3 replies to this topic

[Sloot]

We have two shell and tube heat exchangers that were just put into service in the last month. They are for process and make-up air heating with 45 psig steam in the shell and 40% propylene glycol through the tubes. The heat exchangers are simple horizontal u-tube with ¾” x 0.049” stainless steel tubes. The tubes were rolled into the stainless steel tube sheet and seal welded. One heat exchanger is about 18” diameter x 4 ft long and the other is 12” diameter x ~ 4 ft long.

From what I’ve read both have suffered classic water hammer damage resulting in ruptured tubes. The tubes that were damaged were on top of the tube bundle with a number of tubes appearing to be crushed. These tubes were not under the steam inlet but about half way down the length of the shell.

Google helped me find quite a bit of information about how to prevent water hammer with nearly all saying that the problem is poor condensate drainage. I only found one reference that attempted to explain what actually is happening in the heat exchanger when it is experiencing water hammer. It was on the Armstrong web site at:

http://www.armstrongpumps.com/Data/ioguide...UBE_I&O.pdf

From this reference:

“In a water heater using steam in the shell, when the demand for hot water ends the steam control valve closes, but there is a good supply of steam left in the shell of the exchanger. As this steam condenses, the pressure drops, often below atmospheric or even practically to full vacuum. This prevents condensate from leaving the shell and sometimes even syphons in condensate from the line beyond the trap. Now, when the steam valve opens again and admits steam to the shell, the rapid condensation, as it strikes the cold condensate, causes streams of water to rise, hitting the top of the shell and bouncing onto the top tubes. Sometimes the breaks in the tubes look as though a 4” spike had been driven through the topside. Other times the tubes may be crushed as if with a blunt chisel over lengths of a few inches or up to two feet.”

We experienced this exact type of damage so I am sure that the problem is water hammer but I am having a problem understanding how this could have happened in our arrangement. The big difference is that the capacity control of our heat exchangers is by varying the condensate flow rather the steam flow, i.e. in our arrangement our heat exchangers were designed to run flooded. There is an automated on/off valve on the steam inlet which is programmed to stay open during the heating season. We should therefore have constant steam pressure on the shell with the condensate level varying with the demand.

A good description of the various ways to control the capacity of steam heat exchangers is in an article found at:

http://www.driedger.ca/ce4_sh/CE4_SH.html

So if the automated on/off steam valve is staying open and we have constant pressure on the shell of the heat exchanger I don’t see how we could experience the rapid condensation that seems to cause the water hammer. I want to check the tuning of the condensate valve but I don’t see how even a big change in the condensate valve position could cause a large swing in steam load or condensation rate. The condensate control valve is 1 ½” (or less?) and the steam inlet is 6” I believe. The load on the glycol side should not change quickly as it is for heating outside air.

I’ve only read about this type of flooded heat exchanger control and have not used it before. I realize that many people will immediately say that you will always have water hammer when your heat exchanger is flooded like ours normally will be but I know that this type of control, although maybe not common, is used and used with success.

So where did we go wrong? Is this type of control prone to water hammer? Can it be avoided? Does this water hammer typically happen when the condensate level is low? Or high? We did not experience water hammer when these two heat exchangers were commissioned but it did happen later.

I can’t seem to come up with an explanation for the water hammering other than maybe, for some reason, the automated on/off valve on the steam inlet is opening and closing when it isn’t supposed to. This valve is programmed to open very slowly over 5 minutes but it is a ball valve so even if it is a crack open there will be a large flow.

Any comments, experience or suggestions would be appreciated.

I have attached a couple of pictures.

Attached Files

•  DSC02727.JPG   135.63KB   71 downloads
•  DSC02730.JPG   150.95KB   60 downloads

[JLMONTREAL]

We have two shell and tube heat exchangers that were just put into service in the last month. They are for process and make-up air heating with 45 psig steam in the shell and 40% propylene glycol through the tubes. The heat exchangers are simple horizontal u-tube with ¾” x 0.049” stainless steel tubes. The tubes were rolled into the stainless steel tube sheet and seal welded. One heat exchanger is about 18” diameter x 4 ft long and the other is 12” diameter x ~ 4 ft long.

This topic will be very helpful, if more inputs come. I have some commets as below:

1. Are you sure the heat exchanger is only 12” and 18” diameter? They look bigger in the pictures.
2. Are you sure the inlet steam pipe is 6” and condensate control valve is 1.5”? I don’t know the heat duty. It seems they don’t match each other.
3. Did you find big noise or shocking in the heater before the failure?
4. What is the design pressure and temperature of the tube?
5. I think the on/off valve helped damaging the tubes by causing water hammer. To me, to open this valve (6”?) in 5 minutes is too fast for 45psig steam; especially there may not be sufficient pre-warming before reopening this valve when it was closed by failure after some while.

[Art Montemayor]

Sloot:

This is a very interesting problem that you've run into. You furnish us with a lot of data, but we still need some clarifications and additional information:

• You state "our heat exchangers were designed to run flooded". I don't think that is the case at all. We have to be careful of how we interpret what you say. Your steam heater can't be "flooded", simply because flooding means 100% inundated. If your tube bundle is 100% inundated, you have essentially ZERO heat transfer from the available steam. You MUST have some tube bundle area exposed to the steam in order to heat your glycol. That fact simply provokes other questions like: what is the orientation of the condensate drain? Do you have a condensate pot external to the exchangers where the condensate level is controlled? Or do you control the condensate level within the heater shell?
• Is your tube bundle baffled? If so, how are you draining the condensate?
• Are you varying the condensate level around the bundle with instrumentation, or is it kept constant? Please show us your instrumentation loop.
• What is your heat load or what is your steam or condensate flow rate? I agree that a 1.5" condensate drain valve seem big; but we don't know the load, so we can only guess.
• I presume BOTH heaters were similarly damaged, at the same time, under the same conditions, and both are hooked up in parallel service. Is that correct?

You are referring to a top-notch instrumentation expert in Walter Driedger. I have, and will continue to strongly recommend anyone involved in process engineering to read everything he has written. Please look at the diagrams and sketches in his article and note how he distinguishes between external level control in a condensate pot and internal level control in the heater shell. Walter never uses the word "flooded" to describe the heat transfer he is controlling.

I have used the method you describe to heat reboilers in distillation columns and strippers. It works and it has worked very well in all my applications. It should work well in yours as well. But we need to know ALL the details of the application – especially the instrumentation loop and a detailed and accurate sketch of the installation and the piping.

I have applied this method wherever I have wanted to maintain a constant and steady heat transfer because it does not create a partial vacuum during steam valve throttling of other, conventional means of controlling the steam rate. I am surprised that you are using this system on a horizontal oriented heater. Normally, I would have installed the heaters vertically – as Walter Driedger has shown in his sketches. This is to take advantage of the varying of the heat transfer area that is flooded by the condensate level as fixed in the level controller. This is an advantage of this method of control in that you can turndown the heating rate by varying the surface being heated. ---- which explains why I object to your using the term "flooded" to define how the heater works.

My experience tells me that the heaters should work – and work well – using the type of condensate level control that you apply. However, as even Driedger states, there are some subtle differences that can appear and can cause havoc if the total picture is not addressed in the application. I think you should focus on your equipment orientation and the method and type of instrumentation that you are applying. You can, in my opinion, forget about the type of heat up being a factor. It works.

Additionally, the tubes you are using should be adequate for the design. I would not use a ball valve to control the steam flow. The valve should be throttling and a ball valve simply is not designed to do that. Use a conventional steam control valve.

We removed the damaged tubes and welded plugs into the tubesheet early last week. We then restarted these two systems. The start-up went very smoothly and neither of the heat exchangers made a noise .... but we started with empty heat exchangers, not flooded.

We disabled the two steam inlet valves and now have them held 100% open. These valves were not used for modulating service but just for on/off. The capacity control is from the modulating condensate control valve.

These two heat exchangers aren't in normal building heating service as the primary source of heat is from a process energy recovery system (ERS). This results in the steam being used to top-up the building heating system likely resulting in a number of start/stops in one week.

I wasn't there but there were a couple of occasions on the day after start-up when there was hammering and looking at the trends on the PLC, it coincided with the process ERS being down and the steam HX being called on to provide heat.

We had steam traps installed right above the heat exchangers on the steam inlet so when the steam was called for the heat exchanger would have been fully flooded up to the trap just above the heat exchanger.

I've talked to a few more equipment suppliers and I have now pretty much concluded that hammering cannot be avoided in the start-up of a "horizontal" flooded heat exchanger. I have also learned that others have been able to successfully operate flooded shell and tube heat but only if the heat exchanger is installed "vertically" with steam inlet on the very top of the shell with the bonnet on the bottom. See the attached sketch.

I will try to answer some of your other questions tomorrow. And sorry for the term "flooded" operation. Looks like I need a new (and correct) name for this type of steam heat exchanger capacity control.