This thesis deals with the optimisation and control of economic dispatch for power systems. The economic dispatch problem is modelled in three different algorithms to cater for many physical and system operational constraints. The first and initial modelling strategy is based on a steepest descent technique whose formulation includes load demand with spinning reserve requirements, loading and deloading rates, and limits of generation outputs. The second modelling strategy is based on a Dantzig-Wolfe decomposition principle which produces a master problem solved by revised simplex method, and a capacitated transportation subproblem solved by network flow algorithm. The third modelling strategy is based on a combination of Lagrangian Relaxation and period-by-period decomposition. The Lagrangian Relaxation is solved using subgradient optimisation while the period-by-period decomposotion is solved using priority lists. The efficacy and efficiency of all the above three methods and solution schemes have been demonstrated by case studies. The last two methods are believed to be suitable for on-line dispatch of large power system.