In fluence of Wind Deviation Angle on n-Hexane Evaporation Loss of Internal Floa

Huang Weiqiu; Xu Manlin; Li Fei; Ji Hong; Fang Jie

(Jiangsu Key Laboratory of Oil & Gas Storage and Transportation Technology,Changzhou Uniνersity, Changzhou 213164)

Abstract: With the increasing attention to environmental protection, it is still necessary to strictly control the oil evaporation loss from the IFRT (internal floating-roof tank) to the atmosphere. Upon using n-hexane as a representative of light oil, the effects of the WDAs (wind deviation angles) on air flow distribution, the wind speed, the n-hexane vapor concentration, and the evaporation loss rate in the IFRT were investigated, and the mass transfer of the vapor-air was analyzed. The results are shown as follows: when the WDA is 0°, the vapor concentration in the gas space above the floating deck is the lowest; when the WDA is 22.5°, the oil evaporation loss rate is the largest; when the WDA is 45°, the vapor concentration is the highest,but the evaporation loss rate is the smallest. It is recommended to arrange the vent to the wind direction with an angle of 45°to reduce the evaporation loss and protect the atmospheric environment.

Key words: wind deviation angle; n-hexane evaporation loss; internal floating-roof tank; numerical simulation; wind-tunnel experiment

1 Introduction

Internal floating-roof tanks (IFRTs), as a kind of relatively ideal oil storage containers with less oil evaporation loss and vapor pollution, are often used to store high-quality oil and refined oil. However, an IFRT cannot be sealed absolutely in reality. Meanwhile, the floating deck has a certain degree of ‘drift’ on the liquid surface and the rim seal or waxing board gradually becomes aged or even worn down with the service time[1], which may cause accelerated oil evaporation. On the other hand, the oil film of the tank wall will also be evaporated into the air by the gas turbulence in the gas space. Oil evaporation will inevitably result in increased oil loss and deterioration in oil quality. American Petroleum Institute (API) worked on the wind direction, wind speed coefficient, and wind pressure coefficient on the surface of floating deck of the external floating-roof tank (EFRT), and corrected the loss calculation method[2], but the evaporation loss rate of the liquid storage and the concentration distribution above the floating deck were not measured. Huang, et al.[3]summarized three evaluation methods for the oil vapor emission (OVE), with the loss-calculating formulas refined and the evaluation software improved. Sharma,et al.[4]investigated the static breathing evaporation loss from horizontal storage tanks, and the regression loss formula was found to be different from the AP-42 formula. Zhu, et al.[5]measured the changes of various parameters under different conditions of a 93 RON gasoline, such as weight, viscosity and concentration of oil vapor components. Tamaddoni, et al.[6]studied the emission characteristics of VOCs when oil tankers were loaded with crude oil. The evaporation loss rate is more dependent on the tank size at low wind speed, which is contrary to the common semi-empirical evaporation model[7]. Hou, et al.[8]investigated the heat and mass transfer mechanisms in refueling process by using a numerical model of multiphase flow. Cooling in the tank may condensate the oil vapor in the tank, which causes unconventional changes in evaporation loss[9]. Wind tunnels have been widely used in simulation experiments of the airflow field in various industries[10-11]. Based on the wind-tunnel experiments, Tominaga, et al.[12]found that the distribution of wind speed and the concentration of gas space in the tank are mainly affected by the inlet air flow vent. Ai and Mak[13]found that the distribution of polluted gas is highly dependent on the wind direction,and the diffusion is intensified when the WDA (wind deviation angle) is not equal to 0°. There are some loss mechanisms that need to be researched deeply on the effect of different WDAs on various factors such as the evaporation loss rate, the air flow motion in the tank, the distribution of wind speed, and the vapor concentration.The Fluent software was used to simulate the 1000 m3ratio model IFRT, and the corresponding tank model was placed in a self-made wind tunnel for experimental verification. The effects of three WDAs on airflow distribution, wind speed,n-hexane vapor concentration,and evaporation loss rate in the IFRT were numerically simulated, and the mass transfer of vapor - air was analyzed in this study.

2 Methodology

2.1 Experimental protocols

A small IFRT represents a scaled model (32:1) of a 1 000 m3IFRT. All parameters and experimental conditions were all consistent with the numerical simulation. The tank diameter, rim gap width, tank wall height, and roof height were 344 mm, 6 mm, 395 mm,and 41 mm, respectively. Four vents (B1—B4) were evenly set along the circumference of the upper wall of the tank (as called the wall vent tank, WVT) and (D1—D4) were similarly set for the roof fringe of the tank(roof vent tank, RVT). Three WDAs (0°, 22.5°, 45°) were considered to blow to the vent B1 (upon taking the WVT as an example), as shown in Figure 1. The WDA was set to be 0° in the experimental part. The ‘FDH’s (floating deck heights) were set at 88 mm, 132 mm, 176 mm,220 mm, 264 mm, and 312 mm from the tank bottom to the floating deck. Because the composition of gasoline is quite complicated,n-hexane was taken as a representative of conventional oil owing to its slow and moderate component volatility.

The tank was placed in a self-made wind tunnel (Figure 2). The wind tunnel (DFWT-10) included a test section of 1.5 m (H) × 1.5 m (W) × 3 m (L) to simulate the wind field and the temperature field. The vapor concentrations were analyzed by gas chromatography. The wind speed,temperature, and humidity were measured by a hot-wire anemometer. The oil evaporation loss was automatically measured by a high-precision electronic balance.

Figure 1 Schematic representation for different WDAs

Figure 2 Wind tunnel and schematic representation for the experiment

2.2 Governing equations

The gas space in the tank was filled with a mixture of two components,n-hexane vapor and air. The ambient temperature was controlled at an average temperature of 13 °C. The realizablek-εturbulence model[14]was used to simulate the air flow movement law of the vapor in the gas space of tank, and the wall-function method was used to correct the wall surface. It was simpli fied to a constant ambient wind that did not change with time and height.The continuity equation, the momentum equation, and thek-εturbulent model equations were respectively equations(1)―(4):

in which:ρ(kg/m3) is the fluid density;xi,xj(m) are the moving distances onXandYaxes, respectively;ui,uj(m/s)are the velocity onXandYaxes, respectively;p(Pa) is the absolute pressure;fi(N/m3) is the volume force;μ(Pa·s) is the dynamic viscosity;K(m2/s2) is the kinetic energy;ε(m2/s3) is the dissipation rate;PKis the kinetic energy generating term;Gbis the buoyancy generating term;YMis the compressibility correction;υ(m2/s) is the kinematic viscosity;ωkis the angular velocity. When the direction of shear flow is the same as the gravitational direction,Cε3= 1; and when it is perpendicular to the direction of gravity,Cε3= 0.

2.3 Computational domain and boundary conditions

The ‘FDH’ was set at 312 mm for numerical simulation.Five rows of measuring points (C1 — C5) in the model tank were set as shown in Figure 1(c). The coordinates (in mm) onXZplane were (83, 83), (83, -83), (-83, 83), (-83,-83), and (0, 0), respectively. The heights of measuring points for C1—C4 onYaxis were set at an equal interval of 10 mm from 315 mm to 395 mm sequentially, and the heights for C5 were set from 315 mm to 435 mm because of the space of tank dome (Figure 4). A three-dimensional computational domain (Figure 3) was applied, and the size of the region was 15D(X) × 10D(Z) × 5H(Y) (D:tank diameter,H: total height of the tank). The floating deck rim was set as the mass- flow-inlet, and the mass- flow rates were set experimentally. The wind speed was set at 4.36 m/s along the positive direction ofXaxis. The steady state and the SIMPLE scheme were selected. As regardsn-hexane, its molar mass is 86.18 g/mol, its saturated vapor pressure is 11.91 kPa, and its diffusion coefficient in the air is 7.93×10-6m2/s at 13 °C.

Figure 3 Computational domain of IFRT

Figure 4 Geometric model of gas space

3 Experimeatal Investigation

3.1 Influence of ‘FDH’ and vents position on wind speed in tank’ s internal gas space

Figure 5 shows the distribution of the wind speed at different heights when the ‘FDH’ was increased in the WVT and RVT. It is obvious that the wind speed is the smallest at a height of 400 mm (above the top of the tank)as shown in Figure 5(a). Furthermore, the trend of changes in wind speed for different ‘FDH’s is similar. After the floating deck is raised, the gas space becomes smaller.Then the disturbance of the gas space will be aggravated,and the wind speed of each point will be also increased.

It can be seen from Figure 5(b) that the wind speed decreases from top to bottom of the tank. The wind speed is almost unchanged after the floating deck is raised. The wind speed in RVT is generally lower than that in WVT according to Figure 5. Therefore, the evaporation loss rate of RVT is smaller than that of WVT integrally because of lighter gas disturbance.

3.2 Influence of ‘FDH’ and vents position on n-hexane concentration in tank’s internal gas space

Figure 6 shows the distribution of mole fraction at different measuring points with four ‘FDH’s in two kinds of tanks. It can be seen from Figure 6(a) that the mass fraction in the middle of the gas space is roughly equal to the average value, which is larger than that in the top center. When the floating deck is raised, the vapor evaporates faster, while the inlet airflow is in a smaller gas space, and the thickness of the boundary of the liquid surface becomes thinner[15], leading to an increased average mass fraction. As a result, the mass fraction increases when the floating deck is raised.

The distribution of mass fraction of gas space is relatively homogeneous according to Figure 6(b). The mass fraction decreases after the floating deck is raised. The mass fraction in RVT is greater than that in WVT. When the floating deck is raised, the air flow exchange rate increases,and the fresh air takes moren-hexane vapor away, resulting in a quickened evaporation loss rate of the stored fluid.

Figure 6 n-Hexane concentrations in the WVT and RVT with different ‘FDH’s

3.3 In fluence of ‘FDH’ and vents on evaporation loss rate

Figure 7(a) shows that the evaporation loss rate of WVT and RVT can both increase when the floating deck is raised. However, the evaporation loss rate of RVT is less than that of WVT. Besides, the higher the floating deck,the more obvious the difference between two rates is. The rising of the floating deck will promote the evaporation loss rate of the liquid, and the air pollution will be aggravated. However, the evaporation loss rate of the liquid storage will be decreased. The higher the floating deck is, the more the evaporation ratio curve tends to be horizontal (Figure 7(b)). If the WVT is transformed into a RVT, the evaporation loss rate and the evaporation ratio will be reduced, and the air pollution can be reduced.

4 Simulation Calculation

4.1 Simulation and experimental veri fication

Figure 8(a) and Figure 8(b) show the wind speed (Xaxis)of the gas space and vapor concentration clouds in the tank, respectively, when the WDA is 0°. It can be seen from Figure 8(a) that the ambient wind enters the tank and moves to the leeward side with a largest speed. The more the closer to the leeward side is, the larger the range will be affected by the inlet air flow. The gas that collects on the leeward side will move toward the windward side,which is opposite to the direction of inlet airflow, so convection occurs. It can be seen from Figure 8(b) that since the roof and the middle of the tank body are affected by the inlet air flow, the vapor concentration is lowered.

Figure 9 and Figure 10 show the comparison between the measured and the simulated values of the wind speed andn-hexane vapor concentration at Row C5, respectively.The simulated values are generally consistent with the measured values except a slight deviation between them(Figure 9 and Figure 10). It is reasonable to use the simulation method to investigate the law of evaporation and diffusion of stored liquid in the IFRT.

Figure 7 Comparison of evaporation loss rate and evaporation ratio in WVT with those in RVT

4.2 In fluence of WDA on various parameters

4.2.1 In fluence of WDA on air flow speed in tank

Figure 11 shows the gas streamline and speed clouds for different WDAs above the floating deck. The gas above the floating deck moves with the inlet airflow with a high speed. The farther the air flow moves, the larger the range is affected, but the speed is gradually reduced in the process of gas movement.

Figure 9 Comparison of measured and simulated values of wind speed

Figure 10 Comparison of measured and simulated values of n-hexane vapor concentration

Figure 11 Speed clouds and streamline diagram with different WDAs

When the WDA is equal to 0° (Figure 11(a)), the air flow is continuously dispersed to the both sides while moving against the leeward side of the roof, and it moves to the floating deck vertically after being blocked by the tank wall. Then, the airflow is folded back to the windward side. After that, the airflow moves to the leeward side again along the tank wall. Then the cycle repeats, and an elliptical eccentric vortex is formed, which is almost perpendicular to the floating deck, and the center of the vortex is close to the leeward side. This vortex may cause a pressure difference between the windward side and the leeward side of the IFRT rim seal, and there will be convection or local vortexes in the sealed space for a gas-tight seal of the tank. There is an approximately horizontal clockwise vortex throughout the gas space in tank when the WDA is 22.5° (Figure 11(b)). When the WDA is 45° (Figure 11(c)), the inlet airflow moves to the windward side along the center of roof, then it moves to the floating deck in a direction parallel to the wind direction, and then it further moves along the floating deck and tank wall to the leeward side. Then, the air flow moves upward and flows back to the roof again. Thus two substantially symmetrical vortexes are formed in the tank.

4.2.2 Influence of WDA on wind speed distribution in tank

Figure 12 shows the comparison between the average wind speed of four vents in the tank with three WDAs.The average wind speed of the inlet airflow can be one of the most important factors on the gas space inside the tank. When the WDA is 0°, the average wind speed of B3 is about 0.17 times that of B1. When the WDA is 22.5°,the wind speed of B3 is about 0.32 times that of B1, and the in fluence of the ambient wind on the tank is greater than the case when the WDA is 0°. The wind speed of the inlet air flow is about 0.10 times the ambient wind speed when the WDA is 45°. Therefore, the ambient wind has a slightest influence on the inside gas space of the tank when the WDA is 45°.

Figure 12 Comparison of average wind speed of vents

Figure 13 shows the distribution of wind speed in the 5 rows of measuring points in the tank with three WDAs.It can be seen from Figure 13(a) that the wind speed of 5 rows (C1―C5) decreases in the following order:VC5＞VC2≈VC4＞VC1＞VC3, andVC1-VC4gradually increase with the height of measuring points, butVC5decreases. Afterwards,VC1-VC4increase signi ficantly at the measuring points just above the vents and then reduce signi ficantly at the higher measuring points, presenting an upward cone shape. In addition, the wind speeds for C1―C5 are almost equal at the height of the tank wall. It can be seen from Figure 13(b)that the wind speed decreases in the following order:VC3＞VC2＞VC1＞VC4, when the measuring points are below the height of the vents. The wind speeds are approximately the same at the height of the vents. The wind speed decreases in the following order:VC4＞VC3≈VC1＞VC2, when the measuring points are above the vents. The wind speed along the central axis (VC5) is significantly smaller than that of the tank rim in any case. It is difficult for airflow entering the tank to directly impact the gas of vortex center due to the centrifugal force, because only the viscosity can drive the gas of vortex center to rotate.

When the ambient wind passes over B1 and B2, the gas inside the vent will be driven to the roof along both sides of the tank wall, and the movement of nearby gas is accelerated. The evaporation loss rate is closely related to wind speed above the rim, and the wind speed is the smallest when WDA is 45° (Figure 13(c)). Therefore, the evaporation loss rate is the smallest when WDA is 45°.

4.2.3 In fluence of WDA on n-hexane vapor concentration distribution in tank

Figure 14 shows a cloud of vapor mass fraction in the tank with three WDAs. Figure 15 shows the distribution ofn-hexane vapor mass fraction of C1—C5 above the floating deck with three WDAs. It can be seen from Figure 15 that the curves ofn-hexane vapor concentration and concentration gradients of C1—C5 from the floating deck to the roof are gradually reduced. When the WDA is 45°,then-hexane vapor concentration in the tank is the largest.When the WDA is 0° (Figure 15(a)), the concentration curves of C2 and C4 almost coincide and are signi ficantly higher than the other three curves. The concentration curve of C5 is between that of C3 and C1. Then-hexane concentration of the rim seal surface decreases in the following order:CC3＞CC4＞CC2＞CC1＞CC5. When the WDA is 22.5° (Figure 15(b)), then-hexane concentration of the rim seal surface decreases in the following order:CC4＞CC1＞CC2＞CC3＞CC5. When the WDA is 45° (Figure 15(c)), the concentration curve of C3 is significantly higher than that of other four curves, and then-hexane concentration on the rim seal surface decreases in the following order:CC2＞CC4＞CC1＞CC3＞CC5.

Figure 14 Gas space concentration clouds of n-hexane with different WDAs

Figure 15 Distribution in mass fraction of n-hexane in the tank

After the floating deck is raised, the inlet airflow moves in a smaller gas space. The equivalent film thickness of the concentration boundary of liquid surface will be thinner because of the strong turbulent flow action[3], and the concentration gradient in the gas space increases. The vortex inside the tank passes through the liquid surface,which causes the liquid surface to oscillate. It is easier for the liquid molecules to escape from the liquid surface due to more kinetic energy. The gas exchange rate between the inside and the outside of the tank is increased. The oil concentration in the tank is lower because of vapor being taken away by the fresh air flow. The oil evaporation will be accelerated, but it will be inhibited due to the increasing gas concentration in the tank. The higher the floating deck,the greater the vapor concentration in the tank.

Figure 16 shows the comparison of the average mass fraction ofn-hexane vapor at four vents on the tank with three WDAs. The largest fraction of vapor is at B2 with a WDA of 22.5°. The averagen-hexane vapor concentration of vents B2 and B4 (Figure 16) is about 1.27 times the average concentration of C5 (CC5) when the WDA is 0° (Figure 15(a)), and it is much lower than the largest concentration (Cmax) of the curves in Figure 15. When the WDA is equal to 22.5° and 45°, they are both lower than that ofCC5andCmax, respectively.

4.2.4 Influence of WDA on evaporation loss rate of n-hexane vapor in tank

Figure 17 shows the comparison of the evaporation loss rate of two ‘FDH’s (88 mm and 312 mm) at two wind speeds (4.36 m/s and 6.36 m/s), when the IFRT is blown at three different WDAs. Compared with the oil loss rate at a WDA of 0°, it can be seen that the oil loss rate with the WDAs of 22.5° and 45° is increased by 5.46%,-51.40% (a); 5.72%, -68.48% (b); and 6.84%, -55.42%(c), respectively. It can be also found from Figure 17 that the oil loss rate can be accelerated due to the increase in‘FDH’ and the increase in the ambient wind speed.

Figure 16 Comparison of average mass fraction of n-hexane vapor

Figure 17 Comparison of evaporation loss rate of n-hexane vapor

5 Conclusions

The conclusions drawn from the research results are summarized as follows:

(1) Then-hexane evaporation loss rate increases with the raising of floating deck in a tank. And then-hexane evaporation loss rate of the WVT is greater than that of the RVT integrally. It is recommended to convert the WVT into a RVT to reduce the emissions of VOCs. It is also recommended to consider the effects of the position of the vent, the floating deck and the different WDAs in the further revision of the floating roof tank loss formula worked out by the API.

(2) When the WDA is 0°, 22.5°, and 45°, respectively,then-hexane vapor concentration in tank increases in turn. Then-hexane evaporation loss rate of the WDA with 0° and 22.5° is similar, which is much larger than the scenario with a WDA of 45°. It is recommended to arrange the wall vents of the tank to the wind direction with an angle of 45° to reduce then-hexane evaporation loss and the atmospheric environmental pollution.

Acknowledgements:Financial supports of this work from the National Natural Science Foundation of China (No. 51574044),the Key Research and Development Program of Jiangsu Province (Industry Foresight and Common Key Technology)(No. BE2018065), the Postgraduate Research & Practice Innovation Program of Jiangsu Province (No. KYCX18_2630),and the Nature Science Foundation of Jiangsu Province (No.BK20150269) are gratefully acknowledged.