Army Flow Model

hen he was an artillery officer earlier in his career, Maj. John McKitrick would put in a personnel requisition to replace a departing member of his unit. Then he'd wait. Sometimes the soldier he requested would report for duty; sometimes the slot would go unfilled. McKitrick often wondered why one such request would be honored promptly and another put on the back burner.

Now, as a member of the Army headquarters staff, he knows a lot more about the rules that govern the complex operations of that worldwide organization. In fact, McKitrick is engaged in capturing within a set of computer systems many of the Army's business processes and practices.

McKitrick leads development of a planning and forecasting tool that's painting big pictures of military forces, now and as many as seven years in the future, for the Army's top operations managers. It's called the Army Flow Model.

The Flow Model can produce static pictures, such as a portrait of the active-duty Army in 2005 if today's funding trends continue. More usefully, it can generate a series of alternative pictures, showing the impact of alternative funding levels or recruiting strategies or other changes in resources and policies.

These are big pictures in the sense that they aggregate important management information and trends from millions of pieces of data entered at Army installations. But they're not big in another sense-they can be displayed on PC screens for any of several hundred Flow Model users on the secure Defense Department intranet.

Until recently, the information that feeds the Flow Model had to be dug out of one or more of the Army's 37 operations databases. There are separate stores of data on personnel, logistics, base facilities, budgets, and force structure. Many of these systems rely on old technology, and they store data in different computer languages.

Before the Gulf War, headquarters staff officers responded to the information needs of military decision-makers-colonels and generals-by delving into the databases and coming up with a report on say, the total costs and operational implications of fielding a new weapons system. The report would include information on training the users of the new weapons and the people who would maintain them, redistributing the previous generation of weapons, or perhaps reassigning forces.

Trouble was, the databases weren't synchronized. Some of them covered different years, and others started with different months. So the decision-makers sometimes got incomplete or inaccurate views of the situation. Two reports on related topics might present inconsistent information-"trying to answer the same question with data that's different," McKitrick says.

Resolving inconsistencies took time. These days, time is a seldom-enjoyed luxury for the armed forces. As the military's operating tempo has increased, decisions have to be made faster and faster. There's less time to debate the accuracy of numbers, to negotiate for someone else's data and to reconcile conflicting reports.

Two-Pronged Attack

Now the Army is speeding up the flow of information from its components through the staff and up to the decision-makers.

The new approach starts with a data warehouse at Army headquarters that compiles the service's major databases in a single repository. Though that seems like a simple notion, its implementation has been revolutionary within the Army, freeing many planners and analysts from time-consuming data collecting.

The Army's Director of Information Systems for Command, Control, Communications and Computers, who also is the Army's chief information officer, is building the Flow Model based on data readily available from the warehouse. The model incorporates the business rules and processes that establish relationships among the pieces of data.

For example, if a tank's engine dies and it goes out of commission for extended repairs, there are rules about how to replace it. The Army unit must first seek another tank from another unit in its own command, rather than looking for the nearest available replacement in terms of distance.

The Flow Model applies those rules to resource data, letting the Army staff organize the enormous quantity of information in the warehouse and develop projections of future situations. The model also allows analysts to test "what-if" analyses of different scenarios at a single point in time.

Force Management

The Army Flow Model is actually a set of models. Three models crunch personnel data, taking into account factors such as retirement and other attrition rates, skills and specialties, and distribution of Army personnel worldwide. Seven models produce logistics analyses, looking at such factors as the acquisition process (the Army has a seven-year procurement cycle for major weapons systems), ammunition stores and the distribution, inventory and maintenance of equipment.

Other elements build models of budgets, the Army bases that add up to the service's infrastructure, and training. At the end of the chain of models is readiness-the goal to be achieved if people, equipment and other resources are in place. "We tie everything together with a common start line," McKitrick explains.

It's obvious that such a system is a manager's dream. It does sophisticated integration of extremely complex and large data sets, producing spreadsheet-like reports. If the user wants to dig further into the information, the underlying data can be displayed with a few mouse clicks. A great deal of detail is available, based on the freshest data available. Precise equipment inventory and location information can be retrieved for 6,000 major items, such as M-1 tanks or Blackhawk helicopters.

The same information is available from the source databases, of course, but nowhere else can the analyst match up so much disparate data.

If the analyst uses the full capabilities of the model, getting a report of this kind is not instantaneous. They take hours or a couple of days. But for urgent tasks, data can be downloaded quickly to a spreadsheet and turned into a PowerPoint presentation with which to brief a general or colonel.

New Missions, New Strategies

The ultimate goal of the Flow Model's creators, McKitrick says, is to support the Army's top staff officers as they seek to reshape their fighting forces. The Army is one-third smaller than it was in 1989, and it has been engaged in overseas missions-for which it was often not fully prepared-for a decade. The deployments played havoc with some scheduled maintenance, military training and other plans for improving readiness.

These and other fundamental challenges require Army headquarters to become more flexible. It must develop new management strategies faster and more efficiently. That's why "business is getting pretty good" for the Flow Model, McKitrick says with a grin. More headquarters staff members are using it, and this summer it will be made available to some staffers in the major Army commands. An unclassified version is in the works for those without access to the secure DoD intranet.

McKitrick also hopes to provide a Web browser interface for the model soon so it will be more nearly interactive. Now it requires special software to access, although it looks to the user like a member of the Microsoft Office software family. At the technological heart of the Flow Model are Sun Microsystems Inc. Unix workstations and Oracle Corp.'s relational database management system.

Jeffrey Lerner, a logistics staff officer, is one user who says he's looking forward to being able to access the Flow Model through his Web browser. He describes the model as "the single source of information for everything we do here" and says he'd like to see it take on a new role in the future. Lerner envisions using the flow model to plan and test improvements in Army processes and business practices-that is, the rules that form the model's backbone.

The Army's Strategic and Advanced Computing Center is building the model with a staff of fewer than 20 employees and substantial support from Science Applications International Corp., a Beltway technical services contractor. The computing center began the project around 1993 with no formal funding. Funding stabilized in 1998 once the program found a sponsor: the deputy chief of staff for operations.

Now Maj. McKitrick is in charge of development, which will continue into 2005, and Maj. Lisa Keller is in charge of production. On the production side, the model is "run"-that is, updated stem to stern--several times a year. Each run can take as long as one week.

Although many of the model's benefits are in the difficult-to-quantify category of better management, it also has yielded some sizeable cost reductions. On a mainframe computer system in Huntsville, Ala., the Army used to produce updates of its 10-year Total Army Equipment Distribution Plan (TAEDP) five times a year, at an annual cost of nearly $1 million. Now the Huntsville data center staff simply loads data into a Flow Model module, a job that costs about $50,000 a year.

In fact, this year the Army expects that the measurable benefits of the Flow Model will begin to exceed its costs. As the model expands and matures, and as new requirements confront the Army's top echelons, the benefits are projected to multiply.