Graphs - how to cut everything below/above a threshold

Timbones · 2022-09-18T21:29:03.232Z

If you set the Offset of the graph to -500 and then limit the Y-axis minimum to 0, then I think you’ll get what you’re after.

Robert.Wall · 2022-09-18T22:27:29.816Z

That doesn’t seem to work for me - the statistics now include the hidden negative part.

I think it will have to be done in the Input processing with + and Allow positive etc and/or the conditional processes - something like

If >, skip next [value = ] 500
Reset to zero
Log to feed

though I’m not completely sure what Nikolas wants. Is it that everything between about 400 W and 650 W is “noise” and should not have been recorded?

nikall · 2022-09-18T22:39:33.787Z

Thank you both for the reply, I really appreciate your prompt response!

Well to make the long story short, what I want to achieve is to find “approximately” the consumption of a specific load (let’s say a washing machine, a boiler e.t.c.) for a given period of time in the graph.

@Robert.Wall yes let’s say that in the case I described above, the “static” consumption (those 400W) of my home is just “noise” and I want to get rid of it and get only the additional consumption (and the average of it in KWh).

Robert.Wall · 2022-09-18T22:40:13.062Z

See what I just added…

nikall · 2022-09-18T22:58:37.177Z

Robert,

This looks to work but it creates a problem. If I do that in my current input (Power1) then real time consumption shows 0 (since it’s below the threshold). Can I do it in the Power1+Power2 input and get the result in a separate graph alone ?

thanks in advance for your help!

Robert.Wall · 2022-09-18T23:00:02.051Z

Even better, after your present input processing, do Reset to Original and start again… logging to a new Feed.

nikall · 2022-09-18T23:03:45.065Z

Hmmm… I’m not sure how to do what you just suggested

Robert.Wall · 2022-09-18T23:04:43.085Z

What were your original process steps?

nikall · 2022-09-18T23:06:19.251Z

I did what you suggested in your initial response and when I saw that the real time power was affected I just removed the processing steps. Should I add them again ?

Robert.Wall · 2022-09-18T23:08:27.727Z

What did you have when you posted the first question 2 hours ago?

nikall · 2022-09-18T23:12:14.410Z

nothing except the default feeds/input/graphs of the emoncms (I’m not sure what you ask…it’s probably because it’s 2 AM here and my brain is sleeping already )

Robert.Wall · 2022-09-18T23:17:57.235Z

I think you probably had “Log to Feed”. This is the minimum you need.
Let’s say it was

Log to Feed Watt real time

You add Reset to original, then the steps as suggested, so that you have

Log to Feed [value = ] Watt real time
Reset to original
If >, skip next [value = ] 500
Reset to zero
Log to feed [value = ] My Big Load

And My Big Load on a separate graph.

Timbones · 2022-09-19T09:04:26.165Z

You could also do it with a virtual feed, with the advantage that it takes up no more space, and can process all previous data as well.

Click ‘New feed’ at the bottom of the Feeds page, add a virtual feed, then check the box next to it and click on the spanner on the top bar.

Add these process steps. Note there’s no need to log the result.

nikall · 2022-09-19T15:38:12.360Z

Guys thank you very much for your replies. You really make a difference in this forum by helping people like me.

Well, solution by @Timbones works better indeed. It gives me all the big loads using historical data as well. To be honest this works better for what I have in my mind to use these data for.

The solution by @Robert.Wall works as well meaning that it filters the big loads but it doesn’t process the historical data.

Now I have a new question which arises by comparing the two solutions.

image1205×1024 50 KB

If you look at the statistics tab you will notice that the same load for the same period of time has different Wh values depending on how I have implement the post-processing. Any clues and which one has the correct value ?

The only thing I can notice immediately by looking at the stats is the quality difference in the measurement.

Thanks in advance for your help!

Robert.Wall · 2022-09-19T15:58:04.589Z

In the context you have, Quality is meaningless. It refers to the number of data points (not counting NULL values) as a proportion of the possible number plotted across the visible width of the graph.
Notice “the possible number plotted” because the graph algorithm, if there are too many points requested to plot across the width, selects individual points to plot - so if it happens to pick several missing (NULL) values, the Quality is no longer representative of the quality of the original data.

What you need to do to properly compare values is look at the original data in enough detail to be able to manually calculate the energy (Wh) in that spike, then compare it with the number that each method gives you.

Note ‘my’ method includes the base load in the value passed on. To exclude it (as a fixed value - you can’t do any better) you must obviously subtract it (add “+ [value] -500” between “Reset to zero” and “Log to feed”.

nikall · 2022-09-19T18:27:53.526Z

Hi Robert,

I added the additional step in the post processing you suggested. Now looks much more similar to Tim’s solution.

image1264×1075 70.6 KB

I used the Watt Real time Feed which is the default and measures all the loads as reference. Using the zoom in function of the graphs I have isolated the part where I did the measurements. I used as a load a microwave oven which is about 1500W and fired it up for 1 min. So taking the Watt Real Time graph we can see that during this 1 min the difference is 1548 Watts which is very close to the nominal power of the oven.
So we have, 1548 Watts x 0.0167 Hours (1min) = 25.85Wh which looks very close to both yours and @Timbones solution.

But as I said, Tim’s solution serves better my needs as I wanted to know historically the big loads in order to size a PV system which I’m considering as a solution in order to save money from energy bills and provide an alternative power source during grid outages.

Well thank you both for your precious contribution especially you Robert because you have helped me in the past so many times.

Well, If I decide to install the PV I will come back here to ask about integration in the emoncms.

P.S.

One more question to @Timbones , how can I do the same for small loads ?

Timbones · 2022-09-19T19:42:23.882Z

nikall:

One more question to @Timbones , how can I do the same for small loads ?

Have you looked at the histogram feature? Here’s my consumption for today, showing how long was spent (seconds) at each wattage level. Maybe that has the information you’re looking for?

image1047×570 24.1 KB

Also try the “kWh at Power” option at the top, to get a different perspective.

nikall · 2022-09-19T20:23:10.856Z

Yep! It’s just fine. Sorry for my obvious ignorance of the functionality of emoncms but I hope this will drastically change in the future especially if I install the PV.

Thanks again for your time and help!

Robert.Wall · 2022-09-19T22:51:05.412Z

nikall:

But as I said, Tim’s solution serves better my needs as I wanted to know historically the big loads in order to size a PV system

With every respect, you failed to mention that in your first post. We can only know what you want help with if you tell us.

nikall · 2022-09-20T08:08:56.121Z

You are totally right Robert, it’s my fault. I just wanted to write a short intro post without going into details about how it was useful to me. Lesson learned…

Thanks again for your help and understanding.