Nanofluids performance counts on nanoparticle configuration. The evolutionary topology optimization algorithm is applied in this work for direct nanoparticle configuration optimization. The initial random distribution of nanoparticles inside base fluid is found to well simulate practical nanofluids. The optimized nanofluids configuration proves to be continuous strips assembled by nanoparticles with width comparable to size of nanoparticles and length comparable to size of heat -transferring system. System overall thermal resistance is significantly reduced by the optimized nanoparticle configuration. Future nanofluids should be fabricated with strip-shaped nanoparticles in order to perform better. (C) 2017 Elsevier Ltd. All rights reserved.