Process Safety: Summertime Slip

“The issue is that heat can cause metals (and other materials) to expand – and in the cold, to contract – which in turn can impact whether a part is in or out of spec.”  — Barbara Osborne

You probably understand the idea of thermal expansion: when materials warm up, their atoms vibrate faster, so the material expands. Different materials have different volumetric coefficients of thermal expansion, α. The bigger the α, the more the material expands when heated.

[Note: The linear coefficient of thermal expansion, α_L, is usually one-third of the volumetric coefficient of thermal expansion, α_V.]

Liquids have a higher α_V than solids. Gasoline, for example, has an α_V of 950 x 10^-6/°C. Water is less but in the same order of magnitude, with an α_V of 207 x 10^-6/°C.

On the other hand, inorganic non-metals typically have the lowest α_V, two orders of magnitude lower than liquids. Fused quartz has an α_V of 1.8 x 10^-6/°C and borosilicate glass has an α_V of 9.9 x 10^-6/°C.

Falling in between, common metals each have an α_V about an order of magnitude smaller than liquids, but about an order of magnitude larger than many inorganic non-metals. The α_V of common metals typically falls in the range of 25 x 10^-6/°C to 70 x 10^-6/°C. Carbon steel, for instance, has an α_V of 32.4 x 10^-6/°C.

Why does this matter to process safety?

Expansion Joints

An obvious problem is process equipment that goes from ambient temperature when idle to a very high temperature when operating. A long steam pipeline will grow in length when it heats from idle up to operating temperature. Consider a carbon steel 50-psig steam line that is 500 feet long. The temperature of saturated 50-psig steam is 298°F. If ambient is 70°F, that line is going to expand over eight inches:

(298°F – 70°F) x 1°C/1.8°F x 0.0000324/°C / 3 x 500 ft = 0.684 feet

With nowhere to go, eight inches of additional pipeline will do some damage. The same concern applies to the structures that support process equipment. Fortunately, structural engineers and piping designers are acutely aware of these issues and include expansion joints in their designs, so that thermal expansion does not become a process safety issue.

A Very Bad Day

Thermal expansion, however, can be a process safety issue after the design is done and the installation is complete. Especially if the installation was in cold weather. Flange connections in pipelines are subject to thermal expansion and contraction. In the wintertime, bolts and nuts contract. The bolts become shorter and the nuts become tighter. (Nuts are tighter not only because shorter bolts pull them tighter against the flange, but because they shrink around the bolt.) If the bolts and nuts were installed correctly in the summertime – tightened in the correct order to the correct torque – the torque in the wintertime may subsequently exceed the specification. This probably won’t result in uneven stress across the flange face, but it may damage the gasket, so that when the bolts later expand, the gasket leaks.

On the other hand, if the bolts and nuts were installed correctly in the wintertime – again, tightened in the correct order to the correct torque – the bolts and nuts will expand when the warmer weather of summer comes. Fortunately, the gasket will not be damaged. But it will loosen. And a loose gasket may leak.

The especially frustrating aspect of this phenomenon is that it won’t just be one flange that leaks. It is entirely possible for every flange installed at one time to experience the same kind of leak at the same time.

Which makes for a very bad day for the maintenance department.

Mechanical Integrity

The mechanical integrity element of the Process Safety Management (PSM) standard, 29 CFR 1910.119(j), explicitly identifies “piping systems (including piping components such as valves) as one of six specific types of process equipment that must be inspected and tested at a frequency consistent with good engineering practices. While it is perfectly appropriate to test the thickness of a piping system every 5 or 10 years, depending on a number of factors, the tightness of flange bolts should be tested more often, given the seasonal changes in temperature.

It’s not just the frequency that matters, though. The torque on flange bolts should be checked, at least qualitatively, at the hottest part of their thermal cycle, as well as at the coldest part of their thermal cycle, especially when prior operating experience indicates that loosened flanges are a problem.

Check Your Flange Bolt Tightness

In case you are wondering, this warning about the impact of thermal cycles on the torque of flange bolts is not just idle speculation. Like many safety warnings, this one is written in blood. So, check. It is much easier to check than it is to recover from the exposure to simultaneous, spontaneous leaks.

Fortunately, the process of checking can be easier than taking a torque wrench to every bolt on every flange in the plant every summer and every winter. It is sufficient to test a statistically representative sample using the √n + 1 test plan. If you have 5,000 flanges in the facility, test 71 (the square root of 5,000) plus one, chosen at random. If they are all sufficiently tight, you’re good. If they’re not, aren’t you glad you found out before you had a release?