What are the potential dangers of using split flanges in process pumps?

[rollingstein]

I was reading an accident report where a process pump leaked from its delivery line through a "split flange assembly".

The report further mentions: "Split flanges are not a good feature because they expose twice as much surface to the effects of corrosion"

What exactly is a split flange?

Googling reveals something like this but I don't see the increased surface exposure here.

 

[UltrafastPED]

An ordinary flange is a continuous ring; when it is correctly tightened it forms a uniform seal.

A split flange is only used when the geometry of the system prevents use of an ordinary flange - for example there is an obstruction which prevents use of the continuous ring without dismantling the system.

This flyer on "flange management" may shed some light:
http://www.jameswalker.biz/en/pdf_docs/11-flange-management-guide

and then: http://www.olympus-ims.com/en/applications/crevice-corrosion-testing-sealing-surface-flanges/

 

[AlephZero]

I think the basic point (which is implied in your picture, if you notice the bolts), is that a simple flange has only one mating face between the components being joined, but a split flange has more than one.

"Exposed" faces may be a bit misleading. Bad stuff can get between the mating surfaces because of capillary action, internal pressure, vibration (fretting), insufficient bolt clamping loads, poor geometrical tolerances and surface finish, etc. "Exposed" doesn't necessary mean ""the surfaces that you can see".

 

[rollingstein]

I think the basic point (which is implied in your picture, if you notice the bolts), is that a simple flange has only one mating face between the components being joined, but a split flange has more than one.

.

I see. But if you mean the area where the semicircular parts will mate, isn't that tiny in comparison?

 

[256bits]

I was reading an accident report where a process pump leaked from its delivery line through a "split flange assembly".

The report further mentions: "Split flanges are not a good feature because they expose twice as much surface to the effects of corrosion"

[/PLAIN]

Does the report mention whether the connection failed due to corrosion or incorrect assembly, and give reasons why to point in either direction. If the report writer is not backing up the method of failure, then that statement would be just his own personnal unnecessary opinion.

Split flanges are more susceptible to uneven gasket load due to overtightening and uneven tightning of the bolts and their rated load is thus less than a welded flange. As for the corrosion part, contaminants can get in at the back end of the flange surrounding the pipe wheras a welded flange has that part sealed, and also the inner part of the split flange is not welded to the pipe either. For that reason, you do have extra areas where such modes of failure as Alpha Zero mentioned that could lead to failure.

 

[rollingstein]

Here's a link to the relevant section of the original report

Trvor Kletz What Went Wrong

He seems pretty unequivocal about "split flanges are a bad feature". He also says they expose twice as much area to corrosion. So maybe I'm missing something.

 

[256bits]

He could be talking about just that type of design which seems to have three gaskets ( or some type of o-ring ), which I assume are the whitish objects.

The bottom flangs is the split-ring which when bolted to the top flange compresses a gasket against the ( what looks to be ) plastic mating part of the pump body. I imagine the top flange might be connected to a pipe in a similar manner. Bolting the two solid flanges together would seal the pipe to the pump assembly. So rather than just 2 places where a gasket seals two solid flanges, this one has 4 places on either side of the solid ring flanges. In which case I believe he is just talking about a particular design. At the very least that is my interpretation with just a picture to go on and would have some percentage of error.

I was thinking more of this type of design:
http://www.marzolf.com/index.html
click on How It works for a tiny video.

In any case, it seems that the failure even if at that connection location, was due more to irregular maintenance, lack of record keeping, and a lack of identification of critical elements. Assembly of that part just lookin at it seems cumbersome and painstaking to get it right. An overtightening of a bolt, or loosening of a bolt at the flange via corrosion, and it looks like that the connection would be misaligned.

As a side note it says that the backup or spare recycle pump failed after 25 minutes. From what the author states, it makes me wonder when the last time previous to that event that the backup pumping system had ever been serviced, even if not in use, to check for corrosion or gasket integrity.

Hopefully that helps. It would have been nice of the author to have provided a reference, instead of trying to second guess him, and I will admit that my admission may be out in left field, as I do not have a reference either.

 

[rollingstein]

It would have been nice of the author to have provided a reference, instead of trying to second guess him, and I will admit that my admission may be out in left field, as I do not have a reference either.

There is a ref., a 1977 report but I couldn't find it.

Your analysis does help. I'm chewing on it now.

 

[256bits]

Here is a report i believe of the incident from HSE.
http://www.hse.gov.uk/comah/sragtech/caseoctel94.htm

And a debate in the House of Commons ( farther down under Ellesmere Port (Explosion)
)
http://www.publications.parliament.uk/pa/cm199394/cmhansrd/1994-02-11/Debate-9.html

A full report from HSE
http://www.icheme.org/resources/safety_centre/products/~/media/Documents/icheme/Safety%20Centre/HSE%20reports/FireAtAssociatedOctelCompanyLtd.pdf

Which I found from
http://www.icheme.org/resources/safety_centre/products/hse_accident_reports.aspx

Item 63 state that the corrosion of the flange could have been caused by the releasing of the chemical and the raised temperature of the fire. they decided to go though with a priori corrosion of the flange after doing some tests.

Item 64 discusses misalignment and vibration issues with pump-motor drive, improper clamping of the motor to its base, and improper housing assembly.

Item 65 describes the joint as being flexible with a PTFE bellows between the flanges. It insists upon a proper maintenance and inspection..

Item 66 discusses that HSE choose to list the flange as the source of failure rather than the PTFE bellows, although Item 124 mentions either could have been the culprit.

Item 123 gives an assessment of the underlying cause of the fire onwards from the initial release of chemical.

 

[nvn]

He also says [split flanges] expose twice as much area to corrosion. So maybe I'm missing something.

rollingstein: For a split flange, I currently got ~2.11 times as much surface area exposed to corrosion, compared to a solid flange.

I tried doing a comparison in the attached file. I only show one half of the pipe cross sections, due to symmetry. Figure 1 shows a solid flange. And figure 2 shows an equivalent split flange. I highlighted the relevant surfaces, for comparison between figures 1 and 2. The cyan surfaces are an exact match. The green surfaces differ.

Each surface is labeled with its relative length. I assumed the gasket interface surfaces do not leak, and thus do not corrode; therefore, I did not include the gasket interface surfaces in this comparison. I also did not include flange outside diameter ends, nor split flange ends, because these ends would need to corrode for a long time, before they would have much effect.

Instead of trying to show the split in the split flange in figure 2, I show the split in figure 3. I currently assumed each split covers (0.10*2*pi) rad = 36.0 deg.

Underneath each figure title, I show the total surface area of the highlighted surfaces in each figure. Notice that 2*pi*r divides out of each total. Therefore, the relevant, exposed surface area of figure 2 plus 3, divided by the surface area of figure 1, is currently, ratio = A2/A1 = (626 + 49.6)/320 = 2.11.

Of course, you could get a more accurate ratio if you perform more detailed, careful calculations than I did, or if you draw a particular connection design on cad, and then sum the relevant, exposed surface areas on cad.

 

[rollingstein]

@nvn

Very nice! Thanks.

Out of curiosity, what did you sketch this in. Looks very neat.

 

[nvn]

MS Paint.

 

[rollingstein]

MS Paint.

duh! Must have been painstaking work.

 

[nvn]

Some split flanges use a different bolt for the split flange overlap tangs, as shown in figure 3 in post 10 (and as shown in post 1), whereas other split flanges do not use a different bolt for the split flange overlap tangs, as shown in figure 3 in my attached file, below, and as shown on the pump in post 6.

I would say the split flange overlap tangs might create a weak spot, causing uneven clamping force on the gasket or seal. I think 256bits alluded to uneven gasket loading. And because the overlapping tangs are already thin, a small amount of corrosion of these tangs could make them even thinner and weaker rather quickly.

Aside: Those split flange bolts in figure 8-3 in post 6 seem to have significantly less than 1*D thread engagement length, threaded into possibly soft material, perhaps mild steel, in which case it might be easy to start to strip those tapped threads, if you overtorque the bolts, further contributing to uneven gasket loading.