Managing government funded scientific consortia

In recent years, it is becoming apparent that good science not only requires the talents of individual scientists, but also state-of-the-art laboratory facilities. These faculties, often costing millions to billions of dollars, allow scientists unprecedented opportunities to advance their knowledge and improve the quality of human life. To make optimum use of these experimental facilities, a significant amount of computational simulations is required. These mega-projects require large-scale computational facilities and complementary infrastructures of network and software. For physical sciences in US, most of these research and development efforts are funded by the US Department of Energy (DOE) and National Science Foundation (NSF). Universities, US National Laboratories, and occasionally industrial partners work together on projects awarded with different flavors of government funds managed under different rules. At Fermilab, we manage multiple such collaborative computing projects for university and laboratory consortia. In this paper, the author explores important lessons learned from my experience with these projects. Using examples of projects delivering computing infrastructure for the Lattice QCD Collaboration, the author explains how the use of federal enterprise architecture may be deployed to run projects effectively. The author also describes the lessons learned in the process. Lessons learned from the execution of the above projects are also applicable to other consortia receiving federal government funds.