Reverse flow / vortexing question 1

[MidstreamEgr]

I have a system that has two 4" gasoline truck loading-arm metering assemblies fed by a single 6" pump header. Each arm may or may not be operating simultaneously for truck loading. In the 4" branch from the header for each, we are injecting ethanol. This injection is downstream of the common header, and upstream of the metering assembly (meter, control valve, arm). Since we need to make sure the correct amount of ethanol is injected in each, regardless of how many arms are operating and when, the conceern was raised if we have any way to make sure that we would get the "right amount" of ethanol in each stream -- for example, we want two E10 streams, not am E5 and an E15.

THe concern is that injected fluid from one of the two 4" pipes could migrate to or affect the flow of the other. The injection of ethanol would be from a higher pressure source (to ensure it's able to inject), but since it's injecting into an "open pipe", the pressure should equalize instantly with the primary fluid (gasoline).

I have no concern about this myself, but can't assemble the right technical reasons or math to prove it won't be an issue. Energy required to backflow against an active fluid stream, potential of vortexing at one tee and "sucking" up fluid, etc. None of this seems possible especially since each arm is metered, but it's only experience and gut, not mathematical equations. Of course, I could be wrong too.

Any thoughts. Am I off-base? Is there a way to calculate minimum distance from common header to the injection point to avoid vortex, or "injection plume", etc?

 

[rutherford703]

It depends on how you control the ethanol flow and also injection points. If you can accurately control the ethanol injection rate, then based on the flow meters in the loading arms, you can get accurate ratio of gasoline/ethanol. Otherwise, you may end with one arm with more and the other with less ethanol.

As to the back mixing or vortex, I don't see it is a problem. The flow will always go to the lower pressure direction - the tank provided the valve to the tank is open

 

[MidstreamEgr]

The ethanol streams are metered, as are the combined ethanol/gasolione streams - which is where the lack of ability to confirm is. For example, if the total flow is 600gpm for each arm (See sketch) ---
- 60gpm of ethanol into injection A for meter A
- 600gpm at meter A of combined gasoline/ethanol
- 60gpm of ethanol into injection B for meter B
- 600gpm at meter B if combined gasoline/ethanol

But since the gasoline itself isn't metered separately prior to ethanol introduction, we may have 600gpm of gas and 0gpm of ethanol on meter A, and 480gpm of gas and 120gpm of ethanol on meter B. Meters won't tell us how much otherwise...

But that isn't a concern (that we can't meter it separately), as long as we know that physically the ethanol has no way of flowing "upstream" to the common header and into the wrong meter. Looking at injection point A - is there a way the ethanol from A (which is only injecting if that 4" meter run is operating) to get into the B meter run?

 

[bimr]

You are not really "injecting into an "open pipe" because you have the flow meter after the injection point. One would assume that the meter is not measuring air but liquid.

You may have problems if the head loss downstream of the injection point causes flow to back up.

One can envision a scenario where one of the discharge valves is not fully open which then causes flow to back up in the other one.

 

[BigInch]

You're probably OK if you can stay closely to steady state flows in both the A and B lines. You may have some cross-mix flow potential, especially as you start up flow in one meter while you are closing off flow in the other. If, for example, the A line was shut off quickly enough to cause backpressure to build up in the header, it would attempt to equalize by filling into the accelerating B line. Then B would be injecting 10% ethanol into a stream that already had 10% ethanol mixed into it. It may not be a problem practically speaking, as I imagine there is not much potential for large quantities to backflow and cross-mix in this small system, but theoretically it presents some mathematical difficulty in the QA/QC control of the mixtures, as it can't be stated for certain exactly what mixture might be flowing where during those situations. Any backflow across the injection points would nearly double the ethanol injected into that volume as it passed by the injection point again. There may also be some slight cross-mixing potential when the systems are both off and temperature changes in the lines shrink or expand. Again, not much would occur in this small system, but the QA/QC aspect of all of that gets very sloppy. A couple of check valves would tend to help with positive control of the start-up/shut-down scenario. Perhaps one upstream of each injection point.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[bimr]

A better piping scheme would have injected the ethanol downstream of the valve and meter.

 

[MidstreamEgr]

Injection is controlled by flow meter and control valve just as the overall flow is controlled by flow meter and control valve, from a centrifugal pump so that flow is always "forward".

Bimr, sorry for the confusion but when I said "gas" it's short for gasoline, not gas/vapor. inject downstream is not an option due to process requirements to meter the combined flow with temp correction.

With primary flow always moving "forward", I think BigInch is right that start/stop may be an opportunity for "reverse" flow to travel upstream against primary. To avoid this we can have the injection start after and stop before primary (gasoline) flow starts and stops.

The only other I can think of is if flow thru the tee could siphon flow from injection A if it's "too close" to the tee, and relative flow rates between primary A and primary B differ enough. What I don't know is how to calculate that potential, or to figure out how far from the tee the injection point needs to be in order to ensure we are avoiding that.

 

[TD2K]

Given you are loading 600 gpm of ethanol/gasoline mix, the 60 gpm of ethanol can't go anywhere but into the truck along with the flow of gasoline.

I did a design for a terminal similar to this but I'm pretty sure we metered the gasoline and the ethanol separately and then mixed the two streams. I'll have to see if I have any old P&IDs to be sure.

 

[BigInch]

To ensure positive control of the mixture, each stream must be metered separately and then the mix would be metered again (for writing the sales ticket) at the point of load-out. Ethanol injection could be "metered" by a PD injection pump.

If you had the ethanol injectors 5 diameters downstream of tees, possibly 3, I would think vortex mixing would have no chance to occur. My only worries here would be for the transient cases and the lack of positively assured mixture control. Those two check valves would go a long way to enhancing that. Normally you would have another check directly on each sales meter as well to absolutely-positively prevent backflow there. Although you could say that you have a reasonably good idea of what is in the mix, your QA/QC Dept., or your client's laywers, may argue with you forever about that.

Centrifugals do not always flow forward; when stopped, leakage through to suction can occur. Is there a check valve at the pump. Does the header have a relief valve, or another outlet that might cause backflow on depressuring, or some other backflow on depressurizing scenario. In my start-up/shut-down scenario I did not assume that you would have high enough flow velocities on shut-down to worry about. Is there an emergency stop valve somewhere that might cause "reverse" waterhammer, vaporization and backflow when refilling the vapor pockets.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[MidstreamEgr]

thanks to all! I am investigating vortex/siphon as a potential, and will arrange the operation of the system to avoid the transient cases. I'm considering the check valves, but given that 90% of the flow will be thru the check valve, the 10% ethanol that is a concern it seems wouldn't be an issue (if there's enough gasoline flow to keep the check valve is open, it's very unlikely the ethanol flow could independently flow upstream thru the check)

 

[BigInch]

Correct.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[bimr]

Regarding

"sorry for the confusion but when I said "gas" it's short for gasoline, not gas/vapor. inject downstream is not an option due to process requirements to meter the combined flow with temp correction."

There is no confusion. You said in your first post "injecting into an "open pipe" which was an incorrect statement and would imply some sort of drop pipe. Your sketch clarified the matter.

Not sure there is a requirement to meter a combined gasoline and ethanol mixture. Surely your control system can add up the fuel totals if the fluids are metered separately. Since you already have separate meters, why wouldn't you use separate meters for better accuracy?

Both of the attached articles indicate that the ratio method of blending is more common as well as more accurate than the sequential method of blending that you are doing.

http://www.ethanolproducer.com/articles/8340/terminal-error

http://help.intellisitesuite.com/Hydrocarbon/papers/2030.pdf

Also note the piping diagram in Fig. 4.

 

[MidstreamEgr]

Thanks BIMR. We're not actually sequentially blending (which each stream is flowing one at a time). It's sidestream blend, which is sub-classification of ratio blending, where the two streams are actively flowing at the same time and combine upstream of the primary meter. The primary reason for this is to allow the owner to account for the chemical growth that results from the gasoline/ethanol mix (see

http://info.smithmeter.com/literature/docs/tp0a015.pdf

). This is done ~50% of the time at terminals blending ethanol into gasoline at the rack, for them to take advantage of the incremental gains (tiny percentages over huge volumes).

 

[BigInch]

A+B = C
C <> A+B. A common problem with blending HC liquids, but an advantage if you meter it.
I like that doc.
Note they did show one check valve on the skid, Fig 8.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[bimr]

Fig 8 in an unloading skid, not a loading arm.

One would think that you should have some concerned then if you are trying for that accuracy. Every one of the Typical Ethanol Blending Arm Configurations are shown with check valves on each of the product lines where the lines meet.

 

[BigInch]

Now what is it.

If it ain't broke, don't fix it. If it's not safe ... make it that way.