Protecting Sweet Pipeline from back flow when tying into Sour

[BarrySand]

It has been standard practice at our company to use two check valves on a sweet pipeline when tying into a sour pipeline in order to protect against back flow. I've been looking through all the codes I can get my hands on to find this as a requirement but can't find it written anywhere explicitly.

Does anyone know a code which would explicitly state the need for this double check valve to protect against back flow?

I've talked with the ERCB and they stated that they only required one unless the tie-in was specifically for the purpose of blending (at which point other regulations come in) but everyone here seems certain a code calls for two.

 

[BigInch]

You won't find that in any code because it is not a consideration of design for safety. The two check valves are there for operational reasons only. Nobody wants to contaminate a whole lot of sweet gas with sour. Since two check valves are usually considered to have a low enough probability of failure at the same time such that the possibility of backflow will be held to an acceptable risk level, you won't need three.

Let your acquaintances be many, but your advisors one in a thousand’ ... Book of Ecclesiasticus

 

[zdas04]

I just finished a project where we used two check valves for a spec break (which is really what you're doing with the sweet/sour separation). To satisfy company requirements for the spec break we had to be able to test the check valves periodically. We came up with block valves outside the pair of check valves and a vent/fill valve between each pair of valves (i.e. one vent between the upstream block and the first check, one vent between the two check valves, and one between the second check and outlet block). With that piping configuration you can shut the block valves and put pressure between the outlet block and the second check and test the pressure between the checks, then you can put pressure between the checks and test the pressure between the inlet check and the inlet block.

For safety-critical spec breaks, this seemed like the most cost effective way to be able to verify integrity in a place with no access to electricity or pressurized gas (the line was produced water). Without the means (and procedures, and schedule) to test the check valves there is no way I'd sign off on it being a spec break.

David

 

[BarrySand]

Would this not be a safety issue as there is a possibilty of contaminating parts and piping rated for sweet service with H2S if not properly protected?

I assumed this would atleast be would in a some set of recommended practices like API.

 

[KiwiMace]

This is highly analogous to domestic cold water connections in what David describes being equivalent to a double check backflow prevention assembly (like a Watts 774 for example). Other assys automatically bleed from between the two checks.

 

[zdas04]

Barry,
While I don't want a building to fill up with sweet gas, I REALLY don't want it to fill up with sour gas. Either one would kill folks (either from an oxygen deficient atmosphere or from poison gas).

I was thinking that I have significantly more stringent design criteria for a sour pipeline than for a sweet pipeline. If I can provide a spec break between the two lines, then I can build the sweet line to the sweet spec and the sour line to the sour spec. Spec breaks are by definition safety critical.

David

 

[BigInch]

Its not an immediate safety issue for strength, or containment, the primary concern of design codes.

Could be a wall thickness pressure safety issue over time, if you don't address a significant corrosion potential by one method or another.

It might be a product-material compatibility issue, if you chose your materials poorly.

Breathing 100% oxygen, or closing a 2" valve might be a safety issue too. Neither of those addressed by design codes either.

In any case, don't look for the two check requirement in the design codes. You won't find it.

Let your acquaintances be many, but your advisors one in a thousand’ ... Book of Ecclesiasticus