Homemade Controller

MDNLansing

Has anyone ever tried, successfully or not, to build a residential boiler control unit using the Arduino platform? I am considering creating a unit that is more accurate and predictive for thermostat functions. For example, the unit would call for heat when needed, but predict if the target temp can be obtained by a single boiler cycle and the residual heat that cycle creates. So instead of kicking back in when the low limit is reached on the pressuretrol, just leave the system off and let the radiators finish radiating. Even if the target temp hasn't been obtained yet. Something far more advanced than just schedule and programming like most current wifi thermostats.



I know there are some high end units and commercial controllers that let you do things like this, but I can't find anything for residential installations. Thus, the nerd in me has emerged.



Thoughts?

MDNLansing

Never heard of ot

I am not familiar with the Phidgets platform, but will certainly investigate it. Thanks!

MDNLansing

Great Links

I love that project you did MarkS. Although I'm not looking to modulate my burn rate, I am sitting here thinking why not. I'm looking at a new install and will have 6 months to do it. I certainly will build a custom control unit, and I think my project might be getting a little deeper now. I love the concept of using the burner to regulate steam pressure. I don't think you could possibly get more efficient than that. The wheels are turning in my head...

PMJ

Homemade Controller

I have run my system with a homemade controller for about maybe 10 years now. My purpose was to do exactly what you are looking for - allow the heat from each fire cycle to radiate into the room and not continue firing all the way until the stat is satisfied which always results in overshoot of some amount. I don't use any pressure controls at all - I have an ultra low pressure 2 pipe system - about1000 EDR. I only change the cycle fire times and off times based on how much time from the last call for heat and then during each call I steadily increase the length of the on cycle vs the length of the off the longer it takes to satisfy. I do 3 complete cycles an hour during a call for heat under all conditions and my boiler never runs more than 50% of the elapsed time during a call - ever. It is simple too - no new sensors to maintain and only on and off times to manage as you learn your own system and structure.

I have written a number of posts about this in the past. The result has been dramatically evened out heat with rads gently warm all the time instead of going from dead cold to piping hot as was the case with the run till satisfied approach.

MDNLansing

No Sensors

So PMJ, you're doing this without taking into account heat loss and outside temperature? Does your program adjust for longer of more frequent cycles the further into winter you get? If I'm reading your post correctly, you're basing the cycle times and frequency based on time models and not actual conditions correct?

PMJ

Adjustments

I'm happy to go into detail about this if you are interested but I will try to summarize quickly here. If my system is off more than 4 hours it will run flat out for 20 minutes - the minimum time I have determined I need to get the top of all my rads warm 2-3 columns in from fill valve from a dead start will all piping at room temp. From there I start at 5 min on and 15 min off for 3 cycles. If the stat is still not satisfied there it switches to 8 min on and 12 min off for another 3 cycles. If still not satisfied it switches to 10 min on and 10 min off which will surely satisfy my structure at -20F outside.



A single call for heat may not be satisfied for 3-4 hours. But the temp is steadily rising the entire time inside one degree and I never see the stat show anything but the setpoint - it doesn't show to tenths. You get the idea - I am just spreading out the whole process which evens out the heat a whole lot. When it is really cold the boiler is never off the 4 hours and the first hour after a call it only runs 15 min total - but the pipes are all hot and with the system in natural vacuum the steam from those short first 5 min cycles shoots to the rads instantly and the structure is still warming the whole time - slowly.The system adjusts itself to the conditions. when it is in the 40's outside it never gets to the 10 on/10 off High mode.



Hope this makes sense - I'll provide as much detail as you wish. I have done it for years now and love it - and I know I am saving money.

PMJ

Trying to calculate

instantaneous heat loss based on outside temp looks like a project to me as sun loads and wind change the load dramatically at the same temp. I tried modulating manually with the main gas valve and found more uneven heating at low fire levels. Also, I was concerned about flue losses leaving the chimney wide open for so much longer a period than what I do now and I don't have a direct outside combustion air supply.





The really nice thing about what I am doing is that there are no extra moving parts - just the simple programmable logic controller in with the tstat as an input and the main gas valve as the sole output. It turns out that you can develop a system that adapts itself very well to the actual current conditions just based on call times(both calling and not) and run times. You aren't locked into anything either - you can just keep tweaking. Just plug in your laptop and modify.

PMJ

Science Project

You definitely have taken it to the science lab level - I admire your energy.

I wanted to stay in something without too many variables and took care of itself. I have been pleasantly surprised at how much improvement I could get so simply. I even have an sheet tacked up with instructions on how to easily go back to on/off control - should my wife ever need to call a pro when I am not in town. What does your system do when there is no internet connection?

Yes, there are function blocks and analog I/O. I nice platform really at a low price with free software.



I have a standard atmospheric burner and standing pilot. Never looked at combustion efficiency. Frankly, I like knowing I can have heat with no power at all. It is a trade-off but I think I am doing great on gas with a system this size.

One thing I am considering is to look at gas cfm/EDR/degree heating day as a way to compare the efficiencies of two different structures. What do you think?

MDNLansing

Mix It Up

I am envisioning a combination of things here. So for instance, on warm days, start at short drawn out run times and cycles. But on cold days start more aggressive with longer run times and more frequent cycles. I like the idea of modulating the fire rate, and modulating the cycles as well. I'm working on some math models now to try and predict the run and cycle time needed for the next say 4 hours. So if it's warn out the radiators just constantly stay partially warm, when it's cold they fire hot and slowly fade to warm. 'm also looking at a separate function to maintain the water temp for days that will need more cycles. My idea is to use a small separate burner (or only a portion of the main burner) and pump water through the boiler and maintain water temp. How efficient this is, is still up in the air. I don't have enough math done to know if this is feasible or not.



I have a vacuum vapor system from 1900 and I'm fitting all of this into all the original piping. I will be assisting the vacuum though by using an air pump when the final cycle completes to speed up the vacuum process. Is amazing how fast the radiators heat when you go from 1.5 psi to 15" of vacuum in 30 seconds. In the spirit of the old dead guys I'm using the steam pressure to draw the vacuum so I can avoid using an electric pump.

PMJ

Calculation

I ran my Oct/Nov data with your formula and got 8.5 using sqft and 34.4 using EDR. OK I guess for 1926 brick construction - 2 floors with unfinished  attic- not one shred of insulation anywhere, and I run a large rad in a 20x22 garage. The boiler is 1950's Bryant 460,000 BTU input - steam mains and dry returns are insulated.

vaporvac

Steam under Vacuum

I'll leave to those doing these experiments to comment more fully on your ideas. I just wanted to point out (as you probably know) that the vacuum determines the temp at which steam is made. For instance, 15.94 " of HG vacuum would give you steam at 176deg. So perhaps you could somehow modulate your vacuum to get your target water temp.

You could get a low fire if your burner has a multi-stage burn rate. Of course Mark's Midco burner is the ultimate, but $$$. It is the perfect thing for the vaporvacuum systems as it can most closely approximate a coal fire. C

MDNLansing

Give and Take

Something often overlooked about vacuum is actual temp. While it's true you can generate vapor at lower temps, it also means the steam is a lower temp. My theory right now is to use the vacuum to boil at a lower temp in order to extend steam production. So we boil at 212, then lower the pressure and continue to boil down to 175. What I am testing right now in my shop is what the average steam and radiant temp is when this happens. Are the radiator temps on the high end, low end, somewhere in between? Not sure yet.

MDNLansing

Scale Tests

I am starting some scale tests and various experiments to get some data and algorithms for the controller I will be building. I am wondering if anyone has done this before? My main concern is regarding steam pressure and scale. My test equipment will progressively get more complex and begin to mimic a true steam system. However, I am unsure what direction to take with the scale of steam pressure. For example, if I build a 15:1 scale model, should I scale the steam pressure 15:1 as well, or use normal levels. I am thinking not and that actual pressure should be used regardless of the size of the system, but I am curious if anyone has done this before. Since my house runs on approximately the same pressure as large NYC buildings, my though is that scaling pressure is not needed. Any feedback on this would be appreciated.