Application of distributed computing for resource scheduling

Conference · Fri Nov 01 04:00:00 EST 1996

OSTI ID:389933

Abdul-Rahman, K H; Thiyagarajah, A; Aganagic, M ^[1]; Shahidehpour, S M ^[2]

---

1. Siemens Energy and Automation, Brooklyn Park, MN (United States)
2. Illinois Inst. of Tech., Chicago, IL (United States). Dept. of Electrical and Computer Engineering

The short term resource scheduling is one of the most important tools for optimal operation of power systems. The objective of the resource scheduling is to minimize the total system operation cost including production, maintenance, start up and shut down costs subject to system and individual unit constraints. Due to its economic implications, the resource scheduling problem has been an active research subject for several decades. Although a number of solution methodologies have been proposed to this complex problem, consistent generation of optimal schedules for power systems of practical sizes has been extremely difficult. Other approaches for short term resource scheduling based on the augmented Lagrangian relaxation (ALR) technique were proposed. The ALR is obtained by adding quadratic terms to the objective function of the standard Lagrangian relaxation (LR). The ALR algorithm has proven to be effective in improving both convexity and convergence properties of the resource scheduling solution. In this paper, the authors introduce a distributed processing approach coupled with an efficient ALR method to further improve the solution quality and speed up the resource scheduling algorithm. In the commitment stage, subproblems corresponding to individual units are solved using distributed processing on several processors. DP is used to solve the individual unit subproblems. The distributed processing implementation is based on the Parallel Virtual Machines (PVM) software that is capable of integrating different computer configurations. PVM is a software that enables machines with widely different architecture and floating point representations to work together on a single computational task. Under PVM control, a user defined group of computers appears as one large distributed memory machine eliminating the need for expensive hardware and software required for supercomputers.

• OSTI ID: 389933
• Report Number(s): CONF-960426--
• Country of Publication: United States
• Language: English

Related Subjects

24 POWER TRANSMISSION AND DISTRIBUTION
DISTRIBUTED DATA PROCESSING
ECONOMICS
PLANNING
POWER GENERATION
POWER SYSTEMS