It may be time for the U.S. Postal Service to go back to the drawing board on the Flats Sequencing System, which faces trouble on two fronts.

USPS is pressing ahead with installation of the football-field-sized FSS machines, even though the system recently fell short of standards for the second time in an acceptance test. The throughput rate improved over a test late last year but was still 23% short of the standard, USPS's inspector general's office recently revealed.

The inspector general's report recommended installing only one more FSS machine "until the system demonstrates operational stability and successfully passes the field acceptance test." The machines are supposed to automate the labor-intensive process of handling such flat mail as catalogs and magazines. (See "The Unofficial Guide to Flats Sequencing" for more information.) Only a handful of the 100 machines in Phase I of the FSS program have been installed.

USPS management acknowledged that FSS "did not meet the key contract performance requirements" and said it would conduct another acceptance test, but indicated it would not slow deployment of the machines.

"When fully deployed, the expected annual FSS savings are $600 million. As such, it is in the best interest of the Postal Service to take advantage of every available opportunity to sustain the deployment while ensuring that it does not adversely affect the forecasted savings and/or increase operational burdens," wrote David E. Williams, acting vice president at USPS, in response to the inspector general's report.

The math seems to be on Williams' side. Achieving annual savings of $600 million with an investment of only $1.4 billion looks like a good deal; even if the machines last only 15 years, that would represent a 40% internal rate of return. That gives the Postal Service and lead contractor Northrop Grumman plenty of incentive to fix the machines.

But then there's the second problem with FSS: Not only is it not yet living up to its design, but the design may already be obsolete.

The system was designed under the assumption that flats volume would continue growing into the future. Large postal rate increases, competition from other media, and the economic recession have reversed that trend, causing flats volume to decline at an annual rate of about 11% the past two years.

USPS management has responded by increasing the territory that will be served by Phase I machines. A few days ago, it added more ZIP codes to the program, bringing the total to 2,314 instead of the approximately 1,300 in the original design.

That means each machine will serve far more addresses than originally intended, some in ZIP codes that did not qualify for the program at first. And some of the machines will go to locations that didn't originally pass muster for Phase I. All of this suggests that the $600 million estimate, which has been around for at least a year, is no longer valid.

Insiders say postal executives have already concluded that the machines are too big and need to be tweaked in other ways. But estimating the level of flats volume 5 or 15 years from now, or what type of machine could best handle that, could be quite a challenge.

Failure of the machines to meet contractual standards so far might be a blessing in disguise for the Postal Service: That could give it leverage to force Northrop Grumman to make desired changes to the design. Or even grounds to cancel the contract and go back to square one.

To anyone who has heard postal officials talk enthusiastically about FSS, the idea of USPS abandoning the program seems extreme.

But consider this fact: Much of the work on and learning about FSS has been done at Dulles, VA, where three machines are running and a fourth is being altered. Five days ago, the Postal Service announced that it has begun a feasibility study "to determine if efficiency could be increased by consolidating mail processing operations" from Dulles to the nearby Northern Virginia facility in Merrifield.