Parallel processing strategies for simultaneous modular flowsheeting

Conference · Fri Jan 01 04:00:00 EST 1988

OSTI ID:5587531

LaRoche, R D; Chen, Y T; Subramanian, D ^[1]

---

1. Pennsylvania State Univ., University Park, PA (USA). Dept. of Chemical Engineering

This study explores the use of the parallel processing capabilities of advanced computer architectures in chemical process flowsheeting. Widely used sequential modular software techniques are adapted using the simultaneous modular concept where parallel processing can be employed. This paper presents a formulation of the simultaneous modular approach to exploit multiprocessing. Experience with the Pittsburgh Supercomputing Center Cray X-MP/48 and the Cornell National Supercomputing Facility IBM 3090/600E is presented. Optimization test problems are presented where the interconnection equations are solved at the flowsheet level using a modified Han-Powell successive quadratic programming (SQP) algorithm. This paper develops methods to maintain well balanced module level tasks in a multiprocessor environment. The stream vector and objective function gradient (Jacobian) computation phase is a major focus since this phase typically accounts for 70% of the total optimization problem computation. Both petrochemical and electrochemical examples are presented to illustrate the significance of highly nonideal thermodynamic property effects.

• OSTI ID: 5587531
• Report Number(s): CONF-881143--
• Country of Publication: United States
• Language: English

Related Subjects

02 PETROLEUM
020400 -- Petroleum-- Processing
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY
400400 -- Electrochemistry
99 GENERAL AND MISCELLANEOUS
990210* -- Supercomputers-- (1987-1989)
ARRAY PROCESSORS
CHEMICAL ENGINEERING
CHEMICAL PLANTS
CHEMISTRY
COMPUTER CODES
ELECTROCHEMISTRY
ENGINEERING
INDUSTRIAL PLANTS
OPTIMIZATION
PARALLEL PROCESSING
PERFORMANCE
PETROCHEMICAL PLANTS
PHYSICAL PROPERTIES
PROGRAMMING
THERMODYNAMIC PROPERTIES
USES