Disc breaking and parametric instability in warped accretion discs 2024 Disc breaking and parametric instability in warped accretion discs – 3 Loren E. Held, Gordon I. Ogilvie Abstract We present the first local simulations of disc breaking/tearing in a warped accretion disc. Warps can arise due to a misalignment between the disc and the rotation axis of the central object, or a misalignment with the orbital plane of a binary (or planetary) companion. Warped discs can break into rings, as found in observations of circumbinary protoplanetary discs and global simulations of tilted discs around spinning black holes. In this work we isolate the mechanism of disc breaking in high-resolution, quasi-2D, local (shearing box), hydrodynamic simulations of a Keplerian disc. We consider the evolution of a free (unforced) warp in the wavelike ( α < H / r \alphaH/r ) or wavelike regimes ( \alpha1 , so we will omit a detailed discussion for the sake of brevity. The key results are that the disc becomes parametrically unstable and does not break, though it does produce some shocks at early times. By the end of the simulation the disc in the A=9H_{0} run looks similar to the disc shown in the bottom panel of Figure 10 for the A=2H_{0} run. The similarity of this run to the \psi_{\text{max}}\sim 0.14 run suggests that the transition from parametric instability to disc breaking likely occurs around \psi_{\text{max}}\sim 1 , at least for warps of wavelength \lambda\sim 90H_{0} . 5.1.2 Large warp amplitudes: \psi_{\text{max}}\gtrsim 1 Next we compare our fiducial simulation with A=28H_{0} ( \psi_{\text{max}}\sim 1.95 ) to a run with a somewhat smaller warp amplitude of A=18H_{0} ( \psi_{\text{max}}\sim 1.26 ). Note that the latter run has almost the same maximum warp amplitude as the large warp amplitude simulation in Lodato & Price 2010 . In Figure 11 we show the volume-averaged time-series of the kinetic energy from these runs (black and gold curves for A=28H_{0} and A=18H_{0} , respectively). 12 12 12 We have multiplied each time-series by the simulation’s corresponding vertical box size ( L_{z}=48H_{0} for the A=18H_{0} run and L_{z}=68H_{0} for the A=28H_{0} run) in order to eliminate the box-size-dependence in the vertical average. The time-series for the large-warp amplitude runs have similar shapes: the radial kinetic energy increases initially as the sloshing motions grow, and decreases after around 8 orbits as the sloshing motions (and the warp giving rise to the sloshing motions) are damped by shocks. We do not observe parametric instability in these runs. At these large warp amplitudes the horizontal sloshing motions are so strong that the parametrically unstable modes are likely sheared out. Interestingly, the peak in the radial kinetic energy (as well as the dissipation of vertical kinetic energy) is larger between around orbit 2 and orbit 9 in the A=18H_{0} simulation compared to the A=28H_{0} simulation. This could be because in the A=28H_{0} case the sloshing motions are more effectively damped by the stronger shocks or by the stronger bouncing. The flow field and orbit-averaged radial profiles in the A=18H_{0} simulation (not shown) are qualitatively similar to those in our fiducial simulation at A=28H_{0} (see Figures 8 and 9 ). We see transient steepening of the warp at early times where the warp amplitude is maximum (at x/H_{0}\sim-45,0,45 ), and the horizontal sloshing motions periodically converge into shocks at the peaks of the sine curve ( x/H_{0}\sim-22.5,22.5 ). As in our fiducial simulation, the disc flattens around these locations and eventually breaks into four rings. However, in the A=18H_{0} run the gaps are not as deep: we measure \langle\Sigma\rangle_{\tau}\sim\rho_{0}H_{0} at orbit 25, compared to \langle\Sigma\rangle_{\tau}\sim 0.5\rho_{0}H_{0} at the same time in the A=28H_{0} run. The warp has also been noticeably damped by this stage, with \langle\psi\rangle_{\tau} peaking at around 0.5 in the gaps in the A=18H_{0} run compared to \sim 2 in the A=28H_{0} run. 5.2 Viscosity study So far, our simulations have not included explicit viscosity. Given the high resolutions that we can reach in the shearing box (our highest resolution run was at 128 cells per scale-height H_{0} ) the numerical dissipation is likely small in our inviscid runs and they should be in wavelike regime ( \alphaH_{0}/r (diffusive). In the shearing box it is tempting to use the radial wavelength of the warp \lambda in place of r . In our simulations this is equal to the radial box size L_{x}=90H_{0} , giving H_{0}/\lambda\sim 0.01 , which suggest that our inviscid and small viscosity runs are the in wavelike regime ( \alpha\lesssim 0.003\ll H_{0}/\lambda ), while our large viscosity run is in the diffusive regime ( \alpha=0.03\gtrsim H_{0}/\lambda ). However, an implicit assumption of comparing \alpha with H/r is that radial wavenumber of the global warp k=2\pi/\lambda is roughly 1/r . If that is not the case, the relevant comparison is in fact of \alpha with kH . In that light, kH_{0}\sim 0.07 for our set-up, and so all of our warps are actually in the wavelike regime ( \alphakH_{0} . Here k=2\pi/\lambda , with \lambda=L_{x}=90H_{0} , is the radial wavenumber of the sinusoid used to initialize the warp. Thus simulations with \alpha\lesssim 0.07 fall within the wavelike regime. Run Box Size Res \alpha Regime \psi_{\text{max}}\,(A/H_{0}) zBC Riemann Reconstr. Runtime Outcome FreeWarp22 [90,0.08,68] 32/H_{0} 0.003 wavelike 1.95 (28) reflective HLL parabolic 50 breaking FreeWarp23 [90,0.08,68] 32/H_{0} 0.03 wavelike 1.95 (28) reflective HLL parabolic 50 breaking inhibited FreeWarp20 [90,0.08,48] 64/H_{0} 0.003 wavelike 0.14 (2) reflective HLLC parabolic 50 instability FreeWarp21 [90,0.08,48] 64/H_{0} 0.03 wavelike 0.14 (2) reflective HLLC parabolic 50 damped bending wave