Demand Planning
Written by Peter Buxbaum
MLF 2011 Volume: 5 Issue: 9 (October)
Military organizations, like their commercial counterparts, undertake laborious efforts to forecast demands for items like commodities, supplies and spare parts. That way, they can carry just enough inventory to meet warfighter needs and streamline costs and operations.
There’s just one problem. The forecasts are always wrong. There is always too much or too little of any given item, never the exact amount.
That doesn’t stop the forecasting, however. Organizations like the Defense Logistics Agency, as well as the other military agencies and services, are constantly trying to fine-tune the forecasting process by deploying better technologies, implementing more effective business processes, and developing collaboration strategies designed to generate better information upon which to base forecasts. The improvement in forecasting accuracy is of increasing importance in this day and age of shrinking federal budgets.
“We do a lot related to forecasting,” said Bob Carroll, the planning process division chief at DLA. “We’re trying to determine what we should buy, at what quantity, and at what location it should be stored. “Forecasting is imperfect,” he added, “but certainly we seek to improve it and provide the best overall logistics value and cost to the taxpayer.”
There are two basic approaches to go about forecasting. “You can look at the past and try to predict the future,” said Carroll, “or you can come up with intelligence on current requirements and cook that into the projection. We actually try to balance an analysis of past data with programmatic information.”
Developing programmatic information involves collaboration with DLA customers. “We work with customers to get a feel for their expectations of future demand,” said Carroll, “and we balance that against historical trends.”
The DLA enterprise is huge, spending $38 billion per year to manage 95 percent of the repair parts procurement for all of the armed services as well as 100 percent of the food, fuel, medical supplies, clothing and construction equipment across the Department of Defense. DLA activities reach 126 nations with 520,000 shipments annually and 54,000 requisitions in any one day. It manages over 5 million items in eight supply chains across 26 distribution depots. DLA employs over 160 demand planners, some of whom are deployed to theater at any given time.
“What we seek to do with forecasting is to make sure the customer has stock on the shelf,” said Carroll. “But another part of the logistics process is mitigating the risk of having too much inventory. The idea is to close the variance and manage the risk. If I err, should I err on the side of having too much or not having enough?” Different situations and different items require different calculations in that regard.
“You don’t want to be burdened with an iron mountain of stuff,” said Alan Manning, a logistics and supply chain expert at SAS. “That would produce an overweighted logistical tail. It’s important to have confidence in the forecast. That way you will plan more precisely to the forecast and you won’t be tempted to take out insurance in the form of extra inventory.”
More accurate forecasts would reduce not only inventory costs, but transportation costs as well, noted Charlie Fletcher, senior manager of the Strategic Operations Group at Alion Science and Technology Corporation. “You want to be able to position stocks forward so that you can be more responsive,” he said. “Accurate demand forecasts will allow you to move material by ship rather than air. Shipping costs one-tenth [the cost of] air.” Fletcher, a retired Army major general, formerly served as director of operations and plans at the U.S. Transportation Command and as the commanding general at the Military Surface Deployment and Distribution Command.
Measures DLA is taking to improve demand forecasting include standardizing its information technology systems, deploying specialized demand planning tools, and encouraging collaboration with customers. At the technology level, DLA decided in the late 1990s to replace its decades-old legacy systems with an SAP Enterprise Resource Planning (ERP) system. In addition, DLA integrated with SAP a planning and forecasting module from Manugistics, a company that has since been absorbed by JDA.
“We made the decision to move to commercial off-the-shelf technology but we had to modify it,” said Carroll. “The Department of Defense is different from the commercial sector. We are not in business to make money. We are not out to optimize DLA operations but DoD as a whole. Putting DoD needs first means we might spend more money but DoD can save.”
The JDA forecasting tool can use several different statistical models to make a forecast and is especially useful for situations where past experience is not an effective predictor of future demand. DLA started feeding the JDA tool with data in 2003. In 2006, all items subject to demand forecasting were in the system. DLA has more than 5 million items in their catalog and at any given time about 1.5 million are active. Of the active, about 300,000 were replenishment and therefore relatively easier to forecast.
What the JDA tool does is to gather up all the historical demand data and apply these through algorithms to several different categories of items that DLA carries. “The system comes up with a projection of what we need to order over time, how much is needed and where, and when to put out a purchase request,” said Carroll. “We can also mix past demand with current programmatic intelligence to make adjustments to the forecast. We can also add into the forecast other elements such as seasonality and cyclicality.”
The key to the accuracy of any demand forecast, for Manning, is data. “You want as much data as you can get. The more the better,” he said. “It should be as granular and as detailed historical information as possible.”
This data includes not only raw demand data, but other information that provides context to the manner in which equipment, parts and supplies are used. “Information such as the geography where the items are to be used, whether it is the Arctic or an arid desert, or the mission equipment is deployed to, such as peacemaking or hot combat, is important,” said Manning. “All of these things can also be swept into the demand analysis.”
Other military components have also implemented tools that help with demand forecasting. Alion is working with the Army Operational Test Command in Fort Hood, Texas, developing software used to integrate platforms and show their relationship in a network. “Because of our background in modeling and simulation,” said Fletcher, “we are able to emulate the performance and activity of networks without all of its nodes actually being activated. These tools are used to perform supply chain projections for networks that have not yet been deployed.”
The tool is analogous to diagnostics systems on automobiles that warn drivers to check systems or components before they break down. “These tools allow the Army to understand the characteristics of radios and networks before they are put on the ground, forecast demand for major sub-assemblies and repair parts, and set up supply networks,” said Fletcher.
SAS provides analytical tools that have been used by military organizations, as well as commercial enterprises, to perform demand planning and to reduce inventories. The Army and Marine Corps have used SAS tools in connection with reset activities in Iraq. The Air Force has used the tools to plan for the deployment of manpower and personnel in connection with the troop surge in Afghanistan last year.
In the commercial world, Honda Motor was able to sustain a 99 percent fill rate for spare parts while reducing its global inventory by 50 percent. “Honda realized that if they did a better job of forecasting, they would realize gains in terms of a smaller inventory footprint,” said Manning. The SAS demand forecasting tool is capable of being integrated with leading ERP systems such as SAP and Oracle as well as with over 80 other data sources.
Matching up supply and demand is not merely a question of technology, however. Closer and more collaborative relationships among the DLA and its customers have also proved to be key to the changes DLA is seeking. DLA has sought collaboration with its key customers in an effort to get the best possible data for its forecasts. By getting granular data from documents such as bills of materials and repair schedules, DLA is better able to meet the needs of the service repair facility.
“Collaboration with our customers is a vital aspect of demand planning,” said Carroll. “Sometimes it is difficult to forecast what you think your usage will be.” This is especially the case when it comes to ramping up for war or ramping down for withdrawal from theaters of operation. “The demand picture just explodes when you go to war,” said Carroll. “You are trying to figure out what will explode and ramp up quickly for those items. On the other hand, when peace breaks out, the question becomes how you ramp down quickly from wartime buying rates.”
Communications and collaboration are equally important in both of these scenarios. DLA sits down with its customers to discuss their understanding of demand right down to the nut, bolt and screw. “They give us their best guess for what the demand for these items will be,” said Carroll. “When you are relying on human intelligence for demand planning, it is best to get information from those who are as close to the warfighter as possible.”
Improving supply chain visibility is a continuing and ongoing process that will always inch forward—but may not actually achieve—full demand and supply synchronization. But this continuous improvement is important.
“For me the real untapped opportunity is to merge the supply chain procurement functions of DLA with TRANSCOM’s movement capabilities,” said Fletcher, “and put that together the way major commercial firms do it today.”
DLA’s performance measures include metrics like demand and planning accuracy. “We are looking at criteria such as absolute percent forecast errors, material availability to customers, and customer wait time,” said Carroll. “We are also talking to customers what the most appropriate measures are for demand planning and forecasting.
“Forecasting has always been wrong and it probably will never be perfect,” Carroll added. “The trick is to constantly review the process. The past has been a good teacher, but we are looking at better ways to manage more erratic demand. There is no perfect solution but there are ways to make improvements to tools and processes. We try to find ways to get the best value out of them and to make sure that they work from both the statistical and collaboration parts of the process.”
The tightening of the U.S. defense budget is one thing that DLA and the rest of DoD will have to come to grips with. As Congress moves to cut defense spending, demand forecasters will have to become more cognizant of the costs of their planning and decision-making.
“There has been a decided shift toward the recognition of the financial implications of decisions over the last two years,” said Fletcher. “As organizations have flattened, commanders in the field have become more and more aware of the costs of doing business. There are finite limits on what you can spend so have you have to make some choices. Now that war supplementals are ending, the services are going to have to learn to live within their own budgets. Better demand planning and forecasting is an opportunity to reduce costs.”
“DoD has had some constraints historically in that achieving the mission was always number one,” said Manning. “But now with budget restraints being applied, focusing entirely on the mission is not feasible going forward. All of the services will have to consider efficiency as well as effectiveness. They can use their gut intuition and experience as a starting but they will have to back that up with clear cut metrics and analysis. That is the way of the future.” ♦