Damage to HX with sudden loss of flow 1

[NovaStark]

Hi all,

I am trying to quantify the effect on a kettle type heat exchanger in the event that it loses shell side flow and if I need a device to ensure a minimum flow to the shell side. So the details are as follows:

Tube Side:

This side cools hydrogen gas from around 100F to 40F at about 200,000 lbs/hr and due to the cooling, water condenses out. This stream then goes to a knock out drum and then towards a compressor. Pressure is 500 psig. Design = 300F/585 psig

Shell side:

Uses refrigerant which facilitates cooling via latent heat (phase change) at 34F with about 20000 lbs/hr of refrigerant (6% vapor and rest is liquid). Pressure is 50 psig. Vapor at the outlet then goes to KO drum as well and then to a compressor. Design = 300F/225 psig.

Both tubes and shell are Carbon Steel (SA-179 and SA-516-70 respectively).

So now if I assume that there is a sudden loss of flow to the shell side, all the refrigerant boils out and I am just left with hot tube side gas going in and coming out. So this should just reduce my compressor's efficiency. I am not sure if components within the compressor can get damaged here, similarly to how hotter pump product can cause internal damages.

Just based on design pressures and temperatures, the heat exchanger should be okay but how would I quantify the overpressure if any occurs? I don't believe any should occur but I could be wrong as the shell side is solely protected by and RV on the KO drum which is sized for tube rupture on a different exchanger I believe.

The only issue I can foresee, is if flow is lost and you have the tubes hot, and you reintroduce colder product, thermal shock may occur. Not sure how to determine if it actually does occur given or say it won't cause a failure without FEA or similar.

Any ideas/suggestions/tips on how I can quantify these effects ? I guess the exercise is similar to a HAZOP study but I just don't know if my thoughts are sufficient to say that another lay of protection is needed.

 

[LittleInch]

Looks like two separate things here.

Not seeing the P & ID it's difficult, but not having a bespoke relief valve on the shell side looks very strange and unusual.

Introduction of cold fluid looks quite complex to me and probably needs to be looked at for a variety of cases, but then the boil out case would also have some tubes not in the liquid refrigerant and hence hotter than those at the bottom, thus introducing all sorts of impact on the tube sheet no?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[SnTMan]

From what I can understand of the situation, (and speaking only of the exchanger) I'd probably be most concerned about the tube-tubesheet joints, if the are expanded only.

Regards,

Mike

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[georgeverghese]

For the tubeside, sounds like the RV on the connected compressor suction side KOD could be used as credit for tubeside tube rupture if there are no restrictions in the relief stream flowpath.
You've quoted the upper design temps for tube and shellsides but not the lower design temp. What could be the lowest refrigerant temp ( probably lowest at startup / shutdown and depressure to zero psig). This should also be the lower design temp for the tubeside, and this LDT should be applied coincident with UDP (585psig) to see if there is risk of low temp brittle failure.

 

[NovaStark]

Looks like two separate things here.

Not seeing the P & ID it's difficult, but not having a bespoke relief valve on the shell side looks very strange and unusual.

Introduction of cold fluid looks quite complex to me and probably needs to be looked at for a variety of cases, but then the boil out case would also have some tubes not in the liquid refrigerant and hence hotter than those at the bottom, thus introducing all sorts of impact on the tube sheet no?

Yes it does sound that the tubesheet to tube joints would be negatively impacted but what could I used to quantify that or at least accurately/qualitatively say that there would be a negative impact ?

From what I can understand of the situation, (and speaking only of the exchanger) I'd probably be most concerned about the tube-tubesheet joints, if the are expanded only

I'll have to relook at the drawings to see how the tubes are expanded.

For the tubeside, sounds like the RV on the connected compressor suction side KOD could be used as credit for tubeside tube rupture if there are no restrictions in the relief stream flowpath.

You've quoted the upper design temps for tube and shellsides but not the lower design temp. What could be the lowest refrigerant temp ( probably lowest at startup / shutdown and depressure to zero psig). This should also be the lower design temp for the tubeside, and this LDT should be applied coincident with UDP (585psig) to see if there is risk of low temp brittle failure.

The MDMT for both the shell and tubes are -10 degF if I recall correctly.

I uploaded the relevant P&ID to give a clearer picture. So the loss of flow would be due to the control valve on the inlet to the suction chiller failing closed, so no flow. There is an RV on the KO vessel that it goes to on both tube and shell side however I don't have the detailed design calculation for these RVs to see which tube failures they are designed for. When you see the shellside flow path you'd see what I mean.


With respect to brittle fracture, I am only familiar with going through API 579 to do Brittle Fracture Assessment, this is what you're referring to correct ?


This entire exercise also assumes that the system stays online throughout the loss of low scenario. What may happen in actuality is that the compressor may trip on high suction temperature which would shut the systems off. So no hot side flow.

 

[georgeverghese]

The shellside PSV set at 225psig is on the 4th stage refrigerant flash drum 120CF4, but there is a check valve in the relief path. This would in some cases be considered sloppy design, but it may be permitted provided this check valve is classified as process safety critical and is included in the plant preventive maintenance schedule for equipment that must be inspected annually for proper operation. What type of check valve is this, and is it oriented horizontally at this high point in the piping?

The 3% of SP inlet pressure loss in the HP side of the relief path to this PSV would still be applicable, and this should include the dp across this check valve, the demister pad on the stage 4 flash drum vapor exit and other vessel nozzles. It should be checked if this demister pad can withstand the relief flow, which may be much higher than the normal operation refrigerant vapor flow leaving this stage 4 flash drum. The PID doesnt show a demister pad on the chiller 130 shellside exit nozzle??

Given that the refrigerant is ammonia, the lowest temp this syn gas compressor suction chiller 130 would experience should be approx -35degC (-31degF) during startup / shutdown operations, coincident with the UDP for each side of this HX, so this should be reflected in the MDMT or LDT for this unit.

Agreed, the refrigerant level in this kettle should always be above the top of the tubesheet ( tube bundle fully immersed in refrigerant ) so that differential thermal stresses in the tube bundle that can cause mechanical damage of the bundle are avoided. At the least, this low level trip ( LALL) can be configured off the existing shellside LIC on this chiller. The syn gas compressor should be tripped if level drops below LALL. Conversely, compressor startup should also be inhibited if level is below LALL.

 

[georgeverghese]

Thermal duty on this syn gas suction chiller show a syn gas stage 1 KOD TIC cascading on to a PIC which varies kettle refrigerant pressure / temp, and not by level control. So shellside level remains constant even as syn gas flow varies. So it should be possible to set up this LALL on this LIC without losing thermal duty control at this chiller.

Check that the impulse lines feeding the dp cell picking up level on the shellside are both remote sealed, else you'd get the wrong level readout at the LIC.

 

[LittleInch]

Great response George and I agree.

I still think the RV is too far away from vessel 130C for my liking and too many opportunities for blockage.

Unless the HX is specifically designed to accommodate a differential level then loss of level below the tube height should be a trip / shutdown event, IMHO.

And I would argue that in a HAZOP.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[georgeverghese]

Standard process safety practice would be to install at least one instrumented safeguard to address the tube rupture risk at chiller 130 (to complement the PSV on the stage 4 refrigerant flash drum vapor exit), so another trip loop would be to configure a high pressure trip ( set at say 180-190psig) off PIC1114 in DCS which controls kettle refrigerant pressure. This approach may not go down well during a formal SIL exercise since there is a risk of loss of containment associated with overpressure here (unlike the LALL which maintains mechanical integrity of this chiller and doesnt directly result in a loss of containment), so there is a risk some process safety puritan may insist on a dedicated high pressure trip transmitter for this safeguard.

 

[NovaStark]

Hi all,

The shellside PSV set at 225psig is on the 4th stage refrigerant flash drum 120CF4, but there is a check valve in the relief path. This would in some cases be considered sloppy design, but it may be permitted provided this check valve is classified as process safety critical and is included in the plant preventive maintenance schedule for equipment that must be inspected annually for proper operation. What type of check valve is this, and is it oriented horizontally at this high point in the piping?

It's a wafer type valve and it is horizontally mounted at the high point. Don't believe that we have it as a critical check valve but I can always recommend this.

The 3% of SP inlet pressure loss in the HP side of the relief path to this PSV would still be applicable, and this should include the dp across this check valve, the demister pad on the stage 4 flash drum vapor exit and other vessel nozzles. It should be checked if this demister pad can withstand the relief flow, which may be much higher than the normal operation refrigerant vapor flow leaving this stage 4 flash drum. The PID doesnt show a demister pad on the chiller 130 shellside exit nozzle??

Just to make sure that I have the same terminology as you, SP means set pressure correct? How exactly would I go about checking to see what DP a check valve can withstand without physically checking it? The specs might give me the rating but not what the seat can take before yielding etc. Is there a simplified way to do this? Also the drawings for 130 don't show a demister but I am guessing that the in 120C the demister pad there is sized for this shell side flow (assumption)

Given that the refrigerant is ammonia, the lowest temp this syn gas compressor suction chiller 130 would experience should be approx -35degC (-31degF) during startup / shutdown operations, coincident with the UDP for each side of this HX, so this should be reflected in the MDMT or LDT for this unit.

I double checked with my oem dwg which does show 10 degF MDMT for the tubes and shell. I'll have to check to see what conditions the unit is started in but I do agree that it should be -31 degF. So I do find it strange.

Agreed, the refrigerant level in this kettle should always be above the top of the tubesheet ( tube bundle fully immersed in refrigerant ) so that differential thermal stresses in the tube bundle that can cause mechanical damage of the bundle are avoided. At the least, this low level trip ( LALL) can be configured off the existing shellside LIC on this chiller. The syn gas compressor should be tripped if level drops below LALL. Conversely, compressor startup should also be inhibited if level is below LALL.

I was also thinking of this. But I'd have to request the HAZOP to see why it wasn't implemented in the first place.

Great response George and I agree.

I still think the RV is too far away from vessel 130C for my liking and too many opportunities for blockage.

Unless the HX is specifically designed to accommodate a differential level then loss of level below the tube height should be a trip / shutdown event, IMHO.

And I would argue that in a HAZOP.

I didn't see anything in the mechanical calculations for 130 to say that it is designed for diff level and subsequent DT on the tubesheet. So I doubt it's designed with that in mind.

Standard process safety practice would be to install at least one instrumented safeguard to address the tube rupture risk at chiller 130 (to complement the PSV on the stage 4 refrigerant flash drum vapor exit), so another trip loop would be to configure a high pressure trip ( set at say 180-190psig) off PIC1114 in DCS which controls kettle refrigerant pressure. This approach may not go down well during a formal SIL exercise since there is a risk of loss of containment associated with overpressure here (unlike the LALL which maintains mechanical integrity of this chiller and doesnt directly result in a loss of containment), so there is a risk some process safety puritan may insist on a dedicated high pressure trip transmitter for this safeguard.

Loss of containment in the sense that once the RV lifts on 120C it goes to the vent correct whereas trip via low level just shuts the system down?

You've given me a lot of different perspective on this George, far more than I thought!

 

[georgeverghese]

Agreed that wafer style dual plate check valve should have been classified as process safety critical.

SP = set pressure. dp across check valve is the frictional pressure drop as a result of gas flow - best to get these off the manufacturer's flow charts - in this case, we'd be using the relief flow.

Have my doubts if that demister on the stage 4 flash drum vapor exit nozzle is meant to mechanically withstand tube rupture relief flow. Agree with LI there seem to be just too many restrictions / mechanical failure risks in this relief flow path, so this arrangement should have been justified with an inlet pressure loss calc which should be somewhere in the plant designer's process engineering calcs.

For a lower design temp of -31degF, ideally you'd be using A333, fine grained killed carbon steel impact tested at -31degF or lower.

Suspect TIC1363 (which cascades onto PIC1114) may be a poor actor in this control loop unless special features are enabled at the field installed temp transmitter. Temp sensors show considerable response lag and this results in a slow acting TIC which then later overeacts to compensate - very often, TIC loops show erratic and oscillating output signals as a result. This may be the reason why you have these large level- pressure disturbances at this chiller 130 ?
To improve temp transmitter response, suggest the following (your instrument engineer may have more ideas)
a)Install the TT / thermowell in the piping at a location where there is high fluid turbulence, like facing an elbow head on - the PID seems to show such an arrangement on the syn gas KOD dwg, but check if this is actually how it is set up in the plant.
b)Use thermocouples instead of RTD type TT and keep the measured range as narrow as possible to increase sensitivity.
c)Replace thick wall thermowell with thin wall finned thermowell
d)If you have larger flow disturbances in syn gas flow at this chiller, consider taking the risk of snipping off the tip of the thermowell and have the sensor probe tip naked in the flow stream.

 

[NovaStark]

SP = set pressure. dp across check valve is the frictional pressure drop as a result of gas flow - best to get these off the manufacturer's flow charts - in this case, we'd be using the relief flow.

I'll have to request from the manufacturer. Once my calculated DP > OEM specified DP, it's undesirable. Got it.

Have my doubts if that demister on the stage 4 flash drum vapor exit nozzle is meant to mechanically withstand tube rupture relief flow. Agree with LI there seem to be just too many restrictions / mechanical failure risks in this relief flow path, so this arrangement should have been justified with an inlet pressure loss calc which should be somewhere in the plant designer's process engineering calcs.

Hopefully this is in our documentation but I doubt it as usually the plant designer did not give us much about the design of certain components.

For a lower design temp of -31degF, ideally you'd be using A333, fine grained killed carbon steel impact tested at -31degF or lower.

Agreed since most of the ammonia piping is A333.

Suspect TIC1363 (which cascades onto PIC1114) may be a poor actor in this control loop unless special features are enabled at the field installed temp transmitter. Temp sensors show considerable response lag and this results in a slow acting TIC which then later overeacts to compensate - very often, TIC loops show erratic and oscillating output signals as a result. This may be the reason why you have these large level- pressure disturbances at this chiller 130 ?
To improve temp transmitter response, suggest the following (your instrument engineer may have more ideas)
a)Install the TT / thermowell in the piping at a location where there is high fluid turbulence, like facing an elbow head on - the PID seems to show such an arrangement on the syn gas KOD dwg, but check if this is actually how it is set up in the plant.
b)Use thermocouples instead of RTD type TT and keep the measured range as narrow as possible to increase sensitivity.
c)Replace thick wall thermowell with thin wall finned thermowell
d)If you have larger flow disturbances in syn gas flow at this chiller, consider taking the risk of snipping off the tip of the thermowell and have the sensor probe tip naked in the flow stream.

We've never really had any pressure disturbances in this vessel as far as I know. This query came out of a hazop revalidation study.

 

[georgeverghese]

Well you're lucky there with this TIC cascade loop then - I seem to run into problems with TICs' just about every time I face these.

 

[NovaStark]

I'll keep that in mind. Much appreciated!