By Andrew Perks

Where do the months go? What a year 2020 was, coronavirus, USA elections, lock downs and all the usual life-issues. 2021 is going to likely be a struggle year but surely it holds more promise than the one we have just endured.

I have mentioned before some of the advantages of lockdown for me. One of these is the technical webinars that are now available. Being an IIAR member I have access to loads of information. I watched an interesting webinar on hydraulic shock in ammonia systems which taught me a lot. It has always been my contention that we need to listen to our plants because when they start making weird noises we need to ask why.

There was an incident in August 2010 at the Millard facility in Alabama where the system suffered from hydraulic shock releasing 14 500kg of ammonia. I had already looked at the incident but the IIAR webinar went a lot deeper. The piping at an end cap of a 300mm suction header catastrophically failed.

Now, I thought how does that happen when it is not in the flow path? I have seen end caps come off on hot gas headers due to condensed hot gas forming liquid in the header where due to velocity pressure acts like a liquid slug. That’s all pretty understandable and easy to design out of a system, but this header failure was something different.

Ammonia has an expansion rate of 850 to 1. That is 1ℓ of liquid will form 850ℓ of vapour – pretty massive and is why we don’t want to lock liquid up in a system. I have heard recently of two condensers that have been locked up full of liquid and left to stand. They didn’t do too well, Both failed and it was estimated that they developed internal pressures of between 80 to 100 Bar prior to failing. As I said, that is understandable.

On the other hand, we need to consider the condensation of vapour. That said, 850ℓ of vapour when condensed will result in 1ℓ of liquid. This creates a vacuum. I never really appreciated the forces at play when this happens.

At Millard, similar to a hot gas header with a liquid slug being driven by a hot gas stream, the mentioned suction header had a slug of liquid driven by a vacuum when the plant conditions permitted the suction gas to condense. Same difference, but the pipework was sitting at -40ºC. That’s a different set of circumstances. Now we are looking at brittle fracture, and did it fracture!

You might ask how did this happen? We have plants all over South Africa and this doesn’t happen to us. Well, I beg to differ. I was recently at a plant where the end cap of a blast freezer coil ruptured releasing 4 000kg, and it’s not the first. These events were also covered in the IIAR webinar. Sometimes we just don’t really appreciate the forces at play in our systems. Quickly pushing pressure and temperature up in a cooling coil then rapidly reducing the pressure and temperature has its issues.

As I said, a well-designed ammonia plant is not supposed to moan and complain, especially not after or during a defrost cycle. The event at Millar is reputed to have been caused by a plant operator interrupting a defrost cycle where a whole load of different forces was put into play resulting in a hydraulic shock incident.

There are some interesting circumstances around this event. The plant had a power outage for 7 hours and on start up a particular room went onto defrost. The plant operator interrupted the defrost cycle to get the room back onto cooling, and was probably told to get that room on temperature and how did he do that? You got it. He went into the PLC programme and did a few modifications with disastrous results. Operator error, lack of risk assessment, bad instruction from someone who should have been aware of the possible consequences and a whole lot more.

Again, the lack of training and appreciation of actions, and their consequences.

As we are still using old-world technology in South Africa most of our defrost systems are hard wired and follow a specific sequence so this type of incident is highly unlikely but not impossible. It’s up to us as an industry to ensure that plant operators and supervisors fully understand the consequences of their actions and that before we institute any changes or new procedures, a full risk assessment is carried out to quantify the risks and avoid rushing into responses.

Funny how there is never enough time to get it right the first time but there is always enough time to fix something.

I have recently been involved with an MHI report on a plant in Cape Town and I have been looking at the hazard analysis – some interesting issues to discuss next time. So be safe until then. 

Click here for the latest issue of RACA Journal