Cross-Dock Modeling And Simulation Output Analysis Mehdi Charfi
Abstract Cross-docking is a consolidation practice in logistics that facilitates the transfer and sorting of products from suppliers to distribution centers, eliminating warehouse holding, minimizing costs, and allowing for the realization of more efficient deliveries. There are several logistic and integration problems inherent in the cross-docking process, including scheduling uncertainty, inefficient staffing and record keeping, and limited integration of statistical data. The previous phases of this project have addressed the design of a computer-based discrete event simulation model (Intelligent Transportation System) of a freight cross-dock facility. In this phase of the project, the discrete event simulation model will be exercised with realistic data in the form of matrices that describe originating truck arriving volumes. Upon creating a realistic model of the cross-docking facility, the model and simulation will be analyzed to determine suitability and rationality. With this foundation in place, simulation and optimization models can assist a given cross-dock facility in realizing optimization benefits in areas including door assignment, expansion, and productivity. 2
Background and Introduction Cross-docking is a logistics technique used in the retail and trucking industries to rapidly consolidate shipments from disparate sources and realize economies of scale in outbound transportation. Cross-docking essentially eliminates the costly inventory-holding functions of a warehouse, while still allowing it to serve its consolidation and shipping functions. 2 As illustrated in Figure 1 below, the idea is to transfer shipments directly from incoming to outgoing truck trailers, without in between storage. Goods typically spend less than 24 hours in the cross-dock, sometimes less than an hour. With the process of moving shipments from the receiving dock (strip door) to the shipping dock (stack door), bypassing storage, cross-docking reduces inventory carrying cost, transportation cost, and other costs associated with material handling. 2 Figure 1. Cross-docking Diagram 3
Cross-docking presents an extremely valuable advantage for many companies. It has become a front-line business strategy for those looking to move products in and out of the warehouse in the most cost-effective, efficient, and timely manner possible. The proper processes, systems, and supply chain relationships must be in place to successfully cross-dock on a large scale. Among these necessary systems include: automated material handling, warehouse management systems (WMS), order processing systems, quality controls systems, and strong relationships between supply chain partners. Ultimately, several factors that produce effective cross-docking are the sharing of information, clear communication, confidence in the quality and conformance of goods, and product availability. 3 In all, the outcome that all parties involved desire are driving costs out of their operations while satisfying customer demand. While cross-docking renders many benefits for the supply chain companies and vendors involved, the process contains several disadvantages. First, it is crucial for the involved partners to have the necessary storage capabilities and an adequate transport fleet to operate cross-docking. 4 Without these essential facets, the process will become overwhelming and fail to meet logistical needs. In addition, an adequate information technology system is fundamental to operating the cross-dock facility. Without the appropriate system in place, the sharing of information will be extremely complex. This could lead to the loss of crucial information and be detrimental to the communications between partners. Furthermore, cross-docking eliminates buffers in the supply chain. 4 While overall inventory is reduced, it becomes difficult for suppliers to have enough flexibility to react to demand changes without hurting customer service. Lastly, it is important to understand the 4
shipping costs required to properly operate through cross-docking. It might be necessary for some companies to reduce their inventory to balance out the increased costs of shipping. Overall, the ease and effectiveness of cross-docking depends on the supplier-vendor relationships and how significant the process is in the company s business model. The purpose of this portion of cross-docking research is to develop a simulation and to be able to generate truck arriving volumes, through modeling techniques, which will be used as input data for the simulation. The previous phase of research involved the design of an Intelligent Transportation System (ITS) of a freight cross-dock facility. This system was derived from the need to improve the efficiency and transparency of the cross-docking process. The system was structured around a centralized database named The Efficient Dynamic Display of Incoming Exchanges (EDDIE). A modular interface was developed and integrated with an existing container management system and data terminals at the freight checking stations. 1 The database established data connections that streamlined daily operations, increased transparency, and performed more accurate labor calculations for the cross-dock facility. By increasing the facility s access to archived data, the system established the necessary groundwork for the optimization endeavors of the current phase of this research project. 5
Methods After understanding the background information of the project as well as its relevance, several objectives were set for this research project. These objectives are listed below: Researching the cross-dock system Gathering realistic data Construct a statistical model using the realistic data Formatting realistic data to the inputs required by ExtendSim software Applying statistical techniques to ensure that simulation results and conclusions are adequately verified The primary aim of this project was to develop and format a model of the crossdocking facility to be used as input data for a simulation which was created using ExtendSim (simulation software). The model should be able to be randomized so that it can produce sample days, and weeks, of operations. Given a sample matrix of data representing incoming item volume from one day of cross-dock operations, a statistical model was developed through assumptions, intuition, and statistical techniques. Realistically, this sample data represents the average item volumes of a given cross-dock facility. However, to render the best quality of data, the cross-dock facility will provide numerous data sheets so that this data can be properly analyzed and used to create sample data sets with more accurate means and standard deviations. Therefore, with the absence of such a source, it became essential to create a model that makes the most sense in the absence of data. 6
After examining a sample daily volume flow matrix, several statistical distributions were considered in creating a model for the data. Among these were the Gaussian (normal) distribution, the exponential distribution, and the Weibull distribution. The Gaussian distribution was found to present a more accurate representation of the sample data through trial-and-error methods, so the other two distributions were ignored. A significant assumption made was that each item in the 50-by-50 matrix (Figure 2) serves as the mean for the specific origin-destination point. Through Microsoft Excel analysis, the standard deviations were found by treating each set of data (column in matrix) for a particular destination as the determining factor. The standard deviation for the set of destination values was used as the standard deviation for each item in that set. While this is somewhat unrealistic, it can be very difficult to accurately determine the performance of the cross-dock facility without facility-specific data. Now with all the necessary statistical information in place, Microsoft Excel can be used to create the model. The formula below was the key to forming the model, as it inputs the means and standard deviations while outputting a sample data point for each origin-destination item. =INT( NORMINV( RAND(), MEAN, STANDARD DEVIATION )) As shown in Figure 2, the columns represent different destinations used by the crossdock facility, while the rows represent different origins. Each value shown represents a specific mean for that explicit origin-destination pair. Standards deviations for each item were found by computing the standard deviations of each column. Using these means and standard deviations, as well as a normal distribution, sample data sets can be formulated to complete the statistical model. Using Microsoft Excel, and the formula shown above, a new 50-by-50 7
matrix can be developed to serve as a sample day of operations for the cross-dock facility and input data for the ExtendSim simulation. Figure 2. Portion of 50-by-50 Daily Volume Flow Matrix between Origins and Destinations Results Due to insufficient realistic data, the data from the 50-by-50 daily flow matrix has served as the mean for the desired sample data. Furthermore, the standard deviations have been calculated from the data to assist in creating new sample data sets. Using Microsoft Excel and several statistical assumptions, a model was formed to randomize a day s worth of input data. To validate the model, 200 new columns were generated to replicate the incoming truck volume for the first destination. With these columns in place, 200 sample one day data sets have been created for a specific column. Ideally, the means of each row in the column will match the original data that was received. As shown in Figure 3, the average of 8
each row was taken and compared to the original data. Furthermore, to simplify the recognition of accuracy, the difference between the two columns is shown. According to what was produced, the means of the rows of the sample data is very similar to the original column for incoming truck volume. These similarities validate the methods used in creating the model in accordance with the assumptions made throughout the process. Figure 3. Validation of the Model 9
Discussion Throughout the course of this research, several significant issues developed that have proved to be difficult obstacles to overcome. The most noteworthy of these challenges is the insufficient amount of data received to aid in the modeling process. Through in depth research on the cross-docking technique, along with the use of statistical theories, a sample volume flow matrix was used to create sample data sets replicating one day of operation. Ideally, numerous data sets will be retrieved in the future so that the model could be more realistic and accurate to the cross-dock facility s performance. However, with this obstacle, more emphasis had to be placed on a theoretical approach to the problem. Several theories were discussed until our methods became reasonable. However, further work can certainly be done to improve the current model pending collaboration with the cross-dock facility and the collection of more data from that particular facility. The more data availability from the cross-dock, the more accurate our model and simulation can be. Future Work Using computer programming techniques, the developed model can be used as input data for the ExtendSim simulation. The simulation will be able to input specific data and run a realistic animation of the operations of the cross-dock facility. Further communication with the cross-dock facility is crucial to improving the current model to develop a more realistic model. Once the model is synchronized with the simulation, the final simulation and model will allow any given cross-dock facility to realize optimization benefits in areas including door assignment, expansion, and productivity. 10
References [1] Improving Cross-docking Efficiency through the Use of ITS Technology. Jeffries, Jacci, Kennedy, Ciara, Zheng, Lisa. 2011. [2] Algorithms for the Cross-dock Door Assignment Problem. Hahn, Peter M., Zhu, Yi-Rong, Guignard, Monique, Liu, Ying, Pessoa, Artur A., de Azevedo, Guilherme H.I. 2010. [3] Cross-Docking: Bypassing Storage. Vink, Jerry. April, 26, 2006. Multichannel Merchant. [4] Bartholdi, John J., and Kevin R. Gue. "The Best Shape for a Crossdock." Transportation Science Volume 38.2 (2004): Pages 235-44. Acknowledgements Dr. Peter Hahn Dr. Monique Guinard Mr. Frederick Abiprabowo Ms. Sarah Murphy Ms. Cora Ingrum 11