Seeking resources about crude oil flow in pipelines

[Internetguy]

Hello! I've inherited a project where I have to modify an existing leak detection system for a crude oil pipeline. It's a Volume Balance system (if volume departing a site is different from the volume arriving at the destination, then the difference is seen as a leak). The problem is, the system compares volumes at a moment in time.

We have a few pipelines that aren't packed, and so there is a time delay between departure and arrival, usually between 30 minutes to an hour, depending on the line. This throws the leak detection system off, and I want to predict the amount of time that it would take for the crude to reach it's destination.
Can anyone recommend a textbook or other resource for learning about crude oil pipelines?

 

[bimr]

The API has a reference library:

http://www.americanpetroleuminstitu...-statistics/publications/publications-catalog

 

[LittleInch]

No leak detection system will work with pipelines that are not"packed"

I find it difficult to believe you have a working system like this.



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

 

[Internetguy]

bimr:
Thanks for the link, I'll spend some time with that!

LittleInch:
Sorry if I miscommunicated. The leak detection system we have works for "packed" lines, but does not currently work for lines that aren't "packed". What I'm trying to do is to predict how long it takes for a shipment to reach its destination, given a specific flow rate and pipeline.

 

[BigInch]

Pure Volume balance does not care if a pipeline, or any transportation method, is running full. 10 one-half truckloads leave during a week. 9 one-half truckloads arrive during the same week. 1/2 truckload missing = presumed leak.

It doesn't depend on the volume in the pipeline, only what goes in and what comes out. The integral of flow in over time equals the same coming out over the same time, as long as the time is long enough to account for holdups within the line and give the holdups time to deplete again. Holdups are just noise in the signal. Sometimes the noise lasts a long time, however if the noise is getting louder, a leak can still be presumed. If next week two 1/2 truckloads are missing. That's probably a 1/2 truckload / week leak.

 

[LittleInch]

Pure volume balance will still only work if the start condition and end condition over your time period is the same. With a non packed pipeline that's difficult to achieve with any accuracy.

Why can't you operate these pipeline s in a packed condition at all times??

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

 

[Internetguy]

BigInch:
The issue with that approach in the troublesome pipelines is that the flow trickles down to a crawl in between shipments, but it doesn't stop. I'll have to go back and look for any large gaps between shipments where the flow might come close enough to zero to be called "stopped", but I believe that these periods are far and few between.

LittleInch:
The troublesome pipelines would be operated in a packed condition if they were equipped to, but they are not equipped to do it. I don't fully understand the reason why, but if I recall correctly, the pipes that operate in these conditions are short in length, and have their sources on a higher elevation. Gravity drives the flow, as long as a certain pressure is met at a checkpoint.

 

[BigInch]

Start and end conditions do not have to be the same. In fact let there be no start and no stop. Keep it clear that this is not a transient leak detection method. Start begins when you start counting time. Transients are noise that over time will equal net zero, unless there is a leak. "Over time" means after the transients have settled out. Steady state is left and if Vol in <> Volume out "over time" and [&Delta;]V keeps increasing... its probably a leak. Pipeline doesn't have to be packed either. If you run an unpacked pipeline, as LittleInch said, It's going to be tough to model it hydraulically, plus you will have to incorporate extended noise limits making the alarm system notoriously unreliable. The operators will eventually learn to ignore it.

This is a completely statistical method. It needs to know nothing about hydraulics. Just algebra. If there is no leak, all volumes going in = all volumes going out, over time. If it doesn't, it's a leak... (or an ingress to the pipe, in the cases of very high hydrostatic pressure outside the pipe).

A 1% leak is picked up in around 10 time steps.

See attached spreadsheet.

 

[BigInch]

That 1% in 10 time steps is when leak rate is equal to half the noise. If noise is a greater percentage of steady steady state, it will take a bit longer.

 

[BigInch]

InternetGuy, your noise is too loud. You will not be able to operate that leak detection system when the noise is loud. Don't start the leak detection system until you have nearly equal inlet and outlet flow rates, and stop it before slowing down the flow rate again.

Your best bet for a leak detection for system like that will probably be a leak sniffing or physical sensor cable string of some kind.