1 Abstract—Increasing variable renewable generation penetrations will cause increased cycling operation for conventional generating plants. Not all of these plants are necessarily well suited to such operation. Traditional long-term generation planning frameworks often neglect these operational characteristics and therefore do not reflect the operational constraints and costs associated with cycling of generating plants. Using a detailed generation dispatch model in PLEXOS, this study assesses the potential impact of short-term operational constraints and costs on future ‘high renewable’ generation portfolios obtained from a long-term portfolio planning framework. A case study of the Australian National Electricity Market (NEM) with different renewable penetrations, ranging from 15% to 85%, suggests that the technical and cost impacts associated with the operational constraints modelled are moderate even at high renewable penetrations. The extent of the impacts also depends particularly on the level of carbon price and the mix of generation technologies within the portfolios. Index Terms— Generation planning, operational constraints, cycling, flexibility, variability, renewable generation I. INTRODUCTION ARIABLE renewable generation, particularly wind and solar photovoltaics (PV), are fast becoming major generation sources in a number of electricity industries. Given their variable, somewhat unpredictable and partly dispatchable nature, there are concerns over the potential operational and economic impacts of integrating such renewable sources into power systems. In particular, they will increase the variability of net demand to be met by conventional dispatchable generation necessitating more frequent cycling (ramp up/down, startup/shutdown) operation of these units [1]. Large thermal generation has particular technical limitations and costs associated with such operation. As the shares of wind and solar generation are projected to substantially increase in the coming decade, it is important to ensure that this can be accommodated by future electricity systems [2] with large thermal plants. Hence, long-term generation planning and investment modeling needs to reasonably capture actual operational characteristics of generating plants and their ability to respond to rapid and frequent changes in net demand [3]. Key thermal plant This work has been supported in part by Australian Renewable Energy Agency (ARENA) and the CSIRO Future Grid Project. The authors acknowledge Energy Exemplar and FICO for providing academic licenses of PLEXOS and Xpress-MP solver used in this paper. The authors are with the Centre for Energy and Environmental Markets and School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney, Australia (email: peerapat@unsw.edu.au). operating characteristics include minimum generation levels, startup times and costs, ramp rate limits, and minimum up/down times. Given the associated complexities and the long time horizon involved, traditional long-term generation planning frameworks often ignore these operational constraints, and the costs associated with them. However, this may mean that generation portfolios calculated to be optimal under long-term planning frameworks may not be operationally viable or economically optimal in practice. This paper aims to assess how the technical limitations and economic impacts of generator operational characteristics, especially at high renewable penetrations, might impact on the least cost generation portfolio outcomes obtained from long- term planning tools. In particular, it compares overall future industry costs obtained from a long-term generation portfolio planning and investment modelling tool, MC-ELECT [4], against these industry costs when operating costs are obtained by solving a detailed inter-temporal constrained dispatch in PLEXOS (a commercial power market modelling tool) [5]. II. GENERATION PLANNING AND INVESTMENT MODELS A. Monte Carlo based Generation Portfolio Modelling Tool A long-term generation portfolio planning tool, ‘MC- ELECT’, developed in previous work, extends load duration curve (LDC) optimal generation mix techniques by using Monte Carlo simulation to incorporate key uncertainties [4]. The expected costs, cost risks and CO 2 emissions of a range of generation portfolios in a given future year are obtained from several thousand repeated scenarios with probabilistic input parameters. The outputs can be described as an ‘expected’ (mean) future value of annual portfolio costs. The cost spread can be represented by standard deviation and is referred to as ‘cost risk’. Financial portfolio techniques is then used to determine an Efficient Frontier that contains optimal portfolios in terms of expected costs and associated cost risk [6]. MC-ELECT was employed to analyze future generation portfolios with different renewable energy penetrations in the Australian National Electricity Market (NEM) for 2030 [7]. Fig. 1 illustrates the previously calculated optimal generation portfolios for different renewable penetration levels, ranging from 15% to 85%. This result forms the basis of the case study presented in this paper (Sections IV. V. , which explores the impact of short-term operational constraints on those portfolios determined to lie on the efficient frontier. The main limitation of MC-ELECT is inherently linked to its use of LDC, which removes chronology and prevents the Impact of Operational Constraints on Generation Portfolio Planning with Renewables P. Vithayasrichareon, Member, IEEE, T. Lozanov, Student Member, IEEE, J. Riesz, Member, IEEE and I. MacGill, Member, IEEE V