Calculation & Shortcut DESKBOOK PUBLISHED BY AL INCERING *UBLICACIONES PERIODICAS C.A. APARTADO POSTAL 10354 TELF: 77922-TEL/FAX: 78034 MARACAIBO - E00. ZULIACALCULATION & SHORICUTTABLE OF CONTENTS PHYSICAL or CHEMICAL PROPERTIES; STOICHIOMETRIC CALCULATIONS; MATHEMATICS Quick conversion of NOx and CO emission rates Flammability of low-Btu gases Properties of mixtures via parametric analogy Mentally converting from degrees F to degrees C Estimate relative volatility quickly Caloulate kinematic viscosity of liquid water Improved Z charts Nomograph for moist-air properties Calculate thermal conductivity for unassociated liquids FLOW and STORAGE of LIQUIDS and SOLIDS An exact solution to the Colebrook equation Sizing pipe after steam traps A quick estimate for centrifugal-pump efficiency Unusual problems with centrifugal pumps Check pump performance from motor data Shortcut methods for determining optimum line sizing in Sl units Operating performance of reciprocating or positive-displacement compressors Protecting against backflow in process lines Predict gas/liquid-flow pressure drop Startup of centrifugal pumps in flashing or cryogenic liquid service Power consumption of double mechanical seals Estimate maximum allowable pressures for steel piping HEAT TRANSFER Use these guidelines for quick preliminary selection of heat-exchanger type Assessing fouling in heat exchangers Rtwvmaoaw+ 15 16 18 19 22 24 26 27 30 31 33 36 39 41Monitor fouling graphically Weighted mean-temperature-difference for exchangers with phase changes Conserving steam Operating boilers intermittently Find the most compact surface condenser Does your surface condenser have spare capacity? Oil removal from condensate Improve efficiency of electric heating How to predict batch-reactor heating and cooling Purging prevents condenser corrosion How effective are finned tubes in heat exchangers? MASS TRANSFER and SEPARATION OPERATIONS Relating transfer units and theoretical stages A new correlation for sieve trays Using CAD systems to prepare McCabe-Thiele diagrams Evaluating turndown of valve trays Quickly determine multicomponent minimum reflux ratio Extend the life of slurry agitators Critical-velocity plots for moisture separators How to simplify batch crystallizer design PROCESS MEASUREMENT and CONTROL How to keep track of instruments Don't overspecify control valves Analyzing batch process cycles Control disturbances? Check grounding Ensuring the security of field control cabinets. Avoid erratic control operation by suppressing relay arcs A control algorithm for batch processes Simplify batch temperature control Making small-scale additions: 1. Meter small amounts of problem liquids 2. Measure small gas-flows Try this simple pressure tap trick 63 66 70 71 73 74 75Maintain maximum liquid level in pressurized vessels Determine viscous-reaction endpoint easily Onstream calibration of pressurized gas-flow controller Desuperheater control-system cools gas Prevent plug valves from sticking and jamming OTHER ENGINEERING TECHNOLOGY Reducing equipment-cleaning wastes The Pareto chart. . . .Tool for problem solving Improve production efficiency via evolutionary operation Directly determine reaction order Determine saturation and stability indexes graphically Protect turbomachinery from magnetism Correctly position jam nuts Quickly estimate steam turbine power Electrical circuits in hazardous locations 99 100 102 104 106 109 113 116 119 121 123 125 126 129PHYSICAL OR CHEMICAL PROPERTIES; STOICHIOMETRIC CALCULATIONS; MATHEMATICSQUICK CONVERSION OF NO, AND CoO EMISSION RATES V. Ganapathy, i Abco Industries, inc, Abilene, Tex. | | oniorng combustion enix || M Sion produets — NO, and CO combinedeycie plants or cama | produce both ofthese ematnn nations perahity thes te tery sre. Because of ths station ce9 wa olor nr ts ftom mats fowrates GU/H) a out dry bai whichis heearehwee oct | they are reported to regulatory bod: ——————__| | Nomenclature | | NO, (ppm); 1%, volumetric | | ratio of CO (ppm) {| jw flowrate of a constituent gs in the exhaust stream, || | Ibyh Wo = total exhaust stream | flowrate, Ib/b 1] [F — =converon factor | MW — = molecular weight of ex-| | | haust stream, Ibs | Hs0 = percent volume of water | in the wet exhaust stream | || en in the exhaust stream Conversion factor, F 0, ='percent volume of oxy- | | 32 28-24 ts BO H20, vol. 9 sections of the vertical and diagonal lines yield conversior | | calculating the votumetic ratios (in ppm) of CO and NO. 4 | n factors forthe usual 1 industry With the enclosed chart it i possible to accomplish this purpose, The chart provides a conversion factor, , which allows both the lb/h-to-ppm(r) eonver sion and the reverse Underlying theory Io, in units of tb/b, is the flowrate of NO, (usually reported as NO,) in a turbine exhaust stream of WF Ib/h flowrate, the following equation gives the volumetric ratio on a dry hasis. (The molecular weight of NO. is 46.) F = [owls (wearin) x 100/(100 ~ 8,0) @ In this equation, 40 is the volume of water vapor in the exhaust stream, and J/Wis the molecular weight of the gases. V, must further be corrected for the 15% oxygen, dry basis, as follows: Vi, fin pomiy)} = 10°71 15) 3 TOT eAOT xz, ® In this case, %0, is the pereent volume of oxygen in the wet exhaust gas. Similarly, the CO emission rate in Pom), 15% Os, dey bass, is Ve Lin ppmiyi] = 1.8827, 3) Figure 1 solves for the concentrations of both NO, and CO; the conversion factor, F, is obtained from it as a fune- tion of 4,0 and %0,. A molecular weight of 282 was used for the con- struction of the chart. Then, Va fin ppm(v]] = (w,/ 7) x 108 x Fd) V, fin pp(y)} = 1.642 2 (wel W) x 10x F 6 Sample problem Determine the NO, and CO coneentra- tions, in ppm(y), 15%, dry basis, if 25 Ib/ hhof NO, and 15 Ib/h of CO are present in 560,000 lb/h of turbine exhaust gas. The percent volume of HO is 10, and the percent volume of Q, is 11 in the wet exhaust gas Solution: From the Figure, Fat 10% HO and 11% 0, is 0.48. Then, V, equals (25/550,000) x 10° x 0.46 = 21. Also, Ve equals 1.642 x (15/550,000) x 10° x 046 = 206, .When gos streams need to be incinerated, estimating flammability can be difficult. Here are a table and a procedure to help. Flammability of low-Btu gases John J. Clowson Engineering Consultant very engineer whose responsibltiesinelnde furnaces, fuels, incia- erators or processes involving ofgases is occasion ally confronted with an offgas or 8 waste stream that needs to be incinerated. Such streams are sometimes so low in their combustible content that one cannot even be sure that they wil burn; in such cases, there is frequently a voice in the background r questing an immediate es mate of the amount and value of primary fuel required to n- cinerate the gas stream Estimating flamability Accurately estimating. the flammability of low-Btu gases has always been problem On such occasions, we pull a handbook ofthe’ shelf that contains « "Flammability Ta ble," examine the data for sev eral ofthe most common com- pounds in che mi, notiee that differences between om pounds seem radial, then start searching fora common denominator. If some of the tama on the at “ora with an endothermic reaction, the problem becomes even more sizable. If you're unlucky | quantity, or minimum flammabil tnough to own wo handbooks with amma abe, you | may end vp wishing that you were on vaeation, because references don't always seem to prove the same answers. This article will examine this confusion and attempt to provide a more satisfactory solution Tb fluegas/Ib fu Endothermic cracking of methane and other gases that a ing process, oxidation will not occur until after the In the eas the era For constituents undergo- ing cracking reactions that ab sorb heat, the heating value of the fuel is: (ib carbon/tb fuel x 14,098 Beu/Ib) + (Ib _hydrogen/Ib fuel x 51,623 Buu/lb))~ heat of formation Thus the low heat value (LHV) of methane is: (0.7487 > 14,098) + (0.2513 < 51,628) ~ 2,005.98 = 21,5184, ‘The maximum fluegas quanti ty or minimum flammability condition for methane is: Wb fluegas/Ib- fuel = 21,518/ (carbon fraction x 745) 8858 lb, where the carbon fraction represents active combustion carbon, Oxidized ‘or partially oxidized carbon should be treated as inert in this equation For constituents underge- ing eracking reactions that re- lease heat, the heating value of the fuel is carbon energy plus hydrogen energy, plus the heat of formation. Thus, the low heat value of acety: Tene is: (0.9226 x 14,098) + (0.0774 x 1023) + and the maximum fluegas for acetylene is: (T42I(TAS. x 0.9226) = 30.18 Tb bsorb heat inracking resetion has taken place. Then, if the fluegas is, or execeds, 150 Btu/lb, the hydrogen oxidizes, triggering the eurbon reaction. If, after the carbon has been oxidized, the fluegas stream enthalpy equals or exceeds 745 Btu per pound of free carbon, the combustion reaction is self- sustaining. Recognizing that most flammability charts are + /~at least 2.5% of the minimum quantity, ths criterion ean ne easily confirmed within a reasonable tolerance. ‘Table — Minimum conditions for the combustion of fuel gas CO reaches 550 Beu/lb CO; however, any hydrocarbons in the mixture will be only partially oxidized at best. ‘Combustion flammability table ‘The table of combustion flammabilities, accompanying this article, is not intended to replace flammability charts that were designed for safety considerations. Rather, itis a tool for the combustion engineer who is seeking ‘to achieve ‘complete combustion, ‘The following. coriments are appro- priate in understanding or using. the ‘Minimuni chart: cua, ME agg maletiometne | “TA fuel components ae in eva a —— me — or st it T2P. If fuel is in a liquic Component traston Sur tuegeso asl vy lum tamparati, + | or sate a T2F If fue i ina Tid Methane 0.7487 21,520 98.58 456 1,956 ‘subtracted from the low heating value | ethane ayes noge 3409 280 gies of the component. Propane —Datyt sea values Jn, hydrogen reline —Slake Haase aha ‘8 Zige | .2'Te values for carbon, hydro | Rotuane oaz00 ieee eat of formation are taken ae ‘from a well-known chemical-engineer- n-Pentane. 0.9929 19,517 3147 1.29 jing handbook; however, if one attempts | 'so-Pentane = 0.8323 19,478 ‘att 130 to caleulate low heat values from pub- NeoPentane 08229 19888 91.28 ee lished component values, the low heat Eityene 0.8563 202951 323 values do not always agree with pub- lished low heat values in the same Propylene o.e563 19,691 so 7 ce handbook. These discrepancies have n-Butene os ae = been reconciled by assuming that LHV isoutere 08569 18082 3038 te | mPomone Osea ege3 8038 ie values are correct and then adjusting | Benzene 0.22 17,480 25.40 1.49 ‘the heat of formation. 3. A\loss of energy from the reaction Touoe 0912s i720 2800 rr 2494 | prior to completion af the reaetion (ie, Ayo 06017760 28.a8 i Bee aac Seat cli ek Atsiyene 0.9228 20778 90.18 388 ear : bape laphthalene 71 16.708 23.93 097 2502 sink, ete.) can prevent the completion of ee eikaivedtion i, contlioes ares boris Caton line. In those cases, the energy input monoxide 4347 9.00 10.05 See must be increased to compensate for Note The e-seum and lehteam omoeres ee assured oe 72F fort acstons. the las Exothermic cracking ‘Those constituents that release energy in the cracking pro- 38 do not appear to obey the same rules as Uhose that do not. Values in the published literature suggest that those compounds oxidize ata much lower thuegas enthalpy level. should be remembered. that the published Sammability arts ure basically a safety tool intended to prevent explo- ons. Their purpose isnot to define the minimum fraction of fuel required for complete combustion, Published flammability levels identify the concentration at ‘whieh the eracking reaction and hydrogen oxidation begin to vent. A complete oxidation reaction, ineluding the free carbon, oeurs atthe same enthalpy level stated for the other srses, Because all components react in the same manner, mix: ures also react in like manner, and the minimam combus- tion-Bammabilty-level of a mixture i: ax, Ih uegas = (LHV/Tb fuel/(eaebon fraction x 745)) ‘Tae important consideration is the defrition of the C frae- tion. CO, is not combustible earbon and should be treated as sn inert. CO is not combustible carbon but an entity unto itself, CO will burn when the fluegas enthalpy per pound of 4. The procedure described in this ar- ticle establishes fuegas enthalpy as the criterion for a complete combustion reaction. Temperature is a result, not a criterion. For example, a fluegas stream resulting from the combustion of CO gas containing a significant amount of water vapor will burn at a temperature well below 1,706, ‘The minimum enthalpy value required to maintain com bbustion ean be partially provided by preheating the combus- tion-air stream or the fuel stream, or both. A preheat enthal- by treatment for either or both streams is, for example, very ‘common when using low-Btu gases for process heating purposes. In such eases, the minimum enthalpy limit can be used to define the minimal supplemental fuel requirement. Roy V. Hughson, Biitor The author ohn J. Clawson S17 Wark oud, Pns City OK 4, ap egigeer ing consultant spealiang ip beat transfer, bred heaters ard balers Bor te bee cnc conrlant(e eee nd Cie Boos bez Conca canslce ie ute to Gen a fever gat ee BE sean” Ste Gane mee ‘rowhe untety tc Ua Sic Contandon hcnae Rowe are flout of elon suena ae ote acy eh ead atta cantata Us be st Bic a fp Sa hehe Pete i Ein bel lta eee ed Tea of omper erecting bape combust fd tester potProperties of mixtures via | parametric analogy Matthew Pottick and Kurt F. Kaupisch* Dictrrminingstysica property datas eral in designing [chemical processes. For binary mixtures, it is often dificult to find values of physieal properties over the range of temperatures and eoneentrations that will occur through- out a process. This is so even for mixtures of some fairly ‘eornmon compounds. Quick estimates of properties are needed during the early stages of design. This can often be done, and with reasonable accuracy, by employing what is known as the method of the parametric analogy. This method is based on ideas presented ‘earlier by one of the authors [1] Parametric-analogy method Assume that we are dealing with a two-component, mutually soluble system, Also assume that we have a curve of a property vs. concentration at one temperature. The eurve can be plotted from published data, ‘Now, assume that, for both compounds, there is no deviant behavior owing to changes in either temperature or concentration. ‘Then, for other temperatures, the shape of the eurve will remain essentially constant. The only difference willbe in its relative position as temperature varies, To determine the temperatureddependent gositions, use date found in the literature for pure components. That is, the endpoints of these new, shifted curves will be fixed by Knowing the properties of the pure eomponents at each temperature Example ‘Toillustrate and clarify this method, we shall determine the surface tension of ethanokwater solutions at various ‘temperatures: Fig. 1 2] presents the variation of surface tension with concentration, at 25. For pure water and pure ethanol, the variation of surface tension with temperature appears in Fig. 2 (7. These two szraphs fix the endpoints forthe eurves developed in Fig: 8 The shape ofthe curves remains essentaly the same; only the endpoints sre different. ‘This method can be used for determining values of many other physical properties, ineluding heat capacity, enthalpy, boilingpoint rise and viscosity. References 1 Kaupiach KP, Method of Digits} Computer Data Retveval, Chom Bagh daly BL aD 2 Goniercat Solvents Corp, “Industrial Ext Alhs.p on Corp. 8. Lange N. A. ed. “Handbook of Chemistry” 0th dp. 151, Mea Hat Sow ek oat on a “an Corp. Brenerng Di 080 Andro Way, Cninat O 6, New Vor, ‘010230 a0 0708090100 Ethanol concentration, wr. Figure 1— 5) wsing paramovie analogy, tis graph ‘of surface tension can be extrapolated to other temperstues a Facweter For ethanol ‘02040600100 120 140 Temperature, °C Figure 2 — Pits of the surface tension of water and ethanol ere used in Fig. 310 determine endocints 7 3 Surface tension, yrelem ‘0 i020 30 40 80 G0 7080 9D 100 Ethanol coneentration, wt 5 Figure 3— This predicion of euraoe tension ie urate enough to uso for protminayKazem Abnary™ Hire Ss ie sd accurate media to assomplish shi | of a calculator or paper and conversion witha peneil. Taking the standard formala: C= (F-218 om) and substituting 2 for 18, the formula becomes: O=rH F216 @ ‘he error in using this approximation is: (C-Cye = 031 = 1% @ ‘This error can be decreased to 1% by using: O=C+uC o Bech! Epcbering Det, Fahy of Energy Unienaly ofTeran Mentaily converting from °F to °C ‘To illustrate this methad, we will convert frum 86°F Ww °C 1. Divide *F by 2: 86/2 = 43, 2. Subtract 16: 43-16 = 2 B.Add 10%: 27 4 27-— 29.70 An exaet conversion yield 0.070. IF the temperature is an | odd number, add 1 to it to make it even and follow the above procedure. In this case, adding 1 to the temperature will, to some degree, cance! out the 1% error, making the result more accurate. As an example, convert 93°F to °C 1.98 + 1 = 94 94/2 = a, 3.47-16 = 31 4314 81-3470. A more-exaet ealeulation yields 33.0. Estimate relative volatility quickly MP. Wegte" Avis greiner ten needs an estimate ofthe relative volatility of two components. Here is a simple expres sion that relates the average relative volatility to the normal boiling points and the latent heats of vaporization of the two components, in the temperature range of their boiling points; so[ocsieax where: a = relative volatility between the two components in the temperature range Ty, to Trg Ty, = normal boiling point of Component 1, K Tug = normal boiling point of Component 2, K £, = latent heat of vaporization for Component 1 at Ty, keallkmole tent heat of vaporization for Component 2 at Th, kealmole If compound's latent heat is not known, it can be esti- ‘mated from the normal boiling points and molecular weight, M, using the following equations’ For alcohols and aciis: Ty #8 + 9.1269 log 7, ~ 6.1589 — 9.388 + 3.1269 log 589 5 + onasne F-62055 x0 9(28)] a | For other polar compounds: 1 = ninco +3084 7+ oma 2 r + (1.085 « 10-9) )= a6 x 10-74] or hyrenrbne rafino1 + 27 og IT), ~ (2630). ge } Example ‘The average relative volatility of benzene and toluene ean be determined using the following data: Ty, = 353.3 K, Tu = 888.8 K, Ly = 7,362 kealvkmole, and Z, = 7,930 keallkmole (where the subscripts and ¢ denote benzene and toluene, respectively), Substituting these values into Eq, (4) above, wwe find that a ra. 7a0)]= 2975 This compares with a value of 2421 for a determined using Vapor-pressureitemperature charts, scr Pn 4,38 Hg Nt ei Kot al, "Te Properties Geran Lids ed, Mera, New York, it. 21h ar x (TiCalculate kinematic viscosity of liquid water AE. Bostowissi™ ing water without looking up its viseosity | ‘An accurate and simple correlation to calculate the od in terms N&= ‘you can calculate the Reynolds number for flow- kinematic viscosity of liquid-water (1) is express of temperature as y= 0168 Tomi where, T= liquid water temperature, °C Eq. (1) provides the required kinematic viscosity of liquid water in a one-step caleulation at any temperature between 25 and 250°C, ‘The accuracy ofthis correlation lies within 4%, as shown in the table below. Also, Eq. (1 finds applications in uid mechanies and heat transfer (e.g. for internal water flow) in calculating the Reynolds number, Nee = DU/», which ean be written as: ( Nne = 5.95 DUTOSS @ ‘Paahy of Bagnerng, Mansur University, BLMansursh, Faye 25°C 1%.” To solve this and other problems, argon was chosen as the reference fs once again. By analyzing the extensive saturation ‘em. perature tables [4], the author was able toad nes for liquid and vapor to the original diagram, It was found that the curve marked “Deviation > 1%" fell very cloce to the eurve deduced for saturated vapor Lice change to the superheat region was found to beFigure 4— Extended 2 chan reduced volume) < 0.75. Note that eritcal-region anomalies are not included on the chart, Values of the compressibi vapor appear in the table. ‘The author is indebted to Prof. E. F. Obert for allowing him to reuse the original chart, which was amended as required except near the critical point, 7, >0.7, vy (critical | ity factor for the saturated indicated previously. le References jNekon, L5G and Ober, Generalized PVT Properties of Gases, Troma 3B V0L Te Ns Wy EE os Ge ey, 1\C, aod Obert. Lawe of Corepanding Sse, ACh J Well, Not pp ach ss, Penn Angus, §. an Armstrong, TUPAC Chemie! Dasa Sarioe No. 5 neal Fas modyghie Fables oe Pa hes Sto Ne ‘iktrwords, Loo Rabineyich. V.. “Mhecmophysial Properties of Neon, Argon, Ki ton and Keron’ Sandands Prva, Nareke Be aNomograph for moist-air properties | " P. Candhidasa® (1 When engineers need to evaluate the values of moiscair properties they gencrallyreler to psychrometric charts or tables. However, reading a chart can be difficult and time-consuming, and using tables often requires double interpolation. ‘The namograph presented here can be used t estimate several moist- air properties at a glance, if any two pieces of information are known. Standard atmospheric pressure is assumed. To use the figure, simply connect the two points representing the known quantities, and read the values of the other properties directly from the nomograph Example ‘An air sample has a dry-bulb tem- perature of 95°F (35°C) and a wet- bulb temperature of 80°F (26.7°C). Determine the relative humidity, the dew-point temperature, the humidity ratio and the vapor pressure. Solution: On the nomograph, con- nect t= 95°F on the center axis with f= 80°F on both axes. Read the following approximate property val- ues: RH = 53%; y= 75°F (24°C); 130 grains’ of moisture per pound of dry air (0.0186 kg moisture per kg dry air); and P= 0.87 in. Hg @2.1 mm Hg)Calculate thermal conductivity for unassociated liquids q Clayton P. Kew Here sz cation or aeulatng the thermal conduc tivity of many unassociated industrial liquids. Tt is better than many other methods because itis more accu rate and, more importantly, the input data are easy to obtain: parameters such as density, eritieal temperature, critical pressure, boiling point, molecular weight. It is suitable for vither polar or nonpelar liquids, but cannot be used for assceiated liquids such as water, aleohols, or Nomenclature M_ Molecular weight, gie-mole Pe Critieal pressure, atm T Temperature, K 1, Normal boiling point, K 1. Critical temperature, K organie acids. re i For spherical unassociated liquids (either polar or non- || 7 Dimensionless temperature polar), the dimensionless thermal conductivity, A*, is a || 4, Thermal conductivity, calfexemXi0) : A* Dimensionless thermal conductivity fanetion of dimensionless temperature, 7%, density, p*, 3 oe 4 Dipole moment, debyes and dipole moment, 4*(1]. Here, the shape limitation ig removed by using the Pitzer accentrie factor, w [2] x Dimensionless dipole moment ‘The Pitzer acentrie factor is ¢ property of pure fluids || ? Density, giem* a and has been widely tabulated (for example, see Ref. || @* Dimensionless density (2D. Te can be estimated in several ways, one of which [2] || @ Pitzer acentric factor, dimensionless Boltzmann's constant He = 3.20048 x 10-* calK is given in the Nomenclature. x N Avogadro's number R Aopecie eben te inimad slug Shan yield the best fit (standard error of 4.3%) of the data: oe 2 aone + 9:0 x10-% N= 6.023 x 10"/mole Ideal gas law constant R= 1.9872 call(g-mole\(K) Equations for dimensionless quant T= Tyr To. + (0.3788 ~ 0.17460) ‘ M60 Tt wp + 08 w= 1.58n( FE) — 1.490 22 G.04 e159 (54-) - 1.490 tl Note that this is « dimension equation, ao that the Parameters used mst have the units shown in the Nos rencltare listing Reterences WH ek Tay Ot i Mtl Sach apt naar a, a SERA Saaha tt ermine ma, | pr oe eee ms Met aay, mcFLOW and STORAGE of LIQUIDS and SOLIDSpeo An exact solution to the Colebrook equation Hsisien Chent hhemical engineers are familiar with the Fanning (or Darcy) friction factor, f the Moody chart of fvs. Reyn= olds number, 2, and how all ofthis fits together to caleulate pressure drop for a given fuid flow in a given sized pipe. The friction factor is calculated frum the Colebrook equation: LVF = -2 logi(e.7D) + (@2.5UR.VA) a where cis the absolute pipe roughness, D is the inside diam- eter ofthe pipe, and Ris the Reynolds number for flow. Eq, () serves as the basis of the Moody chart. Since the Colebrook equation is implicit in , it often has to be solved bby numerical methods, such as the Newton-Raphson, For many years, numerous investigators have proposed explicit approximations for Eq, (1) as ean be seen in a sum- mary by Serghides (1), Serghides compared nine explicit approximation formu- las, Various formulas gave good results but none exactly compatible with the Colebrook equation, Here is a new strategy to solve this equation. The method uses direct sub- stitution repeatedly; the results are quite accurate. The method Let us begin with an intial substitution for 2.5U/VF of Eq. (D, As an estimate, we ean use the famous Blasius equation: f= 0.79, ® ‘where 4 converts from Darey to Fanning factors. Ba, ()is valid for Reynolds numbers between 2,100 and 10° for by. rately smooth pipe (iD = 0) This simple equation is usefil for making estimates fir a range af OOS oy = (0467. Using an average valu of 0.0383 fr fields a nave forthe term 2.51/V fof about 14. Combining tis vais wisy Eq. (1) gives: UVF = ~2 logile.7D) + WR) @) Subsitating a, into Ba, (1), we obtain the result of fst iteration (where: A = €8,%D, and B= 8.028) ~2 log ~ B log + 14/R,)) a ‘ves of New Meso, Dept of Chemie and Noda Engnecrng, Abu! ‘gene, Si ssh ae By repeated direct substitution, we obtain the result of fifth iteration: UVF = ~2 loge ~ B log(A ~ B log(a ~ B log — B logiA ~ B log(A + 1WRQ))) 6) Although Eq. (5) seems lengthy, itis easy to compute due to its nested form. Itis identical to the evaluation of polynomi- als by Horner's rule—the optimal way to arrange a polyno- mial for rapid evaluation without peeforming. substantial ‘computations during rearrangement. The evaluation of a polynomial of: PO) = aye" + age + stages o requires n(n + 1)2 multiplications and m additions. Horn- er’ rule is achieved by rewriting P(x) as: PCat) = dines + ty + Oy + 2 > Og + a2) (1) which takes only n multiplications and » additions; a; to as, are constants, the accuraey of Eq. (6), a matrix of 70 points was compared with Eq, (1). This comparison is siilar to one by Zigrang and Sylvester (2), but also covers the transition region (2,100 < R, < 4,000), This test matrix consists of seven Reynolds numbers by ten relative-roughness values. ‘The values of Rare: 2.5 x 10°, 4 x 10, 3 x 10S, 10°, 108, 107 and 10%, The values of eD are:4 x 10°8, § x 10-8, 2 x 10-4, 6X04 1.5 x 10-9, 4x 10-8, 8x 10-4, 1.5 x'10-%, Bx 10°? and 5 10°, ‘The results? In’all cases, the accuracy agrees within at least six decimal places with the Colebrook equation. Thus, Eq, (5) is valid for all values of R, and elD. In fact, the cotivergenee is so good that any degree of precision can be obtained simply by inereasing the number of the iteration References 1. Seis, 7K. “Bana Pin Factor Acute” Cher. Eng, Va eoesc 2. Zipang D4. an Sivenor ND. Exit Apprexingion tothe Solon of ‘CoEtevts Fin Pator Eaton, SICH Te Val Noe p Sie ES 1516 Sizing pipe after steam traps Michael V. Cologera* and Arthur W. Brooks! ndersized condensate return-linesereate one ofthe most common problems encountered with process steam traps. Hot condensate passing throngh a trap orifice loses pressure, which lowers the enthalpy ofthe condensate. This enthalpy change causes some of the condensate to flash into steam. The volume of the resulting two-phase mixture is usually many times that of the upstream condensate. ‘The downstream piping mast be dequately sized to effec- tively handle this volume, An undersized condensate return- line results in a high flash-steam velocity, which may cause ‘waterhammer (due to wave formation), hydrodynamic noise, premature erosion, and high backprossure. The latter condition reduces the available working differential pres sure and, hence, the condensate removal capability of the steam trap. In fact, with some Lraps, excessive buekpres sure causes partial or full failure. Due to the much greater volume of lash steam eompared with unflashed condensate, sizing ofthe return line is based solely on the flash steam, itis assumed that all Mashing oe- curs across the steam trap and that the resulting vapor- liquid mixture can be evaluated at the end-pressure condi- tons. To ensure that the condensate line does not have an appreciable pressure-drop, a low flash-steamn velocity is as- sumed (60 fs) [1 Propet A, Hackensack, ND 0 ering in Line eth Ptr, N89.‘The flash-steam volumetric fowrate too | is then determined Qe= Wary, @ Based on the assumed velocity, the required cross-sectional area is caleus lated as 2G Are = 3,600 x 50 a 20 | Tosimpify the nomograph, the ow area is converted to 2 nominal pipe di- ameter. For cases of low pressure drop or high subeooling, it may be necessary to size the condensate line bused on the liquid velocity. Generally, a velocity of 8 fus is acceptable, 60 For flash-steam velocities other than Upstream pressure, Py, pig Nominal pin siz, Og, 50 fs, the nomial pipe size may be sa | 270Tonimated by: LUD >. 5 4a Vi 2 Additional advantages to the 20 | nomograph 1. Te yields a single result; decision ‘making is not required, as with other Figure — This nomograph datermines recommended pipe ze methods, Soteieet sown er = 2, With minimal training, field ‘maintenance personnel ean use it. Nomograph 8. Not only is it applicable to new construction but, more ‘The accompanying nomograph quickly sizes the recom- | importantly, can be used to check existing line sizes when mended condensate return-line. trap performance is questionable. Fae nomegraph employs an enthalpy balance atthe up- | 4. It builds on the eoncept in Ref 2] to provide a com- stream and end-pressure conditions to caleulate the weight | plete caleulation procedure’ ppereentage of flash steam that is formed and the lash steam flowrate, Example An evaporator is condensing 5,500 lbh of steam at 150-psig @) } supply pressure. During normal operation, a control valve maintains a pressure of 85 psig upstream ofthe steam trap. ‘The condensate is returned to a vented tank. What line size is recommended downstream of the trap? On the chart, connect upstream pressure of 85 psig through vented tank pressure (0 psig) to the pivot line. From pivot line, connect through condensate flowrate of 2 Nomenciature al Neel a ae hee 5,500 Ib/h to find 4 in. as the closest size, 1» Nominal pipe size, bated on velocity w, in, Do Nominal pipe size, based on 60 fs in References ho Canderete enthaipy at wptream’precure, P,, | F “Pens Mana Co, rene, No Gey, el, po th, Eandensate enthalpy at end-pressue, P, Btutb || Pu. 1 Sh Conte Vara, che Bro. May, 284, ‘4h Latent heat of vaporization at P., Btub ae @, Flash-steam volumetric flowrate, At "New flash-steam veloity, fs tue Flash-steam specie volume at Py, 1b Wy Condensate formed at P, Tbh W, Flash steam formed at P,,lbvh Flash steam, wt % s "7A QUICK ESTIMATE FOR CENTRIFUGAL-PUMP EFFICIENCY Alejandro Anaya Durand, | sttuio Mexicane dei Petrsieco | power (ie., brake horsepower) | needed for & centrifugal pump, the | design engineer must incorporate the expected pump efficiency inthe calcu lations. This value is provided in the pump performance curve, However, in the early stages of system design, itis convenient to estimate a value for the effcieney. Final values, of course, will depend on the pump that is spect fied, at the operating conditions that will actually be encountered ‘The accompanying ehart provides a | visual means of obtaining this esti- mated efficiency. The curves are based on the following equation [1] | n onder to culeulate the necessary B = 80 - 0.2855H43.18 x 10°QH 88 x LG + 5.39 LOE 6.39 WW QHE +40 108G ump efficiency, flow, gal/min H = developed head, ft Range of application ‘The plotted values agree wel reported values from pump-perfor mance curves, especially over the range: @ = 100 to 1,000 gal/min H = 50 ta 300 | For flows in the range 25-99 gal/ min, a rough estimate ean be ob- tained using the determined effcien ce at 100 gal/min, und then subtract: ing 0.35 (pereentage points) per gal! min, times the difference between 100 gal/min and the low-flow gal/ min figure, To begin the pump-specification process, use this chart and then derive the pump’s power rating sin stoxe Head (tt) Brake horsepower calculation | horsepower for a flow of 600 gal/min of ‘To use the derived efficiency value in a brakeshorsepower (Blip) computation, convert it from percentage figure to a decimal (ie., 50% = 0.5). Then use the following standard equation | | where enon. is the specifie gravity of | the material being pumped, at the ap- propriate temperature Example: Caleulate the pump brake 18 water at OU'F (sp.gr. = 1), assuming a head of 60 ft. Solution: From the chart, obtain an value of 72.5%, convert to nd calculate: (3,980 x 0-725 = 1045 Reference 1 Branan Cart Th Process Engince's Pocket Hondbcok (Out Publishing Co, Hothmer Unusual problems with centrifugal pumps Solutions to these puzzling ump problems may help you to resolve some of the difficufies that you encounter with pumps. 8. Yedidian, Centtitugal Pump Consultant ‘ first sight, a centrifugal pump seems to be one of the simplest of machines. In practice, however, itis capable of posing an enormous spectrum of difer- ‘ent problems [/]- Occasionally, one comes across problems that seem to defy everything we know about centrifugal pumps. We shall report on four such eases. Case No. 1 —The twin pumps In a chemical process plant, two identical pumps were in stalled side by side, to iransfer liquid from the same source into the same pressurized eontainer. Each pump had been provided with a separate suction and ¢ separate discharge line. Also, the two pumps were never used simultaneously. While one ran, the other served as a standby. Everything about these two systems seemed tobe ident- cal—except that one pump performed perfectly, whereas the second operated with great noise and vibration, The troublesome pumping system was dismantled several tines, bbut nothing wrong could be found ‘The author realized that the successful pump’s pipe loop had a 2-in-dia. discharge, with a reducer connecting directly toa 1%n. pipeline. However, the troublesome loop had a 6ft length of of 2nd, pipe connected to the pump's discharge, and only after this length was the line reduced to 14 in, When the 2in-dia, section of pipe was replaced by 134n, Pipe, the probler-causing pump operated satisfactorily ‘The reason can be explained by referring to Fig. 1. Up toa certain ertical fowrate, @., the net-positivesuetionhead (NPSH) requirements of a centrifugal pump increase, ap- proximately, as the square of the fiowrate [2], Above d., however, the NPSH requirements start to inrease ata mach faster rate — shown schematically in Fig. 1 Inthe particular case described above, the frictional losses in the pipelines constituted a very significant part ofthe total heed against which each of these two pumps had to operate, Inthe discharge line that consisted exelusively of 1! n-dia, piping, the resistance to the flow was adequately high, This kept the total head — against which the pump had to oper. ate— well above Ha the critieal pumping head. This, in turn, limited the flowrate to well below @.. Inthe other pipeline, however, the reduced resistance of the 2in-dia, pipe section, brought down the total head to well below Hz. This increased 19 Flow, 0, vs. NPSH the NPSH requirements of the pump well above the avail able NPSH. Consequently, cavitation developed within the ump. This, in turn, gave rise to the noise and vibration, Case No. 2—Head In a storage tank A somewhat related case was encountered with a pump installed in an oil depot. The pump transferred fuel from a storage tank to oil-delivery trucks. Whenever the ol level in the storage tank was low (Le, when the available NPSH was low), the pup operated satisfactorily. However, when the storage tank became full (.e., when the available NPSH was high), the pump operated with extreme noise and vibration, ‘The total head against which a pump operates is defined ag the difference between the total head existing at 2 pamp's outlet, and the total head available at its inlet. In this particular application, the discharge head tended to be pract- cally constant. This meant that an increase in the available NPSH automatically redueed the total head against which the pump had to operate ‘Thus, in acsordance with what has been explained in comneetion with Fig. 1, the author coneluded that when the storage-tank was full, the pump operated well below thecritical head, 2 This, however, meant thatthe pump dalv fered a flowrate significantly higher than @.. However, the flowmeter thet vas installed in the pipeline, as well as measurements of the time required to fil a a fuebtruck’s tank, indicated that the flowrate was well below @.. There seemed to be no soltion tothe puzzle However, the author recalled a study he had made earlier cn the effects of flowrate on NPSH requirements [J], Ae cording to that study, the most important factor that deter- nines the NPSH requirements of a pump at given speed is the rate of flow through the impeller. The flowrate through an impeller is usually slightly great er than that through the pipeline, owing to leakage through the wearing rings. However, if @ wearing ring is missing, ‘Curve shows NPSH Curves 8,6,D, show total head lea leo Ea zo Eo too S fea jo f : f 4 o ob et ts Pauses ago nara ts SCE AS NE rere re Wig Ko | i | | 20 this short circuits the impeller discharge to the impeller eye. In such a ease, the flow through the impeller may easily be 0 to 40% higher than the flow through the pipeline. In our particular case, this would have brought the total flow through the impeller well above the critical flowrate, Qe ‘The pump was opened, and the front wearing ring was found to be missing. A new wearing ring was installed in the casing, which solved the problem. Case No, 3—A recessed-hub Impeller ‘A pump that had operated satisfactorily at the NPSH values presented in Fig, 2, Curve A, showed satisfactory performance at 40 gpm, when operating at an available NPSH of 5 ft, The same pump, however, continually failed at tuch lower flowrates (Curve A, dashed line), although prev ‘ous tests (Curve A, solid line) showed that it could operate, at these fowrates, at NPSH-values significantly lower than the available 6 ft Tts performance at these times is shown by | Curve B in Fig. 2 (4) In order to eliminate this problem, the impeller was re- placed by another of identical design. This time, the pump produced satisfactory results (Fig. 2 Carve C), But the (question still remained: Why did the first impeller fail? A close inspection of the two impellers revealed that the failing impeller had a recessed hub, as shown in Fig. $ at A, while the second impeller had a solid hub (with no recess), (The same series of tests was repeated with a second, identical pump, with the same results) In order to verify whether this difference in the castings was the real cause of the observed differences in perfor mance, the recess of one of the impellers was filled with | epoxy and redrilled, as shown in Fig. 3at B. This immediately | restored the full suction capabilty of the pamp, as seen by the test results presented in Fig. 2 by the dashed Curve D. ‘The tests were performed on pumps ordered by a custom- et, and there was no time to find out what caused the above effect. It seems, however, that the balancing holes acted here as resonators, The natural frequency of such resona- tors depends upon their length, Most probably, the shorten- ing ofthe balancing holes brought their natural frequency in unison with the periodical shedding of vortices, which are known to appear at low partial flowrates. This, in turn, ‘caused the early appearance of eavitation ‘This problem occurred ata pump-manufuevarer’s plant 4), bbut such a case can easily occur in the field when a worn-out impeller is replaced by @ new one. This is especially true when the spare part has not come from the original source, Case No, 4— The Inconstant pump This problem, too, was encountered at a pump factory [5], and is also one that ean sometimes occur inthe field. ‘A pump designated to operate at 50% ofits best efficiency point, when provided with 3 ft of NPSH, was tested at a constant value of NPSH and found to operate perfectly. The ‘pump test was then rerun in the presemee of a witness. This time, however, the pump failed complately. ‘The pump was tested still another time, this time at the constant specified flowrate and variable NPSH. The test produced a twolevel eurve—Curve A of Fig. (5) In view of the unsatisfactory and confusing results ob-tained from the tested purmp, both the installation and al test | procedures were carefully scrutinized, However, there was no clue as to the source of the observed inconsistencies. As a final check, the entire loop was put under vaeuurm, and held for 20 min. The whole pumping system was proven to be adequately airtight. AS a last resort, it was decided to find if che problem could have been due to air dissolved in the pamped liquid ‘A vacuum pump, located at the top of the airtight suction tank, was allowed to ran and, with the aid of regulating valves, air removal from the tank was adjusted to such arate as to keep @ constant vacuum of 15 in. Hg. After holding this vacuum for about 15 min, to allow a part ‘of the dissolved air to escape, the pump was tested again, Be OTP Ta PSH, fe 1618-20-25 26 Figure 4— Time efects of cissolved ai cn pump's NPSH OF 4 ee 02a 6 Te ae PSH, fe Figure §—How dissolved ai causod stil another elute ‘This test produced satisfactory results (Fg. 4, Curve B). Using the same procedure, the test was repeated in the presence ofa witness, and the pump failed again! This time, the head vs. NPSH eurve had a shape as shown in Fig. 5 by Curve A. & series of additonal tests were then carried out, but the results were never consistent. In one particular case, they even produced a eurve such as that in Fig. 5, Curve B After a significant amount of testing, the answer to the mystery was finally found ‘The observed inconsistencies inthe tet results turned out to be actually due to the air dissolved in the liguid. Tn addition, however, they were found also to depend upon the number of steps in which the NPSH was lowered to 3 ft as well a5 upon the time lapse between two sets of readings. ‘The first (preliminary) test was carried out at a constant NPSH, andl varying flowrate. The tank had been evacuated to 27 in. Hg, and kept under that vacuum until the test engineer was ready to make the readings. Tis took enough time to allow the vacuum pump to remove any excess of ait 2 that had been liberated from the water. Consequently, the pump performed as expected. With regard tothe tests at constant fowrate and varying NPSH, the time effect can be explained as follows: The initial removal of air was carried out ata vacuum of 15 in. Hg. Consequently, below this pressure, ditional air was being liberated from the liquid this ar appeared as.a mass of tiny bubbles, dispersed throughout the volume of water. ‘This lowered the specific weight of the pumped mixture. ‘The amount of air liberated each time the pressure had been lowered depended, of eourse, on the magnitude of the drop in pressure: The greater the reduction in pressure, the more air was liberated When the reduction in pressure was carried out in many small steps, as in the case of Fig. 4, Curve B, only a small amount of air was liberated with each step. This allowed enough time for the vaeuum pump to remove the liberated air during the periods between the measurements. There- fore, the centrifugal pump produeed satisfactory results, When the measurements were carried out in large steps, as in Fig. (Curve A) the rate at which the air was liberated rom the water exceeded the rate at which it was removed by the vacuum pump. This resulted in a gradual inerease in the amount of free air dispersed throughout the pumped liquid, and caused a continuous drop in pressure. Sometimes, at an intermediate number of steps, a state of| equilibrium cecurred between the rate at which air was liberated from the liquid, and the rate of air removed by the vacuum pump. In such a case, the pressure readings re- ‘mained constant, but were lower due to either a reduction in the specific gravity of the liquid eaused by the presence af ait ‘bubbles, or partial air blockage of the impeller inlet, or both U1). This produced a bilevel curve, like Curve 4 in Fig In the case shown in Fig, 5 Curve B, the available NPSH ‘was lowered, in large steps, down to about 8 ft This caused the air to be liberated from the pumped water at a higher rate than that which was removed by the vacuum pump, Thi caused a significant drop in the total head. Below 8ft NPSH, however, the mechanic started to reduce the available NPSH in significantly smaller steps. This enabled the pump to start to recover a part of the ost head Roy V. Hughson, Biitor References 1. Vedi, S. “Centityga! Pump Problems, Caves and Cues" Penn Well Book Go, Tulsa, Ota. 1a 2 Yediia, S; Some Observations Relating to Suction Prforsance of Inducers ard Pam, are Zags Sept 4 Nesidah, &, Effet of Scale ant Sued on Cavitation in Centitugal mg, ASME Syinpesium oa Fit ieshinie the Pectacaee toys 19 4 Yedidah, §. 4 Poesble Explanation to Sone Puan Catton Pa omens, ASME Cavtaton dnd Palypace Fw Foray GAG 8 Yedhdt, S Time Mets of Air on NPSH Tens, ASME C DMutinhate Flow Por, The author & Yesidiah ig centrifusal pump consutan 9 Sure, Kee Sane Ste tc sein eager air algo arta Hang ares ara the author gf -CoutePa roblems=Chases und Cire pabnhed by. Pon Well Boake He ss'memter af the meray Soa Mchanial Engineers sedis a nogetcred prokccenalCheck pump performance from motor data \iGanapathy* Piss names tte need tigre the ow or head af pumps or fans Such information is needed to eheck per: formance or to see if meter readings are aceurate; for meters, the flow orifice right be plugged oy the meter might be incorrectly calibrated Motor data Eleetricalcurrent readings provide a good indication of the Now. The following analysis is developed for pumps, but it ray also beextended to fins, since they | aah & 400 have simile eharuetristi eh ‘The power consumed by pump is: or Sf =f Bur = QSPUTE) a) | Eo ‘The head may be related tothe differ ential pressure by He WaPip Converting Ea, (1) into kilowatts P= 0.0OnEQAPIE, ofthe motor is given by @ ‘The power out P = O00TT2EIeo8bE, @ Equating Eqs, (8) andl QAPIE, = QHpIMAE, = 408BTe086Ey (5) 1 pump data ean be related to motor data, ‘Comments ‘The folowing shold be kept in mine 1, Motor elfcieney does not vaxry much with load. ‘The variation is probably 2-5 for mediumsized motors {60-500 ‘np) and for practical parpnses can be considered eonstant, So can the power factor and voltage ADCO Iatr, n Nomenclature Bur Beale hrseposver. hp E Voltage Ey, By Elfcieney of pump and motor, a Head, st i Current, amps, Pp Power consumed, kW @ Flowrate, galimin Greek letters AP Pump differential, pst » Density, Ibn? cosh Power factor Figure 1 — Typical characteristic curve for 2 mulstage centitugal pump uses to prepare table 3 | a LX] ele eed vi estos Pt TTT ote 300- ose Feo cans. | | | Cure ope 8 8 220] HOT) 00 5a 00 0 owes, sli Figure 2 — Using motor data, this plot can be constructed and then used to check pump an meter performanceTable — Typical operating data for a pump 2. The head, H, and effcieney, Ey, of the pamp vary para- bolicaly — see Fig. 1 for a typical pump curve, For a given fluid and flowrate, Qip/448, is a curvilinear function of Q. ‘Thus, Eq. (6) may be written as £Q =f) © For a given pump and motor, if B, cose and Ey, are known, @ can be determined as a function of ewrrent. For the pump curve in Fig. 1, we shall determine Bq, (6) 23 First, atable is made of @, H, and Hy. QH/B) is caleulated (see the table) Assume E = 460V, Bm = 0.95, coss) = 0.9 and p = 62 Ibi, Substituting these data in Eq. (6: QHIB,, = 4.08 x 460 x 0.9 0.95 x L482 = 3,6801(7) ‘By plugging in values from the table, ean be found as 2 fietion of @. This is shown in the table and plotted in Fig. 2 Using Fig. 2, one can easily relate the flowrate (or head ot pump efficiency) to motor earrent, This sa eonvenient way of checking the performance of @ purmping system or of double checking meter readings. Thus, having a relationship between flow and current permits monitoring of any significant devi- ations. For example: What if at some fowrate, the meter shows a current reading that differs from that given by the curve forthe samne flowrate? Then either the owmeter has to be checked Gf the variation is greater than about 3%) o the pump has to checked for misalignment, overheating, ete, (The same method may, of course, be used on fans.) ‘The equations take into acount fition and heat losses (of both the pump and motorShortcut methods for determining Notbert J. Bel Phis atl ofersshortet equations and nomographs that determine exact optimum pipe sizes, and approximate ws ‘wes (usally within 107) of the preg yraiont fo Under turbuient-low conitons, The pressure drop ‘The basic pressure-drop equation (the Darey formula) for low ‘hough a pipe can be written ast Ap = 500 fLpv%id @ The usual practice is to work in terms of the pressure sradient, whieh is the pressure drop per linear unit of pipe, For Instance, substituting L~ 1m into Eq, (1) yields Sq = 500 fovtid @ ‘The task ofthe piping designer sto size the line so that the flow of Suid will create a pressure gradient that is near the economical optimam, The economic optimum conditions take into account both the capital cost of equipment and operating «osts of pumping, energy, ete Forsny given low, the optimum pipe diameter, doy, willbe related to optimum pressure gradient by modifying Bq, 2) so SCP Bran Ca, 1am Sth Frnt Avene, Aunt, CA 800580 timum line sizing in SI units that both variables d and Apa are at optimam conditions, Dosim values for both optimtm pressure grarient and the velocity linit for a new potroloum relinery are given in the table. However, these values may be used for other chemical processing applications as well In many instances, designers try to stay below these pres- ssure-gradient criteria, This is really not necessary; rather, ‘coming close to the optimum pressure gradient” is better. ‘Therefore, one should select the pipe size that — after satistying the necessary structural requirements — has the standard inside diameter closest to dys. The resulting, pres- ‘sure gradiont will be closest to the optimum pressure gradient. But, in some eases, the actual pressure gradient eould some. what exceed the calculated optimum pressure gradient, Fluid velocity For cincular pipes, the average velocity is 1.278 x 0? gid? @ Pipe designers normally work with velocities that will not | reate undue vibrations, and erosion in the piping. Thus velocie limits, such as those provided in the table, are used as a further criterion for optimum pressure gradientsin pipe sizing. 1,000 Lis 0.00001 0001 0.001 001 on iomte | Flowrate,m3/sor Ls Nove: Vates or ors aint re aes ony for wat w= Vinal and =, 000 gm Figure 1 — Determine optimum flow conditions from this char, but for fluids other than water, use correction facor fom Figure 2 24eee —_— Shortcut pipe-sizing equation Substituting Eq, (3) into Eq. @), and solving for dope, we get: doe = 959 [fq?/SP ope? suggested f= 0.2161Re02 where He = pain, Nomenclature D Nominal pipe size, in, a Pipe inside diameter, mm f Friction factor, dimensionless b Straight pipe length, m Ap Pressure drop, Ps Apa Pressure gradient, um 4 ‘Volumetric flowrate, m/s Re Reynolds number, dimensionless » Average fluid velocity, mis B Dynamic viscosity, (mPa)(s) e Fluid density, kgimn® Subscript opt Economical optimum @ | Aowrate in ‘To use this equation, an estimate of the friction factor is | Units needed. For turbulent flow in elean steel pipes, Simpson* has | ® | As shown in Fig. 1, the optimum line to carry 0,02 m/s 20 | Us) of water ata pressure gradient of 100 Pulm comes out toa Dominal size of 6 in. (152 mm). For convenience, a scale for Lis has been added to allow input in these flow Alko, velocity-limit lines have been pleted. In order to use Fig. 1 to determine pressure gradient for fluids other than water, it will be necessary to use a pressure- ‘gradient maltipler (PGM), according to the following equation: AP = (APrawatert!%p/10008 o ‘Table — For dosign purposes, tis not et ‘essential to "say below’ the pressure-oradiont criteria Ligiid Type atline __Isehnlagn fal Welocyinit ms Wer Hee wm 4s tate ms 4S Hydration: Pump sein, 5 (O81, Todas" Iaxinain = 25 Pam suction wt Pam seen, moo tore ike, | ame chars, mle | Pg scar, wr s2, Bei ‘rain = | ein Inet atone se Rate Ont Pott s foun fe promt patent i Bos] 19 05) 03| oF 15 Ia fin POWe 2 Dora, bur 10 705 Viscosity (balls) Figure 2 — Use this chart to find the comtacton factor for prescure-gradient values obtained from Figure 1 age = 516 HOSEA MITA (Arg) 28 6 Shortcut nomographs Nomographs based on Eq. (6) can be constructed that permit auiek sizing of pipes for different liquids. For instance, Fig. 1 «an be used for water at normal temperatures, when the Aid onsity is 1,000 kgim?, and the visoasity is 1 (mPa}G), which is | also equivalent to 1 eP. ‘Sige, LL, Cho Br, Api 4 190, 5p A, ‘To assist in using this equation quickly, another nomograph (Fig. 2) was constructed, in which POM = 02%p/1000)08, Therefore, in order to find a pressure gradient for any fd whose viscosity isnot equal to 1 (mP&s) andor whose density is not 1,000 kg/m’, we must multiply the values from the two omograptss 4pm = (Par, 1 X PEM g 2 ® For example, for a fluid having a viseosity of 10 (mEaXs) and with a density of 1,200 kgim*, we find from Fig. 2 that PGM = 182 25Operating performance of reciprocating or positive-displacement compressors Miche! Polonyi" ir pressure increases when air flows into a system, and | \decreases when air lows out. Using this principle, here is | a simple method to determine air-compressor performance and compressed-air load for plants using reciprocating or positive- displacement compressors (see figure). Accuracy for this testis approximately 10%. ‘This technique has proven useful in eases where (1) there is no flowmeter for the compressor; (2) the compressor is fst put into service; (3) the unit is running for a longer period than normally expected, or does not stop (in such eases, there is usually no way of telling if this eondition is due to higher air consumption by the system, or lower compressor yield); and/or ) the plant air load is not known. Plant-air or service-air applications are almost exclusively supplied by using reciprocating or positive-displaeement com- pressors. Since these compressors pump an almost constant amount of air, they must exceed the needed plant-air eonsump- tion in order to maintain the required air pressure Satery valve Tovsers Tenparature From other J oF Prosare ne campresor Figure — This compressor pedormance lest can ‘be run for any given pressute interval or ime duration Accordingly, air pressure varies between two setpoints, At the lower setpoint, the compressor starts pumping, aad at the higher pressure, i stops. These setpoints are adjusted via an ‘10 West Tht, Now Va, NY 10 26 air-pressure switch, but they eannot be too close — otherwise the compressor would start and stop too often. On the other hand, ifthe setpoints are too far apart, pressure fluctuations Would be too great. Normal values are in the range of 10% to 20%e, However, in order to perform the test, its not necessary to wait to reach these setpoints (the extremes of the system). Any pressure or time interval in which data can be taken accurately will suffice. Ifthe volume of the system is known, one ean measure the time interval it takes for the pressure to reach a specific value ‘Then apply the following equation: Flow = Volume x (P;~ P,)Py * time where: Volume = System volume Py Discharge (atmospheric) pressure P, = Initial pressure By Final pressure ‘time = interval to reach the final pressure Guidelines to pertorm the test 1. Determine the alr volume of the system, This volume includes the air receiver and the piping. A measuring tape should be sufficient to determine the dimensions of the air receiver. Piping volume need only be estimated. 2. Use any calibrated pressure gage to perform the test. (Te more accurate the gage used the better the test results.) With the compressor runring, measure the time interval for either a fixed pressure change (try in the range of 10-20), or ‘wait until the compressor starts and stops pumping and use that pressure range 3 Perform two tests. The first should be withthe eompres- sor running, which will show the amount of excess air in the ayatem. When the compressor is stopped, perform second test that will provide the data to determine the aetual air being used. The compressor yield is equal to the sum of the two calculated flows. 4. If more than one compressor is running, perform the test cn the one that starts and stops, and move on in sequenee, to the next one, In caleulating, make sure to use the volume for the whole system. The yield for each of the compressors requires the solving of m equations, where n is the total number of compressors actualy in serve at the test time 5. Accuracy of results can be inereased by including a tem- perature correction factor, so that the corrate flowrate is equal to the measured low times the ratio of absolute ambient ‘temperature to absolute compressed air temperature. JP... Jr...Protecting against backflow in process lines Backflow —a flow-reversal, by which liquids _ get into places they do not belong—ean be serious, Here is how to prevent it. | ‘Stephen J. Emery, Hersey Products, Inc [7 Backflow is most simply defined as a reversal of the normal direction of flow of piped liquids, It happens when, for one reason or another, upstream pressure falls, below downstream pressure. Backflow can result from a pump failure, heavy upstream drawoff, downstream pressure rise, ruptured lines and many other conditions ‘What makes backflow a major cause of concer is that reverse flow of liquids often has highly undesirable consequences. In mixing tanks, for example, backflow of | mixed product to feedstock holding-tanks can result in contamination and costly waste. Backflow can also create hazardous conditions, particularly where toxic or explo- | sive fluids are involved. Backflow presents its_most serious problems, of course, when human health may be affected. For exam. ple, where a single water-supply system furnishes both potable water and process water, great care must be taken (0 protect potable water from possible contamina: tion. Backflow of water containing impurities or toxins into the potable water supply is, obviously, intolerable. In fact, safe drinking-water legislation and osia (US. Occupational Safety and Health Administration) re- quirements have made it mandatory to provide backflow Protection on all cross-connections between potable and onpotable water supplies. Ideally, water or other fluids that should be kept separate would have separate supply systems, In practice, this is often not practical. In some ‘cases, certain facilities, equipment and processes cannot operate satisfactorily without direct cross-connections. The problem is how to maintain control over fluids, so as to prevent backflow or backsiphonage (which is pressure reduetion at any point in the water distribution system to atmospheric or lower, that could draw water through the system in a direction opposite to normal), Open air gap: Best protection is generally agreed that the best protection against backflow is the open “air gap.” An air gap is created by interrupting” a piping system. Upstream piping is sepa. rated from downstream piping by an air space (the recommended gap is at least twice the diameter of the supply pipe). Typically, fluids pass from the supply line toa holding tank and thence to the downstream piping system. The air gap between supply line and tank makes Open air gap is best, but it can be rendered useless by bypass piping Be eet i IN ) metric loop provides protect backsiphonage but not backpressure 277 8 | Vacuum broakor provides some protection, but isnot recommended for hazardous liquids ~vscuum breaker Fig. 3 ‘Single check valve is entirely dependent ‘on the leakproof integrity of the seal Double check provides more protection than single, but still dopends on seal integrity Double check with atmospheric vent Fig. 4 Fig. 28 backfiow into the supply system virtually impossible. The drawback to the airgap method of backflow control is that line pressure is lost at each air gap and ‘must be reestablished by means of pumps or towers. Ina plant with many cross-connections, this would be an extremely expensive proposition. Moreover, because the open air gap is often impractical, codes requiring it may sometimes be ignored ‘in many cases, bypass piping is installed while a tank is being repaired or refurbished (Fig. 1). The result is no protection atall. The existence of open tanks may also be undesirable, particularly where fumes or airborne con- taminants pose problems Barometric loop Other methods of backflow prevention include the barometric loop (Fig. 2), which operates on the principle that a perfect vacuum can lift water only 34 ft. This method requires the construction of a piping loop that rises 85 ft above the highest fixture in the piping system. While highly effective against backsiphonage, the baro- metric loop will not provide protection against backpres- sure. Itis also cumbersome, susceptible to freezing, and vulnerable to bypass arrangements. Vacuum breakers Vacuum breakers (Fig. 3), of both atmospheric and pressure type, also are used to control backflow. They are at best a partial solution, offering some protection against backsiphonage but none against backpressure. Single check valve can leak Another method for controlling both backsiphonage and backpressure is a single check valve (Fig. 4). In theory, 2 check valve should provide adequate protec tion, Its, however, entirely dependent on the leakproof integrity ofits seals and, as such, is not acceptable for use in situations where backflow into potable-water supplies could include hazardous, noxious or otherwise undesir- able materials. Double check valves: A more effective method of protecting against back- flow is the double check valve (Fig. 5). Should upstream pressure drop below a certain level, the two check valves close to protect against backflow. When pressure is restored, the valves open. ‘As with the single check valve, the effectiveness of double check valves depends entirely on their sealing integrity. Should the seals develop leaks, undetected backflow could occur. For this reason, double check valves should only be used where backflow could be objectionable—but not hazardous—and should never be used where backflow could cause a potential health hazard or other serious condition. artial answer Double check valve with vent This device, generally known as a continuous-pressure type, is basically a double check valve with an atmospher- ic vent located between the checks (Fig. 6).Ina backpres- sure condition (downstream pressure exceeds supply pressure) or a backsiphonage condition (supply pressure drops to atmospheric or lower), both checks close and