Strengthening your supply chain one link at a time.
Having constructed network models for over 16 years it has been my experience that they are overwhelmingly distribution oriented with a sensitivity or two reflecting high level production changes. However, I have built some production line models which provide a more robust manufacturing solution. In order to do this you need a few data points that are not required for a distribution network study.
Manufacturing Lines – the name and plant location of each line.
Line Capacity – ideally for each line and plant combination you want to obtain the unit capacity (cases, eaches, etc.) and maximum operating hours for the period (usually a year).
Product Eligibilities – this is required for a typical network study (at a plant level), but for a manufacturing study you need the specific lines that can make the product.
Production Cost – the cost to produce each product at a certain plant on a certain line.
With the above data inputs you can construct a manufacturing network model that will not only indicate “what should be made where”, but indicate the specific lines that should make each product. Various supply chain optimization software may be used to accomplish this such as Supply Chain Modeler – formerly Supply Chain Guru (Coupa/Llamasoft), LogicNet Plus (Coupa), Cosmic Frog (Optilogic) and others.
Each manufacturing line needs to be identified for inclusion in the model. I have built a model with every single line delineated as well as aggregating lines within the same plant that have the same capacity – either method works when applied appropriately.
With respect to capacity you must define the upper production limit for each manufacturing line. This would be defined as maximum units (cases for example) that can be produced or production run time limit. For example, if a line can run eight hours per day for five days per week and 50 weeks per year then the capacity is 2,000 hours. This calculation should be performed for every plant/line combination. If aggregating lines then the capacity should be multiplied by the number of lines represented.
Next, the capacity needed to produce each item at each plant on the eligible lines is defined. Consider the line above that runs for 2,000 hours maximum per year. If this line can produce 50,000 cases in that time then the capacity required is .04 hours per case. Every item would have an associated capacity needed to produce which is relative to the overall capacity for the line.
Finally, the cost to produce an item on each line at the respective plant would be added to the model. The trade-off between producing at a low cost plant (in Mexico for example) and shipping inbound to a US distribution center would be weighed against producing at a high cost US plant thereby potentially eliminating long distance transfers (depends on the plant and DC locations).
With the above elements added to the model, the optimization will detail which items should be made on which lines at specific plants. It’s common to have similar lines at different plants as well as different lines at the same plant with the same production eligibilities. Reporting within the optimization software reveals where every item was made as well as the utilization of each line with respect to the capacities you have defined.
Production sensitivities are common in the network models that I have built over the years, but a manufacturing study adds some interesting elements to the optimization. Aside from optimizing production within the existing manufacturing footprint you could locate a new plant with some indication as to the specific line types that would be required.
—Dan Gunter, St. Onge Company
