A Description of the COBRA-IE Code and Its Assessment Against Void Fraction Data


October 9, 2015

David Aumiller
Senior Advisor
Bettis Atomic Power Lab

Adjunct Associate Professor
University of Pittsburgh

Thursday, October 8, 2015, 2:00 pm
800 22nd Street NW, SEH B1220
Washington, DC 20052

Hosted by: Dr. Philippe Bardet ([email protected])

Abstract:
A variant from the COBRA-TF code series has been developed that incorporates many new features and enhances computational capability and accuracy.  The new code, referred to as COBRA-IE, has been developed as a general purpose thermal-hydraulic analysis code, with an emphasis on analysis of Loss-of-Coolant Accidents (LOCA) when used in conjunction with an integrated code system using the R5EXEC, RELAP5-3D, MELCOR, and CONTAIN programs.  This presentation provides an overview of the COBRA-IE code describing the advanced physical models and capabilities that have been developed and implemented in the code.  Of particular note are the inclusion of the ability to treat non-condensable gases, the development of a three-field counter-current flow limitation (CCFL) model, implementation of mechanistic dispersed flow film boiling (DFFB) models, the development of RELAP5-3D compatible water properties, and the creation of several nonlinear solution algorithms.  Finally, implementation of thermal-hydraulic, neutron kinetic and control system coupling in COBRA-IE is described.  The overall accuracy of programs such as COBRA-IE is tied to the ability to predict void fraction.  As such, a comprehensive assessment has been made using one-dimensional void fraction data.  The results of this assessment are provided.  The assessment utilizes data from 9 different experimental facilities.  It includes data from air-water and steam-water facilities, heated flow, adiabatic flow, subcooled boiling, saturated boiling, co-current up flow and co-current down flow.  Approximately 1100 data points are evaluated and included in this assessment.  Overall, COBRA-IE was able to predict the void fraction with an average error (predicted – experimental) of less than 0.04.  Plots describing the relationship between the error in the prediction and parameters such as pressure and flow are also provided.  The presentation will close with a discussion of the Rickover Fellowship Program in Nuclear Engineering (http://scuref.org/rfp-01).

Biographical Sketch:
Dr. Aumiller is the lead author of the COBRA-IE analysis code at the Bettis Atomic Power Laboratory.  In his 19 years at Bettis, he has published works in the areas of coupled code systems, code assessment, scaling, CHF model development, annular flow models, dispersed flow film boiling and CCFL.  He is also an Adjunct Associate Professor at the University of Pittsburgh where he teaches 2 classes in the graduate nuclear engineering program.