You are on page 1of 34
Liquid Flow by Hongteng Gao Ashley Harrell Jonathan Knight Conor Lynch Soa-Sin Sher Monk ce Blue A Aaliy Submitted to Renee Hale ChE 253M - Measurement, Control and Data Analysis Lab Department of Chemical Engineering The University of Texas at Austin Spring 2014 iquid Flow Abstract (& hhe purpose of the Liquid Flow lab is to quantitatively and qualitatively understand the theories and applications of different flow regimes using various flow meters. In orderdo . Ly do s0, all flow meters were calibrated against a primary standard, pressure drops were ° measured against a secondary standard, and visual observations were regpyded. The =. ys turbine flow meter was determined to be the most precise with a precision measurement J" '?.. = of 0.007105. However, the magnetic-inductive meter was found to be the best meter wert, abs wverall for the liquid flow system because it had the best accuracy, and its precision ay varied only slightly from the turbine flow meter. When observing pressure drop information, the orifice meter followed the expected trend of pressure drop increasing as <2 the flow rate and Reynolds number increased, but the venturi meter did not. This is She pS possibly due to the inaccuracy of the flow meters in the transition flow regimes, ifthe — \\! flow rate being used was not fully laminar or fully turbulent, On the other hand, the discharge coefficient, Co, of the venturi meter followed the trend of Cp increasing as the Reynolds number increases, while the discharge coefficient of the orifice meter did not. Instead, the discharge coefficient of the orifice meter decreased as Reynolds increased. Furthermore, some of the Cp values calculated for both meters were greater than 1. which wash $ impossible. It is possible that the ratio of the orifice diameter to the pipe diameter (tere “ contributed to the negative trend of the discharge coefficients forthe orifice meter, but \ {aby ©" there must have also been experimental errors that contributed to the discrepancies. Lastlf, the “ending laminar” transition was determined to be at a Reynolds number of 2533, and the “resuming laminar” transition was determined to be at a Reynolds number of 3022. The “ending laminar” Reynolds number was to be expected. outthe “resuming alee laminar” Reynolds number was much lower than what was expected This was belie’ zack to happen because of the inconsistencies in visual observation such as the green dye pearing tobe turbulent while the red dye appeared to-be laminar at a single flow rate. In addition, the different radial positions of the dye inlets also.affected the appearance of the liquid flow. This i> a gook, stash but it's too vagus. 9 Some. CLES ani too Speevhe Wrote, (achde ts acount of S Ne byonKWor ae Yhyey me .Us t bine mere ait ust thas “twerb os Sithoud Gest stshag That Fe sth Whet other meters OFM Soren Touling this abstract \ Sun Sook expe const belo LAME bud you WS He mest precise 1h \ 4 Long Le a eupniae wamt ae > mader corth TR YO SyStem. Don't bhindside your Re either woerpacted informahen . Trtduce thom then tea Anough- & 2 Contents Safety Introduetion ‘Theory and Sample Calculations Methods Experimental Results Conclusions and Recommendations References Appendices Appendix A List of Tables Table 1: Statistical Properties Table 2: Physical Properties Table 3: Experimental Parameters Table 4: Calculated Primary Standard Flow Rates Table 5: RMS Errors Table 6: Precision of the Meters Table 7: Tabulated Venturi and Orifice Data Table 8: Experimental Pressure Drop Values Table 9: Tabulated Visualization Data Table A-1: Calibration Raw Data List of Figures Figure 1: Experimental set-up of liquid flow experiment Figure 2: Magnetic-Inductive Measurement Curve Figure 3: Rotameter Calibration Line Figure 4: Turbine Calibration Curve Figure 5: Ultrasonic Calibration Regression Figure 6: Instrument Accuracy Figure 7: Flow Rate and Pressure Drop Correlation Figure 8: Laminar Flow Example Figure 9: Ending laminar Flow Figure 10: Turbulent Flow Example Figure 11: Resuming Laminar Flow Figure A-1: Magnetic-Inductive Confidence Interval Figure A-2: Turbine Confidence Interval u 12 16 19 19 2 24 24 28 32 28 Figure A-3: Ultrasonic Confidence Interval Figure A-4: Rotameter Confidence Interval as 1 x or Liquid Flow ” Ss wry be 7 Safety whe pe reov we ‘There are three main Safety hazards to pay attention to while performing this So Res i experiment: water, electrical, and glass. Water is a potential slipping hazard if it spills” \\ Jo onto the conerete floor, so try to maneuver the water in away to minimize spills. (fern! Furthermore, this goes hand in hand with the eléctrical hazard, By keeping the work Soe space dry and clean of spills, you will be preventing any electrical shock or instrument malfunction that may occur if electrical equipment is exposed to water. Finally, be careful with glassware in order to prevent cuts and scrapes. This is especially significant when doing the orifice and venturi meter calibrations because you will be handling glass three-way valves Introduction The objective of this experiment was to understand the theory and practical Ne applications of six commonly used flow meters and to determine flow by yar t2™" sacd} (quantitatively measuring flow rate and qualitatively measuring’ flow regime. Chemical que - . engineers play large roles in designing and supervising the transport of liquid flow “ through uring the flow rates of liquids. Therefore, it is extremely ye the various advantages and disadvantages > 4 Portant to understand how liquid flow Sod ae of the most commonly used flow meters. In order to do so, the team studied a piping system with an inlet water valve and © WS 3° an air operated water flow control valve that pumped water through a magnetic inductive see flow meter, a turbine flow meter, a rotameter, an ultrasonic flow meter, an orifice meter, 5° and a venturi meter, as well as a section of clear tubing set up for dye injection into the es 19 © ligud steam, The team used a primary standard to calibrate the flow meterggand also see OF praed om understanding obetmaon moter by measuring ho pressure drab acros the | 3 orifice and venturi meters. The team then injected dye to visualize the flow patterns | x geo created by laminar, transition, and turbulent flow - 0 descr i In this experiment, the students watered primary standard measurements and “70° An readings of four different flow meters at four different flow rates. They then used statistical analysis to create calibration curves for each flow meter and determine the relative precision and accuracy of each one. This report uses this information to make a recommendation on the best flow meter for this system. A key equation in liquid flow is Bernoulli’s equation, which describes the. conservation of mechanical energy balance for any basi fluid flow system, and yt be used further to calculate flow rates based on the préssure drop. A defining relationship in liquid flow is also the effect that the Reynolds number, which is proportional to flow rate, has on the flow regime and vice versa‘, These two fundamental ideas form of m.. 8 liquid flow theoey, and will be used to explain the behavior of the o and venturi {v Mneters, as well as the flow visualization. This report will also compare experimental data ‘Sq. for the orifice and venturi meters, such as pressure drops and discharge coefficients, to eae the results one would expect based on theory. Similarly, the observations made about flow patterns and flow regime at given flow rates will be used to calculate the Reynolds numbers for the beginning and end of the transition region between laminar and turbulent flow. These values will then be compared to those predicted by the Moody Chart ‘Theory and Sample Calculations {| 5) ‘ poe Flow Regimes “Xte-2V‘F> ecko boot arabe egos s Yon eh Melle Oo AN oe museca Hein ‘The purpose of the Liquid Flow Measurement Experinient is to elucidate certain characteristics of fluid mechanics. Specifically, this, experiment. emphasizes two ‘characteristics, First, fluid flow may have'deteriinistic behavior oF dyriamic (or chaotic) behavior. Second, the criteria by which a fluid is deterministic or dynamic depends on whether the fluid is in a laminar flow regime, transitional flow regime, or turbulent flow regime. Finally, the Theory section is organized by the nature of fluid flow, then the ory behind the Flow Visualization Experiment, and then the theory and design of doe \¢ © metering equipment. The theory and design of the metering equipment will then be used “jg to explain uncertainties in measurements. AN Bo be hue!To begin, deterministic behavior means that any changes within a system may be Ww 3 2X5 se" accurately predicted given sufficient initial conditions and boundary conditions. All of \oo aes Ge "s- Newtonian physics are deterministic (Swendsen, 2012, pps. 3-4). For example, one may uae accurately determine an object's position in space given an accurate vector function for yx © its velocity during a period of time. A laminar flow regime has deterministic behavior o> because instantaneous fluxes of momentum and mass throughout the moving fluid are predictable (McCabe, Smith, Harriot, 2005, pp. 46) On the other hand, final states of dynamic systems may not be accurately predicted from the initial and boundary conditions. A differential change in one initial or boundary condition could have no effect on the system at all, or it could change the system’s final state by orders of magnitude. Turbulent flow is chaotic in nature because there is no method of predicting momentum and mass distributions throughout the fluid. Only statistical approximations may be made at best (MeCabe, Smith, Harriot, pg. 57) (Stewart, 1969). The connection between different flow regimes was made in the late 19th century by Osbome Reynolds through his experimentation and subsequent formulation of the EEEPEEn dimensionless Reynolds’ number, nteth oe pow ee enon Re = one ve a . eo eaunsert where D is the hydraulic diameter, V is the mean fluid velocity, and v is the kinematic viscosity (Kline, 1969). The hydraulic radius is defined as the quotient of the or sectional channel area with the wetted perimeter. The hydraulic diameter is simply twice this value. The mean fluid velocity is defined as the quotient of volumetric flowrate and sectional area, and the kinematic viscosity is the quotient of dynamic viscosity and The Reynolds’ number is a dimensionless group which measures the ratio of inertial forces to viscous forces, where inertial forces are external forces that induce fluid flow, such as gravity or pressure. Viscous forces are molecular interactions within the fluid, like hydrogen bonding, ionicity, or any general Van der Waals force. A small 6 BEEP ERR HEEB EEEREHEEEEEEE NEA Yor Unkarsiand Ua BS of what yor corte atthe: Thisis oll way foo detedluk T dco need a \echua on the of laminar os. tucbulent flow, and TE lm rot couler scale behakor molecu) > coppre nied te cretion Ck effort oP cesenrching Varah Reynolds’ number reveals that intermolecular forces are more preValent than inertial forces and the molecules within the fluid maintain some degree of order. This called laminar flow, and Oswald Reynold’s demonstrated this phenomenon in the late 19th, 2 century (Kline, 1969). Ceadsy praca sdy £ The flow visualization experiment in this lab is precisely the samb ax the experiment that helped make Reynolds famous. A fluid moving with a low Reynolds’ number will have a continuous velocity profile in the radial direction. The shape of the profile is determined by the shape of the pipe. Circular pipes that were used in this ‘experiment give a parabolic velocity profile, increasing from zero velocity at the walls of the pipe to a maximum in the center of the fluid. This results in “sheets” of fluid that continuously slide over one another, each one containing slightly more velocity than its neighbor closer to the wall. These “sheets” form as a result of a small momentum gradient in the fluid that allows molecules to remain ordered despite bulk motion. Reynolds’ visualized these smooth sheets of fluid with a dye, as was also done by the experimenters. During laminar flow, the stream of dye did not show radial mixing with the fluid. It remained in a smooth line and did not change shape or density to the naked eye (Kline, 1969), This is exactly the same behavior observed by the experimenters in the Flow Visualization Experiment. Reynolds was also able to use dye to demonstrate the chaotic nature of turbulent flow. At high Reynolds’ numbers, the intermolecular forces within a fluid are dominated by inertial forces. This leads to radial cross-mixing within the fluid and therefore no smooth velocity profile exists for the bulk fluid motion (Stewart, 1969). Reynolds’ demonstrated this behavior by increasing the flowrate of his laminar stream containing the smooth “sheet” of dye. The dye then reached a transition period in which the “sheets” of fluid begin to break and mix with each other. Shortly after the transitional flow, a clear break in ordered fluid motion was observed as the dye mixed with the entire stream and the flow became turbulent (Kline, 1969). All in all, this brilliant experiment showed the contrast between deterministic, laminar flow and dynamic, turbulent flow within a fluid. This difference may be seen as inertial forces are increased in a fluid until it reaches a small transition period, and then complete order of laminar flow becomes the complete chaos of turbulent flow. As an aside, some degree of order may be seen in turbulent motion if an experimenter looks on a small enough scale. Fluid molecules in turbulent motion will form eddies. Eddies are small vortices. That is, they are aggregates of molecules having circular, laminar flow that vary in size but are typically on the scale of nanometers or microns, Although the study of vortices and how energy is exchanged between them, itis beyond the scope of this material and the reader may see the references for further reading (Stewart, 1969). Measuring Devices and the Secondary Standard ‘The experimenters chose the magnetic-inductive (MI) flowmeter as the secondary. standard because it had the smallest degree of error between its measurements and the theoretical flowrate calculations. This may be seen in the results section in table 6, page’ 20. This instrument has the smallest degree of error because its design is the least dependent on mechanical characteristics of flow. The meter consists of a small metal coil BEB EB EEE RRR REE EEE EE EE {0 Wand turbulent flow follow entirely different empirical power laws for circular pipes. ‘\ .g” in which a magnetic field is induced when a conductive fluid flows over it. Conductive ss © __ flvids-are-vitheronie-or-covalent-with strong dipole character, so water or an acid would iss, oe so work with this meter. (However, nonpolar. chemicals like hydrocarbons or ethers_are (°0“\> ato coe ineffeetive)- This magnetic field then emits electromagnetic radiation, and the frequency \\spo" Se of this r i 2 The flowrate may then be determined because it is. oN“ roportional to the induced field frequency. The MI meter is highly accurate because the} . inductive coil is very insensitive to minor fluctuations in fluid properties. These include ("= ug instantaneous, local changes in pressure or density for turbulent flow, or exaggerated al (©, __ momentum distributions for laminar low. (Vel"t, Mushkin, 1996). Now ey? The next device is the ultrasonic flowmeter. The ultrasonic flowmeter is a highly © ¢¥°“* : : Ses the . Accurate device because it uses multiple acoustic pulses to determine volumetric flowrate. Th Set O02 However, the complex physies involved in the measurements cannot eliminate all of the yga-nt wt YO >* 5 error. It has been shown that velocity profiles in laminar and turbulent flow may affect — jaye. bt “XS? ultrasonic pulses (Filatov, 1996). The re, there are numerous correction factors used in (OS AG, the design of these devices. Filatov +his'SRown that the correction factors for laminar flow Seat os Because of the rapidly changing behavior between flow regimes, correction factors for Me™» > the transition region are still an area of research interest (Filatov, 1996). The device used in this experiment used frequency of pulses to determine flowrate, These types of ultrasonic devices have been shown to be more accurate than other devices which measure flowrate according to the Doppler effect, or other acoustics phenomenon. (Kivilis, Reshetnikov, 1965) he turbine flowmeter is a ‘The third measuring device is the turbine flowmeter ct constant cross-sectional area tube whose sensing element is a stationary” turbine blade —bU © nal to the jucersto The rotary motion induced on the turbine by the moving fluid is propor liquid’s mean velocity. It is also proportional to the liquid’s volumetric flowrate because the geometry of the pipe and the turbine blades never change. The device measures turbine rotations by an electrical signal that is sent to a digital sensor every time the turbine completes a rotation. The frequency of these signals, read by the device in kHz, is proportional to the volumetric flowrate of the fluid (Flow Technologies, 2004), ‘The mechanical nature of the turbine meter can allow multiple sources of error that are worthy of note. One source of error may be mechanical vibrations along the apparatus coming from people or other experiments. Another source of uncertainty could be due to minor cavitation around the turbine leading to infrequent rotations. However, the experimenters observed no substantial error for the turbine meter. The fourth metering device is the rotameter. As mentioned in the method's section, the meter consists of a buoy that adjusts cross-sectional area with pressure, Fluids at higher pressures will displace the buoy more and the meter therefore allows more fluid to flow through. (ABB, 2001). The experimenters simply set the rotameter to 60%, 45%, 30%, and 15% of maximum volumetric flow, and observed predicted decreases in flow rate in accordance with the decreased cross-sectional flow area. " Uncertainties in rotameter measurements are similar to turbine uncertainties because both Z~ ° ker] devices are mechanical. Roto The final two metering devices belong to a class of devices called obstruction 5.05. °° meters, named so because they obstruct the fluid and use the resulting pressure change t0 g poe qe-Q CX determine flowrates. One type of meter the experimenters used is called an orifice meter, cated or" ercor and the other is called a venturi meter. Both meters use the Bernoulli equation to calculate the pressure drop across their obstructions. The Bernoulli equation is a differential mechanical energy balance that includes kinet potential energy from gravity, and mechanical work from pressure. The solution to the energy balance is othe fornrof Equation 2: ne TF rWAcblks ae, Kaliceed inte, ©) eaten you tak cee the M19, where P is the pressure, p is the density, Vis the velocity, Wis the sievation, gis Hetert gravitational acceleration, and g. is the masg to weight conversion factor (Willson, Friedman, 2012). Ohad ort tin APS Vande? Note that terms for mechanical work loss due to friction have been neglected. As ‘a result, kinetic energy will be lost as heat from friction and the flowrate calculated by the obstruction meter will be less than the flowrate predicted by the Bernoulli equation. This prompted the inventors of obstruction meters to define a discharge coefficient, @) which is the quotient of the flowrate according to the meter to the flowrate according to the Bemoulli equation. ‘There is a specific discharge coefficient for each Reynold’s number (Daniel Co., 1997). There is one major problem the experimenters observed when calculating a discharge coefficient for their orifice meter. ‘The discharge coefficient for a Reynolds number of 15,000 was slightly greater in m ge coefficient fora jy Reynolds number of 20,000. The ‘acual Reynolds” mumbo) 0.64 and-0.00,- Sur CS respectively. This is counterintuitive because a higher Reynolds number means a greater mean fluid Velocity which would then imply a greater volumetric flowrate. However, this, is not what the experimenters observed at a Reynolds number of 20,000. A second set of measurements displayed the same behavior and therefore confirmed the existence of some unexpected phenomenon in the orifice meter. The most feasible explanation is that there is a transition region for the orifice meter as a result of the design. The inereased pressure drop could be caused by expansion of the orifice throat, or flowback in the connection from the meter to the gauge, or any number of other possibilities. The difference in measurements made by the Venturi meter and the orifice meter are strictly due to differences in their design. The orifice plate always allowed a greater pressure drop than the Venturi meter. This is because the orifice meter uses a discontinuous change in cross-sectional flow area and the Venturi meter uses a continuous change in cross-sectional flow area. ‘The sudden opening in orifice meters allows for greater fluid expansion and therefore a greater loss of pressure. The smooth closing and opening of the Venturi meter allows for less fluid expansion and therefore less pressure loss. (McCabe, Smith, Harriot, 2005, pps. 225, 228), 9 CE ee R00 bt Yor ovens viouSly « Flow Meter Calibration ate peers "onal oA Sa iN previous Combine cur supers SNaless text The outputs of a magnetic-inductive flow meter are in kilohertz (kHz), which are proportional to the system flow rate. The measurements of a magnetic-inductive meter can be highly accurate, but it only works for conducting fluids, such as water in this experiment. The turbine flow meter contains a rotor, which rotates when fluid passes through; the rotating speed of the rotor is proportional to the flow rate. The number of rotations per second is then converted into a frequency reading in kHz, For a flow rate less than 150 cc/see, one can calculate the flow rate by plugging readings into the following equation, Q = oWe @) where Q is the volumetric flow rate in ce/see, w is the output in Hz, and We is the Witte constant, 115 cc/see. Turbine meters are usefull for particle-free fluids, and the ‘measurements can be fairly accurate (Willson, n.d.) The ultrasonic flow meter measures the difference in time of an ultrasonic wave travelling upstream versus downstream. Then one can calculate the fluid velocity, and thus the flow rate, by the following equation, yt) 2Xtuta 6 where L is the path length, X is the axial spacing between the transducers, 4, is the upstream time, and fis the downstream time. However, there might be some small error in the laminar flow regime because of the parabolic velocity profile (Willson, n.d.). The rotameter measures flow rate by allowing the cross-sectional area the uid travels through to vary, causing some measurable effect. It consists essentially of a gradually tapered glass tube mounted vertically in a frame with the large end up and a float moving freely in the tube (McCabe, 2007, p. 232). The float rests at different positions of the tube due to different flow rates. The flow rate calculation is based on the following equation, 2acve Joi “AE Pw Qaetuat = Awl (6) where Q is the volumetric rate of flow, ft'/sec; Vr is the volume of the float, ft; pe and py are the densities of the float and the fluid, Iby/ft"; dy andy are the area of the float and the annular orifice, ft”; and Cp is the discharge coefficient (Willson, n.d.). In this experiment, the team used a rotameter to control the flow rate when primary and other flow meters were running. 10 4 EK oll VS HERE REE EEE EEE EEE MeN OE Sioa IO chro In order to calculate parameters for the magnétic-inducNon meter linear regression, the students caleulated the sum of squares and calculated the parameters using their definitions. Flow Meter Calibration Calculations aot. ysaxrtb 7) =e en)? 4 Sex = D(X — X)° = 7083 (< (8) =e Syy = DRa(% -¥) = 614 kHz? (9) Say = DEAK VX —¥) = 2086 ke (10) 00 siz Be = 0.294% ay — bX = -2.58 kz (12) For the standard error of estimate for the linear regression, the students used the sum of the squares error as shown below. SSE = Syy ~ (2) = 614 - (22) = 0.00831 kHz? (13) 00644kHz | a4) The 95% confidence predicts the ce ‘model and in which Wis the populatio standard deviation, ainty with which the true parameter is within the , X'is the average of the data, and o is the Table 1, Sample Calculation Statistical Properties Statistical Property Value X (Average Flow Rate) 90.2 mLs N (Population Size) 4 3.182 RHEE E HEHEHE HEHEHE HR HR REE EE EE (15) a = 0.294 + 3.182 x 0.0644 |1+ 22 = 0.2944 0.242 (16) B bexs, [E any B 58 + 3.182 x 0.0644 | = -2.58 + 0.00243 kz (18) = Gem? fr, omar 5 YH WEEK se [OSs yj + 3.182 x 0.0064 | + SE (19) \ For the first data point, y; C08 wet 91 = 40.8 + 0.02 kilz (20) Orifice and Venturi Meter Calibration Calculations as The Reynolds number &s shown below is a dimensionless number that relates the velocity, diameter of the pipe, density of the fluid, and the viscosity of the fluid flowi through the pipe. The Reynolds number predicts the flow behavior of the fluid through the pipe, such as whether it is laminar or turbulent. ‘The example below is a sample calculation determining the flow rate of the fluid at the orifice meter throat at a Reynolds number of 10,000. ‘Table 2. Physical Properties for Sample Calculations. Property Value Density of Water 0.9984 g/mL Diameter of Orifice Throat 0.508 em Viscosity of Water 0,01002 g/(sec-em) Specific Gravity of Blue 1.75 10,000 22) HBEEHEHHEH HEHEHE EEE ELSES SS _40,000)(1.002%1077= 25) _ | em @% = “Tassos )osoacm) see 23 A= BE = MBC ~ 0.203 cm? (24) + + Q = 40.0 Te (25) ,, 4 ‘alt Bernoulli’s equation, which is shown below, models the flow of a liquid in a pipe between two points by incorporating the pressure, height, and kinetic energy difference ho bet the twe ts. —~ \ pss tween the two points. Ze Now bent HO PP WE-VD—Ce= Tg cepa Cuno o-2i c elroy Pr--P2 2G Ie sous ob - inte Ceper nce ‘The following equation measures the ideal flow rate in an obstructed pipe assuming no TeFe-re friction losses, and is a rearrangement of Bernoulli's equation assuming a horizontal pipe 4 ye and no height difference. Hore £4 \ 26) | Qideat = VoAr Given the imperfect nature by which a liquid flows, especially with an obstruction such as an orifice plate or venturi meter, the ideal volumettic flow rate fails to accurately model the flow of a liquid, and, as a result, a drag coefficient is incorporated to account for the pressure losses caused by the instrument used. cy = Saat Z Qiaeat Qactuat = Code | me 7) in which P is the ratio of the orifice or venturi diameter to the diameter of the pipe preceding it. AP is the pressure drop calculated in the following equation where p is the density of the fluid by which pressure is being measured. Plugging these values into the Qigca equation yields the following: (29) (oven) (ease )(i—(2setens)") (0.203 cm?) 38.282 (30) Qiaeat Using the calibration curve for the magnetic inductive meter, Equation 31 determines the actual flow rate flowing through the pipe as the student measures the pressure drop. 20 ez = eee 31 0615 2 & Qactuat in which fis the average frequency in kHz that the student measured over a period of a minute during the pressure drop readings for each Reynolds number. lz 2) om actuny = 2S%PAHEY2#809 HE) _ 3g 9g om? «2 02045 see Asa result, the drag coefficient for this Reynolds number is found through the following equation Qactuat — 1.02 (33) Qicat a9202 Flow Visualization Calculations @ and 332 calculate the Reynolds for For the flow visualization we used equati the laminar to turbulent transition. at aassaknice25909Ktt2 _ ge om? cea, = ne = 08.7 G4) 045 sec = Sacsuat — 9875 on i Ve nate Isstemt 272 sec 65) (572%) aasem)(o9ee4-2 (i002 Re= = 2533 36) Experimental Methods — qe Section Experimental Set-Up whe Ser + ‘The experimental set-up forthe liquid flow experiment consisted of a piping system with an inlet water valve andr operated water flow control valve that pumped“ pu “ho? water through the system. Water flowed through six different flow meters: a magnetics induntive Row miele, a tubing flow mater retiiten, an altmonionow meter a SSS het orifice meter, and a venturi meter, in that order, as seen below in Wee Zo pun Pressure gauge panel, including) Figure 1. Experimental set-up of liquid flow experiment (taken and edited by authors) A pressure gauge panel was included in order to measure the pressure drop across the orifice and venturi meters. By using a combination of three-way valves, the team was able to control whether the pressure was being measured across the orifice plate, the / venturi throat, the entire orifice meter, or the entire venturi meter. A section of clear pipe with syringes inserted at fixed points to inject dye was included near the end of the piping loop for observations of flow visualization. The water in the system discharged through an open end of tubing into a trough. Specific parameters of the orifice and venturi meters, the manometer, and the flow visualization tube are alse reported below in Table 3. 15 ‘Table 3 Experimental Parameters (Willson, n.d., p.1) Property Value Diameter of the orifice throat 0.20 in Diameter of the venturi throat 0.25 in Diameter of the pipe upstream of the orifice meter 0.55 in Diameter of the pipe upstream of the venturi meter 0.50 in. Diameter of the flow visualization tube 1.75 in Specific gravity of blue manometer fluid L75 Flow Meter Calibration The first part of the experiment measured the volumetric flow rate of the water in the system with a magnetic-inductive flow meter, a turbine flow meter, an ultrasonic flow meter, and a rotameter, and calibrated the readings of each meter against a primary standard. The primary standard was determined by using a graduated cylinder and a calibrated stopwatch to measure the volume of watydisgharged from the system for a given time interval, which the students tsiode keer Sh ten seconds. The team then calculated the true volumetric flow rate of the system by taking the ratio of the volume measured to the time, jnterval, ‘Phis-tve-volumetrieftow-rate-served-as-the Prmary standard-S Ae The team used the rotameter readings as baselines for when to take flow rate (Sood | measurements. Therefore, in accordance with the lab manual, the team took a set of (oo ¢ qahon readings of the other three flow meters and primary standard measurements when the 6° rotameter read 60% of the maximum flow rate (Willson, n.d., p. 2). Because the magnetic-inductive, turbine, and ultrasonic flow meters measure instantaneous flow rates, students recorded 10-20 readings of each flow meter over about five minutes, and |“ reported the average reading for each flow meter at a rotameter reading of 60%, Similarly, the students took five measurements of the true volumetric flow rate, and reported the average as the primary standard for a rotameter reading of 60%. Because the rotameter reading varied minimally with time, the students only took a rotameter reading at the beginning, when the flow rate was first set, and at the end, after the team had taken all of the other readings and measurements. The team repeated this procedure at rotameter readings of 45%, 30%, and 15%. For the primary standard measurement at the 60% rotameter flow rate, the team measured the volumetric flow rate five times, and all of the data points were reasonable. E However, for the 45% rotameter flow rate, the first measurement seemed to be abnormally higher than the second and third measurements; therefore, the students decided to take a fourth measurement. Fortunately, the fourth measurement was consistent with the last two measurements. Students only took three measurements at each rotameter reading of 30% and 15% because each set of measurements were consistent. In accordance with the lab manual, the team took the magnetic-inductive flow meter as their secondary standard flow meter, and plotted the magnetic-inductive flow meter readings against the primary standard readings (Willson, n.d., p. 2). The team then used the calibration curve in subsequent steps of the experiment. 16 properties of water at 20°C, the team calculated the corresponding flow rates through the orifice plate in milliliters per second. They then calculated the associated Reynolds numbers at the venturi meter throat based on the calculated flow rates (Willson, n.d... 4). Based on the calibration curves generated in the first part of the experiment, the team determined the required readings of the magnetic-inductive meter needed to achieve the Nalculated flow rates. The lab manual designated the magneti Orifice Plate and Venturi Calibration Using the given Reynolds numbers of 10000, 15000, and 20000, and the e& <4 correlation coefficient, R®, suggesting that the dafa best fit the expected linear trend Boe ‘og The experimental setup contained a pressure gauge panel consisting of four pressure gauges, with the fourth gauge being a manometer. The student conducting the 2 experiment used the magnitude of the pressure drop to determine the appropriate pressure | | ‘gauge to read from. When reading from the manometer, the student measured the pressure by reading the difference in blue fluid height in the manometer. Due to the fact instantaneous property, and that these readings easily varied within a range of at least 0.2 kHz, it was nearly impossible to set the meter readings to the exact values calculated from the calibration curve. However, the team made every effort to set and adjust the meter to within an acceptable range of the desired reading, usually around 0.1 kHz. For cach desired flow rate, one team member was responsible for recording 20 meter readings over the course of approximately five minutes for an average flow rate measurement, while the other team members adjusted the three-way valves to measure the pressure drop actoss the orifice plate, upstream and downstream of the orifice, across the venturi throat, and upstream and downstream of the venturi. Upon reviewing the preliminary data, the team found that the pressure drop upstream and downstream of the venturi meter did not increase between the middle and highest flow rate values as expected. To verify the results, the team repeated the experiment for the highest flow rate value. Viswal Determination of Turbulence In the final part of the experiment, the team introduced the dye into the flow visualization tube by opening the stopcock of the appropriate buret of dye. The experimental set-up introduced green dye at the beginning of the flow visualization tube, while the red dye entered near the end of the flow visualization tube, as seen in Figure | Once the dye was flowing and visible, the team varied the liquid flow rate with the control valve in order to observe the approximate laminar, transition, and turbulent flow rates and practiced varying the flow rate of the dye to optimize its visualization. Some experimenters believe sending the dye in “puffs” yields the best visualization of the dye, while others believe trying to match the flow rate of the dye with the flow rate of the liquid yields the best visualization of the dye. The students noticed that it was difficult to ‘maintain consistent lengths and velocities of “pufls”, and decided to match the flow rate of the dye with the flow rate of the liquid for optimal visualization, In order to match the flow rate of the dye with the flow rate of the liquid, the students slightly opened the 7 that the magnetic-inductive meter readings changed every second, beeause-it-is—an~ inductive meter as the team’s secondary standard flow meter, which had the highest value of the squi LY Gook stopcock for laminar flow, halfway opened the stopcock for transition flow, and fully opened the stopcock for turbulent flow. ‘Once the students were confident with the ability to vary both the liquid flow rate and the dye flow rate, the stopeocks were closed and the liquid flow rate was returned to ‘an approximately laminar flow. The red and green dye stopcocks were both partially opened, and their flow paths through the flow visualization tube were observed, making sure to observe the entire length of the tube in case any mixing occurred toward the end of the tube. Students adjusted the liquid flow rate until the flow paths of the red and green dye both appeared to be laminar, and then recorded the liquid flow rate at which both flow paths appeared to be laminar and took pictures of the laminar dye paths. The flow rate was recorded by taking 20 magnetic-inductive readings periodically within five minuteggand averaging the values; the magnetic-inductive flow meter was used based on previous results and conclusions. The students slowly increased the liquid flow rate until the dye flow paths began to transition from laminar flow to turbulent flow. At this point, the dye’s flow path no longer appeared as smooth as its laminar flow path; instead, it appeared slightly wavy or mixed. Using the same procedure previously described, the team recorded the liquid flow i i i i i i i i rate at this point, and took pictures of the dye paths. This point is the “ending laminar” ' t ' ' ' ! I I I I I point, but it can also be considered the beginning of the transition region. The liquid flow rate was then increased past the transition region and into the turbulent region established at the beginning of the lab. The stopcocks of both dye burets were opened fully to match the dye flow rate with the liquid flow rate, and when both flow paths appeared turbulent, the students recorded the liquid flow rate and took pictures. In order to determine when laminar flow resumes, which can also be considered the end of the transition region, the students slowly decreased the liquid flow rate until the dye flow paths began to transition from turbulent to laminar flow. This was determined when the dye flow paths were no longer disordered and completely mixed, but rather slightly ordered and mixed more slowly, Once again, the students recorded the liquid flow rate at this point and took pictures, before closing the dye stopcacks and water valve Results Flow Meter Calibration At each rotameter reading, the team measured a set of volumetric flow rates, and the average ofeach set was taken tobe the primary standard flow rate at that rotameter. bey reading, as reported below in Table 4, The volumetric flow rates decreased wher - rotameter readings decreased; therefore, the overall primary standard is reliable, {but OS PO Seymore wy z you re? \ere gene “ ess Saag es \ oust Me ere - douse ni EO to 4 2 eek The alteteilty coms from howd condont you oF . s remand Fochmiguxs arb whether ' Aa qouLr meee Fe chniqurd , ote yo tee ope che Noro salcdrsrip teak Bub waily, Ye = Yoo a a ae neTIned Orin, ant ViCeyia 1A. SO fer to singe | HBHEHHREHEHEHEE EEE 8 8 EE Eee Table 4. Calculated Primary Standard Flow Rates assumed that the errors of the instrument readings were much greater than those of the primary standard. This is because linear regression already assumes that the errors on the y-axis are greater than those of the x-axis (Willson, n.d.). To verify this assumption, the students calculated the average standard deviation, which is a measure of the error, of the primary standard measurements, and found it to be 1.31:)On the other hand, the calculated root-mean- square errors (rms error), or standard errors of estimate (S,) of the instrument ‘measurements, which is a measure of the error of the instrument readings, are tabulated below in Table 5. Three of the four instrument reading errors were actually lower than the error ofthe primary standard measurciyents, gel iy violating the assumption, oe RMS Errors confidence intervals for each of the flow meters are in the Appendix. Steg OF Udy Flow Meter RMS Error | yt wy po? zs Magnetic-Inductive (kHz) 0.06443 o Turbine (kHz) 0.004537 o Ultrasonie (mL/min) 34.5252 “Rotameter (%) oo0117 J) gos 1 Raaliee Gor eats Nonetheless, the calibration curves for each of the four measured flow meters are shown \ you below in Figs. 2-5; and the explicit data points used for the calibration curves and the BHEHERHEEHEHHEHE HEHE SEER eee feep the caphons fr Yee RoureS oor Ack the seme if 4 they ge the sont iMter machen. Ta this cast, sepinhon ee & bad gromnar: We caphons Sor Hpse Source p> S hold at 35 Numb S> crak nae oR re metees- Frequency (kH2) o 50 100 150 200 ‘Average Volumetric Flowrate (Primary Standard) (ml/sec) HeteF @Aiorabion Figure 2. Magnetic-Inductivd Measurement ‘Curve. The calibration curve for the magnetic-inductive meter has an error of 0.0644 kHz for the frequency measurements and the primary standard had an error of 1.31 mL/sec. Rotameter Calibration 40% Rotameter Reading 0 2 4 6 80 100 120 140 160 ‘Average Volumetric Flowrate (Primary Standard) (mL/sec) Figure 3. Rotameter Calibration Line. The calibration curve for the rotameter has an error of 0.001917% for the instrument measurements and 1.31 mL/sec error for the primary standard measurements. 20 Turbine Calibration (0089-00497 R= 0.9999 Frequency (kH2) 0 20 40 «60 ©68 ©=— 100-120 140360 ‘Average Volumetric Flowrate (Primary Standard) (m/sec) Figure 4. Turbine Calibration Curve. A linear regression line for the turbine has an error of 1.31 mL/see for the primary standard and 0.004537 kHz error for the turbine measurements, Ultrasonic Calibration 9000 8000 ‘7000 6000 5000 4000 3000 2000 1000 ° y=57.847*- 108.07 R®=0.9999 Ultrasonic Reading 0 2 4 60 80 100 120 140 160 ‘Average Volumetric Flowrate (Primary Standard) (m/sec) m@EE EEE HERE EE Ee ee Figure 5. Ultrasonic Calibration Regression. The linear regression the team calculated has an error of 1.31 mL/see and 34.525 mL/min, Lor ches In order to accurately determine the true flow rate in the system by comparing the primary standard to the meter reading, the students needed all of the data in the same units. Therefore, they converted all-of the.raw data collected among the flow meters and SG used the calibration curves to convert the data into units of mL/s. These values were then os pa) es a be ws \ Ree" \2 2! ao * Cocoemmne™ BEEBE EREHEHE HEHE EEE EE EEE used to tabulate the precision of each meter in Table 6, where a lower pre corresponds with a more precise meter. -now Table 6. Precision ofthe Meters ~ so. C2" %o 45% 30% 15% te . Rotameter__Rotameter__Rotameter__Rotameter (hats 2 0.01072 0.008126 0.009243 0.01410 0.01055 inductive ‘Turbine 0.007404 0.003456 0.004804 0.01276 0.007105 Ultrasoni 0.005221 0.008605 0.009408 0.01132 0.008638 Rotameter 0.008264 0.01111 0.01639 0.03226 0.01701 After comparing the average precision of the flow meters at different flow rates in Table 6, the team concluded that the turbine meter was the most precise, with the lowest, 4 average precision number of 0.007105. However, since the average precision numbers of | (oC the magnetic-inductive and ultrasonic flow meters were also fairly small and very close eee to the turbine meter’s precision number, the team concluded that there was not a significant difference in the precision of the three flow.meters, although the rotameter was clearly the least precise. The students then calculated experimental flow rates (Qeyx) of each meter, as determined by cach calibration curve, and its difference with standard flow rate (Qeayi— Qua). Using this data, we plotted (Qeapr— Qua) VS. Qua in Fig. 6. Instrument Accuracy = Fos Fos Ear 7 Bos a @ Magnetic Inductance Sos gos wmuitrasonic | Bos . z aTurbine }g] 6 F gy -otameter z $ | do 2 a | 0 so 10050200 ‘Actual Flow Rate (ml/sec) er plot qualitatively demonstrates that the struments, Figure 6, Instrument Ac magnetic-inductive meter is the most accurate of the four In Figure 6, one can see in two of the four data sets (34 and 108 mL/s) the (exer ~ Qsrn) of the magnetic-inductive meter is smallest, which means around these flow rates the magnetic-inductive meter is the most accurate flow meter. However, at the flow rate near 2 BEEBE HEHEHE RHEE REE ee ee 147 mLis, the magnetic-inductive meter is less accurate than ultrasonic, and at the flow / rate near 71 mL/s, the magnetic-induetive meter is ess aceurate than the rotameter. When _// considered as a whole, however, the magnetic-induetive meter is still the most accurate herefore, the magnetic-inductive meter (the flow meter assigned as the -condary standard) is the best option because the accuracy is an important factor, and can be used to differentiate between the turbine, ultrasonic, and magnetic-inductive flow meters, which all had similar levels of precision. Besides precision and accuracy, there are other factors that are also significant in choosing a flow meter for a particular application, For example, the magnetic-inductive meters are useful merely for conducting fluids, and the turbine meter is only useful for particle-free fluids and flow rates below 150 mL/s (Willson, n.d.). This limits the use of both the magnetic-inductive meter and the turbine meter. Common applications of the magnetic-inductive meters and turbine meters are pharmaceutical applications and leak detection in pipelines respectively (Ma inductive, n.d.; Hoffer, 2013). On the other hand, the rotameter is advantageous to chemical engineers because it can measure a wide variety of flow rates (Willson, n.d.). Engineers often use rotameters to measure steam Sa flow in boiler control (Variable, 2013). When it comes to orifice and venturi flow meters) D& “G chemical engineers can evaluate them hand-in-hand, This is because they both measure | >\ O°“ pressure drops due to the liquid flow rate. They only differ slightly in manufacturing and! quan pumping costs. The manufacturing of orifice meters is cheaper in comparison to the ren‘ ‘manufacturing of venturi meters, but orifice meters cause a higher pressure drop and, __ therefore, higher fluid pumping costs than venturi meters (Willson, n.d.). The final flow” \-¢ meter to consider is the ultrasonic flow meter. This flow meter is advantageous due to its negligible pressure drop, but it is disadvantageous due to its high cost compared to the other flow meters (Bengtson, 2010). Engineers often use ultrasonic flow meters i wastewater applications (Introduetion, n.d. os Using the average magnetic-inductive flow meter reading for each ideal Reynolds number and ideal flow rate, the team calculated the actual flow rates and the actual Reynolds numbers at the orifice plate and venturi throat, shown in Table 7. The team found that the Reynolds numbers at the venturi throat were lower than the correspondin; Reynolds numbers at the orifice plate. This is the expected relationship between the two sets of Reynolds numbers because if one rewrites the Reynolds number as a function of flow rate, rather than velocity, the Reynolds number becomes inversely proportional to diameter, and the diameter of the venturi throat is larger than the diameter of the orifice plate. HEE HEHE HEHE HR RRR RRR Eee Table 7. Tabulated Venturi and Orifice Data ‘Comparison of idea! versus actual Reynolds numbers and flow rates, and tabulated discharge coefficients, for orifice and venturi meters Ideal ‘Actual Nre Quast Qiteat (ML/5) Co Nee _| Orifice | Venturi | (mL/s) | Orifice | Venturi | Orifice | Venturi 10000 | 9728.57 | 7787.46 38.96 36.28 42.25 1.074 0.9222 15000 | 15293.68 | 12242.18 | 61.25 62.55 71,32 | 0.9792 0.8588 20000 | 19699.27 | 15768.74 | 78.89 85.53, 73.39 | 0.9224 1.0749 20000 | 19258.93 | 15416.25 | 77.13 84.30 71,98 | 0.9150 1.0715 The measured pressure drops across the orifice plate were significantly higher than the corresponding pressure drops across the venturi throat; similarly, the measured pressure drops upstream and downstream of the orifice meter were significantly higher than the corresponding pressure drops upstream and downstream of the venturi meter. This can be scen in Table 8. This data supports the expected differences between orifice and venturi meters because venturi meters reduce the pressure loss caused by obstructing the liquid flow, as previously discussed. The fixed conical shape of the venturi meter controls the shape of the jet of liquid flow downstream of the venturi throat, while the sharp edges of the orifice plate allow for the liquid stream to separate just downstream of 4 the plate and create eddies, resulting in the additional loss of pressure head (McCabe, “>! 2007, p. 226). Practically speaking, this also means that orifice meters are cheaper to Pas oy manufacture but require higher fluid pumping costs to overcome the larger pressure drop) (McCabe, 2007, p. 227). Table 8. Experimental Pressure Drop Values Pressure Drop Orifice Plate Orifice meter Venturi Throat Venturi Meter Ideal Nre_inH20_cmBlue_inH20 cmBlue cm Blue cm Blue 10000 - 213 - 16 13 8.2 15000 - 63.3 - 52.2 32.2 22.9 20000 35 1836-29 98.07 34.1 218 20000 34 114.9828 94.69 32.8 21.1 Bernoulli’s equation shows that as flow rate or velocity increases, the magnitude of the pressure drop at each point should also increase. The data in Table 8 shows that while the orifice pressure drops duly increase with increasing flow rate, the pressure drop did not increase significantly across the venturi throat when the ideal Reynolds number increased from 15000 to 20000. It actually decreased slightly across the entire meter. The team took a second set of data points at the ideal Reynolds number of 20000 to veri! ify the unexpected data points, and produced similar results. Manufacturers design simple flow meters to measure laminar and turbulent flow, but not necessarily transition flow, 24 However, the experimental values for Cp did not correspond to expected theoretical trends. While the overall trend for the experimental venturi Cp is increasing as the Reynolds number inereases, the overall trend for the experimental orifice Cp is actually decreasing, The experimental venturi Cp for the middle flow rate was slightly lower than the one calculated for the lower flow rate, but this can be explained by the error in pressure drop trend, which was discussed above. The experimental venturi Cp was also lower than the corresponding orifice Cpat the low and middle flow rates. Several of the Cp values calculated for both orifice and venturi meters were also greater than 1, which is‘impossible, and all of the Cp values for the orifice meter were greater than 0.9, which is significantly-greater than the constant Cp of 0.61 expected for high Reynolds numbers. re ‘These discrepancies are potentially due to experimental error when reading the pressure drops on the manometer. However, because the discharge coefficient is also a funetion of B, the ratio of the orifice diameter to the pipe diameter, itis possible that the specific B for this orifice meter contributed to the negative trend. It has been found for certain B values, a negative trend was also recorded for discharge coefficients relative to Reynolds numbers, although no reports have been published on this orifice meter’s of 0.36 (Ramamurthi, 1999, 135). Other sources of error that may have contributed to these significant discrepancies between experimental and theoretical trends and values include experimental error when reading the pressure drops on the manometer; the accuracy of the calibration curve which was used to determine the actual flow rate: and experimental errors in calculating the ideal flow, although it should be noted that no matter how the pressure drop was converted units-wise for Equation 26, the negative trend in the experimental orifice Ci persisted. Visual Determination of Turbulence ‘The first observed flow regime was a laminar flow path, as shown below in Fig. 8. The dye is flowing in an orderly manner without mixing in the visualization tube, which is characteristic of laminar flow. It should be noted that the red dye had a'tendeney to appear more laminar thar the green dye because the red dye was closer to the tube wall, ~\(25 which slows down the dye’s velocity, and thus lowers the Reynolds number. — 7 \lon voor} havc laminas” _\ Ploco athe wall br _— hachulcat in the mt Aye noeete Misherbe not Ple the Re was high ddle (qenere Wy. The 5 & athe Plow He of its \veeton, mB EB BE BB EB EE HE EE = Sa Sf a = ss SE EP EB ERE ER ER RE EE RE Figure 8. Laminar Flow Example. A picture shows a stream of red dye flowing along a visualization tube, After the students increased the flow rate, the flow path began to change (see Fig. 9). In this case, the flow is primarily ordered and unmixed after first being injected, but it becomes more wavy and mixed as it flows further down the visualization tube, representing the beginning of the transition region from laminar to turbulent flow, or the “ending laminar” point, \ Figure 9. Ending laminar Flow. The picture shown above illustrates the red dye becoming disrupted by the onset of turbulence in the tube. When the students increased to flow rate significantly to ensure that they had reached turbulent flow, the dye flow path became completely disordered, with complete mixing in the visualization tube, as shown below in Fig, 10. The green dye was injected into the middle of the flow pattem, and was quickly dispersed, reflecting the expected turbulent flow pattern, Figure 10, Turbulent Flow Example. The dye mixes with the water almost instantly indicating a turbulent flow field. mere Qe BS 2 BS 2 2 2 2 2 SS Se ee Be After the students decreased the flow rate, the flow path began to change, as shown below in Fig. 11. The flow path is primarily wavy and mixed throughout the visualiza tube, but a clear dye path can still be observed, unlike the complete mixing of the completely turbulent flow (see Fig. 10). This flow path and corresponding flow rate represent the transition from turbulent to laminar flow, which is also the end of the transition region, or the “resuming laminar” point. Figure 11. Resuming Laminar Flow. The turbulence within the tube begins to become more ordered and less sporadic, and the dye does not axially mix along the tube. Magnetic-induetive flow meter readings taken at both transition phases were averaged, and used to calculate a Reynolds number for the beginning and end of the transition region from laminar to turbulent flow. Because the magnetic-inductive flow meter reading is proportional to flow rate, and thus velocity, itis expected that the reading for the beginning of the transition region, the “ending laminar” point, is lower than the reading for the end of the transition region, the “resuming laminar” point. In addition to the magnetic-inductive flow meter readings, a Reynolds number was also calculated and recorded for each transition point, as shown in Table 9. Table 9. Tabulated Visualization Data, Transition Point Average MI flow meter reading (kHz) Reynolds Number “Ending Laminar” 23.52 2533 “Resuming Laminar” 28.55 3022 According to the Moody Chart, the “ending laminar” and the “resuming laminar” flow transitions can be expected at Reynolds numbers of 2000-3000 and approximately 10000 respectively. The “ending laminar” Reynolds number recorded by the students, 2533, matches this particular rule-of-thumb. However, the “resuming laminar” Reynolds number recorded by the students, 3022, does not. This could possibly be due to

You might also like