Victaulic Couplings in Data Center

[tivester]

Twice within the last 5 years at two different data centers we have had a bolt fail on a Victaulic coupling of our chilled water piping. Both failures resulted in massive leaks and water damage. Luckily the damage was not to the data center floor itself, however similar piping is used under the raised floor of the data centers to the CRAC (computer room AC) units. As this is the chilled water line, all piping is fully encased in thick insulation with hard protective plastic surrounding the insulation which makes inspection of the couplings impossible to the naked eye. The chilled water piping system is now 15 years old. The bolts that failed appeared to be stressed at time of installation and eventually gave way, however this is a superficial and amateur evaluation of the bolt failure.

My concerns are:

1. What is the typical failure rate and mode of Victaulic couplings on chilled water piping?
Some forums depict grooved piping systems such as Victaulic as notorius leakers when compared to welded systems, while others speak to their life-long reliabilty. In either case there is very little data I can find about the actual life span of such piping systems. One site did relay an estimated failure rate for mechanical, rigid shaft clamp/compression type couplings of 14 failures per million hours (Weibull shape parameter = 2, charateristic life = 9 years, replacement interval = 15 years).

2. Is there a way to inspect couplings without destruction of the insulation?
Most CUI (corrosion under insulation) tools focus on corrosion of the pipe rather than the coupling. Inspection of the most recent pipe where the coupling/bolt broke showed absolutley no signs of deterioration. Whether there was moisture outside the pipe to deteriorate the bolt prior to the leak is hard to say. Some inspection using infrared has been attempted previously, but did not appear to show anything conclusive.

Thank you in advance!

 

[EnergyProfessional]

AFIK the rubber seals are wear items and need to be replaced every so many years. This obviously they don't tell you when you install them....
the only acceptable location for Victaulic or threaded fittings are sprinkler systems since they don't have moving water and don't have temperature changes.

In HVAC I only accept welded steel or soldered copper. Contractors always bring up those press fitting for copper and Victaulic for steel... and they all fail. Threaded pipe also is critical since the installer often is sloppy and especially with glycol you get leaks every once a while.

Life-long reliability.... whose life? the life of a fly, or the life of a turtoise? If they refer to the life of the fitting, yes, they lasted as long as the fittings lived.

If you are unlucky all of the fittings are of similar wear-age and situation. You probably want to remove insulation and visually inspect. Maybe an IR camera could help detecting cold spots outside the insulation (leaks of water reduce insulation value).

this also seems an installation issue, so the same guy may have over tightened the pipe more often. Or they don't have proper expansion loops?

 

[RossABQ]

First of all, were the failed couplings/joints truly Victaulic (brand), or something similar? I am surprised that this type of coupling would have been allowed for use under the RMF.

I would not use failure rate data for large chilled water systems in general facility use as a metric for your situation. I assume your mechanical joint piping to CRAC units is mostly 3 and 4 inch, and supported on sleepers on the floor?

 

[tivester]

HerrKaLeun: Thanks for your feedback. IR has been attempted in the past, but I am not sure of the quality of the camera used. I will have to check into the use of expansion loops.

RossABQ: Thanks for your feedback as well. They are Victaulic brand couplings. RMF? I am unfamilar with the acronymn. On Victaulic's website they have many case studies of data centers with HVAC chilled water systems installed with their product including under the raised floor. The original engineering firm for our data center and at least one other of the case studies on Victaulic's website was Bruns Pak.

As far as the failure rate goes, I was just wanting some general idea of typical age when failures occur and common types of failure. Since I was getting anything, scientific data or general rules of them would greatly help.

It's is 4" to the CRAC units. In the location where it broke this time, it happens to be an overhead line in the hallway between the two main switchgear rooms. Under the data floor I am not sure of the support system, I want to say it is a hanging type.

 

[RossABQ]

RMF = raised metal floor

Use of mechanical joint systems under the floor has some definite advantages (accommodating future changes) but the potential consequences of a leak ruled them out on all the jobs I worked. In normal facility systems (non-data centers), I don't remember ever seeing a joint failure on a system that was properly installed, properly supported, and operated within its design parameters (pressure).

I don't know that I would expect actual expansion loops on a system that doesn't normally have temperature changes, other than the change from ambient (75 deg) to CHW temps on start-up. Some accommodation may exist in a very long run of pipe. That would be an area to monitor if it did exist.