. this purpose are receiv- Using DIERS two-phase equations to estimate tube rupture flowrates The method is more accurate than the traditional ‘separate’ phase approach B. Elliott, Stone & Webster Engineering Corp., Houston, Texas SME Section VIII and API 521 discuss the A need to design and protect heat exchangers against overpressure due to a tube rupture, API 521 specifically addresses requirements and approaches to solve this problem. The first solution is to raise design presture ofthe low-pressure side toa minimum of two-thirds of the design pressure ofthe high-pressure side, Howover, this may not be desirable due to economics or other design reasorie, "The second solution s toad a relieving device, like a relief valve or @ rupture dise. To size this relieving device, flow through the tabe rapture from the high-pressure side to the low-pressure side of the heat ex. changer must be calew. lated (Fig. D. Equations developed by the Ameri= can Institute of Chemical Engineers (AIChE) for ing widespread support. In many instances, high-pressure liquid flashes to a much lower pressure through the tube rupture, thereby requiring a method for calculating two-phase flashing flow. Equations that have been developed in the past two decades for two-phase flashing flow in relief valves and flare header piping can be used for determining flowrate through the tube rupture. Historically, the process industry has sized relief valves whose inlet and/or discharge is two-phase by separating out the phases into pure liquid and pure vapor phases, determining a required area for each phase, and adding the two areas together for a total required relief valve area. While this has no theo- retical basis, it was seen as giving a reasonable “ball. Fig. 1. Tube rupture diagram, —___ — park” estimate, without going into complicated math- ematical and semiempirical techniques. There is evi- dence that using this method can lead to incorrectly sizing relief valves Over the past 15 years, The Design Institute for Emergency Relief Systems (DIERS), an arm of AIChE, has been looking into the area of two-phaso flow in relief valves and flare headers. It has devel- oped various caleulation methods to deal with these problems. These methods are receiving widespread support and use within the industry. Since a tube rupture, as defined in API 521, is just flow through a tube (one side of the break) and flow through an orifice (tubesheet side of the break), the | DIERS two-phase mass flux methods can easily be applied to these prob- lems when there are two-phase or flashing fluids, DIERS equations for physical properties of the fluid. A generalized equation for nozzle dis- charge is used for sizing the relief valves for two-phase flashing flow. Hence, it can be used to approximate flowrate through the DIERS mass flux equations have also been devel- oped for pipe flow with choked discharge. These equations can be used for the tube side of the tube twice the flow through the tubesheet, assuming that the flowrates will be reasonably close. This would Jess accurate) than the other method. HYDROCARBON PROCESSING /AUGUST2001 49 Bad FEequations assume steady-state and equilibrium condi- ons, and they do not account for transients. An esti- nation of peak pressures caused by shock waves that boeur immediately after the tube rupture would have to be evaluated using a different method. ‘The method for estimating tube rupture flowrate is as follows: @ Determine mass flux through the tubesheet side of the broken tube. ‘© Determine mass flux through the tube side of the broken tube. ‘ Calculate tube discharge area. @ Calculate flowrate through the tube rupture. ‘To determine inlet composition of the relief valve on the low-pressure side, calculate any additional flashing and mixing that occurs after the rupture. Methodology using DIERS HEM equations. There are various specific DIERS methodologies for calcu- Jating two-phase flashing flow. The specific equations presented here were developed by Leung.! Their Strength is ease of use and that they only require inlet conditions, However, they are not necessarily as accu tate with fluids close to the critical point. An alterna- tive sot of equations was developed by Simpson? These equations are stronger when dealing with critical flu- ids, but require more physical data.* ‘Determine mass flux through the tubesheet side of the broken tube. ‘Stop 1: Calculate the compressible flow parameter, o: ‘The DIERS methods treat the two-phase fluid as a single homogeneous “compressible” fluid. Leung's method does this by calculating a compressible flow parameter, o. This parameter is a measure of the flu- Jas “compressibility.” The larger the value of «, the ‘more the fluid behaves like a compressible fluid. Values for o fall into these categories: Flashing flow: worl Gas/vapor flow: onl Nonflashing flow: O 0.6.8 At values < 0.6, Eq. 3 should be solved numerically. Step 3: Calculate the dimensionless mass flux, 0*: If the flow is critical: G=n/ot © If the flow is not critical: Gra fsfotan+(o-0e-nlf hoon 2}+1] © and n= Ps/Po oO where 'P; =low-pressure side “relieving pressure”, psia P)= high-pressure side pressure, psia Step 4: Caleulate actual mass flux, G (b/hr/in ?): Gy =17006* (oo) © inlet pressure (psia) overall density of the two-phase fluid at inlet, Thin? Continued etDetermine the mass flux through the tube side cof the broken tube. Flow through the tube side can be / approximated by using the DIERS mass flux equations for horizontal pipes (or vertical pipe, ifthe exchanger is vertical) For consistency, we have continued to use Leung's methods. Use the samo w as was calculated in the previous Step 1. The method consists of three equations, one for tube inlet, one for the flow through the tube and one for tube discharge. If an exchanger doesn't comply with the API 2/3 rule, discharge from the tube will almost certainly be choked, therefore, the discharge nozzle equation can be assumed to be critical flow. ‘Step 1: Inlet “nozzle” © where: n, = the ratio of the pipe inlet pressure to the “high-pressure side” pressure. n= Pal Po ao Step 2: Pipe coca _o _,(-e)y+e WO Ge wo er -softcopere( a] as (Q-a)ny +002 J where: nz =the rato ofthe outlet pressure to the “high- pressure side” pressure. ny Bi a2 where: f= friction factor = tube length D=tube diameter Step 3: Outlet nozzle: of = Te. as) ‘These equations are combined and then solved numerically to obtain a G*. Step 4: Calculate actual mass flux, G (Ib/hr/in.), Gq = 1,7006* (Pyp)# aay where: * Po = inlet pressure, psia wwerall density of the two-phase fluid, Ib/ft? Calculate tube discharge area, A, (in.*) Ay = nD? ap where: Dy = tube inner diameter, in, Or use a cross-sectional area cited in a reference. Caleulate flowrate through the tube rupture: We 4,(41+G3) as) At this point, total flowrate through the rupture has been calculated. The flow then needs to be eval- uated at the low-pressure side pressure (presumably relieving pressure) to determine if there is any addi- tional flashing. This will determine inlet conditions of the relief valve being used to relieve the tube rupture flowrate. Sample calculation. As an example, consider a slurry/HP steam generator in an FCC unit. The high- pressure (boiler feed water) side has a design pres- sure of 600 psig. The low-pressure (FCC slurry) has a design pressure of 200 psig (PSV relieving pressure = 220 psig). The tubes are 1-in. 10 BWG tubing (length = 16 ft, inside diameter = 0.782 in, inside cross-section = 0.4208 in, friction factor = 0.00575). Calculate flowrate through a fall rupture of e single tube. xo =0 p= 49.599 Ibyn? py = 1.992 Ibn? pr =49.599 1hjn? Cy =117 Beu/lbF 1) =9488R Fy =614.7 psia L=7283 Bey/lb Continued THE GENUINE INTELLIGENT STEAM TRAP REDUCE COSTS... of maintenance and inspection, Up to 15% eneroy saving. STOP LEAKS...!! Continuous leak and failure detection on steam traps and valves. “BiTherm SmartWatch™™" A New Concept of Intelligent Valves Visit us in CHEM SHOW 2001 - New York 23 - 25 October / Booth # 2473 INTELLISENT CONTROL SYSTEMS (1.C.5.) Web: wwwbitherm.com Email: [email protected]| Distributers in: Argentine, Australia, Eaypt, Tly, Korea, USA, Portugal, Singapore, Brazil, India. Circle 78 saflow parameter, w:(Eq. 1): = (0 49.809/1.302/1- 614.7/0.7728.* 1.992) + 0.28505° 1.17" 948.8 614.7" 49.539 [[v1.ss2- y49.539)/728.3)" = 6.29 Since w > 1, the flow is flashing (which is as expected) ‘Step 1: Caleulate compressible. (Eg. 4): Step 2: Calculate the pressure ratio, Tig = 0.6055 + 0.1356(In6.29) ~ 0.0131{1n6.29)” ag =0811 Check with Ba, 3: sit? +(.20?~2+625)1-0s14)? +2*625%In(0812)+ 2*6.29%(1- 0811) Song = 0.811 is acurat Step 3: Calculate dimensionless mass flux through the tubesheet rupture, G;* (Eq. 5): Since the actual pressure ratio is 234.7/614.7 0.382 and the critical pressure ratio is 0.811, the flow is critical. Hence, of =081y(6.29)8 Gf =0823 Step 4: Calculate actual mass flux through the tubesheet, G (Ib/br/in.?) (Eq. 8 og G, = 95,820 Ib/hr/in.? : So total mass flowrate through the tubesheet would be (Eq. 16): W; = 95,820 Ib/hr/in.? + 0.4208 in.” W, = 40,821 Ijin, 700* 9.329(614.7* 49.539)? Determine mass flux through the tube side of the broken tube. The compressible flow parameter is the same: = 629 Step 1 9): [e.29ian, +(6.29) ea ay Step 2: Pipe (Bq, 10): 1+ 523 | 280 "629-529; fese-sfu (a > (seem a] —f} 29, 629- zn ‘54 HYDROCARBON PROGESSING/ AUGUST 2001 Step 3: Outlet Nozzle (Bq. 11): Substitute Eq. 11 into Hgs. 9 and 11 to eliminate Gz, and solve numerically (e.g., on a spreadsheet): 78 nzs0art Gj «0390 Step 4: Calculate actual mass flux, G (Ib/hr/in.”) (Eq. 14): Gy = 1,700" 0,190 (614.7" 49.539) g = 56,422 Ihfhe/in? So total mass flowrate through the broken tube would be (Eq. 16): Wp = 86,421 lb br/in® +0.4208 in? We = 28,742 lyr ‘The total mass flowrate is : W = 40,821 + 23,742 = 64,063 Ib/hr Ifflow was estimated by just doubling the tubesheet flow, the total would be 80,642 Ib/hr, which is an addi- tional 25%. This would be a conservative answer while still remaining in the ballpark, and most probably not affecting the size of the relief valve that would need to bbe purchased. ‘To determine the vapor/liquid split for the relief valve inlet, flash the mixture at the relief valve reliev- ing pressure (220 psig). In this example, the mass, fraction of flashed steam would be about 13%, and x, for the relief valve sizing calculation would be 0.13 except that in this example, the hot slurry side will vaporize some of the remaining water, giving a higher x, (perhaps as high as 1.0). This will need to be eval- uated separately. 7 “tee. tone cd PpAiit Ae ar Sj “etn ve Paseo is Salty Devine Chil Brinn, age i oto. * Late 61 enon Corin Onegin Honsgscnus Bulan tig Chats Hon ATCAE Jura Oc 1588 “Lung 6, “Dashy Sas Rea Bre ad ipog he Tro Foae Flom” Chena Regiirg Prgms Deca 8. 280 1 AS ices ii ests a 1 De te epee Sey Rae le Sf wo Phase Fr” a ne. Signa con, rane le Send re sence nnnnennteesrernninmcrtiee