Anomalous particle pinch and scaling of v_in/D based on transport analysis and multiple regression

Predictions of density profiles in current tokamaks and ITER require a validated scaling relation for v(in)/D where v(in) is the anomalous inward drift velocity and D is the anomalous diffusion coefficient. Transport analysis is necessary for determining the anomalous particle pinch from measured density profiles and for separating the impact of particle sources. A set of discharges in ASDEX Upgrade, DIII-D, JET and ASDEX is analysed using a special version of the 1.5-D BALDUR transport code. Profiles of rho(s)v(in)/D with rho(s) the effective separatrix radius, five other dimensionless parameters and many further quantities in the confinement zone are compiled, resulting in the dataset VIND1.dat, which covers a wide parameter range. Weighted multiple regression is applied to the ASDEX Upgrade subset which leads to a two-term scaling rho(s)v(in)(x')/D(x') = 0.0432[(L-T,L-c((x) over bar')/rho(s))(-2.58) + 7.13 U-L(1.55)nu(e*)((x) over bar')(-0.42)]x' with x' = rho/rho(s), effective radius rho and average value (x) over bar'. The rmse value of the scaling equals 15.2%. The electron temperature gradient length LT, is the key parameter of the anomalous particle pinch which yields the main contribution. A further parameter is the loop voltage U-L which introduces the electron collisionality parameter nu(e*). All exponents are statistically significant. The parameters U-L and nu(e*) suggest a new anomalous particle pinch term driven by the Ohmic inductive electric field. The nonlinearities in the two-term scaling show that quasilinear theory is disproved by experiment. Regression analysis of the whole dataset VIND1.dat from four tokamaks shows that the L-Tc/rho(s) scaling covers the dependence of rho(x)v(in)/D on the effective plasma radius. It is further found that the rho(s)v(in)/D values from transport analysis do not respond to a change in collisionality regime and are not clearly related to the prevailing turbulence type. The new scaling law predicts for ITER high values of rho(s)v(in)/D and peaked density profiles, caused by the L-Tc/rho(s) term and central heating due to alpha particles. The density peaking improves the energy confinement by s ome 20%.