MIT team finds that the ratio of component atoms is vital to performance.
The Hanford facility in Washington state stores millions of gallons of hazardous chemicals and radioactive waste--the legacy of past weapons-production activities. Complete disposal of the waste will take decades. In the meantime, the Hanford facility must maintain and operate the waste-storage tanks so as to ensure environmental and human safety.
For the past two years an Energy Laboratory team has been developing a computer model that will help managers at Hanford find ways to do the job faster, cheaper and better.
The model, which simulates Hanford operations using about 2,600 components to represent people, tasks, costs and other factors, should help Hanford managers develop strategies for achieving their goals more efficiently and for responding to external and internal changes with minimal disruption.
The researchers have performed a number of case studies using the model. For example, in one study analyzing the impact of a proposed budget cut, the model showed that focusing the cut on selected activities and personnel would delay the ultimate waste cleanup by some years but that imposing the cut across the board would postpone cleanup indefinitely.
The model is being developed by Professors Kent F. Hansen and Michael W. Golay of the Department of Nuclear Engineering; the Energy Lab's Sangman Kwak, a technical staff member, and Malcolm A. Weiss, a senior research staff member.
THE HANFORD PROBLEM
In the past, the Hanford plant produced weapons-grade plutonium and with it vast amounts of hazardous waste. Much of that waste, about 60 million gallons, is now stored at Hanford in 177 underground tanks. The US Department of Energy (DOE) views operation of those waste tanks as its most important safety problem. About a third of the tanks are known or suspected to have leaked waste into the ground. And the chemical composition and physical state of much of the waste is unknown.
Some activities at Hanford relate to permanent disposal of the waste, but much work focuses on nearer-term "waste tank safety and operations" (WTSO). The WTSO program includes four tasks: ensuring safe operations at the tank site, managing the waste safely, complying with environmental regulations and characterizing the waste. In 1994, the budget for the WTSO program was $412 million. The work force included more than 1,900 employees. And the WTSO program will continue (at some level) until the ultimate disposal of the waste is complete, at least 40 years hence.
Formulating a model that can simulate Hanford operations is a challenge. Traditional models for simulating such operations are static, using a spreadsheet approach to describe a system. But Hanford operations are complex, dynamic and subject to outside influences such as Congressional budget cutting.
These features make traditional approaches to modeling unsuitable. The MIT researchers therefore turned to system dynamics. This approach constructs a quantitative description of the elements of a system, including both physical quantities such as work accomplished and employment level and "soft" components such as attitudes and worker morale. Elements that interact are linked by time-dependent feedback loops.
To develop the model, the researchers had to identify all the critical components and feedback loops at Hanford. They then assigned starting values to the components and rates to the feedback loops that affect those starting values. To formulate and quantify the model, they used information gathered in interviews with key Hanford personnel. The completed model describes the important activities and features of the WTSO program such as work flow, personnel and productivity. Finally, the model provides for changes and events that may affect the behavior of the system. These include changes in the organization of the program and in technology.
Among the case studies conducted using the model, one explored how reorganizing the Hanford system would affect productivity. For this study, the researchers assume that an emergency in one part of the Hanford facility requires extra managers, so the experienced WTSO managers are removed and replaced by less experienced people. Not surprisingly, the model shows that the productivity of the managers drops dramatically.
But the impact does not end there. Other workers continue to perform their usual tasks, but their productivity depends on being guided by the managers. The inexperienced managers are not familiar with their new jobs so are slower to make decisions and offer guidance. The productivity of the general workforce also drops significantly. The results emphasize that any reorganization leads to an initial loss in productivity, which translates into lost money. Thus, if managers reorganize to achieve some benefit they must be sure that that benefit will exceed the initial cost. The model can help managers assess such trade-offs.
The WTSO program is only one of many possible applications of the system dynamics methodology to waste management. Other activities at Hanford include disposing of the waste in the tanks as well as disassembling and cleaning up plutonium-producing reactors. Cleanup is also needed at the rest of DOE's "nuclear weapons complex," a total of 14 sites across the country. Appropriate models could help managers at each of those sites be more effective. Models could also be developed to represent the entire DOE nuclear weapons complex.
This research was supported by the DOE through the Los Alamos National Laboratory and the Pacific Northwest Laboratory.
A version of this article appeared in MIT Tech Talk on March 13, 1996.