EUROPEAN TRANSACTIONS ON ELECTRICAL POWER Euro. Trans. Electr. Power 2005; 15:371–380 Published online 11 April 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/etep.46 A high-speed algorithm for unified simulation of large-scale power system dynamics K. Al-Anbarri, R. Ramanujam* ,y , P. Rajesh and K. Kuppusamy Department of Electrical and Electronics Engineering, College of Engineering, Guindy, Anna University, Chennai 600 025, India SUMMARY An algorithm for unified and economic simulation of short-term and long-term dynamics of a large-scale power system is described in this paper. The algorithm is a combination of (i) a high-speed algorithm with significant improvements and modifications for large time step width requirement, and (ii) a variable time step algorithm which employs artificial damping that damps out faster interunit electromechanical oscillations. Automatic switchover between short-term and long-term modes, user control of step widths and the numerical stability for large step widths make the algorithm ideally suitable for voltage stability analysis involving slow dynamics and real-time simulation for training of system operators. Copyright # 2005 John Wiley & Sons, Ltd. key words: real-time simulation; long-term dynamics; training simulator; scenario-building; artificial damping 1. INTRODUCTION Power system stability studies range from short-term studies which involve faster synchronizing oscillations to long-term studies which include study of slow dynamics of over-excitation limiter, on- load tap changing transformers, boiler dynamics and automatic generation control. A continually variable step approach with step size determined by accuracy and numerical stability has been proposed in Reference [1] for unified simulation of fast and slow dynamics. Such an algorithm is ideal for off-line studies but will not be suitable for real-time unified simulation. In the simultaneous implicit approach [2], the differential equations for the internal voltages of synchronous machines and state variables of control systems are discretized using the trapezoidal rule, and combined with algebraic equations of the network to yield equations of the form ½Y½U¼½I ð1Þ The non-zero entries on the right-hand side of the above equation correspond to generator nodes. Repeated solution of Equation (1) and constraint on step width significantly increases the total Copyright # 2005 John Wiley & Sons, Ltd. *Correspondence to: R. Ramanujam, Department of Electrical and Electronics Engineering, College of Engineering, Guindy, Anna University, Chennai 600 025, India. y E-mail: maanju_ra@yahoo.com