We have computed six two-dimensional simulations with two cospatial, interacting fluids that represent the stars and gas in a disk galaxy. The two fluids interact through star formation, mass loss, stellar winds, and supernovae, and the gas includes an optically thin cooling function. Our previous simulations have been able to produce the multiphase nature and the overall topology of the cold and warm interstellar medium (ISM) and here we extend the previous work by allowing for million-degree gas and an impulsive form of heating that represents supernovae. The six simulations differ in the form and rates of energy injection; for simulations with heating only from stellar winds and from both supernovae and stellar winds, we have run three simulations with low, moderate and high energy injection rates (with a factor of 4 difference between adjacent rates), where the moderate energy injection rate corresponds to that of the Galaxy. The three values of Edot in our simulations correspond to 2 (low), 8 (moderate), and 32 (high) × 10<SUP>41</SUP> ergs s<SUP>-1</SUP> for a disk of 16 kpc and are equivalent to supernova rates of 0.0075, 0.03, and 0.12 yr<SUP>-1</SUP>, respectively (assuming E<SUB>SN</SUB> = 6 × 10<SUP>50</SUP> ergs, and a ratio of energy released from supernovae to that from stellar winds of 3:1). We use a grid that is 2 kpc across and ±15 kpc in the vertical direction and impose a constant gravitational potential along this direction. <P />Our simulations create a three-phase medium with filaments of dense, cold and warm gas surrounding bubbles of hot gas, which are usually hundreds of parsecs across and can exceed 1 kpc in size. This filamentary topology is very similar to that inferred for our Galaxy and others, based largely on H I observations. The evolution of the cold, dense filaments is dominated by a loss of identity from filament-filament collisions, although in higher energy injection cases a Rayleigh-Taylor instability can cause a filament to fragment. <P />We calculate central densities, scale heights for the density, and filling factors of the three phases of gas, and demonstrate that the vertical density distribution of each phase is usually best fit with more than one component. The calculated central densities, scale heights, and filling factors for each gas phase reproduce the observational values made for the Galaxy in the moderate energy injection rate simulations. Also, the computed median pressure and pressure scale height best reproduce the Galactic values in the moderate energy injection rate simulations. <P />Velocity information for the cold (or neutral) gas is analyzed at one or two times in each simulation. The occasional multicomponent nature of H I emission profiles and the holes of H I in position-velocity plots occur in the simulations, although we fail to recreate quantitatively the amount and velocities of high-velocity gas that are observed in the Galaxy. Specifically, only the highest energy injection rate cases have significant amounts of neutral gas (more than 0.25% of the mass of cold gas) at |υ| ≥ 50 km s<SUP>-1</SUP>, and the maximum velocities are smaller than those observed in H I (≳100 km s<SUP>-1</SUP>) We attribute these shortcomings to numerical viscosity in the simulations. <P />Further analysis of the cold gas velocities reveals an anticorrelation between the column density of cold gas and its velocity dispersion for a galaxy viewed face-on. Also, one calculation of the net mass flux of each gaseous phase as a function of height demonstrates that most of the hot gas is rising, while cooler gas is falling, which is consistent with the galactic fountain model. <P />Varying the rate of energy injection has a large effect on the nature of the ISM, especially the extent of the cold gas, so we can constrain to a relatively narrow range the energy injection rate within the Galaxy. Our results suggest that neither the low-Edot simulations nor the high-Edot case can reproduce the distribution (i.e., central densities, scale heights, and filling factors) of the multiphase ISM in the Galaxy, although the high-E simulations are more effective at reproducing the high-velocity H I observed in the Galaxy.
Read full abstract