understanding of hot air recirculation …...bundle of the main cryogenic heat exchanger (mche),...

15
1 UNDERSTANDING OF HOT AIR RECIRCULATION PHENOMENA IN AIR-COOLED BASE LOAD LNG PLANT Siti Farhana Bt M Shaari Malaysia LNG Sdn Bhd Bintulu, Sarawak, Malaysia E-mail: [email protected] Kei Kubota JGC Corporation Yokohama, Japan Email: [email protected] KEYWORDS: air-cooled heat exchanger, hot air recirculation, computational fluid dynamics, fluid flow, liquefied natural gas ABSTRACT With the push towards air cooled systems in modern liquefied natural gas (LNG) plants, the need for better methods for analyzing the dispersion of hot air exiting the air-cooled heat exchangers (ACHE) is becoming increasingly important. One focus for improving the LNG plant production rates is the control of hot air recirculation (HAR) around the ACHE units and gas turbines. Site weather conditions such as wind speed and direction are one of the major factors that need to be considered when evaluating the HAR phenomenon. Current computational fluid dynamics (CFD) methods for modeling HAR in LNG plants do not sufficiently consider the site weather conditions for the input parameter, resulting in a decreased accuracy of the CFD models. This paper highlights the need for CFD models to include detailed long term weather data in order to accurately simulate the HAR phenomenon. This paper provides case studies which clearly demonstrate the importance of including detailed weather data for CFD modeling. Results have shown close agreement with actual temperature data taken around ACHE and gas turbine air intake locations. This paper also demonstrates how detailed weather data combined with CFD modeling allows plant performance to be evaluated under changing weather conditions. This provides the owner/operator with reliable data that can be used for improving the overall LNG production rate. INTRODUCTION Malaysia LNG (MLNG) Tiga is an air-cooled, gas turbine driven LNG plant which consists of two LNG trains (Train 7 and Train 8). The plant utilizes Propane Pre-cooled Mixed Refrigerant (C3-MR) by Shell/Air Products and Chemicals (APCI) with expanders for the liquefaction process. As an air-cooled base load LNG plant, the performance of the air-cooled heat exchangers (ACHE) is critical in ensuring sustainable LNG plant production. One of the main contributors to the production capacity being constrained for MLNG Tiga is the rise in the local air temperature caused by rejected heat from the large banks of ACHE units. This has impacted the performance of the ACHEs and the gas turbine drivers of the refrigeration compressors for the two-train LNG plant.

Upload: others

Post on 28-Feb-2020

8 views

Category:

Documents


0 download

TRANSCRIPT

1

UNDERSTANDING OF HOT AIR RECIRCULATION PHENOMENA IN AIR-COOLED BASE LOAD LNG PLANT

Siti Farhana Bt M Shaari Malaysia LNG Sdn Bhd

Bintulu, Sarawak, Malaysia E-mail: [email protected]

Kei Kubota JGC Corporation Yokohama, Japan

Email: [email protected]

KEYWORDS: air-cooled heat exchanger, hot air recirculation, computational fluid dynamics, fluid flow, liquefied natural gas

ABSTRACT

With the push towards air cooled systems in modern liquefied natural gas (LNG) plants, the need for better methods for analyzing the dispersion of hot air exiting the air-cooled heat exchangers (ACHE) is becoming increasingly important. One focus for improving the LNG plant production rates is the control of hot air recirculation (HAR) around the ACHE units and gas turbines. Site weather conditions such as wind speed and direction are one of the major factors that need to be considered when evaluating the HAR phenomenon.

Current computational fluid dynamics (CFD) methods for modeling HAR in LNG plants do not sufficiently consider the site weather conditions for the input parameter, resulting in a decreased accuracy of the CFD models. This paper highlights the need for CFD models to include detailed long term weather data in order to accurately simulate the HAR phenomenon. This paper provides case studies which clearly demonstrate the importance of including detailed weather data for CFD modeling. Results have shown close agreement with actual temperature data taken around ACHE and gas turbine air intake locations.

This paper also demonstrates how detailed weather data combined with CFD modeling allows plant performance to be evaluated under changing weather conditions. This provides the owner/operator with reliable data that can be used for improving the overall LNG production rate.

INTRODUCTION

Malaysia LNG (MLNG) Tiga is an air-cooled, gas turbine driven LNG plant which consists of two LNG trains (Train 7 and Train 8). The plant utilizes Propane Pre-cooled Mixed Refrigerant (C3-MR) by Shell/Air Products and Chemicals (APCI) with expanders for the liquefaction process. As an air-cooled base load LNG plant, the performance of the air-cooled heat exchangers (ACHE) is critical in ensuring sustainable LNG plant production.

One of the main contributors to the production capacity being constrained for MLNG Tiga is the rise in the local air temperature caused by rejected heat from the large banks of ACHE units. This has impacted the performance of the ACHEs and the gas turbine drivers of the refrigeration compressors for the two-train LNG plant.

2

Figure 1. MLNG Tiga Train 7 and Train 8

The performance of the ACHEs and gas turbines can be significantly influenced by the temperature of intake air into the ACHEs and turbine drivers. The performance of the propane condenser, which accounts for nearly half of the ACHEs for MLNG Tiga, is especially sensitive to the intake air temperature since the propane refrigerant is condensed from gas to liquid at this location. A rise in the intake air temperature caused by hot air recirculation (HAR), the phenomenon in which the warm air from the ACHEs flows back into the ACHE intake or the intake of other equipment, is therefore a critical issue that the plant owner/operator faces.

The ACHE intake air temperature is strongly influenced by the weather conditions at the site such as wind direction and wind speed. The following are the two modes of recirculation, which are heavily dependent on the weather conditions at the site.

Recirculation of hot air from the exhaust of an adjacent train when the wind blows perpendicular to the LNG trains

Recirculation of hot air within the same train, or self-recirculation

An in-depth understanding of the relationship between the weather conditions and degree of HAR can considerably benefit the operation of an LNG plant. A better understanding of HAR can also provide clues to mitigating this issue for a plant in operation.

This paper introduces a study which combines a thorough analysis of actual plant data collected at the MLNG Tiga plant with results from CFD analysis. Data from 152 temperature sensors installed near the air intakes of the ACHE units and wind data from two weather stations were used to verify the accuracy of the CFD simulation. CFD is shown to be an effective engineering tool in large-scale fluid flow applications such as in this study. This paper also proves how effective this combined approach is in understanding the HAR phenomenon.

3

OWNER / OPERATOR PERSPECTIVE

HAR in MLNG Tiga is normally observed during the day time around 10 am to 4 pm, and at night from 9 pm to 11 pm. The impact of HAR on the plant production is significant since it largely occurs around the propane condenser, which has more than half of the total cooling duty provided by all the ACHEs in the liquefaction unit. The impacts to the plant condition can be summarized as follows.

a. Increase in the discharge pressure of the propane compressor which can lead to a relief at the discharge line of the compressor; and

b. Increase in the compression power required from both the MR and C3 gas turbines which can lead to a plant trip should the gas turbines be overloaded.

Figures 2 and 3 below show the impact of HAR toward the discharge pressure of the propane compressor and compression power required from both the MR and C3 gas turbines, respectively.

19

20

21

22

23

24

25

26

27

28

29

0

5

10

15

20

25

30

35

40

20-07-12 20-07-12 20-07-12 20-07-12 20-07-12 21-07-12 21-07-12 21-07-12 21-07-12 21-07-12 22-07-12Di

scha

rge

Pre

ssur

e (b

arG

)

Air I

ntak

e Te

mpe

ratu

re (d

egC)

Effect of Hot Air Recirculation Towards Discharge Pressure of C3 Compressor

Air Intake Temp (C3 Subcooler) Air Intake Temp (C3 Condenser) Air Intake Temp (MR aftercooler)Air Intake Temp (MR aftercooler) Discharge Pressure of C3 Compressor Set Point 914-RV-095Set Point 914-PICA-230

Figure 2. Impact of HAR on Discharge Pressure of Propane Compressor

4

0

5

10

15

20

25

30

35

40

MR

and

C3 G

as T

urbi

nes P

ower

Mar

gin

Air I

ntak

e Te

mpe

ratu

re (d

egC)

Effect of Hot Air Recirculation Towards MR and C3 Gas Turbines Power Margin

Air Intake Temp (C3 Subcooler) Air Intake Temp (C3 Condenser) Air Intake Temp (MR aftercooler)Air Intake Temp (MR aftercooler) C3 Gas Turbine Power Margin Power Margin (Low Alarm)MR Gas Turbine Power Margin

Figure 3. Impact of HAR on Compression Power Required from both MR And C3 Gas Turbines

Experiences and Challenges

The HAR issue has been a major factor in a number of plant slowdowns and trips ever since the 2003 commissioning of MLNG Tiga. However, as the panel operators have become more experienced in handling the plant operation during HAR conditions, the number of plant trips due to HAR has dropped significantly. The last plant trip was recorded in 2007.

Currently, effective interventions by the panel operators play a vital role in handling the plant operation when HAR occurs. This includes ensuring the discharge pressure of the propane compressor stays below its relief valve set point and that the available power of the gas turbines is within the minimum allowable limit. As the ambient temperature starts to increase, key parameters such as the trending of ACHE air intake temperature measurements, the power margin of the MR and C3 gas turbines, and the discharge pressure of the propane compressor are closely monitored. Critical interventions required include the optimization of the MR composition and propane kettles level, adjustment of the Heavy Mixed Refrigerant (HMR) flow to the middle bundle of the Main Cryogenic Heat Exchanger (MCHE), followed by the Light Mixed Refrigerant (LMR) flow to the top bundle of the MCHE, and finally the LNG flow until the discharge pressure of the propane compressor and available power of the gas turbines return to an acceptable level. Once HAR subsides, the operator will normalize the process conditions and increase the LNG production back to the target production.

Although plant trips due to HAR have been eliminated, plant slowdowns are still a concern in order to sustainably maximize the LNG capacity from MLNG Tiga. Figure 4 below shows a typical trending of the plant slowdown due to HAR for MLNG Tiga.

5

90

95

100

105

110

115

120

0

5

10

15

20

25

30

35

40

20-07-12 20-07-12 20-07-12 20-07-12 20-07-12 21-07-12 21-07-12 21-07-12 21-07-12 21-07-12 22-07-12

LNG

Prod

uctio

n

Air I

ntak

e Te

mpe

ratu

re (d

egC)

Plant Slowdown due to Hot Air Recirculation

Air Intake Temp (C3 Subcooler) Air Intake Temp (C3 Condenser) Air Intake Temp (MR aftercooler)Air Intake Temp (MR aftercooler) LNG Prod LNG Target

Target

+5%

-5%

Figure 4. Plant Slowdown Due To HAR for MLNG Tiga

MLNG Tiga is designed with an HAR margin of 2°C. However, in reality, the temperature rise due to HAR was found to be much higher. A study conducted in 2012 showed that an increase of 1°C in the ACHE air intake temperatures can contribute to about 35 ktonnes of LNG production loss per train.

WEATHER CONDITIONS

A thorough assessment of the site weather conditions, particularly the wind direction and wind speed, is critical for an air-cooled LNG plant. Collecting accurate long term wind data is essential in understanding the relationship between the wind conditions and HAR. For this purpose, wind data from the plant was collected for nearly one year from April 2011 to January 2012 using two weather stations. The weather stations were installed near the MLNG Tiga area to measure wind direction, wind speed, and ambient temperature. Measurements were taken at a height of approximately 1.5m from the ground level. Figure 5 shows the installation locations of the weather stations. A summary of the collected wind data is included in the following section.

6

Figure 5. Weather Station Installation Locations

Wind Data Analysis

The wind data was divided into 16 wind directions and 6 wind speed ranges, with true north set at 0 degrees. Plant north is oriented 29.5 degrees in the counter clockwise direction of true (or magnetic) north. All measured wind directions are based on the true direction. Figure 6 below shows the frequency of occurrence (or wind rose) for each of the 16 wind directions measured at the two weather stations.

Weather Station 1 Weather Station 2

Figure 6. Wind Rose for Entire Measurement Period

Wind measurements for Weather Station 1 appear to be influenced by the presence of Train 7, which may have acted as a barrier, thus limiting plant easterly winds from being recorded properly. Therefore, the Weather Station 2 wind data is likely to be more accurate for easterly winds. Although the wind roses for the two weather stations are slightly different due to the locations which they were installed, the diurnal wind pattern appears to be similar. Winds near the seashore often exhibit a pattern of interchanging land and sea breeze throughout one day [1]. Figure 7 illustrates this point clearly.

7

(a) 21:00 to 9:00, Land Breeze (b) 9:00 to 13:00, Land to Sea Breeze Transition

(c) 13:00 to 20:00, Sea Breeze (d) 20:00 to 21:00, Sea to Land Breeze Transition

Figure 7. Wind Rose Transition throughout the Day

The plant runs perpendicular to the coastline, which is located north of the plant. Land breeze and sea breeze are clearly observed at the plant as shown in Figure 7. From 21:00 to 9:00, the wind is predominantly blowing from the land to the sea, whereas from 13:00 to 20:00, the wind is predominantly blowing from the sea to the land. The transition from land breeze to sea breeze occurs between 9:00 to 13:00, and the transition from sea breeze back to land breeze occurs between 20:00 to 21:00. No major seasonal trends were noticed from the monthly wind data.

ACHE INTAKE TEMPERATURE MEASUREMENTS

Wireless temperature sensors, which were installed specifically for this HAR study, were used along with permanent sensors to monitor ACHE and gas turbine air intake temperatures throughout the entire measurement period. A total of 152 temperature sensors were used to record temperatures at an interval of 10 minutes. The temperature sensors were distributed evenly throughout Train 7 and Train 8 so that an accurate temperature contour for the intake temperatures could be obtained. A detailed comparison with the CFD results can be made as a result of the large amount of temperature data collected. Figure 8 shows the installation locations for each of the ACHE air intake sensors for Train 8 (identical locations for Train 7).

8

Figure 8. Installation Locations for Train 8 ACHE Temperature Sensors (72 Sensors per Train)

Temperature Data Analysis

As mentioned earlier, a major issue that the plant operator has experienced is the sudden rise in the Train 8 propane condenser air intake temperatures in the day time due to HAR. Although the night time effects of HAR on plant performance cannot be ignored, HAR has a much more detrimental impact during the day since temperatures are already elevated and close to the ACHE design limit. Based on the temperature data collected at the weather stations, the mean daily maximum ambient temperature was found to be approximately 30°C. Ambient temperatures generally reached a peak around 14:00. For this reason, the focus for this paper was on the day time ACHE air intake temperatures.

Figure 9 shows the Train 8 average propane condenser temperature rise under various wind directions for steady wind conditions. The wind was categorized as stable if there were no changes in the wind direction for more than 1 hour. Elevated ACHE air intake temperatures were clearly seen from the data as the wind blew more perpendicular to the trains (cross wind). The local temperature rise for several bays were much higher than the averages shown in Figure 9, with some temperatures exceeding 6°C under cross wind conditions.

Figure 9. Train 8 Propane Condenser, Average Measured Temperature Rise (Intake Minus Ambient) under Day Time Wind Directions

9

COMPUTATIONAL FLUID DYNAMICS

CFD is commonly used in order to understand the complex 3-dimensiontal nature of fluid flow for individual equipment within an LNG plant. With the growing availability of more powerful computers, CFD can now be used to evaluate the fluid flow at a larger scale as in the study presented in this paper. For this study, the flow characteristics of the hot air from the ACHE outlets were determined using CFD under different wind conditions.

CFD Model

The flow and temperature fields for the MLNG Tiga plant were simulated using the CFD software CFX 14.0 developed by ANSYS. The k-ε turbulence model was used to simulate the effects of turbulence. The effects of the atmospheric boundary layer were considered in properly simulating the wind for this analysis [2, 3]. A steady state analysis, which does not consider the effect of gusts or the transient nature of the wind, was conducted. The mean daily maximum temperature at the plant, which was measured to be approximately 30°C, was used as the ambient air temperature for the CFD analysis.

The CFD model includes buildings, compressor enclosures, tanks, columns, vessels, drums, stacks, heat exchangers, and other equipment aside from the ACHEs and gas turbines that may significantly influence the air flow or temperature distribution within the plant. Concrete/steel structures, piping, and other small equipment were included as porous regions in the model. Several views of the CFD simulation model which was used for this study are shown in Figure 10.

Figure 10. MLNG Tiga CFD Model

10

Understanding the HAR Phenomenon

The measured wind data shown in Figure 6 was used as the basis for determining the conditions for the CFD simulation. In order to study the effects of day time winds, two different wind directions were chosen for this study. The following two wind cases were considered.

NW wind, which represents the prevailing day time wind direction (see Figures 6 and 7)

WNW wind, which represents a relatively frequent cross wind direction

The W wind and WSW wind directions (included in Figure 9) were excluded from the CFD analysis since these winds were rarely observed at the site. Also, since wind speeds were mostly between 1 m/s and 3 m/s as shown in Figure 6, an intermediate wind speed of 2 m/s at a height of 1.5 m from the ground was used for all of the cases. HAR generally does not occur under calm wind conditions.

The results from the CFD analysis were used to understand the HAR phenomenon. Outputs from the CFD analysis such as the temperature contour and isothermal surface results are shown in Figure 11.

Figure 11. Temperature Contour Plot and Isothermal Surface at 10°C above Ambient (WNW Wind)

A comparison between the streamlines for the NW wind case and the WNW wind case is shown in Figure 12. Both self-recirculation and recirculation of hot air from the upstream Train 7 is observed for the WNW wind case, leading to higher intake temperatures compared to the NW wind case. Figure 11 also shows a large volume of hot air from Train 7 flowing toward Train 8 for the WNW wind.

NW Wind: Minor Self-Recirculation WNW Wind: Both Modes of Recirculation

Figure 12. Streamline Results Showing the Different Modes of Recirculation

11

The major obstructions in the pipe rack underneath the ACHEs such as the compressor shelter result in a large pressure drop across the pipe rack. Due to this pressure drop, a large volume of hot air from Train 7 flows in a downward direction at the wake of the train for cross winds. This hot air eventually reaches the Train 8 ACHEs and raises the air intake temperatures mainly on the west side of Train 8 as illustrated from the WNW wind streamline results in Figure 12.

Site Data vs. CFD Results

In order to assess the impact of HAR, the intake temperatures to the ACHEs and gas turbines were calculated using CFD. The temperature rise above the ambient temperature at the intake of each of the ACHE bays and gas turbines due to HAR were determined. These results were used to identify locations of ACHE air intake hot spots and compared with the site temperature data.

The Train 7 and Train 8 ACHE and gas turbine air intake temperature rise results from the actual site temperature data and CFD analysis for a steady WNW wind are shown in Figures 13 and 14, respectively. The average temperature rise for each of the ACHE bays is provided, with several bays exhibiting a rise of more than 5°C.

Figure 13. Temperature Rise above Ambient, Site Measured Data (WNW Wind)

12

Figure 14. Temperature Rise above Ambient, CFD (WNW Wind)

A comparison between the air intake temperature rise results from the site data and CFD for the ACHEs on the west half of Train 8 (including propane condenser) is shown in Figure 15. Temperature results have been plotted from the north end of the train to the south end. The CFD results show a very close agreement with actual site temperature data as shown in Figures 13 through 15. The average temperature rise for the Train 8 propane condenser from the CFD analysis was 2.0°C, nearly the same as the 2.2°C temperature rise based on the site data. Both the site data and CFD results show local hot spots toward the southern part of the propane condenser (E-91445), with a number of bays approximately 6°C above the ambient temperature.

Figure 15. Temperature Rise for ACHEs on West Half of Train 8, Site Data vs. CFD

13

A few discrepancies are seen between the site data and CFD results, especially for the east half of Train 8, where air intake temperatures appear to be underestimated. The reasons for the discrepancies seen between the site data and CFD may include the following.

Actual ACHE performance not consistent with the design specifications used for the CFD model

Inaccuracies associated with the measuring devices

Influence of gusts on the recorded temperatures

Heat sources near ACHE intake areas that were not accounted for in the CFD model

High temperature measurements for the gas turbine intakes due to the sensors not being properly shielded from the sun

Approximations used in the CFD model such as the pipe rack congestion

The results have proven the effectiveness of CFD in understanding the HAR phenomenon at an actual plant. However, CFD alone cannot explain everything since there is a limitation in simulating nature and the wind. Therefore, an extensive amount of site experience and data is crucial in assessing the CFD results and ultimately dealing with the issue of HAR.

HAR MITIGATION MEASURES

Realizing the significant impact to production loss due to HAR, MLNG has implemented several initiatives in 2012 to minimize the impacts of HAR. As explained earlier, Train 8 exhibited elevated ACHE intake temperatures due to HAR. In order to reduce the amount of HAR and lower ACHE air intake temperatures for Train 8, various mitigation measures were proposed. Modifications to the hardware and ideas to optimize the process were considered. The hardware modifications that were carried out for MLNG Tiga are the following.

a. Installation of a permeable windscreen on the west side of Train 8 to divert the hot air away from the air intake area (Figure 16)

b. Changing the ACHE pulleys to a larger size in order to increase air flow rate

c. Wrapping insulation on the heat transfer fluid pipes underneath the ACHEs to reduce the impact of any additional heat source near the intakes

14

Figure 16. Installation of a Permeable Windscreen on the West Side of Train 8

HAR mitigation measures such as the permeable windscreen has shown promise in reducing the degree of HAR while at the same time maintaining a sufficient amount of intake air to the ACHEs. As of December 2012, the production gain in Train 8 as a result of the hardware modifications is estimated to be 7 ktonnes. The success for Train 8 has justified proceeding with similar modifications for Train 7 in 2013.

CONCLUSION

An increase in the air intake temperatures of the ACHEs and gas turbines due to HAR can lead to a detrimental loss of production for an air-cooled LNG plant. When HAR occurs, the operator is faced with the difficult challenge of trying to maintain the LNG production rate, but in reality a loss of production is inevitable. Rather than relying on the skills of an operator to minimize the effects of HAR whenever it occurs, effective solutions at the site to mitigate HAR must be considered. As explained in this paper, HAR cannot be dealt with lightly, and understanding when and how it occurs is of utmost importance for any plant owner/operator in solving this issue.

The combination of actual data from the plant site and CFD analysis has proven to be effective in understanding the HAR phenomenon at MLNG Tiga. Since weather conditions such as the wind direction and wind speed strongly influence the degree of HAR observed, accurate long term measurements of the wind must be taken. In order to get meaningful results from the CFD, the weather data must be carefully evaluated first before using it as input to the CFD model. In this way, the reliability of the CFD simulation can be ensured. The CFD results have been shown to match well with actual site ACHE data from MLNG Tiga. Evaluating all of this information is not an easy task, which is why a countless amount of site investigations and data regarding HAR is critical in handling this issue. Effectively dealing with HAR only comes with years of experience since there are an overwhelming amount of factors to consider.

The best course of action against HAR can be proposed following the method explained above. In the case of MLNG Tiga, hardware modifications to Train 8 were implemented to minimize the impact of HAR. An

15

improvement to the LNG production rate was seen for Train 8 as a result of these modifications. The method presented in this paper in analyzing HAR was demonstrated to be effective for an existing LNG plant. However, constraints for a plant once it is constructed can significantly limit the mitigation measures that can be considered, which was the case for MLNG Tiga. By taking into account the impact of HAR in the early design phase, a plethora of ideas based on extensive HAR experience can be proposed to optimize the plant for maximum LNG production.

REFERENCES

1. Wallace, J.M. and Hobbs, P.V., 2006, Atmospheric Science: An Introductory Survey, Second Edition, Academic Press, Burlington.

2. Garratt, J.R., 1992, The atmospheric boundary layer, Cambridge University Press, Cambridge.

3. Richards, P.J. and Hoxey, R.P., 1993, “Appropriate boundary conditions for computational wind engineering using k-ε turbulence model”, Journal of Wind Engineering and Industrial Aerodynamics, 46 & 47, pp.145-153.