Overpressure Protection In Pilot Plants and Laboratory Units
Richard Palluzi
Pilot Plant and Laboratory Engineering, Safety, and Design Consultant at Richard P Palluzi LLC
Providing overpressure protection for all pilot plant and laboratory equipment is a critical safety issue often overlooked in research for a variety of reasons. The equipment is considered too small. (Ignoring the fact that it has enough stored energy to seriously injure anyone close to the failure.) The operation is considered fully attended. (Which it never is as operators invariably take breaks, rest room visits, or get distracted by calls, emails, or other personnel.) The operator is assumed to never make a mistake. ?(Ignoring the truly wide-ranging potential for human error. See Scared Safe: The Importance of Human Error when Evaluating Research Operations for Safety, https://www.dhirubhai.net/pulse/scared-safe-importance-human-error-when-evaluating-research-palluzi for a more detailed discussion.) The potential for something being over pressured is not recognized. (Assuming a regulator or similar component can never fail or be set wrong.) These and numerous other ill-advised reasons lead to many situations where component failure is a very credible risk and over pressure protection is required.
Over pressuring a component can lead to numerous issues. Loss of containment including spills, sprays and leakage resulting in personnel injury or exposure if the fluid is toxic or hazardous, hot or cold. Fires or explosions if the material is flammable or combustible and released to the environment. (Recognizing that mists and sprays of flammable and combustible liquids are often easier to ignite than standing amounts.) Equipment rupture resulting in equipment damage, fragmentation or catastrophic failure is certainly possible. These failures can result in fires or explosions if the material is flammable or combustible. Fragmentation or catastrophic failure can result in serious injury or death. Fragments, fires or explosions can also damage or penetrate other equipment of piping resulting in a cascading series of failures. Environmental releases can result in discharges to the environment and/or exceeding permitted or legal limits with attendant fines, legal actions, or costly remedial actions can be the result. ?Equipment damage can result in the need to repair or replace costly components with attendant downtime. So ,the potential hazards are serious enough that care should be taken in evaluating the need and assessing the risk.
Providing over pressure protection is typically done in one of three ways:
Providing over pressure protection by design is an inherently safer approach. It avoids the need for other types of relief devices, eliminates the potential for premature releases ,and saves relief device maintenance costs. However, it is important to fully analyze all the potential failure modes which might occur. It is often easy to think that a reactor cannot be over pressured because it is designed for more than the liquid feed pump or gas compressor can reach. However, if the reactor is overheated, resulting in a lower pressure rating, if the reactor is corroded resulting in a weakened spot, if an unknown reaction results in a higher internal pressure, if an external fire results in a lower pressure rating, if a pump is replaced with a higher pressure model, ?or numerous similar situations arise then the reactor may fail. Hence, diligent hazard analysis and risk assessment often shows that providing an overdesigned system becomes more expensive and less certain than originally conceived. My experience is that most of the systems that rely on design for over pressure protection are often flawed and fail to provide the safety required for all potential situations. Low pressure laboratory systems, those designed for 15 psig or less are particularly prone to problems. Often the full water, air or nitrogen pressure is not considered, any of which is capable of over pressuring the equipment unless a relief device is provided.
Often the capacity of the system is badly underestimated. A glass reactor rated for 5 psig has little ability to relieve excess pressure unless a larger line is provided. Often these lines are determined “by inspection” and upon analysis are found to be much too small (2-4 times is not uncommon). Similarly, events that are used to develop the required capacity are often poorly identified. Assuming the maximum regulator design flow is the bias fails to address sonic flow in the regulator when a diaphragm fails. The maximum flow from a site utility system often fails to consider periods of higher supply pressures or the consequences of a valve been opened too far or a regulator set too high.
Providing electronic overpressure protection is often desirable in pilot plants for a variety of reasons. It avoids the potential of leakage inherent in a mechanical relief device. ?The pressure of the system is often monitored and so many of the most expensive components are already present. If the system is prone to clogging, this approach can often be more reliable as it may be able to be configured to avoid the potential for clogging. There is also a perception that active intervention at the appropriate time can result in less problems (e.g. loss of data, contamination, etc.). Highly problematic in my experience but desirable.
Typical means of providing electronic over pressure protection is through use of a pressure switch or electronic transducer with a high pressure shut down. A pressure switch is an inexpensive and straightforward device. When the desired set pressure is reached a contract is actuated and appropriate measures taken. The device is generally highly reliable. An electronic transducer is more expensive but gives more data than a simple switch. It allows monitoring and recording of the system pressure which is often desirable for research purposes. A high alarm or alarms may be set to take appropriate action. A common approach is to set a high alarm to warn the operator that the system pressure is above normal limits and allow them to take appropriate action. If the pressure rise continues to rise, a second alarm, called a high-high alarm, typically takes automatic action. Typical actions may include venting the system, shutting off the feeds, cutting off the heat (and/or cooling the system) or similar actions.
However, use of an electronic over pressure protection requires a careful assessment of the reliability of the automatic safety shutdown system. Any potential failure in the automatic shutdown system such as an improperly entered set point, a stuck contact, a transmitter or switch failure, an improper calibration, or other potential failures could result in over pressure. For this reason, many organizations are reluctant to allow this approach or use it only after a detailed hazard analysis and risk assessment of all potential failure modes of the electronics, associated systems, and the process. Most tend to consider it a higher risk solution due to the increased risk of failure.
Providing over pressure protection by mechanical devices, relief valves or relief devices, is the most common form of over pressure protection. Mechanical devices have a long history of reliable and successful performance. Maintenance, while required, is normally not onerous and ranges from annually to every 10 years. These devices are typically set and forget unless activated, in which case an inspection to confirm proper operation and setting is normally, but not always, required.
?
The two most common types of mechanical relief devices are spring loaded relief devices (relief valves) and rupture disks. Two less common types are pressure/vacuum conservation vents and hydrostatic head devices (bubblers or liquid seals).
The flow rate (capacity) of a normal pilot plant or laboratory scale ?relief valve is dependent on the pressure drop across the valve as with any orifice or restriction. This is different from larger, plant scale relief valves that essentially fully open to allow their full flow capacity immediately upon opening. A standard pilot plant or laboratory relief valve may be able to be set at 50 psig, for example, and to relieve 10 SCFM but not simultaneously. At a setting of 50 psig, it may only be able to relieve 1 SCFM; it may need a 100 psig inlet pressure (i.e., a 50 psid pressure drop) to achieve 10 SCFM. Always confirm the release rate by looking on the actual capacity curve for the specific valve to be sure of the proper specification.
Spring loaded relief valves are available in a wide range of pressures, materials and end connections.? Tube fitting connections are available from several manufacturers which are often highly desirable in pilot plants which utilize tubing and compression fittings extensively. They will stop the release when the process pressure becomes lower than the relief device’s setpoint allowing the process to continue operation. Smaller sizes, common in pilot plants, are relatively inexpensive.
There are some concerns when using spring loaded relief valves.
领英推荐
Sizes below ?” NPT or 1/4” OD tubing are more difficult to obtain. This can cause problems in smaller systems. Lower pressure settings, typically below 25 psig, are problematic on smaller valves, typically those 3/4” NPT and less, as the relatively small springs tend to have a wider variability and more difficulty in opening reliably. Hence a wider range of potential opening pressures are possible. This makes finding spring loaded relief valves for low pressure service (such as glass or plastic) difficult unless relatively larger sizes are used. These often become difficult to install and are often operationally undesirable.
Leakage at pressures below the relief valve’s set point is always possible. Typically spring loaded relief valve manufacturers talk in terms of the maximum process pressure to relief valve set point ratio that the device should be used. A spring loaded relief valve with a maximum recommended ratio 0f 85% should not be used for a process that will normally reach higher than 85% of its setting (e.g. 85 psig for a 100 psig setting). However, all spring loaded relief valves have the potential to leak prematurely and will, occasionally, leak at very low levels (25-50%). ?Typical spring loaded relief valves should not be used to more than 80-85% of their setting unless recommended by the manufacturer and after careful consideration of the effects of a premature opening. Lower ratios, i.e. operating even further away from the set point, are always more desirable. This creates problems for the inexperienced designer or researcher as if components are purchased rated to say 1,000 psig, this constraint limits the maximum normal operating pressure to 800 to 850 psig. For a true 1,000 psig operation typically higher pressure components rated above 1,000 psig/85% or 1,176 psig are required to give a “cushion” for normal operation without leakage.? These ratios can be lowered, dramatically, by pressure surges or changes in temperatures.
Spring loaded relief valves also almost always have a change in flow direction and a restriction at the seat. This makes them more likely for clogging or plugging when using dirty, fouling, or agglomerating fluids. They also become bulky and expensive in larger (1” NPT and above) sizes.
Small size relief valves can also take a semi-permanent set over time as the sealing gasket tends to adhere to the seat and plunger and requires significant extra force to open after an extended period. The problem is most noticeable on spring loaded relief valves smaller than ?” NPT and set below 100 psig. Typical unexpected jumps in activation can range from +50% for a 25 psig setting to+ 20% for a 100 psig setting. The adhesion seems to start in 6 months below 100 oF and in 3 months at temperatures above 100 oF. Endemic to all manufacturers, it has no proven solution except activating the relief valve more frequently in place, inspecting the relief valve more frequently, or accepting a higher possible activation value. The last is the most common approach; many organizations require a higher pressure rating for all components for these smaller valves. Larger valves and higher set pressures appear to be essentially exempt from the problem. Coating the relief valve with a non-stick material seems to slow the onset but not significantly (9 months perhaps). See Performance Analysis of Small Size Pilot Plant and Laboratory Relief Valves, R P Palluzi, Process Safety Progress, Sept, 2003 for more details on spring loaded relief valve performance and inspection intervals.
Rupture disks are solid plates designed to fail (rupture) at a specific pressure. The failure may be an opening or pealing up of the disk or a complete fragmentation if frangible materials such as graphite are used. They are available in metals, plastics, and plastic coated metals in a wide variety of materials and pressures.
?Rupture disks are less likely to leak as their only seal is a static gasket. As a result, rupture disk process to relief ratios are higher than spring loaded relief valves, typically 85-95%. Again, these can be adversely lowered by pressure surges, pressure cycling, corrosion, or temperature cycling.
Rupture disks produce large flow rates for a given size as they provide essentially open path conditions. Thus, they are often favored for larger relief rates. They also have no bends or restrictions inherent in the design so their use on dirty, fouling or agglomerating service is common. However, they may still clog or plug, particularly if the fluid cools and hardens or sets up below the rupture disk over time.
They are available in lower pressures than spring loaded relief valves and can be rated for vacuum protection. However, these lower set pressures are not always available in smaller sizes. A 5 psig rupture disk is readily available from numerous manufacturers but not usually in sizes smaller than 2” NPT for example.
Rupture disks, once activated cannot, reseal, and must be replaced resulting in a process shutdown. The releases are sudden, and occasionally violent, at higher pressures. Vent systems must be designed to withstand the resulting sudden pressure surge. Recognizing a rupture disk has failed after the event is not always easy unless the system pressure is easily verified. Manufacturers do provide systems for detecting and alarming upon rupture albeit at a higher cost.
Rupture disk settings are very temperature dependent. Hence, they come rated for a specific operating temperature. On small pilot plants it is often challenging to determine exactly how close to the process the disk will end up being installed. Hence it can become difficult to determine the rupture disk temperature in advance. This is, however, critical as too high a temperature and the rupture disk will fail prematurely (effectively have a lower set point) and too low a temperature the rupture disk may not relieve when intended (effectively too high a set point). Actual operating temperatures should always be checked and confirmed within design limits during the initial startup. Prepare to often be surprised.
Rupture disks may be flat or have a bulge up (forward acting) or down (reverse acting). Either way can be designed effectively and each has some different advantages. This obviously raises the potential for a disk to be installed backwards. All manufacturers clearly label their disks for the proper direction; many also provide assemblies which only allow the disk to be installed in the right direction. While safer, these assemblies can usually be installed upside down! Other manufacturers design their disks so that if installed improperly they will either fail below their listed setting (i.e. prematurely but safely) or within a predetermined amount over the setting such as 10% which can often be accommodated in the design.
Rupture disks also have manufacturing tolerances.? The manufacturing range is a pressure range within which the rupture disk must relieve (burst) to be acceptable to the purchaser for a particular process. A 100 psig rupture disk with a 5% manufacturing range can be supplied with a stamped set (relief) pressure of anywhere between 95 and 105 psig. So called zero manufacturing range rupture disks are available at a higher cost. In this case the burst pressure and the tag pressure will be the same. (The manufacturer does this by having to do additional testing on each lot to achieve the – in effect – tighter tolerance.)? The burst tolerance is the range in which a rupture disk should be expected to burst. The ASME Boiler and Pressure Vessel Code, for example, specifies a burst tolerance of ± 5% for burst pressures above 40 psig? and ±2 psig below 40 psig.
Hydrostatic head devices, often called bubblers, rely on a height of sealing fluid to set the back pressure of a gaseous system. Any excess pressure pushes up or bubbles out through the system. Small laboratory scale units are available from many glass supply houses. Most pilot plant units are homemade using pipe or tubing.
Hydrostatic head devices are leak free (unless over pressured) and relatively easy to construct. They are very low pressure devices as even heavy liquids require excessive heights for higher pressures. Care should be taken to ensure their construction is adequate for any pressure surges and that their capacity is properly evaluated. It is also important, although easily overlooked, to have some way to confirm the liquid level as evaporation, leaks and liquid carryover can result in lower liquid levels than required.
Next month’s article will discuss additional issues with providing over pressure protection. You may want to consider Engineering Career Solutions course Safe Operation of Pilot Plant and Laboratory Equipment and Systems (https://ecstechtraining.com/safe-operation) for additional training and effective problem solving.