Abstract: To study the non-uniform spatial and temporal distributions of the early-age temperature of rock-filled concrete (RFC), this study develops a numerical method for simulating heterogeneous RFC, considering the layered filling process of self-compacting concrete (SCC) and discrete rockfill particles. We examine variations of RFC temperature by combining with the data from the prototype experiment of temperature in the Shibahe reservoir. The results show that in the initial stage of SCC pouring and filling, the mixing of concrete with non-uniform placement temperature is completed in several hours. The second stage of the RFC early age, or a synchronous temperature rise process of the rockfill and SCC, takes roughly 2-3 days, during which the rockfill promotes hydration heat absorption with a rockfill temperature rise and a rising rate both lower than those of SCC. After that, the overall RFC temperature tends to be stable and homogeneous with a trend of decreasing slowly. On a lift surface, the rockfill ratio of its different areas may be slightly different, and a higher value occurs in the area with lower average temperature. In summer construction, attention should be paid to the areas near steel formworks, because they feature a high temperature peak and steep temperature gradients due to few rocks in their upstream seepage-prevention layers or downstream triangle areas. Concrete pouring imposes a significant influence on the shallow surface layer of a lower bin, causing a secondary temperature rise of about 6 ℃ and little effect on the layer below the 1.0 m depth. For a lift surface with a short SCC pouring period, an efficient alternative is to use a homogeneous RFC model to simulate the temperature during the construction period, but it should select reasonably an equivalent placement temperature and an equivalent adiabatic temperature rise.

Key words: rock-filled concrete, temperature simulation, heterogeneous concrete, layered filling process, early-age temperature characteristics