1.Which is the preferred solution method to carry out slug flow analysis : Time history or Spectrum (Waterhammer/Slug) analysis. Why?
2.For slug analysis using DLF v/s frequency, slug duration is used to generate the DLF v/s frequency profile. Whether analysis takes care of load duration, once the profile is generated using DLF generator or that slug duration should be provide additionally as load duration in control parameters? However, active control parameter in Water hammer/Slug analysis does not show the option of providing load duration.
3. Slug load is periodic in nature. Whether this periodic nature of slug force can be taken into account in analysis? How?
4. Whether analysis is to be carried out for each elbow/tee individually or that combinations of these load at different elbows are to be considered ?

Here are my thoughts on your "slug analysis" questions.

1. I would prefer a spectrum analysis over a time history analysis for a "slug" problem. The reason is that there is too much about the slug that you don't know and have to guess at. Therefore, a conservative envelope approach is a better alternative to the more exact solution offered by the time history.
2. The slug duration is considered in the generation of the spectrum/DLF - this is the flat portion of the trapezoidal pulse. This is why there is no entry for this value on the "control parameters" dialog. <em>Remember, there are no time/phasing relationships in a frequency domain (spectrum)analysis.</em>
3. The spectrum analysis assumes your loading is an impulse, i.e. not periodic in nature. If your loading is such that a second slug enters the system before the vibrations caused by the first disapate, then perhaps you would be better off looking at a harmonic analysis. (This is equally difficult because you don't really know anything about the slug, but you have to input a force!)
4. Again, because we are assuming a one-time occurrance of an impulse, we only have to consider a single elbow or tee in any one load case. Usually, if you look at the longest 3 or 4 runs of pipe, this is sufficient. If these don't cause a problem, no other part of the system will either, because there won't be enough time to develop a response to the load.

Several times above, I stated that you don't know too much about the slug. For an accurate analysis, you need to estimate the length and diameter of the slug. These values are necessary so you can compute the timing rounding elbows, and the force exerted on the system by the slug. Most users performing slug analysis assume a <em>small slug</em> and a <em>large</em> slug, and assume they have enveloped the actual situation.


------------------
Regards,
Richard Ay (COADE, Inc.)

I'd like to share my experience on 'slug analysis' with us.

I use time-history analysis for slug in piping stress analysis because no spectrum for slug is available in industry or international codes. You may make a spectrum, But it is very hard to get it justified.

The numbers of slug exist in piping system depends on the process nature of the piping system. Normally Process study is required.
when liquid level is low and pressure variation are minor, No slug can take place.

For long slug, you have to consider all elbows within the length of slug. and more strickly speacking they are in different phases.

The frequency and force of slug will determine if fatigue analysis are required.
For many cases, Fatigue analysis can be avoided.

Actually, analysis for whater hammer is somewhat different from these for slug in gas simply because their natures are different.

Regards

Alvin Zhu

As a structural engineer I have a question regarding slug loading.

The scenario that we have is a blowdown header (36" dia.) that should be exclusively gas flow. Unfortunately condensate accumulates in the pipe at times and eventually causes a slug. There has been considerable pipe /nozzle damage as a result of these slugs, and as a result the pipe is being modified with the addition of snubbers. A few years ago one of these snubbers was added to a portion of the line with a design load of 50 kips. This was attached to a massive concrete structure and as such other than the issue of trying to install concrete anchors through a rats nest of reinforcing steel this installation did not cause us any real concerns.

Now they want to add more snubbers further down the line in pairs with a load capacity of 75 kips. Unfortunately these new snubber locations are no longer within the concrete structure, but rather a steel piperack structure. Preliminary analysis of our rack structure reveals that we will be significantly over the capacity of the structure with these loads, and reinforcing of this structure will present major challenges. For example the pile foundation will be overloaded in the lateral direction, which because of the non-linear nature of the soil would increase lateral pile deflections from a permissible 1/4" to a totally unacceptable 1.5 to 2 inches. Adding in extra piles is not really feasible, because there is no room to bring in a conventional pile rig to install a second pile of the same size as the first. The number of micropiles required to provide the same capacity border on the ridiculous, quite aside from the fact that there is no realestate for such a scheme anyway, with pumps and other equipment crowded in all around the rack.

The long and short of this problem is that while trying to determine a solution to the problem, I realized that perhaps adopting techniques from seismic design (ductility ratios) and blast resistant design (ductility ratios and load duration versus stucture natural period) I might be able to reduce the actual load that the structure would actually have to be designed for to resist the slug load.

One of the key components of this solution is the load duration, and so finally getting to my question, is there a range of durations that can be associated with slug flow, specifically of interest to me, an upper bound?

I appreciate that I haven't given you a lot to work with in terms of input data, but our experience thus far in getting data on these events from the client has not been good (for the longest time they denied that they were even getting slugs). Hence why I am just tring to see if there is a range based on general experience.

Slugs are formed as you suggest, when liquid condenses and falls to the low points in the piping system. This raises the velocity of the vapor above, until the surface friction forces begome large enough to grab the liquid and move it along. When this reaches the terminal equipment, all hell can break loose, since you have a large mass of liquid moving at normal vapor transport velocities.

Estimating the total liquid volume that can be contained in a slug, as well as the velocity to be expected when a slug is carried along with the vapor, can be very difficult. There is commercial software available, usually used in the petrochem industries. It's very expensive, and apparently you will not enjoy learning to use it. Unfortunately, you really need a reliable estimate in order to come up with a sensible answer, although "pipe full of liquid" for so many diameters is conservative (which your foundation designer may not appreciate since a pipe full of liquid moving at 5-10 times normal liquid flow velocities is a hell of a transient load).

Once the slug hits an obstruction (a change in direction, or a piece of terminal equipment), the forces generated are governed by the momentum equation that you all learned in Fluid Mechanics 101 in your sophomore year in college and, most likely, haven't used since. I had a slug problem to work about a year ago, and in going back to my textbook found that the sample problem in the text was solved incorrectly!

A pragmatic solution to the problem is to make sure the pipe slopes down toward the offending obstruction. Then, if you have small slugs, there will be time for gravity to accelerate the liquid and thereby get back to a "pipe half full" flow problem, which is usually much less exciting.

Have fun!

CraigB

I am the structural engineer in this situation. The line is existing, and changing slope is not an option. All that they are trying to do is reduce the damage to the piping/equipment nozzles by introducing the snubbers. Unfortunatly this adds huge forces to the supporting structure (via slugs hitting 90 degree elbows in multiple locations - fortunately only one at a time).

Taking the snubber reactions as a normal static structural load with an impact factor creates loads that are (based on preliminary estimates) anywhere from 700 to 2000 percent of the available spare capacity in the structure based on the weakest links: piles, precast base columns and the dywidag bars connecting the piles and columns - none are easily upgraded. The modular steel structure above the precast has more residual capacity, and can more easily be upgraded, but would still present a major challenge.

My initial reaction was "go away this can't be done". Basically upgrading the structure in the only practical way by creating a side structure to take these huge loads is not feasible because there is no (read zero) available realestate available for such a solution.

While racking my brain for a solution it occurred to me that since the slug event is somwhat like a gun shot, with a brief transient load, that techniques for designing structures for blast loads might be applicable. I have designed structures for blast loads ranging from 10 psi to as high as 100 psi (the latter 1800 times normal wind loads), and with the short duration of the blast load versus the natural period of the structure it was possible to design for significantly smaller loads. Part of the reduction was due to allowing significant yielding of the structures, not an option in this instance, but allowing a ductility ratio of 1.5, and using a short duration for the slug event I could conceivably only have to actually deal with loads at 160 to 400% of available capacity. I might be able to deal with these kinds of numbers, although not with some pain.

This is why I was looking for some guidelines on what sorts of duration might be reasonable for a slugging event (i.e a range). I did read somewhere about 100 ms. This would help me. Lower values would be better. A duration approaching 500 ms would leave me having to design for pretty much the full capacity of the snubbers.

There was a recent event at the client's site and we are hoping that we can get access to some decent data so that we can better model the event. This would help our stress engineer better determine a realistic slugging load, and for me to better assess the structural impacts of this load.

Regards
AJ

Isn't there a maximum slug length, say, for example, 5ODs? With that and speed you can set maximum duration.
I think that, with dynamics, load requirements drop when you allow things to move a little. So rigid restraints will carry big loads and more flexible restraints will share load with the piping "structure". The problem then is, how much of this slug load goes through the piping to sensitive connections?

Alvin,

I have used the following simple expression to estimate the duration of slug load at an elbow:
bend radius x bend angle (in radian) / estimated slug velocity (anywhere between 0.5 to 1 time the gas velocity)

Fortunately, I have not had any issue with pipe or structure failing due to slug loads in the 2-phase lines that we have designed - so, guess the estimation may not be unrealistic.

Good luck

Sanjay

Correction to my earlier posting.

The duration for which the slug load is sustained should actually be: (estimated slug length- bend radius x bend angle)/ estimated slug velocity.

The expression in my previous posting (ie, the second part of the above equation) is actually the time it takes for the slug force to develope from and dissipate to zero. Thus the average value of those periods need to be considered also for your impact assesment.

Your process engineer colleagues should be able to give you an indication of the slug length (conservative estimates in terms of pipe diameter is common).

Hope everything is as clear as mud now

cheers,

Dave & Sanjay

Thanks for the input. That was what I was looking for.

Dave your formula would give me a slug length of about 150 ft (yikes), and at near sonic velocity that would translate into a duration of about 130 ms. Compared to a structure period of 500 ms I could get a good dynamic reduction of the forces. Actually our structure period was estimated based on a bare steel structure, so the weight of the piping would probably increase the natural period (which also helps), but restraints on the piping may reduce the period.

Sanjay, your formula would appear to shorten the duration depending on the elbow geometry, so this can only help me.

At least conceptually I think I have a solution to the huge forces that stress has given me.

This is all at a front-end stage. In a detailed analysis, we hope to get better data from the client (considering I have proposed structural reinforcing that will effectively block access to a number of pumps and partially block a roadway we may finally get them to take our request for data seriously). The stress guy will need to revaluate his assumptions too, because the stress engineer on another coker currently being designed for the same client has come up with much smaller numbers. I have talked to both of them, and each doesn't understand how the other came up with their numbers, so I have asked the two of them to talk to each other (strange concept considering they're both working for the same client, and you'd think that the new project would like to learn from the lessons of the older project).

For the rack natural period we will need to do a detailed check of the rack and piping. But at least there is a glimmer of hope that we can make this work.