READY FOR FLOATING LNG QUALIFICATION OF SPIRAL WOUND HEAT EXCHANGERS

Transcription

1 READY FOR FLOATING LNG QUALIFICATION OF SPIRAL WOUND HEAT EXCHANGERS Thomas Lex R&D Engineer, Dr.Ing. Klaus Ohlig R&D Manager, Dipl.Ing. Linde Engineering Pullach, Germany Arne Olav Fredheim Sr. Research Adviser, Dr.Ing. Carl Birger Jenssen R&D Engineer, Dr.Ing. Statoil Trondheim, Norway ABSTRACT A floating LNG project has several technical challenges. One is the influence of marine tilt and motion on the thermal and hydraulic performance of the Spiral-Wound LNG Heat Exchanger (SWHE). To learn more about the performance of the SWHE at different tilt angles and movements, two test facilities have been erected in a laboratory facility, one small scale test plant to study adiabatic fluid flow behaviour and one semiindustrial sized test facility to study thermal and hydraulic performance. The design, construction and experimental test program were conducted within the Statoil-Linde LNG Technology Alliance. For the test program, the test heat exchanger has been positioned on a hydraulic, movable platform to represent accelerations at rough sea conditions. The tests were performed at a wide range of operational conditions. Based on the experimental results a CFD model for calculation of the shell-side liquid distribution has been developed. From this project, data describing the influence of permanent tilt and one-dimensional oscillation on thermal and hydraulic behaviour has given sufficient information and knowledge to design the SWHE for FLNG. A long term test on the mechanical behaviour has also been completed to verify the mechanical design procedures. PO-27.1

2 INTRODUCTION All base-load natural gas liquefaction plants today are located onshore. Different studies on production of LNG on offshore, floating installations have been performed during the recent decades. To date no floating LNG plant has been built. To be able to install an LNG plant on a floating installation several challenges need to be addressed and technical solutions developed. The performance of the Spiral Wound LNG Heat Exchanger during permanent tilt and movements has been one subject of uncertainty. SWHEs are very large units and a complete heat exchanger can be up to 50 meters high having a diameter of 5 meters and a weight of 250 tons. The performance of the SWHE is vital for the operation of an LNG plant. The ability to develop good designs of such units for floating production plants is very important. Currently there is a lack of experience on how tilt and movement will affect the hydraulic and thermal behaviour. Previously, tests on liquid distribution for a model SWHE have been performed at Heriot-Watt University [1][2]. These tests were performed with water and aqueous solutions as test fluids. In the Heriot-Watt project, the liquid distribution was measured but it was not possible to derive the effect on performance of the SWHE. By contrast, model hydrocarbon test fluids at ambient temperature or above, with physical properties close to real fluids at low temperature, were used in the test program described below. The main driver for executing this qualification program is the Statoil position as potential operator for block 218 in the NnwaDoro project offshore Nigeria and Linde as a possible technology provider for the planned FLNG plant. The project was established in order to investigate the thermal and hydraulic behaviour of the SWHE during permanent tilt and dynamic oscillations. The project was organised within the framework of the Statoil-Linde LNG Technology Alliance and was executed jointly by the Research Departments of Statoil and Linde Engineering. The project was divided into different activities: Fluid Flow Test Heat Transfer Test Numerical Modelling Mechanical Testing Two test facilities have been erected in a laboratory facility, one small scale test plant to study adiabatic fluid flow behaviour and one semi-industrial sized test facility to study thermal and hydraulic performance and mechanical behaviour under tilting and moving conditions. The semi-industrial test facility consists of a complete SWHE bundle, where hydrocarbon mixture is used as test fluid. The hydraulic and thermal tests are completed and the mechanical tests will finalize the qualification program by mid The fluid flow test served as a basis for developing a hydraulic CFD model to describe the flow effects of permanent tilt. The first version of the CFD model is completed and has been tested and compared with the results from the heat transfer test. PO-27.2

3 LABORATORY SCALE FLUID FLOW TEST Test Facility In these tests the liquid distribution at varied operational conditions has been investigated for test sections with straight inclined tubes. The tube diameter, tube spacing and winding angle are in the same range as for industrial SWHE. Tests have been done both for single-phase liquid and two-phase flow. Test fluids used were iso-hexane for the liquid phase and nitrogen saturated with iso-hexane as gas phase. The physical properties for the gas and liquid phase are similar to industrial process fluids, the main exception being the density of the gas phase. A flow diagram of the test facility is shown in Figure 1. Figure 1: Flow Diagram of Test Facility The liquid phase is pumped from the liquid reservoir to the level controlled overhead container. From the container the liquid is fed to a liquid distributor system at the top of the test section. Below the test section, the liquid is directed to the liquid collection system by flow directors. The gas phase is circulated by the compressor(s). The gas outlet is on the side of the collection cells some distance above the liquid outlet. Liquid droplets in the gas are separated out and sent to the iso-hexane tank before nitrogen is recirculated back to the compressors or to atmosphere. The test section can be inclined along the tube direction, across the tube direction or combination of both (see Figure 2. The maximum tilt in both directions is 8 deg and the minimum tilt is 0 deg. The liquid is collected in cells below the test section. The level in each cell is measured and plotted to obtain a picture of the flow distribution. A flow director is installed to cover the distance between bundle and collection cells without affecting flow distribution. The cells are arranged on boards with 37 cells in each board (see Figure 2). PO-27.3

4 Measurements Figure 2: Test Section and Liquid Collection System An example of the distribution of collected liquid for 100% liquid flow at 0 deg and 8 deg tilt along the tube direction is shown in the following figures. 0 deg tilt along tubes / 0 deg tilt across tubes 10 deg tube inclination, 100% liquid % of average collected flow Board no Cell no Figure 3: Liquid Distribution at 0 Tilt of Test Section PO-27.4

5 8 deg tilt along tubes / 0 deg tilt across tubes 10 deg tube inclination, 100% liquid % of average collected flow Board no Cell no Figure 4: Liquid Distribution at 8 Tilt of Test Section The difference between 0 deg tilt and 8 deg tilt is clearly visible. At zero deg tilt, an almost uniform liquid distribution in the middle section is observed. Since liquid collects at the walls, distribution certainly experienced a peak at these locations For the case of 8 deg tilt the liquid collects more to the tilted side and the influence of the spacers (centred at cell no 11 and 26) becomes more visible. The areas with the lowest amount of liquid are below the the spacers. Tests were performed at different sets of tube inclinations and spacing between tube layers. The effect of geometry on the liquid distribution at 10 wt% gas fraction is shown in Figure % gas 0,8 Fraction at left side 0,75 0,7 0,65 0,6 0,55 10 deg, thin spacer 10 deg, thick spacer 15 deg, thin spacer 15 deg, thick spacer 0,5 0,45 0, Tilt along (deg) Figure 5: Effect of Geometry on Liquid Distribution PO-27.5

6 It can be observed that the amount of liquid collected at the left half of the test section (the test section is tilted to the left) increases at increased tilt angle, however, the influence is higher for tube inclination of 10 than for 15. The measurements of liquid distribution showed the following overall trends; The liquid maldistribution tended to increase with increasing tilt angle Tilt angles normal to the tube direction showed a larger effect on flow distribution than tilt angles perpendicular to the tube direction A winding angle of 15 indicated less liquid maldistribution than a winding angle of 10. This tendency was most pronounced at higher tilt angles. The effect of the presence of spacer bars tended to have most influence at higher tilt angles. The effect of space bar thickness was small. Increased gas fraction tended to reduce the liquid maldistribution, except for the highest gas fraction at 15 winding angle. A general observation is a significant increase of liquid maldistribution if the tilt angle becomes close to the winding angle. Consequently, the design of industrial SWHE for offshore LNG plants needs to consider expected tilt angles. SEMI-INDUSTRIAL SCALE SWHE TEST Spiral Wound Heat Exchangers are used for the liquefaction and sub-cooling of the natural gas in LNG processes. In each application the natural gas and other media (LNG, refrigerant to be cooled) pass through the tube bundle in an upward direction releasing heat to the refrigerant, which evaporates in the shell side of the heat exchanger. The refrigerant enters the top of the SWHE in its liquid state, or with a small fraction of vapour, and leaves the heat exchanger as superheated vapour at the bottom. Depending on the process, the refrigerant as well as the natural gas have different thermodynamic states covering temperature ranges from ambient down to -160 ºC For the semi-industrial scale test, the realistic conditions had to be simulated in a reasonable experiment. But not all of the full-scale conditions can be included in the test. Particularly, the overall size of the heat exchanger and the operating temperature and pressure as well as the mass flow rate of the tube side and shell side fluid had to be scaled in order to reamain within the lab facility and cost constraints. A basic process consisting of two loops and an experimental scaled SWHE was therefore developed (see Figure 8). Process Conditions. The operating shell side pressure of the SWHE was set slightly above higher than ambient conditions. The shell side operating temperature depends on the selected test fluid whose boiling range should preferably be in the range of 40 C to 80. To adjust the mass flow rates of the experiment, the momentum flux similarity was used. Appropriate model fluids representing the refrigerant and the tube side stream had to be selected in order to meet relevant operating conditions. Compared to the shell side heat transfer, the tube side heat transfer was assumed to be less sensitive to marine tilt and motion and thus negligible for a detailed experimental investigation. For the experiment, attention was focused on the heat transfer phenomena occurring in the shell side of the SWHE. Therefore the tube side fluid of the experiment simply had to act as a PO-27.6

7 heat source. Liquid water was selected as the tube side fluid for the tests. Unlike in industrial size SWHEs, a change of phase does not occur. Geometry. The refrigerant originally enters the SWHE as a liquid, or liquid with a few percent of vapour, and gradually changes its thermodynamic state to a superheated vapour during the process of evaporation. However, in a lab-scale experiment the complete process of evaporation cannot be simulated reasonably at designated mass flow conditions with the momentum flux similarity. Therefore the experimental SWHE has to cover different sections of the full-scale heat exchanger which are labelled for the experiment as Upper Section, Middle Section, and Lower Section. The section to be simulated is defined according to the vapour fractions at the inlet and outlet of the model heat exchanger. The model SWHE is scaled in terms of height as well as diameter and tube layers, respectively. Compared to a full scale heat exchanger, the height ratio is approx and the diameter ratio is approx However, the tubes' diameter and spacer thickness are identical to a full scale heat exchanger. Test Fluid Selection. The fluid representing the refrigerant was chosen in such a way that at experimental conditions its thermo physical properties are similar to those of the original refrigerant at industrial operating conditions. In particular, the liquid and vapour density, dynamic viscosity, and the surface tension of the shell side model fluid were chosen as the vital properties to be similar to those of the industrial refrigerant. To meet these criteria, a two component fluid consisting of n-pentane and isooctane (2,2,4- trimethylpentane) was selected as test fluid. This fluid has a boiling range from 40 C- 60 C. The vital thermo physical properties of this mixture at test conditions correspond approximately to those of the industrial refrigerants at operating conditions as shown in Figure 6 for the liquid viscosity and density. The liquid dynamic viscosity and surface tension play an important role for the wetting of the heat exchanger surface and, consequently, for the film thickness around the tube bundle. Since the film thickness essentially impacts the shell side heat transfer, these properties were chosen as a measure for the comparability. The similarity of the liquid and vapour densities is important for comparable mass flow and momentum flux conditions in the experiment, respectively Liq. Viscosity [mpa s] Test Fluid Refrigerant 2 Refrigerant 3 Liquid Density [kg/m³] Test Fluid Refrigerant 2 Refrigerant Temperature [ C] Temperature [ C] Figure 6: Liquid Viscosity and Liquid Density of Test Fluid PO-27.7

8 Tilt Angle and Oscillation Frequency. In order to simulate a stationary tilt of the heat exchanger due to wind forces or continuous movement due to rough sea conditions, it had to be installed on a platform which is able to tilt and move the SWHE. Based on previous studies at Statoil regarding tilt angles and oscillation periods of a floating barge under rough North Sea condition, a special tilting mechanism has been installed which is able to either tilt the heat exchanger to a defined angle or to realize continuous movements with varying amplitudes and frequencies. The worst case accelerations at the highest level above the deck, corresponding to the top part of an industrial SWHE, were used to set the oscillation period giving the same accelerations for the model SWHE. Process and Instrumentation Figure 7: Tilted Spiral Wound Heat Exchanger Basic Process. The process applied in the experiment consists of two closed loops representing the natural gas and the refrigerant cycle (Figure 8). The loop representing the natural gas is hereafter referred to as Loop 2 or Water Loop and the loop representing the refrigerant is referred to as Loop 1 or Model (Test) Fluid Loop. The fluid of Loop 2 enters the experimental SWHE at the bottom and flows upwards in the tube side of the heat exchanger, releases heat and leaves the device at the top. The fluid is circulated by pump P210. The heat added to Loop 2 is provided by heater E210. The fluid remains in its liquid state during the whole cycle. The test fluid of Loop 1, in contrast, passes through different thermodynamic states (liquid and vapour) since the different sections of the full-scale heat exchangers and thus different stages of evaporating refrigerant have to be simulated. The test fluid is pumped by pump P110 to heater E120A/B, where it is heated up to a defined temperature. Up to the expansion valve, the fluid remains in its liquid state. The temperature and pressure before the valve and the defined pressure drop across the expansion valve set the thermodynamic state and vapour fraction of the test fluid entering SWHE E100 at the top. PO-27.8

9 liquid expansion valve TI PI TI PI Heater E120A/B Spiral Wound Heat Exchanger E100 liquid/gaseous TI PI liquid LNG Loop (Model fluid 2) Circulation Pump P110 Circulation Pump P210 liquid Heater E210 TI PI Condenser E110 liquid Refrigeration Loop (Model fluid 1) TI PI liquid/gaseous Figure 8: Sketch of Basic Process Depending on the section in the real heat exchanger (upper, middle, or lower) to be simulated, the vapour fraction at the inlet is either increased or decreased by means of the temperature and the pressure drop across the expansion valve. The test fluid flows downwards through the shell side of the heat exchanger and absorbs heat from the water of Loop 2. When leaving the SWHE at the bottom, the test fluid is passed to cooler E110 where the vapour is condensed completely. After that, the liquid model fluid is pumped back to the heater. Instrumentation. Besides the regular process instrumentation, several resistance thermometers and pressure transmitters are installed to closely observe the flow and evaporation behaviour inside the Spiral Wound Heat Exchanger. Forty thermometers located at different levels and different circumferential locations at the bundle's shell side give clear information about the evaporating test fluid. Ten thermometers at different levels and layers are installed on the bundle's tube side. Measurements Generally, nine measurements were taken at each adjusted process condition, six stationary tilt angles and three continuous movements with different oscillation speeds. The process conditions were defined by means of the sections (upper, middle, lower) to be simulated. These sections differed in terms of shell side inlet and outlet vapour fractions. At Upper Section conditions the model fluid entered the SWHE as liquid and left it with a low vapour fraction. At Lower Section conditions the model fluid left the heat exchanger superheated after having entered with a high vapour fraction. The Middle Section condition covered the process conditions in between. PO-27.9

10 55.2 Shell Side Temperature Tilting Angle 10 Temperature [ C] Vertical Tilt Angle [ ] : : : : :00.0 Duration [min] Figure 9: Shell Side Temperature and Tilt Angle at Continuous Movement Figure 9 shows an example of the shell side temperature distribution at a specific location together with the tilt angle at Lower Section conditions. This data was taken at moderate oscillation speed. The influence of the movement, indicated by the tilt angle plot showing the vertical deflection of the heat exchanger, on the temperature distribution can be seen clearly. Furthermore this indicates a local maldistribution where tubes are partly wet or dry depending on the movement. MATHEMATICAL MODELLING A mathematical model for predicting flow on the shell side of a spiral wound heat exchanger using a CFD code has been developed. One of the purposes of the model is to predict the effect of tilt on an SWHE. The model is defined in the framework of multiphase flow. In addition to vapor and droplets, additional phases are defined for falling film over the tubes, one phase for each winding angle or winding direction. The detailed flow around individual tubes is not modeled, but the effect of the tubes is accounted for by momentum source terms calculated using friction factors. The friction factors are computed from the governing equations for a falling film over non-horizontal tubes. These governing equations are an extension of the governing equations for horizontal tubes as described in [2]. Additional correlations are also used for entrainment of liquid in the vapor flow and for the effect of spacer bars used to hold the tubes in place. As an example, the correlation relating the film velocity normal to the tubes to the film thickness is provided. The following correlation for uniform flow with no or negligible gas shear is assumed: 2 U U f i, n ρ + f v, nμl = ρ g 2 l D δ cosα PO-27.10

11 Here, D is the diameter of the tubes, δ is the average film thickness, and U is the average velocity in the direction normal to the tubes andα is the angle of the tubes relative to the horizontal plane. The dimensionless parameters f v, n and f i, n are the friction factors. A similar correlation is used for the flow velocity along the tubes. Finally the correlations are rewritten in volumetric form and included in a CFD code as source terms. Thus the model relies on a number of correlations, based on either in-house work or on methods and data found in open literature. In either case, the correlations contain parameters that have been tuned against the laboratory scale fluid flow test. Heat transfer is currently not included in the model, but ongoing work is directed toward this goal. Further details of the model can be found in [3]. A comparison of some calculated and measured values from the laboratory scale fluid flow test is shown in Figure 10. Here the net transversal flow through the mid-plane of the test section has been plotted for experiments with various gas fractions. For a zero tilt angle this value is zero, as the liquid flows equally fast to the left and right. As the test section is tilted however, more liquid flows in the tilt direction. As can be seen, a fairly good agreement between calculations and measurements has been obtained, and the trend towards less transversal flow for increasing gas fraction has been reproduced. Normalized net transversal flow fx/fy Tilt angle (degrees) 0 % gas, exp 0 % gas, model 22 % gas, exp 22 % gas, model 50 % gas, exp 50 % gas, model Figure 10: Calculated Net Transversal Flow through the Test Section Mid-Plane Compared with Experiments Calculated film thickness in the semi-industrial scale SWHE for two tilt angles are shown in Figure 11. The film thickness shown is the average over the two winding directions. The inlet conditions were set to represent a typical mid section in a full scale SWHE. As stated above, heat transfer was not modeled but the evaporation rate from the film was set to a constant value throughout the bundle. PO-27.11

12 Film thickness (mm) Tilt = 0 Tilt = 6.3 Figure 11: Calculated Film Thickness in the Semi-Industrial Scale SWHE for Typical Mid Section Inlet Conditions MECHANICAL TEST Besides the qualification of the SWHE in terms of process engineering a heat exchanger planned for off-shore service also has to be mechanically qualified. Based on the equipment used for the semi-industrial scale SWHE test, a mechanical qualification procedure was conducted. A theoretical FE analysis preceded the experimental investigations in order to predict the loads occurring under continuous movement. The experimental results were used to verify and modify the FE modelling approach. By this approach a sustainable modelling strategy for full scale industrial floating SWHEs is established. Furthermore, during the experiments the heat exchanger was exposed to a total of 100,000 load cycles. This provided the opportunity for an in-depth investigation of the wear in the bundle due to continuous movements once the heat exchanger was dismantled. Besides the already mentioned issues of interest the experiments were aimed at looking into the following aspects: Validation of bearing concept: The same bearing concept used on the existing semiindustrial scale SWHE will be applied to a full scale floating application. This concept provides for four or more brackets in circumferential direction, which are in the same position along the longitudinal axis. They are the only points where the SWHE is attached to the structural steelwork. This concept is different to the traditional skirt. The PO-27.12

13 advantage of this concept is that the bundle's and mandrel's weight forces are directly introduced to the supporting brackets. The shell has to carry the induced loads only over a short distance. This concept has already been applied to Linde's industrial scale SWHE in Tuha. For mechanical purposes, this concept seems to be the best version of the bearing concept for a floating SWHE. The applicability of this concept to a floating full scale heat exchanger has already been proven by FE analysis. Vibration Coupling: Both, the experimental and the industrial heat exchanger consist of three spring-mass-systems, the structural steelwork, the vessel and the bundle including the mandrel. It is not clear whether these systems are coupled to each other or not. If they are coupled the next question to be answered is whether they oscillate inphase or out-of-phase. In the former case, the three systems will be treated as one which substantially eases the modelling approach. In the latter case, reaction forces might be doubled under unfavourable conditions and might lead to high loads. The complexity and detail of the FE model has to be increased in order to cover these aspects. In this context any information about the eigenfrequencies are of great value. Impact of tilting on wear in the bundle: Under tilting conditions the parts of the bundle might move relative to each other. This might wear the inner parts of the heat exchanger. Dismantling of the heat exchanger after the measurement program and an examination of the inner parts clearly gave information about the tribological strains like the wear rate and areas of wear under tilting conditions. Applying new design concepts and diffenrent materials could then prevent wear phenomena. Level of detail in modelling: The design process of the industrial FLNG SWHE is inconceivable without an accompanying FE Analysis. High standards are applied to the FE model. It shall be able to predict the occurring loads and peak stresses as accurately as possible in order to optimise the fundamental design, the distribution of forces, the vibration behaviour and the material selection. However, at the beginning of a detailed modelling approach it should be clear which parts of the system have to be modelled in detail and which parts can be represented by rough and general mathematical relations. The experiment with the Test SWHE answered this question and therefore will improve the calculation methods. It will mark the first steps towards a detailed and reliable FE model being absolutely necessary for the design of an industrial size FLNG SWHE. CONCLUSIONS AND FUTURE OUTLOOK All base-load natural gas liquefaction plants today are located onshore. Different studies on production of LNG on offshore, floating installations have been performed during the recent decades. To be able to install an LNG plant on a floating installation several challenges need to be addressed and technical solutions need to be developed. The performance of the Spiral Wound LNG Heat Exchanger during permanent tilt and movements has been one subject of uncertainty. A research project has been established within the framework of the Statoil-Linde LNG Technology Alliance in order to investigate the thermal and hydraulic behaviour of the SWHE during permanent tilt and dynamic oscillations. A numerical heat exchanger model has been developed and afterwards tuned and verified by use of data from internal PO-27.13

14 fluid flow tests and heat transfer test. The mechanical integrity of the heat exchanger during oscillating operation was verified in separate tests. Knowledge and competency to modify and adapt SWHE design to FLNG requirements has been established through the qualification program, and the necessary design rules for SWHE in moving or tilted environment has been gained from the theoretical studies and the experiments. Linde as a technology provider and Statoil as an operator have hence taken a large step forward towards qualified floating LNG solutions. REFERENCES CITED [1] Waldie, B., Liquid Flow on Inclined Tubes and Tube Banks for Modelling of Large Coil Wound Heat Exchangers for Floating LNG Production, Proc. 12 th Int. Heat Transfer Conf., p411, Grenoble, [2] Waldie, B., Effects of Tilt and Motion on LNG and GTL Process Equipment for Floating Production, GPA Europe Annual Conference, Rome, Italy, 2002 [3] Kocamustafaogullari G. and Chen Y., Falling Film Heat Transfer Analysis on a Bank of Horizontal Tube Evaporator, AIChE Journal, Vol. 34, No. 9, pp , 1998 [4] Jenssen C.B., A CFD model for falling film in spiral wound heat exchangers, in Proceedings from the 4th International Conference on Computational Heat and Mass Transfer, May 17 20, 2005, Paris-Cachan, France, pp , 2005 PO-27.14