18 replies to this topic

[Narnia]

We have a continuous polymerization process where the heating of entire process is done thru Dow therminol vapors.

Two dedicated heaters (NG and HFO fired) each equipped with a flash tank at the vapor outlet line sustains the entire heat load of the plant. . The vapors at the outlet of the each heater pass through the flash tank before going to the main process. Now each flash tanks has a PSV and a Rupture disc. The rupture disc is installed at the upstream of the PSV and has a very notorious history of bursting without any pressure surges in the system..

Would any one please let me know the reasons of that happening.

Kindly find below the operating parameters and suggest if a change in rupture disc design would rectify the problem


Media in flash tank: Diphenyl oxide/ Diphenyl (C12H10O) (present in Vapor phase)

Operating Pressure: 54~55 Psi
Max Operating Pressure: 57~58 Psi

Operating Temp: 364~365 C^o
Max operating temp: 367 C^o

Lifespan of BD: 15 ~30 days


I suppose the above would be sufficient for assessment.

[fallah]

Narnia,

Seems your information not to be sufficient, then provide the below:

- Is there a gauge pressure between RD and the PSV?
- What are the set pressure values of RD and PSV?
- Material of RD?
- The way you would recognize the RD has been ruptured?
- Is the RD specified (guaranteed by relevant vendor) for the relatively high operating temperature (367 C)?

Indeed, the margins between operating pressure/temperature and maximum pressure/temperature are very low, need to be reviewed.

Fallah

[Narnia]

Dear Fallah,

Thanks for reviewing

Below are the required information’s

Spec of RD

RP 75 psi @ 385 Deg C
72 psi @ 20 deg C

Tolerance +/- 5

Material of construction: Nickel

The trends were checked but no pressure pulses or level fluctuation were found.
The pressure transmitter on the flash tank is the only pressure indicator which shows the running pressure and that’s the pressure at u/s of the BD as well.
There are pressure indicators installed at downstream of the BDs, which start indicating pressure whenever the BD bursts.
The normal level 45 ~55 %. The level fluctuate between 5~ 10 % which is normal.
DOW circulation pump discharge pressure 8.5 kg/cm²
Condensate feed tank pump discharge pressure 4.3 ~ 7.8 kg/cm²

Would revert BACK after reviewing the requested information.

Looking for very prompt response.
Thanks

[GS81Process]

Rupture discs are known to sometimes burst prematurely. However, please check the following things:

Has there been any solid buildup observed on the downstream side of the bursting disc?

Have you compared the measured disc thickness with the specification to ensure you are not seeing corrosion?

Is there any visual evidence of disc corrosion?
Can you confim that the stamped burst pressure (and tolerance) match the datasheet?
Do you drain/vent the space between the rupture disc and the PSV?
Is the pressure transmitter located between bursting disc and PSV? If so, this is proper.

Edited by GS81Process, 25 November 2011 - 12:12 PM.

[Narnia]

Below are replies to your queries
1. Has there been any solid buildup observed on the downstream side of the bursting disc?
No
2. Have you compared the measured disc thickness with the specification to ensure you are not seeing corrosion?
Yes, these are same
3. Is there any visual evidence of disc corrosion?
No
4. Can you confirm that the stamped burst pressure (and tolerance) match the datasheet?
Yes
5. Do you drain/vent the space between the rupture disc and the PSV?
I don’t understand this. There is no vent /drain points between these two.
6. Is the pressure transmitter located between bursting disc and PSV? If so, this is proper.
Yes there is a pressure gauge working properly.
Looking forward for prompt response.

[fallah]

'Narnia', on 25 Nov 2011 - 12:05 PM, said:

Below are the required information’s

Spec of RD

RP 75 psi @ 385 Deg C
72 psi @ 20 deg C

Tolerance +/- 5

Material of construction: Nickel

Narnia,

Set pressure of the PSV?

Fallah

Edited by fallah, 26 November 2011 - 04:06 AM.

[Narnia]

Fallah,

Set pressure of PSV is same as that of RD

[Narnia]

But note that PSV is not popping ... only RD damage

[fallah]

'Narnia', on 26 Nov 2011 - 09:36 AM, said:

But note that PSV is not popping ... only RD damage

Narnia,

It could just be said that the RD selection has not properly been done. Anyway, please provide below information for selected RD may help to find the reason of bursting:

-Disc type (Forward Acting, Reverse Acting,...)
-Manufacturing Design Range
-Specified Burst Pressure/Temperature
-Max/Min Marked Burst Pressure
-Operating Ratio

Fallah

[GS81Process]

'Narnia', on 26 Nov 2011 - 08:05 AM, said:

Below are replies to your queries
1. Has there been any solid buildup observed on the downstream side of the bursting disc?
No
2. Have you compared the measured disc thickness with the specification to ensure you are not seeing corrosion?
Yes, these are same
3. Is there any visual evidence of disc corrosion?
No
4. Can you confirm that the stamped burst pressure (and tolerance) match the datasheet?
Yes
5. Do you drain/vent the space between the rupture disc and the PSV?
I don’t understand this. There is no vent /drain points between these two.
6. Is the pressure transmitter located between bursting disc and PSV? If so, this is proper.
Yes there is a pressure gauge working properly.
Looking forward for prompt response.

I asked about venting/draining of interstitial space between the RD and PSV in case of liquid buildup based on piping design. If this is not an issue then disregard.

My experience with phenols is that they are known to sublime. I have observed in several instances buildup of a white, waxy substance on relief devices in phenolic service. But you indicated that this is not a problem.

You state that there is a pressure gauge between the RD and PSV. How often was this checked? Can you be assured that a pinhole didn't form in the RD and go unnoticed? The preferred installation is to have a pressure transmitter with DCS signal back to the control room, then you can look back on the data.

Have you calculated the theoretical burst pressure of the RD based on UTS or yield strength of nickel and thickness? Does it match? If so, can you send a portion of the disc to lab to confirm nickel composition and perform failure analysis?

[Narnia]

can you please confirm me that Ni material Rupture disk is compatible for above mentioned service. If yes can you provide me some information about their compatibility.

Thanks

[GS81Process]

Unfortunately I cannot confirm the suitability of NI in this service. I am not entirely familiar with your process application. Do you work with someone who could assist in this regard?

[breizh]

Hi ,
Did you talk to Dow or Dowtherm supplier , they should be able to comment and assist?

Can thermal degradation generate a byproduct responsible to damage the RD?

Breizh

[GS81Process]

Breizh has made a good suggestion in contacting Dowtherm supplier to determine if nickel is appropriate for the rupture disk. Probably also valuable for you to confirm via material test that the disk was actually nickel as I suggested earlier. Given that you have ruptured several discs though I suspect it is nickel.

I would also recommend that you check the design and installation of the rupture disc holder to make sure that undue stress is not being imposed on the disk.

Edited by GS81Process, 02 December 2011 - 12:38 PM.

[kkala]

Some relevant notes, asterisk (*) means reference to Cl. Matthews, "Pressure Relief Valves, A quick guide", Professional Engineering Publishing Ltd, 2004.
1*. Creep is not a likely cause of frequent bursting, it could have effect after 12 -18 months.
2. Since Ni yield strength decreases with temperature, one would expect burst pressure to be 75 psig @20 oC and 72 psig at 385 oC. Tolerance is +/- 5% or +/- 5 psi ? But no influence on the issue.
3*. Bursting pressure should be slightly lower than upstream PSV set pressure (here it is same as PSV set pressure). But no influence on the issue.
4. Max operating pressure = 58 psig, max operating temperature 367 oC. Margins adequately far from PSV relieving conditions. PSV is assumed not to open at all, when the disk bursts.
5. Previous posts indicate that cause of bursting disk failure may be rear / unusual, probably such as below.
5.1*. Fatigue (rear). Check for mechanical vibrations on the pipe, near bursting disk.
5.2. Vendor has not properly considered the bursting pressure at 385 oC. Try bursting disk from another vendor.
5.3. Undetected corrosion on the disk. Matter under investigation by other posts.
Having seen rupture disk bursting in ammonia drum (fertilizer company), reason was pressure variations in NH3 handling. Hoping present issue will be solved.

Edited on Dec 04

Edited by kkala, 04 December 2011 - 02:27 PM.

[GS81Process]

kkala point regarding vibrations is worth a consideration. Fatigue to the disk can also be caused by improper design of the disc holder.

[acer_asd]

Dear Narnia
In our company whenever we buy RDs we always buy one extra RD and test it by bursting it at reference conditions. The RD for testing is picked up randomly. Do you do that? As far as i know this is standard practice. It is possible that this RD is not designed for conditions given in datasheets. I suggest you to do this exercise if you are not doing it already.

Hope this helps you in finding the root cause of your problem

Edited by acer_asd, 07 December 2011 - 11:34 PM.

[Narnia]

Would you please elaborate the method for testing rupture disk at reference conditions,
what are reference conditions & how these will be achieved.

Thanks

[kkala]

Having worked in a fertilizer plant (1975-81), I have the impression that each bursting disk (for an ammonia drum) arrived with a stamped bursting pressure, a bit different from piece to piece (range probably +/- 5% ). There were complains that their actual bursting pressure was too variable, but I think sudden instantaneous increase of NH3 (gas) pressure during operation ( plus some hammering) caused the trouble.
http://www.iceweb.co...ecification.htm indicates that +/- 5% bursting tolerance is over the manufacturing range, which can be zero. It also reports
"The testing that is done in the factory determines the actual burst pressure of the batch. If the customer nominates ASME VIII certification, 2 tests are made in an oven at the customer’s coincident burst pressure. The average of these tests must lie in the manufacturing range and is stamped on the disk tab in accordance with ASME VIII. If ASME certification is needed, then the stamping on the disk tab cannot vary".
Consequently tests at the temperature specified for rupture (coincident temperature) can be made in Vendor's shop.
Recommended range of operating pressure for several types of rupture disks is also reported.
I suppose that water or air is used for above tests.