Work in progress. This part of the site is still under construction. More results including graphics and animations will appear very soon.

The Accretion Disc Boundary Layer in Cataclysmic Variables


NEW: Preliminary results.

Accretion onto a star, for a relatively hot gas, therefore demanding less resolution. Here the ratio H/r = 0.15, is about 3 times larger than in Cataclysmic Variables, and is therefore more relevant to accretion onto a pre-main sequence star in Young Stellar Objects. At this stage there is no heating or cooling in the energy equations (no radiation and the heating due to viscous dissipation is not added).

A short movie

shows the evolution of this preliminary models. The density is shown on a logarithmic color scale. The disk is shown edge-on, and only the first quarter (upper right) of the r-theta plane is shown. The star is rotating at only a fraction (0.3) of the Keplerian speed. The matter spreads quickly towards the pole to eventually englobe the whole star. (the brown-yellow-green spot in the upper rigth corner of the picture is for color reference).

The same movie is shown when the initial density profile has been substracted. This enables to see from where in the disk, the matter is initially released to be accreted onto the surface of the star. This movie is slightly shorter than the previous one and stops a little bit before the matter spread completely to the poles.
1. OBJECTIVES

I propose to carry out a numerical study of gaseous astrophysical discs as found in some binary systems (Cataclysmic Variables - CVs), in which a star is losing mass that is transferred to its companion (a white dwarf). This matter forms a flat disc around the white dwarf and, due to viscous dissipation, the matter in the disc is slowly accreted onto the surface of the white dwarf. The main unsolved problem in these systems is the boundary layer region between the fast rotating inner edge of the disc and the slowly rotating stellar surface. In this region the disc is expected to slow down until its velocity matches the velocity of the stellar surface. The disc looses kinetic energy which is expected eventually to be emitted by the inner disc. The luminosity of the inner edge of the disc predicted by one-dimensional simulations[8,17,22,23,25] is large, however, two decades of observations have not shown any evidence of a luminous (and hot) boundary layer[4,18]. I plan here to carry out two-dimensional simulations of the boundary layer region to solve the long standing problem of the missing boundary layer.

2. SIGNIFICANCE, IMPACT AND OUTCOMES

The way the matter is accreted from the disc onto the star has direct consequences on other important processes. The accreted material, after it accumulates onto the surface of the white dwarf, undergoes a thermonuclear runaway - TNR (nova). In order to model correctly the TNR event one has to know whether the material from the disc has spread toward the poles of the star or has remained in the equator[26]. It is not known either whether the equatorial material on the star is accelerated (as some observations suggest[2,5]) or not. Convective currents due to Eckman pumping (or even turbulence due to the shear) in the outer envelop of the star (and in the accreted envelop) could take place, and would mixed the freshly accreted material (Hydrogen-H) with heavier elements (e.g. P,Al,..) found in the star. During a TNR the heavy elements would be ejected in the interstellar medium and affect the composition of the interstellar gas. The problem of the BL is therefore important not only to explain the observations of CV systems, but also to predict the TNR events, the mixing of the elements in the outer envelop of the white dwarf and the heavy elements ejected in the interstellar medium[26], and most importantly this would have significant implications in the theory of STAR FORMATION. In this picture, the heating of the star due to BL radiation and advection of energy would also affect the TNR event[24].

3. PRELIMINARY STUDIES

I have been working on the boundary layer between the accretion disc and the accreting star since I started my PhD[9]. I carried out one-dimensional (time-dependent) simulations of the boundary layers in many star-discs system, ranging from discs around white dwarf in Cataclysmic Variables[8,10] to discs around Young stars[11,13] and around low mass stars[12]. I have shown that the energy in the boundary layer is radiated radially (therefore, heating the star) and consequently the temperature of the boundary layer is lower than first expected[8]. However, the observed overall luminosity (energy released by radiation) of the boundary layer is still much less than in these theoretical models. I also found[15] that the BL energy is advected into the star (again resulting in heating of the star) when the rate of the mass accretion is high. This last result explains partially why in high mass accretion rate CVs system the BL is not observed. However, a theoretical model is still missing, for low mass accretion rate systems.

I have already modeled and written many two-dimensional hydrodynamic codes[6,7,14], and I have also developed and tested a two-dimensional multigrid method[19,20,21] that will be the basis for a new numerical code for the boundary layer problem (more details on the code and the reason to chose the multigrid method are given in the methodology section).

4. METHODOLOGY}

4.1 The numerical modeling

The main task of the research will be to solve numerically the equations governing the boundary layer region between the star and the disc. The equations consist of the Navier-Stokes equations, an energy equation, an equation of state and eventually an equation for the diffusion of the elements. The equations will be written in spherical coordinates (r,theta,phi). I will assume the flow to be axisymmetric around the vertical axis of rotation of the star, and consequently there will be no dependence on phi. I will solve the equation in the first quarter of the plane (r,theta) with the disc located in the theta = pi/2 plane (with a finite thickness). The radius will extent initially from the stellar surface (eventually the outer layer of the degenerate star will be included) up to an outer radius of the order of a few stellar radii. The equations will be written in finite difference form (second order) for the space derivatives, while the temporal scheme will be implicit (a Crank-Nicholson solver will be used). The implicit equations (for the diffusive terms) will then be solved using a multigrid solver for the following reason.

In order to solve the system, one usually needs to invert a matrix, which is computer-time consuming. A two-dimensional numerical simulation of the boundary layer was carried out[16] but the model could be followed only for a relatively short time scale, because of the inversion of the matrix. The multigrid method[1] (a variant of the relaxation method) is faster than the matrix inversion by 1-3 orders of magnitude[19] (depending on the number of dimensions of the problem and the number of grid points - the larger the problem, the faster the relative implementation). Moreover, the multigrid method can easily be implemented to work on a parallel computer and achieves an additional 1-2 order of magnitude in speed[19] over the usual (non-parallel) matrix inversion. Even for a modest performance of the multigrid code, I expect to be able to solve the problem of the boundary layer on the long (physical) evolution time scale, using much less computer time than any matrix solver.

4.2 Plans of the Research

First I will develop a purely hydrodynamical two-dimensional time-dependent code, using a simple polytropic approach (P=P(rho) only, neglecting the energy equation and the diffusion of elements). This will allow to check the formation of an Eckman layer, and check the possibility of an Eckman pumping, an accretion belt, circulation, and maybe even turbulence due to the shear in the boundary layer. This simplistic approach should already be able to show whether the matter has a tendency to spread towards the poles of the star, or if it is confined to the equator due to the centrifugal force[3].

Later, I will include an energy equation with a source term (due to viscous dissipation) and a leak term (radiation), with an ideal gas equation of state. Here I should already get a final answer to the question of the spreading of the matter towards the poles, the formation of an accretion belt, the thickening of the inner disc, and mainly whether the flow is hydrodynamically unstable and produces turbulence and/or short period oscillations (dwarf nova oscillations) in CVs?

Eventually, I will take into account the diffusion of elements and I will include the outer layer of the stellar surface (the outer layer of the white dwarf is degenerate). In this final stage I expect to answer the following: what is the depth of the accretional heating in the outer envelope of the white dwarf versus latitude, and what is its thermal time scale? how much spinup of the star results from angular momentum transfer to the star by the accreted matter? what is the belt rotational velocity? and mainly does convective mixing and dredgeup due to shear mixing occur? \\

REFERENCES

Brandt, A., 1977, Math. Comput., 31, 333

Cheng et al., 1997, AJ, 114, 1165

Durisen, C., 1977, ApJ, 213, 145

Ferland, G.J., Langer, S.H., MacDonald, J., Pepper, G.H., Shaviv, G., & Truran, J.W., 1982, ApJ, 262, L53

Gansicke, Bauermann, 1996, A & A, 309, L47

Godon, P., & Shaviv, G. 1992, "Two-Dimensional Boundary Layer Modeling", Vinia Del Mar Workshop on Cataclysmic Variable Stars, ASP Conference Series, 29, Nikolaus Vogt (ed.)

Godon, P. & Shaviv, G. 1995, "The dynamics of two-dimensional local and finite perturbations in envelopes of rotating White Dwarfs. I", ApJ, 447, 797

Godon, P., Regev, O., & Shaviv, G. 1995, "One-Dimensional Time-Dependent Numerical Modeling of Accretion Disc Boundary Layer ", MNRAS, 275, 1093

Godon, P. 1993, Ph.D. Thesis, "Investigation of the Boundary Layer problem by the Spectral Method", Technion, Haifa, Israel

Godon, P. 1995, "The propagation of acoustic waves and Quasi-periodic Oscillations in Accretion Discs Boundary Layers", MNRAS, 274, 61

Godon, P. 1996, "Accretion disc boundary layers around pre-main sequence stars", MNRAS, 279, 1071

Godon, P. 1996, "Hot boundary layers in Symbiotic binaries AX Persei and CI Cygni", ApJ, 462, 456

Godon, P. 1996, "Accretion discs boundary layers in classical TTauri Stars", ApJ, 463, 674

Godon, P. 1997, "Numerical Modeling of Tidal effects in polytropic accretion discs", ApJ, 480, 329

Godon, P. 1997, "Advection in accretion discs boundary layers", ApJ, 483, 882

Kley, W., 1989, A & A, 222, 141

Lioure, A., Le Contel, O., 1994, A & A, 285, 185

Mauche, C.W., Wade, R.A., Polidan, R.S., van der Woerd, H., Paerels, F.B.S., 1991, ApJ, 372, 659

Meier, D.L. & Godon, P., "Introduction of Multidimensional Finite Element Techniques to Astrophysics", Interim Report on Task \# 673, Director's Research and Development Fund, Jet Propulsion Laboratory, NASA, October, 1997

Meier, D.L. & Godon, P., 1998, in Numerical Astrophysics 1998, ed. K. Tomisaka and K. Shibata, (Dordrecht: Kluwer), in press. "Multi-Dimensional Astrophysical Structural Analysis: A New Field of Numerical Astrophysics"

Meier, D.L. & Godon, P., 1999, "Multi-Dimensional Astrophysical Structural and Dynamical Analysis. II. Fast, Parallel Solution of the Astrophysical Structure Equations", in preparation

Narayan, R., & Popham, R., 1993, Nature, 362, 820

Regev, O, Bertout, C., 1995, MNRAS, 272, 71

Regev, O., Shara, M., 1989, ApJ, 340, 1006

Regev, O., 1983, A & A, 126, 146

Shaviv, G., Starrfield, S., 1987, ApJ, 321, L51