USING CFD AND DYNAMIC SIMULATION TOOLS FOR THE DESIGN AND OPTIMIZATION

OF LNG PLANTS

Tania Simonetti 1, Dominique Gadelle 1, Rajeev Nanda2

1. LNG Department, Process Division, Technip France 2. LNG Department, Process Division, Technip Houston

Keywords: 1. CFD; 2. Dynamic simulation; 3. Hot air recirculation ; 4. Air tower design ; 5. Cooldown

procedure; 6. LNG 1 Introduction and objectives

In recent years, an increase in LNG plant design production capacity and a step out in technology has been observed; newly designed train capacity has risen to 6.3 MTA for OKLNG project and to 7.8 MTPA for Qatar individual trains as examples. Along with this development, equipment sizes have grown to exceed previous common experience while overall plant layouts have evolved towards more spread out or congested configurations due to the need of installing larger and larger trains.

This context amplifies a need for the best possible design tools, capable not only to investigate and prove the proper performance of critical pieces of equipment, but also to optimise capital investment in equipment, piping and layout without compromising the proper performance of the plant.

In parallel, enhanced computing capabilities have widened the domain of application of Computational Fluid Dynamics and Dynamic simulation, allowing these tools to occupy an increasingly important place in terms of verification and improvement of design. Nowadays, these simulato rs are capable not only to describe the performance of a single piece of equipment but also to give a complete picture of an installation and of its response to operation al upsets or procedures.

The purpose of this paper is to illustrate some recent applications of CFD and dynami c simulation, where these simulators have stepped out of their traditional roles and have been employed to validate layouts of specific areas or even whole LNG plants, or used as design tools for pieces of equipment and layout. The use of CFD and dynamic simulation in the applications discussed, in most cases ended up in significant economi c gains.

2 Computational Fluid Dynamics

CFD represents a powerful simulation tool that allows very accurate mechan ical and thermal modelling. CFD is based on numerical solution of equations for the conservation of mass, movemen t quantity, energy (refer to Appendix 1 for equations).

In general, a CFD simulation is built in two parts: - Geometric model definition via a Computer Aided Design tool - Mathemati cal solver based on Navier Stokes equation The simulated domain can be modelled in 2D or 3D: a domain is defined with its boundary conditions. The equations can be solved in steady or unsteady state. The results can be presented in graphic form allowing immediate visualisation and interpretation of hydraulic and thermal profile s. These principles can be best illustrated through case studies developed during some LNG plant projects, such Qatargas II (2 x7.8 MTPA LNG production), Qatargas III (2 x7.8 MTPA LNG production), Yemen LNG (2 x 3.45 MTPA LNG production), OKLNG .(2 x 6. 3 MTPA LNG production), Freeport Terminal LNG.

a. Case study: utilizing Air as Heat Source in Air These air base technologie s are very energy efficient, but a careful evaluation need s to be done

to quantify the advantages based on the specific site conditions. Some of the main considerations are (i) the lower air temperature during the cold months of the year may require a suppleme ntary heat source, that increase capital and operating costs, (ii) the cold air, due to negative buoyancy, may tend to recycle back. Any recirculation would result in reduction in heat transfer area and performance justifying rigorous Computation al Fluid Dynamics (CFD) modell ing, (iii) dealing with fog problem s, as air gets saturated due to its tempe rature reduction, (iv) handling of condensed moisture from the air and resulting water disposal issues, (v) demand s on the control system to compensate for variations in ambient conditions that requires dynami c process simul ations for analysis of the system.

In an air tower or reverse -acting cooling tower, the air exchanges heat with the flowing water by direct contact. The heat transfer mechanism in an air tower is the reverse of a cooling tower. The moisture in the air condenses as air gets cooler, and there is a net production of water in the process. The heat of

condensation makes a significant contribution to the total heat duty. The excess water is disposed of from the air tower sump.

Figure 1 show s how the air tower can be utilized for LNG vaporization. Although various schemes are po ssible to integrate the air tower, one of the typical scheme s is to utilize the shell and tube exchanger for LNG vaporization with an intermediate fluid such as ethylene or propylene glycol flowing in a closed loop circulation. In such a scheme, the circulating fluid circulates through the loop consisting of LNG vaporizers and intermediate exchangers which could be plate and frame type. When the air tower is not operating during the winter, the intermediate fluid is heated in a fired heater. In summer, when no heating is required from the fired heater, the intermediate fluid exchanges heat with water from the air tower. For flexibility, the system would be designed to have part of the heat from the air tower and part from the fired heater. It is important to note that the power consumption is significant in circulating the water by pump s for the system. There is a point of diminishing return to extract the heat from the air tower as winter approaches.

Figure 1: Utilizing Air Tower for LNG Vaporization

The air tower can be designed conceptually by extending the concept from a normal cooling tower with the following details to be addressed:

i. The fill material and type should be tested to confirm that the quantity is adequate. In the case of the air tower, water condensation takes place instead of evaporation as in the cooling tower. It is important that appropriate fill material and quantity are used.

ii. The air that comes out of the air tower is at low temperature and there is a tendency for air to settle down due to negative buoyancy. Computational Fluid Dynamic s (CFD) model l ing is required to confirm the amount of re-circulation and the impact on design. Due to recirculation of cold air, under some conditions the tower performance can deteriorate significantly.

iii. The wind speed and direction have significant impact on the tower performance. Again, the impact can be studied from CFD modelling. The location of the air tower based on the results of CFD modell ing is key to its successful performance and optimization of the design.

iv. The local ambie nt air tempe rature and fluctuation are also impo rtant cond ition s for understanding the duration of reduced performance. These conditions impact the design of the air tower. A backup vaporization system and its design should be also based on the same.

v. There would be a net generation of water in the air tower due to condensation. Thi s wate r quality is generally the same as rainfall, which should be drained off to a suitable location.

vi. The water that circulates in the air tower and the piping system is moderately corro sive. Special metallurgy or internal coating for equipment and piping is needed. Generally, water treatment by dosing chemicals will be very expensive as there is a net overflow of water out of the system resulting in a loss of expensive chemi cals. Moreove r this could also be a permitting issue.

Figure 2: Envelope Indicating Temperature Below Ambient

Figure 2 and 3 illustrate the envelope of low air temperature due to cold air recirculation. The

envelope shows air re-ci rculating back to the inlet of air tower. It is important that the amount of recirculation be computed as it would impact the design of the air tower. The temperature reduction at the inlet of the air tower can significantly reduce the tower performance.

It is important to note that the horizontal fan configuration will not perform well under low wind speed. This is illustrated by the CFD envelope shown below in Figure 3. The low wind speed results in the cold air settling near the tower intake area. The higher wind speed results in more turbulence, better mixing and less cold air recirculation.

Figure 3: Temperature Envelope for Weak Wind Speed

By CFD modell ing the impact of using vertical and horizontal fans in an air tower was studied. In the

final design for Freeport LNG Terminal the vertical fans were adopted after extensive study of local meteorolo gical data, plot plan and the site location. The overall control system was extensively studied and verified using process dynamic simulations. Also tests were conducted to measure and validate the heat and mass transfer coefficients for the actual fill material used in the air tower.

Figure 4: Comparison of Vertical and Horizontal Air Tower Design

The map s of velocity vector and surface temperature reveal the impact due to the presence of other equipment in the plot plan. The interference from other equipment on the air tower performance cannot be ignored.

Figure 5: Velocity Vector Around Air Tower

Figure 6: Surface Temperature around Air Tower

The analysis of the air tower system as illustrated above required the following aspects to be

evaluated in detail: i. Heat and Mass Transfer Mechanism: The heat exchange is in reverse direction

when compared with the standard equipment utilized for similar service. The correlations derived from the cooling tower design required verification through testing.

ii. Air Recirculation: The plot plan and the local meteorological conditions play an impo rtant role in the design of such systems.

iii. Location of Equipment: The location and orientation of the air tower on the plot plan were found to be a key factor to its performance. The effect of wind speed and direction, prevalent wind direction and interference with other equipment is significant.

iv. Temp erature at Site: Average ambient conditions can be misleading for detailed evaluation and design. Detailed evaluation of minimum and maximum temperatures and changes were found to be very important for the final design and optimization of the system. In some case s, the average temperature at best

may be used for initial snapshot studies at the onset of the project. v. Full Backup Vaporizer during the winter: At Freeport, as in many cases, full

backup vaporizers are required for operation during the colder months. A cost benefit analysis is required to justify the initial capital investment against fuel savings and NOx and CO emissions.

vi. Condensation of Water: Excess water would require collection and disposal. Special metallurgical requirements were evaluated, and resulted in improvements such as the internal coating of the water circulation pipe.

b. Case study: CFD application to slug catcher performance assessment In this case study, CFD simula tion has been used to assess gas distribution and incoming liquid

separation efficiency in a large finger type slug catcher consisting of 12 x 48” fingers. Gas distribution was successfully simulated using 3D segregated implicit solver. This gave velocity

(refer to figure 7 here below) and pressure profile throughout the slug catcher.

Figure 7: Slug catcher performance assessment via CFD: velocity profile

Liquid separation efficiency was modelled by injection of liquid droplets. Two droplet injection models were developed; the assumption s underlying each one are the following:

i. No shear: in this case all droplets agglomerate and form liquid film when they enter in contact with any wall in the slug catcher

ii. Shear: in this second ca se only the droplets that enter in contact with fingers wall solely, are trapped and agglomerate.

The range of droplet sizes used in the model varies from 1 to 400 µm. Based on these CFD simulations, the amount of stopped and escaped droplets from the slug

catcher could be computed. In either case, the efficiency of the slug catcher in terms of gravitational separation could be assessed by plotting the curves of percentage of trapped droplets against droplet diameter per each type of droplet injection model.

The actual efficiency of the slug catcher in terms of liquid separation versus droplet diameter is somewhere in between the no-shear assumption case plot and shear case assumption plot (see here below figure 8).

Figure 8: Slug catcher performance assessment via CFD: liquid separation efficiency

The geometry, boundary conditions and fluid zones drawn by using the Gambit software. The hydraulic behaviour simulated with Fluent software.

c. Case study: Hot air recirculation studies in LNG plant In LNG plant, LNG production capacity is directly linked to refrigeration power. This refrigeration

power is dependant upon ambient air temperature because of the influence on gas turbine available power when these are used as mechanical drives, and also because it determines the refrigerant condensing temperature when air is used as the cooling media. Consequently, ambient air temperature directly affects LNG production.

Hot air recirculation studies aim to evaluate actual air temperature at the inlet of both gas turbines and air coolers. Air temperature may in fact be higher than suggested by site meteorological records due to recirculation of hot air from sources such as air coolers plum e and exhaust stacks. The results of such a study are used then to validate the layout and plot plan of the installation.

In the case study described, CFD was used to evaluate hot air recirculation and to validate the layout of a large, two train West African facility. The CFD simulation model included two LNG trains, LNG and LPG tanks.

The model geom etry was built taking into account all large-scale obstacles such as compre ssor houses, driers, substations, technical rooms, etc and significant details in congested areas, e.g. cable trays, piperack, zones below main compressors (nozzles, pipes, auxiliaries). On the other hand, downwind units were considered to have a lesser impact on air circulation and were excluded from the model.

The model took into account atmospheric conditions, e.g. prevailing wind directions, ambien t temperature, wind velocity and turbulence profiles.

The sources of hot air for this application were the gas turbines and waste heat recovery unit exhaust gase s. Refer to figure 9 for views of the model and hot air sources.

Figure 9: CFD application to hot air recirculation study: two trains model for CFD simulation. The study led to the following results:

i. Air temperature rise observed at each air cooler and gas turbine inlet for all the selected wind conditions.

ii. Therma l amplification observation: gas turbines exhaust led to local temperature rise higher than 50C whilst air coolers caused l ocally tempe rature rise of 25C. This allowed identifying specific areas where the air temperature rise with respect to forecast ambient temperature may impact the design of equipment. Figures 10 and 11 offer a visual appreciation of the configuration and associated air temperature rise.

Figure 10: CFD application to hot air recirculation study: thermal amplification.

EXHAUST AT 480K

EXHAUST AT 827K AIR COOLERS INLET 300K

AIR COOLERS OUTLET 310 TO 337K

GAS TURBINES INLET

Figure 1 1: CFD application to hot air recirculation study: two trains resultin g heat plume.

Validation of the configuration via air recirculation stud y: The air temperature rise was determined for actual site conditions and the performance of affected

equipment was able to be checked. In conclusion, the air recirculation study allowed the efficiency of the LNG plant to be confirmed and to validate the selected plot plan.

d. CFD applied to other engineering studies CFD has proved to be a valuable tool for a number of other engineering studies such as vapour/

liquid separation, compressor suction line hydraulics, optimisation of compressor suction piping layout, optimisation of piping routing upstream of critical separators.

i. Case study: CFD application to vapour/ liquid disengagement in large LNG trains propane evaporators.

Propane evaporators are at the core of the LNG plant, and good plant performance requires the pressure drop to be minimised. A critical issue in these evaporators is the good separation of liquid droplets in the chillers. The objective of this study was to evaluate hydraulic behaviour with respect to pres sure drop and separation efficiency, good separation efficiency translating into homogeneous and optimal velocity across the evaporator mesh.

The evaporators studied were HP, MP, LP and LLP chillers of Feed Gas and Mixed Refrigerant in a large baseload project using Air Products C3/MR p roce ss.

Simulations were built so as to examine only the gaseous phase above High Liquid Level. The geome tries of the evaporators including nozzles, headers and wiremesh mist eliminators were fully described in the models.

The results of this case study indicate d that the operation of these chillers is satisfactory: i. Pressure drops were all within a percentage of operating pressure that is

acceptable for this type of equipment. ii. Mesh velocities were in a range that is judged acceptable for liquid separation.

Therefore, the study led to the confirmation that the operation of these chillers is satisfactory. In addition a number of recomm endations for the design of the evaporators were able to be made,

such as the preferred use of a header with elbows rather than a T to reduce pressure drop. Such results are especially valuable for large capacity equipment that represent a step out with

respect to experience and referenced equipment sizes.

ii. Case study: CFD application to verific ation of compressors suction line hydraulics Verification of line hydraulics with CFD answers a need for design and optimisation of critical lines,

with a tool that is versatile and user friendly. In LNG plants minimisation of pressure drop in compressor suction lines brings significant gains in terms of compressor power, which in turn allows an increase in available refrigeration duty and LNG production.

In view of this objective, CFD has been used on a number of recent projects to screen compressor suction lines thoroughly for pressure drop and velocity profile at the compressor flange and validate modifications.

In the case study six compre ssors suction lines were verified: Low pressure, Medium Pressure, High Pressure Propane and Low Pressure, Medium Pressure, High Pressure Mixed Refrigerant.

The CFD geometry took into account the line itself and detailed geometry of conical filters, venturis, butterfly and non-slam check valves, drums and associated internals (vane type distributors; me sh pad), kettles and outlet manifold.

The CFD output consisted of complete pressure drop and velocity profiles. These results have allowed the following:

i. Verification of compliance with compressor vendor requirem ents for pressure drop,

ii. Analysis of distribution at compressor nozzles and impleme ntation of modifications to layout and piping when necessary.

The most remarkable modifications that resulted from the study were as follows: i. LP MR suction line: CFD showed that pressure drop exceeded vendor

requirements. In addition poor distribution was observed at the compressor flange. CFD was then applied to different line configurations and sizes leading to a solution where the diameter was increased to 64” from 60”. In figure 12, the results for gas distribution across compre ssor nozzles can be compared for the two configurations, the former based on standard (good) engineering practice criteria applied to line sizing and the latter designed using CFD. In addition, as a final result the resulting pressure drop was reduced by 45% with respect to the ini tial configuration.

ii. MP MR suction reducer geometry was re-specified to improve the velocity distribution across the inlet nozzle. This allowed a reduction in pressure drop of 12%. Figure 13 shows the CFD output from the different tests that lead to the finally retained arrangement. It can be seen that CFD output allows a quite straightforward interpretation of results.

iii. The CFD study first showed pressure drop to be significant. From the velocity and pressure drop profile, it was possible to identify the most effective modification that consisted in increasing the distribution pipe diameter at the kettle manifold: a comparison of the two configurations and resulting pressure drops and velocity profile can be seen in figure 14.

Figure 1 2 CFD application to compressor suctio n line hydraulics verificati on: LP mixed refrigerant suction line.

Figure 1 3: CFD application for compressors suction line hydraulics verification: MP Mixed refrigerant suction line.

Figure 14: CFD application for compressors suction line hydraulics verifica tion: modifications in LP

Propane suction system. This study shows firstly how CFD simulation can accurately picture the pressure and velocity profiles

of a given system, allowing modelling of fittings such as control valves, strainers, check valves and internals. Pressure drop calculations are far more accurate than with standard engineering tools.

Furthermore, CFD allowed several tests to be performed in order to find the optimum solution. The improvemen t in flow pattern resulting from each modification could be easily visually appreciated from the output profiles.

iii. Case study: application of CFD for layout design for refrigerant compressors suction lines

CFD can be used to optimise compressor layout for minimum investment cost while respecting pressure drop constraints.

On one project a CFD study of the Low Pressure Propane Compressor suction line was carried out to find the minimum compressor table height that still met the specified maximum pressure drop thus achieving substantial economic savings. In fact, through CFD it was possible to identify an improved arrangement for suction line fittings (elbows, strainers) to meet the allowable pressure drop and with an acceptable flow pattern. Figure 15 shows the effect on velocity profile at the compressor inlet nozzle for a difference of one metre in compressor table height. Figure 16 shows the nozzle velocity profile with the final geometry. In this case study CFD proved to be an effective layout design tool, allowing a clear basis for disc arding a costly configuration.

Figure 1 5: Impact of raising compressor suction table height on LP Propane case study: CFD output.

Figure 1 6: Impact of revised geometry after CFD study for LP Propane case study.

iv. Other case studies: CFD application to verification of critical separators performances

CFD allows the proper operation of separator internals to be verified. Such performance verifications are usually carried out with the purpose to ensure the correct operation of the separator, nonetheless CFD can also be used to relax piping routing criteria upstream critical drums internals with a view to reducing straight lengths and a more compact installation.

For one LNG project, CFD simulations of critical separators such as the LP MR suction drum and LP C3 suction drum led to the relaxing of straight length requirements between the last elbow and the drum inlet, thus simplifying the piping routing. Figure 17 and figure 18 show the results of these CFD simulatio ns: it can be easily seen that the velocity profile at the mesh inlet is rather homogenous, indicating good distribution across the section and efficient droplet separation despite the reduced straight lengths upstream the drums.

Figure 1 7: LP MR suction drum separator performance: velocity distribution at mesh inlet.

Figure 1 8: LP Propane suction drum separator performance: velocity distribution at mesh inlet.

CFD applied to separators may alternatively lead to separator size reduction, bringing substantial investment savings. As an example, separator volume reduction after CFD verification is summarised in the Table 1: data are taken from an optimisation study aimed at reducing drum investment cost while maintaining gas liquid separation performance. In this case study all the CFD study drums are fitted with multi-vane inlet device and wiremesh mist eliminators. Table 1: reductions of critical drums

Drum number Weight of initial Weight after Weight Material

design (Tons) optimization (Tons) Reduction (Tons) 106 -V - 101 206.7 171.7 35 304L SS 106 - V - 102 148.5 120.2 28.3 304L SS 106 - V - 103 172.6 79.4 93.2 LTCS 106 - V - 104 66.6 60.1 6.5 LTCS 106 - V - 105 63.9 57.6 6.3 LTCS 106 - V - 106 58.5 52.3 6.2 LTCS 106 - V - 107 53.8 48.7 5.1 LTCS

The total weight reduction for the stainless steel drum was 63 tons on each LNG train .

3 Dynamic simulation Dynamic simulation allows modelling of transient behaviour in processes bringi ng new information

useful for system design that cannot be represented with static simulations. Dynamic simulation can address many aspects of process plant design.

In the case studies presented, the problems successfull y studied include cooldown procedure for a LNG pipe network, and a typical case of compressor dynamic simulation .

a. Case study: Cooldown dynamic study of a LNG pipe network. The interest of optimisation of cooldown procedure in LNG rundown/loading systems lies in the gain

over LNG and cold gas flows used and in duration shortening for such complex operation.

The case study for this application was the CLP storage rundo wn and loading lines belonging to LNG production facilities different than those providing cold gas and LNG for the first cooldown. It was then of utmost importance to limit the duration and the flowrate of cold gas and LNG taken from outside production entity. In this context, dynamic simulation has proven a highly efficient tool to tailor up the cooldown procedure in the perspective of minimising the use of cold gas and LNG and duration.

In the study for CLP storage cooldown, the network is composed of a rundown system of 4km of 22” pipe, a cross over line 500m of 10” pipe and a loading loop with 16km 36” pipe. The cold gas was brought through a 3000 m long pipe of 6” size coming from existing external production facilities. The limit of the system is imposed by the allowable back pressure at cold gas injection point. Pressure at the other end of the network is set. Figure 1 9 gives a schematic view of the network configuration. It is apparent that the 6” line segment represented the controlling section with respect to allowable pressure drop and allowable velocity during the transient for the cold gas.

Figure 1 9: Cooldown dynamic study of LNG pipe network: simplified view of the studied network.

The selected cooldown procedure consists of:

i. Step 1: rundown system cooldown with 20t/h of cold gas during 24 hours. Maximum allowable back pressure is respected while cooling down the system to -90°C. Loading loop starts to cool down.

ii. Transition: cold gas flowrate reduced to 0t/h between t=24h and t=25h to allow LNG injection: back pressure is decreased to 1.1 bara and LNG available at 6.7 bara.

iii. Step 2: once cold gas flowrate is reduced to 0t/h, LNG flowrate is increased up to 28t/h. The rundown system quickly cools down, and loading li nes progressively finishes their cooldown in parallel. When LNG arrives liquid at berth 5 (refer to picture 1 9), parallel circulation must be stopped and loop circulation is required. At that time, fluid is about -150°C and pipes wall temperature are aroun d -80°C. Jetty drum can be partially filled (up to 10-20%) so that pipes wall temperatures decreases below -100°C.

The overall profile of reached temperature versus time obtained via dynamic simulation is shown in figures 20 and 2 1 for cooldown of rundown lines and cooldown of loading lines respectivel y.

Figure 20: Cooldown dynamic study of LNG pipe network: overall cooldown of rundown lines chart, achieved wall temperature versus elapsed time.

Figure 2 1: Cooldown dynamic study of LNG pipe network: ov erall cooldown of loading lines chart, achieved wall temperature versus elapsed time.

The use of dynamic simulation allows several tests to be run on a built model, in this specific context this made evident that it is not required to cool down the loading loop with cold gas before letting in LNG injection. By this method, cooldown duration and involved flowrates of cold gas could be optimised.

Conclusivel y, dynamic simulation applied to network has proven to be an efficient and flexible means of validatin g and customising an operating complex procedure.

Figure 22 Cooldown dynamic study of LNG pipe network: overview of CLP cooldown network.

b. Case study: Dynamic study of flash gas compressor in LNG plant In this case study, a dynamic simulation applied to this flash gas compressor was performed on the

end flash gas compressor provided downstream the liquefaction facilities in order to confirm the transient phase operati on such as the start-up sequence of the compressor and the compressor behaviour during trips. The result s of such simulation permitted to precisely identify the suitability of selected material such as antisurge valve s size, the need for additional valves, e.g. hot gas bypass, in case of surge during transient, verification of the proper cooldown rate procedure during start-up in order to protect machine at all operating conditions. Therefore the outcome of the dynamic simulation has a direct impact over installation and provided instrumentation.

The studied flash gas compressor was a three stage fixed speed machine, feeding the HP fuel gas network. Flash gas compressor compress LNG flash gas from 0.24barg up to 28.5barg. First stage is provided with inlet guide vane valves and suction drum. Each stage is provided with air cooled aftercoolers to reduce interstage temperature to 45 °C.

The dynamic simulation has been carried out with Hysis dynamics; the model has been filled in with vendor compressor curves depicti ng polytropic height versus suction flow. The pressure drop in pipe s and air cool ers is rated based on design conditions. The valves included in dynamic simul ation are filled in with the installed Cv, including anti -surge valve data from compressor vendor.

i. Dynamic simulation: start -up sequence development The flash gas start -up sequence has been defined as follows: i. Initial conditions correspond to the following configuration: outlet shutdown valve is closed;

inlet suction control valve is fully open; inlet guide vanes are closed at 70°; antisurge valve is in manual mode and fully open position; pressure control valve to flare is closed; compressor loop is pressurised at suction pressure, i.e. 0.24 barg.

ii. Step 1 - motor start -up: in this phase dynamic simulation showed that required load torque curve providing compressor acceleration stay behind the available torque curve at 70% and 100% available voltage. Therefore no problem is encountered to start up the compressor.

iii. Step 2 - compressor cooldown: the compressor has reached nominal speed. A start-up pressure control val ve is provided to flare gas from discharge, thus allowing to cooldown the system and place the compressor on line. During this step this valve to flare is opened in manual mode while cold gas is allo wed in compressor.

iv. Step 3 – closure of antisurge valve and opening of antisurge bypass valve: the antisurge controller is switched to automatic mode, therefore antisurge valve closes down. The relevant closure time is defined from vendor data and specification. Antisurge bypass valves open so as to compensate the closure of antisurge valves. The antisurge bypass valves are used to regulate recirculation so as to slowdown or increase the cooldown rate. The cooldown rate is set in accordance with machine vendor requirement.

v. Step 4 – closure of antisurge bypass valves: after cooldown the antisurge bypass valves are set back to automatic mode and close down. No impact on compressor operating parameters is observed. In this phase the cooldown of the compressor is continued

vi. Step 5 – At the last phase of cooldown, the antisurge bypass valve s are closed and compressor suction temperature reaches -63°C.

vii. Step 6 – Comp ressor on line with process: the IGVs are opened manually from Control Room in order to increase the discharge pressure of the compressor. The IGVs are gradually open so as to reach the required pressure to discharge into high pressure fuel gas network. When this is achieved, the pressure control valve to flare is set to automati c mode.

The dynamic simulation permitted then to ensure the following: • Compressor motor torque is suffici ent for compressor to reach nominal speed with suction

valve open. • Compressor can be adequately cooled down and put in line via provided pressure control

valve to flare .

ii. Dynamic simulation: compressor behaviour upon trip In the dynamic simulator, it is p ossible to carry out a trip scenario once the start-up scenario is

stabilized. The assumption behind a trip dynamic simulation was the blocked outlet case; this assumption was selected because for the actual design and configuration represented the worst case scenario.

In picture 22, one can see the evolution of process conditions during transient.

Figure 2 3: Dynamic simulation: flash gas compressor trip under blocked outlet, exampl e of process parameters variation.

It is observed that upon blocked out let, as soon as the discharge pressure rises up to 30 bara, the compressor is shutdown by the process safety logic, then at this point the antisurge valves are fully open ed (2 seconds time) and inlet suction valve (10 seconds time) is closed. The study showed that each stage compressor is subject to surge in case of high speeds prior to motor complete stop.

Based on compressor vendor data, the mentioned surge phenomenon under trip conditions may bring the machine to reverse rotation in case the settling pressure is not reached before total stop of the motor. If this takes place, the remedy to protect the machine is the installation of a hot gas bypass valve. However, by analysing this specific dynamic simulatio n data, it was observed that settle out pressure is reached in 21 seconds after trip, when the motor was still at 22% of nominal speed; therefore the surge under trip conditions does not pose a threat in view of lifetime of the compressor according to manufacturer.

Conclusivel y, the dynamic simulation not only permitted to observe real behaviour of the machine upon process upsets and to identify the concerns to machine integrity during transient phases (surge), but also allowed to take the most suitable countermeasures to ensure the safety of equipment.

4 Conclusions and Way forward

From the above case studies, one can observe that CFD and dynamic simulation offer the following advantages:

• Accurate depiction of flow patterns and response of the systems that are not available

throughout static simulati on • Precise identifications of deviations from process requirements in performance, either in one

single equipment or with respect to overall LNG plant layout • Quantification of such deviations • Several alternates feasible for a given design problem thus allowing to find the optimum

solutions from economic and performance point of view • Easy visual interpretation of flow pattern and consequence of design modifications • Realistic simulation of the dynamic response of a given system, being a unit or the overall

plant, thus leading to tailor procedures (cooldown, but also start –up and turn- down are some of the potential examples) so as to minimise flaring and improve operability of the plant

All the above findings permit to integrate the CFD and dynamic simulation as design tools in projects.

This leads to the following:

• Confirmation of plant efficiency and LNG guarantee production • Confirmation of layout

• Confirmation of plot plan design • Achievement of CAPEX savings in equipment sizes • Achievement of CAPEX savin gs in piping length • Increase in compactness of installation • Reduction in weight of equipment • Reduction of flaring due to tailored operating procedures

All the above benefits are even more attractive in the near future development of LNG plant ,

expected to be on moving supports, where gains in compactness, weight, piping lengths and layout design are even more crucial issues to designer and investor.

Acknowledgements Matthieu Chambert, Technip Process Division Phil Hagyard, Technip LNG product line Jocelyne Launois, Technip Process Division Julien Metayer, Technip Process Division Henri Paradowski, Technip Process Division