Riser (Layout C) In Y (Layout B) In Z (Layout A) Max Mz (= 2 * Table 4)
Loop Optimisation Wizard
Return to the model and use File > Save As… to create a new iteration of the model with an expansion loop. After performing the Save As… the analysis must be run once again because the Loop Optimisation Wizard uses existing results in modifying the layout.
Run the analysis and return immediately back to the input. The Loop Optimisation Wizard button will now be available (if this button is ever greyed out then there are no results available for use – rerun the analysis).
Select element 75-80 as this is where the loop will be placed and click the Optimisation Wizard button to access the wizard.
The Loop Design Wizard will appear and the data required can be input in order for CAESAR II can design the expansion loop.
The wizard will create iterations of a loop setup in order to focus in on a specific Stress value or Restraint load. We wish to reduce the Restraint Load on node 10 to below the allowable for API 610 table 4. We will reduce the load to 3300Nm (currently the load is around 5150Nm).
The loop will be located on what is currently the element 75 – 80. This will therefore be as in Loop Layout A.
The final thing to define is the space allowed for locating the loop. A cube of space can be defined which the loop will fit inside. The Wizard will create the largest loop possible in the available space, and if this is below the specified target load, the wizard will continue iterating to create the most efficient loop possible, as close as possible to the target value.
In the Loop Design Wizard, select the OPErating Load Case and choose the target data to be Restraint Load.
As we already have element 75-80 selected, this element will be selected anyway.
The table will be filled in with the results data. Double Click in the MZ cell on the Node 10 row. This will fill in the Node and Type fields. Enter 3300 as the data in the Load field.
Specify the Loop Type as the centre loop type – this matches the type A that we have already determined is the most efficient. The final option in the loop type section will allow the wizard to evaluate all of the Loop Types and determine the most efficient. This takes longer as 8 loop types are defined.
Also in the loop type section, change the Height to Width ratio to <none> to allow the height to vary as needed.
Finally the space available for the loop must be specified. Click the Draw Cube button. A cube of space will be shown in the model. Currently this will be facing the wrong way. Click and Drag the point labelled PT3 to reposition the cube in the correct orientation. On doing this the “Major Direction” field will change to –Z. Increase the size of the cube in the –Z direction to ensure that there is enough room to design the loop.
Once finished, click Design. The wizard will run through a number of iterations and converge on the defined target value.
Once complete the loop will be added into the model automatically. A confirmation message of the total length of pipe and number of bends is displayed
The loop wizard has added an additional ~3070mm of pipe. Split between the two legs to the loop, each leg is now ~1535mm longer. Our quick hand calculation indicated a length of just over 1m, but as noted previously this did not take into account rigid elements or elbows, nor does it consider any intermediate supports such as the guide at node 75. In addition, we used a target value slightly lower than the Max Mz as in the hand calculation.
For simplicity, round up the length of each of the two new legs to 1600mm. With even more flexibility, the MZ moment at node 10 will be lower still.
Re-run the analysis.
Code Checks:
SUS –Max stress is now 16% at node 68. Node 68 is the top of the riser. Recall how CAESAR II adds intermediate node points around the bends as discussed in the TURBO example.
EXP – Max stress is 34% located at node 78. This node is the far end of the elbow at the start of the long Z run
Hanger Sizing:
Carpenter & Paterson DV70 size 11 hanger is still selected. The hot load has decreased slightly.
Pump Load:
Re-run these loads through the API-610 processor once again
This time all loads pass on the discharge nozzle. The local My (global Mz) is now 1.85 times the allowable (we have used the 2x table 4 approach) and so now passes.
We have reduced the load on the pump by adding flexibility into the system in the form of an expansion loop. The addition of this loop required an extra 1.6metres of space. What if this space was not available?
Fix Model – Part 4
We will now return to the model without the expansion loop and once again attempt to reduce the load on node 10, this time assuming that there is no space available for addition of an expansion loop. In this case we will add in an expansion joint instead to add in flexibility.
Open TUTOR.c2 (the file before the expansion loop was added) and save as Tutor_Expansion_Joint_Check.c2
We already know that the issue with the pump is the Mz moment. This moment as we have seen is caused by the thermal growth of 70-75. This horizontal displacement at node 70 causes the bending moment.
As such we wish to prevent this horizontal growth from being applied to node 10.
We will include an expansion joint to absorb this horizontal growth. Adding the expansion joint just above the pump will best absorb this growth.
What type of joint should be used? As we are only trying to absorb movement by lateral deflection only and there is no axial deflection or relative bending rotations at the joint ends, a tied expansion joint will be suitable.
First of all we need to know the horizontal deflection that we have to absorb. We will use CAESAR II to determine this by breaking the system above the pump and viewing the displacements report. The value we obtain from this can be used to select the number of convolutions in the expansion joint.
Select node 20 to 10 and change to 21 to 30
The system will now have two sub-systems sharing the same origin. We need to reconnect 20 and 21.
Add a Y restraint at node 20 and specify node 21 as a CNode.
We also need to prevent any rotation at this point as well. Specify three rotation restraints (RX., RY, RZ) at node 21, CNode 20
Leave the transverse directions X & Z free to move. The system near the pump connection should now look like the following:
Using the EXP case Displacements report, we need to calculate the change in position between Node 20 and Node 21. Nodes 20 and 21 will move together in Y, RX, RY, RZ because of the Node/CNode restraint definitions.
DX = 33.4mm
DZ = 11.446 – 1.150 = 10.3mm
This results in a relative horizontal displacement of 35mm.
This also shows a Mz moment of 4900Nm. This is quite a high load – as the system is completely free to move in the X and Z direction, resulting in the largest displacement. If we were to introduce some stiffness (as would be in the expansion joint itself) this displacement would decrease.
Using the Senior Flexonics/Pathway expansion Joint catalogue, we will select a 3.5kg/cm2 class 150, 8” expansion joint. The catalogue shows a 20 convolution expansion joint provides 38.8mm lateral deflection. This satisfies our requirement. However this expansion joint also adds a lateral stiffness of 6kg/mm or 58.7N/mm.
If we introduce this stiffness, the deflection would reduce. Reduced deflection drops the required number of convolutions and, in turn, increases the stiffness between nodes 20 and 21.
This iterative process can continue until the deflection test fails or the pump load becomes too high. Add the final two restraints (X and Z) between 20 and 21. Set stiffness to 58.7 N/mm
Re running the analysis and viewing the EXP case displacements shows the new relative lateral displacement of around 21mm.
The OPE restraint loads show that the MZ has now decreased drastically and is now around 25% of the previous value.
However, the MY is excessive at over 6000Nm. This moment will also place torsion on the expansion joint and this torsion may also be excessive.
If this observation did not stop the iteration, how would this process proceed?
• Test 16 convolutions - 16 convolutions allow 24.8 mm lateral deflection and has K = 118 N/mm
• Reset X and Z restraint stiffness to 118 and reanalyse
• Check travel limits for the proposed joint and the load limits for the pump. If 16 convolutions is OK and overall joint lateral translation drops, test a shorter (i.e., fewer convolutions) joint.
In conclusion, because of the large global My on the pump and the torque on the expansion joint, the proposed length and location of this joint should be reconsidered.
For the purposes of this exercise, analysis of a 20 convolution, tied expansion joint will be evaluated. For this length, a tied universal joint would probably be preferred; consult manufacturer for other options.
Model the Expansion Joint Assembly
The flanged expansion joint would be located between the discharge nozzle and the existing weld neck flange. To save time in this examination, the expansion joint will be placed between the flange and pipe rather than between the nozzle and flange. The error introduced will be small.
Return to Tutor.c2 and rename as Tutor with Expansion Joint.c2 Select Element 20-30 and access the Expansion Joint Modeller.
Select to create an expansion joint with the following properties:
Pressure 50 pound
Style Tied
Convolution Material 304SS # Convolutions 20
After clicking OK, CAESAR II will split the selected element to fit in the expansion joint. Which end of the element to place the joint must be specified. We will split at the From end (Node 20).
The temperature of the element in question is 315°C. Apply this temperature to the joint, which will subsequently cause the stiffness to be adjusted.
The expansion joint modeller will finally confirm the creation of all the elements to be used in creation of the joint. The stiffnesses will also be displayed, along with the Allowed Movement.
Click build and CAESAR II will attempt to define the expansion joint using the data supplied/obtained from the catalogue.
Re analyse the system.
Using the Expansion Case displacements, calculate the lateral deflection between nodes 21 and 22 – the nodes on either end of the expansion joint.
DX = 20mm; DZ = approx. 12mm. The overall deflection is therefore around 23mm.
The Flexonics catalogue shows that this displacement is acceptable for a 20 convolution joint. There is minimal Axial deflection (DY = approx. 0.3mm) and Angular Rotation (RX and RZ = 0mm). Torsion in the joint is 0.3 degrees. The catalogue actually shows that this torsion is excessive.