Oil spills in harbour area: a conservative approach to risk assessment and emergency planning

Oil spills in harbour area: a conservative approach to risk assessment and emergency planning E. Palazzi, F. Currò & B. Fabiano Chemical and Process Engineering Department DICheP G.B.Bonino, Genoa University, Italy bstract The study of dispersion and transformations following an open sea oil spill has been faced by several authors, on the basis of mathematical models of different complexity, so as to quantify the different phenomena involved in the release evolution: convective transport by sea currents, possibly in connection with wind action; turbulent diffusion; buoyant action on dispersed particles; wave action; superficial spreading of the oil due to different density between sea water and hydrocarbon; eoration and emulsification rates. Dealing with a release in a port area, besides the environmental consequences, the most severe hazards are linked to fires and explosions due to the formation of a flammable cloud by eorated hydrocarbons and air and subsequent ignition. If a cloud of sufficient size is formed and ignition occurs instantly, a large fire, jet flame or fireball may occur, but significant blast-pressure damage is unlikely. Instead, the blast effect produced by our cloud explosion (i.e. cloud formation and ignition delayed typically by some minutes) can vary greatly depending on the speed of flame propagation and can result in extensive widespread damage. We considered such types of events, focusing our attention to individual intervention times and types more suitable to contain the risk into an acceptability level. Considering that intervention time and dispersion time are comparable, the basic variables considered in developing the mathematical model were oil spill spreading rate, hydrocarbon eoration rate, cloud transport and dispersion into the atmosphere. The developed model allows the separation of the risk areas and intervention times in confined regions and ports of relative simple geometry. The same model can be applied, as well, to more complex geometry, typical of port areas, obtaining conservative results for a preliminary risk evaluation. Keywords: area risk; toxic and flammable release; emergency, domino effect. Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

360 Coastal Environment, incorporating Oil Spill Studies 1 Introduction The problems arising from hydrocarbon releases in port areas and in open sea are very different. In the latter case, the main damage can normally be of environmental kind and the most effective protective actions, as release containment and recovery, are mainly based on the knowledge of the long time evolution of the oil spot. Dealing with this topic, various mathematical models were proposed [1-4], focusing attention on the different parameters involved in transport, dispersion, depletion and transformation of the spot, as sea currents, wind and wave action, turbulent diffusion, spreading of the bulk and climbing of oil drops due to the buoyancy, eoration and emulsification rates. Dealing with a release in port area, besides the environmental impact, the most severe consequences can derive from the development of flammable and/or toxic clouds, due to the hydrocarbons orization. Specific safety measures are to be adopted, based on the knowledge of the extension of the risk areas so as determined from the relatively short time evolution of the cloud. In particular, toxic effects are mainly due to the duration and mean concentration of human exposure to the volatile pollutants, whilst the flame development is related to the instantaneous concentration of the cloud coming in contact with an ignition source. Then, in both cases, the knowledge of the distribution in space and time of the concentration of the hydrocarbon ours in the cloud C(x,y,z,t) is needed, to planning the proper safety action. mong these, one can distinguish: prevention measures, aimed at reducing the accident frequency, e.g. the study of ship courses, suitable procedures for maintenance and control, and so on; protection measures addressed to mitigate the consequences of the dangerous events, e.g. division of tank in wartight compartments, hole repair, suction and partial recovery of oil from the tank and/or the spot, mechanical containment and/or sorption of spot by means of idoneous materials, reduction of eoration rate of the hydrocarbons by foaming and possibly freezing the liquid surface, warnings and evacuating the risk areas. Based on the aforesaid considerations, this paper deals with a conservative, simplified approach to assess risk and set-up emergency planning in connection with oil spills in harbour areas. General behaviour of an oil spill et consider an accidental release of hydrocarbons coming from a tank T into a port area. In the most general case, the release originates an oil spot, a relatively dense cloud C and a passive plume P, as indicated in Figure 1. The hydrocarbons can be released partly instantaneously and partly according to a continuous flow-rate; moreover, they can directly split into the different subsystems, C and P, since both liquid and our phases could be involved in the release. Taking into consideration the rate of hydrocarbon mixing, eoration and transport into the passive plume by means of the atmospheric turbulence, and specifying an appropriate set of boundary conditions, the Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

Coastal Environment, incorporating Oil Spill Studies 361 material balances for, C and P can be solved to get C(x,y,z,t), and then to map the risk areas. Usually, only numerical solutions can be obtained, since the flow rates generally vary with the time, depending on the release and weather conditions and on the geometrical parameters characterizing each particular situation. Moreover, the accomplishment of the safety actions further complicates the problem, by putting additional terms and/or modifying some of the flow-rates in the material balances. s an example, repairing the tank damage stops the release before the tank emptying, suction of the hydrocarbons from the tank reduces the mass and duration of the release, spot containment hinders the oil spreading and then the rate of eoration, and so on. Nevertheless, to the end of an optimal emergency planning, only the most dangerous situations can be analyzed, which allows to make use of simplifying assumption, so to obtain most useful results. From this point of view, it is convenient to treat in different way the cases of windy or calm whether conditions, as shown in the next section. 3 Different scenarios following an oil spill 3.1 Windy meteorological conditions 3.1.1 Instantaneous releases of gaseous and liquid hydrocarbons at normal boiling point s well known, the accidents involving pressurized and refrigerated liquid hydrocarbons generally involve the maximum risk levels. Indeed, a considerable part of the release almost instantaneously orizes, giving rise to a dense cloud, where the hydrocarbon concentration normally exceeds the dangerous levels [5], and a liquid spot, both at normal boiling point. In such instances, the only effective safety action near the release is limited to avoiding the contact of the cloud with the target and must be realized in very short times. s regards toxic effects, it is required at least the immediate availability of breathing masks and/or air-tight shelters, whereas the risk of fires and explosion can be avoided only in absence of ignition sources. Far from the release, the extent of the risk areas is mainly determined from the meteorological conditions. Usually, under the wind action, the quasi - instantaneously developed cloud is transported and diluted more and more, up to reach no dangerous concentrations. arious mathematical models [6-10] are available in literature to calculate the behaviour of our concentration, C(x,y,z,t), and then to map the risk areas. lso at some distance from the release, relatively short time are available in windy conditions to put in action typical safety measures, e.g. sheltering or evacuation in the case of a toxic cloud. In this case, the most effective protective action consists in planning and designing ship courses and docking points conveniently far from sensible targets, so to avoid possible domino effects. The cloud tails deriving from the eoration of the liquid fraction of the release usually involve a lesser level of risk, which anyhow can be evaluated according to the model developed in the section 4. Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

36 Coastal Environment, incorporating Oil Spill Studies 3.1. Continuous releases of hydrocarbons at normal boiling point and instantaneous or continuous releases of hydrocarbons far from the normal boiling point In the situations here considered, the behaviour of the release is initially dominated by the spreading of the liquid spot on the sea. Increasing the spot area, the eoration rate also increases. Then, the volume of a spot deriving from an instantaneous release decreases with time, down to vanish. On the other hand, a continuous release of liquid hydrocarbons gives rise to a liquid volume increasing as long as the eoration rate does not exceed the source flow-rate, later it again reduces down to zero. Owing to windy conditions and to the absence of an initial cloud, it is reasonable to assume that the hydrocarbon ours deriving from eoration and those directly released be immediately transported in the plume, without formation of a significant dense cloud. In the model developed in section 4, specifically tailored on these situations, only the phase of liquid spreading and the rate of eoration will be considered. 3. Calm weather conditions Whatever the kind of release may be, in the case of absence of wind and atmospheric stability, typical nightly meteorological conditions [6], owing to spot eoration, the cloud continuously grows and spreads on the sea, even attaining the entire port area if no corrective action is performed. Changing meteorological conditions, for example following sun rise, a great mass of hydrocarbons can be transported in short time by the wind, originating a cloud even more dangerous than the instantaneous one. The description of these situations requires to take into account also the spreading of the dense cloud. However, in opposition to the case discussed at the point 3.1.1, relatively long times are available and different safety measures can be established to reduce the risk into acceptable levels, just because the growth of the cloud in absence of wind is usually very slow, so that the probability of the aforesaid extreme situations can reasonably be excluded. Figure 1: Schematic spill dispersion. Figure : Cylindrical spill spreading. Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

4 Behaviour of the oil spot 4.1 Model structure Since the cloud dimension is strictly connected with the eoration rate, and then with the area of the oil spot, also the evolution of the last one must be considered in risk analysis. However, its mathematical description can be greatly simplified with respect to the case of releases in open sea, as the behaviour of the spot is initially dominated by the spreading due to the density difference between oil and sea water, which allows to neglect the effects of the viscous forces and, in particular, the mixing between liquid hydrocarbons and water. In particular, the behaviour of the spot of cylindrical shape depicted in Figure can be approximately described by means of the mass balance and of a relatively simple laws for spreading [6] and orization [-4], according to: d = = (1) dt d 1/ 1/ 1/ = u fb f = ( ghe ) π = s () dt where: u f = spreading velocity of the spot [ms -1 ]; b f = unrestricted boundary of the spot [m]; = specific orization rate [m 3 m - s -1 ]; s = [πg(1-ρ /ρ w )] 0.5 spreading liquid coefficient [m 0.5 s -1 ]; = spot volume [m 3 ]; = spot area [m ]. The last term in eq. was obtained by combining the equations =πr and = e + i = (h e + h i ), with the following condition of hydrostatic equilibrium : ρ (h e + h i ) = ρ w h i. From eqs (1) and (), one obtains: d = (3) 1/ d s which, integrated in case of general semi-continuous release, of constant liquid spill rate and specific orization rate, gives: 3 + 4 Eq. 4 means that under the condition: > = Coastal Environment, incorporating Oil Spill Studies 363 / 3 [( 0 ) ( ) ] (4) 3/ = 0 s 0 M i.e. when the losses due to eoration ( ) exceed the liquid spill rate for the whole duration of the release, the volume of the slick decreases as increases. Obviously, this trend is always verified when dealing with the instantaneous release case. Conversely, increases up to its maximum value corresponding to: (5) M = 0 1 4 / 3 (6) Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

364 Coastal Environment, incorporating Oil Spill Studies that is reached when the slick area corresponds to the value M resulting from eq. (5). Parameters 1 and in eq. (6) are explicated in Table 1, together with the values of the maximum of the slick area, for the different release scenarios: M = M + (7) ccording to eq. (), one can notice that these values are obtained when =0. It must be observed that, in any situation, the resulting value of M represents a conservative estimate of the maximum extension of the spill area, since the model do not take into account the effect of the viscous forces on the spreading rate. Table 1: Maximum values of the volume and area of the slick. Release scenario Semicontinuous (, 0, 0 ) Continuous ( 0 = 0 = 0) Instantaneous = 0 1 4 3 s 1/ 3/ 4 0 0 4 3 [ 1 +( M - 0 ) ] 1/ M ( 1 + 0 ) 1/ M 0 1 4 / 3 3 4 s / 3 s 1/ 0 M M + M ( 1 + 0 ) 1/ 4. Complete spreading time of the spill Planning of appropriate safety measures, particularly intervention time and techniques, requires the estimation of the time t M needed by the spill to reach the area M, corresponding to the maximum orization rate and, therefore, to the maximum extension of the hazardous area. The time evolution of is obtained from the integration of eq. (), as follows: t = t [F(θ 0 )- F(θ)] (8) where: η 3 F ( ϑ) = cos 1/ ϕdϕ (9) 0 and the parameters t, θ 0 and θ are defined in Table. The characteristic spreading time t M can be obtained by solving eq. (8) under the condition θ = θ M = - π/, corresponding to = M, as follows: π t M = t F( ϑ 0 ) F( ) (10) 3/ 4 0 Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

Coastal Environment, incorporating Oil Spill Studies 365 Since π F = π F 1.73, in the particular case of continuous release, eq. (10) can be written as: 1/ 3 3 t.8 / (11 M, c ( ) ) s On the other hand, in case of instantaneous release, it results θ 0 0, being 0 <<< 1 in all the situations of practical interest; then eq. (10), at last, gives: 1/ 4 t 1.5 1/ (1 M, i 0 ( ) ) s Table : Definition of the model parameters for the different release scenarios. Release scenario t Semicontinuous (, 0, 0 ) 4 3 s 1/ 3 Continuous ( 0 = 0 = 0) 4 3 M s 1/ 3 Instantaneous = 0 ( ) 4 + 3 1 0 s 1/ 1/ 3 θ 0 arcsin M 0 π arcsin ( + ) 1/ 1 0 0 θ arcsin M arcsin 1 M arcsin ( + ) 1/ 1 0 5 Results and discussion To the purpose of risk analysis, the hypothesis of circular slick, valid in the absence of obstacles, wind and current effects is a conservative one, in that it corresponds to the maximum of the velocity of the spill area increase (d /dt) and, therefore, of the liquid orization. The most general formulas here presented are referred to a semi-continuous release of liquid hydrocarbons, ( 0 >0; 0 >0), characterized by a constant flow rate and infinite duration. When dealing with situations different from the above-mentioned, the approach must be modified: when the release is characterized by a finite duration t r, due to tank emptying or a safety intervention, we can apply the formulas previously presented in connection with t t r. Then considering t t r, we can apply the formulas calculated for an instantaneous release starting from the values of and previously calculated with reference to t = t r. In Tables 3 and 4 are reported the results of the main parameters of the release behaviour of toluene and n pentane, respectively. It was assumed an ambient release temperature equal to 5 C, so that toluene is rather far from its normal Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

366 Coastal Environment, incorporating Oil Spill Studies 0 boiling point (T b,tol = 383.8 K; 0 ptol y tol = 0.039), while n-pentane is p atm 0 n pent p proximate to it (T b, n-pent 309. K; 0 y n pent = 0.70). In both situations, we patm considered only a liquid release characterized by a total volume r in the range 300-30.000 m 3. The values of reported in the above-mentioned Tables 3 and 4 and taken from [11], correspond to an average wind velocity in the logarithmic boundary layer of and 6 m s -1 respectively. Table 3: Toluene liquid release (T = 98 K; p= 1 atm; ρ = 866 kg m -3 ; s = 4.1 m 1/ s -1 ). Wind r [m 3 ] M,c [m ] t M,c [s] M,i [m ] t M,i [s] Continuous Instantaneous u = [m s -1 ] 300 1.5 10 5 4946 1.9 10 5 348 1000 3.7 10 5 668 4.6 10 5 4631 = 1.9 10-4 3000 8.4 10 5 8794 1.0 10 6 6094 [m 3 m - s -1 ] 10000.1 10 6 1188.6 10 6 836 30000 4.7 10 6 15638 5.9 10 6 10837 u = 6 [m s -1 ] 300 9.1 10 4 308 1.1 10 5 098 1000. 10 5 409.8 10 5 836 = 5.0 10-4 3000 5.1 10 5 5385 6.4 10 5 373 [m 3 m - s -1 ] 10000 1.3 10 6 776 1.6 10 6 504 30000.9 10 6 9576 3.6 10 6 6636 Table 4: n-pentane liquid release (T = 98 K; p= 1 atm; ρ = 659 kg m -3 ; s = 6.5 m 1/ s -1 ). Wind r [m 3 ] M,c [m ] t M,c [s] M,i [m ] t M,i [s] Continuous Instantaneous u = [m s -1 ] 300 4.0 10 4 835 5.0 10 4 578 1000 9.9 10 4 117 1. 10 5 781 = 3.4 10-3 3000. 10 5 1483.8 10 5 108 [m 3 m - s -1 ] 10000 5.5 10 5 005 6.9 10 5 1389 30000 1.3 10 6 638 1.6 10 6 188 u = 6 [m s -1 ] 300.4 10 4 511 3.1 10 4 354 1000 6.0 10 4 690 7.5 10 4 478 = 9.0 10-3 3000 1.4 10 5 908 1.7 10 5 630 [m 3 m - s -1 ] 10000 3.4 10 5 18 4. 10 5 851 30000 7.7 10 5 1616 9.7 10 5 110 In case of instantaneous release, 0 = r by definition, then t M,i is directly obtained by means of eq. (1). M = M,i was calculated by neglecting 0 in the corresponding formula in Table 1, so that it results: Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

Coastal Environment, incorporating Oil Spill Studies 367 1/ 1.15 1/ 3/ 4 M, i 1 s r (13) When dealing with a continuous release lasting a finite time t i, the analysis is less simple owing to the presence of one more variable (t r or ). First of all, it is convenient to calculate the liquid spill rate (here defined as virtual spill rate in correspondence of which the release ends exactly at the time when the slick area reaches its maximum value, so that: r 1/ 3 / 3 = t =, =.8 ( ) (14) r t M c s Combining these equations, together with the one of M reported in Table 1, one obtains the value of the critical spill rate and of the other spot parameters as a function of r : 1/ 3/ 4 0.46 (15) t M, c M, c = t ( s ) r.( ) r = s 1/ 1/ 0.9 s 1/ 4 r 1/ 3/ 4 r By comparing eq. (13) with eq. (17) and eq. (1) with eq. (16), respectively, one can observe that M,c 0.80 M,i and t M,c 1.44 t M,i. Making reference to a n-pentane release and to an average wind speed of m s -1, the locus: M,c =0.091 s t r 3 (16) (17) (18) obtained by eliminating r from eqs. (16) and (17), is drawn in Fig. 3. Considering a particular release volume, e.g. r = 10.000 m 3, in the same figure is depicted the trend of all kinds of liquid spill releases considered in this work, as a function of the release duration. [m] P i M,i M, M,C i P P M,1 1 P 1 t=t/ t r r M,i t r t=t r r t[s] Figure 3: Continuous release of finite duration. Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

368 Coastal Environment, incorporating Oil Spill Studies Curve, in particular, corresponds to a critical spill rate according to which the release ends exactly at the point P (t r ; M,c ) on the aforesaid locus. On the contrary, curve 1 represents the behaviour of a a release with a spill rate <, so that it terminates at a point P 1 (t r1 > t r ), some time after the slick area has reached the maximum value. s the liquid spill rate decreases, the risk connected to the spreading of the liquid decreases as well, down to an asymptotic value of zero when 0. Curve corresponds to the situation where >, consequently the release ends at the point P (t r < t r ). It can be noticed that, dealing with the last situation, increases a little for t > t r, due to the spreading of the slick that lasts until =0. s the liquid spill rate increases, the hazard connected to the spill increases as well, owing to the increase of the maximum area of the spot and the decrease of the time at one s disposal for safety intervention. The limiting value is obtained under the condition, corresponding to the situations of instantaneous release (curve i). 6 Conclusions theoretical framework to assess risk connected to hydrocarbon spills in port area is presented. The model allows the attainment of cautious values of the maximum spreading area, corresponding to the maximum orization rate, as well as of the complete spreading time. These conservative results are to be considered in setting-up emergency planning with appropriate response equipment, fire-fighting and lightering resources References [1] Muňoz, J.F., Gmez, C., ittle,c. Model for simulation of movement and biodegradation of hydrocarbons in saturated and unsaturated soils. Oil and Hydrocarbons Spill III, ed. C.. Brebbia, WIT Press, Southampton, U.K., pp. 3-11, 00. [] Comerma, E., Espino, M., Daniel, P., Doré,. & Cabioch, F. n update of an oil spill model and its appliction in the Bay of Biscay: the weathering processes. Oil and Hydrocarbons Spill III, ed. C.. Brebbia, WIT Press, Southampton, U.K., pp. 1-, 00. [3] Pinho,J..S., ntunes do Carmo, J.S. & ieira J.M.P. Numerical modelling of oil spills in coastal zones. case study. Oil and Hydrocarbons Spill III, ed. C.. Brebbia, WIT Press, Southampton, U.K., pp. 35-45, 00. [4] Fragoso M. da R., Torres jr..r., Mehdi, N., Franca, G.B. & ima, J..M. Oil spill fate forecast system for Ilha Grande Bay, Rio de Janeiro, Brasil. Oil and Hydrocarbons Spill III, ed. C.. Brebbia, WIT Press, Southampton, U.K., pp. 47-56, 00. [5] Kaiser, G.D. & Walker, B.C. Releases of anhydrous ammonia from pressurized containers. The importance of denser than air mixtures. tmospheric Environment, 1, 1978. Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8

Coastal Environment, incorporating Oil Spill Studies 369 [6] an Ulden,.P., On the dispersion of a heavy gas released near the ground. oss Prevention and Safety Promotion in the Process Industries 1, ed. C.H. Bushman, Elsevier, msterdam, pp. 1-6, 1974. [7] Reijnhart, R., Piepers, J. and Toneman,.H. apour cloud dispersion and the eoration of volatile liquids in atmospheric wind fields I Theoretical model. tmospheric Environment, 14, pp. 751-758, 1980. [8] Eidswik, K.J. model for heavy gas dispersion in the atmosphere. tmospheric Environment, 14, pp. 769-777, 1980. [9] Eidswik, K.J. Heavy gas dispersion model with liquefied releases. tmospheric Environment, 15(7), pp. 1163-1164, 1981. [10] Zeman, O. The dynamics and modeling of heavier-than-air cold gas releases. tmospheric Environment, 16(4), pp. 741-751, 198. [11] Reijnhart, R., Rose, R.. apour cloud dispersion and the eoration of volatile liquids in atmospheric wind fields II Wind tunnel experiments. tmospheric Environment, 14, pp. 759-76, 1980. Coastal Environment, incorporating Oil Spill Studies, C.. Brebbia, J. M. Saval Perez &. Garcia ndion (Editors) 004 WIT Press, www.witpress.com, ISBN 1-8531-710-8