Pipe Temperature Drop Due to Cold Fluid 1

[chindy]

Hi,

I'm doing work on a system to determine a control loop response time to protect a recovery skid. The scenario is low temperature tripping the skid, diverting the flow to the main flare. I need to find how quickly the downstream pipe (not designed for low temperature) will cool to its temperature limit of -20 F. The following are the assumptions:

1. The system is initially at equilibrium of 80 F (pipe temperature is the same as the fluid temperature).
2. A sudden rush of cold fluid displaces the warmer fluid, so that the pipe contents are now at temperature T.
3. There is no external heating (air convection or solar).
4. The flow rate of the cold fluid is constant.
5. It's a large pipe, so the fluid temperature is constant.

This setup assumes the temperature change is sudden, so the TT will instantly trigger the alarm. The key is finding the maximum time I need to design for between when that alarm triggers and the downstream system is tripped, which is based on how long it will take for the pipe to cool from 80 F to -20 F. I've been having difficulty determining how long it will take for the pipe to cool, however. I can say that for a representative foot of pipe of "m" mass, m*(heat capacity of pipe)*(change in temp of pipe) = Q, the total Btu's needed to cool that one stretch of pipe. However, for the ethylene side, how do I determine the time needed to absorb that amount of heat Q? I've done a few different things with wildly different answers and don't think any of my methods are actually sensible.

 

[pmover]

a simple diagram of the system process with fluid temps will aid others in providing responses.

 

[georgeverghese]

Process control loops set up in an attempt to avoid low temperature brittle failure risk are not accepted as a process safety measure in most oil/gas companies. It may not pass a process safety audit / HAZOP.

 

[shvet]

Risk of brittle failure is a common case in liquified gases industry for example in N2 evaporators. I came across 1oo3 trip in this case.

This setup assumes the temperature change is sudden

Discourse around "cooling time" is not correct because of short time local overstress caused by rapid cooling of pipe segments. In your case it is not guaranteed that local stress has enough time to dissipate.

https://www.youtube.com/watch?v=2sPStGlFOts

 

[LittleInch]

Well this is a transient thing so not easy to calculate manually.

If the pipe is big and the length relatively short and a decent flow velocity (> 3m/sec), then the cold gas will probably loose very little of its temperature from one end to the other.

The mass of the pipe is not big and hence the amount of heat energy not large.

I would work on something like an internal heat transfer rate of 150 to 200 W/m2/K.

You don't say what the temp of the gas is?

But I think you're talking tens of seconds if the gas is something like -50C or lower.

But George is correct as usual, this is simply a poor design and open to challenge in safety reviews. It has been a noted cause of failures and fatalities in a number of locations and is just not the way to design something.

Plus, if you stop this flow when the pipe has started to cool with very cold gas in, it will continue to cool and potentially crack / rupture.

I think you're not approaching this whole scenario in the right way. IMHO.

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

 

[pierreick]

HI,
Thanks to share with us a PID or PFD with relevant process data.
The concern is certainly the resistance of the line, you may have a concern on the operation side (freezing for example).
Did you check the hazop study, this case should be part of utility failure and consequences?
Pierre

 

[shvet]

for info
Shock chilling should be considered as temperature difference is 100F. See para. 3.3.3.e API 579-1-2016.

 

[chindy]

Hi all,

Thanks for all the really great responses. I had left out most of the numbers since I was trying to avoid just getting an automatic answer and was hoping to better understand more on the approach of this type of system, but I'll provide some more details to help drive discussion. I also added a simple flowsheet to outline the system.

The system is a flare header designed with low temperature carbon steel (in this particular case, -150 F/-101 C). In an attempt to reduce flaring, a flare gas recovery system is being installed. The piping spec for the vendor skid is standard, vacuum-rated carbon steel, but it is not low-temp (min temp of -20 F / -29 C). There's a system of valves that either direct the flare header to the flare (emergency or startup operation) or to the recovery skid (normal operation). The LOPA scenario is concerned with low temperature flare gas (-134 F) embrittling the skid/associated piping, as the flare gas can be quite cold during certain upsets. I'm assuming uniform temp of the flare gas because of a high velocity and large diameter piping. I haven't done any calcs on how much the gas might heat up before reaching the section of pipe in question, but I imagine it won't be a significant amount.

I'm taking over from previous, leveraged work and using their rationale. They tried to pick a worst-worst case to analyze how much time our system would have to react to prevent a process upset. Some of my main concerns with the scenario were:

1. Isn't it an issue as soon as any part of the regular CS pipe sees the cold fluid, not just once the pipe is uniformly below it's spec? This would make a major difference in the time to respond.
2. LittleInch brought this up as well, but if the system isn't evacuated and isolated once the cold fluid is in there, then it will continue to cool anyways and the risk of pipe damage, while slightly reduced, is not eliminated.

 

[georgeverghese]

Ok, so this is a trip loop whose process safety time to close the recovery skid shutdown valve is required. Also ask the recovery compressor vendor what the max permissible chill rate is (ie dT/dt in degF per sec) may be for all T less than 80degF. Also note the TT trip sensor has its own response lag with the thermowell it would be sitting in. Chilling would be fastest with smaller line sizes, and any inline small bore piping / tubing, since thermal inertia is less and gas velocity is higher leading to higher heat transfer rate to the pipe or inline tubing.
An air operated single acting actuated 18inch SDV would typically close in 10-15seconds. The trip TT standard thermowell may have a response lag of say 10seconds. That sums up to some 20-25seconds of cold gas at < -20degF past the 18inch SDV into the recovery skid.
By the way, is the recovery compressor suction side mechanical design pressure > max backpressure that may be seen on the main flare header ( assuming the control PCV to flare has failed close or has some poor control response)?
All heavy chemical components in this flare gas (in all operating scenarios) must be immiscible in the lube oil used at this compressor, else oil viscosity will be otherwise compromised. In a naphtha based cracker, there would be a wide range of these heavy components in the flare gas at startup.

 

[SJones]

Your exercise is fitness for service (FFS) of the vendor skid, or the piping leading to it. To that end you will need to refer to API 579, as advised by shvet a few posts above. Also noting, as directed by shvet, the implications of shock chilling. In addition, you will need to take a close look at the materials of the skid. The mention of -29 deg C instantly smacks of A106 GrB, A105 etc. Quite a few operators would not permit such a low MDT for these materials owing to a lack of provenance, and poor fracture performance even at temperatures above 0 deg C. Good luck, and give the person who ordered up the skid a good talking to.

Steve Jones
Corrosion Management Consultant

http://www.linkedin.com/in/drstevejones

All answers are personal opinions only and are in no way connected with any employer.

 

[georgeverghese]

@SJones, Agreed, A333 Gr 6 (impact tested at pipe mill) would have been a better choice for lower design temp of -20degF. And -100degC lower design temp probably leads one to suspect 3 1/2 nickel steel is the current selection, given this is in the US, which has gone out of fashion for many years now in Europe, ME and APAC.

 

[shvet]

@chindy
Thermowell response time seems the major problem as cooling rate depends on temperature difference (trip minus fluid) and response time can become very high.

Also stratification effect should be considered:
- relief gas -134°F might become much denser than recovered gas
- 48" of pipe is high enough
Relief flowrate could be not sufficient to provide uniform velocity profile => gas can stratify => there is a risk of thermowell will not contact cold flow. Was TT thermowell installed upright of pipe? Stratification is hard to predict as I know (correct me please if I am not right).

We provided dynamic modelling in Ansys for such transient relief cases. It is not so difficult to find a person/contractor for such kind of modelling - modern software offer you can combine flow, mixing, thermo and stress calculation in one dynamic model so human factor is the only issue. It is not so expensive, complicated and long-lasting as seems.

Anyway only one issue has a critical impact on a result - you are the engineer, will you totally believe in such modelling /calculations (any kind of - software or manual) based on anonym forum or handbook? You personally? What will be accuracy and confidence probability of such calculations/modelling? What are boundaries of not widely accepted and proven practices? What is your own assessment of failure risk? What would be consequences of failures and are you ready to share responsibility of those?

 

[LittleInch]

In response to chindy above.

Yes /I believe that no part of the system, whether 1m long or 100m long should be exposed to temperatures below its design temperature.

As George says, small bore connection and fitting are notorious for getting very cold very fast and then failing.

Other issues can include the stratification and uneven bending forces as one part of the pipe cools faster than the other.

I've just done a real simple calculation to see what we're looking at and my answer tells me that if you drop that really cold gas in, you have about 1 to 2 minutes before you drop below the DT.

That's not a lot of time for all the protective devices to recognise the issue and respond accordingly. I suspect you're heading towards a SIL 2 or SIL 3 given that flare lines are normally pretty thin and if they break can lead to catastrophic damage in a plant.

Given that this is gas, I don't think having cold gas in there will drop the temp when not flowing by more than a degree or two. But constant flow will chill it really very rapidly indeed. but these things can be simulated in the right program such as CFD analysis and give you a great colour plot to show the managers...

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