Fan Speed vs. combustion temp

[Nofossil]

I finally got around to plaing with direct control of the blower on my Eko 25. I'd like to write a smarter control algorithm than what's in the Eko controller. I haven't installed my lambda sensor yet, but I do measure combustion and flue temperatures. Combustion sensor is a bit away from the actual secondary combustion to preserve sensor life.

Seems like the first question is "What's the relationship between fan speed and combustion temp at different points in the combustion cycle?"

Having lots of opinions but no data, I did a fire today with lots of refuel events, and varied the fan speed at various points to see what happened. Here's a chart - fan speed varied from 30% to 100%. Not sure I can draw many conclusions from it, except faster is definitely not always hotter.

Any thoughts or insights? Combustion temp and flue temp are divided by 10 to fit on chart.

 

[BoilerMan]

GREAT DATA Nofossil!!

It looks like 50-60% is the sweet spot, but it really seems to not have as much bearing on flue temp as I'd have thought. Generally the flue temp mimics the combustion temp pattern, and I've been looking at this too long as it is............

TS

 

[Nofossil]

Note: Didn't mention that combustion and flue temp are divided by 10 to fit on the chart. Edited post to reflect this.

What's really confusing is this: I absolutely KNOW that reducing fan speed will at some point reduce combustion temperature. By that token, increasing fan speed MUST in some conditions increase combustion temperature (seems like should be most conditions).

Looking at the chart, that seem to be a little true right after reloading. I'm d&^%$(*d if I can see a consistent response to increasing or decreasing fan speed otherwise.

My in-depth but possibly naive model of gasifier combustion has always been this:

Phase I - Ramp up. Fire has been started and there are some coals. Wood gas is generated, and is a rich soup of carbon monoxide, hydrogen, and complex hydrocarbons. More air = higher volume of wood gas blown into secondary = more heat to a point, but you COULD provide too much air for the volume of wood gas. In that case you'd just be blowing extra air up the chimney, maybe suppressing secondary combustion as well. During this phase you want to increase temperature as quickly as possible. You could have either too much or not enough air.

Phase II - Full burn. Unburned fuel is present. Wood gas is generated in copious quantities. More air = higher volume of wood gas blown into secondary = more heat. During this phase you want to modulate air to maintain a high but not excessive combustion temperature - I was aiming for 1100 degrees at my probe location.

Phase III - Burn out. The complex hydrocarbons have been baked out of the remaining fuel, so all you have is coals - pretty much pure carbon. The wood gas that's being generated is mostly carbon monoxide, which doesn't need as much secondary air. There's still lots of heat being developed, but a much higher percentage is in the primary chamber. Secondary combustion is pretty quiet. Increasing fan speed doesn't make much difference in secondary temps, and too high a fan speed is just blowing extra air up the chimney. During this phase you want to slow the fan down gradually to maintain as high a combustion temperature as you can get, but combustion temperature is going to drop regardless.

The data doesn't seem to support this very well.

So.... given this, what would be a good fan speed control algorithm? My Eko controller runs at 100% until the boiler output temp gets close to the setpoint, then slows the fan a bit. If output temp exceeds the setpoint it goes into idle - no fan at all, with brief periodic fan activations to purge wood gas and prevent explosions.

I have to believe I can do better.

 

[Fred61]

I don't know if I comprehend all the data presented in the graph nor can I see if it answers any of your theories but what I think I see is higher efficiency at higher combustion temps. Is that obvious to you or am I reading it wrong?

 

[TCaldwell]

Nofo,
Batch burning is a moving target, outputs always changing to maintain a given setpoint throught a burn. by varying the fan speed to the degree that is shown is like starting and stoping the fire, not what would happen if you were trying to control it. The results might make more sense if you burned at static fan output and monitored temps, say a burn at 70% fan op, then 80% and so on. this would reference a temp curve to a given fan speed, the temps sec burn and flue will indicate the best max/min fan outputs. Also you can set a target sec burn temp,[your 1100] and manipulate fan speed to try to maintain that for as long as possible, excluding startup and char. One thing, o2 and temp do not have a reliable relationship, espically at startup. Any o2 reading below 1200degf[gasification] is best not considered for repeatable control, if you look at the manuals of most o2 controlled boilers, the actual o2 control starts when a given flue temp is met that represents gasification upstream. At higher secondary burn temps o2 and temp will roughly track. these correlations will differ with different boiler mfgr combustor designs. You maybe could monitor differential pressure, flue/ambient and establish a optimum dp value and manipulate fan speed to maintain dp.It is time to install your o2 sensor!
These thoughts are based upon experience with my cobbed garn, that is not right yet, but I am getting closer. i am currently working on a cascade,[ 2 controller] one o2 and the other temp, in the near future I will be posting some data.

 

[BoilerMan]

All of the combustion peaks are very close to one another reguardless of fan speed (to a point of course). I think you'd need a combustion analizer to determine what your excess air is during different phases of combustion, as you said I'd think it would be most critical during Phase III. A CO ppm would indicate this best. I think to the best fan speed you need more than combustion temps, but it is interesting that the fan seems to not have much bearing on the temps.

It would seem that there would be an optimal fan speed for the heat exchanger to x-fer the heat to the water most efficiently. But also one would want the combustion temp as high as possible without the flue temp rising excessively, indicating not enough heat exchange for the given surface area.

TS

 

[hiker88]

In my unit, fan speed seems mostly correlated to boiler set point. So, with the safety door closed, the fan will run at 85% right up to the boiler set point, and then it will start ramping down to maintain. If the boiler temp gets below setpoint, fan speed will come right back up as high as it needs to to get the boiler temp back to setpoint. As the fire burns down, fan speed comes up and will stay there until flue gasses are below 135c, at which point the burn is complete and the fan shuts off.

Now saying that, I have a programmable set point for flu gas as well and it is set to 250c by default. I've never seen flu gasses above about 228c, so I suppose it is possible that the boiler could be below set point and if the flu gasses were at 250c, the fan would slow down. I believe the description of the data point in the manual says "the temperature in degrees beyond which efficient gassification can't be maintained" or something like that.

I hope this helps.

 

[Nofossil]

There's an unknown offset between actual secondary combustion temps and the combustion temp shown on the chart. I expect that 1200 on the chart corresponds to 2000 degrees or so under the nozzle.

What really struck me is that I'd look at the combustion temp - it would be a bit low so I'd increase fan speed a notch or two. a few minutes later I'd check back. Sometimes it would have gone up, other times it might have gone down. Based on the results, I'd try changing the fan speed again. Sometimes the temp would go up, sometime it would go down. No real pattern, although low fan speeds seemed to give every bit as much output as higher fan speeds on average, except for the periods right after refueling.

I'm thinking this is my approach until I get my lambda sensor installed:

1. After refueling, run at 100% until target combustion temp is reached.
2. Modulate fan to maintain target combustion temp as long as possible.
3. After falling below target temp for 20 minutes, drop fan to 40%.
4. After combustion temp stays below 400 for 20 minutes, turn off fan.

 

[TCaldwell]

the positive and negative temp to changing fan speeds is related to the p/s air settings and how much wood is left in the combustion chamber, if you changed the p/s settings you would see a different temp change related to fan speed change. I think your initial approach is good but if you could monitor o2 also it might show correlations.

 

[TCaldwell]

Ok Nofo, silence usally means the wood is burning, how is the temp/fan control going?

 

[jebatty]

Are you measuring fan speed or cfm moving up the stack? And I'm assuming a relationship between fan speed and cfm similar to pump head and gpm, as well as a fan curve similar to a pump curve??? Maybe I'm giving too much credit to the engineers who design a particular gasifier, in that the system is "tuned," so that fan speed, firebox design, noxzzle design, tunnel design, 2ndary chamber design, firetube design, etc. are all integrated, and therefore simply changing fan speed, or any other single part of the system, may not produce an expected result due to interaction with the entire system.

 

[TCaldwell]

Merry Christmas to all, Jim what you are talking about is very important, linearization if the controller output to the final control element, in nofo's case that the when the controller asks for 40% output it is actually getting 40% of air volume not always reflective of 40% fan speed. This relationship becomes important because the above mismatch if allowed can drastically change the pid tuning gain value. linearization is offered on some pid controllers anywhere from 10/25 points along the existing curve to desired curve values can be input, usally through milliamp values. I did it with a eurotherm q498 signal conditioner module, linearizing controller output to p/s damper cfm, instead of damper position. I am using the stock static speed fan on the garn, linearizing made a huge diference. I used a hotwire anemometer to monitor cfm in the p/s combustion air ducting, also noticing a significant total airflow difference with turbulators and without. With the flue temp and cfm reduced, effectively reducing the btu output but keeping the transfer effiency high, garn's turb might not be as restrictive. The lambda boiler manufacturers would have encountered this stuff, and probably have proprietary controlls to handle, i am not sure the manually adjustable p/s designs can reach the sensitivity, but operate well in a given range

 

[BoilerMan]

I used a hotwire anemometer to monitor cfm in the p/s combustion air ducting, also noticing a significant total airflow difference with turbulators and without. With the flue temp and cfm reduced, effectively reducing the btu output but keeping the transfer effiency high

This is very interesting, hotwire would give nice airflow readings without any restriction in flow. Interesting to see the turb effect though, lower flows, therefore reduced output, but better heat transfer, so it could be a wash of sorts, with reduced fuel consumption as a side effect (in theory).

What type of actuators or stepper motors are you useing for p/s adjustment? Nofo, same question, do they come with your controller, or is it something that you are pieceing together? Overall I'd be interested in the wood reduction, I would imagine that wood quality and consistency would have alot to do with this. Seems like the more consistant the wood, the better the initial tune would best for fixed adjustment, like a carburator. But with variaing species, MC, and size, a "fuel injected" closed loop self-adjusting carburator would have benifits, I'm just curious as to how much overall efficiency is gained as a rough number. Seems as the "moving target" could gain significantly with closed loop feedback.

TS

 

[Nofossil]

My variable speed controller has a non-linear response that's supposed to roughly match typical fan curves so that 50% input results in 50% of peak flow. I have no airflow measurement sensor, so it's a guess at this point. My p/s ratio is set with the standard Eko mechanical gadgets and does not change.

There are some confounding effects - zero fan does not equal zero airflow due to chimney draft, for instance, and I expect that draft affects secondary flow more than primary flow due to nozzle geometry.

Right now I'm adding periodic purge to my PID control.

 

[Nofossil]

Here's a shot of this evening's fire. Still don't have purge cycles on this controller, but it does work pretty well. PID control is courtesy of my Version II NFCS. The wood boiler circ trace is a test - not actually controlling the circ that way yet.

 

[TCaldwell]

Taylor, sorry for the radio silence, the damper actuators are belimo 4-20mA modulating, same as the output of my controller, hard to tell if i am actually saving wood as the controlled burn charachteristics are different from a non lambda. Really the chosen o2 setpoint determines how hot and for how long the burn lasts, a lower residual o2 burns will burn hotter and faster.
Nofo, it looks like the early 100% fan speed started to cool the burn down, too much secondary air , maybe a gradual fan speed ramp for the start

 

[Nofossil]

Here's a burn cycle that exercises the purge cycle logic. There's a 'fan disable' switch for refuelling. When the controller sees that the fan is no longer disabled by the switch, it runs the fan at 100% until target combustion temperature is reached - only a few seconds in this case.

Then the PID logic in the controller tries to maintain 1100 combustion temp, but is not allowed to reduce fan speed below 30%. If it detects that fan speed has been low (below 50%) for three minutes, it will invoke a 20 second full-speed purge cycle.

If the controller is unable to reach target temperature for an extended period of time, it will drop fan speed to 50%. In this case, that actually results in a sustained rise in combustion temperature - presumably because less cold air is being blown through the boiler.

Data is logged only once every 30 seconds, so some purge cycles are missed. I'm pretty happy with this approach until such time as I can add my lambda sensor.

 

[TCaldwell]

What do you think your actual secondary burn temp is with regards to the 1100deg monitoring location, does your flue temp correlate to a high burn? have you tried to reduce the output limits of the fan on the high side, possibly tempering the rate of temp increase to avoid bang bang control, another way also to try would be a fan speed ramp to temp setpoint. with my damper setup the 90 deg stroke time is 60 sec, not sure if this could be equated to 0-100% fan speed in 60sec. but I am able with a pid controller gain=.75% and a integral of 53sec to roughly control temp.