Escolar Documentos
Profissional Documentos
Cultura Documentos
a r t i c l e i n f o a b s t r a c t
Article history: Single and double-plate floating roof oil tanks are two types of widely used floating roof oil tanks in
Received 21 November 2018 petrochemical industry. However, the differences in flow and hear transfer characteristics of waxy crude
Received in revised form 1 February 2019 oil inside these two tanks have been studied insufficiently. Finite volume method is employed in this
Accepted 24 February 2019
research to study the temperature drop and gelatinization processes of waxy crude oil as well as the dif-
Available online 9 March 2019
ferences in single and double-plate floating roof tanks. Based on a comprehensive consideration of the
atmosphere, soil, floating roof oil tank as well as the tank structure and the variations of the waxy crude
Keywords:
oil state and rheological behavior, general physical and mathematical models are established. In the
Single and double-plate floating roof oil
tanks
model, wax precipitation and gelatinization processes of waxy crude oil are described by the enthalpy-
Temperature drop porous media method. Non-Newtonian behavior is described by the Power law equation. Turbulent nat-
Wax precipitation and gelatinization ural convection is described by the LES method. SIMPLE algorithm is employed to couple pressure and
Non-Newtonian fluid velocity. Taking 1000 m3 single and double-plate floating roof oil tanks as examples, the evolution of
LES oil temperature and flow behavior is studied and the variations of the gel oil thickness and heat flux
are analyzed. Moreover, the differences between these two tanks are also discussed. Results show that
due to the structure difference of the tank roof, the temperature drop rates are 0.018 °C/h and
0.007 °C/h respectively in single and double-plate floating roof tanks in the case of this research.
Secondly, for the growth of gel oil on tank bottom, in double-plate floating roof tank, gel oil thickness
keeps growing and fluctuating, while in single-plate tank, the original gelled oil layer disappears firstly
and then increases gradually. Thirdly, although for both tanks, tank roof is the main part of heat dissipa-
tion towards the atmosphere, the maximum heat fluxes are respectively over 2.0 kW and 0.4 kW for sin-
gle and double-plate floating roof tanks, and the total average heat fluxes respectively are 1.49 kW and
0.59 kW.
Ó 2019 Published by Elsevier Ltd.
1. Introduction the total, oil would gel, which seriously threatens the process
safety and may even cause accidents. Therefore, it is of great signif-
Almost 80% of crude oil produced in China is waxy crude oil icance to study the thermal and gelatinization characteristics in
which is characterized by high pour point, high viscosity and com- the single and double-plate floating roof oil tanks during storage.
plex rheological property. Thus, in the practical production pro- There are two methods to study the characteristics of heat
cess, heating is always necessary. Single and double-plate transfer of waxy crude oil in an oil tank: experimental study and
floating roof oil tanks are the widely used oil tank in petrochemical numerical study. Due to its advantages in low cost, short research
industry. During storage, oil temperature drops due to the heat period and comprehensive research results, numerical simulation
transfer between oil and environment. When oil temperature is method is preferred by more and more researchers in recent years.
below wax precipitation point, wax crystals start to precipitate in Li et al. [3,4] investigated the temperature drop process of crude oil
the oil. When wax precipitation amount reaches 1–2% [1,2] among in a 10 104 m3 double-plate floating roof tank using experimen-
tal and numerical simulation method. The effects of crude oil prop-
⇑ Corresponding author. erty, tank roof type, initial oil storage temperature, solar radiation
E-mail address: yubobox@vip.163.com (B. Yu).
and liquid level on oil temperature drop were analyzed and
https://doi.org/10.1016/j.ijheatmasstransfer.2019.02.082
0017-9310/Ó 2019 Published by Elsevier Ltd.
458 M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469
discussed. However, in this study, phase change and variation of tionship between average oil temperature and storage time, oil
rheological behavior of oil during wax precipitation were neglected level and tank radius, total heat transfer coefficient, Ra number,
and the results can be applied only to the temperature drop char- Pr number were fitted. However, due the too many assumptions
acteristics of crude oil with low wax content. Yang et al. [5] and simplifications, the results are much different from the
adopted experimental method to study the characteristics of tem- practice.
perature drop in a 10 104 m3 floating roof tank in northeast In conclusion, although there have been some researches on
China. The profiles of the oil temperature in the core region and temperature drop process of waxy crude oil in floating roof tank,
along vertical direction in the tank were obtained and analyzed. many shortcomings exist as well. (a) Previous researches usually
However, in this research, the reason for temperature drop of oil treat oil state and rheological behavior change of waxy crude oil
in the tank was not investigated systematically. By considering during temperature drop process as the change of physical prop-
the relationship between oil property and temperature, Zhao erty. Also, the interaction between wax crystals and liquid waxy
et al. [6,7] studied the gelatinization process of waxy crude oil in crude oil are not considered accurately. (b) Previous researches
the floating roof tank. The gelatinization process of waxy crude commonly focus on double-plate floating roof tank while seldom
oil was described with the classical phase change model proposed on single-plate floating roof tank or compare both. (c) Previous
by Voller. However, during simulation, release of latent heat due to researches make much effort in the study of temperature drop of
wax gelatinization was regarded as heat capacity at constant pres- waxy crude oil inside the tank, while systematic exploration on
sure. Moreover, during the study of the gel oil growth, natural con- the mechanism of heat transfer and the growth of gel oil inside
vection was simplified into heat conduction. All these factors limit the tank is lacking. Therefore, in this study, with the consideration
the application of the research results. Wang et al. [8] applied of the variations of oil state and rheological behavior, general phys-
numerical approach to study the temperature drop characteristics ical and mathematical models of the temperature drop and gela-
in a model tank which was geometrically similar to 10 104 m3 tinization of waxy crude oil in single and double-plate floating
double-plate floating roof oil tank with the similarity ratio was roof oil tanks are established. 1000 m3 single and double-plate
1:10. Furthermore, the effects of physical properties of waxy crude floating roof oil tank are taken as examples to study the character-
oil on the temperature drop process were analyzed. However, this istics of the temperature drop and gelatinization of waxy crude oil.
research can only be applied to small-size floating roof tank. The evolution of the gel oil thickness and heat flux are studied and
Besides, in this research, neither the increase of gel oil thickness analyzed in detail. Furthermore, the difference between single and
nor the heat transfer process was analyzed. Wang et al. [9] inves- double-plate floating roof oil tanks on the characteristics of heat
tigated heat transfer characteristics of waxy crude oil in a transfer and gelatinization are also compared and analyzed
10 104 m3 floating roof tank for long-term storage using both systematically.
experiment and numerical simulation methods. On the aspect of
experimental investigation, temperature drop processes of waxy
crude oil at four different wax contents were analyzed, also the 2. Physical and mathematical models
variation of vertical temperature distribution against time was
obtained. On the aspect of numerical investigation, standard k-e 2.1. Physical model
model was applied to study the gelatinization process due to tem-
perature drop, as well as the melting process due to temperature The physical model of single and double-plate floating roof oil
increase. However, the author still didn’t analyze the gelatinization tanks contains the soil layer under the tank, air layer and steel
process of waxy crude oil deeply. Besides, the results of this layer in the tank roof (for single-plate floating roof oil tank, there
research were only applicable to double-plate floating roof tank. is no air layer), waxy crude oil in the tank, steel layer on the tank
Sun et al. [10] carried out the numerical study on the flow and heat bottom, steel layer and insulating layer on the tank wall. Thus, heat
transfer characteristics of waxy crude oil in the cooling and heating transfer in the tank involves the coupled heat transfer of gas-
processes during static tank storage, and the effects of the atmo- liquid-solid (for double-plate floating roof oil tank) or liquid-solid
sphere temperature, solar radiation on the temperature field were (for single-plate floating roof oil tank), heat conduction between
analyzed. Results showed that the cooling and heating processes tank and soil and forced convection between tank body and atmo-
can be divided into several stages and each stage displays different sphere. Predictably, heat transfer in both single and double-plate
characteristics. Moreover, solar radiation has great influence on the floating roof oil tanks is complex.
temperature fluctuation, and the atmosphere temperature has a The state of wax crystal and waxy crude oil state varies depend-
great influence on the temperature drop rate. Oliveski et al. [11] ing upon oil temperature [16]. Specifically, at the initial moment,
studied the temperature drop process in a small tank in 15 h using the oil temperature is relatively high and oil appears as pure liquid
both numerical and experimental methods, in which the Ra num- and shows Newtonian behavior. When oil temperature drops to
bers were fixed at 4.3 109 and 3.1 109 respectively. Oil temper- below wax precipitation point Tw, wax crystals begin to precipitate
ature distribution along the tank height at the center of the tank and solid-liquid dispersion system begins to form. At this time,
was obtained and compared with the experimental results. The waxy crude oil still shows Newtonian behavior. When oil temper-
comparison showed that these results agree well with each other. ature continuously drops to below abnormal point Ta, waxy crude
Oliveski [12] also studied the cooling process of the fluid with high oil begins to show non-Newtonian behavior. At this stage, the
Pr number fluid in tanks of different sizes using finite volume solid-liquid dispersion system still exists. When oil temperature
method. Correlations between the Nu number versus ratio of is below thixotropy appearance point Tt, wax crystals in the oil
height to radius, dimensionless total heat transfer coefficient, Ra begin to cross-link with each other and form wax crystal porous
and Pr numbers were obtained. However, these researches focused media [16], oil begin to gelate. Table 1 presents variations of the
on the small-size tank, without considering either phase change or state of wax crystal and waxy crude oil and rheological behavior
non-Newtonian property of crude oil. Besides, variations of heat with oil temperature.
flux and gel oil growth during temperature drop were not involved. Furthermore, the size of the oil tank is large and the turbulent is
Some researchers [13–15] also focused on small-size model tanks violent. For instance, the radius of the 1000 m3 floating roof oil
and simplified the turbulent natural convection into a laminar nat- tank is 6 m [17,18], and the Ra number reaches up to 2.0 1013
ural convection. Finite volume method was adopted to simulate when the temperature difference is 30 °C. Therefore, the computa-
the temperature drop process of crude oil numerically and the rela- tional burden and calculation time are unaffordable. Based on this,
M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469 459
Table 1
Variations of the states of wax crystal and waxy crude oil, rheological behavior with oil temperature [16].
the actual floating roof oil tank is simplified into a 1 rad one in this @ qui
study with the consideration of the axisymmetric of the oil tank, as ¼0 ð1Þ
@xi
Fig. 1 shows.
" !#
2.2. Mathematical model @ qui @ qui uj @p @ @ui @uj @ sij
þ ¼ þ l þ þ qg i þ sui
@t @xj @xi @xj @xj @xi @xj
General mathematical model of single and double-plate floating
roof oil tanks is established within the two-dimensional cylindrical
ð2Þ
coordinate system involving the non-Newtonian behavior
!
[16,19,20] and wax precipitation (phase change) [21–26]. Consid- @ qcp T @ qcp ui T @ @T @ cp qj
ering the coupled gas-liquid-solid (double-plate floating roof oil þ ¼ k þ sh ð3Þ
@t @xj @xj @xj @xj
tank) or liquid-solid (single-plate floating roof oil tank) heat trans-
fer process in the tank, solid (steel layer, insulating layer) is partic-
ularly considered as fluid with infinite viscosity to realize an sij ¼ q ui uj ui uj ð4Þ
integration solving. Besides, LES method is adopted to calculate
the turbulent flow and heat transfer of waxy crude oil in the tank.
The governing equations for the floating roof oil tank above the qj ¼ q uj T uj T ð5Þ
ground, which contains steel and air on tank top, waxy crude oil,
steel on tank bottom and steel and insulating layer on tank wall, where q, b and cp are mass density, volume expansion coefficient
are shown in Eqs. (1)–(5). and specific heat respectively. ui , p and T are the velocity, pressure
Fig. 1. Schematics of floating roof oil tank: (a) single-plate floating roof oil tank; (b) double-plate floating roof oil tank (front view); (c) floating roof oil tank (top view).
460 M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469
and temperature after filter. sui is source item caused by the binding l
sui ¼ ui ui;s ð11Þ
effect of wax precipitation. sh is the source item of latent heat of Kd
wax precipitation, which will be discussed later. sij and qj are the where ui, ui,s are the oil velocity and wax crystal velocity respec-
sub-grid scale stress tensor and sub-grid scale heat flux, which tively. Kd is the permeability of wax crystal porous media, which
can be determined by Eqs. (6) and (7) [27–31]. can be determined by the Kozeny-Carman equation [21–25], as
1 shown in Eq. (12).
sij ¼ 2lt Sij þ dij skk ð6Þ
3
ð1 g s Þ3
Kd ¼ K0 ð12Þ
g 2s
lt @ T
qj ¼ ð7Þ where K0 is the permeability constant and is set as 109 in this
Prt @xj
paper [22,23]. gs is the amount of the wax precipitation.
In Eqs. (6) and (7), Sij is the rate-of-strain tensor for the resolved Besides, latent heat would release during the temperature drop
process. In this research, the released latent heat is characterized
@u
scale, which is calculated by Sij ¼ 12 @u i
@xj
þ @xij [27–31]. Prt is sub-
by the source item sh in energy equation.
grid Prandtl number, which is fixed as 0.87 [26,27]. lt is sub-grid
scale viscosity. The model of the SGS viscosity is critical for LES, @ ðqDHÞ
sh ¼ ð13Þ
and the Smagorinshy-Lilly model [26–29] is applied in this @t
research in order to account for the flow near the wall, as shown where DH is latent heat. In this paper, DH = (1 gs)L, where L is the
in Eq. (8). total phase change latent heat.
lt ¼ qL2s S ð8Þ 2.3. Boundary and initial conditions
8
2.3.2. Initial condition < 0; T P Tw
Before computation, the oil tank is considered to be kept empty gs ¼ ð18Þ
: 5 Tw T
statically for a long time, during which the heat exchange between 9
%; T < Tw
tank and the environment reaches a balance. Then hot waxy crude
According to the correlation between the latent heat DH versus
oil is rapidly pumped into and fills the tank. Therefore, in this paper
the wax precipitation amount gs and the correlation between the
it is supposed that the initial temperature of the waxy crude oil, air
wax precipitation amount gs versus oil temperature T, it can be
layer, steel layer, insulating layer are all 30 °C. Soil surface temper-
derived that when oil temperature is below wax precipitation
ature is considered as equal to atmosphere temperature outside
point, f1(DH) can be determined by Eq. (19) [8,9].
the tank at the initial moment, which is 1 °C. Soil temperature at
different depth is obtained by linear interpolation between con- (
1
T w 180 1 DLH ; DH < L
stant temperature layer and soil surface temperature. f ð DH Þ ¼ ð19Þ
T; DH ¼ L
3. Numerical method and model validation
where wax precipitation point Tw, abnormal point Ta and thixotropy
appearance point Tt are 27 °C, 25 °C and 23 °C respectively.
3.1. Numerical method
Finite volume method is adopted to discretize the governing 3.2.2. Properties of the floating roof oil tank
equations, where transient term, convective term and diffusion In this research, numerical study is performed on the tempera-
term are discretized by the fully implicit scheme, GAMMA scheme ture drop and gelatinization characteristics in 1000 m3 single and
and central difference scheme respectively. Furthermore, SIMPLE double-plate floating roof oil tanks, the sizes of which x1, x2, x3,
algorithm based on non-uniform collocated grid is adopted to solve x4, x5, x6, r1, r2 and r3 are 5.0 m, 0.01 m, 8.01 m, 8.02 m, 8.52 m,
the discretized governing equations. To improve the computational 8.53 m, 6.0 m, 6.11 m and 11.11 m respectively [17,18]. Thus air
efficiency, multigrid method is introduced in this paper to solve the layer in double-plate floating roof tank, waxy crude oil, soil and
soil temperature under the tank bottom. The update of the latent insulating layer have x r respectively being 0.5 m 6.0 m,
heat content DH is crucial for calculation and the method in liter- 8.0 m 6.0 m, 5.0 m 11.11 m and 8.53 m 0.1 m. For a better
ature [21] is applied in this research. balance between computation accuracy and efficiency, tank roof,
tank bottom and tank wall are discretized with dense mesh, while
3.2. Properties of the waxy crude oil and floating roof oil tank tank core region is discretized with coarse mesh. According to trial
calculation, the meshes (x r) for air layer, waxy crude oil, soil and
3.2.1. Properties of the waxy crude oil insulating layer are respectively 65 101, 100 101, 66 138
A kind of waxy crude oil produced in an oil field in northwest- and 178 13. Taking the waxy crude oil region for example, the
ern part of China is selected and the properties of waxy crude oil minimum grid size in direction and in r direction are respec-
and wax crystal are obtained by experiment. Properties of air, steel, tively 5.78 104 m and 2.42 104 m. Besides, after a series of
soil and insulating layer are shown in Table 2 [3,4,8,9]. selection, time step in this research is set to be 2.0 s.
With the consideration of the variations of rheological behavior
and gelatinization process, the viscosity of waxy crude oil is calcu-
3.3. Model validation
lated by Eq. (15) [8,9].
8
> 3.3.1. Verification of grid sensitivity and time step sensitivity
>
> 0:058e0:0371 T ; T P Tw
< Based on the discretization introduced in Section 3.2.2, grid sen-
l¼ 0:058e0:0371 T ð15Þ
; Tw > T P Ta sitivity and time step sensitivity of the computation results are dis-
>
>
1g s
>
: cussed. Waxy crude oil in 1000 m3 double-plate floating roof tank
la ; T < Ta
during a 100 h temperature drop is taken for this discussion. The
Apparent viscosity la of waxy crude oil is determined by Eq. results are shown in Figs. 3 and 4. Note that in Fig. 3, the mesh
(10), where consistency coefficient K and flow behavior index n information is given in form of x r for the waxy crude oil domain.
is fitted by the experimental data [8,9], as Eqs. (16) and (17) In Fig. 3, totally three sets of grid systems are used for grid sen-
show. sitivity study. Based on the three sets of grids, temperature drop
processes of waxy crude oil in double-plate floating roof tank dur-
K ¼ 176:4424e0:354 T ð16Þ ing 100 h are calculated. Core region temperature and average oil
temperature inside the tank, total heat flux of the tank are com-
2
pared for the three sets of grids. The results based on the grid sys-
n ¼ 0:0157ðT T a Þ þ 1 ð17Þ
tem of 120 121 are relatively close to those on 100 101 grids,
Perform the DSC experiment on the waxy crude oil to obtain the which are much different than that on 80 81 grids. Therefore,
correlation between wax precipitation amount gs and oil tempera- in this research, all the results are based on grid system of
ture T, as Eq. (18) shows [8,9]. 100 101.
Table 2
Properties of different materials [3,4,8,9].
Material q cp k b l L
kg/m3 J/(kg °C) W/(m °C) °C1 Pas J/kg
Waxy crude oil 869 2100 0.14 7.53 104 Eq. (15) 2.0 105
Air 1.25 1009 0.025 3.7 103 1.77 105 –
Wax crystal 869 2100 0.14 – +1 2.0 105
Steel 7860 475 46 – +1 –
Soil 1700 753 1.6 – +1 –
Insulating layer 55 1880 0.02 – +1 –
462 M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469
Fig. 3. Comparison of the results calculated on different meshes: (a) oil temperature; (b) heat flux on each time step.
Fig. 4. Comparison of the results calculated on different time step: (a) oil temperature; (b) heat flux on each time step.
Fig. 4 shows good consistency in both the core region temper- 4. Results and discussion
ature and the heat flux based on the three different time steps.
However as with the profiles of average temperature, results 4.1. Temperature drop and gelatinization characteristics in the tank
based on time step of 1.0 s and 2.0 s agree quite well with each
other, and both have some deviations from the result at time step Temperature field in 1000 m3 single and double-plate floating
of 4.0 s. Therefore, in this research, time step of 2.0 s is finally roof tanks at different moments are shown in Figs. 6 and 7. From
selected. figures it can be seen that when waxy crude oil in the tank is not
totally gelled, there is a domain in both types of tank where oil
3.3.2. Model validation temperature keeps uniformly distributed as oil temperature drops.
In this subsection, the general model developed in this study is This domain is named as the core region of the oil tank. While
validated by the experimental data. Literature review shows that when oil in the tank becomes fully gelled, obvious temperature
most of previous experiment researches are for 10 104 m3 gradient is observed in the core area for both two tanks. The reason
double-plate floating roof tank. In our research, experiment mea- for this phenomenon is that at the initial moment oil temperature
surement is made on oil tank in northwestern and northeastern in the tank is 30 °C, while both the atmosphere and the soil surface
region of China [4,32]. are only 1 °C. Heat transfers from oil towards the atmosphere and
In this case, both oil tank radii are 40 m, liquid heights inside soil, forming turbulent natural convection in middle-upper region
the tanks are respectively 17.7 m and 19.44 m. The other informa- of the tank, which is named the core area. At the same time, the
tion can be found in Refs. [4,32]. Therefore, during validation, the bottom region of the tank presents a layer-distributed tempera-
computational domains of the two tanks are discretized by the ture, as shown in Figs. 6(a), (b) and 7(a)–(c). As time goes on, oil
same set of mesh. Grid systems for air layer, waxy crude oil temperature in the core area decreases continuously. When oil
domain, soil and insulating layer are respectively 97 241, temperature drops to below wax appearance point (27 °C), wax
160 241, 82 302 and 266 13, and the total number of grids crystals begin to precipitate. When oil temperature drops to below
is 95504. Based on trial calculation, time step is selected as 5.0 s. thixotropy appearance point (23 °C), the precipitated waxy crystals
Model validation is performed by comparing experiment results cross-link with each other and wax crystal porous media forms.
with numerical calculation results of the average oil temperature Waxy crude oil begins to gel and the core area shows obvious tem-
at the gauge hatch (r = 38.7 m). From Fig. 5 it can be seen that perature gradient, see Figs. 6(e) and 7(e).
numerical results and experimental data agree well with each Although oil temperature distributions in single and double-
other. plate floating roof tanks are similar to some extent, there are also
M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469 463
13 43
present 42 present
Li et al. [4] Xu [32]
12
41
40
11
39
10 38
0 100 200 300 400 500 0 100 200 300 400 500 600 700
time (h) time (h)
(a) (b)
Fig. 5. Comparison of the present results with experimental data [4,32]: (a) oil tank located in northwestern region of China; (b) oil tank located in northeastern region of
China.
Fig. 6. Temperature distribution in 1000 m3 single-plate floating roof oil tank: (a) 10 h; (b) 50 h; (c) 200 h; (d) 300 h; (e) 400 h; (f) 600 h.
some differences in temperature drop and gelling process. (1) Oil much smaller, which is only 0.007 °C/h. (2) During gelling, core
temperature drop rates largely differ. In single-plate floating roof region variation process is different. For double-plate floating roof
tank, oil temperature in the tank presents an averaged drop rate tank, the core area always exists and is located in the middle-upper
of 0.018 °C/h, while in double-plate floating roof tank, the rate is part of the tank, as shown in Fig. 7(a)–(c). In single-plate case, at
464 M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469
Fig. 7. Temperature distribution in 1000 m3 double-plate floating roof oil tank: (a) 10 h; (b) 200 h; (c) 700 h; (d) 900 h; (e) 1200 h; (f) 1600 h.
the initial moment, the core area is also in the middle-upper region weakened. At this time, for single-plate floating roof tank, heat dis-
of the tank, as shown in Fig. 6(a)–(b). After 200 h temperature drop, sipated from tank roof and tank wall can’t be quickly supplied from
the core area covers the whole tank, as shown in Fig. 6(c). The rea- the core region due to the weak natural convection of liquid oil,
son is that the double-plate floating roof tank has two steel layers forming irregular interfaces between cold and hot oil regions,
and an air layer between the plates, as shown in Fig. 1. With a bet- shown as Fig. 6(e). While for double-plate floating roof tank, series
ter heat insulation effect, waxy crude oil in double-plate floating vortexes of air fill the air layer on tank top, which are clockwise
roof tank flows weakly and steadily, with turbulent natural con- alternative with anticlockwise, as Fig. 8(a) shows. Heat transfers
vection remains in middle-upper part and heat conduction in bot- from waxy crude oil through the air layer inside tank roof to the
tom part. Comparatively the single-plate floating roof tank has a atmosphere. Thus there appears obvious interface between cold
high rate of heat exchange with environment. Turbulent natural and hot oil region in tank roof region, as Fig. 7(e) shows.
convection in middle-upper part of the tank and heat conduction The flow and heat transfer of the air layer on tank roof in
at the bottom only exists together at the initial period of tempera- double-plate floating roof oil tank are studied in this subsection
ture drop. As the oil temperature drops, turbulent natural convec- due to the significant influence on the heat transfer between waxy
tion develops towards the tank bottom until it covers the whole crude oil and atmosphere. Fig. 8 shows the air temperature, veloc-
tank after 200 s from the beginning, and a layer-shape temperature ity vector and stream function after 700 h temperature drop in
distribution disappears. At 300 s, the layer-distribution of the tem- 1000 m3 double-plate floating roof oil tank. It can be seen that
perature field at tank bottom appears again due to the small tem- the whole air layer is filled with vortexes, which are alternatively
perature difference between waxy crude oil and environment and positive and negative and the maximum air velocity is about
stable temperature distribution in air layer on tank roof and soil 0.4 m/s, as shown in Fig. 8(a).
under the tank bottom, as Fig. 6(e) shows. In Fig. 9, waxy crude oil temperature fields, velocity vector
When crude temperature drops below Tt, wax crude oil appears fields and stream function for both single and double-plate floating
as a porous structure built by wax crystals and liquid oil filling in it. roof tanks are shown when oil temperature in core area is 23.5 °C.
Flow strength of the liquid oil in the porous structure is largely In these two types oil tank, flow fields are largely different. In
M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469 465
Fig. 9. Oil temperature and velocity vector (I) and stream function (II) in 1000 m3 floating roof oil tank: (a) single-plate floating roof oil tank at 200 h; (b) double-plate floating
roof oil tank at 700 h.
466 M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469
Fig. 11. Oil temperature distribution on r = 5.52 m in 1000 m3 floating roof oil tank: (a) single-plate floating roof oil tank; (b) double-plate floating roof oil tank.
M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469 467
Fig. 12. Gel oil growth at r = 5.52 m on tank roof and tank bottom in 1000 m3 floating roof oil tank: (a) single-plate floating roof oil tank; (b) double-plate floating roof oil tank.
Fig. 13. Variations of the heat flux on tank roof, tank bottom and tank wall in 1000 m3 floating roof oil tank: (a) single-plate floating roof oil tank; (b) double-plate floating
roof oil tank.
transfer through tank wall is concerned, due to the existence of the From Fig. 14 it can be observed that the total heat in both two
insulating layer on both tanks, heat flux is relatively slow, and the tanks increases monotonically and in this process the increase rate
mean value around 0.15 kW. drops gradually. The average heat fluxes for single and double-
The main difference in heat transfer process between single and plate floating roof tanks are respectively 1.49 kW and 0.59 kW.
double-plate floating roof oil tanks lies in the heat transfer on tank Especially, a fast increase stage of the total heat can be seen from
roof. From Fig. 13, it can be seen that heat flux on tank roof in Fig. 14 at the early temperature drop period for both tanks. This
single-plate floating roof oil tank is much higher than that in phenomenon can be explained as follows. Initially, there is a large
double-plate one. Specifically, according to the condition in this temperature difference between the oil and the soil, which drives a
study, in case of single-plate floating roof tank, when the temper- rapid heat transfer from the oil to the soil, with an instantaneous
ature drop time is less than 210 h, waxy crude oil is not completely heat flux over 5.0 kW, as shown in Fig. 13. As time goes on, the heat
gelled, and turbulent natural convection is dominated in the tank. flux through the tank bottom reduces while the heat fluxes
At this moment, the maximum heat flux on tank roof reaches as
high as 2.0 kW. When gel oil begin to appear on tank roof, heat
transfers through wax crystal porous media to environment (sim-
ilar to heat conduction), and the maximum heat flux being 0.7 kW.
For double-plate floating roof tank, an air layer with a thickness of
1.0 m is contained between the double steel plates, indicating a
slower heat transfer process with the maximum heat flux being
0.4 kW. Furthermore, heat flux on tank roof, tank bottom and tank
wall fluctuates periodically with time, which is caused by the peri-
odical fluctuating of atmosphere temperature during storage.
The heat of waxy crude oil is transferred through tank roof, tank
bottom and tank wall respectively to the air layer, soil and insulat-
ing layer and finally to the atmosphere. Thus, studying the heat
transfer characteristics at different parts of the tank has great guid-
ance meaning to the design of tank heat insulation. Variations of
the total heat at different time instants are studied against the
temperature drop time. The results are given in Fig. 14 for both sin- Fig. 14. Variations of the total heat against temperature drop time.
gle and double-plate floating roof tanks.
468 M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469
through tank roof and tank wall increase gradually. Thus the total tive Teams and Teacher Career Development for Universities and
heat appears to rise gradually for the subsequent temperature drop Colleges under Beijing Municipality (No. IDHT20170507).
period.
5. Conclusions References
A general model of the temperature drop and gelatinization [1] M. Kane, M. Djabourov, J.L. Volle, J.P. Lechaire, G. Frebourg, Morphology of
paraffin crystals in waxy crude oils cooled in quiescent conditions and under
process for waxy crude oil in single and double-plate floating roof
flow, Fuel 82 (2003) 127–135.
oil tanks is established in this paper with the consideration of the [2] J.M. Letoffe, P. Claudy, M.V. Kok, M. Garcin, J.L. Volle, Crude oils:
variations of the oil state and rheological behavior. Taking 1000 m3 characterization of waxes precipitated on cooling by D.S.C. and
single and double-plate floating roof oil tanks as examples, the thermomicroscopy, Fuel 74 (1995) 810–817.
[3] W. Li, Q.Q. Shao, J. Liang, Numerical study on oil temperature field during long
temperature drop and gelatinization characteristics of waxy crude storage in large floating roof tank, Int. J. Heat Mass Transfer 130 (2019) 175–
oil are studied and the gel oil growth and heat transfer character- 186.
istics are analyzed. The following conclusions are obtained: [4] W. Li, Study, on numerical simulation methods and regularities for
temperature field of large floating roof oil tank Dissertation of Doctor
Degree, China University of Petroleum (Beijing), Beijing, 2013.
(1) There are many similarities in temperature drop and gelling [5] L. Yang, J. Zhao, H. Dong, J.Y. Liu, W.Q. Zhao, Research on temperature profile in
processes in these two tanks. Firstly, there is a core area in a large scaled floating roof oil tank, Case Stud. Thermal Eng. 12 (2018) 805–
816.
both tanks where oil temperature is approximately uniform [6] J. Zhao, H. Dong, Q.M. Lei, J.J. Guo, L.X. Wei, Research on heat transfer
due to turbulent natural convection when oil is not gelled. characteristic of waxy crude oil during the gelatinization process in the
When waxy crude oil is completely gelled, heat transfers floating roof tank, Int. J. Therm. Sci. 115 (2017) 139–159.
[7] J. Zhao, Heat transfer research and process scheme optimization of oil storage
through wax crystal porous media to environment (similar in alpine cold region Dissertation of Doctor Degree, Northeast Petroleum
to heat conduction), and obvious temperature gradient University, Daqing, 2013.
appears in the core areas of both tanks. Secondly, the gelled [8] M. Wang, X.Y. Zhang, G.J. Yu, G.J. Yu, J.F. Li, Numerical study on the
temperature drop characteristics of waxy crude oil in a double-plate floating
oil layer on tank roof and tank bottom for both tanks pre-
roof oil tank, Appl. Therm. Eng. 124 (2017) 81–90.
form an exponential growth. Thirdly, single-plate floating [9] M. Wang, B. Yu, X.Y. Zhang, G.J. Yu, J.F. Li, Experimental and numerical study on
roof tank has a similar heat transfer process on tank bottom the heat transfer characteristics of waxy crude oil in a 100,000m3 double-plate
and tank wall with double-plate one. And both tanks have floating roof oil tank, Appl. Therm. Eng. 136 (2018) 335–348.
[10] W. Sun, Q.L. Cheng, A.B. Zheng, T.Y. Zhang, Y. Liu, Research on coupled
fluctuating heat fluxes on tank roof along with the storage characteristics of heat transfer and flow in the oil static storage process under
time. periodic boundary conditions, Int. J. Heat Mass Transfer 122 (2018) 719–731.
(2) There are also obvious differences in temperature drop and [11] R.D.C. Oliveski, Correlation for the cooling process of vertical storage tanks
under natural convection for high Prandtl number, Int. J. Heat Mass Transfer
gelling processes between these two tanks due to the differ- 57 (2013) 292–298.
ent heat insulation capacity on tank roof. Firstly, tempera- [12] R.D.C. Oliveski, M.H. Macagnan, J.B. Copetti, A. de La M. Petroll, Natural
ture drop in single-plate floating roof oil tank is much convection in a tank of oil: experimental validation of a numerical code with
prescribed boundary condition, Exp. Therm. Fluid Sci. 29 (2005) 671–680.
faster, and the average rate is as large as 0.018 °C/h, which [13] I. Rodríguez, J. Castro, C.D. Pérez-Segarra, A. Oliva, Unsteady numerical
is only 0.007 °C/h in double-plate floating roof oil tank simulation of the cooling process of vertical storage tanks under laminar
according to the condition in this study. Secondly, the natural convection, Int. J. Therm. Sci. 48 (2009) 708–721.
[14] M. Cotter, M. Charles, Transient cooling of petroleum by natural convection in
growth of gelled oil is different. In double-plate floating roof cylindrical storage tanks-I. Development and testing the numerical simulator,
oil tank, gelled oil layer keeps a fluctuating growth, and gen- Int. J. Heat Mass Transfer 36 (1993) 2165–2174.
erally the growth becomes faster as oil temperature drops. In [15] M. Cotter, M. Charles, Transient cooling of petroleum by natural convection in
cylindrical storage tanks-II. Effect of heat transfer coefficient, aspect ratio and
single-plate floating roof oil tank, gelled oil layer succes-
temperature-dependent viscosity, Int. J. Heat Mass Transfer 36 (1993) 2175–
sively goes through three stages with time. At the first stage, 2182.
the gelled oil layer grows quickly, and then diminishes [16] X.H. Yang, Oil Pipeline Design and Management, China University of Petroleum
slowly until the disappearance. At the second stage, there Press, 2006.
[17] G.C. Guo, W.L. Dong, Z.L. Zhang, Design and Management of Oil Depot, China
is no gelled oil layer. At the third stage, the gelled oil layer University of Petroleum Press, 2006.
grows rapidly again. Thirdly, heat flux on tank roof in [18] C.F. Huang, Transportation Technology of Crude Oil Pipeline, Petrochemical
single-plate floating roof oil tank is much larger than that Press, China, 2005.
[19] Md.A. Ikbal, S. Chakravarty, K.K.L. Wong, J. Mazumdar, P.K. Mandal, Unsteady
in double-plate floating roof oil tank. In single-plate floating response of non-Newtonian blood flow through a stenosed artery in magnetic
roof oil tank, the maximum heat flux on tank roof is over field, J. Comput. Appl. Math. 230 (2009) 243–259.
2.0 kW, and the average heat flux after gelling keeps around [20] S.W. Yuan, Numerical simulation and experimental study of power-law fluid
Dissertation of Master Degree, East China University of Science and
at 0.7 kW. While in double-plate floating roof oil tank, the Technology, China, 2014.
largest heat flux is only 0.4 kW, and the average heat flux [21] V.R. Voller, C. Prakash, A fixed grid numerical modelling methodology for
after gelling is only 0.27 kW in this study. On the whole, convection-diffusion mushy region phase-change problems, Int. J. Heat Mass
Transfer 30 (1987) 1709–1719.
the total average heat fluxes of single and double-plate float- [22] K.E. Omari, T. Kousksou, Y.L. Guer, Impact of shape of container on natural
ing roof tanks are respectively 1.49 kW and 0.59 kW. convection and melting inside enclosures used for passive cooling of electronic
devices, Appl. Therm. Eng. 31 (2011) 3022–3035.
[23] E. Assis, L. Katsman, G. Ziskind, R. Letan, Numerical and experimental study of
Conflict of interest melting in a spherical shell, Int. J. Heat Mass Transfer 50 (2007) 1790–1804.
[24] B.J. Jones, D. Sun, S. Krishnan, S.V. Garimella, Experimental and numerical
study of melting in a cylinder, Int. J. Heat Mass Transfer 49 (2006) 2724–2738.
The authors declared that there is no conflict of interest.
[25] W. Kowalczyk, C. Hartmann, A. Delgado, Modeling and numerical simulation
of convection driven high pressure induced phase change, Int. J. Heat Mass
Transfer 47 (2004) 1079–1089.
Acknowledgement
[26] ANSYS FLUENT Theory Guide, ANSYS Inc Press, 2011.
[27] G.E. Lau, G.H. Yeoh, V. Timchenko, J.A. Reizes, Large-eddy simulation of natural
This research is supported by the Fundamental Research Funds convection in an asymmetrically-heated vertical parallel-plate channel:
for the Central Public Welfare Research Institutes of China (Grant assessment of subgrid-scale models, Comput. Fluids 59 (2012) 101–116.
[28] Z. Tao, Z.Y. Cheng, J.Q. Zhu, D.S. Lin, H.W. Wu, Large eddy simulation of
Nos. TKS180403, TKS160223), National Natural Science Foundation supercritical heat transfer to hydrocarbon fuel, Int. J. Heat Mass Transfer 121
of China (No. 51606117) and the Project of Construction of Innova- (2018) 1251–1263.
M. Wang et al. / International Journal of Heat and Mass Transfer 136 (2019) 457–469 469
[29] Y.H. Qi, T. Ishihara, Numerical study of turbulent flow fields around of a row of [31] J. Ruiz, A.S. Kaiser, B. Zamora, C.G. Cutillas, M. Lucas, CFD analysis of drift
trees and an isolated building by using modified k-e model and LES model, J. eliminators using RANS and LES turbulent models, Appl. Therm. Eng. 105
Wind Eng. Ind. Aerodyn. 177 (2018) 293–305. (2016) 979–987.
[30] L.T. Zhang, S.M. Luo, Y.P. Zhang, W.X. Tian, G.H. Su, S.Z. Qiu, Large eddy [32] X.Y. Fang, Study of oil storage at lower temperature in large floating roof tank
simulation on turbulent heat transfer in reactor vessel lower head corium of Qinhuangdao pump station Dissertation of Master Degree, China University
pools, Ann. Nucl. Energy 111 (2018) 293–302. of Petroleum (Beijing), Beijing, 2000.