Knock sensor deployment on industrial engines

[Lou Scannon]

What is the conventional wisdom on deployment of knock sensors on industrial engines - e.g. number of sensors per bank, location of sensors?
On automotive engines, 1-2 sensors for the complete engine is considered quite sufficient for up to 8 cylinders. I understand that a higher concentration of sensors is typically used on industrial engines. What drives using more than 1 sensor/4 cylinders (assuming a capable knock DSP with the necessary windowing & processing speed)?

 

[Johnstew]

The Caterpillar G3516 engines we have use one sensor on each side of the engine. They're rated 850 kW at 1200 rpm with sixteen cylinders. They seem to work fine, and we've had them detect knocks from mechanical problems in addition to combustion knocks.

/signed/ Hammerhead

 

[dcasto]

The larger CAT 3600 series have one on each cylinder

 

[Lou Scannon]

My question is, what is the technical driver for not sharing one knock sensor among up to 4 cylinders as is done for automotive engines?

 

[dcasto]

On such large machines, there can be a significant difference because of mechanical problem between cylinders. A bad exhaust valve on a little engine gone undetected might do $10,000 damage, these engines cost $1.5MM each. Not only does CAT monitor and adjust for detenation, they also compensate AFR, exhaust temperatures, and there is technology to measure BMEP inside the cylinders. So whats a $50 knock sensor vs $1.5MM and have that engine run 4400HP 24/7?

 

[Lou Scannon]

So to summarize your answer, unlike automotive applications, there is no justification for doing engineering work to reduce the knock sensor count, due to the insignificant cost of the sensors relative to the whole machine & its life cycle cost.

 

[dcasto]

OK, I said on larger engines, the small cost of sensors compared to damage that goes undetected will pay for the installation of knock sensors.

In real life, if a knock sensor detects a problem, it will automatically adjust timing to that cylinder first, then give a warning, then shut the unit down.

Cars do not have an ignition systen that can individually adjust timing (spark timing)unless we put in a more complex programing, so that protection would not be available without more cost to the automobile. Next, we would see more "check engine" lights and customers would be mad.

Finally, would we want our cars to shutdown? So we live with our cars having undetected knock and failures. As a consumer, if an engine fails because of knock, the car company will ponit to the fuel, fuel maker to gas stations, gas stions to the engine, and the consumer is stuck holding the bag. Its the whole risk management game.