Why Physical Testing is Vital in Certifying Valves for Critical Applications
By Cal Cardiff, Dave Bala and Ludwig Haber, Ph.D., Alden Research Labratory
It is easy to take valves for granted. They have been used since the days of the Roman Empire and designs are pretty routine, so people tend to assume that a standard valve will do the job. But that complacency can't apply to critical applications where safety and reliability are paramount.
Occasionally, valve failures make the headlines, such as with the Deepwater Horizon leak last April when a bent pipe kept the valves in the blowout preventer from closing. Six months later, the Tennessee Valley authority found that a flow control valve failed to open while attempting to establish shutdown cooling during a refueling outage at the Browns Ferry Nuclear Plant. An investigation determined that the valve failure was due to a manufacturing defect resulting in undersized disc skirt threads. The valve stem to disc separation had occurred prior to November 2008, so the unit had operated for at least two years without discovery of the faulty valve.
When designing, manufacturing or purchasing a valve, it is important to ensure that a valve meets real world requirements, not just computer calculations, something that can only be achieved through physical tests to validate its performance.
For routine, low-cost, low-value applications, valve testing may not be needed. Manufacturers typically offer to replace any valve that fails, and the valves can be oversized to provide some margin of safety. Those manufacturer warranties don't, however, cover the cost of lost production nor lost customers as a result of unscheduled downtime.
Offshoring offers valve manufacturers a way to lower costs, but it also makes it harder to verify that their valves made by contract manufacturers in other parts of the world are up to the company standards. And, even if the manufacturing is done right, it can be difficult to get documentation to prove that the metallurgy is just right or the hydro-testing was done in a way that matches real world conditions. Having an independent facility test the valves provides both the manufacturer and its customers' peace of mind in knowing that the valves will perform as expected.
Valve customers or systems integrators may also want to conduct their own testing before installing a valve. For one thing, there is the matter of proper valve sizing. It is easy to just specify a valve with enough margin of error to ensure to performance the job. That margin comes at a cost, however, both in terms of a larger, more expensive valve and a larger, more expensive actuator. And even then the extra margin doesn't guarantee reliability.
With safety applications, in particular, it is important to be able to document that the valve will work. Nuclear plant operators are running into this situation now as they seek to extend the life of their plants beyond the original forty year license. The rough calculations that were done decades ago, if they still exist, are not enough satisfy a regulator. Even if the original data could be located, we now have much better tools to evaluate the valves and performance than engineers had 25 years ago, and also a much better understanding of materials science and the long-term effect of heat and radiation on equipment. As part of their relicensing applications, nuclear operators need up-to-date validation that the valves will continue to perform flawlessly through an additional twenty years.
Testing is also necessary to ensure that design specs meet real world operating conditions. Three years ago, for example, Valve Automation and Control (VAC), a business unit of W&O, which is a wholly owned subsidiary of PON Holdings B.V. of the Netherlands, took on the job of changing out 60-year-old firefighting systems on some of the U.S. Navy's T-AKE dry cargo/ammunition transport ships. The ships are designed to carry fuel, ordnance and other supplies in support of naval fleets, so a robust fire suppression system is essential. The work was being performed by VAC under a contract General Dynamics Nasco had with the U.S. Navy. The plans called for using 8-inch four-port ball valves from Montreal Bronze with actuators from Emerson's EIM unit in Houston. The design criteria for the valves showed they were adequate if there was a fire in a single hold, but what if there was a catastrophic event with fires in multiple holds and all the fire pumps kicked in at the same time, a scenario that could easily occur on a ship filled with ordinance and fuel coming under enemy attack? There was no data to verify that the valve was adequate to meet that situation, so testing was needed.
When testing valves, there are several points to keep in mind. To begin with, computer modeling alone is not enough. Running a Computational Fluid Dynamics (CFD) analysis should be done as a first step and can help guide the parameters for physical testing, but, by itself it cannot accurately determine whether a valve will meet expectations. For example, when VAC needed to validate the valves before installing them on the T-AKE's it first went to a nuclear engineering company in Houston that had software to convert the data on flows and pressure into the torque values required to open and close the valves. When the valves and actuators were run through tests at Alden Laboratory, however, the actual torque values were about 40 percent lower than the theoretical torque values predicted by the engineering software, which can make a huge difference in costs.
Another point is that the valves and the actuators must be closely matched. This applies whether it is an automatic or a manual actuator. An oversized actuator can power its way through a sticky valve, but it may mask a valve problem and eventually shear the valve stem. An undersized actuator won't be able to operate the valve if the performance declines over time, which is likely to occur. The resistance of the valve when first installed won't be the same as it is after 10,000 cycles, so the actuator must be adequate to accommodate changing resistance without covering up problems that need urgent attention.
The testing should also match the real world operating conditions, including flow rates, pressure, viscosity, temperature and contaminants or abrasives entrained by the fluid. At a minimum, particularly when dealing with isolation valves, the pressure differential on both sides of the valve needs to be known for a proper test design.
Finally, the tests should be run through enough cycles to verify that the valve will continue to operate successfully throughout its expected lifespan.
Following these guidelines and executing a properly designed physical test of a valve can build confidence that the valves will meet the expectations of customers in actual application. It is far better to show them actual performance results than some theoretical calculations.
Testing the T-AKE Valve
In April 2011, Alden Laboratory conducted physical testing onone of the Montreal Bronze valves slated for installation in the fire suppression system on a U.S. Navy T-AKE dry cargo/ammunition transportation vessel. The tests were done to both verify the amount of torque needed under operating conditions and to verify that the valve would operate properly through an extended series of operations.
Conservative analysis conducted prior to testing showed that hydrodynamic torque could be a significant component of the overall required valve torque. Testing was employed to put the valve in a realistic operating scenario and measure the required torque.
While the exact relationship between pressure and flow experienced by the valve on ship is difficult to calculate, two operating points can easily be identified. The first is the pressure of the system against which the valve is seated and then has to open. The second is the terminal flow rate through the valve under fully open conditions.
Two operating conditions are important to the T-AKE valve installation on the sprinkler system. The first is the test condition where sprinkler associated hardware is periodically tested for performance and timely response. The flow rate for these conditions is expected to be less than 1,000 gpm. The second condition is the operating condition where flow is supplied to the sprinkler headers during an emergency event requiring sprinkler actuation. The flow rate for this condition was identified to be 4,570 gpm. While many test cycles are expected for the valve, very few operating condition cycles are expected. The testing, therefore, consisted of 500 operating condition cycles and 1,500 test condition cycles.
Testing was conducted over a one week period in mid April 2011 using an open test loop design (Fig. 1). The primary loop pump consisted of a 200 HP centrifugal pump. A second boost pump was installed in parallel with the primary pump. A swing check valve installed downstream of the primary pump allowed the second pump to work in series with the first when the test valve was closed. Loop operation was fully automated with the valve controller dictating the pace of cycling. Downstream of the pump, different restrictor plates were installed to provide varying levels of system resistance depending on whether the cycles were conducted for valve operating or valve test conditions.
Three measurements formed the core important data related to the test: flow rate, valve inlet pressure and valve actuator torque. The latter was measured by monitoring the current supplied to the actuator. The actuator had a known relationship between current and torque. A limited number of cycles were also conducted using a manual override hand wheel. Torque measurements for hand-operated cycles were also obtained using a slip-ring / strain-gauge based torque sensor. Flow was measured using a venturi flow meter and valve inlet pressure was monitored using a piezo-resistive electronic sensor. Data collection was continuous throughout all cycles, recording one second data averages throughout the test.
Valve performance was maintained throughout all cycles with no detectable increase in required actuator torque during the testing. The current profiles (proportional to torque) obtained during opening and closing for high flow show a distinct hydrodynamic contribution (Fig. 2). The unseating and reseating torque is generally expected to constitute the torque peak in a valve cycle. However, the torque added to the ball and stem due to the redirection of flow through the ball at partially open positions can cause even greater resistance. Note that hydrodynamic torque is only a contributor during the opening cycle, whereas during closing, hydrodynamic torque works with the actuator. The peak in torque on the closing cycle is observed at the reseating of the valve.
Testing was able to show that while hydrodynamic torque can play a significant role in the peak actuator torque requirements, the valve and actuator package were properly designed to power through the added load and no adverse performance was observed even after extended valve cycles.
About the authors: Cal Cardiff founded Valve Automation and Controls and has continued to work for the company after selling it to W&O. Ludwig Haber is a principal engineer at Alden specializing in fluid dynamics (experimental and computational), and turbomachinery performance.
Power Engineerng Issue Archives
View Power Generation Articles on PennEnergy.com