analysis of oil tanker deck under hydrocarbon fire

INTERNATIONAL JOURNAL OF MODELING AND SIMULATION FOR THE PETROLEUM INDUSTRY, VOL. 8, NO. 2, DECEMBER 2014 17

Analysis of Oil Tanker Deck Under Hydrocarbon Fire

Manco, M. R.1, Vaz, M.A.

1, Cyrino, J.C.R.

1 and Landesmann, A.

2

[email protected], [email protected], [email protected], [email protected]

1Ocean Engineering Program, COPPE / UFRJ, Rio de Janeiro, Brazil 2Civil Engineering Program, COPPE / UFRJ, Rio de Janeiro, Brazil

Abstract. The present paper presents a numerical study of the

behavior of a deck panel of an oil tanker under fire. The fire

condition assumed in the simulation is that of a hydrocarbon

burning process according to the time-temperature nominal

curve in the part 1.2 of EUROCODE 1 and 3 (EC1 and EC3,

2004), which also specify the variations of mechanical and

thermal properties of the steel with temperature. A finite

element model taking into account the initial imperfections

recommended by the ISSC, 2012 was developed through the

use of ABAQUS , 2011 commercial software. The thermal

and mechanical analyses were uncoupled, so the temperature

transient field caused by the fire condition was analyzed first

and then, this thermal load was applied to the structure to

evaluate its mechanical behavior. Once the fire scenario was

defined, it was possible to assess the development of the

temperature field as a function of time. The induced thermal

loads were considered in the analysis of the structural model

together with the pre-existent operating loads, allowing

assessing the behavior of the panel. Thus, the objective of the

present paper is to present a methodology for assessing the

structural behavior of a steel panel in the event of fire.

Index Terms stiffened panel; fire; finite element

1 INTRODUCTION

The design of steel structures for the offshore exploration and

production of oil and gas provides for the consideration of

different scenarios related to severe accidents, including: large

waves, extreme winds, earthquakes, collision between vessels

and fires. For all these conditions, the integrity of the facility

must be ensured and, in some cases, damages must be limited,

ensuring the maintenance of safe operating characteristics for

a given period of time. Due to the presence of large volumes

of flammable materials (liquid and gas), equipment operating

at high temperatures, active flames (e.g., flares) and human

lives, living and interacting in confined spaces, the manufac-

turers and operators of these facilities are required to follow

strict fire protection criteria (deemed the most severe among

all industrial facilities).

The occurrence of a fire in this type of structures is considered

one of the most unfavorable conditions, for the combustion of

hydrocarbons presents a very high rate of temperature increase

in the initial stage of the fire, causing a very rapid loss in me-

chanical properties of steel, thus generating the possibility of

deaths and economic and environmental damages.

Tragic examples of this kind of accidents are those which took

place in the Piper Alpha platform in the UK, in July 1989,

killing 167 of the 229 occupants in less than 22 minutes and,

most recently, at the Deepwater Horizon platform, in the Gulf

of Mexico in April 2010, where 11 people disappeared and a

large environmental impact was generated.

Temperature changes generate degradation of the mechanical

properties causing effects that significantly change the failure

mode forecasted in the panel design (at room temperature).

Such effects are related to the introduction of axial and shear

forces, as well as bending moments due to the thermal gradi-

ent, causing the local buckling of the web (WB) and the flange

(FB) at the ends of the stiffener and its consequent collapse

because of the emergence of plastic hinges and subsequent

membrane behavior of the panel plate until the complete fail-

ure of the panel (Manco et al., 2013). In addition, the adopted

boundary conditions (described in item 3) facilitate the prob-

lem modeling due to the consideration that the properties in

the restraint do not change with temperature, but generate a

numerical problem in the mechanical analysis which is solved

by considering an additional region called Rigid End (Has-

sanein, 2011).

The numerical analyses are performed using the commercial

code ABAQUS, 2011, according to the finite element meth-

od (FEM), taking into account the structural and thermal ef-

fects resulting from the proposed fire. Variation in thermal and

mechanical properties of materials in case of high temperature

conditions are taken into account in the analysis, in accord-

ance with the applicable standard recommendations, such as is

the case of part 1.2 of EC3, 2004. The fundamentals of the

applied analysis model are described in item 2 below. A case

study is proposed and briefly described in item 3 hereto. A

tanker ship deck panel, with the geometry employed by the

ISSC, 2012 in their benchmarking studies, is submitted to a

fire scenario, caused by the burning of hydrocarbons (Part 1.2

of EC1, 2004), allowing to evaluate the thermo-structural

behavior for different instances of the fire. The main results

obtained with the numerical model are presented and critically

assessed in item 4 of this work, taking into account the ful-

fillment of security requirements. The main conclusions ex-

tracted from the analyses performed are mentioned in item 5,

indicating that this methodology allows assessing qualitatively

and quantitatively the behavior of the panel, and can be used

in the improvement of current regulations related to the safety

of structures in the event of fire.

Manuscript received Jan. 26, 2014. Corresponding Author: Miguel Renato

Manco Rivera (E-mail: [email protected]).

MANCO et al.: ANALYSIS OF OIL TANKER DECK UNDER HYDROCARBON FIRE 18

2 ANALYSIS METHODOLOGY

The analysis is carried out using FEM, the model includes a

direct and rigorous consideration of nonlinear physical and

geometric effects on the numerical formulation, allowing the

estimate of the possible structural collapse modes.

The proposed analysis procedure begins with the review of the

panel layout with the selection of the fire scenario. Then, the

thermal analysis is performed, which purpose is to determine

the variation of the temperature in the elements exposed to

fire. The main numerical formulation aspects of this stage are

addressed in item 2.1. The final stage of the procedure aims at

determining the structural behavior as a function of the

elapsed time of fire, in other words, depending on the thermal

conditions of fire exposure and applied external loads (me-

chanical). Computational characteristics adopted in this final

stage of the numerical simulations are briefly described in

item 2.2 hereto.

2.1 Thermal Analysis

The numerical model used FEM to solve the two-dimensional

transient heat conduction problem, as shown in Cook, 2002,

Skallerud and Amdahl, 2002, Lewis et al., 2004 and Landes-

mann et al., 2010. The DS4 element made with 4 nodes was

used to represent the panels.

The partial differential equation which expresses the tempera-

tures (in degrees Celsius) (x,y,z,t) is shown in Equation (1), subject to a temperature field defined in its contour s, which is represented in this analysis by fire-temperature curves (Part

1.2 of EC1, 2004).

tc

zk

zyk

yxk

x

..... (1)

Where is the specific mass of steel (assumed temperature independent), =7850 kg/m, c is the specific heat and k is the thermal conductivity. In this paper, the thermal properties of

steel, as a function of temperature, are provided by part 1.2 of

EC3, 2004 and shown in Figure 1.

When prescribed temperatures are different from temperatures

on the surface, a heat flux qn with two portions is generated:

(i) one due to convection and another (ii) resulting from radia-

tion, which can be written in a single equation through the

linearization of the radiation portion, as below:

gsqqq .rcrcn (2)

where: gsgs ... 22rrr , r is the resulting emissivi-ty, defined as 0.8 (for steel); r is the Stefan-Boltzmann con-

stant (5.67 .10-8

W/mK4); and c is the convective heat coeffi-

cient adopted as 50 W/mK (part 1.2 EC1, 2004). This lineari-

zation of the portion of radiation is necessary given that the

FEM only solves a system of linear equations. Denoting C as

capacitance matrix, Kl and Kc are conductivity matrices (Kt=

Kl+Kc), fb as vector of nodal flux due to convection, Equation

(1) can be rewritten:

Figure 1. Specific heat and thermal conductivity of carbon steel as a function

of the temperature

)()(.)(

. tftKt

tC bt

(3)

Solution of Equation (3) is based on FEM, being possible to

determine the temperature at time n+1 based on data at time n:

bnbnbnntnt ffftKtCtKC 11 1 (4)

where t is the time interval, is the temporal integration factor (taken as 0.9) and the initial temperature throughout the

solid is assumed to be equal to 20C (o). Analyses presented here use the nominal fire curve corre-

sponding to the burning of hydrocarbons (Part 1.2 of EC1,

2004), as given by Equation (5):

tt eet .5.2.167.0og .675.0.325.01.1080)( (5)

where: t is the elapsed time of fire (in minutes), a is the tem-

perature in the middle (in C) and o is the initial temperature

(equal to 20C).

2.2 Structural Analysis

Since the variation of the temperature field was established in

the previous analysis stage, the finite element mesh used, i.e.,

the nodal coordinates, the elements connectivity and the re-

sults for heat fluxes are used in the simulation of structural

behavior under the postulated fire conditions. The procedure is

initialized by the application of external loads, including the

own structural weight, fluid action and other operational loads.

At this stage, deformations and their respective stresses, corre-

sponding to normal operating conditions of the panel, can be

seen. The variation of the temperature field determined in the

thermal analysis is imposed to the structural model along with

other external loads applied.

In building the mesh of finite elements for the structural anal-

ysis, the S9R5 element is used for the panel simulation. This

element is composed of 9 nodes and 6 degrees of freedom per

node (translations and rotations around global axes X, Y and


Z), with capacity for developing nonlinear physical and geo-

metrical analyses. The complete Newton-Raphson solution

process is adopted to update the matrices and the linear solu-

tion of equations. The von Mises criterion is adopted for de-

termining the element plastification criterion.

Apart from the thermal deformation imposed on the structural

model, variations in the mechanical properties of steel as a

result of temperature, as shown in Figure 2, are also taken into

account, including reduction: of yield strength (y,), modulus of longitudinal elasticity (Ea,) and yield point (p,) obtained based on recommendations of part 1.2 of EC3, 2004.

Figure 2. Stress-strain relationship for carbon steel at elevated temperatures.

Defining y,20 as the characteristic yield stress and Ea,20 as the modulus of elasticity, at environment temperature, the degra-

dation factors as well as the coefficient of linear thermal ex-

pansion of steel, as a function of temperature (), are shown in Figure 3.

Figure 3. Reduction factor for the stress-strain relationship and thermal

expansion coefficient for carbon steel at elevated temperatures.

3 CASE STUDY

We studied a stiffened steel panel, which is part of the deck of

an oil tanker, subjected to fire caused by hydrocarbon burning.

The temperature of the hot gases inside the compartment un-

der fire (on the side of the stiffeners) is described by Eq. (5),

while the external temperature was considered constant and

equal to 20 C. In the heat exchange process between the up-

per side of the plate and the environment we considered a heat

exchange coefficient (including radiation and convection) of 9

W/mK according to part 1.2 of EC1, 2004 recommendations

as shown in Figure 4. The geometry of the panel was chosen

similar to the ISSC, 2012 benchmark study and is shown in

Figure 4. The initial geometric imperfections of the panel were

based on the model recommended by ISSC, 2012 according to

Eqs. (7) and (8), where oplv is the imperfection in the plate,

osw is the imperfection in the stiffener and is a parameter

that defines the level of imperfection according to Smith et al.

(1992) (slight =0.00025, average =0.0015 and severe = 0.0046). In the case study we employed the three levels of initial imperfections of the stiffener to assess the influence on

the behavior of the panel under fire. Figure 5, presents the

initial imperfections magnified fivefold.

Figure 4. Case study and initial panel geometry


puopl

b

y

L

xmv

sinsin1.0 2 (6)

uwuos

L

x

h

zLw

sin (7)

The computational mesh used is composed of three regions.

The first region, in blue in Figure 5, is called Rigid End and

simulates the panel support element (bulkhead or deck trans-

verse) the mechanical properties of which are independent

from temperature. This end is used to avoid numerical errors

caused by the application of boundary conditions in the me-

chanical analysis. The second region, in gray in Figure 5,

covers length uL /12 from the rigid end and has a higher mesh

density to allow assessing the WB and FB of the stiffener that

will occur in that area. Finally, the third region in green in

Figure 5 is considered with a less dense mesh to avoid work-

ing with very large matrices. As boundary conditions we con-

sidered the left cross-section of the Rigid End (see Figure 5) as

restrained, while for the right end of the region with less mesh

density the symmetry condition was considered to work only

with half the panel and reduce the computational cost. Finally,

in the side edges we considered the boundary conditions that

represent the continuity of the panel, as shown in Figure 5.

Other boundary conditions employed in similar problems can

be seen in the studies of Heninisuo & Aalto, 2008 and

Vimonsatit et al., 2005, among others. As loads, we consid-

ered only the structure own-weight and three values for side

pressure of 0.01, 0.02 and 0.03 MPa for each of the initial

imperfection levels considered in the stiffener.

Figure 5. Initial geometric imperfections and boundary conditions

4 RESULTS

In all the cases analyzed the failure on the deck presents the

same behavior, only varying the severity of the stiffener dis-

tortion and the time it takes to occur as described below.

4.1 Temperature Field

As expected, the stiffener web is the element that heats more

quickly, because it has a greater massivity factor, i.e., presents

a larger area exposed to fire with a relatively low volume,

when compared to the plate or the stiffener flange. The flange

has an almost uniform heating as evidenced by the little tem-

perature difference between the points D and E. In the plate,

the temperature difference between points A and B, indicates

the presence of a heat flow towards the stiffener. Note that the

temperature on the plate is always lower than on the stiffener

due to the heat exchange with the environment.

Figure 6. Temperature in points A, B, C, D and E, in view of time

4.2 Stress-strain Field

Since we choose a fire scenario caused by hydrocarbon burn-

ing, considered the most severe possible fire, among the sim-

plified models proposed by the part 1.2 of EC1, 2004, the

temperatures of the different panel components increased very

quickly, originating a pronounced drop of the mechanical

properties of the structural elements. According to Yang and

Gao (2004), those temperature increases, besides changing the

thermal and mechanical properties, generate temperature gra-

dients in the panel constitutive elements, resulting in non-

linear forces and moments which, combined with the increas-

ing of imperfec-tions due to thermal deformation, change the

behavior of the panel. Thus, in all the analyses developed from

the data on panel temperature field variation, we determined

the status of stress and strain at different times of the postulat-

ed fire.

Figure 7, presents the stress fields at the beginning of the heat-

ing, i.e., after considering the effect of gravity and the corre-

spondent lateral pressure L.P. It was observed that the initial imperfection does not significantly affect the results, changing

the stress magnitude by less than 2%. In this figure, the cases

with an average stiffener imperfection for each of the side

pressures considered (0.01, 0.02 or 0.03 MPa) are presented.

During the heating, the panel suffers non-uniform thermal

deformations that change the initial stress field, giving rise to

regions where the plastic regime is reached, generating plastic


hinges in the area close to the restraint and in the middle sec-

tion of the panel. Those plastic hinges begin with the stiffener

WB and subsequently with the FB.

Figure 7. Stress fields after the application of gravity and L.P. considered

(panel with average imperfection)

Figure 8 shows the distribution of von Mises stresses ( ,vM ),

axial stresses in the S11 and S22 directions ( ,11 and ,22 ,

respectively) and the shear stress ( ,12 ), normalized in rela-

tion to the yield stress (at temperature determined at the

point at that instant) in the stiffener web. The continuous,

dashed and dotted lines represent the stress distribution for

times of 0 (beginning of heating), 5 and 10 min., respectively

and the colors black, red and blue define the cross-sections

located at 2/uLx , 36/uLx and 72/uLx , respective-

ly. For not having trouble viewing (since the stiffener de-

forms) we used a local coordinate on the vertical axis at the

initial time ( LoY , , the origin of which is the junction of the

plate and the web). In this figure we observe that the von Mis-

es stress has a similar configuration as the axial stress ,11 ,

indicating that it prevails over the others ( ,22 and ,12 )

and that after the appearance of the plastic hinge (graph ,11

black dashed or dotted line) the panel supports only tensile loads and basically shows the panel yield (membrane behav-

ior) until it breaks.

Figure 9 shows the geometric configuration of the panel in the

region close to the restraint ( 36/uLx and 72/uLx ) and

in the middle of the panel ( 2/uLx ) at different times, for

the case of a severe imperfection in the stiffener and a L.P. of

0.03 MPa. This figure evidences WB and FB of the stiffener in

the region close to the restraint (plastic hinge) and maximum

vertical displacement (MVD) as well as maximum transversal

displacement (MTD) in the middle of the panel.

Table 1, presents the analysis times, until panel failure for

each of the cases evaluated. We found that neither the imper-

fection level nor the magnitude of the side pressure (the value

of which was triplicated) significantly changes the analysis

time, varying 5% at the most. This happens because the prob-

lem is governed by temperature and not by the geometric

configuration or the load. MVD and MTD magnitudes are also

presented showing that their values are directly affected by the

S.P., on the other hand it also shows that the level of imperfec-

tion has a very large influence on the value of MTD and on the

stiffener final configuration.

Figure 10 shows the final geometric configurations of the

stiffener web for all the cases assessed, measured in a local

reference system, i.e. in a system with origin in the junction of

the plate with the web in the respective x coordinate. It is

worth emphasizing the important vertical displacements suf-

fered by the panel that surpass 60 cm in the middle section,

anyway, the lateral displacements of the stiffener indicate loss

of its stiffness and the behavior of the plate as a membrane. It

should be mentioned that the MTD occurs in some cases in the

middle of the panel and in other cases in the region near the

restraint.

The same Figure 10 also shows that the magnitude of the

imperfection slightly alters the final geometric configuration.

In this manner a slight imperfection in the stiffener originates

a distortion in the opposite direction to the initial imperfection,

however for average and severe imperfections the slope of the

final configuration accompanies the direction of the stiffener

initial imperfection.

11

22

12


Figure 8. Stress fields after the application of gravity and L.P.= 0.03 MPa (panel with average imperfection)

Figure 9. WB and FB Figure 9 in the region close to the restraint (L.P.= 0.03 MPa and average imperfections).

Plastic

hinge


Figura 10. Final configuration of the stiffener web in all cases analyzed

TABLE 1: ANALYSIS TIMES UNTIL PANEL FAILURE, MAXIMUM VERTICAL DISPLACEMENT (MVD) AND TRANSVERSAL DISPLACEMENT (MTD) VALUES FOR ALL THE

CASES EVALUATED.

Lateral

Pressure (MPa)

Level of imperfection in the stiffener

Slight Average Severe

Time Failure

(min.)

MVD

(mm)

MTD

(mm)

Time Failure

(min.)

MVD

(mm)

MTD

(mm)

Time Failure

(min.)

MVD

(mm)

MTD

(mm)

0.01 39.1 215.4 5.4 39.1 215.2 7.6 39.0 214.9 15.1

0.02 42.5 396.2 6.5 42.2 396.0 11.5 41.2 395.9 25.2

0.03 42.5 574.7 36.3 43.2 574.7 36.3 44.6 621.8 235.0

5 CONCLUSIONS

The numerical-computer methodology for analyzing the be-

havior of steel structures under fire conditions presented in

this paper was applied to evaluate the behavior of a stiffened

panel of the deck of an oil tanker submitted to a fire scenario.

Despite the idealized load conditions the thermo mechanical

behavior could be observed at different times of the postulated

fire, but it should be mentioned that the results obtained are

valid only for the load and boundary conditions taken into

consideration. In a real situation, when the structure suffers the

action of the waves and other loading conditions, a different

behavior can occur. Anyway, as shown by Manco et al., 2013

the influence of the outline on the longitudinal edge affects the

behavior of the stiffener facilitating its distortion. Due to the non-linearity of the distribution of the stresses pre-

sent in the panel during the fire, it is evident that such an anal-

ysis is very complex and cannot be addressed by simple ana-

lytical formulations and thus the use of a numerical method is

necessary. Anyway, we proved that the problem of the ana-

lyzed cases is governed by the temperature and not by the ge-

ometrical configuration or the load.


From the results we concluded that is necessary to apply ele-

ments of passive protection. It should be highlighted that con-

siderably severe fire conditions were forecasted by assigning

the standardized curve for hydrocarbon burning. More refined

studies can be applied in order to make the simulations of fire

scenarios more realistic, as for instance employing CFD

(Computational Fluid Dynamics) models and, consequently,

the mechanical behavior of the structure can be estimated in a

more reliable way.

The conclusions obtained through the numerical simulations

indicate that the methodology presented in this paper can be

applied to assess the structural behavior of offshore structures,

with different load conditions and different materials for ther-

mal protection and has a potential use in the reduction of

structural passive protection without impairing the levels fore-

casted for global safety. The layer of thermal protection, usu-

ally employed in this type of structure, delays the heating of

the protected element, helping to maintain the mechanical

properties for a longer time, improving the panel structural

behavior.

ACKNOWLEDGMENT

The authors wish to express their gratitude to the National

Petroleum Agency of Brazil (ANP), PETROBRAS and

COPPE-UFRJ for their support for the development of this

work.

REFERENCES

ABAQUS, Hibbitt, Karlsson e Sorensen, Version 6.11-3, 2011.

Cook, R.D., Malkus, D.S., Plesha, M.E. and Witt, R.J., Concepts and Applications of Finite Element Analysis, 4th Ed., John Wiley and Sons, New York, 2002.

Cullen, L., The Public Inquiry into the Piper Alpha Disaster, HM Stationery Office, 1990.

European Committee for Standardization. EUROCODE No. 1: Ac-

tions on Structures, Part 1-2: Actions on Structures exposed to

Fire, ENV 1991-1-2, British Standards Institution, London, UK,

2004.

European Committee for Standardization. Eurocode No. 3: Design of

steel structures, Part 1.2: Structural fire design, ENV 1993-1-2,

British Standards Institution, London, UK, 2004.

Hassanein, M.F. Finite element investigation of shear failure of lean

duplex stainless steel plates girders, Thin-Walled Structures

Journal, Vol. 49, pp. 964 - 973, 2011.

Heinisuo, A. & Aalto, A., 2008. Shear Buckling of Steel Plates at

elevated Temperatures, Tersrakenteiden tutkimus- ja kehit-

yspivt, Finland

ISSC. Report of Specialist Committee III.1 Ultimate Strength, Pro-

ceedings of the 18th International Ship and Offshore Structures

Congress (ISSC 2012), Edited by Wolfgang Fricke and Robert

Bronsart, Rostock, Germany, Vol.1, pp.285-363, 2012.

Landesmann, A., Mendes, J.R., Ellwanger, G. Numerical Model for

the Analysis of Offshore Structural Elements under Fire Condi-

tions, Proceedings of XXXIV Jornadas Sudamericanas de Inge-

niera Estructural, San Juan, Argentina, 2010.

Lewis R.W., Nithiarasu, P. and Seetharamu, K.N. Fundamentals of

the Finite Element Method for Heat and Fluid Flow, John Wiley

and Sons, England, 2004.

Manco, M.R., Vaz, M.A., Cyrino, J.C., Landesmann, A. Behavior of

stiffened panels exposed to fire, Proceedings of IV

MARSTRUCT, Espoo, Finland, pp. 101 - 108, 2013.

Skallerud, B. and Amdahl J. Nonlinear Analysis of Offshore Struc-

tures, Research Studies Press Ltd., Baldock, Herforshire, Eng-

land, 2002.

Smith, C.S., Anderson, N., Chapman, J.C., Davidson, P.C. and

Dowling, P.J. Strength of Stiffened Plating under Combined

Compression and Lateral Pressure, the Royal Institute of Naval

Architecture, Vol. 133, pp. 131 147, 1991.

Vimonsatit, V., Tan, K. H., Ting, S.K., Shear Strength of Plate Girder Web Panel at Elevated Temperatures, Journal of Con-structional Steel Research, 2005.

Yang, X.J., Gao R., 2004. Factors Affecting the behavior of Steel Structures in Fire, Proceedings of NASCC 2004, California, EE.UU, 2004.

analysis of oil tanker deck under hydrocarbon fire

Documents