Insert Plate Minimum Reinforcement Dimension 1

[NEBarkley]

I am building a horizontal vessel with 0.375" thick shell walls in Compress and the customer requires large nozzle loadings to be applied to each nozzle. For a particular 30" ID nozzle, the nozzle loadings are so large that the reinforcement pad required is outside of allowable Code reinforcement limits per UG-37, and the only practical solution I know of is to scrap the 0.375" plate to a thicker shell can, 0.75" thick, for that section.

My question is: Would it be possible to make an 0.75" insert plate sufficiently large in distance from the nozzle edge so that I can legally take credit for the plate as though the entire shell were 0.75" thick? If so, how would I determine that dimension? Use the Code mandated limit of reinforcement as the minimum distance the insert plate must meet?

Thanks in advance for any guidance

 

[JStephen]

I believe you can use a repad larger than the "limit of reinforcing" specified in UG-37. The excess material isn't used in the reinforcing calcs, but could be used in the nozzle stress calculations. In that case, I don't know that you'd gain anything by using an insert plate rather than a repad of similar size.

 

[SnTMan]

Agreed

 

[jtseng123]

NEBarkley,
Here are our solutions to this problem coming all the time. Don't confuse the pressure requirement per UG-37 vs. external nozzle load requirement. They are two different things.

1. If using insert plate, which is not uncommon, making sure the width of the insert plate, measured from outside of the nozzle, is not less than: square root (RxT) plus width of the taper section (which in your case will be 1.125" for a 3:1 taper). The insret can be square with 2"-3" radius round corner.

R is the vesssel mean radius of curvature, T is the insert plate thickness.

By doing that, you can run the local stress analysis just at the nozzle to shell junction with shell thickness as 0.75". Square root (RxT) is the distance the local bending stresses in the vessel dissipate to the point of being negligible.

2. If using re-pad, you can use the same square root (RxT) and ignore the stress at the edge. (T is the combined shell and pad thickness). If you are using WRC107, you may end up much wider re-pad becasue WRC107 is too conservative by assuming the whole re-pad is rigid, which is not true. So when we provide sufficient width as stated above, We ignore the stress at the edge. However, if you do want to use the width from WRC107 without showing overstress at the edge, and if the re-pad is wider than 16 times of the pad thickness, an intermediate welds or separated pad sections are required to maintain the integration of pad and shell.

The cost and schedule will be driving which option to pick. I will pick insert since it is clean and less welding in your case.

 

[NEBarkley]

Jtseng123,

Thank you for the excellent response. That is exactly what I was looking for. I believed that this was possible but did not know the exact procedure.

 

[jtseng123]

NEBarkley, Compress has FEA / WRC107 for nozzle local stress analysis. You can use it.

 

[chicopee]

Aside from the good response from jtseng123 , it appears to me that a thicker wall vessel may be more appropriate due to the efforts expanded in fabricating, welding, testing the insert plate and having high related costs. My company used to fabricate bridge girders and you'll note that as you drive on interstate highways, many of these girders have stiffeners which drove the cost of the girders. These girders were fabricated during the construction of the interstate highways and it was probably cheaper than selecting thicker steel for web material. At the end of the20th century because girders with stiffeners were costly, as I understood it, CT DOT selected thicker steel to make these girders without stiffeners. Cost is a driving factor that you should consider.