15 replies to this topic

[Bruno Credidio]

Dear Art,

 

I read some posts in this forum about atmospheric tanks' overflow line design and than I read Mr. PD Hills article, but I can seem to relate this article to overflow lines. I'd like to know how to apply Hills' article to overflow lines.

 

Sorry for disturbing you and for the poor English, I'm from Brazil.

 

Thanks.

 

 

[katmar]

Bruno,

 

What really do you expect Art to do for you?  Do you want him to re-write the Hills article in simpler terms for you to understand?  How is Art to know which are the aspects you are having trouble with?

 

The members here (and especially Art) are always generous with their time and expertise, but they want to know that you have made a reasonable effort before you throw up your arms and say "Art, you do it for me".  Show us the problem you are facing.  Explain what calculations you have done and where you are having difficulties. Make a flow sheet to illustrate the problem.  Then we know that you are serious and where your problem lies.

 

The most important lesson any engineer can learn about problem solving is that the problem statement must be clear.  If you cannot define the problem then you cannot solve it.  If the problem is well stated, it is half solved already.  Take the time and effort to state your problem clearly and you will find that your understanding improves significantly.  And people will be able to help you.

[Bruno Credidio]

Katmar,

 

Sorry about my poor post, in no way I was trying to ask you guys to do my work, I'm just trying to learn something from you, thanks for the advice, I'll try to make myself clearer.

 

I'm interning in Chlor-alkali plant in my city. There are 4 equal atmospheric tanks that receive refrigeration water from HCl reactors to pump it back to the cooling tower, since the reactors are equal the tanks receive the same flow of water, 447,3 m³/h in a line of 12''.

 

But here is what I found strange, 3 of those tanks have an overflow line to a drain of 8'', and the other has a line of 14''. So I decided to try to redesign those lines to see if they fit with the inlet flow. So I came here and discovered about the Hills' article. If i understood it right I should use equation 4, right?

 

"Side-outlet piping - Coming off from the side of a vessel, piping should be sized such that:

 

J_L<0.3 ; d> (4.Q/(0.3.pi.g^1/2))^{0.4 "}

 

I used excel interative calculation to calculate the minimum diameter needed to fit the condition of J_L<0.3 and I founded a ID=19,3''.

 

Does it sound right?

 

The tanks are like the attached drawing.

 

Thank you very much.

Attached Files

•  Tank.jpg   4.28KB   17 downloads

[Art Montemayor]

Bruno:

 

It’s good that you agree that clear communications is the proper way to consult with other engineers.  If you are trying to learn, that is the fastest and most efficient way for you to carry that out.

 

When it comes to Fluid Mechanics, Katmar is the recognized expert.  But even Katmar can’t contribute a useful or meaningful response or comment if you don’t furnish ALL THE ENGINEERING DETAILS involved in your application.  No matter how simple the application may seem, it still requires a detailed explanation complemented with illustrations or diagrams.  Your latest post still lacks the necessary details to allow an expert like Katmar to analyze and give you his valued comments on your query.  For example:

• Why haven’t you furnished a detailed flow diagram showing the refrigeration water flow into each of the four holding tanks, their locations, the size of each line, the dimensions of each line, and the locations, heights, etc.  Your sketch of one tank gives no useful information.  We are not interested in the tank's level gauge.
• What are the details of the overflow outlets on each of the tanks?  Is an eccentric reducer used as the outlet nozzle?  Or is a box-type of outlet used?  What are the sizes and types involved?  Is the overflow to be un-flooded and self-venting?
• What is the amount of overflow water used in the worst case scenario calculation for each of the tanks?  Are all the overflows identical in type and configuration?
• Are the four holding tanks connected together through piping located in their lower portion?
• You state that “the tanks receive the same flow of water, 447,3 m³/h in a line of 12””.  But does this mean that each tanks receives 447,3 m³/h of water from a common header of 12”?  A detailed sketch would be a simpler, detailed, and accurate description.

I am attaching a copy of my formatted version of P. D. Hill’s article and you can use this excel workbook to generate a detailed flow diagram, together with your engineering calculations.  Title it as Revision 1 when you submit it to the Forum in your subsequent reply.  You say you have made the calculations, so please submit them in spreadsheet form for our Forum members to review them and comment.  This is what engineers do and how they communicate.

 

 Designing Piping For Gravity Flow - PD Hills 1983.xlsx   101.74KB   204 downloads

[katmar]

Bruno, the way an engineer reacts to advice is often a good indication of their abilities and their self-belief in those abilities.  Your positive response to my criticism suggests you have a successful career ahead of you.

 

Art has indicated what the ideal formulation of the question would be and you should strive for that level of excellence. However, with the information you have provided we can certainly start the conversation and work towards an answer.

 

Without actually repeating your calculation, it looks correct to me that you would require an ID of at least 19.3 inch to satisfy Hills Eq 4.  But you need to take a step back and ask yourself if you really need self venting flow here.  I'm not saying yes or no - just saying the question should be asked.  You have a tank with non-toxic (I presume) cooling water and the tank has a vent to atmosphere. If the level in the tank rose above the side outlet, would that be a problem?

 

A simple Darcy-Weisbach type calculation for the 8" outlet (assuming an entrance, exit, 2 elbows and 10 m of pipe) shows that you would need about 2 m of head above the outlet to provide that flow.  It is unlikely that a tank would have this much free board above the overflow outlet.  What you certainly do not want is for the water level to go up into the vent of a flat bottomed tank.  You will suddenly find the bottom of the tank becoming dished and the tank behaving like the leaning tower of Pisa.  The 14" outlet would require only 0.25 m of head and this would be more likely to be satisfied in a typical design.

 

On this very simple analysis it would appear that there is a potential problem with the tanks with the 8" lines.  Ask the operators if the tanks have ever overflowed.  Are the bases dished? This would be a good indication of previous overflow situations.

Edited by katmar, 06 November 2015 - 02:21 AM.

[Bruno Credidio]

Katmar and Art,

 

Thanks so much for your time and answers.  I've been reading posts from you for a couple of months and all your advice is of great value to me.

 

I think I'm getting ahead of myself here.  Can you answer some questions before we can continue to discuss the topic?

1. You linked the non-toxicity of fluid with the kind of flow.  Are there any recommendations for the kind of flow in overflow lines?
2. You said that I could use the Darcy equation to calculate the head of liquid over the line. Would it be the same if I used Eq. 3 from Hills' article?

Hope I'm not asking too much of you,

 

Thanks again.

Edited by Art Montemayor, 07 November 2015 - 11:20 AM.
spelling; composition

[katmar]

The only reason I mentioned the toxicity of the fluid was that it means that there is not a problem if you have a bit of breathing in and out the vent and drain. If you have read some of the previous discussions on tank overflows here you will have seen that sometimes it is important to keep the vapour sealed from escaping.  If you did provide a 20" overflow to satisfy Hills' Eq 4 then it would not matter if the discharge end was not sealed. For toxic gases that need to be retained in the tank then the discharge must be sealed (eg with a gooseneck) and allow only the liquid to flow out.  The internal pressure of the tank also has a bearing on all of this.

 

Hills' Eq 3 is a different consideration from what I was trying to do with D-W.  Eq 3 gives the criterion for the minimum level to ensure that the outlet is flooded.  I was looking at the level required to drive the liquid out the side outlet.  The height calculated by D-W could be more, or less, than the Eq 3 height.  Eq 3 involves the flow rate and outlet pipe size only. To calculate the head required with D-W you need the full details of the outlet pipe (length, changes in size, elevation change, fittings etc etc)

 

A potential problem when allowing the level to go above the side outlet is shown in Hills' Fig 1 (although he is using a bottom outlet).  If the level gets high enough it may cause sufficient flow to initiate a siphon.  In this case the level would be rapidly drawn down because the out flow would increase.  This can result in cyclical behaviour, which may or may not be a problem.  I know of a filter tank that has been cycling like this every 30 seconds for the last 25 years, with no adverse effect on the process at all.  In other cases it may be important for the out flow to be steady.

[Bruno Credidio]

Katmar,

 

I think I understand it now. As it's only refrigeration water that is stored in my tank there's no need to guarantee a self-venting flow, I'd be only spending more money into a larger overflow diameter.

 

About the cyclical behavior, I guess that it won't affect my tank, because it won't occur in normal operation conditions. I've asked the operators and the tanks have never overflowed.

 

I'm attaching a spreadsheet with my calculations using D-W e Bernoulli's equations.

 

Thanks so much for the support!!

 

 

Attached Files

•  Overflow.xlsx   14.85KB   127 downloads

[samayaraj]

Dear Mr. Katmar,

 

I have seen Bruno's excel. In that, overflow pipe head (H₂ as indicated in attached excel sheet) will be added as driving force for liquid to come out. I found out maximum velocity as V = sqrt ((H₁ + H₂ - H_loss) x 2 x 9.81). Am I right? I have considered the case as flooded. Please look at the attached excel.

 

(H₁ = Water head above outlet nozzle, H₂ = Height of overflow pipe)

Attached Files

•  Overflow line sizing.xlsm   32.94KB   104 downloads

Edited by samayaraj, 11 November 2015 - 04:59 AM.

[Bruno Credidio]

Samayaraj,

 

I think that only the head of fluid above the overflow counts as driving force. This driving force is the pressure exercised by the fluid above the nozzle (Stevin's Law),the amount of fluid below the nozzle doesn't matter.

[samayaraj]

Bruno,

 

There is a difference if the overflow nozzle is let out horizontal and down. Since, the water in the down side will fall due to gravity and an equal amount of vacuum will be created in the nozzle area. This helps in driving the water. If the vacuum falls below the vapor pressure of water at opt temperature, vapor formation will obstruct the flow. I believe this is the correct approach. Experts correct me if I am wrong.

[Art Montemayor]

Bruno & Samayaraj:

 

Fellows, it's great that we have constructive discussion over this engineering application, but lets communicate like engineers do: in a logical, accurate, and documented manner.

 

I recommended to Bruno to generate his Excel calculations on the submitted PD Hills workbook and label them with a revision number 1.  This was not done.

 

What I was trying to avoid was a generation of various workbooks within the same thread, with different titles, and no continuity or direct access to the referenced PD Hills article.  Now we have 3 workbooks with different titles trying to deal with the same topic.  This is NOT the way engineers organize their on-going discussions and resolutions - together with referenced information and calculations.  They generate ONE document that is continuous, referenced, and - most importantly, labeled with chronological modifications, changes, and additions.  That way, everyone is talking about the same thing and there is continuity - and hopefully combined resolution.

 

I have taken the liberty to combine all three submitted workbooks into the one, Rev2, that is submitted below.  I hope we all use this common document and means to organize and compile our discussions, calculations, and resolutions.  Any ongoing revisions should be so titled with the sequential rev number.  Lets get organized and consolidate our efforts.  What I'm trying to do is avoid any confusion and a variety of documents over just one topic.  If we agree, I can delete the other 3 workbooks, leaving this one as the central, solitary document.

 

Bruno:

I think what Samayaraj is trying to state is that the type of nozzle used on your tank to allow an overflow under flooded conditions is one that will not trap air.   That is why I mentioned an ECCENTRIC REDUCER type as the outlet nozzle.  When installed properly, this will allow any air to be purged or replaced with ease.  This is a suggestion to ensure flooded liquid outflow through the nozzle and down the outlet pipe.

 

I await your comments, objections, criticisms, suggestions, etc., etc.

 

 Overflow Line SizingRev2.xlsm   124.77KB   174 downloads

[samayaraj]

Mr. Art,

 

Thanks for you valuable comment. What you have done is absolutely correct. Your comments helps young engineers to work more professional. Further corrections we will make in the same excel as you mentioned. Thanks again for correcting our mistakes.

 

 

Bruno:

I think what Samayaraj is trying to state is that the type of nozzle used on your tank to allow an overflow under flooded conditions is one that will not trap air.   That is why I mentioned an ECCENTRIC REDUCER type as the outlet nozzle.  When installed properly, this will allow any air to be purged or replaced with ease.  This is a suggestion to ensure flooded liquid outflow through the nozzle and down the outlet pipe.

 

 

What I'm trying to say is, the overflow line head H₂ will create a vacuum and that helps in driving the water. Please correct me if I am wrong.

[katmar]

The question of whether or not to take the vertical part of the overflow line into account is a bit tricky. Bruno has taken the length of this pipe into account as far as the friction losses are concerned, but has not taken the credit for the extra head that it would develop. Sam on the other hand has included both the losses and the head developed in the vertical section.

Who is correct? In a sense both are correct. Sam is correct in terms of the hydraulics, but in many cases engineers would agree with Bruno's approach because it is the more conservative one. In order to take Sam's solution the engineer would have to do a bit more calculation to verify it, and in many cases they would prefer to avoid the extra work and take Bruno's conservative answer. In post #5 above I took Bruno's approach because we did not have all the details of the overflow pipe at that stage and I made some conservative assumptions.

The way to consider this problem is to look at what happens over time. Let us imagine a situation where the level in the tank is below the overflow outlet. At this stage the outlet pipe (including the vertical section) has no water in it and is full of air.

Now assume that the pump removing the water from this tank fails but the water coming in continues at 447 m3/h. The level in the tank will increase. Imagine that the level in the tank has reached 1 cm above the bottom of the outlet nozzle on the side of the tank. Some water will flow out the outlet. The horizontal part of the outlet piping is not full, and the vertical section will also not be full. The water flowing down the vertical section will be in free-fall and will not fill the pipe.

How high does the level in the tank have to get before the vertical pipe is kept full of water? One way to approach this is to take the criteria for designing a siphon. In a siphon the water has to flow fast enough to carry any air in the pipe down the vertical section. There is not perfect agreement on what the required flow rate for this is, but pretty much everyone agrees that if the Froude Number is 1.0 or greater the air will be flushed out.

Remembering that Fr = V /((gD)^0.5), to ensure that Fr > 1.0 we need V > (gD)^0.5. For a 0.2 m pipe the required velocity would be 1.4 m/s and the flow flow would be 158 m3/h. This is a promising result because we are working with an inflow of 447 m3/h.

But we cannot assume all is OK yet.  We have to work out how high the level has to get before this flow rate is achieved. To do this we ignore the vertical section completely. Why do I ignore it? Initially, the water in the vertical section is in free fall so there is no friction loss, and because the pipe is not full we cannot get any head recovery. There will come a point where the pipe is almost full and there will be some friction and some recovery, but my experience tells me that in vertical downflow the head gained is greater than the losses except under extreme velocities - so I will ignore the vertical section.

Now we must calculate the head required to force 158 m3/h through a 0.2 m diameter pipe consising of an inlet, 0.5 m of pipe, an elbow and an exit loss. This requires a head of only about 0.2 m - which would be measured from the centre line of the outlet nozzle and would mean the level is around 0.1 m above the top of the nozzle. At this point we can be sure that the outlet pipe is full and we can take credit for the extra head developed in the vertical section, but must provide for the friction losses as well.

As Sam has shown, at this point the outflow is much greater than the inflow and the level in the tank would drop rapidly. Once the level in the tank went low enough to no longer keep the vertical section full the driving force would decrease, and the flow would drop off. This often leads to the cyclic behaviour I mentioned before.

Now we have a complete picture of what could happen, and we must decide what size overflow pipe to use. In a situation like this the overflow pipe is a small part of the overall cost and if this were a new design I would probably select the 12" or 14" option. But in this case we are faced with an evaluation of an existing setup and we have to decide whether it is safe to leave it as-is. If my calculations above are correct, it is probably safe enough. At this stage it goes beyond an engineering question and becomes a risk analysis question where we have to consider what could go wrong if the overflow line did not cope.
 

Edited by katmar, 18 November 2015 - 05:53 AM.

[samayaraj]

Dear Mr. Katmar,

 

Now my insight on overflow line sizing got increased. Thanks a ton for your detailed explanation.

[Bruno Credidio]

Dear Messrs. Katmar, Samayaraj and Art

 

Thanks so much for your knowledge. Like Samayaraj said, my insight on overflow line sizing has increased too.