|
Elimination of Noise from Charts [ Read Responses | Return to the Index ]
Posted by Bill on Friday, 3 July 2009, at 10:34 p.m. Bob said in another thread: "On a different note, several months ago you suggested that I should using averaging for my long term graph to eliminate the noise. So how do I do this?" Bob, what I do is set up Filtering on key variables that are 'noisy' (such as EWT, Loop Water Delta T, and Heat of Extraction/Rejectio, among others). For EWT I use 'f' and 'F' accumulation codes to give me the hourly, daily, and 30 day rolling averages. You can see plots of all three of these in my charts (WEL0043). EWT is pretty noisy because the water heats up quickly in my 135° (F) attic in the pipes when water isn't flowing. The constant starting and stopping of the GSHP units results in noisy instantaneous EWT values. Loop Water Delta T, also pictured as one of my charts, is very noisy, for the same reason. In this chart, though, I'm wanting to be as close to real time, with the smallest amount of time lag but without huge spikes. So I qualitatively settled on a 12 min. time lag. See the instruction manual on how to do this (set up a Time Constant with appropriate value). And HE/HR gets noise for similar reasons. This one I had to 'dampen' by 30 min. to 'chop down' the spikes to something reasonable. See the chart. Hope this give you some ideas to start from. Best regards, Bill My Web Energy Logger Site (WEL0043)
This site is all about building a cool, energy efficient house, that makes maximum use of earth sheltered design, passive solar heating and cooling, geothermal exchange energy management, and right sizing of the house for it's designated use. The home's placement is on a south-facing hillside in Deep Creek Lake, Maryland. This site describes the design process, the technologies used and the expected results. We also have a comprehensive Links Page for anyone who is also interested in designing a similar project. |