The timescales of capital investments, and therefore the turnover dynamics of capital stock, have limited representation in macroeconomic modelling. This hinders analysis of the economic inertia produced by these timescales, which is particularly important in the context of a rapid net zero transition in which vast quantities of long-lived investments may need to be prematurely abandoned. We set out to determine the minimum model that is required to accurately represent the turnover dynamics of fixed capital. We develop a quantitative framework for estimating the turnover time of fixed capital assets in the US economy, and derive the annual distribution of both total fixed capital stock and new investments across timescales. We find that these can be effectively aggregated into three major timescale components which can be easily incorporated into integrated assessment models.