Thermowell Vibration Calculation 1

[Scotsinst]

I searched keyword and FAQ but couldn't locate what I was looking for.
1.Is the Thermowell vibration calculation to Power Test Code ASME VOL 81 by J.W. Murdock or similar available on the web?
2. As I understand it, the calculation is based on a screwed thermowell, in a weldolet style fitting where the thermowell length is the same as the immersion length. I think the calculation is independent of fluid which seems odd since I assume a TW would be affected more by 32ft/s mercury than 32 ft/s air.
3. In the refining industry, hydrocarbon TW's are flanged, mounted on, typically, a 6" long nozzle plus the pipe wall thickness. If a thermowell was 6" long it would not project into the process flow.
4. A contractor calculated that, per the above test code calc, the TW needed to be 5" long which is absurd since it would not project into the pipe.
I would appreciate any suggestions as how to modify the calc to take into account the approximately 6" of dead space in the nozzle.

 

[imok2]

Try this:

http://www.thermosensors.com/catalog/thermowells.pdf

 

[jsummerfield]

The resonance phenomenon is related to the insertion length and well pattern (heavy duty tapered or straight, etc.) Flanged wells that require more insertion to get into the flow profile are more subject to failure than shorter wells. Keep the wells as short as possible.

The web reference

http://www.thermosensors.com/catalog/thermowells.pdf

is great. It reflects similar data to my Excel spreadsheet. I will compare when I have an opportunity. I spend months going through metallurgical data to obtain my stress values.


John

 

[hacksaw]

Seems that there has been a lot of work in this area in the last tem years. You might try

There are several other papers that might be of interest:

1. Flow induced vibration of thermowells, ISA Trans. 38, 1999, pp. 123-132.
2. Static and dynamic stresses of practical thermowells, ISA Trans. 39, 2000, pp. 133-142.
3. Thermowell design and selection, Hydrocarbon Proc., Nov. 2001.
4. Are your thermowells safe?, TAPPI Journal (Solutions), April 2002.
5. Thermowell Integrity in pipeline services, O&GJ April 2002.


You might try the Linda Hall Library(web site). Worked with a fellow that dug into that stuff. He indicated that Murdocks paper is not commonly available.

I believe that a sizing code has been written. At one time it was available on the internet for free at temperatures.com. it included density effects and tended to be more conservative than the 19.3 procedure and claimed to be based on a finite element calcualtion. i do not know if it is still being offered.

The last paper in the ref. list included several case study failures, that were not predictable by 19.3. Two reasons: corrosion and the fact that 19.3 is based on an incomplete stress model.

 

[hacksaw]

the short an sweet answer to the last question in the orig. post: you have to use L as the unsupported length in figuring the critical frequency.

seems like the Japanese discovered that the ptc rating in liquids is off by a factor of 50% in liquids. they have a special publication that deals with thermowell calculations and is available from JSME. It comes with a cd and shows high speed videos of thermowells experiencing two types of vortex shedding.

 

[Scotsinst]

Thanks for your very helpful input. Some of the sites are unavailable but I found some useful info without locating any software yet.
I understand that we need to use L as the unsupported length to calculate the resonant frequency.
My query is based on the fact that 6" of the unsupported length is not exposed to fluid velocity (is in a "dead" zone). I assume the calculation bases resonant frequency solely on fluid velocity at the TW tip.
In Refinery FCC units, the Reactor Vapor fluid velocity is typically 90-120 ft/sec. We have no recorded TW failures (in about 100 operating units over 60 years) but recently some vendors have declined to supply TW's for this service without velocity collars. Velocity collars are not suitable due to coking potential.
Another issue is that units are revamped many times, sometimes operating at 200% of initial design flow. If the TW's are designed for a certain flowrate, they would need to be evaluated for each flowrate. I know this is not typically done in the refining industry.

 

[hacksaw]

A number of refiners are indeed performing velocity checks on all thermowells, especially those where the flow rates have increased. Two issues: equipment damage and fire. They generally find that about 30% of there wells are exposed to stress levels that are un-acceptable from a piping design aspect.

Where you don't have a rating method, but you know that the peak velocity is less than 150 ft/sec (compressible flows), you simply use a maximum thermowell length of 4".

With a 6" nozzle, we set the length to allow no more that 1-2" of the thermowell tip to protrude into the flow. Safe thermowells do not impair measurement accuracy.

 

[hacksaw]

Have located the thermowell stress analysis code mention previously. It is at

http://www.temperatures.com/twellwakes.html

Evidently, restricted to steam and condensate applications, but allows a good range of alloys.

If I remeber by fluids well enough, it seems that for a given service temperature the flow induced stresses are proportional to density, so being limited to steam does not prevent a estimate for other fluids.

Believe the code also includes an estimate of the tip erros for a given amount of pipe insulation. The instructions are a bit brief, and being a stress analysis it does not give you a go-no-go response, i.e. you have to exercise engineering judgement.

 

[Scotsinst]

Thanks, I'll check it out. My main concern is to have the Thermocouple tip far enough from the pipe wall to negate any wall effects on the measurement.
Having a TW poke 2" in a 48" pipe is not acceptable from a measurement standpoint. I don't care what the Reynold's number is, from our field experience the wall effects affect the measurement.
Maybe we need to go with a very thick TW?

 

[hacksaw]

There are several issues here: stratification, conduction errors, and radiation errors.

You control stratification errors in part by placement, but no single thermowell/sensor will solve that issue, regardless of the length.

Conduction error is not controlled by thermowell length nearly as much as by the temperature of the pipe wall. The best way to control it is by insulating the pipe.

Radiation error, can be minimized by insulating your pipe, since it to is a function of the wall deficit -(process temp-wall temp). You can use finned wells and/or radiation baffles in clean fluids to reduce the radiation error, but then you have an entirely new set of design issues to consider.

The convective heat transfer coef. between the fluid and your sensor, is more than sufficient for the velocities you are describing. It is a problem only at very low pressures and low velocities.

The testing I've seen, suggests that conduction error is not a problem once the thermowell is greater than 2-3 tip dia. in length. In our cases, safety (personnel & equip.) tended to way heavily in our decisions to opt for conservative mechanical designs. We never saw any more than 1-2 Deg error even in the worse cases.

Thicker wells do not always translate into successful designs unless you are simply going for corrosion allowance. The larger the od, the greater the aerodynamic cross-section and drag forces, the lower the resonant frequencies; so in some sense increasing the od can acerbate the problems.

All of the severe service cases we looked at, length was the controlling factor, not diameter; and it was minimized subject to adequate tip placement. We abandoned length selection on the basis of pipe diameter, especailly where high velocities and rotating equipment were involved.

You can use longer wells in large ducts, but it is a tricky design issue and very dependent on the actual flow conditions. It is not generally advised.

If you ever walk up to a thermowell in a Class I, Div. 1 area and see flames playing out of the connection head, you realize how serious proper fitting selection can be. But it is an even bigger problem if you fill your conduit system with combustable gases...

The production hands get real concerned about such on blue water platforms; there are not many safe places to hide in a worse case scenario there. More than a few produces have recently upgraded their thermowell standards as a result.

Depending one your degree of criticality and the fluids being handled, you can make some decisions about your design priorities, but it pays to avoid high velocity applications where the thermowell is forced into resonance. The last estimate I came across was that about 30% of the thermowells in refinery services are mis-applied, i.e. exposed to severe vortex shedding stresses. They don't all fail, but the number that do, is pretty surprizing.

 

[Scotsinst]

Thanks again for your detailed response. I haven't looked up the links yet but I hope to get to it this week.
Unfortunately the Reactor Vapor Temperature is critical for control of the FCC.
The piping is refractory lined because of entrained catalyst.
The effects you name become more of a factor if the TW is close to the pipe wall and we are very hesitant to reduce the length.
The TW's are stellite coated and erosion is a concern. TW's in this service have failed due to erosion but none to my knowledge by resonant-frequency induced failure.
Cheers.

 

[jsummerfield]

I think that some FCC process licensors have thermowell designs specific to the reactor or regenerator area. I bought something very special with a builtup stellite barstock for an FCC in a Texaco plant. This was a long time ago.

John

 

[Scotsinst]

I hate to 'fess up but I work for an FCC process licensor and we do have a special TW construction for Catalyst service.
It is because a vendor/contractor came back to say that the design we have used for a long time is not suitable due to vibration that made me post. When I find time I will follow up on Hacksaw's recommended sites. I printed out the thermo sensors bulletin you indicated but I haven't had time to read it completely.
Thanks again.