Spiral heat exchanger failure

[robsalv]

Hi folks,

We [located in Australia] have a spiral type heat exchanger that has failed in an outer spiral, mostly likely due to fatigue/cold work of the stainless sheet around a row of channel spacer pins. Crack is approx 400mm long, as noted by a fibre borescope visual inspection.

Anyone ever seen a spiral hX repaired before? The vendor has to send someone out from France to effect a repair... only thing is, due to the nature of the spiral hX, we're not sure we've found the only crack! And it's a way complex and involved repair...

Comments, experiences, thoughts?? Can a spiral hX ever be OK in a cyclic service?

Thanks in advance.

Rob

 

[finman]

Hi Rob

You havn't fully described the hX enough for me to be of much use to you, however I have built, installed and maintained tubular, spiral hX's and I guess when you think about it, all hX's are cyclic in one form or another.

Ensure that the design and construction are upto scratch especially any heat treatment that may or may not have been done.

Make sure that the unit has freedom to flex during operation.

Interrogate the vendor as to how they built the unit, this is where hX's start to fail!!!!!

 

[robsalv]

Thanks Finman.

This isn't a tubular spiral hX. This is a sheet spiral hX. Check out the following link:

<

http://www.process-heating.com/CDA/...story/BNPCoverStoryItem/0,3154,18383,00.html>

But I take your point!

The hX is about 5 years old, all wetted parts are 304ss/316ss and the hX is approximately 1.5m [5'] diam by 2.5m [8'] long. The channel is made from 2.5mm sheet. The channels are separated by 5mm diam spacer rods.

The hX is in a heat recovery service. We're guessing that poor tower bottoms level control resulted in excessive operating pressure/temp cycles causing the stainless channel sheets to flex around the channel spacer pins and the material has failed due to cold working and fatigue. The local Alfa Laval rep has only seen this once before in Australia, which required an Alfa Laval France tech to come out and effect a very long and protracted repair using a window welding technique.

My question is to anyone who has spiral operating experience, have you had one fail? Did you repair it? How?

AlfaLaval recommend a new hX with thicker channel sheet... but I don't think that will make much difference in resisting cyclic service. My thinking is that given the amount of restraint required to keep spiral channels from collapsing and the spiral fixed to shell, spiral hX's are not good for cyclic service.

Thoughts.

Taa

Rob

 

[unclesyd]

I just checked, several years ago we had a project that installed a large spiral plate exchanger in a service where the process was feed by a &quot;Bran-Luebbe&quot; 5 high pump. The initial exchanger lasted less than a week before leaks developed. We made no attempt to repair as the manufacturer wanted to replace it. It was replaced by one that had about twice the footprint of the original. It failed in about 6 months service. It was scrapped as it couldn't be cleaned.
A tubular spiral exchanger of an in-house design is now in service and has been for 3 years.

It was quite evident that the plate type spiral exchangers were not suitable for our cyclic service. The process conditions were 150°C in 178°C outlet; 25 psi in 250 psi out. I’ll try to get the pump rate.

 

[pvcguy]

Rob

At our facility (PVC plant) we too have had failures of our spiral exchangers so I can relate. Previous failures include cracking of welds on the centre spirals, cracking of welds at the inlet header and pinhole leaks at the spacer fusion welds and in the plate closure welds. In some of these instances we have had successful repairs and others were not. Once pinholes were found we scrapped the exchanger.
At present we have one unit (also an Alfa Laval, 76” x 72”, all wetted areas SA-240-S31803 2205) in service with 2 cracks in it. A 1000 mm crack in one location and 250 mm crack in another. Both of these are along the length of the exchanger and are at welds where differing thickness transition. Our investigation found this to be a more common problem with spiral heat exchangers. If during the bending of the plate contact is not made with all the spacer pins (especially at transition areas) the plate will flex in service and cause fatigue cracking of the welds in these locations as we have experienced. The persons I talked to was present at the bending process and made the manufacturer add longer pins where required.
We have a repair plan in place (as per Alfa Laval) to cut open the exchanger and “excavate” to the crack location, replace the material around the crack location and weld the plates back in place. While this is a time consuming venture the alternative (a new exchanger) is quite pricey. The biggest issue was to find a local repair facility that understood what we were talking about. The manufacturer is 5 days away and downtime is an issue so we plan to do this work with a local repair company. Even though we have had some issues with these exchangers its impossible to get the heat transfer or the clearances we need for the slurry service out of other designs.
I agree, due to the nature of the design you may not have seen all the cracks, that’s the risk you take.

 

[robsalv]

Thanks UncleSyd and PVCguy.

I'm not sure we're going to go the excavate and repair method on this thing. It turns out that the original assumptions were very very optimistic and the hX has underperformed since day one. Not sure what the organisation want to do but it looks like I have a few hundred kilos of scrap stainless.

Do alfa laval understand the limitations of their own design?? The spacer pins issue at metal thickness transition is an interesting one. I was thinking that the spacer pins needed to be a different design to avoid such a small pressure/pivot point. Don't know. Thicker sheet isn't really going to help fatigue issues too much either. Pity. Small footprint with huge surface area... it seems to me that spiral hX's are ideal if they fit into the very narrow design brief that they are ideal for. We have another two onsite that a merrily doing their job... but this particular one is heading towards the hearse.

Thanks again for the feedback to date.

Rob

 

[pvcguy]

Rob

As already wrote I spoke with an other company who had a representative at the Alfa Laval plant when the rolling took place to ensure the pins were long enough so there was no gap. The inspector there indicated this worked very well. Just the fact that Alfal Laval has a repair plan for this type of failure tells me they get more of this type of failure. It seems to me there are some design changes due to learning’s that are being incorporated into new designs. One of our other units had failed at the inlet nozzle where the long rectangular spacers had straight ends on them and the parent metal failed at these locations due to fatigue cracking. The new design calls for a radiused end to remove the stress riser.

Good Luck

 

[robsalv]

PVCguy, wouldn't a radiused end provide a smaller contact point thereby increasing contact stresses and possible fatigue issues??

My initial thought was to make the pins have a larger diameter, but this would then impact on pressure drop. The next thought was to have an hour glass shape, large diameter at the bearing surface, smaller in the flow path.

Don't know... it's a bit academic now though - still this is an interesting discussion.

Any additional thoughts?

Cheers
Rob

 

[pvcguy]

Rob

The radiused end was on the rectangular hold-down spacer bars found inside headers. These hold the shell together and provide strength for the header / nozzle. The ends of these were designed straight on a previous unit and now have been changed.
As for the spacer pins with larger diameter we had not considered that. We have seen the weld fail around these with pinholes not cracks. This was in our original 316L SS unit. We had repaired a crack at the inlet header and when we put the unit on test it looked like a humidifier. Scrap metal now.
Sometimes I think a lot of these issues could be solved with better manufacturing (usually welding) controls.

Just some thoughts eh

Larry (aka pvcguy)

 

[mk2000]

Well in reading the Alfa Laval website, they say that the shape of the hX promotes turbulence to achieve self cleaning action. This will contribute to the cyclic loading.