My swimming pool pump is now being controlled with my home automation and it keeps the water temperature between 20,5-21C. It only pumps during day time 09-23. Because it's only controlled by water temperature, the colder it gets, the more pump is used. So examining how many hours the pump is used daily, I can roughly estimate how much the pool uses heating energy.
So I collected the daily pump usage.
This is the raw data. There are some spikes and dips which have to be removed first.The baseline seems to be around 4 hours. The dips happen when I forget to tell home automation I've finished swimming. After that there is always a spike because the pump has to do a 'double-shift' to get the temperatures back to normal. There is also a couple of days with 14 hours of pumping and that was because I didn't use pool cover in those days. It's winter so air is dry and lot of pool's heat will go away by evaporation if the pool cover is not used. My heat exchanger is also old and I have no idea how efficient it is now and how efficient it should be.
I also changed the target temperature from 21C to 22C in 27.1 so the last spikes come when the pool temperature is rising.
At first I removed all the spikes and dips manually.
Looks a lot better but there's still a lot of variation between days.
My home automation also records hourly temperatures from Finnish Meteorological Institute so I took an average for daily temperatures and added those to chart. I reversed axis for temperatures so it's easier to compare with pump times.
There's a slight correlation but not much. The pump hours start to rise after 1.1. Of course the last week is different because the water temperature is now 22C. Actually the reason why I started to examine these temperatures was to find out how much more heating energy is needed after I raised the water temperatures. But there are too many variables in this equation.
This graph shows correlation between air temperatures and pump hours. The correlation is weak, R2 is 0,24 but there is at least some correlation.
But from temperature graph can be seen that pump hours start to rise even when the temperatures stay the same. I think the problem is I don't measure the ground temperatures. The pool has two ways to loose heat, into to the air by evaporation and into the ground by conduction. The pool has cover so only heat is lost into the ground.
As I said, I don't (yet) have any thermometers measuring ground temperatures but I found a graph that shows how ground temperatures change in Finland yearly ( actually in Estonia but it's almost the same in southern parts of Finland ) .
This graph shows the ground temperatures are the lowest in February which makes sense because there's a lot of mass that has to be frozen/heated. Same applies in summer, June is not the warmest month but July and August.
What I was trying to do was to find function that gives me pretty good estimation of what heat loss will be if the function is given enough information. It seems the function needs the ground temperatures more than air temperatures. If I can create a function that estimates the heat loss, I can also create alarms if there's something wrong in pump. I already have same kind of estimation for my computers water cooling. The function uses CPU usage percentage and estimates what the CPU temperatures should be.
The green line is CPU%, red is estimated temp(scaled), brown is actual temp and blue line actual temp divided by estimated temp(scaled). If blue line is something other than around 60, then there's something wrong with water cooling. As you can see, it's fairly flat even when CPU usage and temperature changes.
If I can create as good estimation function for the pool, I can also accurately compare when I change something in my pool system like start using higher water temperatures. It seems a proper way to do this is set additional thermometers into the ground but that has to wait till the summer.
So I collected the daily pump usage.
This is the raw data. There are some spikes and dips which have to be removed first.The baseline seems to be around 4 hours. The dips happen when I forget to tell home automation I've finished swimming. After that there is always a spike because the pump has to do a 'double-shift' to get the temperatures back to normal. There is also a couple of days with 14 hours of pumping and that was because I didn't use pool cover in those days. It's winter so air is dry and lot of pool's heat will go away by evaporation if the pool cover is not used. My heat exchanger is also old and I have no idea how efficient it is now and how efficient it should be.
I also changed the target temperature from 21C to 22C in 27.1 so the last spikes come when the pool temperature is rising.
At first I removed all the spikes and dips manually.
Looks a lot better but there's still a lot of variation between days.
My home automation also records hourly temperatures from Finnish Meteorological Institute so I took an average for daily temperatures and added those to chart. I reversed axis for temperatures so it's easier to compare with pump times.
There's a slight correlation but not much. The pump hours start to rise after 1.1. Of course the last week is different because the water temperature is now 22C. Actually the reason why I started to examine these temperatures was to find out how much more heating energy is needed after I raised the water temperatures. But there are too many variables in this equation.
This graph shows correlation between air temperatures and pump hours. The correlation is weak, R2 is 0,24 but there is at least some correlation.
But from temperature graph can be seen that pump hours start to rise even when the temperatures stay the same. I think the problem is I don't measure the ground temperatures. The pool has two ways to loose heat, into to the air by evaporation and into the ground by conduction. The pool has cover so only heat is lost into the ground.
As I said, I don't (yet) have any thermometers measuring ground temperatures but I found a graph that shows how ground temperatures change in Finland yearly ( actually in Estonia but it's almost the same in southern parts of Finland ) .
This graph shows the ground temperatures are the lowest in February which makes sense because there's a lot of mass that has to be frozen/heated. Same applies in summer, June is not the warmest month but July and August.
What I was trying to do was to find function that gives me pretty good estimation of what heat loss will be if the function is given enough information. It seems the function needs the ground temperatures more than air temperatures. If I can create a function that estimates the heat loss, I can also create alarms if there's something wrong in pump. I already have same kind of estimation for my computers water cooling. The function uses CPU usage percentage and estimates what the CPU temperatures should be.
The green line is CPU%, red is estimated temp(scaled), brown is actual temp and blue line actual temp divided by estimated temp(scaled). If blue line is something other than around 60, then there's something wrong with water cooling. As you can see, it's fairly flat even when CPU usage and temperature changes.
If I can create as good estimation function for the pool, I can also accurately compare when I change something in my pool system like start using higher water temperatures. It seems a proper way to do this is set additional thermometers into the ground but that has to wait till the summer.
Comments
Post a Comment