Introduction to Pressure Relief Valve Design Part 2 – Relief Scenarios and the Relief Rate

This is the second in a set of articles introducing the basics of pressure relief valve design from a process designer’s viewpoint. Read Part 1 here or Part 3 here.

Once you have decided to add a relief valve, know what equipment and pipes it is protecting, and choose the set pressure, you can finally begin to size the valve. The first step, and often the most difficult, is to determine all the different relief scenarios, also called contingencies. A scenario is an event that causes overpressure. You as a designer need to check out all the possible scenarios.

Fire is an extremely common cause of overpressure you must design against. Photo taken by Kirrus at Flickr licensed CC-by-SA.

Here is a list of a few scenarios to consider:

• External fire heats up the contents of pipes or vessels. You need to determine how much of the vessel is engulfed in the flames and what liquid level (if any) is present inside the vessels/pipes, as liquid can absorb the energy of the fire by heating and then vaporizing. If you have liquid boiling off you may have a dynamic calculation on your hands, because the for multi-component fluids the composition and fluid properties can change as part of the fluid boils off. Insulation plays a part in fire calculations. API 521 has some equations you can use to model a fire and also and gives you fire dimensions you can usually assume. (In general one considers up to a 25 ft high fire, in a 2500 square feet pool, as a maximum for most relief valve and flare header sizing)
• A valve is closed, causing a pump or compressor to keep pushing into a non-flowing or “dead” zone. The equipment may keep pushing and increasing the pressure until something breaks. This is an especially large risk if dealing with a positive displacement machine. (Unlike positive displacement-type equipment, centrifugal pumps can only push so far before they “dead head” – at max, they will output the differential pressure that their pump curve indicates at a flowrate of zero)
• A heat exchanger stops working (maybe the cooling tower goes down or a valve gets closed), causing pressure to rise
• A reaction goes out of control, aka a “runaway” reaction
• Too much hot fluid is supplied to a heat exchanger
• Too much fuel is supplied to a fired heater
• A control failure or human error opens or closes something at the wrong time. Be creative to consider what problems could reasonably occur
• Instrument air fails
• Power failures, local or global over the entire plant
• The tube in a heat exchanger breaks open suddenly
• Thermal expansion of liquid. (example: imagine a piece of pipe is closed at both ends when 100% full of liquid. Then the sun comes out and heats the pipe, causing the heated liquid to expand and increasing the pressure). It is a mistake to ignore this problem because for liquids the rise in pressure can be very significant. Plus, the nice thing about this scenario is that the relief valves are usually tiny. Sometimes when you cannot figure any plausible scenario you just buy a ¾”x ¾” thermal expansion valve just in case.
• To get other ideas, read any company standards you have. Brainstorm through what can go wrong. If you have a book about designing equipment (like a distillation tower design book) the book may have a section of relief scenarios that can come up. When you do safety studies like HAZOP meetings make sure that you’ve covered every scenario in your relief valve calculations.

See also Common Overpressure Sources, Protect Plants Against Overpressure.

In general, when making up these scenarios you want to avoid double jeopardy. This means that you do not assume that two totally unrelated mistakes occur at the same time. For example, if I can lose instrument air, assuming that two air-powered  control valves fail at the same time is plausible because they both have the same cause of failure. But you would not assume that a control valve fails at the exact same time as a heat exchanger tube ruptures: this is so implausible that normally you do not have to design for it. (However, use your judgment: if you think it’s possible, and also very deadly, you might opt to design for a double jeopardy case at your discretion). Read this for a longer discussion: http://www.cheresources.com/asiseeit2.shtml

For each scenario you have to determine the conditions at which the fluid will have to be relieved, the flowrate developed, and also calculate the fluid conditions so you can design the relief valve and piping. This can be the toughest part of all as there is no single approved place to go for guidance, unless you have the benefits of extensive standards written by large private companies. (e.g. If you work for Exxon or Shell you may have company guidelines to fall back on). In the public domain, some guidance can be found in standards: API 521 will help you with fire, and many large companies have standards for other scenarios. You can also try reading engineering magazine articles, which may have specific articles describing a method to tackle certain scenarios.

In some cases, you have to do a calculation or a process simulation. If you can do dynamic simulation, where variables are calculated as they change over time, you can often get more accurate results than just using a steady-state simulation to approximate a relief condition. In some scenarios it is really necessary to do a dynamic calculation when it is not clear what is the most dangerous part of a relief scenario. If you only have a steady-state simulator, it may be possible to alter the steady-state model to create several “snapshots in time” that reasonably approximate all the dynamically changing conditions.

As an example, if a multi-component liquid mixture is in a drum, and a drum is heated by a pool fire beneath it, the situation is not simple. The lighter components will boil off first, followed by the heavier components. The different fluid compositions have different properties and different latent heats of vapourization. So you may decide to get the fluid properties at the start of the fire, when the first wisps of vapour boil off. But also look at when 10% of the liquid has been boiled off, 20%, 30%, …, 100%; you do however many scenarios it takes for you to get a good feel for all the possible fluid compositions that your relief valve will have to face. (See Designing for pressure releases during fires – Part 2 by S. Rahimi Mofrad & S. Norouzi from the magazine Hydrocarbon Processing Dec 2007)

If necessary, you can add appropriate design margins onto your scenarios to reflect your level of certainty and comfort. Sometimes companies will invest in things like dynamic simulators just to let them avoid making conservative, simplifying assumptions that lead to unnecessarily large design factors.

If you’re really not sure what to do, get help. Hire experts if need be. Dealing with scenarios is one of the “art” parts of engineering, but it’s also one of the most safety-critical activities out there.

Edit 2010-04-22: Minor rewrite for clarity.

—

Remember you can read Part 1 here or Part 3 here. Coming in Part 3: Designing Inlet and Outlet Piping, and miscellaneous topics for further study.