Software solutions 1

[kepharda]

My problem involves a fuel transfer pump. Each time my client first turns it on (1-2 times a week) to pump fuel from the holding tank to his truck loading rack, the pump creates a pressure surge which is breaking piping mounts. The pipe is visibly moving 6-12 inches depending on the location. Once it has been started, subsequent restarts cause no problems.

Luckily I did not design the pipeline, however, I am interested in what other engineers would suggest in the way of modeling software to root out the real problem. I am not looking for anything free, but really just ease of use and say less than 20K a seat.

The specs are 6"-8" pipeline, about 1,143 feet long. The pump is a Goulds Pump 3196 4x6, 1200 GPM. I walked the pipeline today and there are a couple of things that don't look right, but I would really like to model the thing and see what I can come up with.

thanks for your reply's in advance

-dave

 

[khardy]

I spend almost all my professional life designing and modifying truck loading racks for fuels. Never had this happend to me. However, I have seen this in a fire system during a test. I found out that, prior to the test, the line was partially empty. We opened the high-point vents to purge the air and pack the line with water. Then started the pump - no problem.

So, based on what your describing I suspect the line isn't full of fuel when the pump starts. The slug of fuel races down the line, compressing the air untill the air won't compress any more, and the line jumps as the fluid is brough to a stop.

Your client may be pruging the line between uses. (which is a bad idea for several reasons.) Or there could be a leak somewhere which lets air in and allows product to drain back to the tank. Solutions would be to find the leak (look near high points) and / or to install a check valve at the pump outlet.

Now that I think about it this did happen to me once. On startup i shattered a meter blade because i didn't purge all the air out of the line before starting the pump.

 

[BigInch]

Too much power for that short line. What's the rated power of the motor? Motor's often can produce 1.5 x rated torque during startup. This system needs to be started against a closed valve, then opened up slowly.

I use Stoner Pipeline Simulator by Advantica, but don't know the current cost. Applied Flow Technologies has a series of programs that you might want to investigate. I think "AFT Arrow" is their basic transient analysis package.
You have to sort through the options you want to add to arrive at the pricetag.

http://www.aft.com/products/fathom/

"If everything seems under control, you're just not moving fast enough."
- Mario Andretti- When asked about transient hydraulics

http://virtualpipeline.spaces.msn.com

 

[kepharda]

I am 99% sure the pipeline is full of liquid. It is never purged, I spoke to a maintenance tech and the entire line is above ground, there are no leaks, they never drain it, in fact they have gone to great lengths on their own to ensure it is full of liquid. Liquid should not be draining back to the tank as there is a check valve after the pump. I guess it could be leaking, but again the techs on site have checked the line (high point vents) over and over again and never found any air.

Another thing that struck me as very strange was that about 900 feet away from the pump, the pipeline the line size goes from a 8" to 6" line size. There are no branches it just necks down before going back up to a pipe bridge and over to the truck rack itself.

As for the size of the motor I agree it likly too large, it is a 75 hp motor. Strangely enough after the tech bumped the pump (and the pipe made a horrible noise) he let it sit a moment then restarted the pump and there was no problem. The pump had not been ran in about 7 days.

All in all an interesting problem, you should see the pipe support that our client tried to clamp the pipe down to. The concrete was basically torn from the ground.

 

[JohnBreen]

It has been a big problem with underground fueling systems at airports and air force facilities (the quick closing delivery nozzle valve "snaps" shut when the fill sensor says "enough"). Many facilities have seam welded schedule 10 (or less) stainless steel pipe and the seams were splitting (digging up the buried pipe (below the apron concrete) was a spot of bother). In every case that I know about the problem has been solved by the installation of pulsation (surge) bottles that use a reservoir of air above the fuel to cushion the shock. Bos Fluids (Paulin Research Group) was the software that was used for analysis.

John

 

[BigInch]

That's a very large pump for a very short pipeline. I have done many similar models to rectify exactly this same type of problem that has occured on even longer pipelines. What's happening is the transient generated by the start of the pump moves down the pipeline which is very significant due to the 1.5 torque factor at startup, hits the end of the pipeline and 1/2 of the remaining energy is reflected back through the pipeline to the pump discharge where 1/2 of that energy starts the trip again. EACH TRIP down and back only takes about 0.5 seconds, so basically before the pipe has stopped moving from the first hit, it gets hit again and again and again. That will happen at least 4 or 5 times and probably a number of them will occur before the first wave has a chance to disipate totally as well. It is essential that the pump be started against a closed valve and the valve must be slowly opened such that only a small amount of liquid in the pipeline is pressurized during any given increment. It is also essential that the valve not be opened further until the first wave has had time to reflect and dissipate at least one cycle, maybe two. I would guess that the valve's opening time would need to be at least 30 seconds. Furthermore do not use a ball valve for this as their flow pattern is basically 100% open at 10% or less of the total travel. A starting valve with a nonlinear flow pattern is needed. A control style plug valve or a control valve that can be slowly ramped open and closed is what you should be using to attempt to control this situation. You really don't need a computer program to model this, as it is a very uncomplicated configuration and a very short line with little line pack to complicate matters by damping a hard reflection by any added compressibility time. The Joukowski equation will give you a very good idea of the initial pressure, but won't tell you much in itself about how reflections may be disipated and recombined, but I can assure you that it is a pretty rigid system and short enough that not much energy is dissipated before the next wave hits. You say its a 6" and 8" pipeline, so at that interface you are getting a reflection back to the pump equal to original pressure times the ratio of the area of the 6" to the area of the 8", the remainder of which continues down the pipe and is reflected back to the interface and so on, so depending on where that 6/8 connection is, a pretty good reflection may be arriving back at the pump even faster than what I have said above.

"If everything seems under control, you're just not moving fast enough."
- Mario Andretti- When asked about transient hydraulics

http://virtualpipeline.spaces.msn.com

 

[BigInch]

Why did I say you don't have to model it? I meant to say, you don't have to model it to understand the process, but I do think its a good idea to model it to get an idea of what combinations of waves you might be getting and to see if any add togther to increase the original surge pressure predicted by the Joukowski equation. While it can be a good indicator of potential trouble, its not conservative all the time, unless you know enough to continue the surge and resurge calculation until the wave has mostly disipated and have considered reflection possibilities from reducers, partially closed valves, tees, etc.

"If everything seems under control, you're just not moving fast enough."
- Mario Andretti- When asked about transient hydraulics

http://virtualpipeline.spaces.msn.com

 

[zuccus]

What about adding a VFD to the pump and ramp it up over a little bit of time. Kind of using it as a soft start if you will. Might be easier and cheaper then adding Vball or similiar type valve into an existing system.

This is me just talking out loud and may not be an effictive solution.

Zuccus

 

[khardy]

"Once it has been started, subsequent restarts cause no problems."

and

"Strangely enough after the tech bumped the pump (and the pipe made a horrible noise) he let it sit a moment then restarted the pump and there was no problem. The pump had not been ran in about 7 days."

Biginch, if the motor were too powerful, why would the problem dissapear after a subsequent restart? I assume the restarts are a few minutes apart shuch that the fluid has stopped.

 

[kepharda]

Actually the time between the bump of the motor and the restart was less than 1 minute apart.

And at the truck rack no fuel was moving at all. As far as I could tell all we were really doing was pressurizing the pipeline.

 

[BigInch]

Yes, you were packing the line, which pressures the pipeline.

khardy,

Transients are the result of either converting static head to velocity head = v^2/2/g, or converting that velocity head into a pressure head. Any change in momentum of the fluid must go somewhere. After the first start the line is packed full as product enters the pipeline and the pipeline reaches pump discharge pressure. The second start does not add enough additional fluid volume to the pipeline's already packed and compressed contents to create another velocity with enough momentum to kick the pipe around again. If he depressurized the pipeline than restarted, I'm pretty sure he would get the same results.

"If everything seems under control, you're just not moving fast enough."
- Mario Andretti- When asked about transient hydraulics

http://virtualpipeline.spaces.msn.com

 

[MJCronin]

kepharda,

I agree with the suggestion of John Breen; the installation of a pulsation damper should go a long way toward solving your problem.

Also, your 1200 gpm of liquid going through a section of 6" NPS piping gives a fluid velocity of ~ 13+ feet per second, which seems a little high.

Any chance that a high-range pressure gauge (or two) could be installed on the line to get an idea of the magnitude of the surge ?

Actual pressure measurements could come in handy if you do decide to verify computer analysis done on the system

-MJC

 

[kepharda]

MJC

a high-range pressure guage would be great, but like most clients they would only like to know real answer to the problem if it does not cost them additional $.

I am guessing that I will probably be installing a VFD or control valve to eliminate the symptoms and changing out to a smaller impeller.

 

[khardy]

kepharda

I’m assuming like most loading racks, the pump starts, pressurizes the line, and a moment later a valve at the loading station is opened either automatically via a controller or manually. Is this the case? Also, is this a centrifugal pump or PD? Is there a pressure relief valve in the line?

 

[kepharda]

Khardy,

Your assumtion is correct.

Centrifugal pump. There is a pressure relief in the line, it is near the pump/tank and is connected downstream of the check valve and connects back to the Pump outlet.

I downloaded a free demo of AFT Arrow and AFT Fathom to see how they actually work. As soon as I get a chance at least.

thanks,

dave

this message has been approved for citizen to elect kepharda 2008

 

[BigInch]

The relief is connected back to the pump's outlet? Which outlet?

The demo only works on their included example problems. You can't set it up to run your's. But you can see how it works.

"If everything seems under control, you're just not moving fast enough."
- Mario Andretti- When asked about transient hydraulics

http://virtualpipeline.spaces.msn.com

 

[khardy]

BigInch,

I agree that transients are the results of a change in fluid velocity. But just pressurizing the line won’t do it; there has to be a change in velocity. In “subsequent restarts,” the line is packed and starting the pump causes no change in velocity, thus so pressure wave. When you say, “After the first start the line is packed full as product enters the pipeline and the pipeline reaches pump discharge pressure. The second start does not add enough additional fluid volume to the pipeline's already packed and compressed contents to create another velocity with enough momentum to kick the pipe around again.” you are baisically restating my theory. What I don’t understand is how the line is getting “unpacked” when they secure from loading for a few days.

Kepharda,
After they are done loading for the day, do they close one or more of the pump isolation valves? If so, here is a theory: After securing from the day’s loading the staff closes the pump’s isolation valve(s). During the following days, sun heats the discharge line, expands the product, which in turn is vented through the pressure relief line back to the tank. When the line cools at night the product contracts and pulls air in through a very small leak in a valve or flange at the rack. When they start the pump a few days later for the first time, the fluid rushes down the line, compressing the air trapped in side, until the air won’t compress anymore. The product stops suddenly and the resulting pressure wave causes the pipe to jump.

I know you said the line is packed, but this is the only thing I can think of.

 

[BigInch]

Just a small loss of product will cause what seems to be a very disproportionate pressure loss. It could be bleeding through the pump itself, back to the suction side.

"If everything seems under control, you're just not moving fast enough."
- Mario Andretti- When asked about transient hydraulics

http://virtualpipeline.spaces.msn.com

 

[kepharda]

BigTen,

The relief is connected back to the piping just downstream of the pump's outlet....I don't exactly know what you are asking for sure, there is only one fluid outlet on a pump,....let me know if this is unclear.

Khardy,

The pipeline isn't operated in that manner, it is only run sporadically lets say 3 times a month. Yes, the "sunshine pressure" is created because the pipeline isn't operated continuously. So transmix does sit and depressurize thru the relief back to the tank.

this message has been approved for citizen to elect kepharda 2008

 

[BigInch]

So that relief valve is connected from the pipeline to the pump discharge? Wouldn't that be a high pressure trying to relieve to an even higher pressure (when the pump is runing)? How does that work? I think I would have connected it to the pump's suction (provided the pump's suction is rated to that hi a pressure).

"If everything seems under control, you're just not moving fast enough."
- Mario Andretti- When asked about transient hydraulics

http://virtualpipeline.spaces.msn.com