230 TWO-DIMENSIONAL FLOW APPLICATIONS 8.4 Channel flow with specified end pressures One of the options in DS2V is for “constant pressure boundaries”. ‘This is to allow computations for the flow through a tube or channel when the overall pressure ratio is known, but there is no information on the flow speeds. The “secondary stream” option is used to set a stationary gas at the higher pressure up to the mid-point of the channel and a stationary gas at the lower pressure in the remainder of the channel. “Reservoirs” are established at both ends of the channel. These expand to flow cross-sections that are large in comparison to that of the channel so that the flow speeds at the entry and exit are very small. The number flux of each molecular species is sampled at the mid-point and the versions of DS2V before version 4.5.10 used this flux to set the number of molecules that enter the upstream and downstream boundaries. The first test case is for a large 5:1 pressure gradient and the data report in DS2VD.TXT is: The n in version number n.m is 4 The m in version number n.m is sl The approximate number of megabytes for the calculation is 108 The flow is two-dimensional x Limits of flowfield are -7.9999998E-04 , 3.59999996-03 y limits of flowfield are @.e00eeeeE+ea , 2.¢¢100016-03 The approximate fraction of bounding rectangle occupied by flow is @.2000000 The estimated ratio of the average number density to the reference value is 1.¢¢0000 The nunber of molecular species is 1 Maximum number of vibrational modes of any species is e The number of chemical reactions is @ The number of surface reactions is e The reference diameter of species 1 is 4,1699999¢-10 The reference temperature of species 1 is 273.0000 The viscosity-temperature power law of species 1 is 0.740000 The reciprocal of the VSS scattering parameter of species 1 is 1.000000 The molecular mass of species 1 is 4.6500001E-26 Species 1 is described as Nitrogen Species 1 has electrical charge of e Species 1 has 2 rotational degrees of freedom and the constant relaxation collision number is 5.000000 The number density of the stream or reference gas is 2.5@@@000E+23 The stream temperature is 300.0000 The velocity component in the x direction is .900¢¢e0E+00 The velocity component in the y direction is @.ee@eeeeE+008.4 Channel flow with specified end pressures 231 The velocity component in the z direction is @.0080000E+00 The fraction of species 1 is 1.0¢0000 There are 1 separate surfaces The number of points on surface nied) 89 The maximum number of points on any surface is 99 The total number of solid surface groups is 1 The total number of solid surface intervals is 98 The total number of flow entry elements is @ Surface 1 is defined by 99 points Surface 1. is comprised of 3 segments Segment 1 is a straight line The segment starts at -7.9999998E-04 1.00@@@00E-03 The segment ends at @.00@000GE+20 1.9999999E-04 The number of sampling property intervals along segment is 16 Segment 2 is a straight line The segment ends at 2.00@0001E-03 1.99999995-04 The number of sampling property intervals along segment is 50 Segment 3 is a straight line The segment ends at 3.5999999E-03 2.0020@01£-03 The number of sampling property intervals along segment is 32 The data on the 98 intervals is in 1 groups Group 1 is a solid surface containing 98 intervals The gas-surface interaction is species independent Diffuse reflection at a temperature of 300.0000 The in-plane velocity of the surface is @.00@0000E+20 The normal-to-plane velocity of the surface is .0090000E+00 For all molecular species The gas-surface interaction is diffuse Rotational energy acconm. coeff. 1.00000 The fraction of specular reflection is @.00000005+00 The fraction adsorbed is 0.00000@0E+00 The side at the minimum x coord is a constant pressure boundary The side at the maximum x coord is a constant pressure boundary The side at the minimum y coord is a plane of synmetry The side at the maximum y coord is not in the flow The stream is the initial state of the flow No molecules enter from a OSMIF.DAT file ‘A flow separation boundary divides the stream and secondary stream at x = 1.00000006-03 The nunber density of the secondary strean is 4.99999995+22 The secondary stream temperature is 389.0000 The velocity component in the x direction is @.0000E+e0 The velocity component in the y direction is @.e¢000@0E+20 The fraction of species 1 in the secondary stream is 1.000000 The sampling is for an eventual steady flow The calculation employs the standard computational parameters232 TWO-DIMENSIONAL FLOW APPLICATIONS S.0+22 e+23 1 250423 Number density (/cubic m) Fig. 8.21 The number density distribution with version 4.5.09. A steady flow was established with a number flux of 5.65%102! molecules per second and the number density distribution shown in Fig. 8.21. However, when the pressure ratio was reduced from 5:1 to values of the order of 1.2:1 there was an indefinite increase on the mass flux with time. This was because a positive fluctuation in the number flux led to an increase in the number of entering molecules that increased the entry pressure and led to a further increase in the number flux. The system was therefore unstable and steady flow was attained in the high pressure ratio case only because the flow was choked with a region of supersonic flow downstream of the channel. The program was modified such that the molecule entry flux was reduced in direct proportion to the increase in pressure near the entry to the upstream reservoir. However, this only slowed the increase in mass flux and, while this would eventually be reversed, the magnitude of the oscillations would be unacceptable. The solution was to enforce steady flow by keeping the total number of simulated molecules fixed to the initial value. This was achieved by setting the number of molecules that entered the downstream reservoir to the number leaving that boundary less the net number change at the upstream boundary. Also, because there had been little change in the molecule number in the earlier calculation and the average entry speed was about 25 m/s, the data was altered such that the velocity component in the x direction of the main stream was 25 m/s rather than zero.8.4 Channel flow with specified end pressures 233 ‘S.e+22 Le+23 1sSe+28 Deed 250123 ‘Number density cubic m) Fig. 8.22 The number density distribution with version 4.5.10. ‘The new boundary conditions lead to little qualitative change but, as shown in Fig. 8.22, the number density in the upstream reservoir is slightly lower and the number flux falls to 5.31x10! molecules per second. The Mach number contours in Fig. 8.23 show the choked flow with largely sonic flow at the exit of the channel. The high speed flow from the tube is unable to cope with the rapid expansion and the flow in the downstream reservoir is dominated by a large vortex. The Knudsen number based on the upstream mean free path and the full height of the channel is 0.0125. This is near continuum, but the boundary layer has a significant effect and the number flux is less than half the inviscid continuum number flux of 1.22102? molecules per second. The degree of influence of the reservoir shape on the number flux would be an interesting study. Fig, 8.23 The Mach number distribution.234 TWO-DIMENSIONAL FLOW APPLICATIONS ao 280. 300. 220. ‘Transiational temperature (K) Fig. 8.24 The temperature variation over the flowfield. 200. 400. ‘600. ke 44,000 Scalar pressure nkTer (N/square m) Fig. 8.25 The scalar pressure contours. Figures 8.24 and 8.25 show the temperature and scalar pressure contours, The latter quantity involves the product of the number density and temperature and it is notable that the pressure contours within the channel are normal to the flow. The large temperature variations are characteristic of flows in which the velocities are at least of the order of the speed of sound. The statistical fluctuations in the velocities are relative to the average molecular speed and this speed differs from the speed of sound by a numerical factor near unity. DSMC is therefore more suited to large disturbance flows and it is frequently claimed that DSMC is not capable of dealing with typical MEMS applications that involve flow speeds that are small in comparison with the speed of sound. To test these claims, the calculation was repeated with the overall pressure and number density difference reduced by a factor of ten. That is a downstream number density of 2.31023 /m?,8.4 Channel flow with specified end pressures 235, 230222356123 24e+2i ‘Number density (euble m) Fig, 8.26 The number density contours. ‘The number density is shown in Fig. 8.26 and the speed distribution over the flowfield in Fig. 8.27. The number flux was 1.27*1021 molecules per second and, at the mid-point of the channel where the number density is 2.39*1025 /m®, the average speed is 26.6 m/s. This was established to reasonable accuracy through a short preliminary calculation and, because the area of the upstream reservoir entry is five times that of the channel, the stream was assigned a velocity component of 5 m/s in the x direction. ‘The results are based on a run that lasted two days on a single processor. There is therefore no problem in the application of the DSMC method to problems with flow speeds of the order of 10 m/s. Problems with velocities less than 1 m/s require a sample of the order of 108, It is shown in §8.9 that this can now be attained, even on a personal computer, and claims that DSMC cannot be applied to MEMS type problems are now outdated. a a a Flow speed (m/s) Fig, 8.27 The flow speed contours.