Postal Service uses software to manage labor pool
- By Florence Olsen
- Apr 19, 2004
In Philadelphia, 2,500 U.S. Postal Service employees work in a mail-processing facility that stretches across five floors. Three floors in an adjacent truck terminal and a nearby rail yard hum with other mail-sorting activities day and night.
The Philadelphia facility — and others like it — present a challenge for USPS workforce managers: how to assign the right workers to the right place at the right time.
To deal with the problem, managers have begun using optimization software, dubbed "the solver" but more formally known as the Labor Force Schedule Optimizer System.
"They need this type of tool," said Walter O'Tormey, USPS' manager of processing operations.
Under pressure to cut costs and offer better service, facilities managers say the planning tool has helped them operate more efficiently at several test sites. Now, as USPS expands its use of the solver, postal executives say they expect to see a substantial return on their investment because it is the most comprehensive tool they have for modeling their mail operations.
The tool gives postal managers a uniform way of calculating the optimum number of jobs in their facilities. It also calculates the time of day when those jobs should begin, the time they should end, and what the off days should be, according to Phillip Pensabene, manager of operations technical support for USPS.
Determining job assignments is nothing new for the postal agency. "The new part is we have a mathematical, standardized approach" for making those decisions, thanks to the scheduling tool's ability to solve linear, mixed-integer and quadratic programming problems, Pensabene said.
The solver is based on CPLEX optimization software by ILOG Inc., a French company with offices in Mountain View, Calif.
In 2001, when USPS officials first used the planning tool in a series of pilot and proof-of-concept tests, they eliminated 1,600 full-time equivalent positions out of 11,500.
Now that operations managers are using the tool in at least 19 postal plants, managers expect to begin scheduling mail-processing work more efficiently in those facilities. As many as 200,000 postal workers who work as package and letter sorters could be affected by the postal agency's use of the optimizer to model workers' jobs, O'Tormey said.
But for the most part, operations managers only in the largest and most complex mail-processing facilities will use the labor optimizer. "We're not looking to go to Wenatchee, Wash.," O'Tormey said. Smaller plants are easier to manage, he said, and a tool such as the optimizer isn't needed.
Within USPS, specialists know what data to enter into the software to create an accurate model of operations within a particular mail-processing plant. Each plant has its own profile data, Pensabene said. The tool is typically used by an operations specialist "who has a very good working relationship with the data and understands both mail flows and equipment utilization," he said.
Nevertheless, the use of tools such as the workforce optimizer are often seen as antilabor, said Omer Benli, associate professor of information systems at California State University at Long Beach. But instead, Benli views the use of such tools as necessary for businesses such as USPS to survive and employ people in the future.
Regarding costs, USPS managers can probably expect a reasonable payback from their use of the optimizer, Benli said. "In general, the return on investment for any of these tools seems to be very attractive, but [it is] sometimes very difficult to measure," he added.
By spending $14.6 million on optimization software for scheduling employee workloads, U.S. Postal Service officials expect to eliminate excess labor costs and save $135 million.
Officials will input:
Mail flow patterns.
Number and types of automated mail-processing equipment.
Numbers and skills of employees.
Postal work rules.
Union work rules.
The software will generate:
Optimal job schedules for sorting letters and packages. No unnecessary workers will be scheduled during slow periods.