Not too long ago supply chain network modeling was done on Excel spreadsheets using latitude, longitude and rough data. These crude models, called �centers of gravity� or �centroids,� estimated the middle between suppliers and customers. Network modeling software grew with the development of computers and now there are powerful tools that can optimize supply chain network structures. Each software has its own unique functionality that can be configured to match most supply chains. However, there are limitations and misunderstandings about what these modeling tools are able to do.
     
    Why Model Your Supply Chain?
    Some of the more common reasons to model your supply chain:
                � Business/company consolidation. The business profile has changed and the network structure may need to be modified.
                � Compelling business challenges, e.g., competitive service improvements or cost reduction pressures.
                � Decisions pending, e.g., lease on a warehouse is coming due. Perhaps it would be better to close or relocate the facility or maybe even outsource the operation.
                � Complex trade-offs are not clear, and it�s too expensive and time-consuming to experiment in the real world.
                � It�s been said that if your supply chain network has not been modeled in the past five years, it�s likely that there are significant improvement opportunities.
     
    How Should You Model Your Supply Chain?   
    Organize the Effort � Understand the business setting and competitive environment, identify the issues and articulate the business questions to be addressed. Assemble the modeling/implementation team. Establish the model scope, e.g., products, processes, geography, sites, customer zones, transportation modes, forecast horizon, etc. Define the model structure that addresses the business questions and develop specifications for the input data.
     
    Set Up the Model � Develop the input data, usually the most time-consuming part of the exercise. Build the model. Run and calibrate the model, including baseline comparison with actual costs.
     
    Generate Results � Define scenarios to test. Run model, compare scenarios and develop �learnings.� Prepare as well as present recommendations.
     
    How Does a Network Modeling Exercise Go Astray?
    If not executed properly, a supply chain network modeling exercise can lead to erroneous and costly results. These are some of the most common problems:
                � Insufficient logistics experience
                � Insufficient modeling experience
                � Questionable data
                � Incorrect interpretation of results
                � Viewing the model as an �answer� machine
                � Rush for an answer
     
    We have seen models that contained unrealistic data. For a manufacturer in Chicago, a rate of $250 per truckload to Memphis, Tennessee was used in the model. Such a rate is so attractive that thousands of shippers would take advantage of it immediately. Because of the rate inaccuracy and its impact in the model, Memphis was wrongly selected as a distribution point.
     
    Another project that developed freight cost estimates based on product and order characteristics projected savings of over $4 million annually. When actual transportation, warehousing and inventory costs were applied to the network design, there were no savings. The problem with this model was inaccurate data at the unit level.
     
    Network Modeling Pitfalls
    A network modeling exercise contains a number of opportunities for failure, and here are some of the most common:
     
    Modeling Tool Limitations or Misunderstandings � Modeling tools are loaded with specific rates, costs, rules and other data. The software is designed to optimize the network structure but is constrained by the data that is available in the model. With a change in the network structure or weight on a lane, for example, a shipper may be able to negotiate a better rate than is loaded in the model. With some difficulty, these possibilities can be explored by varying the rates and data in the model to test the impact on results.
     
    Failure to Properly Understand the Business and Questions � A model is an approximation of your business. You should understand the business issues and know what you want to get out of a model before you begin to build and use it.
     
    Assembling the Team � Don�t expect to assign a modeling project to an analyst, including the responsibility for collecting the data, building and running the model and making recommendations. Unless all affected parties are included in the project throughout, the likelihood of sound conclusions and commitment to implementation is very limited. Ensure you have management sponsorship with authority to spend and implement the change. We have seen many modeling projects completed but never implemented.
     
    Project Scope � Ensure the scope of the project addresses the goals and is manageable. It is better to have a smaller scope that gets implemented rather than a large one that never is completed.
     
    Data Development � The most time-consuming component of a network modeling project is assembling the data, and there is a tendency to short cut the effort. The data required in network models typically is not the information used on a daily basis to manage the business. It�s necessary to take samples of actual data and manipulate it into the format the modeling software requires. Anytime one uses samples, it�s possible that the data is not representative. The devil is in the detail. Bad data equals bad results. All modeling data must be approved by the experts who know the business and can verify its accuracy.
     
    Building the Model � Every modeling tool has a series of templates that require data. Before beginning to enter data, one must understand the software well enough to be able to configure it to accommodate your specific supply chain structure. Do not enter data into templates if you don�t know how that data will be used by the model. Be careful entering data that you do not understand. Be concerned when you hear �just put the data in now, we�ll change it later.�
     
    Run and Calibrate the Model � Since all models are approximations, it�s essential that management develop a sense of comfort with the model. You can do this by forcing the model to replicate the current supply chain or baseline. The results must be compared with actual costs and the differences reconciled. Although the model�s specific manufacturing, transportation, warehousing and inventory costs may not be directly visible in the company financial records, most finance experts who have an understanding of the model structure, including the assumptions, can develop the necessary reconciliation.
     
    Run and Compare Scenarios � Once one has spent a large amount of time developing data and building a model, there is a tendency to view the first good result as the �answer� and be finished. The most important objective of network modeling is not to get an answer but rather to develop �learnings� about supply chain behavior. You need to understand cause and effect. When you push the supply chain at one point, what happens somewhere else? Depending upon the complexity of your supply chain, you may need to run and compare 100 or more scenarios. The resulting �learnings� are then formulated into a set of recommendations.
     
    Implementation � In addition to one-time costs, there will be a lot of operating cost changes when implementing a new network structure. Unless these costs are properly captured in a realistic ROI prepared by the company�s financial experts (not a software provider or consultant), implementation and real benefits are at risk. It�s important to have an actionable implementation plan with enough detail so that the necessary changes can be made.
     
    These are some of the pitfalls that one can face in supply chain network modeling. Don�t expect network modeling to be a simple exercise. It requires time and input from many people, supply chain and modeling software expertise, very accurate data and persistence to come up with good results. When these projects are executed properly, they can lead to savings of five percent to 15% of your entire supply chain cost and better service to your customers. Ensure you invest the time, energy and attention to detail required in these projects.
     
    Giles Taylor is the president of Trans-solutions, Inc., a supply chain consulting firm. For additional information, please visit www.trans-solutions.com. Lee Engdahl is a partner with Global Supply Chain Associates, a supply chain consulting firm. For more information, visitwww.gsca.com.
     

    Follow