9 replies to this topic

[go-fish]

Please refer the attached file. In order to achieve crude oil dewatering in refinery storage tank, a crude oil shipment is allowed to stand for 24 hours. A pump is used to transfer the oily water at the bottom to waste water treatment until the inline analyzer detects oil instead of oily water and sends a signal to stop the pump.

As this is a dewatering operation, the pump is operating only until oil-water interface level. Water is assumed to be max. 1% of the crude shipment, and hence the crude liquid height is much more than water liquid height. Therefore, for NPSHa calculation can the static head available at pump centerline be assumed to be H-h?

Also, Table 5-16a in API 650 recommends drawoff sump dimensions and drain pipe size uptil 6”. In my case, 6” pipe is smaller based on the pump-out rate, therefore, is this going to result in more than one sump with all the drain pipes interconnected to make a common suction or the table in API 650 is just a recommendation and a bigger sump and drain can be used?

If Table is mandatory, and I still want to use one drain pipe, can I use 6” pipe even if it results in 1.5psi/100 ft pressure drop in suction instead of recommended 0.5psi/100ft, because there is till plenty static head (H-h= 15m approx.) available to compensate for high friction losses in suction line. The suction flow velocity will be 2.4 m/s for 6” pipe.

Attached Files

•  Dewatering.pdf   10.69KB   126 downloads

[kkala]

An opinion to the queries is as follows.
As this is a dewatering operation, the pump is operating only until oil-water interface level. Water is assumed to be max. 1% of the crude shipment, and hence the crude liquid height is much more than water liquid height. Therefore, for NPSHa calculation can the static head available at pump centerline be assumed to be H-h?
For NPSHa datum is suction centerline. Liquid pumped is water (density ρ) and overhead liquid is crude (density ρ'). Pumped liquid static head is Hρ'/ρ (minimum possible H to be adopted).
Suction pressure (abs) is 1.013+Hρ'g-ΔPf; this, minus vapor pressure at max temperature, makes the NPSHa, if expressed in height of water. Above g=gravity acceleration, ΔPf=frictional pressure drop along whole suction line (including the vertical part). Assume max hydrocarbon content in water and estimate resulting vapor pressure (sum of vapor pressures for immiscible liquids).
Also, Table 5-16a in API 650 recommends draw off sump dimensions and drain pipe size up to 6”. In my case, 6” pipe is smaller based on the pump-out rate, therefore, is this going to result in more than one sump with all the drain pipes interconnected to make a common suction or the table in API 650 is just a recommendation and a bigger sump and drain can be used?
The tank bottom seems to be cone-up, with peripheral sumps. These (say) 4-6 sumps are understood to be connected to a ring connected to the suction line. A cone-down bottom would probably make things easier in this case (especially if crude does not contain tar or other precipitating material). Suction line would not have size limitation.
I am not familiar with API 650, despite participating in tank design. I feel that any restriction in drain size by code, especially for nozzles close to bottom, would aim at not weakening the steel plates. But this could be cured through "reinforcements". Carefully search for such details in the code, logically drains of 8" size are possible with additional measures.
If Table is mandatory, and I still want to use one drain pipe, can I use 6” pipe even if it results in 1.5psi/100 ft pressure drop in suction instead of recommended 0.5psi/100ft, because there is till plenty static head (H-h= 15m approx.) available to compensate for high friction losses in suction line. The suction flow velocity will be 2.4 m/s for 6” pipe.

A guide recommends less than 3 m/s for 6" discharge pipe, so risk of vibrations or erosion is in acceptable limits for 2.4 m/s. The rule of 0.5 psi / 100 ft evidently aims at saving head for NPSHa, but you have much static head available. In conclusion 6" can be used in this case.

Edited by kkala, 07 May 2010 - 03:54 PM.

[Dev 009]

An another opinion for the crude oil dewatering is as follows.

I would not prefer directly run a pump from tank, rather than will have a collection pit optimally sized based on the quantity of water from crude, and run the oily water pump from it . To prevent the crude oil from interface into drain line, will put a oil detection valve ( based on density difference available in market) which will allow only water to flow out of it.

Regarding draw off sump please follow API 650, 6" dimension of drain line is optimum, also u can size drain header for 100m3/hr which is normally recommended. Size the header based on gravity draining. A 10" header after 6" water draw nozzle will probably be more than sufficient.

Regards

Dev

[go-fish]

The pump is going to experience the fluid which is mostly oily water. Most of which is water settled down by density difference and the pump is going to trip when the online analyzer detects more oil. Then isn't adding vapor pressures an overestimation of the actual vapor pressure at pump suction?

Is the ring for interconnecting the drawoff sumps in large tanks a normal practice? I haven't ever been to a site but if it is a normal practice, I think I can go with more than one sump and keep things simple.

I was talking of 2.4 m/s in 6" suction pipe not discharge pipe. I heard that there is a limit for max. suction velocity since high velocity may result in eddy formation whcih can cavitate pumps even when NPSHa is sufficient.

Edited by Art Montemayor, 11 May 2010 - 06:28 AM.

[ELEMAN]

Hello everybody:

I absolutely agree with Dev.

[go-fish]

An another opinion for the crude oil dewatering is as follows.

I would not prefer directly run a pump from tank, rather than will have a collection pit optimally sized based on the quantity of water from crude, and run the oily water pump from it .

This was the initial scheme. However, client asked for a closed system due to possibility of H2S exposure to operator in case of conventional oily water sewer.

[fallah]

Therefore, for NPSHa calculation can the static head available at pump centerline be assumed to be H-h?

No,because H as per your sketch is MAX CRUDE LEVEL and you should consider MIN CRUDE LEVEL as predetermined low low level of crude for NPSHa calculation.

[go-fish]

Therefore, for NPSHa calculation can the static head available at pump centerline be assumed to be H-h?

No,because H as per your sketch is MAX CRUDE LEVEL and you should consider MIN CRUDE LEVEL as predetermined low low level of crude for NPSHa calculation.

Actually, what I meant was since water is assumed to be 1% of crude shipment, the minimum working height during DEWATERING ONLY will be crude level(H) - water level(h).

I have considered low low level for NPSHa calculation of main crude transfer pumps but for dewatering, separate pumps are used.

[fallah]

In normal operation,you are right.But in maintenance case you may use dewatering pump for emptying of the sump in tank bottom, and you have to consider low-low level as a possible case in pump's NPSHa calculation.

Actually, you haven't any facility other than dewatering pump for full emptying the tank from liquid content.

[kkala]

Opinion to the queries on the subject:
The pump is going to experience the fluid which is mostly oily water. Most of which is water settled down by density difference and the pump is going to trip when the online analyzer detects more oil. Then isn't adding vapor pressures an overestimation of the actual vapor pressure at pump suction?
No, it is not overestimation. Water and oil is practically immiscible (solubility of one to other approaches 0), so vapor pressure of the pumped liquid is the sum of vapor pressures of water and oil at max operating temperature, irrespectively of the proportions. Raoult's law is not applicable here. If oil content were 0, then vapor pressure would be that of pure water.
Even in case of oily water, NPSHa is not expected to rise to a level too high for a pump (seeing that crude is not heated).

Is the ring for interconnecting the drawoff sumps in large tanks a normal practice? I haven't ever been to a site but if it is a normal practice, I think I can go with more than one sump and keep things simple.
The ring seems not to be a normal practice, I have not seen it or heard of it. One sump seems not to be a normal practice, either. Probably cone-down bottom instead of cone-up could settle the issue (see previous post by kkala).

I was talking of 2.4 m/s in 6" suction pipe not discharge pipe. I heard that there is a limit for max. suction velocity since high velocity may result in eddy formation which can cavitate pumps even when NPSHa is sufficient
The previous post meant that since 3 m/s has been recommended as limit for discharge pipe, erosion or vibrations (due to velocity) would be within acceptable limit in suction pipe of 2.4 m/s. Of course 2.4 m/s is not recommended for 6" suction pipe in practices, a limit would be 1.5 m/s or 1.5 psi/100 ft (whichever is stricter) according to a practice. We would accept higher values exceptionally, as long as static pressure was high enough to sustain required NPSHa.
Though not heard of it before, the eddy formation concept can have weight, seeing that eddies can create variation in local pressure and the minimum of these pressures should be considered in NPSHa calculation. Variation may not be wide in the specific case (Re~50000), yet assessment is not easy for me.
However I understand size higher than 6" can be used in this case (previous post).

At any case it seems now that this is a new installation, so some basic design concepts can be determined or revised. E.g. a cone-down tank would eliminate need of sumps and ring (I cannot tell operating difficulties of cone-down tanks versus cone-up tanks though).

Edited by kkala, 15 May 2010 - 01:44 AM.