Blast Charts for Explosive Evaporation of Superheated Liquids

A.C. van den Berg TNO Defense, Security and Safety, The Netherlands; bert.vandenberg@tno.nl (for correspondence)
Published online 17 March 2008 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/prs.10252
Explosive evaporation of superheated liquids is a hazard faced in many sectors of the process industries. The assumption that the explosive evaporation process is as fast as the inertia of the expanding mix of vapor and liquid in the surrounding air allows has been used as a safe and conservative starting point for the numerical computation of the blast. The result represents the maximum blast potential of an evaporating superheated liquid. The modeling has been applied for the compilation of blast charts for some gaseous substances that are usually stored, transported, or handled in liqueed form. Because the blast charts have been directly computed from the gas dynamics of an explosive evaporation process, they enable a more realistic assessment of the blast potential of superheated liquids than the usual methods based on TNT-equivalency. 2008 American Institute of Chemical Engineers Process Saf Prog 27: 219224, 2008
INTRODUCTION

Superheated liquids may evaporate violently. Many substances with a boiling point below ambient, such as for instance: propane, butane, ammonia and carbon dioxide are often stored, handled and transported as liquids in pressure vessels. Through the burst of a pressure vessel, the liquid pressure suddenly falls to ambient by which the liquid gets superheated and starts to evaporate quickly. Under certain conditions, the evaporation process may be fast enough to give rise to substantial blast effects. The burst of pressure vessels containing liqueed gases is therefore a major scenario considered in hazard evaluations in the process industries [1]. These days, large-scale sea port and storage facilities for liqueed natural gas (LNG) are being projected in
2008 American Institute of Chemical Engineers

many places all over the world. LNG is transported and stored as a cryogenic at a temperature of 112 K that is the boiling temperature of methane under ambient pressure. Hazard assessments often consider an accidental loss of containment in a port environment. A fast evaporation process of the cold LNG develops in contact with warm water. The evaporation rate however, is largely reduced and controlled by the rise of a vapor lm between the cold LNG and the warm water. Because a dense liquid on top of a vapor layer is an unstable conguration the cold LNG and warm sea water may get in direct contact locally. In direct contact, the heat transfer is so intense that the LNG readily gets superheated enough to develop an explosive evaporation process. Such local rapid phase transitions (RPTs) are observed to produce signicant blast effects [2]. Very similar phenomena may occur when lava or molten metal comes in contact with water [3]. When a liquid gets superheated either by pressure vessel burst or by intense heat transfer, a fraction of the liquid will quickly evaporate. The evaporation process requires heat, which is extracted from the liquid. As a consequence, the temperature and the vapor pressure of the remaining liquid fall. The evaporation continues until the liquid temperature has fallen to the boiling temperature and consequently the vapor pressure has fallen to the ambient pressure. The quick phase transition from liquid to vapor goes hand in hand with the development of a large volume. One cubic meter of liquid may give rise to hundreds of cubic meters of vapor under atmospheric conditions. The violent development of vapor will move the surrounding air, which gives rise to the formation of a blast wave. A blast wave may do damage to people and property up to a substantial distance from the event. Blast charts constitute a simple and quick means to represent the blast potentials of energetic materials. Blast charts are graphical representations of major
September 2008 219

Process Safety Progress (Vol.27, No.3)

blast wave characteristics dependent on the distance from the blast epicentre. Major blast wave parameters, which largely determine consequences in terms of injury or lethality to people and damage to structures, are the peak overpressure and the positive impulse of the blast wave. Presently, the use of TNTblast charts is still common for the representation of the blast potentials of exploding vessels [4]. Determination of an appropriate TNT-equivalency however, is not very straightforward because of the huge differences in source characteristics between an explosive evaporation process and a TNT-detonation. Typical pressures in a TNT detonation process are of the order of 10 GPa. As a consequence of the extremely high shock strengths initially, a lot of mechanical energy (blast) is dissipated into heat during the early stages of TNT blast wave propagation. A TNT-equivalency for an explosive evaporation process, based on the full amount of work done by the vapor pressuredriven expansion (as recommended in [4]), must therefore necessarily result in a substantial underestimation of the maximum blast potential of superheated liquids at some distance. This study presents blast charts compiled directly from the numerical simulation of the gas dynamics driven by an explosive evaporation process. Starting point has been the safe and conservative assumption that the evaporation process of the superheated liquid is as fast as the inertia of the expanding vapor and the surrounding air allows. This study is further organized as follows. A concise literature review in Chapter 2 reveals that the present knowledge with regard to the explosive evaporation rate as well as to the conditions that allow explosive evaporation of superheated liquids is insufcient. Therefore, a safe and conservative approach in the modeling is appropriate for the time being. Chapter 3 recapitulates and updates the numerical modeling, presented earlier by Van den Berg et al. [5,6]. Chapter 4 nally presents the blast charts compiled with the present modeling.

Venart and coauthors [11,12] speculate that pressure vessel burst and explosive evaporation of the vessels contents would also be possible far below the homogeneous nucleation temperature by a curious interaction between the liqueed gas and a release ow through, for instance, a safety valve or a leak. A release ow leads to a pressure drop in the vessel and the development of vapor bubbles in the liquid. The vapor bubbles make the liquid swell and increase the liquid level in the vessel by which the vent may get choked. Then the internal vessel pressure rises again. The pressure rise would make the vapor bubbles implode. Implosion of vapor bubbles is accompanied by strong local pressure pulses. This would result in a coherent power amplied shock wave through the liquid that may shatter the vessel and explosively evaporate a portion of its contents. The phenomena were indicated as a boiling liquid compressed bubble explosion. The exact conditions however, that would allow such phenomena are not very clear. The blast effects from the rupture of a vessel are fully determined by the exact vapor development dependent on the time. The vapor source strength is determined by the liquid release rate from the vessel and the evaporation rate of the liquid released. The liquid release rate is dependent on the failure mode of the vessel which is largely determined by the vessels structural design and is hard to predict, generally speaking. This concise review indicates that the exact release rate from a failing vessel is hard to predict and that it is presently not at all certain when and under what conditions homogeneous nucleation and explosive evaporation are to be expected. The uncertainties make a safe and conservative approach in the modeling of blast effects appropriate, for the time being.
MODELING

CONDITIONS THAT ALLOW EXPLOSIVE EVAPORATION

According to Reid [7,8] explosive evaporation of superheated liquids can only occur when the liquid temperature exceeds a certain threshold temperature that is the super heat limit temperature or the homogeneous nucleation temperature. Only then, vapor will develop simultaneously throughout the liquid and powerfully expand. The remaining liquid will be entrained and atomized in the violent gas dynamics of expansion. The exact value of the super heat limit temperature or the homogeneous nucleation temperature however, is not very clear. Different threshold temperatures can be calculated starting from different equations of state [9]. Reid [7] for instance, predicts the super heat limit temperature to be approximately equal to 89% of the critical temperature in Kelvin. Salla et al., [10] on the other hand calculated the super heat limit as being the liquid temperature at which the heat balance of the evaporation process allows a 50% ash fraction.
220 September 2008 Published on behalf of the AIChE

Evaporation The evaporation process requires heat, which is extracted from the remaining liquid. The decrease of the liquid mass and the fall of the liquid temperature are related through: M1 C1 dT L dM1 where: Ml 5 liquid mass (kg) Cl 5 specic heat of the liquid (J kg21 K21) L 5 heat of evaporation of the liquid (J kg21) dT 5 liquid temperature fall (K) dMl 5 liquid mass decrease (kg) Both the specic heat of the liquid (Cl) and the heat of evaporation (L) are dependent on the liquid temperature, which can be approximated by the following relations [4]: Cl T CA CB :T CC :T 2 CD :T 3 J kg1 K1  LT LT0 :
DOI 10.1002/prs

Tc T Tc T0

A

Jkg1

Process Safety Progress (Vol.27, No.3)

where: T 5 temperature (K) T0 5 reference temperature (K) Tc 5 critical temperature (K) L(T0) 5 heat of evaporation at the reference temperature (J kg21) CA, CB, CC, CD, and A are constants Blast The sudden transition of liquid to vapor is accompanied by the development of a large volume that violently pushes the surrounding atmosphere aside. The computation of the gas dynamics requires the liquid release rate from the vessel and the evaporation rate of the ashing liquid as a function of time as input. For the time being, a safe and conservative approach in the modeling of blast is appropriate. Such an approach consists in the simple assumptions that the vessel instantaneously disintegrates and that intrinsically the evaporation of superheated liquid could occur innitely fast. Then, the evaporation rate is fully determined by the gas dynamics (inertia) of the developing mass of vapor in interaction with the surrounding mass of air. The evaporation rate is now fully determined by the rate at which the developing vapor can expand by pushing the surrounding air aside. This safe and conservative assumption of expansion-controlled evaporation constitutes the starting point for the computation of the gas dynamics induced by the evaporation process. An explosive evaporation process is now modeled as a vapor release from a source area covering the initial liquid volume. The vapor pressure of the superheated liquid in the source area drives the gas dynamics of the vapor release. As a consequence of the evaporation, the liquid temperature and thereby the vapor pressure fall until the boiling point under ambient conditions has been attained. The evaporation rate of the superheated liquid is fully determined by the rate at which the developing vapor can expand by moving the mass of vapor and the mass of surrounding air. This concept can be framed in a numerical mesh of any geometry. For the compilation of blast charts for rapid phase transitions on the earths surface, a one-dimensional hemispherical mesh is appropriate. The gas dynamics has been computed by a time stepwise integration of the Euler equations that are the conservation equations for mass, momentum and energy for inviscid compressible ow. The source area has been specied over the rst 50 cells around the origin. The vapor pressure of the liquid in the source area and the consequent mass ow constitute the boundary conditions for the developing gas dynamics of expansion outside the source area. The entrainment of the atomized liquid mass will of course slow down the vapor pressure-driven expansion process. Strictly speaking, inclusion of the liquid phase entrainment would require advanced two-phase ow modeling. In the present simplied ideal gas approach however, the inuence of the entrainment of the liquid phase on the expansion has been simply taken into account by including the
Process Safety Progress (Vol.27, No.3)

Figure 1. (a) Blast overpressure versus scaled distance

for explosive evaporation of ammonia. (b) Blast impulse versus scaled distance for explosive evaporation of ammonia. [Color gure can be viewed in the online issue, which is available at www.interscience. wiley.com.]

liquid mass in the vapor mass. The inclusion of the liquid phase is the only aspect in which the present modeling is an update of the modeling presented earlier [5,6]. Blast charts can be compiled by sampling both the maximum blast wave overpressure and the blast wave positive impulse all over the numerical mesh during the numerical simulation of the gas dynamics. Validation Evaporation of superheated liquids is modeled as a vapor pressure-driven release of vapor and liquid. It represents the maximum possible amount of work done by the vapor pressure of a ash-evaporating liquid to the surrounding atmosphere. The resulting blast effects can therefore be considered an absolute upper limit. In accidental or experimental bursting vessel scenarios the blast potential of superheated liquids is
DOI 10.1002/prs September 2008 221

Published on behalf of the AIChE

Figure 2. (a) Blast overpressure versus scaled distance

Figure 3. (a) Blast overpressure versus scaled distance

for explosive evaporation of butane. (b) Blast impulse versus scaled distance for explosive evaporation of butane. [Color gure can be viewed in the online issue, which is available at www.interscience.wiley. com.]

for explosive evaporation of carbon dioxide. (b) Blast impulse versus scaled distance for explosive evaporation of carbon dioxide. [Color gure can be viewed in the online issue, which is available at www. interscience.wiley.com.]

largely obscured by the failure mode of the vessel. In practice, an accidental disintegration of a steel vessel is mostly far from instantaneous. Simple acoustic volume source exercising has demonstrated how the blast effects strongly reduce as the vessel failure takes a small span of time [5,6]. Application of the modeling to a series of bursting propylene vessel experiments by Giesbrecht et al. [13] clearly demonstrated its safe and conservative character [5,6]. In cases however, when the liquid temperature is well above the homogeneous nucleation temperature and the vessel near-instantaneously falls apart, the computed blast effects are well in the range of what has been experimentally observed. Birk et al. [14] for instance, report window pane breakage at 170 m distance from the rupture of a vessel containing 1.46 m3 LPG. This is well in the range of what is being predicted by the present modeling.
222 September 2008 Published on behalf of the AIChE

BLAST CHARTS

Blast charts for both the blast peak overpressure and the blast positive impulse have been compiled for some of the most common substances stored, transported or handled in liqueed form under pressure, namely: ammonia, butane, carbon dioxide, and propane for various initial temperatures (Figures 1 4). To indicate that explosive evaporation is probably to be expected only beyond the superheat limit or the homogeneous nucleation temperature (THN), approximate values estimated as being equal to 89% of the critical temperature, have been indicated in the respective blast charts. The blast curves for temperatures below the superheat limits leave the possibility open for the speculative boiling liquid compressed bubble explosions, suggested by Venart and coauthors [11,12]. In addition, blast charts have been compiled for rapid phase transitions of LNG at the
DOI 10.1002/prs Process Safety Progress (Vol.27, No.3)

Figure 5. (a) Blast overpressure versus scaled distance Figure 4. (a) Blast overpressure versus scaled distance

for explosive evaporation of propane. (b) Blast impulse versus scaled distance for explosive evaporation of propane. [Color gure can be viewed in the online issue, which is available at www.interscience. wiley.com.]

for explosive evaporation of LNG. (b) Blast impulse versus scaled distance for explosive evaporation of LNG. [Color gure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

superheat limit temperature (see Figure 5). Under ambient pressure LNG cannot be further superheated because it explodes beyond the superheat limit. Material properties used for the compilation of the blast charts, namely: the molecular weights, the liquid densities, the boiling temperatures, the critical temperatures, the vapor pressures, the specic heats, and the heats of evaporation have been taken from PGS 2 [4]. The blast charts have been drawn up for an ambient pressure (P0) of 101 kPa and scaled with the initial liquid mass M. The liquid density qL makes them applicable for any initial weight or volume of liquid. Quantities of LNG involved in RPTs during loss of containment scenarios are not a priori known, of course. Experimental data on blast from RPTs however, may give an indication that way. The LNG blast charts may then be helpful in the translation of blast overpressure-time records into typical quantities of LNG involved in RPTs under various conditions.
Process Safety Progress (Vol.27, No.3)

Blast charts describe just free-eld blast effects. They allow the evaluation of the blast load of structures of simple shape only. Through injury/lethality or damage diagrams overpressure-impulse combinations translate directly into consequence levels [15]. For detailed consequence evaluation however, simple blast charts for free space dont sufce. In conned spaces such as tunnels for instance, the explosive evaporation rate of superheated liquids may be inuenced by the reduced possibilities for free expansion. Then sophisticated multidimensional simulation is a way to follow. The present modeling can be straightforwardly framed in a numerical mesh of any geometry, which has been demonstrated in earlier studies [5,6,16].
CONCLUSION

The assumption that the explosive evaporation process is as fast as the inertia of the expanding mix of vapor and liquid and the surrounding atmosphere allows, has been used as a safe and conservative starting point for the numerical computation of the
DOI 10.1002/prs September 2008 223

Published on behalf of the AIChE

blast potential of superheated liquids. The resulting blast effects constitute an upper limit because they are computed as being the maximum quantity of work done by the vapor pressure of ash-evaporating liquids to the surrounding atmosphere. The modeling has been applied for the compilation of blast charts in half space for some of the gaseous substances that are usually stored, transported or handled under pressure in liqueed form. Because the blast charts have been directly computed from the gas dynamics of an explosive evaporation process, they enable a more realistic assessment of the blast potential of superheated liquids than the usual methods based on TNT-equivalency. When, simple and straightforward consequence evaluation through blast charts does not sufce, the modeling can be straightforwardly framed in a numerical mesh of any geometry.
LITERATURE CITED

7. 8. 9.

10. 11.

12.

1. T. Abbasi and S.A. Abbasi, The boiling liquid expanding vapour explosion (BLEVE): Mechanism, consequence assessment, management, J Haz Mat 141 (2007), 489519. 2. G.A. Melhem, S. Saraf and H. Ozog, LNG Properties and Hazards Understand LNG Rapid Phase Transitions, ioMosaic Corporation, Salem (NH) and Houston (TX), USA, 2006. 3. S.A. Schubach and D.F. Bagster, Some aspects of modelling steam explosions. J Loss Prevention Process Industries 9 (1996), 193197. 4. PGS 2 (Yellow Book) (2005), Methods for the calculation of physical effects, http://www.vrom. nl/pagina.html?id520725. 5. A.C. van den Berg, M.M. van der Voort, J. Weerheijm and N.H.A. Versloot, Expansion-Controlled EvaporationA safe approach to BLEVE blast, J Loss Prevention Process Ind 17 (2004), 397405. 6. A.C. van den Berg, M.M. van der Voort, J. Weerheijm and N.H.A. Versloot, BLEVE blast by expan-

13.

14.

15. 16.

sion controlled evaporation, Process Safety Progress 25 (2006), 4451. R.C. Reid, Superheated liquids, Am Scientist 64 (1976), 146156. R.C. Reid, Possible mechanisms for pressurizedliquid tank explosions or BLEVEs, Science 203 (1979), 3. T. Abbasi, B.A. Clarke and S.A. Abbasi, Accidental risk of superheated liquids and a framework for predicting the superheat limit, J Loss Prev Process Ind 20 (2007), 165181. J.M. Salla, M. Demichela and J. Casal, BLEVE: A new approach to the superheat limit temperature, J Loss Prev Process Ind 19 (2006), 690700. J.E.S. Venart, G.A. Rutledge, K. Sumathipala and K. Sollows, To BLEVE or not to BLEVE: Anatomy of a boiling liquid expanding vapour explosion, Seventh International Symposium on Loss Prevention and Safety Promotion in the Process Industry, Toarmina, Italy, 1992. J.E.S. Venart, K.F. Sollows, K. Sumathipala, G.A. Rutledge and G.A. Jian, Boiling liquid compressed bubble explosions. Experiments/Models, GasLiquid Flows, Vol. 165, pp. 5560. ASME, Fluids Engineering Division, 1993. H. Giesbrecht, K. Hess, W. Leuckel and B. Maurer, Analysis of explosion hazards on spontaneous release of inammable gases in the atmosphere, Part 1: Propagation and deagration of vapour clouds on the basis of bursting tests on model vessels, Ger Chem Eng 4 (1981), 305314. A.M. Birk, D. Poirier and C. Davison, On the response of 500 gal propane tanks to a 25% engulng re, J Loss Prev Process Ind 19 (2006), 527541. PGS 1 (Green Book) (2005), Methods for the determination of possible damage, http:// www.vrom.nl/pagina.html?id520725 A.C. van den Berg and J. Weerheijm, Blast phenomena in urban tunnel systems, J Loss Prev Process Ind 19 (2006), 598603.

224 September 2008

Published on behalf of the AIChE

DOI 10.1002/prs

Process Safety Progress (Vol.27, No.3)