Efforts to meet climate change mitigation and energy security targets spur investments in variable renewable energy sources. Their implications for the operation of power plants are frequently investigated drawing on unit commitment and dispatch models. However, the temporal granularity and operational detail these models consider is commonly omitted in the broader family of long-term energy system models. To compensate this short-coming, these two types of tools have sometimes been 'soft-linked' and harmonised for limited simulation years. This paper assesses an alternative approach. We examine an extended version of an open source energy system model (0SeMOSYS), which is able to capture operating reserve and related investment requirements within a single tool. The implications of these model extensions are quantified through comparison with an Irish case study. That case study examined the effects of linking a long-term energy system model (TIMES) with a unit commitment and dispatch model (PLEXOS). It analysed the year 2020 in detail, applying a yearly temporal resolution that is over 700 times higher than in OSeMOSYS. Without increasing temporal resolution (and computational burden) we show that results of the enhanced OSeMOSYS model converge to results of TIMES and PLEXOS: Investment mismatches decrease from 21.4% to 5.0%. The OSeMOSYS analysis was then extended to 2050 to assess the implications of short-term variability on future capacity investment decisions. When variability was ignored, power system investments in 2050 were found to be 143% lower. This might imply that energy policies derived from such long-term models - of which there are many - may underestimate the costs of introducing variable renewables and thus meeting climate change or energy security targets. (C) 2014 Elsevier Ltd. All rights reserved.