HIPS / HIPPS 7

[GreenHawk]

All,

I'm currently working on reducing the relief load from a distillation column to the flare system by installation of a HIPS (or HIPPS) on the steam valve to the column. My understanding was that after installation of HIPS, I have the replace the pressure safety valve by a smaller one which will take of other over-pressure scenarios. However, I have been told by someone lately that according to guidelines of a major US chemical company, it's not required to replace the safety valve with a smaller one and we just have to keep using the existing PSV. My argument is when this existing PSV will open after installation of HIPS, it will release the same relief load (as was before installation of HIPS) as it is a function of the orifice of PSV. Can anyone guide me what is the good engineering practice in this case?

Additionally, if we don't have a smaller nozzle on the column, can we install the smaller PSV on the bigger nozzle by using a reducer?

I will appreciate your comments on these questions.

Thanks,

 

[TD2K]

Is this a new column you are currently designing or an existing column that you are retrofitting with a HIPS system?

I don't see a need to automatically install a smaller PSV on the column. Your existing one is over-sized so it would be subject to chatter if a relief event occurred. However, in addition to the sizing case, there are usually many other ways of reaching relief pressure on a column which are frequently much smaller than the sizing case so these cases are also subject to chatter (you could design a large and small relief valve to cover the various cases but I've not typically seen this). I suppose for a case like this I would be trying to quantify how often this relief valve might be called on to protect the column against overpressure. Relief valves should not be relieving very frequently but is this say once every couple of years or a once in 100 year event?

There's definitely no problem using an oversized tower nozzle for a PSV. The criteria for the sizing of the inlet line is to keep inlet line losses under 3%, a bigger line just makes that pressure drop using the relief valve's rated capacity smaller.

 

[GreenHawk]

TD2K,

Thanks for your reply. This is an existing column which I'm retrofitting with a HIPS system. The intention is to create room for additional relief load from a plant expansion, which would otherwise require an expensive expansion of the relief system.

Can you also guide me that when I model this flare system with multiple relief valves, do I have to use the sizing flow rate of PSVs (i.e., the relief flow rate demanded by the credible relief scenario) or do I have to use the rated flow rate of PSV which is a function of PSV's orifice?

Regards,

 

[TD2K]

Flare networks are usually evaluated using the required relief rate, not the relief valve's rated capacity.

 

[Latexman]

It seems my company has a different standard. Back pressure constraints of relief networks are evaluated using the rated flow rates at 10% overpressure. The sizing flow rate can be used for PRVs with known modulating behavior, if each modulating PRV meets its back pressure constraints when evaluated by itself in the network using the rated flow rate at 10% overpressure.

GreenHawk, I'd recommend you determine your company's policy. It may be dictated by your insurance carrier.

Good luck,
Latexman

 

[don1980]

The industry standard guidance for sizing/rating flare headers is found in API 521. There's an extensive section explaining how to size/model flare headers, and it's easy to read and understand.

Tailpipes are sized for rated flowrate, but, by default, sub-headers and headers are sized based on sum of the required flowrate from each source.

 

[Latexman]

Thanks don1980! Looks like I have to influence a change in our policy.

Good luck,
Latexman

 

[don1980]

One follow-up comment.....the rational for using the req'd flowrate for sizing flare headers makes good sense in most cases, but you need to apply judgment for sub-headers that are fed by just a copule of sources.

The logic for using the sum of req'd flowrate is three-fold: (1) PRVs will be cycling open and closed (all properly sized PRVs have some amount of excess capacity), (2) different vessels will reach their set pressure at different times, and (3) the duration of relief flow will vary from one vessel to amother (small vessels might be over and done before some larger vessels even start).

So when you think about it, it's way too conservative to say that all the affected PRVs will be open and flowing at the same time, and it's easy to recognize that the sum of the req'd flowrates is a much more realistic basis. But, notice that this logic is based on there being a significant number of sources (PRVs). Sometimes we have a sub-header that only has two or three source loads. If so, consider whether or not the general guidance (sum of req'd flowrates) is applicable for that specific subheader. If the source vessels have a similar size and similar contents, then perhaps it's reasonable to say that each PRV is open at the same time.

So, keep in mind that the API 521 guidance is general guidance. It's perfectly safe for most flare headers, but use judgment when you have atypical circumstances (e.g. hdr fed by only 2-3 sources).

 

[123engr]

There are multiple PSVs discharge connected to a common flare header and each with its different relief load. If the flare header size is to be determined, should the total relief load of all the connected PSVs be used? I asked this question based on your comment that " it's way too conservative to say that all the affected PRVs will be open and flowing at the same time".

I am what I am by His grace

 

[Latexman]

It depends. You have to go credible scenario by credible scenario to determine the max. credible relief load. It is not a simple task. It depends on layout, process, safety systems, emergency response, etc.

Good luck,
Latexman

 

[don1980]

123engr,

Refer to API 521 for a complete explanation.

As latexman says, it depends on your scenarios. Flare hdrs are sized for what is called "global scenarios". Those are facility scenarios (e.g. fire, loss of cooling, power failure) that results in multiple PSVs blowing at the same time. So, you can't size the flare hdr until you tabulte the sources (mostly PSVs but can include some emergency dump valves that are triggered during an emergency) and listing the applicable scenarios. Generally, there are only two global scenarios: fire and loss of cooling. There are exceptions, but power failure and loss of cooling are typically the same thing (not independent), because loss of cooling is usually initiated by a power failure.

Identify the sections of the plant that can experience a global scenario. For fire, these are called fire zones. It's highly unlikely that a fire will involve the whole plant. Fire zone are bounded by trenches, curbs, dike wall, peak surface elevations, etc. This is largely a subjective evaluation. API 521 has guidance for the size of a typical fire zone. Some companies lay out fire zones by drawing fixed-sized fire zones, using just the API size. Personally, I think that's too arbitrary. I suggest drawing those zones based on the drainage patterns and the boundaries mentioned above. It's generally an iterative process. Sketch up an initial suggestion on a plot plan and discuss it with others. Look at the size of the indivisual fire zones and compare their size to the API guidance. In my experience, engineers typically draw first-pass fire zone that are much larger than the default API guidance. Look at the zones that are significantly larger than the API suggestion and see if there's justification for making them smaller. Usually, I find that we settle on a size somewhere between the one that was originally drawn and the default size suggested by API.

Do the same for loss of cooling. That's usually a more objective analysis, as compare to fire. In many cases, it's reasonable to assume you're going to loss all your cooling pumps. In other cases, such as those where you have multiple water distribution systems, loosing all the CTW pumps isn't credible. That decision also depends on your electrical distribution design. If you have four CTW pumps, it's common that two are fed by one electrical feeder (and MCC) and the other two by a separate one.

Once all that done/checked/reviewed, determine the flare header sizing by simulating each global scenario using a network flow tool like Flarenet. At this point, sizing the sub-hdrs and main-hdrs is a pretty simple task. BTW, in my experience, the worst case flare load is almost always based on loss of cooling. If the flare load is too high, especially in an existing facility, you need to figure out how to justify reducing it. Fortunately, for loss of cooling scenarios that's relatively easy to do. You can use HIP (or SIS if that acceptable with the owner's risk tolerance) to stop the overpressure from loss of cooling before it happens. That can be done by tripping the heat source before the vessel MAWP is reached). By strategically using HIPS/SIS to "turn-off" one or more high loads, you can get the total load down to an acceptable value. I haven't yet found a case in which we couldn't solve a load problem without expanding/mofifying an existing flare system, and that's good because doing so is generally prohibitively expensive.