High Flow Through 2" Piping/Nozzle 1

[EmmanuelTop]

Good afternoon everybody,

We are in the process of fixing the design error regarding the minimum flow of high-head, multistage centrifugal pumps. In brief, the existing single loop (both pumps have the minimum flow control through a single valve which simply does not work) is to be replaced by separate, individual loops for minimum flow control of each pump. The existing configuration simply uses the minimum flow line downstream of the air cooler at discharge of the pumps, while rest of the flow (flow-controlled) goes to the process.

However, this solution with individual recycle lines leaves us without the possibility to cool the recycle streams which may lead to progressive temperature rise across the pump(s) and subsequent damage. The questions arrisen are:

1. Is it worth considering re-designing the recycle loops without any cooling, and just recycling (partially) directly to the pump suction? Recycle flow varies from 0 up to 50% of total discharge flow of the pump. Personally I am against such solution as we don't know where we may end up.

2. Another option is to have the spillback header connected to the low-pressure column somewhere upstream. I have seen this type of solution successfuly implemented elsewhere but the problem here is that we have only a 2" line and 2" nozzle on the column, for the maximum recycle flow of 170,000 kg/hr. This section is 2 meters long only but according to pressure drop calculations this would create the loss/backpressure of around 10 bar. As the pump discharges at 80 barg and the column operates at 2 barg, having such high pressure drop does not represent a problem just by itself, but the velocity in 2" line and in the nozzle would be more than 20 m/sec which is extremely high.

Can you give your views on these two items? And perhaps come with additional suggestions that might enlighten us.

Regards,

http://antwrp.gsfc.nasa.gov/apod/astropix.html

 

[TD2K]

I'd look at the effect of the maximum recycle rate on the NPSHA for the pump, you'll be heating up the fluid with the recycle stream. I don't like recycling directly to the pump suction because of the heating effect (which will be worse with a high head pump) when you are on full recycle. You say the recycle shouldn't be more than 50% but what about startup or an upset?

I wouldn't accept 20 m/sec in a line for a liquid.

 

[hacksaw]

ar eyou recyling a flammable liquid?

 

[Latexman]

And, is it conductive or non-conductive?

Good luck,
Latexman

 

[EmmanuelTop]

The stream in question is 50% MDEA solution, and the pumps are lean amine charge pumps.

@TD2K: Wouldn't the suction and discharge temperature progressively rise if there is no cooling of the recycle stream? And yes, you are right, startup/upset are another issues which might make things even worse.

http://antwrp.gsfc.nasa.gov/apod/astropix.html

 

[TD2K]

The temperature wouldn't continue to increase unless you were on full recycle because you are still bringing fresh 'cold' fluid that will carry away the energy added by the pump.

Let's say your pump has a 5F temperature rise across it (unlikely to be that high unless it's a very high head pump and low efficiency). Now consider 50% of the discharge flow is recycled back to the suction. The overall temperature rise across the pump will be 10F and the 50% forward flow is removing the same amount of energy from the system (the pump).

Why isn't a single recycle line working as it's currently designed? Do you routinely run both pumps? If you are, how come you need to recycle flow then?

 

[EmmanuelTop]

The existing single recycle line and a single flow control valve do not work well because there is significant difference in frictional resistances of suction/discharge piping of the two pumps, resulting in one pump working harder than the other one. Their Q-H curves are very much flat so only a slight difference in system resistance creates large difference in flows delivered by each pump. Normally we see this problem when switching between pumps, and we suddenly lose forward flow to the process, and the unit trips.

I am not sure I understood your statement regarding the heat out-flow from the system. Just imagining the process in my mind. First run through the pump: 2 degC temperature increase, then we recycle 50% of this flow back to the suction. The suction flow coming from the process has the same temperature, so the combined suction temperature will increase due to warmer recycle stream. Now, we have higher discharge temperature than in the first run, and higher combined suction temperature, and so on... To me it looks like the temperature rise inside the system could continue without any control. Does that make sense?

http://antwrp.gsfc.nasa.gov/apod/astropix.html

 

[TD2K]

Let's look at your example.

Say the cold stream is at 20C. It goes through the pump and exits at 22C, 2C rise across the pump, and goes off to the process.

Now, you recycle 50% back to the suction. The cold fluid is still 20C but it's mixing with the recycle stream which is 24C now (will get to that number next). The combined suction to the pump is the average given a 50% recycle or 22C. You go through the pump and pick up 2C giving a discharge temperature of 24C, the figure I used at the start of the paragraph.

With no recycle, the energy added to the net stream as heat is Q*2 deg C*Cp (Q is the flow and Cp is the heat capacity, don't worry about the units for this example).

For the 50% recycle case, the net flow from the system is 1/2Q*4 deg C*Cp which is the same as with no recycle. It should be the same since the pump flow and its efficiency will not have changed.

 

[EmmanuelTop]

TD2K, thank you for your analysis. I have made a simplified dynamic model of the system and confirmed your observations. As long as there is any forward flow, heat will not keep accumulating in the system.

Since this loop cannot work in 100% recycle due to pump overheating, is there any point to size the recycle line for 100% flow? If I use the design minimum flow of 200 t/hr I get 6" line size required to maintain velocities below 4.5 m/sec (15 ft/sec). For a 4" line the line velocity would go up to 6.5 m/sec. Is this acceptable as a short-term operation? Normally the 4" line will never see velocities higher than 4.0 m/sec.

I have attached the dynamic model, maybe it will be of use to other forum members.

Thanks again.

http://antwrp.gsfc.nasa.gov/apod/astropix.html

 

[TD2K]

I don't have Hysys on my home computer so I can't see the simulation flowsheet.

The recycle line should be designed to meet the pumps' minimum flow requirement which should be on your data sheet or is available from the vendor. Since you say that you operate two pumps at least part of the time, the minimum flow is 2x the individual flow. If you can't take credit for the minimum flow to the process then this needs to be the sizing basis for the recycle line. Then you can evaluate where it goes, directly to the pump suction, to another location where the stream can cool, etc.

 

[EmmanuelTop]

Two pumps run simultaneously only during changeover i.e. switching between the online pump and the standby pump. My reasoning is - since the pump cannot run at 100% recycle on a continuous basis - the minimum flow line should be sized for the "normal" recycle flow, for which the 4" line is more than adequate.

The only true concern is if there are design scenarios when the pump is supposed to run in 100% recycle mode for a prolonged period of time, e.g. startup or upsets. Then recycling without cooling is meaningless as the pump would need to be shutdown after ~5 minutes due to danger of overheating, according to dynamic simulation.

http://antwrp.gsfc.nasa.gov/apod/astropix.html

 

[EmmanuelTop]

A sketch showing the current and the proposed arrangement is attached below, for easier understanding.

http://antwrp.gsfc.nasa.gov/apod/astropix.html

 

[hacksaw]

are you putting "sight glasses" in your recyle lines?

that is a bit unusual given the 20 m/s flow condition isn't it?

 

[EmmanuelTop]

Hacksaw, the velocity should be far below 4.5 m/sec for normal operating conditions as only one pump runs at a time; two pumps run during changeover only, for a very short period.

Maybe I didn't understand your comment in a correct way?

http://antwrp.gsfc.nasa.gov/apod/astropix.html

 

[hacksaw]

just a concern regarding high velocity flows, just make sure the glasses are rated for the service