The incorporation of detailed chemistry in combustion simulations is challenging due to the large number of chemical species and the wide range of chemical timescales. The performance of acceleration methods such as tabulation/retrieval strategies may deteriorate dramatically when large variation in the accessed composition space is present. In this study, a dynamic adaptive acceleration method (DAAM) is proposed, in which in situ adaptive tabulation (ISAT) or dynamic adaptive chemistry (DAC) is dynamically selected for chemistry integration based on the encountered composition inhomogeneity. The principle component analysis (PCA) of instantaneous representative compositions is employed to identify a low-dimensional subspace, in which the composition inhomogeneity of the computational cells is quantified through reconstructing the histogram of composition. ISAT is invoked for cells being in composition regions with high cell/particle numbers to avoid unnecessary tabulations and DAC is employed for the remaining ones by invoking on-the-fly reduction to generate small skeletal mechanisms for local thermo-chemical conditions and therefore accelerates the chemistry integration. A heuristic approach that dynamically adjusts the DAC reduction threshold based on the user-defined ISAT error tolerance has been proposed for DAAM, which enables a single, intuitive error control parameter for the combined use of these two methods and more importantly enables rigorous local error control. DAAM have been demonstrated in internal combustion engine (ICE) model simulations and premixed charge compression ignition (PCCI) engine simulations of n-heptane/air mixture, respectively. DAAM can improve the acceleration performance up to 50% compared to standalone ISAT while maintaining the same level of accuracy in temperature and species. It also shows advantage in speedup performance over the fixed ISAT-DAC method at the same level of accuracy. (C) 2018 The Combustion Institute. Published by Elsevier Inc. All rights reserved.