In this study, combined heat and power units are incorporated in the practical reserve constrained dynamic economic dispatch, which minimizes total production costs considering realistic constraints such as ramp rate limits and valve-point effects over a short time span. The integration of combined heat and power units and considering power ramp constraints for these units necessitate an efficient tool to cope with joint characteristics of electricity power-heat. Unlike pervious approaches, the system spinning reserve requirements are clearly formulated in the problem and a novel charged system search algorithm is proposed to solve it. In the proposed algorithm a novel self-adaptive learning framework, adaptive selection operation and repelling force modeling are used in order to increase the population diversity and amend the convergence criteria. The proposed framework is applied for three small, medium and large test systems in order to evaluate its efficiency and feasibility. (C) 2013 Elsevier Ltd. All rights reserved.