R. Byron Bird, Warren E. Stewart, Edwin N. Lightfoot-SOLUTION - Transport Phenomena, 2nd Edition-Wiley (2001)

Fendmenos WMT MC) Sune TIN OC ale] pees eH Edwin N. LightfootSOLUTIONS TO THE PROBLEMS in TRANSPORT PHEOMENA Second Edition (2002) by R. Byron Bird Warren E. Stewart Edwin N. Lightfoot Department of Chemical Engineering University of Wisconsin Madison, Wisconsin 53706 USA This solutions manual has been prepared by the authors of the textbook for use by professors teaching courses in transport phenomena. It contains the solutions to 500 of the unsolved problems in the textbook. No part of this material may be reproduced in any form, electronic, mechanical, photocopying, recording, scanning, or otherwise. JOHN WILEY & SONS, Inc. New York, New York1A.1 Estimation of dense-gas viscosity. a. Table E.1 gives T, = 126.2 K, p, = 33.5 atm, and yp. = 180 x 10-® g/cm-s for Nj. The reduced conditions for the viscosity estimation are then: Pr = p/pe = (1000 + 14.7)/33.5 x 14.7 = 2.06 T, = T/T. = (273.15 + (68 — 32)/1.8)/126.2 = 2.32 At this reduced state, Fig. 1.3-1 gives u, = 1.15. Hence, the predicted viscosity is p= p/p, = 1.15x180x10-* = 2.07x 10-4 g/cm-s. This result is then converted into the requested units by use of Table F.3-4: pe = 2.07 x 107* x 6.7197 x 107? = 1.4 x 107* Ib, /ft-s1A.2 Estimation of the viscosity of methyl fluoride. a. CH3F has M = 16.04—1.008+ 19.00 = 34.03 g/g-mole, T. = 4.55-+273.15 = 277.70 K, p, = 58.0 atm, and V, = 34.03/0.300 = 113.4 cm*/g-mole. The critical viscosity is then estimated as fe = 61.6(34.03 x 277.70)'/?(113.4)-?/3 = 255.6 micropoise from Eq. 1.3-la, and He = 7.70(34.03)/?(58.0)?/3(277.7)1/° = 263.5 micropoise from Eq. 1.3-1b. The reduced conditions for the viscosity estimate are T, = (370 + 273.15)/277.70 = 2.32, pr = 120/58.0 = 2.07, and the predicted y, from Fig. 1.3-1 is 1.1. The resulting predicted viscosity is = pyfe =1.1 x 255.6 x 107° = 2.8 x 1074 g/em-s via Eq.1.3-la, or 1.1 x 263.5 x 1076 = 2.9 x 1074g/cm-s via Eq.1.3-1b.1A.3 Computation of the viscosities of gases at low density. Equation 1.4-14, with molecular parameters from Table E.1 and collision integrals from Table E.2, gives the following results: For Oo: M = 32.00, o = 3.433 A, é/K = 113 K. Then at 20°C, xT/e = 293.15/113 = 2.594 and Q,, = 1.086. Equation 1.4-14 then gives ¥32.00 x 293.15 (3.433)? x 1.086 = 2.02 x 1074 g/cm-s = 2.02 x 107° Pa-s = 2.02 x 107? mPa:s. p= 2.6693 x 1075 The reported value in Table 1.1-3 is 2.04 x 10-? mPa:s. For No: M = 28.01, o = 3.667A, e/K = 99.8 K. Then at 20°C, xT/e = 293.15/99.8 = 2.937 and Q, = 1.0447. Equation 1.4-14 then gives V28.01 x 293.15 (3.6672 x 1.0447 = 1.72 x 107* g/cm-s = 1.72 x 107° Pas = 1.72 x 107? mPa:s. fe = 2.6693 x 1075 The reported value in Table 1.1-3 is 1.75 x 10-? mPa:s. For CHy, M = 16.04, o = 3.780A, e/K = 154 K. Then at 20°C, x«T/e = 293.15/154 = 1.904 and ©, = 1.197. Equation 1.4-14 then gives V16.04 x 293.15 (3.780)? x 1.197 = 1.07 x 1074 g/cm-s = 1.07 x 107° Pa-s = 1.07 x 107? mPa-s. b= 2.6693 x 107° The reported value in Table 1.1-3 is 1.09 x 107? mPa-s.1A.4 Gas-mixture viscosities at low density. The data for this problem are as follows: Component M Lt, poise x 10° 1(H2) 2.016 88.4 2(CClF 2) 120.92 124.0 Insertion of these data into Eq. 1.4-16 gives the foloowing coefficients for mixtures of H2 and Freon-12 at this temperature: G1, = 82 = 1.0 2 Sy a (14 2 Ty ( 884)" ae a VB 120.92 124.0 2.016 = 3.934 2 1 120.92\ 1/2 124.0\/? / 2.016 \*/4 $o = (14+ ——— 14+(—— ——— VB 2.016 88.4 120.92 = 0.0920 Equation 1.4-15 then gives the predicted mixture viscosities: a= DiS ye = A:= B:= A+B= 1-22 MxpGig YireG2— z1p/ oy e2Kl2/ 2 pmix X 10° pobs,poise X 10° 0.00 3.934 1.000 0.0 124.0 (124.0) 124.0 0.25 3.200 0.773 6.9 120.3 127.2 128.1 0.50 2.467 0.546 18.1 113.6 131.7 131.9 0.75 1.734 0.319 38.2 97.2 135.4 135.1 1.00 1.000 0.092 88.4 0.0 (88.4) 88.41A.5 Viscosities of chlorine-air mixtures at low density. Equation 1.4-14 and Tables E.1, E.2 give the following viscosities at 75°F(= 273.15 + (75 — 32)/1.8 = 297.03 K) and 1 atm: For component 1, (Clz), Mi = 70.91, 01 = A1IBA, €1/K = 357K; hence, KT /e, = 297.03/357 = 0.832 and Q,,1 = 1.754, and V¥70.91 x 297.03 = 2.6693 x 1075 ~——_ i Xi" (@T1B)? x 1.754 = 1.304 x 107* g/cm-s = 0.01304 cp. g For component 2, (air), Mz = 28.97, ¢ = 3.617A, €1/K = 97.0K; hence, kT /e, = 297.03/97.0 = 3.062 and 2,1 = 1.033, and 28.97 x 297.03 = 2.6693 x 107 i “i (3.617)? x 1.033 = 1.832 x 10~* g/cm-s = 0.01832 cp. Eq. 1.4-16 then gives the following coefficients for Eq. 1.4-15 at this temperature: Py, = Bo =1.0 2 by - 24 2 WPT, (0.01304) 1? (28.97) 1/* vB 28.97 0.01832 70.91 = 0.5339 2 boy = (14 28% “MOT, (0.01882)? (70.91 ua VB 70.91 0.01304 28.97 = 1.8360 Equation 1.4-15 then gives the predicted mixture viscosities: m= y= 2 = A:= B= A+B= 1-22 Veep Gig YLxphog tipn/ dl, t2p2/ doy Mmix,cp. x 10° 0.00 0.5339 1.000 0.0 0.01832 0.0183 0.25 0.6504 1.2090 0.005012 0.011365 0.0164 0.50 0.7670 1.4180 0.008501 0.006460 0.0150 0.75 0.8835 1.6270 0.011070 0.002815 0.0139 1.00 1.000 1.8360 0.01304 0.0 0.01301A.6 Estimation of liquid viscosity. a. The calculated values for Eq. 1.5-9 at 0°C and 100°C are as follows: TK 273.15 p, g/em? 0.9998 V =M/p, cm3/g-mole 18.01 AU vap.ty, cal/g-mole= 897.5 x 18.016 x 252.16/453.59 8989. AU vap,7,/RT = 8989/1.98721/T 16.560 exp 0.408AU vap,7,/RT 859.6 Nh/V, g/cm-s 2.22 x 10~* Predicted liquid viscosity, g/cm-s 0.19 b. The predicted values for Eq. 1.5-11 at 0°C and 100°C are: _ TK 273.15 Nh/V, g/cm-s 2.22 x 10-* exp(3.87}/T) 179.7 Predicted liquid viscosity, g/cm-s 0.0398 Summary of results: Temperature, °C 0 Observed viscosity, centipoise[=]g/cm-s x100 1.787 Prediction of Eq. 1.5-9 19. Prediction of Eq. 1.5-11 3.98 373.15 0.9584 18.80 8989. 12.120 140.5 2.12 x 1074 0.0298 373.15 2.12 x 1074 44.70 0.0095 100 0.2821 2.98 0.95 Both equations give poor predictions. This is not surprising, since the empirical formulas in Eqs. 1.5-8 et seg. are inaccurate for water and for other associated liquids.1A.7 Molecular velocity and mean free path. From eq. 1.4-1, the mean molecular velocity in O2 at 273.2 K is _ 8RT 8 x 8.31451 x 107 x 273.2 4 v= Vow= ¥— _3xsn007 = 4.25 x 10* cm/s From eq. 1.4-3, the mean free path in O2 at 1 atm and 273.2 K is RT 82.0578 x 273.2 = = = 9.3 x 107§ cm Vand2pN V2n(3 x 10-8)? x 1 x 6.02214 x 1023 Hence, the ratio of the mean free path to the molecular diameter is (9.3 x 107*/3 x 10-8) = 3.1 x 10* under these conditions. At liquid states, on the other hand, the corresponding ratio would be on the order of unity or even less.1B.1 Velocity profiles and stress components A. Ty = Ty, =—pb, and all other 7, are zero. pv, 0, = pb’y*, and all other pv;0; are Zero. b. Ty = Ty, = —2ub, and all other 7, are zero. (0,0, = pb’y’, p0,0, = pv,v, = pb’ xy, pu,v, = pb’x", and all other pv;v; are zero. c. All t;, are zero (0,0, = by’, pU,0, = pv,v, =-pb’xy, pu,v, = pb?x’ and all other pv;v; are zero. . Ty = Ty = Ub, T,, = —2ub, and all others are zero. the components of pvv may be given in the matrix: (0,0, ={pb'x? pov, =4zpbxy pv, v, =—4pb°xz pyv=| pv,v,=ypb’xy puyv,=fpby* — pu,v, =-4pb*yz p0,0,=—Zpb°xz pv,v,=—tpb’yz — pv,v, = pb*2’1B.2 A fluid in a state of rigid rotation a. A particle within a rigid body rotating with an angular velocity vector w has a velocity given by v=[w xr]. If the angular velocity vector is in the +z-direction, then there are two nonzero velocity components given by v, =—-w,y and v, =+w,x. Hence the magnitude of the angular velocity vector is b in Problem 1B.1(c). b. For the velocity components of Problem 1B.1(c), Ov ov oy Ws 9 and “ys - 94 ox Oy ox oy c. In Eq. 1.2-4, we selected only the linear symmetric combinations of derivatives of the velocity, so that in pure rotation there would be no viscous forces present. In (b) we see that the antisymmetric combination is nonzero in a purely rotational motion.1B.3 Viscosity of suspensions Expanding the Mooney expression, we get (with ¢ = 6/9,) 2 3 Here =14+ 3¢ + 1 39 + 1 3o sees Mo l-e) 2!\1-e 3!\1-é aa ae ee ome B (14-4 145 S949? 25 5.1), 4:(125, 25151), B 2%, 48 4 208 The first two terms match exactly with the first two terms in Eq. 1B.3-1. We can make the third term match exactly, by setting erred 17 whence ¢, = 0.618 and the coefficient of ¢° becomes 125 25 1 1 42 a 48 4 0.618 20.382 — If we try ¢ = 0.70, the coefficients of $* and ¢° become 6.70 and 17.6 respectively. This gives a somewhat better find of Vand's data. {-|01C.1 Some consequences of the Maxwell-Boltzmann equation a. The mean speed is : ae 2p-mu PT [re tee om in b. First rewrite Eq. 1C.1-4 as co ng 2 co | my? /2KT co [ue rus PXT gy fie mus /' du, fie rus PAXT gy oo Sa poo fem PT dy, fre mT ay fie maz [OKT gy x The integral over u, in the numerator of the first factor is zero because the integrand is the product of a factor "u," (an odd function of the integration variable about u, = 0) and an exponential function (an even function), and the range of integration extends equally far in the positive and negative directions. c. The mean kinetic energy per molecule is ee —mu? °° 7 fo ute m™ PRT gy =m 2KE J, o’e Pde f ure Me KT gy m fF eee dé Nie 3 = m Nie Nie erie joojus he 3xT and is thus 4KT for each degree of freedom. {-\\1C.2 The wall collision frequency When we change to dimensionless variables in the second line of Eq. 1C.2-1, we get Zan sez) plage ashe (be eae) ar (bees) =e) (AE. 2 Ea PE v\=n ca |-121C.3 The pressure in an ideal gas a. The dimensions of the quantities in Eq. 1C.3-1 are S [=] VP At [=] ¢ m [=] M f [=] aby)” du,du,du, [=] (L/t)” Using these units, one finds that the expression on the right of Eq. 1C.3-1 has units of M/ Lt? (which are the same as the units of force per area). b. Combining Eqs. 1C.1-1 and 1C.3-1 we get p= 2nn{ xa) EI 2em (us rueud AKT du,du, 20KT 3/2 = 2nn{ =) uze™ rau [PT gy fre mS aay fre mt PT hy 1K oo ooh m( 52) 2 ny" ff ee Pde" ne an. [7 Cee dc = 2anf 2822)" (“#)(Vz)(Vz) = nkT m \-131D.1 Uniform rotation of a fluid a. For the special case that w =5,w, we get v=[Wxr]=2,2 2, 6).5 0x, = W(8, E104 + 8, €3,X) = w(—8,y + 8x) Then using Eqs. A.6-1, 2, 13 and 14, we can get the velocity components in cylindrical coordinates v, =(v-,)= w((—8.y +3yx):3,)= w(-ycos @ + xsin 6) = w(-rsin Ocos 6 + xcos Osin 0) =0 V9 =(v-5_)= w((—3,y +8,x)-39)= w((-y)(-sin @) + xcos 6) = w(rsin @sin 6 + rcos @cos 0) = wr Therefore, the angular velocity of every point in the fluid is v,/r=w, which is a constant, and there is no radial velocity. This is the way a rigid body rotates at constant angular velocity. b. The vector operations are (using the abbreviated notation of §A.9 and the Einstein summation convention) (V-V) = 0,0; = 0; pn m¥n = Ein min = Enmnlm = 0 1~imn™™ mn imn™” m™ in nmn~ m On {Vv}, ={V[w xr]} = OE jn @m%n = mn @mOin= € ij i© jmn™™ mn jn m™ in © jmi = {VW} 5 a {Vv}; and from this last result we see that Vv + (Vv)' =0. c. The results above indicate that for a fluid is a state of pure rotation, the tensor t is identically zero. That is, there are no viscous stresses present in the fluid. This was the assertion made just before Eq. 1.2-4. |- [41D.2 Force ona surface of arbitrary orientation. a. We can specify the surface area and the orientation of the surface of AOBC as ndS. To project this surface onto the yz-plane, we take the dot product with 8,, so that the area of AOBC is (n-8, )dS. b. The force per unit area on three triangles perpendicular to the three coordinate axes are Force on AOBC =6,. Ay, +8 yyy +8, yz Force on AOCA =8,2,, +8, +5, Ay, Force on AOAB=56,7,, +67, +87 c. Force balance on the volume OABC is then m,dS = (8,7. +5 yy +8, %,,\(n-8,)dS +(8, yx +8, Zyy, +8,7,,)(n-3,)d5 +(8, 7, +8, +5,%,,)(n-3,)d or TN, = [n-8,8,7,,]+|n-5,5,7,,]+[n-8,5,7,.] +[n-5,5,7,,|+[n-8,5,2, |+[n-3,3,7,. | +[n-5,5,7,,]+[n-3,5,7,,|+[n-3,5,7,] = EE [n-85)%y]=[0-2 j-152A.1 Thickness of a falling film. a. The volume flow rate w/p per unit wall width W is obtained from Eq. 2.2-25: w _ vRe_ (1.0037 x 107)(10) pw 4 4 Here the kinematic viscosity v for liquid water at 20°C was obtained from Table 1.1- 2. Since 1 ft=12x2.54 cm, 1 hr =3600s, and 1 gal=231.00 in® x (2.54cm/in)?=3785.4 cm® (see Appendix F), the result in the requested units is _ = 2.509 x 107? cm?/s eo 2 a 2 OW 0.02509 cm?/s x 37ahA gal/cm? x 30.48 cm/ft x 3600 s/hr = 0.727 U.S. gal/hr-ft b. The film thickness is calculated from Eqs. 2.2-25 and 2.2-22 as 5- By w 1/3 ~ \gcos 8 pW _ 3v_vRe 1/3 ~\qcosB 4 _ (: x 1.0037 x 107? —2 (980.605)(1.0) (7-209 x 10°") = 0.00361 in. 1/3 ) = 0.009167 cm2A.2 Determination of.capillary radius by flow measurement. Assuming the flow to be laminar, we solve Eq. 2.3-21 for the capillary radius: R=? 8ulLw _ 4/8vLw ~ \ViarpaP” V cAP Insertion of the data in mks units gives ,/8(4.03 x 10-5)(0.5002)(2.997 x 10-3) R (3.1416)(4.829 x 105) = V3.186 x 10-38 = 7.51 x 1074 m =7.51 x 107? cm As a check on this result, we calculate the corresponding Reynolds number: Re= D(vz)p _ 4w _ 2w Be tDu = =aRvp 2 (2.997 x 1073) ~ (7.51 x 10-4)(4.03 x 10-5)(0.9552 x 10%) 66.0 This value supports our assumption of laminar flow. Since the entrance length, L,. = 0.035DRe = 0.35 cm is less than L, the entrance-effect correction to R is at most of the order of |[1 —(Z-/L)]'/4 — 1], or 0.2 percent of R in the present example. Difficulties with this method include: (1) Inability to account for departures from a straight, circular cylindrical wall geometry. (2) Inability to account for in- advertent spatial and temporal variations of temperature, hence of the fluid density and viscosity. A simpler method is to measure the length L and mass m of a small slug of liquid mercury (or another liquid of known density) injected into the tube, and calculate the mean radius R of the slug as (m/[p7L})'/, on the assumption that the slug is a right circular cylinder. This method allows comparisons of mean R values for various intervals of the tube length.2A.3 Volume rate of flow thrugh an annulus. Assuming the flow to be laminar, we use Eq. 2.4-17 to calculate the volume flow rate w/p, with the specifications « = 0.495/1.1 = 0.45 pt = 136.8 (Ib /ft-hr)(1 hr/3600s) = 3.80 x 107? lbm/ft-s (Po — Px) = (5.39 psi)(4.6330 x 10° poundals/ft?/psi) = 2:497 x 10* Ibm/ft- --s? R=1.1 in. =1.1/12 ft Here Appendix F has been used for the conversions of units. With these specifications, Eq. 2.4-17 gives w — (m)(2.497 x 104)(1.1/12)4 a (1=(0-45)2)? p (83.80 x 10-2)(27) [a (0.45)") ey] = (0.49242 [1 —o.oaro1) — 0 92088") = (0.6748)[0.1625] = 0.110 ft?/s As a check on our assumption of laminar flow, we calculate the Reynolds number: _ 2R(1—K)(vz)p _ 2w HB ~ tRy(1 + 6) 2(0.110)(80.3) i110 ~~ (3.1416)(1.1/12)(3.80 x 10-2)(1.45) This value is well within the laminar range, so our assumption of laminar flow is confirmed. Re2A.4 Loss of catalyst particles in stack gas. a. Rearrangement of Eq. 2.6-17 gives the terminal velocity v1 = D*(ps — p)g/18u in which D is the sphere diameter. Particles settling at v; greater than the centerline gas velocity will not go up the stack. Hence, the value of D that corresponds to vt = 1.0 ft/s will be the maximum diameter of particles that can be lost in the stack gas of the present system. Conversions of data to cgs units give ve = (1 ft/s)(12 x 2.54 cm/ft) = 30.48 cm/s p = (0.045 Ibm /ft?)(453.59 g/Ibm)((12 x 2.54)~* ft? /em>) = 7.2 x 107* g/cm? Hence, D 1804 i (18)(0.00026)(30.48) (ps—p)g \ (1.2 —7.2 x 10-4)(980.7) = 1.21 x 10-4 = 1.1 x 107? cm = 110 microns b. Equation 2.6-17 was derived for Re<< 1, but holds approximately up to Re=1. For the system at hand, Dup _ (1.1 x 107?)(30.5)(7.2 x 1074) ar (0.00026) 0.93 Hence, the result in a. is approximately correct. Methods are given in Chapter 6 for solving problems of this type without the creeping-flow assumption. 2-42B.1 Different choice of coordinates for the falling film problem Set up a momentum balance as before, and obtain the differential equation fa = pgcosp Since no momentum is transferred at ¥=6, then at that plane Tz,=0. This boundary condition enables us to find that C, = -pgdcosB, and the momentum flux distribution is Tz, = -pgdcosf{ 1 = *) Note that the momentum flux is in the negative ¥-direction. Insertion of Newton's law of viscosity t;, = -p(do, /dx ) into the foregoing equation gives the differential equation for the velocity distribution: dv, _( pgdcosB (1-3) dx LM 6 This first-order differential equation can be integrated to give op ={ exezsenb (1/2) ; mn 5 2\6 The constant C, is zero, because v, =0 at ¥=0. We note that ¥ and x are related by x/5 = 1-(x/5). When this is substituted into the velocity distribution above, we get oo 8|(-9) 4-273] which can be rearranged to give Eq. 2.2-18. 2-52B.2 Alternate procedure for solving flow problems Substituting Eq. 2.2-14 into Eq. 2.2-10 gives d dv,\ | d’v, __ pgcosB = a 7) =egcosp oF ae be Integrate twice with respect to x (see Eq. C.1-10) and get __ pgcosB = mn +C,x+C, Then use the no-slip boundary condition that v, =0 at x =6, and the zero momentum flux boundary condition that dv,/dx=0 at x=0. The second gives C,=0, and the first gives C, =(pgcosB/2m)6?. Substitution of these constants into the general solution and rearranging then gives Eq. 2.2-18.2B.3 Laminar flow in a narrow slit a. The momentum balance leads to P,-P Py- a, _(Po-Pi) 0-1) and 7, = Po= i) a Prac, dx L L Substitution of Newton's law 1,, =—-y o into the above gives a P,- _ 2 do, __(Po- Pi) or p, = -(Po= PL)" Gi. dx UL Lu 2uL Use of the no-slip boundary conditions at x =+B gives the expressions in Eq. 2B.3-1 and 2. One can also see that C, = 0 directly, since we know that the velocity distribution must be symmetrical about the plane x = 0. b. The maximum velocity is at the middle of the slit and is _ (Py = P, )B z,max — 2uL v0 2 and hence vee 1 -(3) UO, max The ratio of the average to the maximum velocity is then (v,) _ Jo Fo[t-(2/8) arty _ fae? )ag = =(1 1\_2 Ds max [" Ejaxdy fae (1-3)=3 c. The mass rate of flow is (P)-P,)B*_ 2(P)-P, )oBW w = p(2BW)(v, ) = p(2BW)(3) > uL BL w d. In Eq. 2.5-22, set both viscosities equal to . set b equal to B, and multiply by BWp.2B.4 Laminar slit flow with a moving wall ("plane Couette flow") Start with the velocity distribution from part (a) of Problem 2B.3 (in terms of the integration constants). Determine C, and C, from the boundary conditions that v,=0 at x=—B, and v,=V at x = B. This leads to ae eR, (3) Mu This expression can be differentiated with respect to x and then Newton's law of viscosity 7,, = -p(do, /dx) can be used to get the expression for the stress tensor. Notice that the velocity distribution is no longer symmetric about the midplane, so that C, #0. 2-82B.5 Interrelation of slit and annulus formulas From Eq. 2.4-17 we get Bul _ a, (1-G-2) of n(P, o ma (1 (1-¢) )+ In(1-) (1 —1+2¢-¢&? eae eee e-se*-te-—-le 2 =(1-14+4¢-66? +46? - 4) + (4¢? —4e3 + é*) ettertiertlety.. Ul (4¢-6e? + 46° -e*)- = (4¢- 68? +46° - e*)—(4e-6¢? +869 ~tet+.-) This gives, finally, a result in agreement with Eq. 2B.5-1 n(P,-P,)R*p w= (bebe)2B.6 Flow of a film on the outside of a circular tube a. A momentum balance on the film gives Art) + pgr =0 or pe (re ss r dr | 8 Ma ar J 7 Par =° The latter may be integrated to give 0, =- par? 4u Next use the boundary conditions that at r=R, v, =O (no slip) and that at r=aR, do,/dr =0. When the integration constants have been found, we get for the velocity distribution 2 2 v, — PER 1-(4} + 2a? Ina 4 R R b. The mass rate of flow in the film is then 27 paR a w=)", ev.tdrd = 2aR? pf" v, dé in which a dimensionless radial coordinate €=r/R has been introduced. Then = _ fi(1—€2? +20? me) eae - ue AE SS (hE? - 464 + 20"[-42? + +467Iné))[ IE ag — 3a‘ + 4a’ Ina) 8u c. If we set a=1+ € (where € is small) and expand in powers of € using §C.2, we get 2-l02 4 2 4.3 wa gR (6° +0(e*))= 2p gR°e 8u 3u This is in agreement with Eq. 2.2-21 if we make the identifications W =2aR and 6=eR (and furthermore consider only the case that cosB =1. 2-1]2B.7 Annular flow with inner cylinder moving axially a. The momentum balance is the same as that in Eq. 2.3-11 or Eq. 2.4-2, but with the pressure-difference term omitted. We can substitute Newton's law of viscosity into this equation to get a ence v, =-Ainr+C, or ? =-D,n£+D, dr fr L U R That is, we select new integration constants, so that they are dimensionless. These integration constants are determined from the no-slip conditions at the cylindrical surfaces: v,(KR)=v, and v,(R)=0. The constants of integration are D, =0 and D, =-1/Ink. This leads then directly to the result given in the book. b. The mass rate of flow is 2m pR UR? 1 w= ho *[,pv.rdrdo = 270 = . Jin 6)6dé R2 1 R?2 = 2mpoo—_ (36? Iné- 46?)| = 2mp-o— (-4%? In«-}(1- x?)) which is equivalent to the answer in the text. c. The force on a length L of the rod fn a rotten) which gives the expression in the book. d. When we replace k by 1— € and expand in a Taylor series, we get (/xk) KRdOdz = 22KRLUv, Ink T=KR 1 _ 2nLp)v9 P=2aL(-p)v) ——_-—_.——— Hoe rte ade) E (1-4e-pe?--) To get this last result one has to do a long division involving the polynomial in the next-to-last step.2B.8 Analysis of a capillary flowmeter Designate the water by fluid "I" and the carbon tetrachloride by "II". Label the distance from B to C as "J". The mass rate of flow in the tube section "AB" is given by a(P 4 —P,)R*p, _ al(P. — Pz) + prgh]R‘p, 8uL 8uL Since the fluid in the manometer is not moving, the pressures at D and E must be equal; hence Pa t Pgh + Pg] + PSH = Pp + O18) + PuSH from which we get Pa ~ Pp t+ PiSh = (Py - 0; ) SH Insertion of this into the first equation above gives the expression for the mass rate of flow in terms of the difference in the densities of the two fluids, the acceleration of gravity, and the height H. 2-\32B.9 Low-density phenomena in compressible tube flow When we replace no-slip boundary condition of Eq. 2.3-17 by Eq. 2B.9-1, we get C= (Po —p,)R? rn (Po —pr)RE 2 4uL 2uL so that the velocity distribution in the tube is 2, = (Po - Pr )R? -[ x) | =P )RE 4uL R UL Next we write the expression for w, but consider only the flow through a length dz of the tube: w= i "t p(z)v,(r,z)rdrdé = 2an'{E Elbo Ede RT where we have introduced the ideal gas law, with R, being the gas constant (we use a subscript g here to distinguish the gas constant from the tube radius). We have also introduced a dimensionless radial coordinate. When we introduce the velocity distribution above, we get _2mR*( M )af__ dp 2¢ a (Ie p at £)4 leas _aR*| M_\(_, a), 40 ~ 8u (#4) p+ is) This is now integrated over the length of the tube, keeping mind that the mass flow rate w is constant over the entire length L _ _ AR* M at “t) Jywdz = By (2) | P+ dp 2-14_ OR! M_\( po-pi , 460 ~ Su (24 2 + R (Po PL) This gives a HIE Meese A) R 8uL (RTI 2 _ 2(Po~Pr)R*/ PavgM |, 460 8uL R,T Rp avg which leads then to Eq. 2B.9-2. 2-152B.10 Incompressible flow in a slightly tapered tube a. The radius at any downstream distance is R(z) = Ry +(R, — Ry)(2/L) b. Changing the independent variable proceeds as follows: _ aR‘p o(- *\(2)- eel *\ R, a 8u dR )\ dz 8u dR L c. First we rearrange the equation in (b) to get _dP _( 8uw L 1 “dR \ mp J\R,—-R, J R* Then we integrate this equation to get PL 8uw L R, 1 -|_ dP =| —— —dR Jr ( mp le - alle R* whence we can get the pressure difference in terms of the mass rate of flow p.—p. _( Buel Ro -R,° ° EV 3ap JL Ry -R, Next we solve to get the mass flow rate ar a) a ea This is the result, with the first factor being the solution for a straight tube, the second factor being a correction factor. It would be better to write the correction factor as "1— X ", so that the quantity X gives the deviation from straight-tube behavior. The quantity X is then 2-16vee 3 Roe Re SAI (RR) _ ft =(Ri/Ro)MRi/Ro) "RE R2-RS = (Ry/R,) 1 1-(R,/Ry)° _ 3(R,/Ro) __1+#(R,/Ry)+(R,/Ro) —3(R,/Ro)” ~~ 14(R,/Ry)+(Ri/Ro) 1+(R,/Ry)+(Rz/Ro)” which then leads to the desired result in Eq. 2B.10-3. 2-1]2B.11 The cone-and-plate viscometer a. In a parallel-plate system with rectilinear flow, the velocity distribution is just v, /v) = y/b, where b is the plate spacing and v, is the velocity of the upper plate. We now make the following correspondences between the parallel-plate system and the cone- plate system (using @ as the usual variable in spherical coordinates measured downward from the z-axis, and yw as the variable measured upward from the plate surface): 0, Ug; Uy PQr;, bLerrsinyy; yorsiny =ry =r(}2-6) When this correspondence is made, Eq. 2B.11-1 results. b. From Eq. B.1-19, we get for the force per unit area in the ¢-direction on a face perpendicular to the 0-direction sin 2 Y% }- 19 120 jauece) q —— 09 Bal sind r de rao Wo Here we have used the fact that the angle between the cone and plate is so tiny that @ is very nearly 47 so that sin @ is very close to unity. c. The torque is obtained by integrating the force times lever arm over the entire plate area: . Q Q)( R® T, = { incre , A ~ 2af Ha -2n( 2 Wo which leads to Eq. 2B.11-3. 2-182B.12 Flow of a fluid in a network of tubes At A the pipe splits into three pipes, and at the next set of junctions the fluid flows equally in six pipes, and then at the next set of junctions the fluid flows back into three pipes, and finally at B the fluid is all returned to a single pipe. Call the modified pressure at the junctions where three pipes split into six pipes P,_,,, and that where six pipes join to form three pipes P,_,,. Then in each of the first set of three pipes w_ n(P,-P,,.)R*p 8uLw Bae 36 AE De 3 8uL OF WAT 8396 = 3 aR Ap In each of the batch of six pipes w nPs,,-P R‘p 8uLw Boe or Pn Pon gee and in each of the final batch of 3 pipes w_ 2(P,,,-P,)R‘p _ 8uLw = SS or Po3 P= a 3 8uL 32R"p When all the pressure differences are added together, the unknown quantities P, ,, and P, _,, cancel out, and we get 8yuLw (5 3n(P, -—P,)R‘p P,-P,= (2) or — 32(P,-P5)R’p a 2142C.1 Performance of an electric dust collector a. First we solve the problem of the vertical motion of the particle as it falls under the action of the electromagnetic field. The equation of motion for the particle (without gravitational acceleration or Stokes drag) is d’x e& = me This equation may be integrated with the initial conditions that x= x, and dx/dt=0 at t=0, to give _ et? 2m X=Xg From this we can get the time ft, required for the particle to fall to x=-B: t= [2m(B +X) f e& Next we look at the horizontal motion. From Eq. 2B.3-2 and the expression for x(t), we find that (with v,, =(p) — p,)B?/2uL dz _ x) | dz_ Xq — (e6#?/2m) ; a =04{1-(4) p-s,|:-(2=esten) This may be integrated to give 2 2ge244 L= fief | 1 ge[ 23-208! +2 oe Ja] m 4m? 3 20245 _ (Po - Pt) (B? ~ x2) 4, toes me Sits QuL oVF" 3m 20m? 9-20Next, square this expression and then insert the expression for t, above to get 2 _ 2(Po—Pr) |2m(B+ Xo) _ pe 5a ,| é (3B-2x))(B+x) Then, in order to remove the radical, we square this, thereby getting _ 2 Lis AG o— PEI (BB 2x0) (B+ x0)° Next, we set dL*/dx, equal to zero, and this yields 4 values for x): 3B, 3B,B, and B. It is only the first of these that is physically acceptable. When that value is put back into the expression for L, we get finally 12 (py — p,)2mB> |" Lain =| = e 35 wes 9-22C.2 Residence-time distribution in tube flow a. A fluid element at a radius r within the tube will require a time t to reach the tube outlet ee L ao. v,(r) Urnax| 1 (r/R) | All the fluid with a radius less than r will have left the tube at this time. Hence the fraction of the flow that will have left the tube is v= [,"[v.rardo _ {zy (2) —ffFo.rdrda “\ RJ \R When the first equation is solved for (r/ R) and substituted into the expression for F(t), we get P= 1- ad (4 4)- (4) - -(zéx) b, The mean residence time is then obtained by solving the last equation in (a) for t. and substituting into the Eq. 2C.2-1: Onax OV1-F Umax (2, ) ty = [ltd = L f dF 2L L Q-LL2C.3 Velocity distribution in a tube The derivation in §2.3 is valid up through Eq. 2.3-15. If the viscosity is dependent on the radial coordinate, however, Eq. 2.3-16 is inappropriate. Instead we get —_(Po-Py) pr F 0, =~“ ay? +O Application of the no-slip boundary condition at the tube wall gives (Fo -P,) 2L 0=- [nar +, (Fr) This may be solved for the integration constant, and the velocity distribution is then », =! Poet or v,= This is the same as Eq. 2.3-18 if the viscosity is a constant. Next we get an expression for the average velocity 2% ¢R ; d do R : (0,)= te — wh o,rdr = 2{0,ydy _ (PoP, )R® ay Por PUR AT yaya | 7 ore Bo yiudy = as a Joacgy lo way _(Po-P R* a m ”), a: ee d v, Sle ey Then we find the dimensionless ratio () | Ree 2-232C.4 Falling-cylinder viscometer a. Equation 2.4-2 is valid for this problem, but the pressure difference is not known. When Newton's law of viscosity is substituted into Eq. 2.4-2 we get P,-P v, -| tal LP -cInr +e, The two constants of integration and the (unknown) pressure difference can be obtained from two boundary conditions and a mass-conservation condition: At r=KR, v, =—v9; at r=R, v, =0; and j 7 "Te v,rdrd0 = n(KR)* v9. This states that the fluid displaced by the falling cylinder must be compensated for by a net motion upward through the annular slit. These three unknown constants may be obtained from these conditions (lengthy!) and the result is v, __(1-6?)-(1+«?)n(1/é) v ~ (1- K?)—(14 x? )In(1/K) b. The force acting downward on the cylindrical slug of height H is (p) — p)g¢-2(«R)’ H. The difference in the pressures acting on the top and bottom of the slug is an upward force (P) —P,,)-2(KR) = -{" (dP/dz\dz-n(KR)? 4v)H - n(KR)* R*[(1- x?) - (14 x?) In(1/x)| In addition, there is an upward force associated with the frictional drag by the fluid 22(kKR)H(-t,, hen = 2n(xR)Hu{ e} r=KR = aati aa) 9-1When these are equated and the result solved for the viscosity, the expression in Eq. 22C.4-2 is obtained. c. Next put k=1-€, and expand in powers of €, keeping terms up to €°. Use §C.2, and obtain Eq. 2C.4-3. 9°952C.5 Falling film on a conical surface a. A mass balance on a ring of liquid contained between s and s+ As gives (228(sin B)5(s)(v))|. - (278(sinB)6(s)(2))).. =0 Letting As— 0 then gives d d 45 (89(2)) =0 whence, from Eq. 2.2-20 ral 5°) =0 Equation 2.2-20 is valid strictly for a flat plate with constant film thickness, but we apply it here approximately to a different geometry. b. When the equation in (a) is integrated, we get sd° =C, in which C is an integration constant. This constant is determined by requiring that the mass flow down the conical surface be the same as that flowing up the central tube (i.e., w). We hence write (width of film) x (thickness of film) x (mass flow rate), and then use Eq. 2.2-20: w =(2assinB)-5-p(v) = (2assin B) (S" of BC eost From this we get C; C= 3yuw _ 3yw (2msinB)-(p*gcosB) p’gsin2p The film thickness as a function of the distance down the cone from the apex is thus 5- 3yw np’ gssin 2B 2-262C.6 Rotating cone pump a. Inner cone not rotating. For sufficiently small values of B, the flow will resemble very much that for a thin slit (see Problem 2B.3), for which the mass rate of flow is given as the answer to part (b). This formula may be adapted to the flow in the annular space of height dz. as follows, if the inner cone is not rotating and if the gravitational force is not included: _ 2( 2) B°(2azsin 0) 3\ dz i where we have made the identification (py) —p,)/L——dp/dz and W > 2ar=2nzsin@. Across any plane z = constant, the mass flow rate will be a constant. Hence the above equation can be integrated to give Pa 3 pw fl, dz 3. pw L, _[?? dp - 2 Hehe ae py -2 He 4 in Bpsindlh z 7 Pi Pa a BB osind Ly b. Effect of the gravitational force and the centrifugal force The result in (a2) may be modified to include the effect of gravitational acceleration g and the angular velocity Q of the inner cone. The gravity force in the z-direction (per unit volume) is given by F.ay,2 = —pgcosB. The centrifugal force (per unit volume) acting in the middle of the slit will be, approximately, F..4. =o($2r) /r ={ pQ?zsin B, where r is the distance from the centerline of the cones to the middle of the slit. The component of this force in the z- direction is then F..,,, =4Q7zsin? B. Then the first equation in (a) can be modified to give B?(2azsinB)p L we a(-£. 1pQ?zsin? B - pgcosp) This equation can be integrated to give 9-24_ 4nB*psinB 102 ein? RYIT2 72 3uin(L, 7, p,) + (409? sin? B)(L3 - 13)-(pgcosB)(L, -L,)] Many assumptions have been used to get this solution: (1) laminar flow (turbulent flow analog is not difficult to work out); (2) curvature effects have been neglected (correction for this is easy to do); (3) entrance effects have been ignored (this can probably be handled approximately by introducing an "equivalent length"); (4) instantaneous accommodation of velocity profiles to the changing cross- section (it would be difficult to correct for this in a simple way). 7-282C.7 A simple rate-of-climb indicator a. Consider two planes of area S parallel to the earth's surface at heights z and z+ Az. The pressure force in the z-direction acting on the plane at height z and that acting at the plane at height z+ Az will be just the mass of air in the layer of height Az: Sp), ~ SP eas = pgsaz Division by SAz and then letting Az—0 gives the differential equation ap __ dp __| pM dz PS OF dz Je which describes the decrease in the atmospheric pressure with increased elevation; here R, is the gas constant. b. Let p; be the pressure inside the Bourdon element and p, be the pressure outside (i.e., the ambient atmospheric pressure). We now write an equation of conservation of mass for the entire instrument: a. ya. _ *(Pi=Po)R* Mag = Wy; Vp, = Pi = 8uL Pavg d Paygi oR Here m,,, is the total mass of air within the system (Bourdon element plus capillary tube), w, is the mass rate of flow of the air exiting to the outside, p; is the density of the air inside the Bourdon elements, and /,y, is the arithmetic average of the inlet and outlet densities within the capillary tube (see Eq. 2.3-29). The third form of the mass balance written above has made use of the ideal gas law, p=pR,T/M. If we neglect changes in the arithmetic average pressure p,y. and use the abbreviation B= AR *Povg /8uLV, we can integrate the mass balance above and get Di = eF(f Bp,e dt + c) 2-29To get p,(t) we make use of the fact that there is a constant upward velocity, so that ~—At these | = -U,=-A hi =p? Then the mass-balance equation becomes 0 p=e (J BpSe“'e” dt + C) = — a e 4“! +Ce Ft Determine the constant of integration, C, from the initial condition that p? =p? att =0. Then 9 Be“! — Ae® Pi =Po >= and paVizPo. A B-A p Beale) In the limit that t > ©, we get fir B>>A A _v,Mg __ 8VuL P+ a BRT ORR" Payg Hence for p, ~ Payg, the gauge pressure is -p,=n( Sob) Ma Pi~Po 0,( 285 [ae Hence the pressure difference approaches an asymptotic value that varies only slightly with altitude. c. To get the relaxation time, note that P= sal eA) B>>A P.(1 _ e*) whence 2-30P,-P_ =e Peet sothat ty -5- 8uLV mR* Pavg © It is necessary to have t assumption that B>>A. <<100 to insure the plausibility of the rel 9-32D.1 Rolling-ball viscometer The rolling-ball viscometer consists of an inclined tube containing a sphere whose diameter is but slightly smaller than the internal diameter of the tube. The fluid viscosity is determined by observing the speed with which the ball rolls down the tube, when the latter is filled with liquid. We want to interrelate the viscosity and the terminal velocity of the rolling ball. The flow between the sphere and the cylinder can be treated locally as slit flow (see Problem 2B.3) and hence the only hydro- dynamic result we need is dp | 12p(v,) dz oc (*) But we must allow the slit width o to vary with @ and z. From the figure we see that R? =(R-r) +(r’+.0)° —2(R-r)(r’ + 6) cos@ where r’ =r? —z’ . Solving for o we get o=-r’ +(R-r)cos@ +R,1+[(R-r)/R] (-sin? 6) The second term under the square-root sign will be very small for the tightly fitting sphere-cylinder system and will hence be neglected. Furthermore we replace Vr? —z? by VR’ —z? and add compensating terms R-VR?= 2 + (VR? 2 -V? 2") R-r SES) o=(R-r)| cos0+ R-r ae] R-r =(R- pfeoso + R =(R- Ao @+1)+ 9-3210 | = 2R-1) oo i 2(R=7) The omission of the term containing (z/R)” and the higher-order terms is possible, since the greatest contribution to the viscous drag occurs at the plane z = 0, and hence less accuracy is required for regions of larger z. Note that the above result gives correctly o=0 atz=0, 0=2,and o=2(R-r) atz=0, 0=0. Next we assert that dp/dz will be independent of 0, which is probably a good approximation. Then according to (*) (v,) must have the form (v,)=B(z)o* —(**) Next, the volume rate of flow across any plane z will be Q= ['(0,)o(9,z)Rd0 = RB(z)[““[o(0,z)] 40 = 8RB(z)(R—r)° [“[cos? 0+ a] dO=8RB(z)(R-1r)° I(x) in which a@ =(R-VR? -z?)/2(R-r). The volume rate of flow Q at all cross-sections will be the same, and its value will be, to a very good approximation Q = 2R7%%, where Uy is the translational speed of the rolling ball. Equating the two expressions for Q gives MRVo BO)= TR-1) (a) (***) Combining (*), (**), and (***) we get dp — 3mpRv dz 2(R-r)?I(a) The total pressure drop across the slit is then 2-34d odp d p= Veal ae aa into which we have to insert dz/da. Virtually no error is introduced by making the upper limit infinite. From the definition of a =-4(R-r)’ a? +4R(R-r)a The first term on the right is smaller than the second, at least for small z. Then dz~./R(R-r)da/Va, and the pressure drop expression becomes (with €? = a) ap=2 (RR 1 da= AEH Wag SHUR po a(R-r)° Te) 3mpv )R*” =4,/R(R-r) ao F(R—n® | where af? gen 4 yo - yo | J=2f- eye ab? yg (v10 +2) | 0.531 The pressure drop multiplied by the tube cross-section must, according to an overall force balance, be equal to the net force acting on the sphere by gravity and buoyancy 4nR°(p, -p)gsinB = aR’ Ap where p, and p are the densities of the sphere and fluid respectively. Combining the last three results gives the equation for the viscosity po jf legen R-n)" On] 0% R 2-342D.2 Drainage of liquids a. The unsteady mass balance is $(p5W) = (p(v.)W8), -(o(v.)W6) iz+Az Divide by pWaAz and take the limit as Az — 0, to get Eq. 2D.2-1. b. Then use Eq. 2.2-22 to get Eq. 2D.2-1: 05 __ pg 08° __ gd" 5 ot 3y oz ns which is a first-order partial differential equation. c. First let A =./pg/p, so that the equation in (b) becomes: OA 2 OA at “ oz Inspection of the equation suggests that A = ./z/t, which can be seen to satisfy the differential equation exactly. Therefore Eq. 2D.2-3 follows at once. This equation has a reasonable form, since for long times the boundary layer is thin, whereas for short times the boundary layer is thick.3A.1 Torque required to turn a friction bearing. Equation 3.6-31 describes the torque required to turn an outer cylinder at an angular velocity Q,. The corresponding expression for the torque required to turn an inner rotating cylinder at an angular velocity 0; is given by a formally similar expression, K2 T, = 4npQ;R?L 1— kK? derivable in like manner from the corresponding velocity profile in Eq. 3.6-32. The specifications for this problem (converted into SI units via Appendix F) are: 1.000 2 K = T5p5 = 0-998004; _? = 0.996012 K? 0.996012 esd = ooo = 249. (; — =) 0.003988 9.8 # = (200 cp)(10~* kg/m-s/cp = 0.200 kg/m-s Q; = (200 rpm)(1 min/60 s)(2m radians/revolution) = 207/3 radians/s R? = (1 in?)(1 m/39.37 in)” = 0.000645 m? L = 2 in = 2/39.37 m = 0.0508 m p = (50 Ibm /ft3)(0.45359 kg/Ibm(39.370/12 ft/m)® = 800.9 kg/m? Hence, the required torque is T, = (47)(0.200 kg/m-s)(207/3radians/s)(0.000645m7)(0.0508 m)(249.8) = 0.431 kg-m?/s? = 0.32 ft-lbs and the power required is P =T,Q; = (0.32 Ib s-ft)(207/3 s~!)(3600 s/hr)(5.0505 x 1077 hp-hr/Iby-ft) = 0.012 hp : In these calculations we have tacitly assumed the flow to be stable and laminar. To test this assumption, we formulate a transition criterion based on the critical angular velocity expression given under Fig. 3.6-2: QpR(1 _ «3/2 Le Insertion of numerical values for the present system gives Re := < about 41.3 for « = 1. (2072/3 radians/s)(800.9 kg/m*)(0.000645 m?)(1 — 0.998004)3/? R ° (0.200 kg/m-s 0.0048 This Re value is well below the transition value of 41.3 for this geometry; therefore, the foregoing predictions of T, and P are realistic. 3-]3A.2 Friction loss in bearings. The power expended to overcome the bearing friction is 2 2 K P=T,Q; = 4rpQ? RL G 9) in which L is the total bearing length of 2 x 20 x 1 = 40 ft for the two shafts. The specifications for this problem (converted to SI units via Appendix F) are: 16 : ee eee) ; = 0.998751 © = 659x000 0.999375; Ke? 0.998751 (; = =) = dooizag = /99-% pw = (5000 cp)(10~* kg/m-s/cp) = 5 kg/m-s Q; = (50/60 rev/s)(2m7 radians/rev) = 57/3 radians/s R? = (8/39.37 m)* = 0.04129 m? L = 40 ft = (40 x 12/39.37 m) = 12.2 m With these values, the calculated power requirement is P = (4n)(5 kg/m-s)(57/3 rad/s)”(0.04129 x 12.2 m*)(799.6) = 6.938 x 10° kg-m?/s? This result is then expressed in horsepower by use of Table F.3-3: P = (6.938 x 10° kg-m?/s*)(3.7251 x 1077 hp-hr-[kg-m?/s?]~*)(3600 s/hr) = 930 hp Thus, the fraction of the available power that is lost in bearing friction is 930/(4000+ 4000) = 0.116.3A.3 Effect of altitude on air pressure. For a stationary atmosphere (i.e., no wind currents), the vertical component of the equation of motion gives dp ac Pg The air is treated as an ideal gas, meee e< RT with M ~ 29, and with temperature in °R given by T(z) = 530 — 0.003z at elevation z ft above Lake Superior. The pressure p2 at z2 = 2023 — 602 = 1421 ft above lake level is to be calculated, given that p; = 750 mm Hg at z = 0. The foregoing equations give dinp _ ie Mg dz ——_-R(530 — 0.003z) Integration gives ; = Mg 1421 dz n(p2/p1) = mh (530 — 0.003z Mg 1 z=1421 ="R 0.003 ° : Mg ; [4] In{530 — 0.0032] z=0 =~ 9.003R "| 530 Insertion of numerical values in 1b,,-ft-s units gives (29 lb» /Ib-mol)(32.17 ft/s”) (0.003 R/ft)(4.9686 x 104 Ib,,-ft? /s?-Ib-mol R) = —0.0505 In(p2/p1) In (525.737/over530] Hence, P2 = pi exp(—0.0505) = 750 x 0.9507 = 713 mm Hg Since the fractional change in P is small, one gets a good approximation (and a quicker solution) by neglecting it. That method gives pp = 712 mm Hg.3A.4 Viscosity determnation with a rotating-cylinder viscometer. Here it is desirable to use a sufficiently high torque that the precision of viscosity determinations is limited mainly by that of the measurement of angular velocity. A torque of 104 dyn-cm, corresponding to a torque uncertainty of 1%, appears reasonable if the resulting Reynolds number is in the stable laminar range. The geometric specifications of the viscometer are: R= 2.25 cm; KR = 2.00 cm « = 2.00/2.25 = 0.888889; «? = 0.790123 1—r? = 0.209877; (KR)? = 4.00 cm? D8 ot R? = 5.0625 cm? The angular velocity corresponding to this torque value is: T,(1 — Kk?) (104 g-cm?/s*)(0.209877) ° Anp(KR)Y2L ~~ 47(0.57 g/cm-s)(4 cm?)(4 cm) 18.3 radians/s The Reynolds number at this condition is: _ MR? p _ (18.3)(5.0625)(1.29) _ R © pe 0.57 210 Accordng to Fig. 3.6-2, this Re value is well within the stable laminar range; therefore, a torque of 104 dyn-cm is acceptable.3A.5 Fabrication of a parabolic mirror. Equation 3.6-44 gives the shape of the free surface as 2 ene (Z)e The required derivatives of this function at the axis of rotation are dz Pz 77? and wg Setting the desired focal length equal to half the radius of curvature of the mirror surface at r = 0, and using Eq. 3A.5-1, we obtain — log? f= 59/2 Thus, the required angular velocity to produce a mirror with focal length f = 100 cm at standard terrestrial gravity is g ee 2f __ | 980.665 cm/s? ~ Vi (2)(100 cm) = 2.214 radians/s which corresponds to 602/27 = 21.1 revolutions per minute.3A.6 Scale-up of an agitated tank. The specifications for the operation in the large tank (Tank I) are Ny = 120 rpm; M1 = 13.5 cp; pr = 0.9 g/cm? and the tank is to be operated with an uncovered liquid surface. To allow direct prediction of the operation of Tank I from experiments in the smaller system (Tank II), the systems must be geometrically similar and must run at the same values of Re and Fr. To meet the latter requirement, Eqs. 3.7-40,41 must be satisfied. Equation 3.7-41 requires Dy Ni = DN? when, as usual, the gravitational fields for the two systems are equal. Then the model must operate at Nu = Ni/Di/ Dy = 120V10 = 380 rpm and Eq. 3.7-40 requires _ Du\* (Nn\? Wy = Dy Mi = (13.5/0.9)(0.1)?(V/10) = 0.474 ep From Table 1.1-1, we see that this value of 17 corresponds closely to the value for liquid water at 60°C. Thus, the model should opeerate at 380 rpm, with liquid water at very nearly 60°C.3A.7 Air entrainment in a draining tank. As this system is too complex for analytic treatment, we use dimensional analysis. We must establish operating conditions such that both systems satisfy the same dimensionless differential equations and boundary conditions. This means that the large and small systems must be geometrically similar, and that the Froude and Reynolds numbers must be respectively the same for each. Choose D (tank diameter) as characteristic length, and (4Q/mD?) as characteristic velocity, where Q is the volumetric draw-off rate. Then _ 4Qp aDu 16Q? d 4 an r 7 D8g Re Subscripts L and S will be used to identify quantities associated with the large and small tanks, respectively. We take the gravitational field g to be the same for both. Then the requirement of equal Reynolds numbers gives (Gi) (a5) = (Fi) (Fs) = Cea) 2ton = com, and the requirement of equal Froude numbers gives (35) = (Bi) Combining these requirements, we obtain D (32) = (0.02277)"/* = 0.080 Dr ° Hence, Ds = (0.080)(60 ft) = 4.8 ft Qs = (0.080)°/?(800 gal/min) = 1.46 gal/min Therefore: a. The model tank should be 4.8 ft in diameter. b. Its draw-off tube should be 0.080 ft in diameter and 0.080 ft high. c. Its draw-off tube should have its axis 0.32 ft from the wall of the tank. Furthermore, if water at 68°F (20°C) is withdrawn from the model tank at 1.46 gal/min, air entrainment will begin when the liquid level is (4.8/60) of the level at which entrainment would begin in the large tank at its withdrawal rate of 800 gal/min. 3-73B.1 Flow between concentric cylinders and spheres a. The derivation proceeds as in Example 3.6-3 up to Eq. 3.6- 26, which we choose to rewrite as 2 Po = D,+ p,{ =) r r The boundary conditions are that vg(KR)=Q,;KR and vg(R)= Q,R. Putting these boundary conditions into the above equation for the angular velocity gives Q,=D,+D,4 and Q,=D,+D, K These equations can be solved for the integration constants Q, Yx«*| |Q, 1 p, =o 1 I_ K*|_ Ox? -Q, ‘ft aw] fl Tp K-71 : 1 i K? ll OE ake 12 _ F Q,| |? Q,|_ «?(Q,-Q,) ef Vx] fl 1 K?-1 f 1 1K? Hence the solution to the differential equation is Vp (9. 8s") (1-95) aR) rt 1-K? 1-K? \r The z-components of the torques on the outer and inner cylinders are T,=[f."(-t oR), p Rao = 2ntr’| sur (2) ] 3-3= nt RRP 2) KR) ‘(- 225] =—4muiL(2, -9,) = 1-K? Leda 2 d(v T, =(h (+7,9KR),__, KRdOdz = 2nL(KR) nore (2)] KR = +4muL(Q; -Q ee eR) b. In Example 3.6-5 it is shown how to get Eq. 3.6-53 for the velocity distribution. The boundary conditions are : v,(KR) = KRQ;sin@ and 0,4 (R) = RQ; sin @. Equation 3.6-53 can be written in the form The constants can be obtained according to the method of (a) and the final expression is % _(Q,-Q;«°) SS) eR! rsn@ 1-K? pe The torques at the outer and inner cylinders are then T, =f (- T, + rap’ (Rsin @)R? sin odod 1 pn v =f" + we (**)] (Rsin @)R? sin 6dedp =R =-87p(Q,; - 2,)(®, («R)" T.= [0 (+t ) eg (KRsin 6)(KR)’ sin Od0d9 = +82(Q; - Q 3 —33B.2 Laminar flow in a triangular duct a. It is clear that the boundary conditions that v, =0 at y=H and at y=+ 3x. Therefore the no-slip boundary conditions are satisfied. Next it has to be shown that the equation of motion gu Po- Pi, { H0. , Or, at ay? is satisfied. Substituting the solution into the second-derivative terms, we get P-P, a Hf |S Flor y—y°? -3Hx? + Hy *) = “1-7. i i -( aa (6y-6H -6y+2H) and this just exactly cancels the pressure-difference term. b. To get the mass rate of flow we integrate over half the cross-section and multiply by 2: P,-P B w= 2¢{ FoF *(y-H)(32? -y”)dxdy PP, pH y/N3 =f FFL) ye yal ay _ PoP, pay -2y ¥ 4uLH Jk (y a 3372 ay a P-P, \(_H° a V3 (Po - P,)H* 33? uLH J 20 180 uL The average velocity is then the volume rate of flow (w/p), divided by the cross-sectional area H?//3 so that (P)-P,)H? 60uL (0,) = 3-\0The maximum velocity will be at the tube center, or at x = 0 and y = y = 2H/3, so that = (Po -P,)H? Oz max ~ 27 Ul L ofS a)3B.3 Laminar flow in a square duct a. The boundary conditions at x = +B and y=+B are seen to be satisfied by direct substitution into Eq. 3B.3-1. Next we have to see whether the differential equation Po -P av, . 02, oa lPo-P, (28.28) is satisfied. Substituting the derivatives from Eq. 3B.3-1 into this differential equation gives oa (e-Pi)-ofo=PUls (2) -(2)) Hence the differential equation is not satisfied. b. The expression for the mass flow rate from Eq. 3B.3-1 is given by 4 times the flow rate for one quadrant: wag oF oe elt-(3) [-(8) ow PoP IPP Bo (1-e\(1- aden PoP, )B pry, eo) eP ~ PoP (i as] _(Po-P,)B%p a= 0.444(P, — P, )B*p BL : BL3B.4 Creeping flow between two concentric spheres a. From Eq. B.4-3, there is only surviving term on the left side (10 sin @)=0 whence _-v, sin@ = u(r) b . From Eq. B.6-8 (omitting the left side for creeping flow) the only surviving terms are o-14F , 4 (7 Be) or 02-2 2® 1 (7%) do Par dr ~F9e WP sinedr\ ar c. When the equation in (b) is multiplied by rsin @, the left side is a function of @ alone, and the right side contains only r. This means that both sides must be equal to some constant, which we call B. This gives Eqs. 3B.4-2 and 3. Integration of the pressure equation proceeds as follows: l(7_ 1 feap=[r a0 or P, -P, = Bint E) _ py Sot2e 1 & sind Al 1 tanz€ tan>€ From this we get the constant B B= P,-P, _ PLHP, ~Incot?4e 2Incothe Next we integrate the velocity equation “( 2) Br pas C,R } —r°—|=— or u=—}—-——+C, dr dr L 2u\R or where we have selected the constants of integration in such as way that they will be dimensionless: C,;=-K and C, =-«-1 (from the no-slip condition at the walls). When this solution is combined with the expression for B we get: va BRL (1-2)-a(-) fe Pe B12) oe(-8)] 3-13which leads to Eq. 3B.4-5. d. The mass rate of flow must be the same through any cross- section. It is easiest to get w at 0=4}a where v, =u(r)/sin@= u(r)/sind 2 =u(r): w= fy fiePPoly.1,Fdrdp = 2nR of ube - 2 (P,-P,)R a _ (1-4) =2aR 2 er ile é)+K 1 Z Ed = anr%p (Pi-Pa)R 1 _ x) see Eq. 3B.4-6) 4ulncothe 6 ( q_3B.5 Parallel-disk viscometer a. The equation of continuity (Eq. B.6-5) just gives 0 = 0. Equation B.6-5 gives for the @-component of the equation of motion = $22 (rm )}s CA or\roar’ ° On b. When the postulated form for the tangential component of the velocity is inserted into this, we get simply d?f/dz? =0, which has to be solved with the boundary conditions that f(0)=0 and f(B)=Q, which are just statements of the no-slip conditions on the wetted surfaces of the disks. It is easily shown that f = Qz/B. and therefore that 0, = Qrz/B. c. The z-component of the torque exerted on the fluid by the upper rotating disk is ec =f (- Trg? dL . yrirae = 2af| u( Ser) rdr= 27 =f rdr z=B We can now do the r-integration and solve for the viscosity to get the desired formula, u = 2BT,/7QR‘. 3-153B.6 Circulating flow in an annulus a. Neglect of curvature gives for the z-component of the equation of motion dP d*v, dv, dP? -——+u or = dz O= — dr? Pa ae The left side is a function of r alone and the right side is a function of z alone, so that both sides must equal a constant. Therefore d’v, rr C’ with boundary conditions v,(«KR)=v,) and v,(R)=0 Use the dimensionless variables @ = v,/v, and € =r/R; instead of the latter, it is more convenient to use ¢=(& - x)/(1- x) in this part of the problem. Then 2 an =2C, with boundary conditions ¢(0)=1 and ¢(1)=0 Integration of this equation gives G=C,0? +C,0+C, The boundary conditions give C; =1 and C, —(1+C,) so that @=C,6? -(14+C,)O+1 The mass-balance condition i gat =0 gives C, =3, and the velocity distribution is 6=30? -4641 b. We use exactly the same procedure when the thin-slit approximation is not made, but the algebraic manipulations -are messier. The equation to be solved is 3-16tly os) = ae rdr\. dr dz with the same no-slip boundary conditions as before. Use the same dimensionless variables as in (a) and find that the solution has the form $=C,é7 +C,InE+C, The three constants of integration are determined from the no-slip boundary conditions ¢(k)=1 and 9(1)=0, along with the mass- conservation condition { “I. édéde =0. The results are Cc =-c __1-[2x/(1- x?)]in(/x)_ _ 1-2 © 8 (t= «?)=(14+ «?)in(Yx)’ * (1- x?) = (14 x? )In(1/x) With these expressions for the integration constants, Eq. 3B.6-2 follows. 3-1)3B.7 Momentum fluxes for creeping flow into a slot a. Inside the slot, the nonzero component of pvv is 3w \° y a Pv,0, A aa) -(2) | Outside the slot, the nonzero component of pvv are pv. =A 2”) py y = of 20) ty ae mWp) (x2 +y?)*’ yy “Pl Wp (x? +y7)* pv», ae @ +y’)° _ 105 2w \) ~ 768" 2) a The integrals appearing here can be found in integral tables. We conclude that the total flow of kinetic energy across the plane at x = -a is not the same as that in the slit. As a— 0, the flow of kinetic energy tends toward zero, since the fluid velocity tends to zero as x + —co. This emphasizes that kinetic energy is not conserved. e. Eq. 3B.7-1 clearly satisfies the equation of continuity, since for incompressible flow (00, /dx) + (dv, /dy) + (dv, /dz) = 0. When the derivatives are calculated from Eqs. 3B.7-2, 3, and 4, it is found that these expressions also satisfy the incompressible equation of continuity as well. f. From Egs. 3B.7-2 and B.1-1 we get Ov, 2w 3x? Ax* 4wy ve y ¥) \yeo ox Qw 3x? 4x* =2 os - — > =0 oa a) +y? y (x? +y" y L The second of these is an illustration of Example 3.1-1. g. From Eqs. 3B.7-2, 3B.7-3, and B.1-4, we get, after evaluating the derivatives _ Wy | ov, tena 54 7) Ov, Ox |, x=0 3-143B.8 Velocity distribution for creeping flow toward a slot a. For the given postulates, the equation of continuity gives =—(rv,)=0 from which it follows that 9, = i f(0) r Since the flow is symmetric about 6=0, df/d@=0 at @=0; and since the fluid velocity is zero at @ = +47, it follows that f=0 at @=ti2 b. The components of the equation of motion given in Eqs. B.6-4 and 5, appropriately simplified are oP nef aP Qu df o=- 2 HES d o--2 we Pde 7 00° 2 dé c. When the first equation is differentiated with respect to 0 and the second with respect to r and the two results subtracted we get Eq. 3B.8-1. d. The equation in Eq. 3B.8-1 can be integrated once to get d af +4f=C, A particular integral is fp; =+C,, and the complementary function is (according to Eq. C.1-3) fog =C,cos20+C,sin20@ . The complete solution is then the sum of these two functions. e. The integration constants are determined from the boundary conditions. It is found that C, = 0, and that $C, =C,. Then from +7/2 n/2 +7/2 w=-Wof 72 dO = -Wo- a2? ‘fd0 =-2WpC, Van cos’ 6d@ = -WpC,x we get C, =-w/Wo7 and the velocity distribution is given by Eq. 3B.8-2. f. From the velocity distribution and the equations obtained in (b) we can get 3-20a = af Se. cos* @-sin’? 6) and hence r--4 2 Jfcos?o- sin 0)+F(8) (*) r’?\ zWo Furthermore OP Qu df (4) 2w 2 —_— = Pa 907 72 db or > |\ ap cos’ 6+ G(r) Here F and G are arbitrary functions of their arguments. The second expression for the modified pressure can be rewritten as P= -&f Zo. cos 9+(1-sin? @))+G(r) = 4 22. (cos? @-sin? 9) +H(r) (*) By comparing the two (*) expressions for the modified pressure, we see that they are the same except for the functions F and H. Since the first is a function of @ alone and the second a function of r alone, they must both be equal to a constant, which we call P... This is the value of the modified pressure at r =o. g. The total normal stress exerted on the wall at 0= 2/2 is’ (when one uses the result of Example 3.1-1) Teo loena = (p + Tog Vleeaa= (P ~ pSH)| 5 ays =P + 2yw 2uw = ”Wor? nWor? — pgh=p.. + h. From Eq. B.1-11 3-2!Corl op =H ro (22) 42S Sr lo=nj2 or r) roe pond = -ifo -( sit |(-2e0 osine| The first term is zero since v,/r=f since vg =0 was one of the postulates. The second term is zero, as can be seen by using Eq. 3B.8- 2, and the fact that cos$=0. This is agreement with the result in Problem 3B.7(g). i. Since the z-component of the velocity is zero, we can expand the velocity vector in either the cylindrical coordinate system or the Cartesian system thus =0 0=n/2 v=6,0, +850, =8,0, +5,0, Since vy =0 was one of the postulates, when we take the dot product of this equation with 5, we get the x-component of the velocity 3 3 v, =(5, -5,) =, cos =— 2w cos’ @=- ———— mWor mWor nWp(x? +y?) Similarly for the y-component of the velocity _ — ow wy 2y = (By 8) =0, sind = - =F cos Osin® =~ Wort 2wx?y _ nWo(x? +y? y These results are in agreement with Eqs. 3B.7-2 and 3. 3-23B.9 Slow transverse flow around a cylinder a. At the cylinder surface we get by using Eq. B.1-11 Cuv,, cos @ . Pl,er = po — ——+ In(r,/1,) d. When this result is substituted into the @ equation we get fod = Be ao P,-P, . P,-P, 2 =-—1+—_—+, from which g=- Tiny g ORT uln(r,/r,) 2 , dz’ wIn(r/r,) The integration constants are obtained from applying two boundary conditions. We could require that ¢=0 at z=+b, and thereby determine the integration constants. Another method is to recognize that the flow is symmetric about z = 0, and use as one of the boundary conditions d¢/dz=0 at z = 0. Either method will give C, =0, and then C, is easily obtained. The final result is 3-24(P,-P,)v? (2) @ =i 24] 1-] = 2uIn(r,/r,) b Division by r then gives Eq. 3B.10-3. e. The mass rate of flow at any cylindrical surface in the system must be the same. Select the surface at r=r, and obtain 20 (P,- w= f (oe, | a dzr,d0= 2b - +2: Serie Ke -é Me The integral gives 2/3, so that Eq. 3B.10-4 is obtained. 3-253B.11 Radial flow between two coaxial cylinders a. From Eq. B.4-2 we get for this flow, with 7,(r), 1 . ol) =0 whence v,= C where C is a constant r dr r Atr=R, 0,(R)=C/R so that C = Ro, (R). b. The relations in Eq. 3B.11-1 follow immediately from Eggs. B.6-4, 5, and 6 for the velocity profile v,(r) in (a). c. Integration from r to R gives R 1 1 a tec'(gr- This gives, making use of the meaning of C obtained in (a) P(R)-P(r)= ec'{-31,] 2 Per) PER) = oR ROP ( eZ) = bole 00} [1-() | d. The only nonzero components are (from Eqs. B.1-8 to 13) dv 1 v 1 = 2 2" ~s9uC; = 242 = -2yC Uy Mo HES Too aa Wa 3-263B.12 Pressure distribution in incompressible fluids The equation of motion method to get the pressure distribution is correct. On the other hand, the second method gives nonsense, as one can see from Fig. 3.5-1. For an incompressible fluid (the vertical straight line), specifying the density does not give any information about the pressure. 3-213B.13 Flow of a fluid through a sudden contraction a. For an incompressible fluid, Eq. 3.5-12 becomes 4p(03 —v?)+(P, —P1)+pg(hy -hy)=0 or 40(23 -v7)+(P, -P,)=0 If "1" is the large tube and "2" the small one, then the fluid velocity in "2" must be greater than in "1." Then the modified pressure in "2" must be less than that in "1." Thus the modified pressure decreases as the fluid moves from the large cross-section region to the small cross-section region, in agreement with experimental observations. b. For an ideal gas, Eq. 3.5-12 becomes RT 40(v; ~ Pf) + pin + aah -h,)=0 The pressure and elevation terms may no longer be combined. If the elevation does not change, the pressure decreases as the fluid moves into the contracted part of the tube. 3-283B.14 Torricelli's equation for efflux from a tank From Eq. 3.5-12 we get 1 4 (One -0) +5 (Pam —Patm) + ¥(0-h) =0 Here it has been assumed that the velocity at the surface is virtually zero, that the pressure is atmospheric at both "1" and "2", and that the datum plane for the height is at the exit tube. When the above equation is solved for the efflux velocity we get Torricelli's equation. 3743B.15 Shape of free surface in tangential annular flow a. The velocity distribution is given by Eq. 3.6-32, and the equations corresponding to Eqs. 3.6-38 and 39 are: zfs) (8) -n(g)] amt Boon Integration of these equations gives (see Eqs. 3.6-40, 41, and 42) QR) | (RY 2 p= ig 22] -(2) -4inr (7) —pgz+C Now let p=Patm at r= R and z=Zp, where Zp is the height of the liquid at the outer-cylinder surface. Then we can write at r= R and Z=Zp Q,«?R 1-K? 2 Pam = ) [-1-4InR +1]- pgzp +C which is the equation that determines C. When we subtract the last equation from the equation for p, we get Q,x?R) P~ Pam = 46 SE) (-Z-4ing +6") -pe(e~20) The equation of the liquid surface is then the locus of all points for which p = Patms OF 1 (Q,x?RY (1 oye (92°8)(Lsame-e) b. When the outer cylinder is rotating, we can use Eq. 3.6-29 for the velocity distribution to get 3-30_ (Q xR [1f ry 1( KR)? p= 2S) Fea -2inr-3( 2) —pgz+C Then, we select r = R and z= 2p as the point to determine C. Q,KRY1 101) 1 Prim =o 2) 32) ~2ine-3(4) | ee i Subtracting, we get P~Patm = o( et) (() -(2) -4lng -(4) +(0) |e -2) From this equation we can get Eq. 3B.15-2 by setting the left side of the equation equal to zero and solving for zp —z. 3313B.16 Flow in a slit with uniform cross flow From Eq. B.6-1, for this problem we have do, _(Po-P.) , d’v, Po _ 4 .4=0 dy L M dy’ Can dn P% in which n=y/B, ¢=0,/|(P)-P,)B’/uL|, and A=Bo,p/p. This equation has to be solved with the no-slip condition at 7 =0,1. We write the solution as the sum of a complementary function and a particular integral. The equation for the complementary function is a d . . Cc oe - age =0 withsolution ¢ = zen +Cy By inspection, the particular integral is ¢p, = /A. Application of the boundary conditions then gives the constants of integration. The final solution is then (with A = Bu,p/y) e-3{ 3) or 8 eur rt) Al" 4-1 wl OA\B. Oe4—1 b. The mass rate of flow in the x-direction is then P,-P,)Bw, eT ain ~o-FiJF Were 1 15) w= [p f, ev.dydz = uL A + 2 A e4-1 c By making a Taylor-series expansion about A = 0, from (a) we get ¢=4(n-1?)+O(A). When A- 0, this result can be shown to be equivalent to Eq. 2B.3-2. Similarly, A Taylor-series expansion about A= 0 yields from the result in (b) leew alsa a) [(Po-P,)WB%p/uL] A\2 A e4-1 3-321 "Gsasdat+ LAS +-.-)- (oaarareyl mln a Nie >I -1j1_1, 1 Al2 A "AQ gA+dAae A(1+4A+3A?+--) _1f(1_1,1-jA+pA?t- A\2 A A = qt O(A) But the "B" in this problem is twice the "B" of problem 2B.3. If we switch to the "B" of Problem 2B.3, it is found that the answers agree exactly. d. For the coordinate system here, we select as the dimensionless quantities y Ux buyp Y=+; V=———— 4 = bf (8 P,P a Then the differential equation and boundary conditions are dV av —al+-, ith V(t) =0 a ay + WV wi (+1) The solution is then the sum of a complementary function and a particular integral (as before in (c)) v=Se ac 4h a a Application of the boundary conditions then leads to ~—e*’ +cosha+Ysinha asinha V= Then the average value of this over the cross-section of flow is 3-33sie" +cosha+ Ysinha)dY _ -(1/a)sinha + cosha asinha 7 a sinha Then we can form the ratio given in Eq. 3B.16-3: Vv_ e”” —Ysinha-cosha (V) (1/a)sinha-cosha As a check on this we can go directly from Eqs. 3B.16-1 and 2 to Eq. 3B.16-3. From the first two equations we get y_evh-1 A _4\_ Bl pAy/B Dy Pe _ Bet aq__ 2Al(e*-1)y-Ble”” -1)| (v,) wfWB 11, 1 pl(a-2)(e4-1)+24] ee Next, we make the connections between the notations in the two different approaches: y=z+b; B=2b; A=2a (the "y" of part (c) is called "z" here, and ¢ = z/b). Then Dy ar|(¢? * —1)(z+b)- 2b(etw/ner _1)| vy) o|(or-1)(e?* -1) +20] _ (24 -1(S+1)-2(e*e"-1)_ (e*-e°*)(S+1)- 2% -e-*) (1/a) [(@-1) 1)(e?* -1) )+2a] (1/a («- Ile" -e*) +2007 “| _ Se me “)+ _ +e #)— 2eF _ sinha: + coshar—e% (1/ar)| a(e* +e*)-(e* -e*)| cosh a - (1/cc) sinh a — 3-343C.1 Parallel-disk compression viscometer a. The equation of continuity of Eq. B.4-2 for incompressible fluids, taking into account the symmetry about the z-axis is just Eq. 3C.1-6. The equation of motion in Eq. 3C.1-7 comes from Eq. B.6-4 ignoring the hydrostatic pressure, the inertial force terms, and omitting the terms that are small. b. Equation 3C.1-7 can be integrated with respect to z to give _1d > v0, = Du dr — 2? +C\z+C, The constant C, is found to be zero from the boundary condition in Eq. 3C.1-8, and C, is found from Eq. 3C.1-9. c. Integrating of Eq. 3C.1-6 with respect to z from 0 to H 23 1 dp oral” FiGhae—H) a= -f"a, Performing the integrations then gives 12urdr\ dr) ° d. Integration of the equation in (c) then yields _12pv, 1? H? —+C,Inr+C, 4 The integration constant C, must be zero, since the pressure is finite at the center of the disks, and C, is determined from Eq. 3C.1-10. Equation 3C.1-13 is thus obtained. e. The force on the upper plate is then 2n pR3UV,R* r 2 3uDpR™ FO= lb Tap “| -(§) = aR (DP at & HS The integral is 1/4, and this leads to the result in Eq. 3C.1-14. 3-35f- In this situation, the radius of the glob of liquid R(t) and the instantaneous disk separation H(t) are related to the sample volume Vby V = a[R(t)] H(t). Then the force acting on the upper disk is F=f") Rey SHPO ROT [Rey p- (ai) peo [Hy R(t) _ 3mpv,[R(b)] _ 3p0V? ~ 2H — 2H)? g. If, in Eq. 3C.1-14 we replace v, by —- dH/dt, we then have an ordinary, separable differential equation for H(t). Integration gives 2F, far=-f" dH 3muR* ?0 Ho H® whence 1 1 4F ot = + [H()~> Ho 32uR*3C.2 Normal stresses at solid surfaces for compressible fluids First write the equation of continuity for a compressible fluid as o 1 $inp=-(V-v)-2(v-Vp) The normal stress on a surface perpendicular to the z-axis is - 00, Txe|,-9 -(-2u az a (3u ~ K)(V-¥)) - dv, My 9 1 5 a 1 — eee eee epee = _ 4yu— —— — -V, 24 et iy aoa a ve) +(x | ea Al. Q =(4u+ «(Zino 7 The terms containing v drop out by the no-slip condition at the surface, and their derivatives with respect to x and y drop out on the surface as explained in Example 3.1-1. This result shows that the normal stresses at surfaces are zero for compressible flow if the flow is at steady state. 3-373C.3 Deformation of a fluid line The curve at any time t is 6(r,t) = (v9/r)t, which in tangential annular flow is (from Eq. 3.6-32) _((R/r) =1 __2(R?/r?)Q,t O(r,t) (ea 75 Qt and d@= aaa 7 dr The differential element of length along the curve is given by (dl? =(dr)? + (rd0)* =(dr)?| 14 4(Rir)"(Qit)” (yx) - 1) The total length of the curve is then : 2 2p72 4(R/r) (2a) dr= Rf! a” 16x N —d ((yx)° -1) ((¥/x)’ -1) € To get a rough, order-of-magnitude estimate assume that N is large and then the "1" can be neglected and the integral performed analytically l=fidl=f', [i+ _ 4aNx I R 1+k (limit of large N) 3-333C.4 Alternative methods of solving the Couette viscometer problem by use of angular momentum concepts a. By making an angular momentum balance (actually the z- component of the angular momentum balance) over an annular region of thickness Ar and height L we obtain =0 r+Ar (2rL)-(r7,9) 7 (2a(r + Ar)L)-(rt,9)| Dividing through by 2”LAr and rearranging we get 2 2 Po9)| -(r 49) ( ig r+Ar wg Tr Ar =0 whence the second form resulting from taking the limit as Ar > 0. Then using Eq. B.1-11 for the stress-tensor component, we have d 3 d “2 —| =] -2))=0 itr ral r whence c Vg=— a + Cor From this Eq 3.6-20 follows. b. Here we start with Eq. 3.4-1, which simplifies to the following for the symmetric stress tensor [V-{rxc}*]=0 The z-component of this equation is 3-3422 (rrxt})=0 or 7£(r{rxt},,)=0 where, in cylindrical coordinates, r=8,7 +5,z. We now work out the cross product, which is {rxt}, = »y Dy Eni (5,7 +8,2), Tj, =€ 167 Tor +E 29,(O) Tr ij Hence the equation of change for angular momentum simplifies to 1d ara ( 7 Te ) =0 r dr and the development proceeds further as in (a).3C.5 Two-phase interfacial boundary conditions a. This result follows at once from Eq. 3C.5-1, when the viscous-stress-tensor terms are omitted. b. To get the right side of Eq. 3C.5-3, it is evident that Eq. 3C.5-1 had to be multiplied by 1/p'v; . The interfacial-tension term in Eq. 3C.5-3 is then (ere lloml le allee ny—+— 21 =n] =+ = 772 Ry Ry )LL 00 R, Ry) Lh The terms involving the viscous stress tensor are II Wot st [n “Tt ]_ vey] oe [n ‘Y |v, =-[n'-7"] Fea p'N% p'0 pm | lvoe" [ate] ante] alata) tt] ue pt 1,2 L2. 7 1,2 iL ny i il TT pv% pv Pv 0 0%P Pp And finally, the pressure terms are converted to modified pressure terms plus terms involving the gravitational acceleration of Paestpiic ty) eons sch) Soe Se) 1,,2 Tm “ pv P % pive feel) LSE We see that the Reynolds numbers for the two phases, the Weber number (Eq. 3.7-12), and the Froude number (Eq. 3.7-11) appear as well as the density ratios for the two phases. 3d3D.1 Derivation of the equation of motion from Newton's second law of motion a. Equation 3D.1-1 is the statement that the time rate-of- change of momentum is equal to the sum of the surface forces and the gravity forces acting on a small blob of fluid. When the Leibniz formula (Eq. A.5-5) is applied to the left side of Eq. 3D.1-1, we get ay pvdV = f © pydV + J(ev)(v-n)ds= f 2 pvdV + J[n-pvv]as dt vit) vit) ot S(t) Vit) ot S(t) = f o pvav + J[V-evv]dV (using Eq. A.5-3) vit) ot vit) 7 The term containing the stress tensor in Eq. 3D.1-1 can also be rewritten as a volume integral using Eq. A.5-3 to give J 2 ovay =~ J[V-evv]dv - J[V-n]dV+ fogdv V(t) ot Vit) Vit) Vit) Since the choice of the blob volume was arbitrary, all the volume integral operations may be removed, and we obtain the equation of motion of Eq. 3.2-9. b. If the blob is fixed, then we can write a momentum balance over the blob as follows: glen = oe ace ea This states that the rate of increase of momentum within the fixed volume equals the rate of increase because of convective transport, the rate of increase because of molecular momentum transport, and the force acting on the system by gravity. The time derivative can be taken inside, since the volume is fixed, and the surface integrals can be converted to volume integrals. The result is an equation containing only volume integrals over the fixed volume: js ppd = Il [V-pvv]dv - Jiv-m] 1 Jav + Jogdv v9 3-A2Since the volume was chosen arbitrarily, the volume integrals can be removed, and, once again, the result in Eq. 3.2-9 is obtained. 3-433D.2 The equation of change for vorticity Method I: Start with the Navier-Stokes equation in the D/Dt-form, but rearranged thus: Fay W)-Svp+ wW*v+g =-Viv? +[vx[vxv]]-Cvp + wvig Next we take the curl and introduce the vorticity w =[V xv] OW _tyx[vxw]}+ w?w ot or & <[y-wy]-[V-vw]+ w?w Then using Eq. A.4-24 and the fact that (V-v)=0 for incompressible fluids and (V- w) =0 always (since the divergence of a curl is always zero, we get Eq. 3D.2-1. Method ITI: Start now with the Navier-Stokes equation in 0/ot-form = -[V-w]-2p+ wvtg Take the curl of this equation and introduce the vorticity to get ae =-[Vx[V-vv]]+ wW?w or me = [e:{Vv . Vv}| - [v : Vw] + WW2w 3-44Details of the manipulations involved in this last step are given here using the abbreviated notation of §A.9 with the Einstein summation convention: -[V x[V-w ]], =—€.9; (0:00, ) = —€).9;(210;0, — 7,.0,2,) but 9,0, = 0 = Ein (a2, (2,2) + (0,9; )| = €ij(9j24)( 42x) -(2191(€.9;2%)) =[e:{Vv-Vv}].-[v-V[V xv] i i in which € = DDLe,,, 55,0, is a third-order tensor. mnr= m~n~ rT 3-453D.3 Alternate form of the equation of motion Take the divergence of the equation of motion for an incompressible fluid in the form of Eq. 3.2-9, but with the stress- tensor term written in terms of the viscosity and the Laplacian of the velocity. This gives 0=-(V-[V-wv})-4v*p p or 0=-(Vv:(Vv)*)-4v? (Vv:(Vv)*) le Then use the definitions in Eq. 3D.3-2 to get Eq. 3D.3-1. 3-db4A.1 Time for attainment of steady state in tube flow. a. In Figure 4D.2, the centerline velocity comes within 10% of its final value when vt/R? = 0.45, giving t = 0.45R?/v = (0.45)(0.49 x 107* m?)/(3.45 x 107* m?/s) = 0.064 s b. If water at (68°F =20°C) is used, with v = 1.0037 x 10-° m?/s from Tables 1.1-1 and F.3-6, the time required is t = (0.45)(0.49 x 107* m?)/(1.0037 x 107° m?/s) = 22 s4A.2 Velocity near a moving sphere. From Eq. 4.2-14 at @ = 7/2, the fluid velocity relative to the approach velocity falls to 1% of voo. at vg = —0.99vc9 relative to the sphere, giving om fi) with R/r <1. If R/r << 1, the cubic term will be unimportant, giving 3/R oa a5 R 4 Clearly, the neglect of the cubic term at this distance is justifiable. or4A.3 Construction of streamlines for the potential flow around a cylinder. In the following drawing we.show the construction of the streamline V = 3 by the method described in the problem statement.4A.4 Comparison of exact and aproximate profiles for flow along a flat plate. Let Tl, = vz/voo and Y = y,/vq./vz; then Eq. 4.4-18 gives the approximation 3/3. 1/13\3?_, e= 3am" ~ 3 (zen) ¥ = 0.323209 — 0.005002Y* and Fig. 4.4-3 shows Blasius’ “exact” II, vs. Y. These two velocity representations will now be compared. Location, Approx. II,, “Exact” II,, Approx. II, /Exact I, Y Eq. 4.4-18 Fig. 4.4-3 a. 1.5 0.468 0.49 0.96 b. 3.0 0.835 0.84 0.99 c. 4.0 0.973 0.96 1.014A.5 Numerical demonstration of the von Karman momentum balance. a. The integrals in Eq. 4.4-13 are co ‘co = | pvz(ve — vz) dy and h= [ p(ve — vz) dy 0 0 Figure 4.4-3 gives f' = v,/voo as a function of a dimensionless coordinate, Y = yV%c0/(vz) and ve = Uoo in this geometry. Thus, vz = voof' and pdy = ./ppr/voodY, so that these integrals take the forms h= Voie | fi(l—f')dY and lh = Voriat | (1 — f')dY 0 0 Numerical evaluation of the integrals over Y then gives [ f'(1— f')dY = 0.664 and | (1 — f')dY = 1.73 0 0 if an accurate table of the solution is used. The following calculation was made from Fig. 4.4-3: Y 0 1.0 2.0 3.0 4.0 5.0 6.0 a 0 0.34 0.63 0.844 0.955 0.983 1.00 qa-f') 21. 0.66 0.37 0.156 0.045 0.017 0.000 fi(l— f')0 0.2244 0.2331 =—-0.1309 = 0.0430 = «0.0167 + ~—:0.000 Application of the trapezoidal rule gives the values q, = [1/2 + 0.66 + 0.37 + 0.156 + 0.045 + 0.017 + 0/2] x 1.0 = 1.74 In = [0/2 + 0.2244 + 0.2331 + 0.1309 + 0.043 + 0.0167 + 0/2] x 1.0 = 0.65 which agree, within their uncertainty, with the accepted values 1.73 and 0.664. b. Use of Eq. 1.1-2 and the results of a. in Eq. 4.4-18 gives di; Tyz| y=0 = a + (0)I2 -1/2 = 0.65./pyv3,— 5 = 0.325/ puv3,/x c. The force in the x-direction on a plate of width W and length L, wetted on both sides, according to the result in b, is L F, =aw | Tyz|,_ dr [ roelye L = 20 (0.325/p,08,) | ald 0 = 1.30/ ppv3, LW?; The recommended coefficient is 1.328.4A.6 Use of boundary-layer formulas. The data for this problem are: W=10f L=3ft Voo = 20 ft/s From Table 1.1-2 and Appendix F: vy = (0.1505 cm?/s)/(12 x 2.54 cm/ft)? = 1.62 x 10~* ft? /s # = (0.01813 mPa-s)(10~* Pa/mPa)(6.7197 x 107! Ib»,/ft-s/Pa) = 1.218 x 107° Ib, /ft-s p= p/v = 1.218 x 107°/1.62 x 1074 = 7.5 x 107? Ib, /ft? a. The local Reynolds number at the trailing edge (zc = L = 20 ft) is: Re = Lvao/v = (3ft)(20 ft/s/(1.62 x 1074 ft? /s) = 3.7 x 10° b. According to Eq. 4.4-17, the boundary layer thickness at the trailing edge is L §(L) = 4.64,/7— Voo —4 442 = 464 (1.62 x 10-4 ft?/s)(3 ft) 20 ft/s =2.4x 1075 ft c. According to Eq. 4.4-30, the total drag force of the fluid on both sides of the plate is F, = 1.328./ ppLW?v3, = 1.328)/(0.075 Ibm /ft?)(1.22 x 10-5 Ibm /ft-s)(3 £t)(10 ft)2(20 ft/s) = 0.62 Ibm-ft/s? = 0.019 Iby4A.7 Entrance flow in conduits. a. With the indicated substitutions, Eq. 4.4-17 gives vLe Um ax 1 =D = 4.64 3 6 Setting Umax = 2(v) at the end of the entrance region, we obtain the following estimate of L,: vl. _(_D 2(v) — \ 9.28 2 D*(v) °~ (9.282 vy = 0.023DRe or which is similar to the expression given in §2.3, except that the coefficient is about 2/3 as large. b. At the typical transition locus zv./v ~ 3 x 10° for flow along a flat plate, Eq. 4.4-17 gives 5/2 = 4.64, |—— Vook = 4.64(3.5 x 10°)~1/? = 0.00847 and the transition Reynolds number based on the characteristic length 6 is Voeo8 _ Voot d Vv Vx = (3 x 10°)(0.00847) = 2542 _ For flow in tubes, with transition occurring when 6 = D/2 and with v.. = 2(v), the latter result gives D(v)/v=2542 as the minimum transition Reynolds number, in fair agreement with the reported value of 2100. c. For laminar flow between parallel planes, the method in Problem 4.C gives 6 = B and vmax = (3/2)(vz) at the end of the entrance region. Insertion of these results into Eq. 4.4-17 gives whence = 0.070BRe with Re=B(vz)/v4B.1 Flow of a fluid with a suddenly applied constant wall stress a. Differentiation of Eq. 4.1-1 with respect to y gives 00, _ |, 90, O(_ d,\_ a (_ an, ayat "ay oy at ay aL ay Then using Newton's law of viscosity, we get 2 OT, OT yy ot oy” b. This equation is to be solved with the initial condition that Ty, =0 for t<0, and the boundary conditions that t,, = 7% at y=0, and that 7,, =0 at t= 0. c. The solution is exactly as in Example 4.1-1 with appropriate changes of notation, and the solution is given in Eq. 4B.1-1. d. To get the velocity profile, we integrate Newton's law of viscosity: I. do, ae wly OF 0,= of [1- et bo lay Changing variables we get T, 20 0, = aM Vt aa (h- erfu)du The velocity at y = 0 is then v,(0,t) = 2 Av won, a u\n mp 4-34B.2 Flow near a wall suddenly set in motion (approximate solution) a. Integration of Eq. 4.1-1 over y gives y=eo <° OD, 00, d j= ov, pl, 7 hae =p), aa whence 7b pv,dz= ay | Since the velocity gradient is zero at infinite distance from the plate, we end up with Eq. 4B.2-2. b. We introduce the variable = y/6(t). Then when Eqs. 4B.2-3 and 4 are substituted in Eq. 4B.2-2, we find d 1 ae ee 2 1 5 SUE or-(1- 3n+4n° \dn=-pv..(-3+ $n a0 5 Then after dividing by pv,, and evaluating of the integral, we get 3d. 3. 1 | —— 6(t)==v—— Eq. 4B.2- sd =a) «(See Ea 4B.2-5) c. Eq. 4B.2-5 when integrated gives [dd6=4v['dt or 4(t)=8vE Then the velocity profile is given by 0 17. 3f_¥_),1f_y ) Bz 24-3(_¥_),1/_yY f 6 ( R I> cos @ and 30 R Ny sin Integration of these two equations gives P= (#2\(2) cos@+F(@) and P= (#22) cos @ + G(r) In order for the solution to be unique, F(@) and G(r) must be equal to a constant. If we require that the modified pressure be equal] to py at z= 0 infinitely far from the sphere, we then get 2 P = Po — P&Z -(#\4) cos 8 d. The z-component of the force acting on the sphere is F,=-{0" 5 (6.-[8, -{p5+7}))|_R’sinadede 4-10=~2nR? |" (pcos 6+ 1, cos @- 7,9 sin a sin 6d where Eqs. A.6-28 and 29 have been used for getting the dot products of the unit vectors. Then we use the result of (c) and Eqs. B.1-15 and 18 for the components of the stress tensor (along with Eqs. 4B.3-2 and 3) to get the three contributions to the force: 2 Fip= “aa? — P8Z -(#+\(2) cos o cos @sin 6d0 r=R = $0R°pg + $ muRv,, F,,,, = ~20R* (3(-20 & “cos 0) . sin 6d@ sin 0d@=8 mpRv,, T=R = +42yRv,, [5B -cos! 6) r — oq? ("(— 3 (22) 100 Figg = 2mR Jy ( wld r uP 00 (—sin @)sin 6d9=0 r=R When these are added together, we get Eq. 4B.3-5.4B.4 Use of the vorticity equation a. For the postulate v, = v,(x) the vorticity w=[V xv] has but one component w, =—dv,/dx. Then at steady state, the y- component of Eq. 3D.2-1 is (v-v{ 2s} = vw 2) +(w-V)v, But v, is postulated to be zero so that the last term drops out. Also the first term drops out, because v, and v, are zero, and v, is postulated to have no z-dependence. Consequently the equation simplifies to d°v, /dx* =0. Integration gives =1¢C 2 0, = Cx +C,x+C; or v,/v z,max =D, & +D,€+D; where we have redefined the integration constants. From the three boundary conditions it is found that D, =-1, D, =0, and D, =1, Eq. 4B.4-1 results. Then the z-component of the equation of motion becomes at steady state dt 2 v-V)v. =-——- + uV 0=-—— p( jv, LV'0, or Knowing the velocity distribution we can evaluate the second derivative of the velocity and get the pressure distribution thus dP 2 2UV, max 0=— Ft poema(—Gr) and Py PG2) = ms For z = L, this gives ¥, max =(Py — P,)B”/2uL in agreement with Eq. 2B.3-2. b. Since it is postulated that v,=v,(r), the only nonzero component of of the vorticity vector is w, =—dv,/dr. Then the 0- component of the steady-state vorticity equation is: A-12.(® 2 4% 9 14 =| - "Or roo *dz) ® $(2S (rw )) +o ea +S 2 di, alrar’ 9) ae a 7 00 [we (226 *) 2) +| w,—2+W, ++ |+w,—# or ro@ fr oz This simplifies to df(1d(_dv, A a(i4(ra x)= 0 or =1C7r?+C,Inr+C, This is the same solution found in the soution to Problem 3B.6. When the two boundary conditions and the conservation-of-mass condition are used, we finally get the solution in Eq. 3B.6-2. To get the pressure distribution we use the z-component of the equation of motion, which is 0=- ae. tates ‘) dr Dd dr When the velocity distribution of Eq. 3B.6-2 is inserted, we get dP Au, (1 - x?) + 2K? In(1/K) dr R?(1- x?) (1- x?) - 2(1+ «?)In(1/x) This may then be integrated to get the pressure distribution. A-I34B.5 Steady potential flow around a stationary sphere a. The boundary conditions are: (i) as ra, v— 0,8,, or by using Eqs. A.6-28 and 29 as roo, v, =U,,cos@ and UV, = —v,, sin@ (ii) at r=R, v, =0 (iii) at z=O as ra, PoP, b. Since v=-Vo, we have as ro -0¢/or =0,,cos@ and -(1/r) 09/00 =-v,,sin@ When these equations are integrated we find that @=—v.,rcos6, Thus we feel that ¢ = f(r)cos@ may be an appropriate trial function. c. We next write the 3-dimensional Laplace equation in spherical coordinates (for a system with symmetry about the z-axis) a0 (no a(n os 7) rang sin? so) -0 Into this equation we substitute the trial velocity potential and get 1df{od 24" #) 24 24 f =0 whichasthesolution f=C,r+C,r? since this is an "equidimensional" equation of the form of Eq. C.1-14. d. Application of the boundary conditions then gives f=-0 ali 32) and $=-v, A +2(8) loose e. Then from the components of v=—V@. we get Eqs. 4B.5-2 and 3 by differentiation. f. Then from the equation of motion for steady potential flow p(Vi0")=-VP (see Eq. 4.3-2) By integrating the components of this equation we get 4-14—P=pv2 (8) fo 6 +f3( 8) foe alec, The integration constant is then obtained from the boundary condition given in (a): C,;=-®,-pv2. Then when the modified pressure is evaluated at the surface of the sphere, Eq. 4B.5-4 is obtained directly. 4-154B.6 Potential flow near a stagnation point a. At the origin of coordinates (z = 0) the complex velocity dw/dz = -20 92 is zero, which is a stagnation point. b. By taking the real and imaginary parts of the complex velocity, we get from Eq. 4.3-12: 0, = 29x and v, =—20 gy. c. When 1%, is positive, the fluid is flowing toward the surface y = 0 in the upper half plane. The magnitude of vp specifies the speed with which the fluid is flowing: v = v2 + v; = 2vor. 4-164B.7 Vortex flow a. By using Eq. 4.3-12, we find that dw (1) T(z T( x-iy ; 1 = = —j= 7 3 =—0, +1v, Zz 2nm\z) An\ zz) An x +y in which the overbar indicates the complex conjugate. Equating the real and imaginary parts gives r y T (sin@ r x I. ( cos@\ - %=- Z-| a2 |= Tar ee ae ees | eee leet 2n\ x+y 2n\ r 2n\ x+y 2n\ +r The components in cylindrical coordinates are : : In iGal v, = 0, COSA +V, sind =0; Up =-v,sind +0, cos6= + *) b. The forced vortex is given by v, = Qr in Eq. 3.6-37.4B.8 The flow field around a line source a. For this purely radial flow v, =v,(r) and the other two components are zero. Then V*¢ =0 simplifies at once to Eq. 4B.8-1. b. Integration of Eq. 4B.8-1 gives @=C, nr+C,. Then, since v, =—d@/dr, we find v, =—C,/r. Next we calculate the volumetric flow rate per unit length thus [= [,"0,rd0 = -2nC, whence 2%, = The pressure distribution is then obtained from the radial component of the equation of motion pv. do, __ dP or (=-\(- r )--¢ " dr dr a Qar)\ 2nr2) ~~ dr Integration then gives P, Py pf 1 ef r ) p= a pili — J, dP >| r( =} Ole o (ive al 9 4-184B.9 Checking solutions to unsteady flow problems a. Substituting the solution of Eq. 4.1-14 (or 15) into Eq..4.1-5, we have to verify that S(v- Jeera) Eo- dora) We have to use the chain rule of partial differentiation along with the Leibniz formula alee —raglteren 3) or (Ida) ter cea) When the definition of 7 is used, the above is found to be an identity.4C.1 Laminar entrance flow in a duct. a. Calculation of the mass flow rate according to Eqs. 4C.1-1 and 4C.1-2 gives w = plvc)W “af [2(%) - (4) ay + pnw [body = pveW [6 - ;| + pv W[B — 6] = pu W [-§ + a Equating the first and last expressions for w, we get (with 6 < B and A= 6/B), B 3 ve(x) = (v=) B53 = (v2)3—y The following related equations will prove useful in part b: = () dA oe a de » 6 We = (ve)? BS 9BA dA <8 dz"! GA) de 4 (28) = ® ()¢23) val =n [EBA 2 °b. The boundary layer in this system lies between y = 0 and y = 6, so those limits suffice for the integrals in Eq. 4.4-13. Evaluation of the terms in Eq. 4.4-13 (divided by p) according to the results in @ then gives: ove 2vve _ Sv (vz) ay By bro = ~ BA(3— A) a f° dion [ 2 2 : al Vz (Ve — vz )dy = 5026) | [(2u — u*)(1 — 2u + u*)]|du with u = y/6 = (028) [iv — 4u? + 2u3 — u? 4+ Qu? — u4]du = 020 ~ (4/3) + (2/4) — (1/3) + (2/4) - (1/5)} = Biv)? end. sts) 4-Lodv. f°? dv. f} 2 fae _ =v,6—= _ d Gn [ (ve — vz)dy vb [ [1 —2u + u*]du dv. dz = Ve = ue) B og aa A /3) With these substitutions, Eq. 4.4-13 gives A (3— ap z 6vx Q7+9A 2B (.)B? = Biv,)* Go ase [2/15] + 9(v.)°B Blv,)? dA [54+18A+45A ~ (3— A) dz 15 | _ Biv,)? dA [54+63A ~ (8— A) de 15 | 2/3) Multiplication of both members by (5/3)A(3 — A)/(B(vz)? gives 10 (_”_) _ 6A +747 aA (vz)B? ) ~ (3— A)? dz in agreement with Eq. 4C.1-4. c. Integration of the last equation with the initial condition that A = 0 at az = 0 gives ve 1 46s + 7s? (vz)B? 10fp Gs From integral tables we get the formulas | tie =7 Fi [Ina +05) + - <,|+¢ s? 1 a? | oye B a+ br — 2aln(a + bx) — athe +C ( ) which yield the definite integrals [weap = > C5*) +92] Pye-ls A) sin (24) al 4-2and the solution 10 6 fin (254) + 5S5] 47/0 A) on (*5*) al 3-A 6A 63 =14+48In ( 3 4 ep ta-y _ 3-A 6A — 21(3 — A) — 63 =14+48In ( 3 ) ZL A 3-A 27A = 144480 ( 3 tea in agreement with Eq. 4C.1-5. d. Setting A = 1 and z = £, in the last equation gives 1 27 Le=35 F + 481n(2/3) + 3 = (0.1) [7 — 19.462325 + 13.5] = 0.104 in agreement with answer (d). e. Application of Eq. 4.3-5 to the region 6(x) < y < B, with vy neglected so that v = v, there, gives 1 ahve +P = constant Insertion of the result of part a gives 1 it oa aw gPlve) +P = constant 3-A Evaluating the constant at c = 0, where A = 0 and P = P,, we get 1, of 3 \Y., 1, ote? (s2 5) +P = 5plve)? + Pe P-—Py= 50(02)” f 2 (25) | or A_224C.2 Torsional oscillatory viscometer a. The equation for the rotating bob in a vacuum is just "moment-of-inertia times angular acceleration equals the the sum of the torques." In mathematical terms this is BOR _ _ I dt? kOx in which —k6, is the "restoring torque." The solution to this equation is Op =C, cos, +C, sin |f =C, cos Wot + C, sin Wot This states that the bob is oscillating with a frequency , =./k/I, which is called the natural frequency. b. If there is a fluid in the gap, the equation of motion of the bob must contain both the "restoring torque" and the torque exerted by the fluid in the @ direction on the solid surface that is perpendicular to the r direction: dO I 3 = —kOz —(2RL)(R)(-T,6| 2) Here (27RL)(-T,9| _,) is the force, and R is the lever arm. Next one uses Eq. B.1-11 to replace the shear stress by the appropriate product of viscosity and velocity gradient. This gives Eq. 4C.2-1. c. Equation 4C.2-3 is obtained from simplifying Eq. B.6-5. d. The choice of the variable x is convenient, since x = 0 at the surface of the bob, and x = 1 is the inner surface of the cup. There are many ways to select the other dimensionless variables, but the choices we have made allow the viscosity to appear only in the parameter called M. Equations 4C.2-11 and 12 follow immediately. e. From Eq. 4C.2-14 we get for the bob angular displacement Ox = ROR exp(i@t)} = (08, +108, )(cos@e + isinDr) = OR, COSWT— OR, SINDT 4-2Here 6p, and 63; are the real and imaginary parts of the complex amplitude 0g. The angular displacement of the bob can also be written in terms of an amplitude and a phase shift: Oz = Acos(@t - a)== Acos@tcosa — Asinotsina When the two expressions for the bob displacement are equated, we find that Acos@t = 62, and Asin @t = 02; from which we get and tana = OR 3198] Oe K{ogt A= ox +08 =|03 for the amplitude and the phase angle respectively. The ratio of the amplitudes of the cup and bob is then d’° dx? dg° dx x=0 f. ine? =M and (ia@)’ 62 =-02 +M g. The differential equation for $° is then © - [2 = 0 with 9°(0) = Adzi@ and 6° (1) = AO 10 The solution to this equation with the boundary conditions is then 6 i@ \sinh.jio@/Mx g° = Ait cosh | + Aral 03 co oh eT Differentiation of this with respect to x gives Eq. 4C.2-16. h. From Eqs. 4C.2-12, 14, and 16 it follows that t==\2 0 ° MAio LOC ° io (ia) 08 =-0 + * Ginh fia VM Bl om - ozcosh || This equation may be solved for 0g, and Eq. 4C.2-17 results.i. When the hyperbolic functions are expanded in Taylor series and the terms arranged in powers of 1/M, we get Gre — cosh ia , (1- 0") sinh Jio/M OR M MAi® io/M ‘[ealteal) «Saas M i |(1-07) o@ 1-o’)o GF _,-i[0-#) a], fo-3o oe], M| A@ 2!| M 3!A@ 4! From this, Eq. 4C.2-18 follows. The results of part (e) can then be used to get the amplitude and phase angle. 4-254C.3 Darcy’s equation for flow through porous media. (a) In Case 1, the equation of state is p = po. Hence, Eq. (4C.3-1) gives (V ev) =0 Insertion of Eq. (4C.3-2) for vo, with p = pp and g = —V4, gives (v °[Vp+ poV 4) =0 or V’?P =0 in which P = p + po® is the “modified pressure” defined in Chapters 2 and 3 for systems with constant p. (b) In Case 2, the equation of state is p = poe®?. Hence, eo B Expressing the divergence term of Eq. (4C.3-1) via Darcy’s equation, we get Vp = pobe’? Vp = pBVp, so that pVp = —Vp ~(V # pv) = +7 (V ol¥p — pal) al¥ ° [pVp— p’g]) Il “(5V%e ~ (V0 pg) Inserting this result, multiplied by 4f/«, into the smoothed continuity equation, we get EHB Op _ ya, _ 2 mE TY PAY 9 9) (c) In Case 3, the equation of state is p = pop. Hence, Vp = poVp and Vp = (1/p0)Vp. Then Eqs. (4C.3-1,2), with the pg term neglected, give Op _ K ea = (Ve el Vel) K = +(V e p—V ( * Po p) K =—(VepV apo! ° pVp) K 1 —(V e V(=p? po! °V(5e") K = v2 2 2upo ” ALGHence, 2eppo Op _ V2 92 Kk Ot (d) In Case 4, the equation of state is p = pop™, giving p = (p/po)!/™. Thus, 1 _ . Vp = — py /™ pV p m whence kK -1 m pv =—-——py pV p pm so that Kay —(V © pvo) = ange Ewe p!l™Yp) En ™ _Zo(m+i)/m um? (Ve Fear p With this result, Eq. 4C-1 gives (m+ leupy!™ \ Op = V2p(m+t1)/m K Ot 4-24C4 Radial flow through a porous medium a. For the radial flow (in cylindrical coordinates) of an incompressible fluid Eqs. 4C.3-1 and 2 become 1d «dP =~ = (ror) and Vor dr Integrating the first of these equations gives vp, = C,/r. Substituting this into the second equation and integrating gives C,Inr+C, aaa or D,lnr+D,=-P where new constants of integration have been introduced. These constants are determined from the boundary conditions: B.C.1: D,InR,+D,=-P, B.C.2: D,InR, +D, =-P, When the integration constants have been determined, the pressure distribution is found to be P-P, _ In@/R) P,-P, ImM(R,/R,) b. The velocity distribution is then given by using Darcy's law 0, = - X40 -_K(p,-p,)—1 _} « (P,-P) =-= P a Mdr wp In(R,/R\)r ur In(R,/R;) c. The mass rate of flow through the system is 2nkh(P, -P,)p Ww = plvo,(R,)|2aR,h= ain(R,]R,) 4-284D.1. Flow near an oscillating wall The problem is to solve Eq. 4.1-1, with the initial condition that v, =0 at f=0, and the boundary conditions that v, =0 at y=oo and v, =v) cos@t at y=0. When we take the Laplace transform of Eq. 4.1-1 and the boundary and initial conditions, we get 2=- 0, dy? po,=Vv with 0, = 09 pf og at y=0 and 0, =Oat y= This ordinary differential equation is easily solved with the boundary conditions to give B, = pz exp(./p/vy) pe +o’ This may be inverted by using the convolution theorem, or else by consulting a table of transforms (see, for example, Formula #11, on p. 246 of Vol. 1 of A. Erdélyi, et al., Table of Integral Transforms, McGraw-Hill, New York (1954)). The use of the table leads directly to Eq. 4D.1-1. 4-294D.2 Start-up of laminar flow in a circular tube a. The partial differential equation, initial condition, and boundary conditions are a, _Po-P, 13 {Bs} Oot L rar’ ar with v,(r,0)=0, v,(0,t) = finite, and v,(R,t)=0. We now introduce the following dimensionless variables ¢ = v, . é = r bt (®,-P,)R? /4yL’ R pR? Then Eq. 4D.2-1 is obtained, along with the initial and boundary conditions: (€,0)=0, ¢(0,7) = finite, and ¢(R,t)=0. b. The asymptotic solution is obtained by setting the time derivative equal to zero and solving the ordinary differential equation with the boundary conditions. Then the partial differential equation for ¢,(&,T) is 24, .1.2(,98) dt EDEL? a with ancillary conditions: 9,(€,0)=9¢.(&), ,(0,t)= finite, and ¢,(1,7)=0. We now try a solution of the form ¢,(&,7) = =(€)T(r). This leads to two ordinary differential equations dT 2 1d({¥ dé pa —=-a’°T d ——| E= E= de a an aaa which have as their solutions T = Cy exp(-a?z) and B=C,J,(aé)+C,Yy(aé) in which the Cs are constants, and J, and Y, are zero-order Bessel functions. Since Yq is not finite at the tube axis, we must set C, equal to zero. Since & must be zero at the tube wall, this will occur only if 4-20J)(@) =0.This will happen only at a@,,@,,@3,---, that is, at the zeroes of Jy. Hence there are many solutions to the E-equation that will satisfy the boundary conditions: =, =C,,Jo(a@,¢). Therefore, the general solution to the partial differential equation must be 0.= DB, exp(—0727)Jo(0,€) The constants B, are to be determined from the initial condition, (1-£7)= EB, fo(a8) This is done by multiplying both sides of the last equation by Jo(@,&)€ and integrating from 0 to 1: fJo(Omé)E(1- E2 Me = 3B, To OnE Jo (mE )E4E Because of the orthogonality properties of the Bessel functions, the only term on the right side that contributes is the term for m =n. The integrals may then be evaluated using some standard relations for Bessel functions. This gives a m OnJ (a) and this leads directly to Eq. 4D.2-2. 4-314D.3 Flows in the disk-and-tube system a. If the tangential component of the velocity depends only on r and z, then the equation of motion simplifies to d(1a O05 _ (23400 ve) }# a2 We try a solution of the product form: v,(r,@)= f(r)g(z). When this is substituted into the partial differential equation above, and the resulting equation divided by f(r)g(z) we get aed) ees The left side is a function of r only and the right side a function of z, only and therefore both sides must equal a constant; we call the constant -c?. Then we have two ordinary differential equations to solve: “(22 (p))+%F=0 and Hg _ oa 0 dr\r dr daze ° 8 The second equation has the solution g = Acoshcz + Bsinhcz, where A and B are constants. The requirement that the velocity be zero at z = L, gives - sinh[cL(1-[z / L])] coshcL where B is a constant. The first equation can also be written as ee (e-alre which is now recognized as a Bessel equation of first order, with the solution f = MJ,(cr)+NY,(cr), where M and N are constants. But since Y, becomes infinite at r = 0, only the J, term is needed, Since f = 4-320 at at r = R, we must have J,(c,,)=0. There are infinitely many c,, that satisfy this equation. Hence the solution is of the form sinhc,L(1— ¢) v—(E,0) = Bale) coshc,,L The coefficients B, are determined from the initial condition RQE = YB, (Cn8) n=1 This is done by multiplying the equation by J,(c,,¢é)€ and integrating over € from 0 to 1 and making use of the orthogonality relation. This gives ROP Ti (Cn€)E2dE = DB, [oJuCnEIi(CmE)E4E or Io(Cm) _ pp 4 2 ___2RQ RQ Cn =B,, 4 [J (Cm)] or B, CJ (Cm) Thus the steady-state velocity distribution is n= =1CnJ2 Cala (Ca) ) coshe,L A-334D.4 Unsteady annular flows For the tangential annular flow (part (b)), the equation of motion for v,(r,t) is pee =n 8 [22 (m0 ] ot ortror’ ° with v,(7,0) =0, 09(KR,t) = KRQ;, U.(R,t) = RQ,. We introduce these dimensionless variables: pat 7 o= tr V6 Q; Re DR R(Q,-9,) ~* Q,-9, co 1 Then the partial differential equation for $(€,7) becomes a _ djl od ar aE $e) with 6(€,0)=0, 6(«,t)=—Ka, o(1,t)=1-a. For the steady state (i.e., at infinite time), we have the solution 1-a(1- x?) Ke )1_ 1 ¢.-[ FS (eas Hence the time-dependent solution is 0(E,7)= $..(€)- ,(§,7)- Here ¢,(&,t) is the transient contribution, which satisfies the partial differential equation for 9(€,t) , but with boundary and initial conditions: ¢,(«,7) = ¢,(1,7)=0 and ¢,(E,0)= ¢..(€) Application of the method of separation of variables with 9,(€,7) = f(E)g(t) gives 434ldg_ yp _ld(id gat pal eae) where -b is the separation constant. Thus we arrive at two ordinary differential equations which can be solved (the second- order equation is a Bessel function ). Since this is a Sturm-Liouville equation, the complete solution is a sum of eigenfunctions multiplied by the appropriate exponential function: o,(E, t) a SCZ, (by6)e in which Z,(b,&) is the superposition of the two solutions to the Bessel equation: Zy (0,6) = J1(b,E)Y1 (b,«) - Ti(b,«)Y4(b,€) These functions satisfy the boundary conditions in that Z,(b,)=0 at § =k. The conditions that Z,(b,é)=0 at€=1 determine the eigenvalues b,. The C, can then be determined from the initial condition, which is 0.(8)= YC,2a(bé) When this equation is multiplied by Z,(b,,¢)édé and then integrated over the domain of interest, we get " §i[Z,(b,8)) 646 - 4[Z5(b,)- 23 (b,«)] The integrals are performed by making use of a mathematical handbook. The expression above may be simplified by using 4-35i. The expressions for A and B ii. The defining equation for the b,, iii. The relation Jy(x)Y;(x)— J, (x)Yo(x) =—2/ax Then it may be shown that 2 J,(b,«) 21 Z = =e — 0(b,) ab, Tia) and Zo (b, Kk) ar and also J ,(b, )[(1- @)J,(b,«) + oxxJ,(b,, )| Lie.) -Jrb,)] in which @ is the dimensionless angular velocity. Then the complete expression for the transient behavior in tangential annular flow of a Newtonian fluid is C,,(, at) = -2¥ I1(b, \[a- O)J,(b,«) + ax] (by )|Z1(bn5) oe n=1 [J?(b,«)- J? (2, )} Complete tables are given in the original reference. as well as some typical velocity profiles. Also, a Laplace-transform solution is given for small times, for which the expression above converges too slowly to be of much value. 4-364D.5 Stream functions for three-dimensional flow. (a) The divergence of the curl of A is >,(0/0z;)[V x A]; which gives, according to Eq. A.4-10, 2(6/82%) X 2 e04(0/08))As = = » u dfs aa ae 0? Ax OPAL = de eit leas Oz; Ox,;Oz; =0 i phi Thus, the mass flux function pu = [VxA] satisfies the continuity equation for steady flow (or for unsteady flow with constant density). Here use has been made of the relations €;;, = —ejix and €jik = €ixk = 0 contained in Eq. A.2-15. The divergence of the product expression for pv is (V e [(Vdi)x(Vib2)]) = » (sit [Dag 5; Ova xr oa = -ErDe (5; x6) (28 ae) dy. a = LL LE © €jni8i) 5 — 2 (5 a) => SFresalOotaSa) t J k = » > So cain (Distr Oebe + O;1 0;n%h2) 7 J k = €123(012%1 O32 + O21 01342) + €231(023h1 01 th2 + O3%h1 O21 2) + €312(031 1 O22 + 81410322) + €132(O13h1 Oop2 + O31 O12%2) + €321(0321 01 2 + O24) O31 2) + €213(O211 932 + O11 O23h2) = O12 03tho(E123 + E213) + O21 O13 ho(E123 + €321) + 023101 Wo (E231 + €321) + 0341 O21 2(€o31 + €132) + 03141 O2b2(€312 + €132) + O11 Osao(E312 + €213) =0 Thus, the product expression for pu is divergenceless. Here Eq. A.2-15 and the symmetry relations 0;; = 0;; have been used. Stream functions of this form have been used by several authors; see footnote 1 of §4.2. 4-3](b) In each of the coordinate systems shown in Table 4.2-1, the two nontrivial velocity components for incompressible flow are the corresponding components of curl(—63A3)/p = curl(—637)/h3), as may be verified by use of Eqs. G, H and I in Tables A.7-1,2,3. Thus, for Cartesian coordinates with v, = 0 and no z-dependence, hg = hz, = 1, and Eq. A.7-18 (with v read as 534)/h3) gives the velocity components vz = curl,(—A)/p = = and vy = curly(—A)/p = + For cylindrical coordinates with v, = 0 and no z-dependence, hg = hz, = 1 and Kq. A.7-18 (with v read as 637)/h3) gives the velocity components 10 a v, = curl,(—A)/p = 1 and vg = curlg(—A)/p = + For cylindrical coordinates with vg = 0 and no 6-dependence, hg = he = r and Eq. A.7-18 (with v read as 63y/h3) gives the velocity components i _ 10% _ _ 10% v, =curl,(—A)/p = or and vy = curl,(—A)/p = ae az For spherical coordinates with vg = 0 and no ¢-dependence, h3 = hg = rsin@ and Eq. A.7-18 (with v read as 537)/h3) gives the velocity components 1 ap _ _ 1 op rand ag MA te = cule ANP = TG Be v, = curl,(—A)/p = — (c) Consider two surfaces, 71 (x1, 22,23) = Cy and 2(x1,22,t3) = C2, which intersect along a line £. At each point on L, the vectors Vy; and Vy2 are normal to both surfaces, and the velocity vector v = [(Vi1)x(Vw2)] is consequently tangent to £. Thus, the intersection of any such pair of surfaces is a streamline. In Fig. 4.3-1, we may choose 7 = U(r, 6) and 72 = z; the resulting streamlines in a plane of constant #2 are shown in the figure. (d) Read v in Eq. A.5-4 as the vector A whose curl is the local mass flux pv. Then the net mass flow through S is [im [Vx A])dS = fee poyac for steady flow, or for unsteady incompressible flow. A no-slip condition v = 0 on C requires [Vx A] = 0 there, but this derivative condition does not require the vanishing of A nor of the net mass flow. 4-385A.1 Pressure drop needed for Jaminar-turbulent transition. The minimum value of Re= 4w/(mDy) needed to produce turbulent flow in a long, smooth tube is about 2100. Poiseuille’s law, Eq. 2.3-21, holds until this critical Re value, giving = 4 = m(Po — Pr)Rip for Re: bR U: Vo A =y 2Lp 4 ) v, where ¥,, = +/(Py — P,)R/2Lp. Then the velocity distribution in the two regions may be obtained from Eq. 5.3-3: Region <: 2 = Ll Ce, Pr fal we) sas * kK Vv Ke Region >: = Sl BO a re Sel ux O, K Vv “ie y being defined as y =1r-—aR in Region < and y = R—r in Region >. b. The continuity of the velocity profile at r=bR gives a relation between A< and J”:—# i = in| 82 )a-oyme [oa zl Jee ~a)° K v c. The mass rate of flow through the annulus is w= 2npf,,Dsrdr + 2npf,,D2rdr The first integral is: 2npv< fal Sa )* ahi < f(b-a)R 1 yu, < =2 — In| ~~ ]4+a R)di mpv; |, [2 ( }. Juve )dy 2 < = 2moo:{ ie Minx + ale + oRey Jas v; K v : 2 2 - 2npo:{ | [2 Gee) 2 (Se Joeins- x) U; K*\ 2 4 K < (b-a)(Ros/v) gt gc? a(t | 2 v 7 1 2 Rvs = 2R’pvs| —|(b- 2a(b- - mi po:| ( a)? +2a(b—a)}in(b af e) +as[(b -a) +2a(b- a)|- [500 -a) +2a(b- ‘| KpRk| 1 yo. ' 270v; bre | m( =] +A J 5-3= 2npv7 {i rol (2 2 [en 2 > = 2noer( *] oer Em + x (Ae - x) v K v 2 > 2 2 - 2npe:| 2) l 1 Roy («inx-x)--{ Sinx-2] U; Kv Vv Kk" \ 2 4 > (1-b)( Re? /v) (BE ear de? v 2 0 = mR? pv? EE —b)In(1- H( SE) -(1- »] K v 1 2 Roz) 1, _ | 0-vfina-of 222) 20-0] +24 (1-b)-4?(1-b)°| = ak?onz(1- 2) Sanco Ss +a - 4(3+ i) We now combine the two integrals and use the result in (b) to eliminate A“ in favor of A”; furthermore, we use the result in (a) to introduce v,,. This leads to: W = TR PV xq {a[ ssun( B= Row (py 1-F } + r|- a| in which A=a¥?(b?—-a?)” +(1-B2)” Baaie(ot— a?) LF 4 tle (my? S(F4 | The expression for A does not agree with that of Meter and Bird, and we conclude that their expression is incorrect.5C.3 Instability in a simple mechanical system a. The centrifugal force acting on the mass is mQ?r= mQ?Lsin@. The gravitational force acting downward is mg. These forces must have a resultant in the direction of the rod, and therefore _ mQ?Lsin@ mg or cos @ = —S— Q*L tan@ If the angular velocity goes to zero, it would appear that cos @ would go to infinity! b. However, this formula describes the relation among the various quantities when the system is rotating and 820. When as Q decreases, the right side of the equation attains a value of 1 when Q=Q,,,, and then cos@=1 and @ is zero. For for value of Q less than its "threshold value", the value of @ remains at 0. When one starts up the system from rest, @ will always be zero. However, if Q>Q,4, and there is any disturbance on the system, then the system will move up to the stable curve (given by the equation in part (a)). It has to be understood that the graph we have given is only for the steady state, and that to understand the system fully, it is necessary to examine the full unsteady-state equation. c. According to p. 12 of L. D. Landau and E. M. Lifshitz, Mechanics (Pergamon, 1990) the Lagrangian for this problem is L=}mL? (6? +Q? sin? @) + mgLcos@ Then Lagrange's equation of motion a dta@ 00 gives for the system we are considering the following equation of motion ae 21 ; mL = mQ*Lsin @cos 6 — mgsin @ 5-\5d. Consider the lower (unstable) branch in the diagram. For very small perturbations (6,) to the steady state (0)), we have then sin @ = sin 0, = 0, and cos @ = 1. Then the equation of motion becomes d’0, 2 s) =(Q?-2]@ dt? ( L} Now we try a small perturbation of the form 6, = AR{e™} When we substitute this function into the differential equation we get @, = tijQ? -(g/L). If Q? < g/L, then both roots @, are real and 6, oscillates. If Q?>g/L, w, is positive imaginary and e™ will increase indefinitely with time. Hence the branch @ = 0 is unstable with respect to infinitestimal disturbances. e. Next we consider the upper branch for which cos 0) = ¢/Q7L and sin@ =/1-(g/Q°L)?. Then the equation of motion becomes 2 mL oo = mQ?L(sin @, + @, cos Oy )(cos Oy — 0, sin Bp) — mg(sin 6, + , cos 8) =—mQ?L6, sin” 0, where we have neglected terms quadratic in 6,. Hence the equation of motion becomes 2 FO +E (2? +(g/L)(2? ~(a/L))0, We now try a solution of the form 6, = AR{e} and get O,=45\(2? +(g/L))(2? -(g/L)) For the upper branch, Q? > (g/L), and hence both quantities @, are real. Hence the system is stable to small perturbations. 5-i6UNSTABLE Quan. = d9/t 5-"]5D.1 Derivation of the equations of change for the Reynolds stress Multiplication of the ith component of Eq. 5.25 by vj and time-smoothing gives for constant p gives (in the Cartesian tensor notation of §A.9, with the Einstein summation convention and with the shorthand notation 0, = 0/dt): PU;0,0; = —v;0;p ’ 0 0/9,5,0; + U;0,0;,0; + 010),0;0;) + HV;O,9,0; (1) Q) (3) (4) (5) (6) Then we write the same equation with i and j interchanged. When the two equations are added we get, term by term: (1) (00,0; + 070,07) = pa, (0707 + 0707) -p(0/0,07 + 070,07) The second term on the right side is the negative of the left side. The first term on the right side can also be written as 20,0707 ;- Therefore the left side is just pd, v0, (2) -(ofa;p"+ ofp’) (3) -p(09,8,.07 + 079,0,07) = - (070,00; + 075 ,0,07) (used 5.2-10) (4) -o of, of, + 179,075, )= -p(07070,5, + 0/0,0,5, ) (used 5.2-11) =p ata -v aay v; ‘)- p(2 10,004 + 050, 9,07) Note that the third term is zero by Eq. 5.2-11, and that the second and fourth terms cancel giving - (9,0; 050; ) (©) + (010 ,0,0; + 079,0,07) Combining the above gives:pd, 070" = -(00,p" + va;P’) -p(5,9,0704) - 9 09,5, + 070, 9,0;) bvivy —p(9,0;0707) +u(0;0,0,0; + V50,9,0; | or pd, 0:0; +0(5,9,0707) = -9(070,9,5, + U',O,0; ) —p( 9.040707) ~(0/9p” + 0,0;p’) +p (079,00; + O70, 9,07 | The two terms on the left side are the substantial derivative term in Eq. 5D.1-1. The remainder of the terms are set out in the same order as in Eq. 5D.1-1. B145D.2_ Kinetic energy of turbulence Taking the trace of Eq. 5D.1-1, we get 2 w= -2p(v’v’:Vv) - o(V -(v’ ‘v’)v’) Dt -2(v’ . Vp’) + 2u(v’ . V°v’) Modify the third term on the right as follows: 20 Wp) =A) + 207 W) The last term on the right is zero according to Eq. 5.2-11. Then divide the first equation above by 2 to get DO Theos (1 an’2\y" pit? > =-p(v’v :wv)-(V-(}p0 \v ) -(V-p’v’)+ u(v’ Vv‘) which is Eq. 5D.2-1. For an interpretation of this equation, see pp. 63 et seq. in the book by Tennekes and Lumley cited on p. 176. 5206A.1 Pressure drop required for a pipe with fittings. The average velocity at the given conditions is (v) = 4(w/p) _ (4)(197 m*/s) aD? ()(0.25 m)? = 40.1 m/s and the Reynolds number is Die) © (0.25 m(40.1 m/s)/(1.0037 x 107® m2/s) = 9.99 x 10° Vv Thus, the flow is turbulent, and Fig. 6.2-1 gives f=0.0020 for hydraulically smooth pipe. The total equivalent length of the pipe and fittings is L,. = 1234 m of pipe + (4)(32)(0.25) m equivalent for 4 90° elbows + (2)(15)(0, 25) m equivalent for 2 90° elbows = 1274.5 m The required pressure drop, according to Eq. 6.1-4 with L replaced by Le, is then L. (Po — pt) = 25° p(v)’ f _ 512735 m : ; 298m m (998 kg/m*)(40.1 m/s)*(0.0020) = 3.3.x 10" Pa= 4.7 x 10° psi 6-|6A.2 Pressure difference required for flow in pipe with elevation change. In this problem the pipe diameter is (3.068 in.)/(39.37 in./m) = 0.07793 m, the mass flow rate is w = (18/60 gal/s)(3.7853 lit /gal)(0.9982 kg/lit) = 1.13 kg/s, the average velocity is 4w (4)(1.13 kg/s) (>) = pz, (0.07793 m)?(998.2 kg/m®) = 0.237 m/s, and the Reynolds number is D(v)p _ (0.07793 m)(0.237 m/s)(998.2 kg/m?) be (1.002 x 10-3 kg/m-s) Re 1.84 x 104 From Fig. 6.2-2 we find that for this Re value, f = 0.0066 for smooth tubes. Hence, Le Po ~ PL = —pg(ho — hr) + 2 pv)? fF = —(998.2 kg/m*)(9.807 m/s?)((—50 x 0.3048//2) m) + 2[(50 x 12/3.068) + (2 x 15)] - (998.2 kg/m?)(0.237 m/s)?(0.0066) = 1.055 x 10° Pa + 167 Pa = 1.055 x 10° Pa = 15.3 psi The corresponding calculation in terms of lb, ft and s, neglecting the small friction term, is Po ~ PL = (62.4-x 0.9982 Ibm /ft?)(32.174 ft/s?)((50/V2) ft) = 7.09 x 10* Ibm /ft/s? = 15.3 psia6A.3 Flow rate for a given pressure drop. The quantities needed for this calculation are as follows, in units of lbm, ft, and s: Po — pL = (0.25 Ib¢/in?)(144 in? /ft?)(32.174 lbmft/s?-Ib+) = 1.16 x 10° Ibm /ft-s?; D=05f; p= 62.4 Ibm /ft?; L = 1320 ft; uw =6.73 x 10-* lb», /ft-s Hence, the solution must lie on the locus _ Dp |(po —pr)D Ref = 7 Ip _ (0.5)(62.4) /(1.15 x 10-3)(0.5) ~ 6.73 x 10-4 2(1320)(62.4) = (4.64 x 10*)./3.52 x 10-3 = 2.74 x 10° Method B: The last equation gives a straight line on the logarithmic plot of f vs. Re, passing through f = 1 at Re= 2.74 x 10° and through f = 0.01 at Re= 2.74 x 10+, and intersecting the f curve for smooth tubes at Re= 3.6 x 10+. Hence, the average velocity is _. Re _ 3.6.x 104 () = p/p) ~ 464 x 108 and the volume rate of flow is = 0.78 ft/s aD? q=* 0) _ (3.1416)(0.5%) (0.78) = 0.152 ft? /s = 68 U.S. gal/hr Method A gives the same result if the plot of f vs. Re\/f is accurately drawn.6A.4 Motion of a sphere in a liquid. The force of gravity on the sphere is Foray = mg = (0.0500 g)(980.665 cm/s”) = 49.03 dynes The buoyant force of the fluid on the sphere is Fhuoy = (4/3)7R® pg = (4/3)7(0.25 cm)*(0.900 g/cm*)(980.665 cm/s”) = 57.77 dynes a. The resultant upward force is Fhuoy — Fgrav = 57.77 — 49.03 = 8.74 dynes and is balanced, at steady state, by an equal and opposite drag force Fy = 8.74 dynes. b. The friction factor is defined by 1 Fi = (aR?)(2)p02)f Thus, for this system, _ F, _ 8F ~ (FR) (F0v%) 7D? v3, 8(8.74 dynes) ~~ (0.500 cm)?(0.900 g/cm*)(0.500 cm/s)? = 3.95 x 10? f c. From Fig. 6.3-1 we see that f is very close to its creeping-flow asymptote, 24/Re. To the same approximation, D —TeoP ~ 24/f = 0.061 Hence, Doveopf — (0.5 em)(0.5 em/s)(0.9¢/cm?)(3:95 x 10?) 24 24 = 3.7 g/em-s = 3.7 x 10” ep 6-46A.5 Sphere diameter for a given terminal velocity. a. Method A: Replot the f-curve of Fig. 6.3-1 as f/Re (which does not contain D) vs. Re. Then from this curve we can find the value of Re for any calculated value of f/Re, and determine D as Rep /pvco >. ; a. Method B: On the log-log plot of f = f(Re), plot also the locus f = (f/Re)Re, which will be a line of slope 1, and find the desired Re at the intersection of the two loci. This method avoids any need to prepare an auxiliary plot. b. The data of Problem 2A.4 give Voo = (1 ft/s)(12 x 2.54 cm/ft) = 30.48 cm/s p = 0.045 Ib,» /ft3)(453.59 g/Ib,,)(12 x 2.54 em/ft)~* = 7.2 x 1074 g/em=3 Ps = 1.2 g/cm* p= 2.6 x 107* g/ems g = 980.7 cm/s” from which we calculate f/ Re = 49 (24) 3 pvz, \ Pp 4 (980.7)(0.00026) / 1.2 — 7.2 x 10-* ~ 3 1.203, ( 7.2 x 10-4 ) = 27.8 We therefore draw a line of slope 1 through f = 27.8, Re= 1 on Fig. 6.3-1. This line intersects the f vs. Re curve at Re= 0.95. The particle diameter is then calculated as Rep (0.95)(2.6 x 10~*) fe : D=— = = (0. 2 m) = Voop ~ (30.48)(7.2 x 10-4) (0.0112 cm)(10* microns/cm) = 112 microns c. Here voy is 10 times larger, giving f /Re= 27.8 x 107°. This locus intersects the f(Re) curve at Re= 75. Hence, at this gas velocity the diameter of the largest particle that can be lost is (75)(2.6 x 10~*) = Go.4s72 x 10-4 = (0.89 cm)(10* microns/cm) = 0.89 x 10* microns 6-96A.6 Estimation of void fraction of a packed column. The superficial velocity is (244 lb/min)(1 min/60 s)(453.59 g/Ib) ~ (1.2865 g/em*)(146 in?)(2.54 cm/in)? vo 1.522 cm/s According to Eq. 6.4-9 (the Blake-Kozeny equation, developed for laminar flows), é 150pLvo (I—e)? D2 Ap 150(0.565 g/cm-s)(73 x 2.54 cm)(1.522 cm/s) ~ (0.2 em)?(158 psi)(68947 dynes/cm?-psi) = 0.0549 Solving this equation for ¢, we find e = 0.30; hence, D,vop 1 (0.2 cm)(1.522 cm/s)(1.287 g/cm*) 1: 1.00 w 1l-e (0.565 g/cm:s) (1—0.30) which indicates (see Fig. 6.4-2) that it was appropriate to use the Blake-Kozeny equation here.6A.7 Estimation of pressure drops in annular flow. a, The definition of f in Eq. 6.1-1 gives Fr mR?(1—?)(Po— Pr) _ R(1—«)(Po — Pr) f= SURE + w)p(es)2/2 ~ InRLG 4+ n)p(Uy)?/2 ~~ ‘Lp(a, Equations 6A.7-1 and 3 give os ~ f DA —«)(o.)p Insertion of the previous result for f gives K 16Lp(vz)? ig 8Lpu(d:) ®= RO —n)\(Po— Pi) DO =n)Ga)p RAL— we )(Po— Pr) Equation 2.4-16 gives (v2) -{ (| -5 (Po—Pr) 8uL | 1—K? = In(1/«) Combining the last two results, we get the relation 1 ewe | K-iT 2 1—K? In(1/«) needed to make Eqs. 6A.7-1 and 3 consistent with Eq. 2.4-16 for laminar flow. The K values given by this formula are in excellent agreement with those tabulated on page 194 and recommended for turbulent flow as well. b. The data for this operation are pf = 1.139 cp(=mPa:s) = 7.66 x 107* Ib,,/ft-s; p = 62.2 Ibm /ft?; D = 15 in = 1.25 ft; « = 6 in/15 in = 0.4; G = 3.801; H=0.131; ‘K =0.6759; w/p = 1500 ft? /s; (:) 4(w/p) 4(1600 ft? /s) = ft/s; aD? — (nD) ~ a[1.25? — 0.57 fy] ~ 888 Bs Re, = KD Ne —_ 3 o.e 75g (i258 £)(1 — 0.4)(988 ft/s)(62.3 bm/f?) _ 5g agt (7.66 x 10-4) Equation 6A.7-2 then gives: 1 . —s = 3.801 log,9(1.6 x 10" \/f) — 0.131 VF B10( Vf) (1

R. Byron Bird, Warren E. Stewart, Edwin N. Lightfoot-SOLUTION - Transport Phenomena, 2nd Edition-Wiley (2001)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

R. Byron Bird, Warren E. Stewart, Edwin N. Lightfoot-SOLUTION - Transport Phenomena, 2nd Edition-Wiley (2001)

Uploaded by

Copyright:

Available Formats

You might also like