) and see what kind of impact aggregation may have. This gives you a deeper look at the impact of aggregation .

• 244 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

The map in Figure 13.6 shows a set of 30 ship-to locations in the Chicago area, at the five-digit ZIP Code level. Each customer location receives between 5 and 20 full truck- load shipments a week.

Customer Customer # of TLs

10 Zip 1 60016

2 60056

3 60073

4 60085 5 60115

6 60120

7 60123 8 60126 9 60148

10 60188 11 60201 12 60202

13 60302

14 60402 15 60411

16 60435 17 60440 18 60453 19 60505

20 60506 21 60540 22 60586

23 60618

24 60623 25 60629

26 60632

27 60639 28 60647 29 60707

30 60714 TOTAL

Figure 13.6 Customer Locations in Chicago Area, and Weekly Demand in Number of Truckloads

10 10

10 10 10

5 5 5 5 10 10 10 10 15 15

10 5 10 15

15 20 15

15 15

5 5

305

For this validation exercise, we will estimate the transportation costs to serve this week- ly demand from a set of warehouse locations in Indianapolis, Indiana; New York, New York; Atlanta, Georgia; and Chicago, Illinois (see Figure 13.7). We will estimate the costs if these customers were to be served from each of these warehouses as potential options .

• Chapter 13 Data Aggregation in Network Design 245

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Figure 13.7 Map Showing Warehouse Locations and Customer Locations in Chicago Area

We will then aggregate these customer locations at the three-digit ZIP Code level, and recalculate the transportation costs from the same warehouse locations for the same total weekly demand. We can then compare the transportation costs for the two scenar- ios to understand the impact of this aggregation strategy.

Before we begin, let us look at how the transportation costs will be calculated. Because these customers are served by full truckloads, we will use a cost-per-mile rate applied against the distance to estimate the costs. For this example, we will use a cost of $2/mile as the rate for the truckload shipments. As noted in previous chapters, truckload carri- ers also apply a minimum charge for shipments that fall within a very short distance. For this case, we will use a minimum charge of $400 per TL shipment. This means that we will use the larger of the distance-rated cost versus the minimum charge. Because the warehouse in Chicago is one of the options, we can note that the minimum charge will apply because the distances fall within 200 miles.

The table in Figure 13.8 shows the transportation cost estimates by customer location (five-digit ZIP Code). We can see that the costs from Chicago to all customer locations are based on the minimum charge because the distances are less than 200 miles. For example, the cost for customer ID 1 (60016) is $400 * 10 truckloads, or $4,000. We also see that there are a few ZIP Codes that are less than 200 miles from Indianapolis and hence the minimum charge is being applied .

• 246 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Figure 13.8 Transportation Cost Estimates for Customers at Five-Digit ZIP Code Level

Now, we will aggregate the customer locations to the 3-digit ZIP Code level for this exer- cise. The chart in Figure 13.9 shows the demand calculation with aggregating from 5- digit ZIP to 3-digit ZIP level. The total number of truckloads is maintained at 305 loads. We will use the largest 5-digit ZIP within each 3-digit ZIP for geo-coding in the model. So, the number of customer points in the model has decreased from 30 to 8.

• Chapter 13 Data Aggregation in Network Design 247

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

I~ .~ WI T rn;j- ....

•. El!1I ., 7 -iID .~- 1 60016 10 600 19 60505 15 605

2 60056 10 600 20 60506 10 605

3 60073 10 600 21 60540 5 605

4 60085 10 600 22 60586 10 605

5 6011 5 10 601 23 60618 15 606

6 60120 10 601 24 60623 15 606

7 601 23 10 60 1 25 60629 20 606

8 601 26 10 601 26 60632 15 606 607 10 60707

9 60148 10 60 1 27 60639 15 606 TOTAL 305 10 60188 5 601 28 60647 15 606

11 60201 5 602 29 60707 5 607

12 60202 5 602 30 60714 5 607

13 60302 5 603

14 60402 10 604 ITotal 305

15 60411 10 604

16 60435 10 604

17 60440 10 604

18 60453 10 604

Figure 13.9 Aggregation from Five-Digit to' Three- Digit ZI P Code Level

We will use the same rate structure for this model including the minimum charge. The results of the analysis at the three-digit ZIP level are shown in Figure 13.10.

Customer Number of Atlanta Chicago Indianapolis New York City

Zip Truckloads = , 'f _ 'n;."U;l.f';' t' . .. : t. J t. , -m- ~ 'Gmin ~ ~ cml ~- -600 40 728 $ 58,254 42 $ 16,000 237 $ 18,976 854 $ 63,360 601 55 693 $ 76,220 23 $ 22,000 207 $ 22 ,775 862 $ 94 ,772 602 10 699 $ 13,985 13 $ 4 ,000 208 $ 4 ,167 842 $ 16 ,841

603 5 689 $ 6 ,889 9 $ 2,000 200 $ 2,001 848 $ 8,476

604 50 686 $ 68,601 11 $ 20,000 198 $ 20 ,000 848 $ 84 ,816

605 40 691 $ 55,267 42 $ 16,000 210 $ 16 ,830 878 $ 70 ,273

606 95 683 $ 129,776 5 $ 38,000 193 $ 38 ,000 841 $ 159,878

607 10 693 $ 13851 12 $ 4000 204 $ 4076 850 $ 16991

Grand Total 305 $ 422,844 $ 122,000 $ 126,824 $ 520,406

Re s ults with S-digit ZIP Code $ 421 ,633 $ 122,000 $ 127,3 32 $ 522,615

% Difference .0.29% 0.00% 0 .40% 0.42%

Figure 13.10 Transportation Cost Estimates for Customers at Three-Digit ZIP Code Level

The results show that the costs from each warehouse for the three-digit model are very close to that of the five -digit ZIP model. The differences are in the range of -0.29 % to +0.400/0, which is certainly smaller than the error in the demand forecasts or in likely transportation rate changes because of changing oil prices.

We also notice that there is no difference in costs from the Chicago warehouse as the minimum charge applies again for all locations. The differences in distances due to aggregation are eliminated due to the minimum charge. This is a key point to take note when aggregating at a regional level. Even though the five-digit ZIP level model shows a greater level of detail in actual locations, this does not offer any additional advantages in terms of costs .

• 248 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

To summarize, this validation exercise helped us understand how the right aggregation strategy can still provide valid and accurate results even at the regional or subregional level.

Aggregation of Products Product aggregation tends to be more challenging than customer aggregation. With cus- tomer aggregation, it is hard to go too wrong grouping customers by proximity. With product aggregation, most firms already have products grouped into marketing fami- lies. Unfortunately, there is nothing about a marketing family that naturally makes it a good candidate for an aggregate product group for a supply chain model. For example, a retailer may have a marketing family for Women's and Men's Apparel. From a logistics viewpoint, it might be more natural to have the socks be in one product family because they may come from the same vendor, have the same relative size, and require the same shipping and handling methods. On the other hand, the hanging clothes may naturally fit together because they are shipped and handled together.

So products sharing logistics characteristics make for good product families in a net- work design model. The marketing families are often irrelevant.

This means that you, the supply chain team, may need to create new product families and have the people in the organization agree to their validity.

Before we go too deep into product aggregation, there are two key questions you should ask at the start of every project. These questions will save you a lot of time in product aggregation and may help you avoid complicated product aggregation exercises.

The first key question is this: Can you create one aggregate product representing everything?

If you think about this from a marketing point of view, you will never answer "yes" to this question. However, from a logistics point of view, if you care only about pounds, kilos, pallets, cartons, or boxes moving through the supply chain, you should use it. In the examples in this book, we showed legitimate models that relied just on total weight or total packages sent through the supply chain, with no details on the underlying prod- ucts. In practice, many good models are built with one product family.

After thinking long and hard about the first question, the second critical question is this: Ifwe can't create one aggregate product family, will two do?

Of course, this question is really many questions, because you may want to ask about three, four, and so on. You can see that the preceding questions suggest a way of think- ing about aggregation that will help you. We have found, in practice, that it can be easi- er to start your product aggregation with one aggregate product (say pounds) and then grudgingly add more products as needed. This is opposed to starting with all 10,000 individual products and trying to whittle it down. In the latter case, it is always harder to get to a manageable number of product families.

• Chapter 13 Data Aggregation in Network Design 249

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

As you start to add more product groups, or your hand is forced, and you must start with all products and whittle it down, you need to know what characteristics to consid- er. What is important is that you consider the logistics characteristics of the products. The following sections provide some guidelines for how to group products.

Consider the Source of the Products

The source of the products is often the most overlooked category and can cause serious flaws in your model if you do not properly account for it. When including the source of your products, you are grouping products by where they came from. This h~lps ensure the correct flow from the plants to the final customers.

As an example, if you are building a model for the supply chain of apples, you could start by grouping all types and sizes of apples into one product called «apples." However, if your red apples come from Michigan and yellow apples from Washington, you need to include this in the aggregation strategy. Customers will demand both red and yellow apples. If you group products into just «apples;' the customers close to Michigan will get all their apples from the Michigan orchards and the ones on the West Coast will get all their apples from Washington. There would be nothing in the model to force customers to get apples from both Michigan and Washington. To make this happen, you need to create a product grouping that includes the source. For example, the table in Figure 13.11 shows some of the types of apples in this example.

SKU 1 Michigan Michigan apples

SKU 2 Michigan Michigan apples

SKU15 Michigan Michigan apples

SKU 53 Washington Washington apples

SKU 68 Washington Washington apples

SKU 80 Washington Washington apples

Figure 13.11 Example of Apple SKUs Categorized by Source Vendor Location

The correct way to model this would be to aggregate the three SKUs from Michigan as «Red Apples-Michigan sourced," and the remaining three SKUs as «Green Apples- Washington sourced," and allow only Michigan apples to come out of Michigan and the same for Washington. This will now correctly reflect the flow of products through the supply chain.

In more complex situations, a general way to think about categorizing products by sources is to think of sourcing groups. A sourcing group consists of all the SKUs or products that are made from the same set of plants. This structure allows you to capture

• 250 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

the fact that products can be made in multiple locations. As an example, the following table shows how you might go about this task.

The table in Figure 13.12 shows that SKU 1 and SKU 8 are sourced from Plant 1, and therefore placed in Group 1. SKU 7 and SKU 12 are both sourced from Plants 1 and 2, and therefore placed in Group 1-2.

~*1 ~:Jr:n'Ia " ~. • 'lor 1 Plant 1 Group 1 2 Plant 2 Group 2 3 Plant 3 Group 3 4 Plants 1,2 Group 1-2 5 Plants 1,2,3 Group 1-2-3 6 Plant 2 Group 2 7 Plant 1,2 Group 1-2 8 Plant 1 Group 1 9 Plant 2 Group 2 10 Plant 2 Group 2 11 Plants 1,2,3 Group 1-2-3 12 Plants 1,2 Group 1-2 13 Plants 1,2,3 Group 1-2-3 14 Plants 1,2 Group 1-2 15 Plant 2 Group 2

Figure 13.12 Categorization of SKUs into Source Groups

Consider Removing Products with Low Volumes Part of the art of modeling is to separate the significant from the trivial. And, often, when a firm makes or sells thousands of products or SKUs, they will have many very low-volume products. The chart in Figure 13.13 shows a typical example of the number of products sorted from highest demand to lowest and showing the cumulative per- centage of demand.

Cumulative % of Total Demand

100%

'E 90% Q) 80% u Q) 70% Il. 60% Q)

50% .~ n; 40% :; 30% E 20% :::J U 10%

~ .., ~",.. , ,

I I

I , o 100 200 300 400 500

Product 10 Sorted by Highest Total Cases

Figure 13.13 Product Demand Pareto Chart

• Chapter 13 Data Aggregation in Network Design 251

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

'i' If) co u Q; Q.

If)

g 1: en

'Q)

When we analyze this, we immediately find what most firms find with their data: 900/0 of the demand flowing through the network is made of less than half of the total 500 products. Taking that further, we see that 990/0 of the demand is made up of only 100 more products, for a total of 300 products. So, 300 of the 500 products make up 99% of the demand, and we can easily remove the remaining 200 products from our analysis.

Consider the Size of the Products The size of the products can impact transportation costs and warehousing costs. So products of similar size could be grouped together.

A perfectly valid and simple way to do this is to group products by small, medium, and large. As an example, in reality, this translates into products shipped in boxes, those on pallets, and bulk items.

A more sophisticated way is to cluster the products by their weight and cube (or cubic volume). This will help accurately model transportation costs while reducing the com- plexity of the model. The graph in Figure 13.14 shows an example of clustering.

70.0

• 60.0

50.0

• 40 .0

JI'''''~ I. K· 4 ..... rs~. ~

30.0 0 / t..~ -...d. II . .... ::~ .4 ~

20 .0 4 ; \ ~~~ A"I ~ ~ ... ~ •

'. , : .1 ~ ~ . ~ 10.0 1 .411 I~ .. ~ •

a~ • • .. .... 0.0

I _ .. 4 ........,.. 0 .000 0.010 0 .020 0.030

I" 4

4 ~

•

• • •

•

0 .040 0 .050 0.060

Volume (pallets per case)

•

0 .070 0 .080

Figure 13.14 Plot of Products and Their Respective Unit Weight

• 252 SUPPLY CHAIN NETWORK DESIGN

•

0 .090 0.100

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

The graph shows a plot of the products based on their unit weight (lbs per case) and volume (pallets per case). The rectangles shown on the graph illustrate how to cluster SKUs based on similar weight and cube characteristics. There are sophisticated ways you can do this type of cluster analysis that are beyond the scope of this book. However, keep in mind that you want to try to keep your aggregation strategy as simple as possible.

Consider Different Packaging Requirements It is also common to aggregate products based on their packaging requirements that may help easily model manufacturing and transportation costs. For example, the products for a beer manufacturer can be classified into bottles, cans, and kegs (see Figure 13.15). We can further classify bottles into 6-pack, 12-pack, and 24-pack, allow- ing the user to model specific constraints and rules for each package type.

4324 ABC 120z 12pk can 12pk

4141 ABC 120z 24pk bottle 24pk

2553 AB41ime 160z 6pk bottle 6pk

3652 AB4 240z 12pk bottle 12pk

7676 OHJ 120z 6pk can 6pk

6546 OGH O.5bbl keg

3645 OGH 240z 6pk can 6pk

Figure 13.15 Example of Beverage SKUs with Package Attribute

Consider Production Requirements

The same criteria can be applied to production requirements in order to simplify sourc- ing and production constraints and rules. For example, products that require special coating can be classified and tagged, allowing these products to be easily assigned to the specific production lines in the model.

C011;sider Products That Share Components or Raw Materials

Products that share the same components or raw materials can be aggregated together in order to simplify the modeling of bills-of-material and associated business rules for these products. A model for a chewing gum manufacturer would classify products as sugar and sugar-free based on products that use sugar or sugar substitute, allowing the modeling of sourcing rules.

• Chapter 13 Data Aggregation in Network Design 253

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Consider Products That Share Transportation Requirements

Products that require special transportation modes or requirements can be aggregated together to allow modeling of transportation rules. For example, products that ship in liquid form would be separate from the same product that may be put in a barrel first.

Consider Per- Unit Production Costs

In some supply chains, there may be several products within the same product family or group or across product groups that have significant variation in the unit production costs. This may be attributed to variations in some component or raw materia:l, or the differences in the production process itself. Depending on how these other portions (raw material and BOM, production lines) are modeled, it may be easier to group the SKUs based on the unit production costs to simplify the modeling.

Consider Predefined Product Families

In some cases, if the company has experience with network modeling, the existing prod- uct families may work. Of course, if the firm has experience with network modeling, they likely will have gone through the steps listed previously.

For example, a retailer may classify its SKUs into frozen goods, dry goods, fresh produce, liquid beverages, and so forth. Or a beverage producer may group products as shown in the table in Figure 13.16;

; ".

4324 ABC 120z 12pk can Can 120z ABC 12pk

4141 ABC 120z 24pk bottle Bottle 120z ABC 24pk

2553 AB41ime 160z 6pk bottle Bottle 160z AB4 6pk

3652 AB4 240z 12pk bottle Bottle 240z AB4 12pk

7676 OHJ 120z 6pk can Can 120z OGJ 6pk

6546 OGH O.5bbl keg Keg O.5bbl OGH

3645 OGH 240z 6pk can Can 240z OGH 6pk

Figure 13.16 Example of Beverage Products with Key Attributes Useful in Aggregation

If this information is readily available, you should use it.

Testing the Product Aggregation Strategy As a next step, we will look at testing an aggregation strategy in a model with some sam- ple data. For this test exercise, we will build a sample model with the following elements:

• 254 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

• 5 plants

• 25 potential warehouses

• Distance-based service constraints

• Inventory holding costs

• Fixed warehouse costs

• Product aggregation-we will test the model with the following options:

• 46 original SKUs

• 4 aggregated products-aggregated products were created using weighted averages and clustering (similar to the methodology used in a previous example )

The model was run with 46 products, and then with 4 aggregated products. The results of the two scenarios are shown in Figure 13.17. The analysis shows that the aggregated model was 0.0003 0/0 higher than the one without aggregation. The difference is that one model picked Seattle, Washington to cover the northern part of the West Coast and the second model picked a location in Northern California.

Total Cost: $104,006,061 Total Products : 46

Total Cost: $104,032,912 Total Products : 4

Figure 13.17 Results Comparison for 46 Original Products Versus 4 Aggregated Products

So the critical question is, Does the Seattle solution really differ from the Northern California solution? To understand this question, we will run two scenarios with each model: one scenario forcing Seattle and another scenario forcing Northern California. The ·results are shown in Figure 13.18.

• Chapter 13 Data Aggregation in Network Design 25 5

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

• ~ ; ; . . ..

None; 46 Products Seattle $104 ,006.061 0.00%

None; 46 Products N California $104,113,011 0 .10%

4 Products Seattle $104,087,045 0 .08%

4 Products N Californ ia $104,032,912 0.03%

Figure 13.18 Comparison of Results with Seattle and Northern California

If we look at the results for the model without aggregation, we see that there is a differ- ence of 0.1 % between the Seattle and the Northern California solutions. This difference is extremely small considering the margin of error and level of accuracy of the underly- ing data within the model. It is likely that the forecasted demand data has more error than the difference shown here. This helps us conclude that the Northern California solution is equivalent to the Seattle solution based on the cost factors analyzed in the model. This means that the solution requires further analysis evaluating other factors such as labor availability and logistics infrastructure.

This also shows that the aggregated model is still accurate given the margin of accuracy for normal models. The key take-away from this exercise is that with the right aggrega- tion strategy, we can get the same level of accuracy in decision making as one would achieve with modeling without aggregation.

Aggregation of Sites In most models, you have a limited number of plants and warehouses. So you usually keep these as individual sites and do not need to aggregate. The one minor exception is when you may have two to three plants or two to three warehouses that are all on the same campus or within the same city. In this case, as long as the facilities do similar things, you may group them together.

This is not always the case for vendors. Many firms have hundreds and maybe thou- sands of vendors.

If we need to include vendors in our model, we treat this a lot like we did with cus- tomers. We group vendors by geographic proximity and maybe by type. The example in Figure 13.19 shows vendors grouped by three-digit ZIP Code in Georgia, but we can often build good models by grouping vendors by a larger geographic area like a U.S. state, country, or even a port of entry. As another example, in a U.S.-focused model you may aggregate all your Asian vendors to a single point in Long Beach, California .

• 256 SUPPLY CHAIN N ETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Figure 13.19 Example Showing Aggregation of Vendors in Atlanta Area

Aggregation of Time Periods When aggregating time periods, we are simply determining the time buckets we will use for our model.

The industry standard for time-period buckets in network design models is a year. Within a year, you capture the full range of demand across seasons, and it is natural to think of reporting your costs and savings in terms of annual numbers. And the nature of most decisions about when to open or close a facility does not need more granulari- ty than an annual bucket.

With this structure, nothing prevents you from running what-if scenarios with different demand patterns. That is, you can run a model with the expected demands five and ten years out. Even more so than with products, we want to make sure we are careful when we model more than a single annual time period.

There are two basic problems we solve when we include multiple time periods in the same model. The first problem we can address is to determine what year to open and close facil- ities. To capture this, we could model multiple annual buckets in the same model and have the model pick which year we open or close facilities. Note that this makes the model much more complicated. One way to avoid this complication is to simply run different what-ifs with future demand scenarios. You may think you are losing accuracy, but in real- ity, predicting demand several years out is very difficult anyway, so you may be just intro- ducing false precision by including multiple years in the same model.

The second problem we can address is to determine how to handle seasonality. In this case, we may build models that have 4 quarterly buckets to make different decisions in different parts of the year. Or if the prebuilding of inventory is important, we may model in 12 monthly buckets.

• Chapter 13 Data Aggregation in Network Design 257

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

For example, if we look at a consumer products company that experiences a high spike in demand in the summer, we may want to make decisions about when to start building the inventory and where to store it. The graph in Figure 13.20 shows the spike in demand and the need for extra storage capacity.

Current DC

Capacity

Overflow

space used

On-hand

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Figure 13.20 On-Hand Inventory Versus Warehouse Capacity

These are some general rules for thinking about time periods. As with all modeling, you can get creative to solve your specific problem. For example, we have seen other cases in which a firm runs a network model with weekly buckets to decide what product is made where and shipped to which warehouse.

In general, as you add time periods to the model, you need to remove other types of decisions. For example, if you have monthly buckets, you are typically fixing the loca- tions of your plants and your warehouses (except the overflow warehouses).

Aggregation of Cost Types Although cost types may not be commonly thought of as an aggregation decision, you also want to simplify your modeling by grouping similar cost types together.

For example, you may use 100 different full-truckload carriers. Instead of trying to cap- ture each of these, you may want to group them all together and use the average of their rates when they share a source and destination combination. This makes modeling eas- ier and it avoids the false precision of forecasting next year's rates for each of 100 differ- ent carriers.

The same lesson applies to costs in your facilities. You may have different components of fixed and variable costs at your plants and warehouses. In fact, you may have many different workers each with a different salary and benefits package. You do not want to model all that detail. It is just as accurate to roll these costs up to a few small categories .

• 258 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Lessons Learned on Aggregation You will not be able to avoid aggregation. But you also shouldn't worry: You can have very accurate models that work off aggregated data. And, remember, because you are running models for the futurJ\ the aggregated models may actually be more accurate than models with a lot of detail that give a false sense of precision.

When you are aggregating, the most important decisions will be around your customers and products.

And if you are going to model in something other than annual buckets, make sure you carefully understand these ramifications.

End-of-Chapter Questions

1. A large electronics and home-appliance retailer is looking to perform a distri- bution network analysis to optimize its warehouse locations. Using any real electronics retailer that you know as a reference, what are the aggregated prod- uct groups that you would consider modeling? Make sure you consider all the factors, including product sourcing, product logistics, characteristics, and so on.

2. Open the file Aggregating Customers. xis found on the book Web site. This file contains a list of 10,000 raw ship-to points and some characteristics about these data points. How would you go about aggregating the customers? Go ahead and follow your strategy. How many customers do you end up with?

3. Open the file Aggregating Products. xis found on the book Web site. This file contains a list of 50 products. How would you aggregate these products? How few product groups do you need in order to capture all necessary charac- teristics of these 50 products? Which of the 50 products belong to each of your product groups?

4. You are working for a company with 500 unique products. You must aggregate these products. The team has decided that they will build a simple way to group products by coming up with six categories. Each category only has three choic- es. They will apply the categories sequentially. For example, they first break the products into Large, Medium, and Small; then for each size they break into Red, Yellow, and Blue; and for each color they break into three more subcategories; and so on for the six categories. What is wrong with this approach?

5. You work for an online retailer and you shipped to 100,000 unique addresses last year. You are building a network model to determine where to open a new ware- house later in the year. Besides the fact that you don't want to build a model with 100,000 ship-to points, why will these 100,000 points not be the same exact points you ship to next year? How different are the points likely to be?

• Chapter 13 Data Aggregation in Network Design 259

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

•

6. You are building a model to determine the best location for a single pharmaceu- tical plant to make a new drug to serve global demand. Why is modeling every country as a single demand point probably good enough for this problem?

7. You are building a model for a global chemical company that has three plants, each making a unique set of products. They have a plant in Asia, one in North America, and one in Europe. They make hundreds of different products, but the products all have essentially the same size, weight, and handling costs, and ship the same way. What is the minimum number of aggregate products you need? What happens if you have fewer than this number of aggregate products?

260 SUPPLY CHAIN NETWORK DESIGN

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Bibliography

"Chapter 1: The Value of Supply Chain Network Design." In Supply Chain Network Design: Applying Optimization and Analytics to the Global Supply Chain, by Watson, Michael; Lewis, Sara; Cacioppi, Peter; Jayaraman, Jay. pp. 1−22. Prentice Hall /Pearson Education (Upper Saddle River, NJ)*, 2012. (22 pages). WATSON, MICHAEL; LEWIS, SARA; CACIOPPI, PETER; JAYARAMAN, JAY, SUPPLY CHAIN NETWORK DESIGN: APPLYING OPTIMIZATION AND ANALYTICS TO THE GLOBAL SUPPLY CHAIN, 1st Edition, (c) 2013. Reprinted by permission of Pearson Education, Inc., Upper Saddle River, NJ.

"Chapter 12: The Art of Modeling." In Supply Chain Network Design: Applying Optimization and Analytics to the Global Supply Chain, by Watson, Michael; Lewis, Sara; Cacioppi, Peter; Jayaraman, Jay. pp. 217−236. Prentice Hall /Pearson Education (Upper Saddle River, NJ)*, 2012. (20 pages). WATSON, MICHAEL; LEWIS, SARA; CACIOPPI, PETER; JAYARAMAN, JAY, SUPPLY CHAIN NETWORK DESIGN: APPLYING OPTIMIZATION AND ANALYTICS TO THE GLOBAL SUPPLY CHAIN, 1st Edition, (c) 2013. Reprinted by permission of Pearson Education, Inc., Upper Saddle River, NJ.

"Chapter 13: Data Aggregation in Network Design." In Supply Chain Network Design: Applying Optimization and Analytics to the Global Supply Chain, by Watson, Michael; Lewis, Sara; Cacioppi, Peter; Jayaraman, Jay. pp. 237−260. Prentice Hall /Pearson Education (Upper Saddle River, NJ)*, 2012. (24 pages). WATSON, MICHAEL; LEWIS, SARA; CACIOPPI, PETER; JAYARAMAN, JAY, SUPPLY CHAIN NETWORK DESIGN: APPLYING OPTIMIZATION AND ANALYTICS TO THE GLOBAL SUPPLY CHAIN, 1st Edition, (c) 2013. Reprinted by permission of Pearson Education, Inc., Upper Saddle River, NJ.

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016 68

Downloaded by Shillin Chen on 3/20/2016. Ohio State University , Keely Croxton, Spring 2016

Footer
Front Cover
Title Page
Table Of Contents
Chapter 1: The Value of Supply Chain Network Design
Chapter 12: The Art of Modeling
Chapter 13: Data Aggregation in Network Design
Bibliography
Back Cover