Elephant traps for vessel designers

Disclaimer

I am not claiming to be smarter than the average vessel designer. Some of my errors have been equally stupid. I have managed to catch at least some of my mistakes before they became an embarrassment. Maybe I have missed some?

The following are a couple of errors which got through – luckily for me, they are not mine. My finger prints were not on them.

Trap 1 Disappearing nozzle and external loads

All vessel design software calculates the local effect of external loads on the pressure boundary. External loads are imposed by pipework on nozzles. External loads also include pipe supports and any other structural attachments to the vessel shell.

What happens to an external load after it reaches the vessel wall? Does the external load just disappear? Well, if you only relied on some vessel design software, then you might think they did.

Real world external loads do not just disappear when they reach the vessel shell. The loads travel into the vessel until:

a) The loads are balanced out (internalised) by other external loads, either in full or in part. For example, when a pipe that is imposing a nozzle load is supported from the same vessel. e.g. An overhead vapour line on a distillation column.

b) The loads reach the vessel supports.

Do not forget to do a free body diagram of the vessel as a whole. Calculate the combined effect on the vessel supports of

– ALL forces AND moments from nozzles

– ALL forces AND moments from attachments

– Self weight and ALL other imposed loads

Ignoring the effect of nozzle loads on the vessel as a whole might not be a show stopper for big vessels with small nozzles. The error might be lost in the design margins.

However, this might not be the case for small vessels with large nozzles. e.g. Heat exchangers.

You might think I am joking? A few years ago, I was looking for data in a vessel calculation certified by a national body. Some components of the vessel support design loads were less than any of the loads imposed on the four biggest nozzles.

Trap 2 Visualisation of Units

When I was a boy, potatoes on my father’s farm were sold in 1 cwt “hundred weight” 112 lb sacks. The sacks were weighed on a scale with 2 x “4 stone” 56 lb balance weights. At 11 years old, I could lift one of the 56 lb weights above my head. Other experiences that have helped me visualise units are the wheel (lug) nuts on the cars I have owned – around 100 Nm – 140 Nm. Another example would be that 1000 Nm is about the torque I could generate by hanging on the end of a 1 m long scaffold tube extension over a breaker bar. I can therefore visualise a pound lb, a Newton N and a metre m. I can also visualise the difference between 1000 Nm, 10,000 Nm and 100,000 Nm.

What about units of Nmm as used in some vessel design software?

Can you visualise imposed nozzle loads of 1000,000 Nmm, 10,000,000 Nmm or 100,000,000 Nmm?

I cannot visualise a Nmm so I looked for a comparison. A bending moment of 1 Nmm is the equivalent of a bee sitting at the end of a 1 m long branch. In old money, about 3 bees sitting at the end of a branch that is about a foot long. I still cannot visualise 10,000,000 bees sitting at the end of a 1 m branch.

In my opinion, if you cannot visualise a value, then you cannot spot obvious errors.

Be extremely careful with units which create stupidly small or stupidly long numbers!

If you really must use moments in Nmm (and stresses in N/m2), then always use dual representation. One number must be a real world unit which folk can visualise. e.g. “MX = 15.5 kNm (15,500,000 Nmm)”.

(I suggested this to a vessel design software supplier. However, I am not expecting changes anytime soon).

The calculation above with the disappearing nozzle loads had three errors.

A second error was the imposed nozzle bending moments. The bending moment loads had been input with numbers corresponding to Nm rather than Nmm. The bending moments for nozzle design were therefore 1000 times too small.

The third error was a conversion of kgf to N. Self weight in kgf was entered as N. The self weight was 10x times too small.

The unit errors were also missed by the vessel designer and the national certifying body.

Yes, the size of the support brackets did look out of proportion. The calculated local stresses were also unusually small. Unfortunately, nobody noticed.

A Weighell