RSS Feeds
The Case of the Bleeding Bellows
True Stories
Kenneth Russell, Professor Emeritus, MIT, Cambridge, MA -- Design News, April 26, 2004
I keep my water heater set at 140F to provide hot water, regulating my shower temperatures via a rotary mixing valve. Should this control system fail, sending forth a stream of scalding water, I am still agile enough to get out of the shower post haste. Not all shower users are so mobile.
Some years ago a seriously ill man was checked into a Boston area hospital for neurological testing and given a shower as part of his hygiene. A hospital staff member put him in the shower and, inexplicably, left him unattended.
The hospital hot water line had a temperature-sensing mixing valve intended to always keep the water temperature below the scalding range. This device was inoperative when the patient was put in the shower and scalding hot water sprayed over the unfortunate man's body. He was not sufficiently competent to escape or even call for help. The resulting burns delayed diagnostic tests until they were too late to do any good. His family understandably sued.
My friend and colleague, Professor Robert Rose, was retained by the hospital, I by the plaintiffs. We agreed to do a joint study of the evidence to find the cause of failure.
The heart of the controller was a brass bellows about the size and shape of a pint can. Brass end caps had been soft-soldered in. The working fluid inside was (I believe) ethyl chloride, which has a vapor pressure of 1 atm. at 55F and 5 atm. at 150F, temperatures that nicely bracket the desired control temperature of around at 100F. I did a little temperature test and found that 105F is hotter than I want in a shower and 140F was hotter than I could stand for more than an instant.
The bellows was full of water when it was removed from service. There was clearly a leak; the question was where and why. We pressurized the bellows with a rubber squeeze bulb and immersed it in water. Initially we saw no leaks, even at a location in the solder joint where an earlier expert had claimed one existed.
I got out my pocketknife and very carefully removed some of the crud (corrosion product) from the suspect region without damaging either the solder or brass. We repeated the leak test and sure enough, there was a stream of bubbles coming from the cleaned region. My lawyer-client was so delighted he almost hugged me.
The crud was found to be composed of chlorides of zinc and copper (from the bellows) and tin and lead (from the solder). Microscopic study of the joint showed de-zincification of the brass, whereby the zinc corrodes, leaving behind a porous mass of copper. (For a thorough discussion of de-zincification in a past column, see DN 02.02.04, http://rbi.ims.ca/3848-535) I believe that the leak lay at the interface between the solder and dezincified brass.
|
The manufacturer had used a chloride flux during soldering, which may be OK, but the bellows were immersed into a hot water environment with the flux still in place, which is not OK. The joint should have been cleaned of the residual flux, which would not have been a difficult job.
The story gets even worse. The manufacturer knew of a clutch of service failures of this particular regulator and knew of the culpability of the corrosive fluxes. They went to a non-corrosive flux and a more sensitive leak detector, but never had a recall of the earlier regulators.
In addition, the regulator should have a fail-safe design, so that failure would give the unwary user a blast of cold water, rather than hot. The manufacturer had a report on designing the subject regulator to fail safe. It was listed as "not available" in answers to inquiries.
My client told me that the case settled very much to his satisfaction but I have no idea of the dollar amount.
Talkback
DN's Resource Center Get Free Information, Made Easy
-
Understanding a Modular Instrumentation System for Automated Test
National Instruments | White Paper
Request Item -
Advanced Features of High Speed Digital I/O devices : White Paper Series
National Instruments | White Paper
Request Item
