Failure and Life Cycle Evaluation of Watering Valves

Abstract

Automated watering systems provide a reliable source of ad libitum water to animal cages. Our facility uses an automated water delivery system to support approximately 95% of the housed population (approximately 14,000 mouse cages). Drinking valve failure rates from 2002 through 2006 never exceeded the manufacturer standard of 0.1% total failure, based on monthly cage census and the number of floods. In 2007, we noted an increase in both flooding and cases of clinical dehydration in our mouse population. Using manufacturer's specifications for a water flow rate of 25 to 50 mL/min, we initiated a wide-scale screening of all valves used. During a 4-mo period, approximately 17,000 valves were assessed, of which 2200 failed according to scoring criteria (12.9% overall; 7.2% low flow; 1.6% no flow; 4.1% leaky). Factors leading to valve failures included residual metal shavings, silicone flash, introduced debris or bedding, and (most common) distortion of the autoclave-rated internal diaphragm and O-ring. Further evaluation revealed that despite normal autoclave conditions of heat, pressure, and steam, an extreme negative vacuum pull caused the valves’ internal silicone components (diaphragm and O-ring) to become distorted and water-permeable. Normal flow rate often returned after a ‘drying out’ period, but components then reabsorbed water while on the animal rack or during subsequent autoclave cycles to revert to a variable flow condition. On the basis of our findings, we recalibrated autoclaves and initiated a preventative maintenance program to mitigate the risk of future valve failure.

The advent of genetically engineered mice has led to an exponential growth in mouse populations used in biomedical research and to fundamental changes to their health management and physical environment. Animal facilities have become efficient, large-scale operations and have adapted innovative ways to control environmental conditions and increase protection against adventitious pathogens³. As such, the majority of genetically engineered mice are now housed in ventilated, filter-topped, microisolation cages in high-density rack systems, which have extended the cage-changing interval from twice a week to as long as once every 2 wk.⁶ Automated watering systems have simplified operations greatly by removing time-consuming and labor-intensive handling of water bottles; in addition, we have seen a considerable decrease in ergonomic musculoskeletal injuries at our facility.⁵ Genetically engineered mice, which often express unsuspected phenotypes, can be difficult and expensive to create and maintain. Modern mouse husbandry and management must find the right compromise between operational efficiency and individual care of these precious models.

Our barrier facility has microisolation cages on individually ventilated cage racks which are equipped with an automated watering system with removable water valves on the animal rack. The watering system manifold delivers reverse-osmosis–purified water from a central filtration system to each cage port. Since the commissioning of our facility in 2002, the valve failure rate at our facility remained close to the 0.1% manufacturer standard based on the comparison of valves returned for assessment per valves manufactured per year. However, the incidence of valve failure dramatically increased in 2007 with no clear, identifiable pattern, causing animal losses due to both dehydration and cage floods. Here we present the investigation, etiologies, and changes to system management to ensure correct function of and restore confidence in our water delivery system.

Materials and Methods

Facility and animals.

The AAALAC-accredited institutional program supports several facilities including the Salk Animal Facility, which is a mouse barrier facility with a daily census of approximately 14,000 cages for about 30 research laboratories. Biosecurity, operation, husbandry, and health monitoring programs are optimized continually, based on contemporary standards to minimize potential outbreaks from adventitious pathogens and maintain an SPF environment. Briefly, access to the barrier facility is restricted by electronic card scanning and is granted only after a stringent training program. Traffic flow, including room entry order, is regulated strictly, and personal protective equipment (dedicated lab coat or gown, shoe covers, hair bonnet, gloves, and an optional face mask) are required for entry. Mice that leave the barrier facility cannot return. Mice are housed in microisolation cages on HEPA-filtered ventilated racks (Micro-VENT, Allentown, Allentown, NJ) with nonautoclaved, standard rodent diet (Laboratory Rodent Diet 5001, Lab Diet, PMI Nutrition International, Brentwood, MO), automated watering with reverse-osmosis–purified water, 1/4-in. corncob bedding, and enrichment bedding (Enrich'n Nest, The Anderson's, Maumee, OH). All mice are manipulated in an animal transfer station or BSL2 cabinet by using aseptic technique and disinfectant (Clidox-S, Pharmacal Research Labs, Naugatuck, CT). Cages are sanitized though a tunnel washer with 180 ^oF rinse water. Approximately 5% of caging then is autoclaved prior to use for immunodeficient strains. Almost two thirds of the mouse population is bred inhouse in established colonies. All imported animals either are purchased from approved commercial vendors free of adventitial organisms or are rederived and tested prior to release into animal housing rooms.

The health monitoring program consists of soiled bedding transfer to sentinel mice, which are tested quarterly for pathogens according to the following schedule. Each March and September, serology for mouse parvovirus 1, 2, and 3; minute virus of mice; mouse hepatitis virus; and epizootic diarrhea of infant mice virus was performed. These serologic studies were repeated in December, when the panel was extended to include Sendai virus, Mycoplasma pulmonis, reovirus, and Theiler murine encephalomyelitis virus. In June, the serology panel comprised mouse parvovirus 1, 2, and 3; minute virus of mice; mouse hepatitis virus; epizootic diarrhea of infant mice virus; pneumonia virus of mice; reovirus 3; lymphocytic choriomeningitis virus; mouse norovirus; ectromelia virus; mouse adenovirus 1 and 2; and polyoma virus. In addition, ecto- and entoparasite examination, fecal floatation; necropsy and histopathology of lesioned organs; and microbiology for Pasteurella pneumotropica, Mycoplasma pulmonis, Salmonella spp., and Pseudomonas aeruginosa were performed each June. For rederivation of transgenic animals, the serology panel comprised mouse parvovirus 1, 2, and 3; minute virus of mice; mouse hepatitis virus; epizootic diarrhea of infant mice virus; Sendai virus; mouse norovirus; Theiler murine encephalomyelitis virus; and Mycoplasma pulmonis.

Water valves and cage racks.

Microisolation cages in our facility feature a water valve (A160 Shielded Quick Disconnect, QD, Edstrom Industries, Waterford, WI) that allows for easy removal from the rack for sanitation (Figure 1 A). The drinking valve is composed of autoclavable 316-grade stainless steel. It connects to a water manifold installed on the ventilated rack and extends into the mouse cage via a flap-covered port. The valve was designed with 2 integrated moving parts that are visible from the front of the valve, the stainless steel stem and the green silicone shield. Movement of the stem from the center to the edge of the valve front, or toggling, allows water to flow. The silicone shield works to prevent the introduction of materials from the cage that could result in either leaky or no-flow conditions. Within the valve are 2 additional silicone components, the diaphragm and the O-ring (Figure 1 A). The diaphragm, seated on the flat head of the valve stem, applies pressure to the head of the stem. The bottom surface of the stem head is pressed against the O-ring, which supports the shaft of the stem as it passes through the inside diameter of the O-ring. Activation of the valve by animals requires 3 to 5 g of pressure applied to the tip of the valve stem. The shield is flexible and does not affect the actuation force of the valve. Water pressure at the valve is 4 to 5 lb to deliver a target flow rate between 25 to 50 mL/min (equivalent to a slow drip).³ Water is released when the animal moves the stem in any direction by biting or licking. When the stem is manipulated, the stem head comes off of the O-ring seal, allowing water to flow through 6 small holes in the diaphragm, down the shaft of the stem, and to the animal. When the animal releases the end of the stem, the elasticity of the silicone diaphragm located behind the stem head pushes the head back to the closed position, stopping the flow of water. The silicone rubber diaphragm and O-ring is designed to withstand temperatures to 300 °F, allowing autoclaving as a component of a routine sanitizing procedure.⁴ Both stainless steel and silicone rubber are noted for their resistance to the chlorinated or acidified water commonly found in animal watering systems.

Figure 1.

(A) Key silicone features of the A160 valve. (B) Normal diaphragm and distorted diaphragm with loss of elasticity and blister. Effect on O-ring after absorbing moisture and losing elasticity. (C) Valve with normal function. (D) Valve with gap in seal, allowing continuous flow.

Water valves were inspected for appropriate function at the time of cage placement, and all valves were removed from the racks for monthly sanitizing according to the manufacturer's recommendations: washing in a tunnel washer (alkaline detergent with 180 °F final rinse) and autoclaving at 250 °F for 15 min (58 min total cycle time: prevacuum cycle with 3 pulses and 15 min dry cycle). Valves were cycled across the entire mouse population of nonautoclaved and autoclaved cages. To decrease risk of potential horizontal pathogen transmission, all valves arweree autoclaved prior to reuse. The facility and valve processing program were inspected by representatives of the manufacturing company in 2007 and found to be appropriate. Equipment installation, maintenance and processing were according to the manufacturer's standards (Edstrom Industries). Due to the uniform design of the valves, no method was available for tracking of continually aging valves. As such, it was not possible to track the approximate number of wash and autoclave cycles each valve had received.

Ventilated mouse racks are equipped with a factory-installed stainless steel watering manifold rack flush system. The 1/4-in. stainless steel water lines allow for online flushing twice daily to 140 cage ports. Twice daily flushing prevents the build-up of biofilm or debris and provides a clean, potable source of ad libitum water. The watering system also removes reliance on large numbers of water bottles for the housed population and decreases the ergonomic risks due to the workflow introduced by water bottles.²

The problem: water valve variable flow.

Variable water flow (low flow, no flow, and leaky) from malfunctioning valves was difficult to isolate and often resulted in clinical signs and animal welfare issues due to flooding or dehydration. The uniform appearance of the water valves was a confounding factor in determining which valves were problematic. Drinking valves manufactured through 2007 were identical in appearance, having a green shield visible at the activation end of the valve. If a valve failing to provide acceptable flow was not identified and removed from circulation immediately, it was put through the sanitation process and returned to use within a few days. This inadvertent continuous cycling of dysfunctional valves increased the risk of exposure to a nonfunctioning valve.

Prior to the full-scale analysis conducted in our facility (beginning in 2007), several hundred valves were pulled off-line due to flooding and returned to the manufacturer for assessment. When these faulty valves were again placed on IVC racks, all valves were toggled and observed to function properly. Flooding was found in some cages after 24 h of valve placement. In other cages, normal flow was observed at toggling but 24 to 72 h later animals were found hunched, lethargic, and dehydrated. There was no way to reliably determine whether the valve would continue to function normally or develop flow problems. Researchers responded to these failures by demanding bottles to replace drinking valves and compensation for loss of animals and research data.

The manufacturer assessed the returned valves at their factory and showed the presence of macroscopic materials on silicone sealing surfaces that explained the failures of 42.3% of the valves. Various combinations of contaminants were seen, such as silicone flash (residual silicone material from the mold process), stainless steel fragments from the thread cutting process, bedding, and fur. These materials would rest on the surface of the O-ring and break the seal between the O-ring and the stem head allowing water to leak past, but these findings could not explain the cases of no water flow. Moreover, the majority of the remaining set of valves functioned normally in factory tests with no indication of irregular water flow.

Increased reports of valve failure from technicians and researchers in the fall of 2007 prompted a large-scale, in-depth assessment of all valves spanning a 4-mo period. We initiated a drip test assessment to determine whether valves were operating appropriately. In close collaboration with the manufacturer, we worked to find the root cause of the variable flow issues.

Inhouse valve assessment.

We began a systematic valve screening process in December 2007. The process included manual testing of every drinking valve before placing in service (Figure 2). We screened approximately 17,000 valves over a 4-mo period by using a drip test with clean valves attached to a test ventilated rack equipped with the online rack flush system. We plugged in several rows of drinking valves, toggled valves, and captured and measured the flow-through in a graduated conical tube. Valve flow rate was assessed for leaky, low-flow, and no-flow conditions. Leaky valves began dripping water as soon as they were placed on the rack or continued dripping after valve toggling. Low-flow valves provided less than 30 water drops within 10 s. No-flow valves did not provide any water when toggled. Valves that were outside of the 25- to 50-mL/min range were returned to the vendor for assessment. Valves that were within the range for acceptable function were placed into general use.

Figure 2.

10-second drip test: procedural schematic for assessment of drinking valve function by using sanitized valves on a test rack.

Results

The cause of drinking valve leaky-, low-, and variable-flow failures.

In 40% of valve failure cases, internal valve component examination confirmed that the presence of foreign debris (metal shavings, silicone flash, bedding, fur) resulted in the leaky valves. Critical insight into the cause of almost 60% of the remaining balance of undeterminable valve failures was gained when recently failed valves were submitted for factory assessment within 1 wk of removal from use. The unique feature of this set of valves was that the internal chamber was still filled with water and the silicone components were still wet. When the valves were placed on the factory test machine, recreating rack-like conditions for water flow and pressure, they failed and were disassembled immediately for assessment. The internal diaphragm (Figure 1 B) was found to have lost the originally designed shape. At the factory test, 79% of these ‘wet valves’ showed swelling in 2 of the 3 internal silicone components, the diaphragm and O-ring. The diaphragm had developed a distorted, water-filled blister in the center. This bubble and distortion in the center of the diaphragm caused the head of the valve stem to be displaced, losing the seal between the head of the valve stem and the diaphragm. The displacement resulted in a leaky or continuous-flow condition. In addition, the O-ring that completed the internal seal had become swollen in several valves (Figure 1 C and D). The swollen O-ring caused 1 of 2 conditions: it either closed the seal with the diaphragm, preventing water from passing along the stem to the internal chamber and resulting in a no-flow condition, or it became stuck away from the diaphragm, creating a gap at the stem head and allowing constant water flow (Figure 1 D). The internal silicone components, the diaphragm and O-ring, had become compromised, absorbed water, and lost elasticity. The result was an obvious blister in the center of the diaphragm and a swollen O-ring that increased in size. Silicone elasticity of both pieces was assessed at the factory by using an impact test, which evaluated measurements of the distorted material compared with same piece after drying out, as it returned to its normal configuration.

In addition to this finding, the sanitizing–sterilization processes at other facilities were assessed. One hundred thirty-five autoclave cycles from 35 facilities were reviewed, averaged and compared with our program. A key difference in the processing programs between facilities was the autoclave drying cycle. The drying cycle at our facility pulled a deep vacuum relative to that at other facilities that were not seeing these distorted silicone features even on older valves (Figure 3).

Figure 3.

Vacuum pressure in autoclave cycles. Green line indicates autoclave cycle activity with no compromised valves. Red line indicates dry-cycle vacuum deeper than –15 in. Hg, causing damage to silicone components of the valves.

Discussion

Numerous automated watering valve failures over a multiyear period required the implementation of a quality-control program to assess valve function. A quality-control program should cover several critical check points, including systematic assessment of quality, equipment, and independent vendor assessment. In developing our systematic assessment of valve quality, a 3-step flow assessment program was conducted both prior to use with animals and after refurbishment of dysfunctional valves. Valves were tested for flow initially on a test rack, were toggled for flow when they were placed on active animal cage racks, and finally were monitored daily for flooding and animals checked for clinical signs of dehydration. If at any of these 3 assessment points the valves failed to operate appropriately, they were removed from circulation for refurbishment. Failed valves were restored to the 25- to 50-mL/min flow rate through an inhouse, valve refurbishment program. A technician assembly line disassembled failed valves, replaced the 2 internal silicone components that were susceptible to distortion (diaphragm and O-ring), and replaced the shield. The shield, visible from the front of the valve, was coded by the manufacturer with a specific color for each year from 2007 through 2012. This color became our indicator of year of valve refurbishment. Refurbished valves then were washed, reentered into to the testing pathway for reassessment, and autoclaved before return to use. Because all original valves had green shields, systematic assessment of valves will continue for several years until all valves are color-coded for refurbished year. Inhouse refurbishing with available technicians allowed us to minimize replacement costs.

Assessment of facility equipment included verification of autoclave cycle and phase duration, calibration, and certification. The facility autoclave pulled a vacuum of –24 in. Hg during the dry cycle. This vacuum exerted extreme pressure on the silicone components of the drinking valves. In our valve sterilizing process, pulling a deep vacuum allowed cycles to complete faster with dry equipment. The vacuum cycle that reached –24 in. Hg allowed us to complete cycles in less than 1 h compared with a standard 75-min full cycle time. However, this high negative pressure, along with heat and steam, degraded the fillers in the silicone, reducing their elastic properties and causing them to absorb water. Water absorption in the diaphragm, evident as a blister, and swollen O-ring resulted in loss of elasticity and caused both continuous- and no-flow conditions. According to the manufacturer of the valves (Edstrom Industries), studies have shown that at normal operating temperatures ranging from 120 to 275 °C, the principal mode of silicone deterioration is due to cleavage of the backbone by exposure to water.¹ Silica, under the right conditions, is hydrophilic. So cleavage of the silicone backbone occurs in the silica bonds with the water in the presence of steam. Unfortunately, once this physicochemical change occurs, no process is available to return the silicone to its original functional characteristics permanently, and the item must be replaced or refurbished.

The silicone rubber that the manufacturer selected for drinking valves was rated as appropriate for the application because it had a high molecular weight, provided a flexible, but strong compound that withstood extreme temperature changes. Based on these physicochemical properties, this silicone rubber was chosen for this valve application with good success in many research facilities. To maintain an appropriate autoclave cycle for valve processing and ensure that the vacuum was not a factor for the silicone components, we began processing the valves through a liquid sterilization cycle (no drying phase) to avoid pulling a vacuum. This practice was reviewed by the vendor and assessed as acceptable.

Using these compiled data from 35 facility operations, the manufacturer calculated the average time that valves had been in service leading to failures was from 3 to 4 y. Because of the limitation of silicone internal components, the manufacturer recommends valve refurbishment every 3 to 5 y, depending on the conditions to which the valves were subjected. The manufacturer also recommends installation of an electronic controller to prevent vacuum cycles from pulling more than –15 in. Hg in older autoclaves lacking programmable controls. Newer autoclaves should be adjusted similarly to reduce vacuum cycle pressures. A good quality-control program should both assess how accurately drinking valves are delivering water and monitor autoclaves used in the processing of valves for appropriate operation.

Corrective measures in the manufacturing process were initiated to eliminate the deposit of residual silicone flash tags and remove stainless steel burrs. A new diaphragm mold was constructed to decrease this material contamination from silicone flash. For these silicone components, the manufacturer initiated heat curing and cryogenic deflashing: after fabrication, silicone components were placed on a tray and cured at 392 °F (200 °C) for 4 h. The silicone O-rings then were placed in a tumbler. As they tumbled, cryogenic media was added at a temperature between −100° and −150 °C. As the parts tumbled, additional nitrogen gas was added to cool the parts. The stainless steel fragments that lingered on the threaded section of the valve after fabrication were addressed by implementing a thermal deburring process. The manufacturer exposed the stainless steel valve components to extremely high temperature and pressure to eliminate burrs.

To aid with tracking of valve age, 2 important production changes were made. A unique color was assigned for the externally visible silicone component for each production year, and beginning in 2008, all factory-produced or factory-refurbished valves were date coded with both a part number and the date of manufacture. This factory refurbishing program provided new silicone internal components, a thermal deburring process to eliminate stainless steel fragments, a new valve stem, and treatment of the stainless steel component parts for further rust prevention. For the valve-tracking effort, we were able to determine the year of refurbishment with a glance at the front of the valve. We targeted a refurbishment number of 4000 valves per color year. This target helped us to address replacement of the 17,000 valves in our facility in just over 4 y, spreading the lifespan of valves and keeping our budget stable.

Using these processes, flood rates in our facility decreased from a monthly clinical failure rate of 12.9% in 2007 to 0.2% in 2009, which represents a total of 71 incidents of valve floods and dehydrations on cage racks across a 3-mo population of 32,162 cages. As a comparison between valves and water bottles, we reviewed the flood rate for cages with bottles. We used 16-oz, low-profile water bottles (Allentown, Allentown, NJ). These bottles featured a nylon-lined screw cap with a ball-bearing sipper tube. For the same 3-mo period there were 65 water-bottle failures among 2984 cages, resulting in a 2.2% clinical flood rate (Figure 4). In March 2011 our valve refurbishment program was 70.5% complete and reached a failure rate of 0.28%, which approached the 0.1% manufacturer standard.

Figure 4.

Water valve failure rate from 2007 through 2009 compared with bottle failure.

This case report highlights the importance of regular quality assessment of animal care equipment. Over time the silicone components of water valves lost their inert, water-resistant qualities and became susceptible to absorption. Once placed into service, those valves failed to provide controlled water flow to mice. Identification of this etiology was challenging due to the variable nature of the water flow. To prevent widespread watering failures from developing, a facility-wide quality assurance program should be initiated. We suggest a program that monitors 4 levels: assessment of valve water-flow performance; calibration and recertification of autoclaves used in the sanitizing of valves; good communication with animal husbandry staff to highlight and monitor issues with animal equipment, and development of a strong partnership with vendors as colleagues in the research support process.