Is knocking/detonation possible in gaseous fuel fired Otto engine? 1

[ebarba]

Hi all!

We're running an 8.0L turbocharged engine on wood gas to drive a stationary electric generator, rated 100 kW.

Wood gas composition is essentially:
20% H2
20% CO
5% CH4
55% N2

We are experiencing loud "clanking" noises coming from the engine when we approach rated power, the conditions of operation are as follow:

Engine compression ratio: 9.5:1
MAP: 165 kPa (24 psi a)
Mixture type: Very lean (sorry, no exact value as we're using a narrowband O2 sensor)
Coolant temp: 90ºC (194ºF)
Engine speed: 1500 rpm
Power at the engine shaft: 93 kW
Spark advance: 12º BTC

The knocking (or supposed so) develops when we try to increase the AFR to push further the engine towards the expected power target.

I thought it was "impossible" to experience knocking with gaseous fuels, especially with Hydrogen in the mixture and so small an advance. On the other hand, the overall compression ratio (considering boost) is quite high and probably the engine is running a bit too hot.

Spark plugs do not show evidence of knocking or detonation.

Are this noises really knocking?
We will try again with lower advances (down to 5º), is this a good idea?
We will also try with richer mixture from the start (near stoich) next time, keeping it that way while increasing the power. We hope we can reach the target power in this way with a lower boost from the turbo, hence lowering the compression ratio.

Eagerly awaiting for your feedback!

Thanks and regards!
EB

 

[Hoxton]

Sorry, but it is very possible to have knocking on spark ignition gas engines, especially with significant quantities of hydrogen in the fuel!

The factor used is called "Methane Number" - I suggest that you search for it.

Essentially, pure hydrogen has a methane number of 0 (zero) meaning that knocking will occur. Natural gas has a methane number between 70 and 80, depending where it comes from.

You can download methane number calculators from caterpillar website among others.

My guess is that you have a MN of about 50

None of this will help you solve the problem!

The gas engine manufacturer will assign a rating based on the methane number and other gas characteristics. As you have what I guess is a self converted engine (ie the original engine manufacturer did not ship this as a gas engine) then you will have difficulty defining a rating for this engine on this fuel.

All I can suggest is that you increase the power until the engine knocks and then reduce power until it does not knock!

Warning: knocking is due to the gas exploding rather than burning, the peak forces can result in engine damage.

Hydrogen is a difficult gas to deal with, since it is explosive and can pass through mild steel pipe walls, for example.

The appropriate safety precautions must be made

 

[Hoxton]

I forgot to add:

You can buy knock detection systems from some of the engine control and governor manufacturers. The one I know will reduce power in the event of knocking and shut down if it is excessive

 

[140Airpower]

Don't forget that liquid fuels become gasses under conditions of combustion. So, since they can knock, "gaseous" fuels can knock. But more convincingly, all common fuel gases have octane numbers, meaning they have been tested for their knocking behavior.
Wood gas is supposed to have a fairly high octane rating. Is it clean of dust and partially burned soot? Smokey?

 

[romke]

knocking can occur with any fuel that normally needs a spark to ignite when the conditions within the combustion chamber are such that part of the fuel ignites spontaneously before the sparkplug induced flamefont has reached it. whether that happens depends on geometry of the combustion chamber, the position of the sparkplug, the ignition timing and the A/F ratio. in this case the geometry of the combustion chamber and the position of the sparkplug cannot easily be changed, the A/F ratio and the ignition timing can. you could try running the engine a little richer and / or experiment with the sparkadvance.

be aware that you may need a more or less constant spark advance, when measured in milliseconds. when setting the spark advance on the base of degrees BTC it will only be optimal for a particular speed and load combination. you could start out by reducing the sparkadvance when knocking occurs and see whether the knocking decreases.

 

[Lou Scannon]

It seems to me that no one on your team really understands what you are doing, and advice you receive over the internet is unlikely to compensate for that.

"Schiefgehen wird, was schiefgehen kann" - das Murphygesetz

 

[ebarba]

First of all, thanks to all of you for your time and advice, especially hemi, for his incredibly high contribution to my lack of knowledge in this subject.

Back to technical matters, I should have explained more in detail my concern: detonation is technically possible with almost any combustible mixture (hence the "" on impossible in my original posting), but:
[ul]
[li]wood gas should be really very hard to detonate[/li]
[li]there is no physical evidence of detonation on the spark plugs[/li]
[li]there is no noticeable power decrease when we hear the noise[/li]
[/ul]
The above made me suspect maybe there was something else going on.

Some more data on the application:
[ul]
[li]The engine is manufactured by Origin Engines and is specifically built for gaseous fuels (NG and LPG), albeit not specifically for wood gas[/li]
[li]The manufacturer did not provide any rating for the gas, but said something like "you get what you get"[/li]
[li]This generator set is intended to always work at rated speed and power, only +/-10% variations will be present. In any case, the ignition module can handle variable spark advance maps based on speed and load[/li]
[li]The gas is filtered, but probably still contains some very fine dust[/li]
[/ul]

Specific questions:
[ol 1]
[li]Granted this was knocking/detonation, are we better off increasing the fuel from the beginning, then increasing power and finally fine tuning spark advance, or changing the spark advance and then power? NOTE: Controlling power is not very easy in this application, tolerance is more than +/-10%, due to control issues[/li]
[li]When changing the spark advance, any idea of where to set it to begin with? Should we start with a very high advance and try from there?[/li]
[li]We would like to run this engine lean, how does this change the spark advance, compared to a stoich application?[/li]
[/ol]
Thanks again for your help.

 

[Tmoose]

Which engine do you have?

http://www.originengines.com/Engines.aspx

Does yours have "piston oilers" ( I assume that squirt to cool the piston crown.)

http://www.originengines.com/DifferenceMakers.aspx

What is air inlet temperature?

I'd start by pulling the initial timing (and thus the entire advance curve) back 5 degrees (if that can be done by twisting a distributor, etc) to see if the clanking improves or even goes away.

Running richer than stoic with liquid fuels reduces likelihood of detonation in part by the cooling effect of evaporation. I have no idea if there is much detonation resistance with "rich" gaseous fuels.

 

[gruntguru]

Lean mix will require more advance than stoich.

je suis charlie

 

[catserveng]

Here is a good non-manufacturer specific paper discussing hydrogen in internal combustion engines, both as a straight and blended fuel,

http://www1.eere.energy.gov/hydrogenandfuelcells/tech_validation/pdfs/fcm03r0.pdf

Also, here is a Caterpillar Gas Fuel Guide you may find provides some additional technical information,

http://www.toromontcat.com/powersys...PLICATION & INSTALLATION GUIDE - LEBW4977.PDF

I have dealt with a couple of wood and producer gas projects on smaller (at least to me) engines. Your fuel composition can contribute to "knock" in most engines.

9.5 to 1 compression is likely a bit high for a low methane number fuel like yours. Is this a natural aspirated or a turbocharged engines? Inlet manifold temperature will have an impact as noted above.

Instead of guessing at your air/fuel ratio you should probably use a portable emissions analyzer to help tune, especially with a fuel and an engine that were not designed to work with each other.

In general to reduce knock in a gaseous fueled engine, retarding timing and leaning the fuel mixture are two common ways to increase detonation margin. Also reducing charge air temperature helps as well. You may also find that this fuel has a fairly small operating range between lean misfire and detonation, at least in my experience with similar fuels.

What kind of ignition system do you have? I would assume if it truly an "industrial" type engine it would use some kind of electronic ignition. Also you mentioned a "narrow band O2 sensor", is it in conjunction with an air/fuel ratio controller? Or just a monitor?

I have also found on these types of fuel gases that exhaust port temperatures can get extremely high very quickly, and while this is a relatively small engine, since it appears this is a bit of a science experiment then adding some additional instrumentation such as exhaust temperature monitoring and possibly knock detection may help prevent an engine failure.

Finally, as to your load control system, are you independently supplying the load in an island or stand alone mode, or are you paralleled to a grid or larger network? Load instability can also contribute to engine knock if the fuel control can't manage fuel properly with constant load variations, causing it to goo lean and misfire or too rich and detonate.

Hope that helps, MikeL.

 

[ebarba]

This is the exact engine:

http://www.originengines.com/Portals/0/OriginEngines_8Ltechsheet.pdf

As a summary:
8.0L turbocharged, spark ignited engine
Power required: 110 kW at flywheel @ 1500 RPM
Fuel: wood gas (20% H2, 20% CO, 5% CH4, 55% N2)
Ignition: electronic, fully configurable, coil-near-plug
MAP: 165 kPa (24 psi a)
Mixture type: Very lean (sorry, no exact value as we're using a narrowband O2 sensor)
Mixing system: A/F controller, turned OFF during test because it was making it more difficult to reach steady state (PID not tuned yet).
Coolant temp: 90ºC (194ºF)
Engine speed: 1500 rpm
Power at the engine shaft: 93 kW
Spark advance: 12º BTC
Turbocharger discharge temp is about 80ºC (180ºF)
The turbocharger is handling air-gas MIXTURE

About the load: we're paralleling to the grid via an asynchronous (induction) generator. Essentially, we get more load by increasing engine speed and this is the power control implemented right now. The electric slip of this generator is very small (10 RPM), so small variations in speed lead to large variations in power. We are planning to switch from a speed-based control to a power-based control soon to make power variations smaller.

My conclusions from all the info you guys gave so far would be:
[ul]
[li]Start with richer (possibly stoich) mixture and increase power[/li]
[li]If knocking occurs, increase advance[/li]
[li]Fine tune changing A/F ratio (this will require changing the O2 sensor to a wideband and turning ON the A/F controller)[/li]
[li]Use cold spark plugs[/li]
[li]Decrease inlet temp and possibly coolant temp[/li]
[li]Test the engine at different loads to properly map the ignition module (this will require a change in the power controller)[/li]
[/ul]

Does the above sound right?

 

[Tmoose]

" If knocking occurs, increase advance "

Retard the timing to reduce knock

 

[ebarba]

We just made a very quick test at the same stated conditions ONLY changing spark advance (no power or A/F ratio was modified). There was no real change from 20º to 10º spark advance.

We didn't feel comfortable running the engine further, so we shut everything down and decided to wait to fully implement the strategy of my earlier post. Tmoose: I meant to retard timing, not to increase it.

Requiring less than 10º advance on an engine running at 1500 RPM makes me very suspicious... is this a case of pre-ignition or pre-ignition induced knocking?

Am I just becoming paranoid?

 

[Tmoose]

Is 10º the advance at speed, under load, or the initial setting?
What is the trigger for the ignition.
How are you measuring the timing? I'd be using a dial-back timing light and a tach to map out the actual ignition advance curve.

On older engines with timing set with a mark on the damper, if the damper outer ring slips, or a timing cover with a mislocated pointer is installed, "setting the timing" will produce poor results. One of the first things to do is confirm the TDC mark is accurate within a degree or 2.

A severely restricted exhaust can cause issues with running and power.

 

[ebarba]

the 10º advance is at 1500 RPM and under load.

The trigger is a reluctor sensor and the timing is being measured by the ignition module. There is an old-school damper and pointer, which we checked time ago and showed to be coherent with the ignition module's figures, but we have never measured timing with it since.

The only restriction to the exhaust is the turbocharger and the muffler, we removed the cat a few days ago to rule out excessive back pressure.

 

[ebarba]

Testerday we performed a short test and found that we had steady (frightening...) knocking noise at the following conditions:

1500 rpm
165 kPa (absolute pressure)
80°C (165°F) inlet manifold temperature
15° advance timing
30% rated power
Very lean mixture

The condition improved noticeably at 5° advance, albeit some sporadic knocking was heard.

Knocking seemed to disappear at the following conditions:
1500 rpm
125 kPa (absolute pressure)
74°C (165°F) inlet manifold temperature
10° advance timing
30% rated power
Stoich mixture

Any ideas?

More thorough tests on the way on Wednesday.

Thanks!

 

[catserveng]

So based on the info from your testing above, your first test you had audible knock, then you retarded the timing 5 degrees and richened the AFR, right? The drop in manifold temperature and pressure is due to the richer AFR I'm assuming, did you happen to note the throttle position during this testing as well?

Since retarding timing (which should reduce knock) and richening the AFR (which usually increases knock) seem counter productive, you may be experiencing a lean misfire and when a cylinder does fire it is overloaded. Hard to tell exactly with the info you have provided, but probably likely.

Adding throttle position and emissions data, such as O2, CO and Nox, would help determine what is going on. If you are getting into a lean misfire you would see CO and O2 go up rapidly and Nox start to get erratic, at least based on the industrial engines I have worked with.

Have you run your fuel thru a Methane Number calculator?, there are free ones from CAT and Cummins available online, and while they won't be exact for your engine will provide a better insight into your fuel and what to expect.

Hope that helps, MikeL.

 

[ebarba]

So based on the info from your testing above, your first test you had audible knock, then you retarded the timing 5 degrees and richened the AFR, right?

After the first test with audible knock we retarded timing from 15 to 5 deg BTDC, but NO CHANGE in AFR (kept very lean). This led to much less knock, but still some was audible (very faint and random).
On the second run we dropped the power to zero, increased AFR to stoich and from there we began increasing the power, maintaining a stoich AFR, still with 5º advance. There we met no knocking.

The drop in manifold temperature and pressure is due to the richer AFR I'm assuming, did you happen to note the throttle position during this testing as well?

In all the tests the throttle was wide open. We noticed that, with the richer mixture, the turbo boost required for the same output was lower (125 kPa vs. 165 of the previous test). Probably the decrease in manifold temp was because the turbocharger was working on a lower compression ratio plus the richer fuel or both.

Since retarding timing (which should reduce knock) and richening the AFR (which usually increases knock) seem counter productive, you may be experiencing a lean misfire and when a cylinder does fire it is overloaded. Hard to tell exactly with the info you have provided, but probably likely.

Don't know if this makes sense to you, but generally speaking this engine behaves like this:
- while running at part load and lean mixture, knock appears if we move towards rich mixture.
- If we start with a richer mixture and with such rich mixture reach the same load as above, we end up with less boost, the same power as before and no knocking.

The problem is that we haven't been able to increase load past 50% rated power.

I believe this is a control-loop issue in that
- when we start lean, the engine is compensating with higher boost and higher efficiency (i.e. the same power with less fuel), but at such high compression ratios it knocks.
- when we start rich or stoich, the engine needs less boost to get the same output, runs on a lower efficiency but cooler ---> no knocking.

Have you run your fuel thru a Methane Number calculator?, there are free ones from CAT and Cummins available online, and while they won't be exact for your engine will provide a better insight into your fuel and what to expect.

We haven't found a MN calculator that accounts for CO (not CO2) which is 20% of our fuel. Searching the web, a couple of references point wood gas at MN=70 (this is the one that looks more accurate:

https://books.google.it/books?id=Cf...#v=onepage&q=wood gas methane number&f=false)

Tomorrow we'll make more tests, if you have suggestions I'm all ears!

EB.

 

[gruntguru]

When you experience loud ("frightening") knock, does it sound like all cylinders are knocking?

je suis charlie

 

[ebarba]

hard to tell, it's like heavy metal grinding. If I'm not wrong, there should be 100 knocks per second with all 8 cylinders are knocking steadily @ 1500 RPM, but I'm afraid my hearing isn't that educated to tell. If forced to give an answer I would tell yes, but the truth is I don't know.

If we hear that again (and I hope not...) I'll try to record and post it.

Thanks!