Afzal Haque, Stephen Richardson, Graham Saville and Geoffrey Chamberlain*
Department of Chemical Engineering, imperial college, London SW72BY, UK Shell Expro, 1 Altens Farm Road, Aberdeen AB9 2HY, UK Experiments were conducted on the rapid depressurization of large pressure vessels. Measure- ments taken included the pressure, temperatures at a large number of positions both within the fluid phase(s) and on the wall of the vessel, and composition, all as a function of time during the blowdown process. The systems studied included subcritical and supercritical, condensing and non-condensing. From these experiments, an understanding of the physical processes involved during blowdown was evolved. This was incorporated into a mathematical model of blowdown, and implemented in a computer program. The model correctly predicts all the major phenomena observed in.the experiments, as a function of time. (Keyw or ds: pr essur e vessel s; hazar ds; physi c al pr oc esses) Rapid depressurization of pressure vessels, usually called blowdown, is a hazardous operation, at least as far as large pressure vessels are concerned. The hazard arises because of the very low temperatures generated within the fluid during rapid depressurization. This inevitably leads to a reduction in the temperature of the vessel itself and possibly to a temperature below the ductile-brittle transition temperature of the steel from which the vessel was fabricated. In the release of gas from a vessel at high pressure, the thermodynamic path taken by the gas inside the vessel is one of constant specific entropy if the process is adiabatic. This would result in the production of very low temperatures (78 K) for the release of, e.g. nitrogen at 150 bar down to ambient pressure. This should lead to substantial heat transfer from the walls of the containing vessel, and hence to the removal of the adiabaticity of the process. As far as the wall is concerned. there are a number of competing processes: heat transfer by convection to the cold gas inside the vessel; heat conduction through the wall of the vessel; and heat transfer by convection between the surround- ings and the vessel. If the state inside the vessel, either initially or at any stage in the blowdown, is one in which the vessel contains e.g. liquid condensate with a vapour phase above, there is the additional complication of heat transfer from the vessel wall to the liquid by boiling. This will result in the evaporation of material from the liquid phase and its transfer into the vapour phase. All the time this is happening, depressurization is taking place through an exit pipe that could be removing either the vapour phase or the liquid phase, according to its position in the vessel. It is therefore quite difficult Received 12 September I989 Presented at the First Int. Conf. on Loss of Containment, 12-14 September 1989. London. UK 09504230/90/010004-0463.00 0 1990 Butterwort~ &Co. (Publishers) Lfd 4 J . Loss Prev. Process Ind., 1990, Vol3, J anuary to predict how the system will behave. In process plant, particularly in the petroleum industry, there are many large vessels operating under pressure and containing hydrocarbon mixtures. These vessels may be horizontal or vertical, and have inlet and exit pipes at the top, bottom or sides. Depressurization is frequently necessary, and in an emergency it may have to be rapid. It is under these circumstances that the lowest wall temperatures will be observed. The objective of this paper is to describe some of the work performed to predict the behaviour of a vessel during blowdown. Experimental The physical processes taking place during blowdown are a complicated mix of several phenomena. To determine the more important of these, a programme of experimental work was carried out involving the depressurization of three vessels ranging in diameter from 5 to 110 cm, with length to diameter ratios from 10 to 3 respectively. In the case of a gas-filled vessel, experiments showed that the dominant mode of heat transfer at high pressures was natural convection. This was most clearly illustrated by examining the blowdown of a horizontal vessel through an exit on the axis of the end closure of the vessel. The vessel was 1.5 m long, with 0.27 m i.d. and walls 2.5 cm thick: It was instrumented with a pressure transducer to measure the internal pressure, and = 80 bare wire thermocouples to measure the temperature within the gas space in the vessel and the surface temperatures of the vessel wall, both inside and out. Pressurization was with nitrogen to 150 bar, and depressurization was through an orifice of diameter 6.35 mm. A data logger was used to record all the temperatures and the pressure once every 3 s. Blow- Rapid depressurization of pressure vessels: A. Haque et al. TOP BOTTOM Figure 1 Isotherms (left hand side) and schematic streamlines induced by natural convection (right hand side) in a vertical plane parallel to the ends of a horizontal vessel down to atmospheric pressure was essentially complete after = 100 s. Figure I (left hand side) shows a plot of isotherms in a vertical plane parallel to the ends of the horizontal vessel, obtained by interpolation and smoothing of the raw experimental measurements. It indicates the presence of natural convection with streamlines roughly as shown on the right hand side of Figure I . The isotherms are essentially the same on all such vertical planes within the vessel, indicating that the axial forced convecting component, due to flow of gas towards the exit port, is dwarfed by the natural convection circulating component. Measurements made with this vessel in a vertical position indicate that natural convection also dominates in this case. Apart from the region close to the wall, the temperature gradients within the gas space were quite small and the band of temperatures shown in Figure 2 is typical of those observed during blowdown of a gas-filled vessel. Figure 2 also shows the corresponding variation of the inner wall temperature. This too varied little over the whole of the vessel. Multicomponent mixtures behave in a similar manner at high temperatures, where the mixture is single phase gas, but show a markedly different behavjour once condensation takes place. Figure 3 shows the bulk fluid temperature measured within the same vertical vessel when a 70% nitrogen/30% carbon dioxide mixture (molar composition) was blown down from the top, with initial conditions of 150 bar, 20C. At first, when the mixture was single phase, the temperature fell in an essentially spatially uniform manner within the vessel. After = 10 s, the tempera- tures split into two bands, with those recorded in the top part of the vessel continuing to fall and following 0 -20 Y -? , - 40 z b E d - 60 5 rn . c Q - 80 s 5 - 100 - 120 0 20 40 60 80 100 Time/s Figure 2 Variation of temperature with time during blowdown of nitrogen: the upper band refers to inner wall temperatures and the lower band to gas temperatures; the solid lines indicate predictions made using the BLOWDOWN package 20 - 20 d- , - i i 5 z k 8 - 40 ; I I I I I I I I I I 0 20 40 60 80 Time/s Figure 3 Variation of temperature with time during blowdown of a 70% nitrogen/30% carbon dioxide mixture: the upper band refers to bottom zone temperatures and the lower band to top zone temperatures; the solid lines indicate predictions made using the BLOWDOWN package the lower band in the diagram. The temperatures measured near the bottom of the vessel, however, showed first a rise and then a steady fall. This behaviour was caused by the onset of condensation. At first, the amount of condensate formed was small and as it reached the bottom wall of the vessel, which was still relatively warm, it partially evaporated, driving off the more volatile components and thereby raising the J . Loss Prev. Process Ind., 1990, Vol3, J anuary 5 Rapid depressorization of pressure vessels: A. Haque et al. boiling temperature. However, more and more conde- nsate was formed in the upper part of the vessel, due to expansion. This fell to the bottom, and, as the bottom of the vessel itself was cooled, the evaporation rate fell off and a pool of liquid gradually accumulated. Since the pressure in the system was still falling, this liquid experienced evaporative cooling. Clearly, a very com- plex and competitive process is going on in this liquid pool, but the experimental evidence indicates that eventually evaporative cooling dominates the be- haviour and the liquid temperature falls, as shown in Figure 3. Only a few centimetres depth of liquid were formed during this experiment, but the evaporative cooling was sufficient to reduce the inner wall temper- ature at the bottom of the vessel to - 10C (at the top of the vessel the corresponding temperature was 15C) before the pressure reached 5 bar, at which point the liquid layer, which was now almost pure carbon dioxide, solidified to form carbon dioxide snow. Similar behaviour was observed during blowdown of natural gas/propane mixtures from the 1.1 m dia- meter vessel. However, the much closer volatilities of methane and propane, as compared with nitrogen and carbon dioxide, did produce some quantitative differ- ences. Thus, the depth of liquid condensate foimed when depressurizing a 30 mole /o propane mixture was. in relative terms. an order of magnitude larger. Since propane has a triple point pressure < 1 bar. this condensate remained liquid until atmospheric pressure was reached. at which stage its temperature was close to the normal boiling point of pure propane, indicating that almost all of the volatile components had been driven out. The inside wall temperature at the bottom of the vessel was almost identical to that of the liquid. The extensive series of measurements made on the blowdown of the two larger vessels for various com- positions provided the data base for modelling blow- down. Most of the measurements were made on vertical vessels blown down from the top. and the model described applies specifically to this. However, measurements were made on vertical vessels blown down from the bottom and also on horizontal vessels blown down from the end, but these are not discussed here. Computer model A computer package called BLOWDOWN has been developed to simulate rapid depressurization of vessels. The vessel is divided into two zones: the top zone contains all of the vapour together with any suspended liquid-phase droplets; and the bottom zone contains all of the liquid phase that has dropped out of the top zone, forming a pool on the bottom of the vessel. If no liquid is present the bottom zone is eliminated, but it reappears if condensation occurs at any time. Each zone is assumed to be at a spatially uniform temper- ature and composition; all zones are assumed to be at a spatially uniform pressure. 6 J. Loss Prev. Process Ind., 1990, Vol3, J anuary The continuous depressurization process is re- placed in the computer model by a series of discrete time steps. Each time step is sub-divided into a number of simple thermodynamic and heat transfer sub-steps according to the following algorithm: Select a pressure decrement. Perform an isentropic flash on each zone. Calculate the rate of discharge through the choke. Calculate the duration of the time step and the amount of fluid discharged. Calculate the heat transfer coefficients for each zone. Perform energy and mass balances over the contents of each zone and an energy balance over the vessel wall. If depressurization is complete, stop; otherwise, repeat this process. During blowdown from the top of the vessel, the rate of discharge through a choke is calculated by requiring the fluid to follow an isentropic path, the fluid emerging from the choke with a speed equal to the local speed of sound of the gas-phase component. The speed of sound in the choke, V,, is determined by the simultaneous solution of the equations: Hi = H, + :VIC s, = s, where H denotes enthalpy, S entropy, p pressure, o volume and T temperature; subscripts i and c refer to conditions far upstream of the choke (where the flow is essentially stagnant) and in the choke, repectively; and C, and C,, denote specific heats at constant pressure and at constant volume, respectively. A number of possible situations could arise in practice, since the fluid approaching a choke could be one phase (vapour) or two phase (liquid and vapour) and the fluid in the choke could be in a metastable state or in thermodynamic and phase equilibrium. (The fluid approaching the choke is assumed to be in equilib- rium). Any combination of these phase states can be handled, but as a result of comparison with the experimental measurements, the preferred method is to use one phase or two phase, as appropriate for the fluid approaching the choke, and equilibrium for that within the choke. In all cases, once the state within the choke is evaluated, the mass flow rate through it follows from a knowledge of the density and speed of sound in this state, together with the cross-sectional area of the orifice. Heat transfer in the top zone allows for forced convection I and natural convection; in the bottom zone, it allows for nucleate3 and film4 boiling; outside the vessel, it allows for natural convection2. Energy balances over the contents and wall of the vessel are always carried out using standard numerical methodsj and solved by a matrix inversion methodh. A mass balance over the contents is required only if liquid is present, since mass transfer between zones only occurs as a result of evaporation from the liquid pool in the bottom zone, and sedimentation from the top zone. Figures 2 and 3 show predicted and measured temperatures. It can be seen from Figure 2 that BLOWDOWN successfully predicts the temperature of the gas and of the inner wall. Similarly, Figure 3 shows that it successfully predicts the temperatures in the top and bottom zones when condensation occurs. The package for depressurization has also been validated for natural gas and for natural gas containing up to 30% propane. Conclusion The experiments undertaken and the model developed have given a good understanding of the physical processes occuring during blowdown, even for multi- component multiphase systems. This work is now being extended to cover: free water, which often forms Rapid depressurization of pressure vessels: A. Haque et al. during blowdown, especially offshore; pipe lines, over which there is a significant pressure drop during blowdown; and arbitrary combinations of vessels and pipe lines, to simulate real systems more effectively. Acknowledgements This work was sponsored by Shell UK Exploration & Production, under contract number C/22225. References I Chcmicol Engineers Handbook (Eds. R. H. Perry and C.H. Chilton). 5th edition. McGraw-Hill. New York. USA. IY73. pp. 10.12- 10. IS 2 Chemical Engineers Handbook (Eds. R. H. Perry and C.H. Chilton). Sth edition. McGraw-Hill. New York, USA, 1973, pp. 10. IO- 10. 12 3 Rosenhow. W. M. Trutts ASME 1952.74. Y6Y 4 Jordan-. D. P. Advunces Heut Trunsfcr 196X. 5. 55 5 Incropera. F. P. and De Witt. D. P. in Fundamentals of Heat & Mass Transfer. 2nd edition. Wiley. New York. USA. IYRS. chapter5 h Thomas. L. H. C~mnt Pure Marh 1954.7. 195 J . Loss Prev. Process Ind., 1990, Vol3, J anuary 7