Built Under Pressure: Part 2- Field-Proven Compressor Wisdom
The Realities of Compressor Startup When the Pressure’s on
I stared at the monitor display of the DCS carefully. We’ve tried it twice already and time was running out. System pressure was falling rapidly.
This shouldn’t be this difficult.
This was the thought that kept popping up in my head after trying startup the reserve compressor after the primary one suddenly shutdown.
See, it is very common to design a compressor system where to start it up, you just put the compressor on “full-recycle.” It starts off with very little load (i.e. discharge pressure is approximately equal to the suction pressure) and the board operator, field operator, or both will slowly restrict the discharge using the recycle line and build up discharge pressure. When ready, you cut the compressor in. Of course, if you miss the timing or pinch too much, you can build excess discharge pressure and open the pressure relief or safety valves.
In this case, there was a hand-off where the field operator started first while gradually passing control to the board. The valves the operator was using had seen enough wear that control was very difficult.
“Okay — we got it!”
“10–4, take it from here”
It worked I thought with a sigh of relief.
Later that afternoon, we found that the valve stem on the recycle line was sticking. It would over-react and over-shoot the set-point the operator wanted. No wonder.
In Part 1, we covered a lot of important, but basic things. It’ll be much more helpful to you if we start seeing how industry actually implements compressors — it’ll also make the above story make even more sense.
Cardinal Sins to Avoid
First, let’s cover some cardinal sins. Why am I introducing these now? Knowing these sins will give you the intuition to understand, or even design, most compressor systems. Unfortunately, I have seen all of these sins done — some I’ve committed myself.
It is important to never, ever do the following:
- Never dead-head a compressor. That is to say, never block the discharge. You will break something.
- Never block in the suction to a compressor. You will drop the pressure of the suction bottle below design, break through pipe gaskets, and/or overload the first and final stages.
- If the compressor has multiple stages, never block in an intermediate stage. You will break something.
- Do not compress gases, or composition of gases, the compressor wasn’t meant for. At best, you’ll over-amp the motor and trip it. At worst, you’ll surge a centrifugal compressor.
- Do not flood a compressor with a liquid. Knock-out pots are used to capture liquids, but sometimes they overflow. Make sure they don’t.
It is important to note that each principal type of compressor, positive displacement or centrifugal, have different ramifications. We’ll cover how they differ.
You’ll notice that I mentioned something called surge. It is a phenomenon unique to centrifugal compressors and it is frightening. The machine will shake back and forth violently, make ungodly noises, and scare the hell out of you. Compressors have torn themselves apart from surging.
Knowing the above, we can basically understand why any compressor system is laid out the way it is. This also includes what instrumentation we need.
What do I need to be measuring?
No one wants to operate equipment blind. You make do with what you have, but it is hard to run plants and equipment without instruments.
What seasoned engineers think about when handling instrumentation choices, is addressing the three most important aspects:
- What are you measuring
- Where are you measuring it
- How is it measured
Since this isn’t a piece on instrumentation design (maybe one day I’ll write one), we’ll focus on the first two: what and where.
For compressors, you’ll want just a few types of instruments that measure:
- Pressure
- Temperature
- Flow
- Power
- Vibration
- Level
Level is for oil reservoirs and knock-out pots that capture liquid so the compressor doesn’t flood. We’ll cover these later.
Vibration analyzers are pretty interesting, but can be difficult to use. Obviously, they are meant to monitor the vibrations produced by rotating components within the compressor. With centrifugal compressors, they are meant to move in one direction with very little translational movement. With reciprocating compressors though, I’ve never had much luck in using vibration analysis to solve problems. Regardless, true anomalies in vibration patterns can indicate issues such as misalignment, imbalance, or wear, prompting maintenance interventions to prevent equipment failure and downtime.
Power monitors measure the electrical power consumption of the compressor motor. In my opinion, power monitoring is one of the most underrated measurements that provide the most utility. Good power measurements can let you calculate efficiency, mass throughput, and even be used as a cross-check with your flow meters. If you have a service with variable gas composition, it can tell you when the molecular weight of the gas is changing. The higher the average molecular weight, the more power the motor uses.
Get Connor Fernandez’s stories in your inbox
Join Medium for free to get updates from this writer.
Ideally, your instruments will feed their data into an advanced control system, such as programmable logic controllers (PLCs) and supervisory control and data acquisition (SCADA) systems.
If you’re unfamiliar with either these, that’s okay. Most engineering curriculums don’t cover these systems in detail. For now, just know that these are computers that take inputs from instruments and push their outputs to stuff that controls the process.
Typical Compressor Layouts
Regardless of compressor type, these systems have to be designed to handle variable flow demand. However, how variable flows can lead to problems. For example, let’s take the following diagram
It’s a simple compressor. The discharge of the compressor dumps gas somewhere and discharge valve is controlled by a valve reacting to the pressure downstream.
This is an extremely poor design that will fail without constant manual manipulation. If the downstream pressure rises too much for any reason, the pressure control valve could close and commit a cardinal sin — don’t dead-head a compressor.
Instead, what if we do something ingenious?
What if, we take any extra gas that the downstream system doesn’t need, and feed it back to the suction of the compressor? We call this a “recycle” or “kick-back” line.
If the downstream pressure is too high, then open the valve of the recycle line and dump more gas to the suction. If the pressure isn’t enough, begin shutting the recycle line. Easy-peasy.
Let’s take a look.
Boom — we’ve created the perfect design.
Just kidding.
Think about the above flow for just a moment. For every pound of gas we throw through the kickback, we have to recompress it. This seems really inefficient, right? At least some part of the power of the motor is going to gas that is not being used. In fact, the gas is also much hotter than when it entered and over time it may cause the compressor to overheat if this recycle line is continuously used. We would need an intercooler, then.
Acknowledging the efficiency problem here, engineers have tried to come up with a number of interesting ideas. Some of these load control options are quite useful, while others have more niche applications. Some examples are:
- Variable frequency drives (VFD)
- Fixed and variable clearance pockets
- Isolate-able internal compressor valves
The most viable alternative to a recycle line is a VFD. Basically, as a compressor spins faster, it pushes more gas. The opposite is also true. In other words, if your discharge pressure needs to drop, the VFD can lower the speed of the motor. Know that you’re doing a bit of trading here — the motor has a sweet spot where it’s most efficient and the VFD put in control will not focus on operating there.
I’ve seen VFD implementations most often with compressors of < 1000 horsepower.
Regardless of the above, compressors with kickback lines are the most common layout in the industry simply because they are simple and they are reliable.
What about Multiple Stages?
Compressor life is really easy when there’s only one stage. So… why would I ever need multiple?
One of my favorite questions to ask junior engineers is:
“aside from increase in pressure, what can you be sure always happens when you compress a gas?”
The most impactful answer in this context is that it gets hotter. The degree of how hot it gets is based on two things: the compression ratio and the properties of the gas being compressed. By far the most deciding factor on discharge temperature is compression ratio.
If you recall from part 1, the plant I managed operated at 300 bar. Let’s say I had to compress air from an atmospheric ~1 bar of pressure to 300 bar. How hot does it get?
Trust me when I say it’s over a thousand degrees Celsius. A house could burn down and the fires inside wouldn’t get that hot.
“Well so what, I’m compressing with metal machine — metal can handle it!”
Truthfully, it’s really impractical and very expensive to handle such wild temperature changes. It can be done (almost anything can with enough time, money, and patience), but should it? Industrial furnaces and boilers often take 6–12 hours to heat up so the metal doesn’t distort and the refractory doesn’t break from uneven heating.
Instead, we’re going to limit the compression ratio so the discharge temperature doesn’t get so hot. On top of that, we’re going to cool the gas off just before we compress it again. Also, it’s way more efficient to compress gas this way.
As you can see, it’s similar to copy/pasting a single stage multiple times. So why’s it so much more complicated when comparing to a single stage setup? Because of it’s operation.
Give this case study a shot:
A 5 stage reciprocating compressor is showing a step change in the 3rd stage discharge pressure (4th stage suction). It is dangerously close to the 3rd stage discharge PSV set-point, but all other pressures are reading normally. The 3rd stage discharge is also running hot.
If the PSV relieves, not only will the compressor have to be shutdown, but the relief must be reported to the appropriate authorities. Management wants to minimize the downtime, because the plant product is sold out and compressor downtime means cutting production rate. This would lead to unfulfilled customer orders and potential liability to the company.
What is the most likely cause for this behavior?
Unfortunately, I’ve been in this type of situation more times than I ever wanted. Now, to figure this out for yourself, we’ll have to cover topics we haven’t reviewed yet.
The answer is that the 4th stage discharge has had a component failure that is limiting mass throughput of the stage. It is most likely a broken suction or discharge compressor valve, but it can also be the piston ring if the stage is double-acting (i.e. compresses gas on both sides of the piston).
How did I get to this answer?
We’ll cover how in the next part. See you soon.
