ermilab today is a laboratory poised on the brink of the greatest physics run of its life. The trouble is, the brink keeps moving.
When Collider Run II begins, Fermilab's Tevatron, the world's highest-energy accelerator, souped up with the brand-new Main Injector, will produce more data about more particle collisions at the energy frontier than ever before in the history of particle physics. There is every reason to believe that in those data lie discoveries that will change the way we think about the fundamental nature of matter. There is nothing that Fermilab, the physicists who use Fermilab and the agencies that fund Fermilab want more than to get started on the new collider run. Yet the beginning of Collider Run II at the Tevatron sometimes seems exasperatingly hard to attain.
Although the earliest "baseline" plans called for Run II to begin in 1999, more recently Fermilab had announced that Run II would begin with the new millennium, in 2000. Hitting the year-2000 Run II target meant building the new Main Injector accelerator and Antiproton Recycler, reconfiguring the Tevatron and Antiproton Source, and commissioning the resulting new accelerator complex. It also required completing two massive upgrades of Fermilab's collider detectors, CDF and DZero, totally reinvented from their Run I configurations to handle two inverse femtobarns of data from the collider, 20 times more than in Run I. The Main Injector is finished, but completing and commissioning the new accelerator system will take until September 2000. And in September 1999 the detector collaborations broke the news that they would not be ready for full-fledged operation until early 2001.
With so much riding on the Run II start date, what accounts for the detectors' schedule slip? And, perhaps more significant, what now leads Fermilab's management and the collaborations themselves to believe that the current schedule is one they can--no kidding--really achieve?
Collaborators and lab management largely agree on the reasons for the detectors' schedule slip. The upgrades, all acknowledge, are extraordinarily challenging projects that force detector technology to unprecedented new levels. They incorporate unique, new, one-of-a kind systems, each requiring extensive R&D efforts that make it extremely difficult to set realistic schedules in the projects' early stages. They require components that vendors have never built before, manufactured to unheard-of standards of precision. Often suppliers overestimate their capability to produce components and underestimate the time it will take. Other Fermilab commitments--finishing Run I, building the Main Injector, and other projects--competed for scarce manpower and funding. The funding for the detectors felt the overall squeeze on the U.S. high-energy physics budget. Project management structures had to adjust to project realities. And, perhaps most fundamentally, the long-held Fermilab practice of basing schedules on the most optimistic possible assumptions resulted in timetables that were not grounded in realistic assessments by those closest to the work.
What has changed, now, to convince collaborators and laboratory leadership alike that the current schedule, for beginning Run II in March 2001 means what it says? Experimenters and managers cite four main factors. First, the projects are far enough along toward completion that most of the R&D and vendor surprises are likely behind them. Second, Fermilab has unequivocally made the detector upgrades its clear priority and allocated the manpower and resources necessary to complete them. Third, the Fermilab director has established a new kind of working relationship with the collaborations. Last, and perhaps most significant, the term "Fermilab schedule" is taking on a new meaning.
Detectors are difficult.
Everyone agrees that building a 5,000-ton state- of-the-art particle detector, one designed to wring every ounce of physics from the trillions of particle collisions the accelerator sends its way, is a very challenging job. In fact, challenging may not be the word.
"Both of these detectors have brand-new systems that no one has ever built before," said Director Mike Witherell. "The fiber tracker at DZero is an all-new design. The central outer tracker in CDF is a major design advance. Together, the detectors' silicon systems are 10 times bigger than their predecessors'. The front-end electronics for silicon detectors must be completely developed at the first assembly stage, rather than being added at the end, the way we have typically built detectors in the past."
DZero Cospokesman Harry Weerts agreed.
"It would be easier if we could use mass-produced stuff," Weerts said, "but for the majority of components there are lots of technical development steps before we can go into production. In many cases, we are asking vendors for levels of precision that they have never encountered before. It all takes time."
Herding cats.
The collaborative nature of detector building makes it still harder to predict and plan a project's progress. Project managers have the task of coordinating the efforts of 500 independent-minded, independently funded collaborators to design, engineer, construct and commission the detector they will ultimately use to do new physics. "Herding cats," is how they describe it, and it isn't conducive to strict scheduling.
"Accelerator construction projects are centrally managed," Witherell observed. "The people working on the project are all part of the laboratory's line management organization. But detectors are built by collaboration. We have invented the institution of the international detector collaboration to build these multi-kiloton `Swiss watches.' Resources are spread over many funding agencies, and collaborators operate in an environment of academic freedom. Essentially, the workers on these projects are all volunteers. When outside people see this, they can't imagine we can ever build anything. We can, but it isn't easy."
Heard the one about the Fermilab schedule?
Like the term "military intelligence," the saying goes, "Fermilab schedule" is an oxymoron. When he was a user himself, Fermilab director Mike Witherell told the Fermilab users' meeting last July, he knew that if a Fermilab schedule actually referred to the future, it was only a draft.
And, said CDF co-spokesman Franco Bedeschi, the laboratory's sliding scheduling philosophy had a vicious-circle effect.
"When managing an experiment, one has to balance the desire to build the best detector possible with the time it takes to build it," Bedeschi said. "This leads you to make different choices depending on how much you trust the laboratory schedule."
In production at last.
Now, say collaboration leaders, they can see light at the end of tunnel. For one thing, they have reason to believe that many of the unpleasant, time-gobbling surprises of earlier stages are behind them. Most of the R&D is done, and vendors are producing and shipping the requisite components at a rate that permits reliable extrapolation.
"Until you start to build something, you really cannot anticipate all the problems," Bedeschi said. "A schedule is always partly guesswork until then. Now, for most things, we are in production. Disaster is always possible, but we can now extrapolate from production rates with much more confidence."
Across the ring at DZero, Cospokesman Hugh Montgomery reported a similar situation. "We like to think that there are very few areas where we are not currently in production," he said.
Priorities and Partnership
For Fermilab's part, the message to collaborators is clear: completing the detector upgrades is the lab's priority, backed up by the resources to get the job done.
"We are giving the detector collaborations all the resources they need," said Deputy Director Ken Stanfield. "All the resources that can usefully be brought to bear on the detectors to get them ready to run are being brought to bear."
At the same time, Witherell said he expects a new level of partnership between the detector collaborators and laboratory management.
"Successful projects are not the ones without problems," Witherell said, "but the ones in which problems are brought into the open, analyzed, discussed, and worked on together to find solutions. The project management for the upgrades needs to know that we are collaborating with them. In turn, we need to know that the project management is being straightforward with us. We must have a good give and take."
The message is getting through.
"There is a very positive attitude in the lab," said CDF Cospokesman Al Goshaw. "It is clear that the upgrades are the highest priority, and that we should speak up if we need help. Mike Witherell came to talk to the CDF executive committee about his vision for Fermilab and for the detector upgrades. It had a very inspirational effect. We know the lab is behind us, and in turn we are committed and obligated to the laboratory. We have decided that every collaborator on CDF must have a part in getting the detector ready. Last week we began `installation shifts' that will continue through June 2000. Every collaborator is in the shift pool, to come in and work for at least a week. It's a requirement for authorship. Franco and I will be taking shifts along with everyone else. I'm looking forward to it as a vacation from being a spokesperson."
Time to Step up to the Plate
The collaborations have also heard Witherell's message about a change in the philosophy of Fermilab schedules, a signal that is reinforced at biweekly upgrade project management meetings with the director. With input from the Beams Division, DZero and CDF, the Directorate has produced a schedule that Witherell describes as realistic but aggressive, calling for Run II to begin in March 2001.
"It can be hard to separate using a schedule as a motivational tool from using it to plan the work on a timetable that all agree you can make," Witherell said. "I believe we now have a schedule that we have a good chance of meeting. But it won't be easy. From here on, it's not a question of vendor deliveries. The things with the most impact on the schedule are under the collaborations' control. Now it is up to them to step up to the plate."
Associate Director Mike Shaevitz expanded on the new Fermilab definition of "schedule."
"Now it means not the earliest possible date you could conceivably imagine, but the date when you believe, with high confidence, that you can finish," Shaevitz said. "I hope it works. If it does, then people will start believing it."
Out at the collaborations, they believe. Weerts said DZero is now committed to realistic schedules.
"No more taking it as a given that schedules always slip at Fermilab," he said. "Maybe the `optimistic' type of schedule works for some people, but overall, that system isn't working. We have to stop doing business that way. We are determined that the date will not slip any more. Every two weeks we formally review the schedule at DZero. Meanwhile, it's simple. I have one piece of paper with a checklist of milestones. I go down the list continually and find out what's on track and where we might be having problems. When I find trouble, we figure out how to fix it."
CDF's Goshaw agreed.
"The lab has listened to our input, and we have been very frank about our problems. From their experience in project management, they have added some time, to create a realistic schedule. They have made clear that we will have the resources we need. Now-- that's it. That's what we have to meet. If we fail this time around, we all have the message that it would be very damaging not just to Fermilab but to high-energy physics. We have a big responsibility to make this happen."
The new Fermilab "we-really-mean-it" Run II schedule calls for CDF to start a commissioning run in August of 2000, and for both detectors to begin taking data in Collider Run II in March 2001. Will they roll into the Tevatron when they say they will?
DZero's Weerts had the final word
"The only way to restore our credibility is to perform," he said.
