6 replies to this topic

[Olaniyi]

Hi colleagues

I have a quick question. I am looking at a process safeguarding memorandum, specifically the relief device on a natural gas heater (the heated gas is used as defrost gas). The heating medium is an electrical heating element.

 

One of the contributing cases is stated as "relief due to blocking in of the exchanger while the temperature elements are still heating, causes the defrost gas temperature to rise, resulting in thermal relief." (No credit is taken for the functioning of high temperature trips installed).

 

My question: How would you go about calculating a relief load for this? I know formulae (from ISO23251) for calculating relief load due to external heat input (from fire) for both gas and liquid service.

However I cant seem to find anything for or analogous to internal heating in a blocked heat exchanger/vessel in gas service, and I am struggling to figure out how I would estimate the relief load from this case i.e. heating element is on, exchanger is gas filled and blocked in at (I assume) at the operating pressure, which is ~50 bara. The relief valve set pressure is set at 60 bar (with 10% overpressure allowed)...

 

Thanks for any help!

 

Regards

Ogeds

[fallah]

Ogdes,

 

Please upload a simple sketch of the system...

 

Please confirm both isolating block valves on the inlet/outlet lines of natural gas to/from the exchanger are closed in the blocked in condition you described...

Edited by fallah, 15 October 2013 - 03:06 PM.

[Lowflo]

1. Determine the maximum allowable temperature for this piping/equipment.
2. At the relieving pressure, generate a table of values for T versus vapor density (using physical properties program, or a simulation program).
3. On the table, locate the maximum allowable T which was determined in #1, and read the density at that point.
4. The relief flowrate is the determined based on the difference between the initial density and final density (determined in #3), and the volume of the system.

Note however that this PSV won't protect the vessel if the system remains blocked in while the vapor continues to be heated (T continues to rise). Such a  case is analogous to a vapor-filled vessel exposed to fire. A PSV can't stop T from rising. The vessel will eventually reach the yield stress (fail) due to the rising temperature.

 

 Unless the final T is self-limiting, without exceeding the allowable T for the system, then I'd use a HIPPS that shuts off the heat source before it causes overpressure. That's a true protective layer. A PSV only provides temporary protection from this scenario.

[Olaniyi]

Thanks all.

 

@Fallah, attached a simple sketch below. Yes I believe the situation is that both blocked valves are closed - again I am relying on what I feel is a reasonable interpretation of the wording in the PSM which I quoted above - i.e. blocked in = inlet and outlets closed.

 

@Lowflo, thanks - really useful comments! Especially your thoughts on the PSV - I am aware that for a fire case in a vapour filld equipment you would likely have steel failure first before any relief from the PSV, but thought there might be a difference simply from the fact that the heat source is internalthus there is more direct heat transfer tothe gas...

 

Ogeds

 

 

 

Attached Files

•  sketch.docx   46.26KB   62 downloads

[fallah]

ogeds,

 

Thanks for sending the sketch...

 

I think such case isn't absolutely analogous to a vapor -filled vessel exposed to an external fire, because heat source is internal and due to direct heat transfer to the gas the vessel wall temperature couldn't be higher than the gas bulk temperature. Then the relieving temperature in such cases wouldn't so be higher than operating temperature and the PSV would certainly pops up before vessel failure.

 

Relief load would roughly be obtained by balancing the heating rate of heat element to the trapped gas and the heat rate outgoing through the exchanger's PRV by internal energy of the gas flowrate at relieving conditions... 

[Lowflo]

Of course, PSVs will only respond to pressure. That limits their effectiveness in protecting vapor filled vessels from scenarios which create temperature increases. If that temperature increase is caused by something other than fire, then the peak temperature may not be high enough to cause vessel failure. Fire, by comparison, is always hot enough to cause vessel failure. That's the difference between this case and a fire case.

 

Depending on the type of heater, the peak temperature may or may not be high enough to overstress the vessel. Use the combination, peak temperature and relief pressure to determine that stress.

 

If the peak temp at the relief P creates too much stress for the vessel, then the PSV isn't an adequate layer of protection. Remember, the temperature will continue to rise - the PSV does nothing to mitigate that. The PSV may open, but it will quickly re-close, while the temperature continues to rise higher and higher.

 

The best way to mitigate this risk is to turn off the heater when the process side temperature exceeds a certain temperature. That's called a HIPPS.

[Olaniyi]

Thanks all

Attached is the high level approach I have taken so far.. would be curious if you think it makes some sense...

 

1. Calculate volume of exchanger Vo

2. Calculate volume of internal tubes (Vi)

3. Volume available for gas Vg = Vo - Vi

4. Find density of gas based on P and T in the exchanger during normal operation = rho_i

5. Find rupture temperature of the exchanger material (from Plant Mechanical Engineers) as the max temp that can be achieved.

6. Assume the exchanger is at relieving pressure at this time (add allowable overpressure as used in formulae for fire case calculations as well)

7. Find the density rho_f at these new P and T conditions

8. Based on the volume of the exchanger find the mass Mg_o that we need to get rid off to maintain the volume at this new density = (rho_i - rho_f)*Vg

9. To calculate the the time to reach the rupture temperature, use t (secs) = m (kg) x Cp (KJ/kg/K) x delta T (K) divided by Q (KJ/s), where Q is the design duty of the exchanger in KJ/s

10. I get t in secs = 142 secs

11. My relief rate is then Mg_o divided by t which comes to about 0.226 kg/s.

 

Thoughts? Note that at this point this is more for my learning than anything - was just curious about this case from looking into the PSM so please feel free to critique heavily.

 

Thanks

Ogeds