Threaded Thermowell On Discharge of Centrifugal Compressor 2

[McDermott1711]

Hi everybody,
Is it allowable to install a threaded thermowell on the discharge of a centrifugal compressor, or it shall be connected to line via a RTJ flange. The discharge operating pressure will be 90 bars. FYI, thread is M30 and there will be an O-ring under crown(head) of the thermowell.
Thanks in advance.

 

[McDermott1711]

Forgot to add: Fluid is natural gas.

 

[Duwe6]

Poor idea - I'd use the RTJ if it was mine. Second choice would be a RF flange. Vibration does bad things to threads. As for "Allowable", it may be [depends on Code]. But it is not prudent. Natural gas [methane] at 90bar / 1300 psi requires respect.

 

[JJPellin]

We would not allow threads in this service. We would require a flanged thermowell. Our standards specify raised face, not RTJ. We had to shut down a unit once when a threaded thermowell backed out from vibration resulting in a serious leak.

Johnny Pellin

 

[hacksaw]

threaded connections are discouraged for pressures exceeding a 150 psig, in fact specifically prohibited in some plants in all but benign process conditions.

Use a flanged connection, LWN nozzle connection not a built up flange, and use remote mounted connection heads. Use as short a thermowell was you can, and insulate the flange.

You also need to give consideration to the how close to the compressor you are located, and the wall thickness of the piping. Severe problems occur if you locate the sensor in an area where accoustic resonances associated with blade passing frequencies.

 

[McDermott1711]

Thanks a lot for your replies.
To hacksaw: Using LWN makes tip of purchased instruments being more off-center and more out-off-flow. What about that?

 

[hacksaw]

When you perform the thermal calcs., it makes almost no difference as long as the difference between the support flange and the process temperature is small, thus the need for insulation. As long as you have 2-3 cm into the flow you are fine.

Accoustic resonances in the pipe with local pressure rises of 1-2 bar contribute to localized temperature variations(standing waves). These can be a more serious source of error.

You do have a longer well with flanged nozzles, but the LWN nozzles provide adequate support stiffness for your well calculations.

 

[moltenmetal]

I wouldn't use a straight thread and o-ring in this service either, but 90 bar natural gas wouldn't push me toward RTJ flange facing either. RF will be fine.

 

[dcasto]

Just to be a contra, in 40 years I've had two raj thermo wells fail and never a threaded one. There are thousands of gas compressors in service across the US and thousands more across the world with threaded thermal wells. And that's reciprocating units! Centrifugal are less susceptible to pulsation and vibration effects.

 

[jbattershell]

API-614 advises that for pipe threads used on systems with hazardous fluids over 75psig operating pressure, all threads shall be seal welded.

sound advice is to use flanged thermowells for piping containing flammable gas or toxic gas. if you use a threaded thermowell, i would seal weld it once it is installed.

the flange type and gasket is also defined in API-614. for class 300# and below, a non-asbestos flange gasket is ok to use. for flange class above 300#, spiral wound gaskets are required (or an RTJ).

sometimes, a customer will require spriral wound on all flanges, regardless or pressure rating when containing flammable gases.

best practice, no threaded joints on flammable service piping. a root valve which is welded is required upstream of the gauges or pressure sensors. down stream of the welded root valve, threaded connections are allowed and are usually 1/2" NPT with 1/2" ss tubing.

 

[hacksaw]

seal welding probably deserves a separate post.

In practice, seal welds in the field do not have adequate controls and most will crack and leak depending on the temperature and the level of vibration involved.

Commonly the thermowell and the flange are different metals and seal welding creates problems as a result of differential thermal expansion as the temperature increased.

Even when the thermowell is factory welded, you can have difficulties in the heat affected zone even if the well is designed for welding.

Before you decide on how and where you are going to install your temperature sensor, you should check into the several published descriptions of thermowell failures at pipeline compressor stations.

 

[McDermott1711]

Before you decide on how and where you are going to install your temperature sensor, you should check into the several published descriptions of thermowell failures at pipeline compressor stations.

Do you have any accessible reference for those published descriptions?
Thanks for your help.

 

[hacksaw]

Simple internet search works.

You won't find solutions, just a bit of insight into what has not worked and what failed.

 

[moltenmetal]

I teach my young engineers to consider seal welding to be a repair procedure rather than a fabrication technique- most of the time. There are rare occasions related to requirements for precise alignment where a seal weld is preferred over a socket weld. But usually it's a crack waiting to happen.