Turbulent channel flow of drag-reducing polymer solutions is simulated in minimal flow geometries. Even in the Newtonian limit, we find intervals of "hibernating" turbulence that display many features of the universal maximum drag reduction (MDR) asymptote observed in polymer solutions: weak streamwise vortices, nearly nonexistent streamwise variations and a mean velocity gradient that quantitatively matches experiments. As viscoelasticity increases, the frequency of these intervals also increases, while the intervals themselves are unchanged, leading to flows that increasingly resemble MDR.The energy dissipated in turbulent channel or pipe flow of a liquid can be dramatically reduced by low levels of long-chain polymer additives [1][2][3]. The most striking qualitative feature of this phenomenon is the existence of a so-called maximum drag reduction (MDR) asymptote [1]. For a given flow geometry at a given pressure drop (i.e. at a given Reynolds number Re), there is an asymptotic maximum flow rate that can be achieved through addition of polymers. Changing the concentration, molecular weight or even the chemical structure of the additives has no effect on this asymptotic value. This universality is the major puzzle of drag reduction.Turbulent flow in or near the MDR regime displays important differences from Newtonian turbulence. Its most commonly discussed signature is a distinctive mean velocity profile U mean (y) that displays clear log-law behavior well-approximated by a formula given by Virk: U + mean = 11.7 ln y + − 17.0 [1] (Superscript "+" denotes quantities nondimensionalized in inner velocity and length scales τ w /ρ and η/ √ ρτ w ; τ w is the time-and area averaged wall shear stress, η and ρ are fluid viscosity and density and y is distance from the wall.) Additionally, streamwise vortices, which dominate near-wall dynamics in Newtonian turbulence, are significantly weakened at MDR. Low-speed streaks become much less wavy in the streamwise direction and the streak spacing is substantially larger [4][5][6][7]. In addition, the Reynolds shear stress near MDR is substantially smaller than in Newtonian turbulence [8][9][10][11]. Indeed, the smallness of Reynolds shear stress is a central issue in a recent phenomenological model of MDR [12]. Based on these observations, many researchers have suggested that turbulence in the MDR regime is "transitional" [3] or "marginal" [12] in some sense that is not yet well-defined, and that the spatiotemporal flow structures that sustain turbulence in this regime are substantially different from those of normal Newtonian turbulence. In the latter case, one simulation approach that has been fruitful in identifying the self-sustaining structures is the so-called minimal flow unit (MFU) approach [13,14]. This approach identifies the smallest flow domain (at a given Reynolds number) in which turbulence can be sustained. Accordingly, temporally intermittent phenomena can be identified more readily than in a large box, where spatial averages incorporate different regions in ...