2. Numerical Experiment

[7]  A complete system of MHD equations was solved numerically by using a modified PERESVET code [Podgorny and Podgorny, 1996]. An absolutely implicit finite difference scheme conservative with respect to magnetic flux was used. It was solved by the iteration method. Owing to the use of the conservative scheme, the errors due to inaccuracy of approximation of div B by its finite difference analog could be avoided. The numerical method also included automatic multilevel division of the time step in the regions of large gradients of quantities. These capabilities of the scheme allowed stabilization of the majority of numerical instabilities. The code provided for graphical presentation of the magnetic field lines in the three-dimensional space. Anisotropy of thermal conductivity in the magnetic field, compressibility, and dissipative processes were taken into consideration. The dimensionless units of density (r0 = 104 mi=1.67 times 10-20 g cm-3 ), temperature ( T0 = 20 eV), magnetic field ( B0 = 1 G), velocity ( V0 = B0/4 pr0 approx 2times 109 cm s-1 ), length ( L0 = 8 R approx 6 times 1011 cm), time ( t0 = L0/V0 sim 300 s), and current density ( j0 sim 1.3 times 10-12 A cm -2 ) were used. Here, mi = 1.67times 10-24 is the mass of proton, and R= 6.9 times 1010 cm is the Sun's radius. The system of three-dimensional MHD equations in the dimensionless form is

eq001.gif(1)

eq002.gif(2)

eq003.gif

eq004.gif(3)

eq005.gif

eq006.gif

eq007.gif

eq008.gif(4)

[8]  In equations (1)-(4), Rem = L0V0/nm0 is the magnetic Reynolds number, nm0 = c2/4ps0 is the magnetic viscosity for conductivity s at temperature T0, and s is the conductivity, s0/s = T-3/2. The dimensionless coefficient b0 = 8pn0 kT0/B02, where n0 = r0/mi. It should be emphasized that b0 in this form is not the ratio between plasma pressure and pressure of magnetic field at a definite point in space, it is simply a dimensionless coefficient expressed through the dimensionless units used for the calculations. The term including viscosity does not exert a considerable influence on the results; it is important for increasing the stability of the finite difference scheme. Re = r0 L0 V0/h is the Reynolds number, h is the viscosity, Gq = L(T0) r0 t0/T0, L(T dimens ) is the radiation function for ionization equilibrium in the corona, T dimens= T0T. L'(T) = L(T dimens)/L(T0) is the dimensionless radiation function. In the problem being solved, radiation did not have a strong effect; eparallel, eperp 1, eperp 2 are the orthogonal unit vectors parallel and perpendicular to the magnetic field; kdl = k/(Pk0 ) is the dimensionless coefficient of thermal conductivity along the magnetic field; P = r0 L0 V0/k0 is the Peclet number; k0 is the thermal conductivity at temperature T0; k is the thermal conductivity; k/k0 = T5/2; kperp dl = [(kk0-1P-1) (kB k0B-1PB-1)]/ [(kk0-1P-1) +(kB k0B-1PB-1)] is the dimensionless coefficient of thermal conductivity in the direction perpendicular to the magnetic field; PB = r0 L0 V0/k0B is the Peclet number for the thermal conductivity across a strong magnetic field (when the cyclotron radius is much smaller than the free path). Thermal conductivity across a strong magnetic field is denoted as kB; and k0B is its magnitude for temperature T0, plasma density r0 and magnetic field B0; kB/k0B = r2B-2 T-1/2. Gg G is the dimensionless gravitational acceleration. Gg= t02/L0, G is the gravitational acceleration, and g is the adiabatic constant.

[9]  The parameters used for calculations were: g = 5/3, Rem=8times 104, Re=104, b0=8 times 10-6, P=2, PB=2 times 106. (We repeat here, that b0 is not the ratio between plasma pressure and magnetic pressure at a definite point in space, it is only coefficient in equations (1)-(4), which is determined by dimensionless units of pressure and magnetic field). In numerical and laboratory simulation it is impossible to use very big and very little dimensionless parameters. These parameters are chosen bigger or less than unit, but not precisely by the same order of magnitude. The principles according to which the dimensionless parameters were chosen were described by Podgorny and Podgorny [1995, 1996]. For the numerical calculations, the grid 41 times 41 times 41 was used; therefore the magnetic Reynolds number was ~50.

2004GI000077-fig01
Figure 1
[10]  The computational domain was a cube with sides 8 R, in dimensionless units 0 le x le 1, 0 le y le 1, 0 le z le 1. The positions of the Sun, corona, and computational domain in the y = 0.5 plane are shown in Figure 1. The dipole giving rise to the magnetic field B0 approx 0.8 G on the Sun's surface is parallel to the Z axis. The magnetic moment of the dipole in dimensionless units is M1={M1x=0, M1y=0, M1z=9.6 times 10-2}. Its center is at the point R1 = {x1 = -0.217, y1 = 0.5, z1 = 0.5}. On the face x = 0, at the point y = 0.5 and z = 0.5, the magnetic field is 0.15 G. The lines of the
2004GI000077-fig02
Figure 2
initial (dipole) magnetic field are presented in Figure 2a.

[11]  The vacuum medium cannot be described in the framework of the MHD approximation. For this reason, at t=0 in the computational domain, an extremely low concentration of 10-1 cm -3, whose influence on the dynamics of the corona plasma that expanded in the computational domain was negligibly small, was specified. At the initial moment of time, the temperature inside the region was taken to be as low as 20 eV.

[12]  At t = 0, a thermal expansion of the corona began. The corona parameters ( rc/mi = 2 times 107 cm -3, Tc = 200  eV) were specified on the face x=0 in the circle formed by the intersection of this face and a spherical surface with the center ( -0.217, 0.5, 0.5) and radius 2R (Figure 1). The center of the circle was ( y=0.5, z=0.5 ), and its radius was 0.125. The velocity of plasma outflow from the corona was found from the continuity equation, so that the mass flow in the numerical experiment corresponded to the loss of the Sun's mass carried away by the solar wind ( sim10-14 of the Sun's mass per year). From this condition, the velocity of inflow into the region Vx=2.5 times 10-4 was specified at the boundary x=0 in the circle with the center ( y=0.5, z=0.5 ) and radius 0.125. The self-consistent values of r, T, and V automatically established in the process of numerical solution of the MHD equations at thermal expansion of corona plasma. The current of the current sheet could freely flow in and out through the planes y=0 and y=1.

[13]  It should be noted that, formally, the parameter b0 did not correspond to the ratio between pressures at any point. At the point x=0, y=0.5, and z=0.5 the b parameter is 8. In the vicinity of this point in the computational domain, where the condition approximated the vacuum one, b=4 times 10-9 at the initial moment.

[14]  The most delicate problem in specifying the boundary conditions is setting the magnetic field at the rightmost boundary X=1. In the process of calculations, and hence extension of field lines along the X axis, all components ( Bx, By, Bz ) undergo significant changes, and the tilt of the field line with respect to the X=1 plane also changes. Therefore it is impossible to specify the magnetic field at the boundary X=1. Setting j=0 is also unacceptable because, in the process of extension of field lines, the current sheet reaches the right boundary. The best approximation is to assume dj/dx=0. To be more certain that the obtained solution corresponds to the conditions in the solar wind, a layer boundary X=1 which is by a factor of 2-3 thicker than the current sheet can be excluded from the consideration.


AGU

Powered by TeXWeb (Win32, v.2.0).