Groundbreaking for the four-story, 253,000-sf Terascale Simulation Facility (TSF) was held in Livermore, Calif., in April 2002 and the project was completed ahead of schedule and within budget in late 2004. The scope of the project called for the construction of a facility that could be adapted to meet the installation and operation requirements of computers over the next 30 years. The schedule carried with it a simple request to have the facility constructed before the next generation of supercomputers arrived.

The TSF is home to two supercomputers built by IBM to perform trillions of operations per second. The computers are designed to support the Advanced Simulation and Computing (ASC) Program, which is part of the Department of Energy's National Nuclear Security Administration.

The first computer, known as Purple, performs sophisticated engineering and physics simulations to help researchers determine the safety and effectiveness of aging nuclear weapons in the absence of nuclear testing. The other computer, referred to as BlueGene/L, handles challenging ASC-related scientific simulations, such as molecular dynamics for materials science.

Accommodating the powerful machines required the installation of more than 12 megawatts of electricity to operate the computers, an additional 10 megawatts to power the mechanical units necessary to generate more than two million cubic feet per minute of cooling capacity to cool the computer floor and to supply the administrative load.

"By that time, we were bringing in enough electricity to power a city of about 25,000 people," says Anita Zenger, project controls manager for plant engineering at LLNL. "With a project of this magnitude, we had to have a very clear outlook of what the scope, cost, and schedule were going to be. We know that commitment to a proactive risk-management process delivers project success."

Success Drivers

Project success is contingent upon having the proper drivers in place to achieve desired objectives. Factors that fueled the TSF project included a high-functioning integrated team, a clear vision of future project events, a realistic risk-based contingency plan, and the best-value contracting acquisition method.

The integrated team included all stakeholders ranging from the client, owner, contractors, and construction management personnel to the project management team, security employees, and all other involved parties.

"We had very strong owner and client involvement. The client was at all of our weekly meetings and the owner talked to the team on a weekly, if not daily, basis," says Zenger. "All of the key participants were involved in a formal partnering session. This allowed us to have ground rules for discussions, and even disagreements, and then determine how we would resolve those issues. The expectations and terms must be clearly understood by everyone because too many cooks can spoil the soup."

It is also important to be willing to change participants, if necessary, as the project progresses.

Using a best-value contracting strategy allows a contract to be awarded based on various criteria rather than just the lowest cost. Criteria could include proposed key team members, safety record, experience modification rating, history of claims, primary subcontractors, costs, references, quality, and schedule.

"When we got the proposals, we reviewed them all, compared the cost versus all of these criteria, and were able to establish the best-value contractor," says Zenger.

Risk-Management Process

The overall effectiveness of the risk-management process hinges upon the availability of a realistic contingency plan that specifies how each particular risk should be mitigated.

A formal risk-management process includes five steps that begin with the actual planning followed by the identification of risks through a screening process. The defined risks are then analyzed to determine their level of threat—whether it be high, medium, or low. Handling the risks involves developing mitigation procedures and tracking the progress of the risk-abatement strategies. The final step is monitoring and conducting periodic reviews of the process.

"The risk-management process is not only proactive, but it is also continuous," says Zenger. "Throughout the five phases of a management process, you are looking at risk management the whole time. It is not just something done in planning. You do it through the entire life cycle of the project."

The methodology used in the risk-screening process is a list of questions and considerations that are reviewed to determine if a risk falls into a particular category. Examples of risks that were addressed during the TSF project include construction safety, unforeseen or unreasonable claims, changes in supercomputer design that could affect the design of the building, trenching in congested areas, interaction with an outside agency for the trench work, and unusual discoveries that could be made during excavation.

"We had an interesting risk because we were designing and building a facility to house the next generation, 30 years, of supercomputers. The computer that was originally going to go in the building had not yet been designed," says Zenger. "We had a risk: What if they change the base design of the computer?"

The risk-analysis process features a very formal risk identification form that includes the risk category, the date the risk was assessed, statement of risk, probability of occurrence, consequence in terms of cost and schedule if the risk is not mitigated, and determination of risk level.

Applying Risk-Handling Strategies

There are four risk-handling strategies: mitigate, accept, avoid, and transfer. Mitigating the risk lowers the probability and consequence, but there may be a residual risk such as impacting the project cost or schedule. Accepting the risk means there will be no cost for mitigation and the residual risk will equal the original risk. Avoiding or transferring a risk means there may be a potential schedule or budget impact, but there is no residual risk.

Mitigating—Since the laboratory is located on top of an old Navy air station, it was necessary to trench a mile across it in order to tap into a sufficiently large electrical capacity. The risk of trenching in a congested area was mitigated by doing water jet potholing to uncover utilities to verify their depth and exact location.

Construction safety concerns were mitigated by using the best-value contracting strategy to ensure the best contractors were working on the project.

Accepting—Planners accepted the risk that any type of unusual discovery could be made during the excavation.

"A mammoth was found during the excavation for another project near our site," says Zenger. "The mammoth was found very deep and TSF was not going very deep so we were able to accept that risk."

Avoiding—The risk of involving an outside agency on the trenching and crossing a regulated drainage arroyo was avoided by boring under this particular area.

Transferring—The transfer strategy was employed to solve the risk about the potential computer design changes affecting the design of the TSF. The owner agreed to assume the risk associated with potential computer design changes.

"You must have periodic reviews of potential risks," advises Zenger. "As you start going through the life cycle of a project, you might uncover new risks and you want to bring those through the risk-management process, as well."

The Finished Product

The TSF office tower, which accommodates more than 250 employees, features research-and-development spaces for visualization and hardware prototyping, a 150-seat auditorium and visualization theater for unclassified presentations, a second theater for classified presentations, an atrium-like lobby, three small computer rooms to support the facility's infrastructure, conference rooms, a classroom, and an operations hub which controls the computers. The auditorium is located inside what is called the Armadillo, a structure that resembles the armored body of an armadillo on the east side of the lobby.

The computers, situated on four-foot raised floors to allow for the convenient placement of cables, are housed in the two rooms that occupy the second level of the two-story supercomputing wing. A wall separating the two rooms can be easily removed to create one large room if there is a future need for more floor space. A mechanical utility room, located on the ground floor beneath each computer room, includes 28 air-handling units that blow cool air up to the second level through the grated floor tiles at a rate of 80,000 cubic feet per minute. The air continues to rise after cooling the computers and is forced into large return-air plenums for recirculation.

Clear-span construction techniques opened up floor space through the elimination of impediments that could restrict computer components. Electrical and mechanical systems were strategically placed to avoid problems that might be created with the use of copper connector cables that can only span 25 meters.

Plenty of natural light flows into interior offices through floor-to-ceiling windows located throughout the building.

Words of Wisdom

Continually following a specific risk-management process is critical to the success of large-scale projects.

"If you follow the process and have your risk-handling strategies in place, it is a great way to manage those risks. We also found that it is beneficial to put risk-mitigation measures on the project schedule," says Zenger. "Risk management is much like fire prevention instead of firefighting. If you don't put it on the schedule, there is a potential that somebody will forget they were supposed to do this mitigation measure because they are too busy putting out fires."

Issues that rank at the top of the risk-screening list of priorities should be addressed first and then energy should be devoted to handling risks ranging in degree of importance. Project participants should be encouraged to attend risk meetings by ensuring it is a pleasant experience by providing refreshments. Successfully addressing a risk is cause for celebration and may be a good occasion to hold a party.

After one risk is mitigated, it is important to look for the next challenge which must be handled.

"It's similar to a critical path," notes Zenger. "Once you get an issue off the critical path, you see what the next issue is and you determine how to get it off the critical path."

By Tracy Carbasho