Making use of the exact equations for structure functions, supplemented by the equations for dissipative anomaly as well as an estimate for the Lagrangian acceleration of fluid particles, we obtain a main result of the multifractal theory of turbulence. The central element of the theory is a dissipation cut-off that depends on the order of the structure function. An expression obtained for the exponents sn in the scaling relations ( Questions of small-scale universality in fluid turbulence hover around the universality of the scaling exponents ξ n,0 of velocity structure functions defined through relations such aswhere u(x) is the velocity component along the separation distance r, measured at the position x and ǫ is the mean rate of energy dissipation. Here r lies in the inertial range given by η << r << L, where L is the large-scale at which the energy is being injected and η ≡ (ν 3 /ǫ) 1/4 is the dissipation scale, ν being the fluid viscosity. The zero index in ξ n,0 shows that no powers of the transverse velocity increments are involved in this particular definition (1). Kolmogorov [1] assumed that the velocity fluctuations in the inertial range are independent of both L and η, and that ǫ, regarded as equal to the energy flux across scales, is the only relevant dynamical parameter. As is well known, Kolmogorov's proposal yields the linear relation ξ n,0 = n/3. Since, in the limit of vanishing viscosity (or, as η → 0), Kolmogorov's scaling theory combines the exact expression [2] S 3,0 = − 4 5 ǫr, it is reasonable to regard the theory loosely as dynamic. However, experimental and numerical data in three-dimensional turbulence have shown (see Ref.[3] for a recent account) that the scaling exponents ξ n,0 depart from n/3, and that there exists a more complicated nonlinear spectrum of scaling exponents ξ n,0 . Its theoretical explanation for the velocity field has proved to be elusive, though considerable progress has been made for passive scalars [4].In recent past, the problem of scaling exponents in turbulence has been analyzed within a general framework of the theory of multifractal (MF) processes reviewed in Ref. [5]. This approach has led to interesting interpretations and novel work (see [5,6] for incomplete list), but its shortcoming is the lack of connection with the dynamical equations. In this paper, a main relation of the MF theory is derived from dynamical equations, supplemented both by an order-of-magnitude estimate for the Lagrangian acceleration of a fluid particle, and the earlier work on dissipative anomaly [7,8].For background, we review here the main ideas of the inertial-range MF theory, whose basis are the assumptions that (a) the velocity increments δ r u have the formwhere u ′ ∼ δ L u may be regarded as the root-mean-square value of u, and (b) there exists a spectrum of exponents h related to the fractal dimension of their support D(h).
Thus, (r/L)3−D(h) is proportional to the probability of the velocity increment falling within a sphere of radius r on a set of dimension D(h). It is clea...