th

10 International Conference on Thermal Engineering: Theory and Applications

February 26-28, 2017, Muscat, Oman

Weathering of stored Liquefied Natural Gas (LNG)

Calogero Migliore1,2, Amin Salehi1, Velisa Vesovic1

1
Department of Earth Science and Engineering, Imperial College London,
London SW7 2AZ, United Kingdom
2
Repsol, S. A., calle de Méndez Álvaro, 44, 28045, Madrid, Spain

Abstract
A model for predicting the weathering of LNG stored in containment tanks has been recently developed that allows
the temperature of vapor to be different to that of LNG. The model is used to analyze the heat transfer between vapor and
LNG, by means of an effective thermal conductivity. The simulation results indicate that the temperature of the vapor
phase will be higher than that of the LNG, and that increase is a function of the effective thermal conductivity. Thus,
demonstrating that the effective thermal conductivity could be used as a proxy to match the results to measurements.
This has important implications for weathering models used in industry, which currently assume isothermal conditions
within the containment tanks.

Keywords: LNG, boil-off gas, weathering

assumption that the vapor phase above the LNG is at the
same temperature as the weathering LNG. We will refer
1. Introduction to this model as the ‘isothermal’ model. Most, if not all of
the models available so far [4,5 and references there in]
The current consensus is that natural gas will make use of this assumption. However, scant industrial
continue to play an important part in the global energy evidence suggests that this is not the case in the
mix [1]. Liquefied natural gas (LNG), is seen by an commercial LNG storage tanks. Hence, in the later model
increasing number of producers and consumers as a [3], we did not invoke the assumption of thermal
viable option for transportation of natural gas. As a equilibrium and have separated the heat flux entering the
valuable energy resource LNG is stored on a commercial LNG from the heat flux entering the vapor section of the
scale in large, highly insulated storage tanks at its boiling storage tank. This has important consequences on the
temperature and slight overpressure. The LNG is stored dynamics of the heat transfer, as the heat that enters the
at cryogenic temperatures for significant durations and vapor phase leads to an increase in the vapor
inevitable heat ingress from the surroundings into the temperature, while the heat influx into the liquid phase
storage tank will lead to vaporization. The more volatile leads to evaporation of the LNG. If the process of heat
components (methane and nitrogen) will vaporize transfer from the vapor to the liquid is slow a temperature
preferentially, resulting in weathering of LNG. If left difference between the two phases will be established
unchecked, the weathering can render the remaining and the vapor temperature will be higher than that of the
LNG unsellable, because of regulatory requirements. liquid, which remains at its boiling temperature
Furthermore, the weathering increases the overall tank throughout the weathering. We will refer to this model as
pressure and in order to avoid over-pressurization some the ‘non-isothermal’ model. The conference presentation
of the generated vapor is removed, as boil-off gas (BOG). will present the results from both models, illustrating the
The industry is specifically concerned in minimizing the importance of treating both the heat transfer and the
BOG and ensuring that the weathering does not greatly thermodynamics correctly. It will demonstrate that the use
impact on the LNG quality. In particular weathering of a more sophisticated heat transfer model leads to a
prediction is used in planning operations, thus ensuring more realistic estimation of the BOG rate that provides
appropriate allocation of LNG cargoes, its compatibility industry with the possibility of implementing less
with stored LNG and avoiding catastrophic events conservative practices with additional cost benefits.
involving stratification, sudden vapor release and rollover. In this paper we focus on the, so far unreported, role
There is thus a need for developing models that can of using an effective thermal conductivity as a proxy
predict the weathering phenomena. This is not an easy measure of the efficiency of heat transfer between vapor
task, as interplay between a number of simultaneous and liquid that allows for ‘bridging’ between the
phenomena, requires a careful consideration of the heat isothermal and non-isothermal models.
transfer and thermodynamics of the system.
We have recently developed two models of
weathering of LNG in storage, containment tanks [2,3]. In
an early model [2] we make use of a standard

1
 used as an effective parameter one can simulate, in a
simple manner, the effect of convection, if any is present.
2. Weathering Model
In the limit of thermal conductivity reaching an infinite
value the non-isothermal model will tend to an isothermal
Figure 1 illustrates the schematic of the LNG storage one. Furthermore, the effective thermal conductivity
tank with the emphasis on the relevant heat transfer. The offers a straightforward way to match the model
heat ingress from the surroundings, through the lateral prediction to the experimental measurements of the
wall of the storage tank, is split into two components, vapor temperature in the real LNG containment tanks,
QVin, and QLin, that represent heat influx into the vapor when and if such measurements become available.
and liquid phases, respectively. The constant heat influx
from the thermal slab underneath the storage tank will
provide an additional source of heat, Qslab. In the non- 3. Results
isothermal model the LNG will also receive heat from the
vapor phase, designated QVL in Fig. 1. The overall heat
Here we report on the simulations of weathering
ingress will lead to the weathering of LNG measured by
behavior of an LNG mixture in order to elucidate the
the rate of vaporization 𝐵𝐵̇𝐿𝐿 . In order to maintain the effect of QVL, by means of effective thermal conductivity
constant pressure BOG will be released and the rate of acting as a proxy. The commercial mixture in question
vapor removal is designated 𝐵𝐵̇. was termed a ‘light LNG’ in our previous work [2,3] and it
primarily consists of methane, with a small amount of
ethane present. The results of weathering behavior are
reported for a period of one year and refer to a standard
3
165,000 m containment tank initially filled with 160,000
3
m of LNG. We refer the reader to Refs [2,3] for further
details of the tank, simulation set-up and the composition
of the light LNG.
Figure 2 (a) illustrates that if the heat transfer
between the vapor and LNG, QVL, is by conduction only
the vapor temperature will increase by approximately 7.6
K over the period of one year [3]. As we increase the
effective thermal conductivity the vapor temperature
decreases. For values in excess of 100k the temperature
increase is of the order of 1 K which approximately
corresponds to the temperature increase observed for a
case where the heat transfer is by fully-developed natural
convection, as reported in Ref [3]. For higher values of k
the non-isothermal model will tend to an isothermal one
and the vapor temperature will tend to the value of the
LNG boiling temperature. As far as we are aware no
reliable measurements of the vapor temperature in the
Fig. 1. Schematic of the heat exchange between the industrial tanks are available. However, the circumstantial
surroundings, LNG and vapor evidence tends to support a conduction case. Until such
measurements become available it is not possible to fully
The system is governed by a coupled set of validate the model, nor is it possible to use effective
differential energy equations thermal conductivity to perform the matching, in case
such a matching is necessary.
𝑑𝑑𝐻𝐻V Figure 2 (b) illustrates the behavior of the BOG rate
𝑈𝑈V 𝐴𝐴V (𝑇𝑇air − 𝑇𝑇V ) − 𝑄𝑄VL = + 𝐵𝐵̇ℎV (𝑇𝑇V ) − 𝐵𝐵̇L ℎV (𝑇𝑇L )
𝑑𝑑𝑑𝑑 as a function of weathering duration. We observe that
irrespective of the magnitude of the effective thermal
𝑑𝑑𝐻𝐻L
𝑈𝑈L 𝐴𝐴L (𝑇𝑇air − 𝑇𝑇L ) + 𝑄𝑄VL + Qslab = + 𝐵𝐵̇L ℎV (𝑇𝑇L ) (1) conductivity the BOG rate decreases over the analyzed
𝑑𝑑𝑑𝑑
period. As indicated before [3] the observed decrease
where U is the overall heat transfer coefficient, A is can be entirely attributed to a decrease in a wetted
the contact area H and h are the enthalpy and molar contact wall area. As LNG weathers, the liquid level
enthalpy, respectively, while the subscripts V and L refer drops and the heat transfer from the surroundings
to the sections of the tank filled with liquid and vapor, decreases. It is interesting to note that over the period of
respectively. For the isothermal model Eq. (1) reduces to a year the choice of a heat transfer mechanism results in
a change of overall BOG of 170 tons which represents
𝑑𝑑𝐻𝐻V 𝑑𝑑𝐻𝐻L approximately 4.4% of the total BOG generated.
𝑈𝑈𝐴𝐴(𝑇𝑇air − 𝑇𝑇) + Qslab = + + 𝐵𝐵̇L ℎV (𝑇𝑇) (2) Figures 2 (c) and 2(d) illustrate the evolution of
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
vapor-liquid heat transfer and overall heat transfer from
The right-hand side of Eqs. (1-2) contains only the surroundings into LNG, respectively. A comparison at
thermodynamic quantities and their evolution with time, 52 weeks shows that the vapor-liquid heat transfer by
and can in principle be obtained from an appropriate conduction has the potential to decrease QVL by up to 3
thermodynamic model [2,3]. times or the cumulative overall heat transfer across the
If we take the limiting case that the heat transfer from 12-month period by 47%. This decrease could result in a
the vapor phase to LNG is by conduction only, then the notable impact on the overall rate of LNG heat input,
thermal conductivity of the vapor phase governs the rate Figure 2 (b), particularly at higher values of k and at later
of heat transfer and consequently the temperature of the times during the weathering process. Similarly, a
vapor phase. It is clear that if the thermal conductivity is comparison at 52 weeks reveals that QVL contributes

2
between 2.4% to 6.9% depending on the value of k, to
the total heat input to LNG,

Fig. 2. The effect of using the effective thermal conductivity on: (a) Vapor temperature, TV; (b) BOG rate; (c) Vapor-liquid heat
transfer, QVL, and (d) Overall LNG heat input as a function of weathering duration.

4. Conclusion References

Use is made of a recently developed model that [1] International Energy Agency. World Energy Outlook:
allows for a variation in the temperature of the vapor Are we entering a golden age of gas?. Paris:
blanket in contact with weathering LNG, stored in a International Energy Agency; 2011.
containment tank. The model is used in a predictive
mode to analyze the heat transfer between vapor and [2] Migliore C., Tubilleja C., Vesovic V. Weathering
LNG by means of effective thermal conductivity. The prediction model for stored liquefied natural gas
results indicate the effective thermal conductivity is a (LNG). J. Nat. Gas Sci. Eng. 2015;26:570-580.
good proxy for the heat transfer mechanism. The [3] Migliore C., Salehi A., Vesovic V. A non-equilibrium
simulations demonstrate that the vapor temperature approach to modelling the weathering of stored
increases in line with circumstantial industrial evidence Liquefied Natural Gas (LNG). Energy (submitted).
and that BOG decreases, as a function of weathering
time. [4] Miana M., Del Hoyo R., Rodrigálvarez V.
Comparison of evaporation rate and heat flow
models for prediction of liquefied natural gas (LNG)
ageing during ship transportation. Fuel 2016;177:87–
106.
[5] Pellegrini L.A., Moioli S., Brignoli F., Bellini C. LNG
Technology: The weathering in above-ground
storage tanks. Ind. Eng. Chem. Res. 2014;53:3931–
3937.