Lecture 7 Contents • Dynamic Similarity and Model Testing (Continued) • Drag on Bodies in Flows with Free Surfaces • Ship and Hydraulic Model Testing What Did We Do In Last Lecture? 1 page for Drag as function of 2 variables (e.g. velocity and diameter) d increases from curve to curve 1 page for each value of ρ 1 book for Drag as function of 2 variables (e.g. velocity, diameter, density) Shelf of books for Drag as a function of 4 variables (velocity, diameter, density, viscosity) If we want 10 data points per curve, at £10 each experiment, this will cost... 10 × 10 × 10 × 10 × £10 = £100,000 THERE MUST BE A BETTER WAY !?!? What Did We Do In Last Lecture? α β D ∝ V ⋅ d ⋅µ ML T −2 = (L T ) −1 α (L ) β γ (ML −1 T STOKES’ LAW Recall, only valid for low Re! D = const × V d µ DRAG COEFFICIENT CD Drag CD = 2 ρ 2 ⎛d ⎞ V∞ π ⎜ ⎟ 2 ⎝2⎠ ) −1 γ What Did We Do In Last Lecture? For a body in a homogeneous fluid: CD(Re) CD Re = Vd ν • Model Testing requires… • Dynamic Similarity Need… (CD)Model=(CD)Prototype and hence need … ReModel=RePrototype Dynamic Similarity and Model Testing What are the required conditions for model experiments? • Want to test scale model of a prototype in wind tunnel. • Question: Why would any of the measured model data be of any LAST SLIDE FROM PREVIOUS LECTURE relevance to the prototype? After all they are of different size! • So, what conditions need to be satisfied such that wind tunnel data tell us anything about the prototype? One Needs: •Geometric Similarity Model has the same shape as prototype •Kinematic Similarity Additionally it is required that ... Fluid Velocity For Model at corresponding points = Const. Fluid Velocity For Prototype •Dynamic Similarity Additionally to Geo. Sim. and Kin. Sim. it is required that: Non.-dimensional forces are the same for model and prototype, e.g. ... (C D )m = (C D ) p Know that for body in homogeneous fluid drag coefficient is function of Re number o-n-l-y. Hence, Dyn. Sim. only possible if (Re )m = (Re ) p Dyn, Sim. Hard to achieve! Example: We want to test a one-fifth scale model in a wind tunnel under dynamically similar conditions. • Prototype Aircraft: V = 100 m (≈ 0.3 × Speed of Sound ) s Lp 1 • Model Aircraft: Scale → Lm = 5 5 • Assume same air properties for model and prototype Dynamic Similarity Requires: (Re )m = (Re ) p Vm = Vm Lm ν = Vp Lp ν Lp V p = 5 ×100 m = 500 m Lm s s = SUPERSONIC !!! But at supersonic flow speed whole fluid dynamics are different. One gets shocks etc. Will learn about this next year! Hence, in order to achieve dynamic similarity without encountering this problem ... • … one needs a huge wind tunnel (such that one can make VmLm large to get large Rem) • … which can be pressurized (such that one can make νkkkkkk = µ ρ small, which also increases Rem) Photos of Wind Tunnels and Models in Wind Tunnels Continued... The world’s largest wind tunnel ( a 10-storey building) at NASA Ames ... … and a not-so-large wind tunnel. Drag on Bodies in Flows with Free Surfaces • In all previous sections when we considered drag on body assumption was we are dealing with bodies in homogeneous fluid. • Now we will look at the more general case when we have a free surface, as for instance, illustrated in the following sketch... Air V Free Surface L Air Water Air Now we cannot neglect body forces (gravity). Consequently have to assume ... C D = F (V , d , µ , ρ , g ) 1424 3{ As Before New • Following arguments in section ‘Bodies Moving in Homogeneous Fluids’ the drag coefficient, CD, now depends on two dimensionless groups, i.e. Re + Another... Inspections shows that simplest way to make g dimensionless is ... gL V2 m m 2 Units: s 2 ⎛m⎞ ⎜ ⎟ ⎝s⎠ Continued... Traditionally one uses instead ... FROUDE NUMBER: Fr = So ... V gL CD = F (Re, Fr ) Dynamic similarity requires ... (Re )m = (Re ) p AND (Fr )m = (Fr ) p The Froude number is a dimensionless number which expresses the ratio of INERTIA FORCE and BODY FORCE. It is important, for instance, when one considers flows with free surfaces and wave generation. Ship and Hydraulic Model Testing • One has free surface flow. Hence, dynamic similarity requires Rem=Rep AND Frm=Frp • We found earlier that one already gets into tough problems when one only has to satisfy the single condition Rem=Rep . Hence, we do not expect things to get simpler now... Let’s try to test a model ship! •Recall ... Re = VL ν with ν = µ , ρ Fr = V gL • If water used for model and prototype then ... µm = µ p , ρm = ρ p ⇒ νm =ν p • For dynamic similarity want ... (Re )m = (Re ) p Requires Vm Lm = V p L p 2 Vm2 V p = Lm L p (Fr )m = (Fr ) p From (I) Lm = Vp Lp Vm Substituting (III) into (II) gives ... (I) (II) (III) Continued... V p2 Vm2 = Vp Lp Lp Vm Vm3 = V p3 Vm = V p And because of Eq. (II) ... 2 Vm2 V p = Lm L p … then also ... (2) - repeated Lm = L p So, in order to carry out model test that satisfies dynamic similarity need model with same size as prototype and test it at same velocity prototype will be exposed to. In other words,… DYNAMIC SIMILARITY IS NOT PRACTICALLY ACHIEVABLE! Continued... • To overcome this problem Froude proposed to write ... CD TOTAL DRAG C DF (Re ) = + SKIN FRICTION DRAG C DW (Fr ) (A) WAVE DRAG • Skin friction drag assumed to be a function of Reynolds number o-n-l-y • Wave drag assumed to be a function of Froude number o-n-l-y HOW / WHY DOES ONE PROFIT FROM THIS? • To get (CD )m = (CD )p W one requires ... W Frm = Frp ...since wave drag is function of Fr o-n-l-y. But (B) is easy to achieve; simply ensure that ... (B) Continued... Frm = Frp Vm = g Lm 2 Vm2 V p = Lm L p Vp g Lp 2 ⎛ Vp ⎞ L p = ⎜⎜ ⎟⎟ Lm ⎝ Vm ⎠ Example: If … Vm = 1 V p one gets ... 10 2 ⎛ ⎞ ⎜ Vp ⎟ ⎟ Lm Lp = ⎜ 1 ⎜⎜ V ⎟⎟ p ⎝ 10 ⎠ Lm = = 100 Lm 1 Lp 100 Thus, if we use a 1/100-scale model then the velocity in our model experiment has to be 1/10th of the prototype speed. Continued... • From Froude’s assumption of Eq.(A) ... C D = C DF (Re ) + C DW (Fr ) Prototype Model (CD )p = CD − (CD )p (CD )m = CD − (CD )m W m W F Since ... (A) -repeated p F (CD )m = (CD )p W W … which is so because designed experiment such that Froude numbers are equal. ( ) ( ) C Dm − C DF = C D p − C DF m p CD p 123 = C Dm + [ (CD )P − (CD )m ] F F 123 1444 424444 3 Measured in model tests Calculated Total Drag Total Drag Difference between Skin. Fric. Drag Coeff. Prototype Coeff. Model Coeff. of Protot. and Model • A similar approach is used for hydraulic modeling in civil engineering • NOTE the essential role of theory in experiments! Photos of Models in Civil Engineering River Model Napier Harbour Model, Hydraulic Research Station, Wallingford, Oxfordshire Undistorted model of the Van der Kloof Dam (Orange River) Continued... Rotating-Tank Facility at L.E.G.I., Grenoble This facility is used to investigate the influence of background rotation (Coriolis Force arising from rotation of the earth) on large-scale geophysical flows in oceans. The diameter of the rotating tank is 13 metres! The facility was built in the early 1960s when engineers wanted to investigate possible effects of a planned tidal power plant in the north of France on flows in the channel. Example: Application of Froude’s Assumption in Ship Modeling, Greek Trireme Calculate power required for ancient Greek warship. How many oarsman are needed? •Prototype: Length = 30 m, Wetted Surface Area = 320m 2 , Speed = 5 ms−1 •Model: 1 − scale 10 Model Test Data •Water: Density ρ = 1000 kg m-3 , Dynamic Viscosity µ = 0.001Nsm − 2 • Also given is: Prandtl-Schlichting Formula for skinfriction drag of a flat plate. CDF = 0.455 (log10 Re L )2.58 : Re L > 3 × 106 Empirical equation obtained from least-squares fit to large number of experimental data points. Assume you found this in book (e.g. Schlichting, 1987, Boundary-Layer Theory, 7th ed., Mc-Graw Hill) Continued... Outline of strategy to solve the problem (A) Calculate Froude number for prototype. Use necessary condition for equal Froude numbers to infer model flow speed. (B) Determine corresponding model drag from table/plot of model-test data. (C) Calculate Reynolds number for model and prototype (D) Use the Prandtl-Schlichting formula for coefficient of skin-friction drag to calculate its values for model and prototype. (E) Calculate coefficient of wave drag for model (=wave drag of prototype since equal Fr numbers) using Froude’s assumption (Tot.Drag= Wave Drag+Skin Fric. Drag). Use the result to calculate total drag coefficient for prototype. (F) Calculate drag and power required for prototype. MAKE ASSUMPTION ABOUT OARSMEN (strokes/min; length oar handle, pull of oarsmen on handle etc.). Use these to calculate number of oarsmen (G) Work per stroke done by each oarsman (H) Power and useful power generated by each oarsman (I) Obtain an estimate for the number of oarsmen required to propel the ship at the assumed velocity Continued... Solution: (A) Calculate Froude number for prototype. Use necessary condition for equal Froude numbers to infer model flow speed. m Vp Frp = Since ... ⇒ gL p = 9.81 Frp = Frm Vm = 5 s m s 2 = 0.29 30 m Vp ⇒ Lm Vp = Lp gL p = Vm gLm m 1 m 5 = 1.58 s 10 s (B) Determine model drag corresponding to Vm from table/plot of model-test data (see graph next slide). Dm = 22.4 N ⇒ Total Drag Coefficient of model then... ⇒ C D = 22.4 N m 1 ρ Vm2 Am 2 = 22.4 N ( 1 kg 1000 3 (1.58 m )2 0.01 × 320 m 2 2 m ) = 0.0056 Because 1/10th-scale model; so every length, L, reduced by factor 0.1. Thus, area which is L2 reduced by 0.1x0.1=0.01 Continued... •Plot of model-test data from previous slide ... Continued... (C) Calculate Reynolds number for model and prototype Prototype: Re p = Vp Lp ν = ρ Vp Lp = µ Model: Re m= Vm Lm ν = ρ Vm Lm = µ kg m 5 30 m 3 s m Ns 0.001 2 m = 150 × 106 kg m 1 . 58 3m s m3 Ns 0.001 2 m = 4.74 × 106 1000 1000 (D) Use given Prandtl-Schlichting formula for coefficient of skinfriction drag to calculate its values for model and prototype. (This means we are assuming that skin-friction drag for ship is the same/equal to that for a flat plate.) Prototype: (CD )p = (log F 0.455 ) = 0.0020 (log10 4.74 ×10 ) = 0.0034 10 150 × 10 6 2.58 Model: (CD )m = F 0.455 6 2.58 Continued... (E) Calculate coefficient of wave drag for model (=wave drag of prototype since equal Fr numbers) using Froude’s assumption (Tot.Drag= Wave Drag+Skin Fric. Drag). Use the result to calculate total drag coefficient for prototype. Since ... Frm = Fr p ⇒ (CD w )m = (CD w )p Solving the ‘Froude assumption’ for model for (C D w ) m and introducing known values gives ... (CDw )m = (CD )m − (CD F )m = 0.0056 − 0.0034 = 0.0022 … and since … (C D w ) = (C D w ) = 0.0022 then... m p (CD ) p = (CD w )p + (CD F )p = 0.0022 + 0.0020 = 0.0042 Continued... (F) Calculate drag and power required for prototype. 1 ρ V p2 Ap 2 2 1 kg ⎛ m ⎞ = 0.0042 1000 3 ⎜ 5 ⎟ 320 m 2 = 16.8 kN 2 m ⎝ s⎠ D p = CD p PowerRequired = Force × Velocity = D p V p = 16800 = 84000 kg m m 5 2 s s kg m 3 s 3 = 84 kW ENTER OARSMEN ... What we really want to know is … How many oarsmen are required to propel the ship at 5 m/s ? ASSUME: • Each oarsman performs 12 strokes per minute • Each oarsman exerts a pull of 310 N on the oar handle • Each oarsman pulls oar through an arc measuring 1m • Efficiency with which oar converts strokes into useful energy is 55% Continued... (G) Work per stroke done by each oarsman Work = Force × Distance = 310 N × 1 m = 310 J (H) Power and useful power generated by each oarsman Power = Rate Of Doing Work = 310 J × 12 Strokes J = 62 = 62 W 60 s s Useful Power = = 0.55 × 62 W = 34.1 W (G) Obtain estimate for number of oarsman required to propel ship at assumed velocity of 5 m/s Number of Oarsmen = = Total Power Required Useful Power Per Oarsman 84000 W ≈ 2460 34.1 W Does this answer sound realistic? Continued... • The number of 2469 oarsmen seems a bit high. Revise assumptions and go through calculations again … • I found on www that the maximum speed of a trireme was 7 knots... 1knot = 0.51 m s ⇒ 7 knots = 3.57 m km = 12.9 s h Hence, maximum speed was, in fact, lower than the value of 5 m/s we assumed. Also, they probably wouldn’t have been going at full speed all day long. I presume that 0.25-0.35 x max. speed is a sustainable relaxed travel speed for slaves on long-distance trips. Check were the revised assumptions get you to ... • On www I also found that the actual number of oarsmen on a trireme was 170 (in 3 files on each side)