For a coil-wound liquified natural gas (LNG) heat exchanger, an existing 66” and new 72” bolted flange joint have been analysed and reviewed for mechanical loading and multiple thermal load cases in accordance to the 2013 edition of ASME VIII Div.1 and Div.2 by Dynaflow Research Group (DRG) using ANSYS Mechanical.
The flanges operate within the cryogenic temperature range of -195.5°C to 65.5°C. To limit thermal gradients and stresses the flanges are insulated. The composition of natural gas depends on the geological location, consisting primarily of methane and ethane. The exact composition of the LNG was unknown for this analysis and therefore the properties of methane were used as this is the most critical component from a leak rate point of view.
The flange assemblies consist of two dissimilar flanges, one of stainless steel (SA182 F304/F304L) and a mating flange of aluminium (SB247-5083-H112); bolted together using 1.5” stainless steel bolts (SA-193-B8-Cl.2). The dissimilar nature of the flange results in a large difference in thermal expansion coefficients which can impact the sealing of the gasket, a factor that Dynaflow considered in this analysis.
Figure 1 | ANSYS Flange Model
A spiral wound gasket is seated within the flange assembly. The gasket properties and constants were supplied by the manufacturer Flexitallic. However, whilst the nonlinear compression and unloading moduli of the gaskets were also provided, no compression or moduli data were available for gaskets of sizes 66” and 72”. The moduli were therefore scaled with respect to the gasket thickness, based on the constant stress-strain behaviour, resulting in the gasket compression and unloading curve calculated in Figure 2.
Figure 2 | Gasket compression and unloading curves for the Flexitallic low seating stress spiral wound gasket.
The model is loaded with each bolt being initially preloaded to the specified bolt load of 275kN with internal an design pressure of 10.4barg.
Four load cases are considered being:
External Loads
Axial and shear forces alongside bending and torsional moments are applied at the junction of the nozzle to the heat exchanger shell.
Steady State Thermal
Analyses made of the flange with internal temperatures at -195°C and 65.5°C where heat transfer to the environment is considered in both insulated and uninsulated states.
Transient Thermal
Analysis made of start-up and shutdown operations where the interior flange temperature of the assembly changes over time (at 28°C/hr). Due to improved thermal conductivity of aluminium the temperature of the aluminium flange temperature changes more rapidly then the steel, inducing a thermal difference ontop of the differing thermal expansions resulting in high bending stresses.
Thermal Bowing
This transient thermal case reviews the occurance where the cryogenic liquid enters the pipe. During the process the bottom of the internal volume fills with cryogenic liquid at -80°C, whilst the remainng volume fills with vapour at -35°C. This situation creates a temperature gradient between the top and bottom of the pipe at about 45¯C, resulting in thermal bowing as the bottom of the pipe contracts relative to the top of the pipe. Different liquid heights were investigate in order to determine the worst case for structural failure or leakage.
Figure 3 | Temperature distribution for thermal bowing (top) and corresponding stress distribution along the circumference of the stainless steel flange to pipe weld.
Results & Discussion
Bolt Loads
Supplemental bolt assessments were developed, checking that the maximum average stress across the bolt cross-section, neglecting stress concentrations do not exceed two times the allowable stress. For both flanges the bolt stresses after bolt-up remained within the allowable bolt stresses.
Figure 4 | Stress distribution on bolt-cross-section [Pa].
Stresses
The flange assembly stresses were reviewed against plastic collapse of ASME VIII Div.2 part 5. Stress linearization’s were performed at discontinuities to split the stress components in membrane, bending and peak stresses as per ASME VIII Div.2 Annex 5-A, Figure 5 shows an illustration of where the stress linearization was performed.
Figure 5 | Direction and location of SCLs.
For all load cases the existing 66” flange stresses was found to exceed the allowables of both the steel and aluminium flanges by ~180%.
The excessive stresses at the flange to pipe weld and the high stresses at the interface between the flange blade and the hub were the result of the bending of the flanges due to the applied loads. DRG recommended to increase the thickness of the flange material at these locations or by changing the design so as to reduce the loads applied.
For the thermal bowing case, it was found that the stress benefit from increasing the thickness of the piping and flange was counteracted by the corresponding increase in stiffness due to the displacement from the thermal bowing originating from the temperature differential and thus remaining the same, however as the stress due to the other external loads does decrease when the hub and pipe thickness are increased there is an end benefit to the flange stresses from the thickness increase.
Using this investigation into the 66”flange, DRG iteratively designed the new 72” flange until the stresses were found to remain within the allowables in ASME VIII Div.2 for all load cases.
Leak Rates
A leak-rate analysis was performed on both the 66” and 72” flanges based on the rules from the ASME BFJ committee, basing the leak rate of the natural gas on the known leak rates of helium multiplied by the ratio of kinematic viscosities between the two fluids.
The leak rates are determined per load case and are compared to the equivalent tightness class leak rates from ASME BFJ of C1, C2, and C3. The results of which are provided in
Flange | Load Case | Estimated Mass Leak-rate [kg/yr] | Class C1 Allowable Leak-rate [kg/yr] | Class C2 Allowable Leak-rate [kg/yr] | Class C3 Allowable Leak-rate [kg/yr] | Estimated Equivalent Tightness Class |
66” | External Loads | 0.275 | 1870 | 18.7 | 0.187 | C2.92 |
Steady State Thermal Loads | 0.230 | C2.95 | ||||
Transient Thermal Loads | 0.277 | C2.91 | ||||
Thermal Bowing | 0.262 | C2.93 | ||||
72” | External Loads | 0.545 | 2020 | 20.2 | 0.202 | C2.78 |
Steady State Thermal Loads | 0.526 | C2.79 | ||||
Transient Thermal Loads | 0.529 | C2.79 | ||||
Thermal Bowing | 0.515 | C2.80 |
Conclusion
The conclusions of the analysis were:
- The existing 66” flange did not conform to ASME VIII Div.2.
- The new 72” flange conforms to ASME VIII Div.2 with the recommended geometric changes from DRG given in Table 1‑1.
- Both flange designs have leak rates equivalent to a C2 tightness class from ASME BFJ.
Table 1‑1 | Final 72″ flange assembly dimensions.
Design | Units | Stainless Steel | Aluminium |
Bore Diameter | mm | 1796.8 | 1778.0 |
Hub Diameter | mm | 1930.4 | 1930.4 |
Flange Diameter | mm | 2108.0 | 2108.0 |
Flange Blade Diameter | mm | 165.1 | 215.9 |
Pipe Thickness | mm | 16.0 | 25.4 |
Hub Length | mm | 241.3 | 241.3 |
Bolt Circle Diameter | mm | 2032.0 | 2032.0 |
Bolt Diameter | mm | 38.1 | 38.1 |
No. of Bolts | – | 72 | 72 |
This project shows the importance of looking critically at unique situations and dissimilar materials within designs. The difference in flange materials alongside the varying temperature gradients over the flange induces large bending moments, causing overstresses.