help DATA TO BE ANALYSISED BUT DO NOT KNOW HOW?

[dhyperman]

Hi
The is a theatre design I have built, with two levels on it and two platforms either end, at mid height. picture a long retangular space, divided into two levels with aplatform at each end. this is all one enclosed space. I have a temperature data for several areas inside it. The room has to perform in a specified ranges. the readings i have are for every 5 seconds when the ac starts from cold. I need so way of analysing this data using stats. the majority of the said data for the temperature is out side the desired range. what should I do...

 

[GregLocock]

Plot it each reading time. Plot the average of all the readings against time. Calculate the time it takes for the best, worst and average reading to get within spec.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[corus]

Depends on what you want to know. If you're wanting to know the temperature after a certain period then put a trend line through it and project the curve forward. If the temperatures are too high then turn up the ac or install more units.

corus

 

[PSE]

If you are interested in the temperature "map" of the space and have the capabilities of generating 3D charts, you could plot the sensor location relative to the space and then the temperature values as the ac came on. Some programs can generate a "quilt" or best fit surface based upon the individual data points and you would be able to visualize the temperature map for the space.

Prior to doing any of that, I would agree with GregLocock and corus to take a look at the initial basics of what your data is telling you over time. Does one point come into spec before others? Is response fast enough? Are the room sensors placed in a location that makes sense for controlling room temp?

Regards,

 

[IRstuff]

Are you sure that your A/C is sized sufficiently?
Is the air actually getting to where it's supposed to?
Is the control system even working?
Is the temperature at the control sensor even correct?

Sounds like you have design or implementation problem, not a statistics problem.


TTFN

FAQ731-376

 

[dhyperman]

its for a train really and not a room. the duct work etc has been placed. The ac unit is supposed to reach 18 degrees all over the trains within 30mins from a start either in cooling or heating cycle. the majority of temp sensors record the temperature being over the allowed temp. 18 degrees is the set point or target the tolerance is of =or- 1 degree celcius. the train has done a trial run and people complaine thats too cold. But the operating temp where the train is maintained is 18-23 which is a narrow band in my opinion. there are two ac units to service the carriage. the unit when they reach temperature cycle up and down but not 100 percent between the band of 18-23, and definitly not at 18+/-1 . I want to run a test to see first off to get an even temp through out the carriage. then run it so the we can compare what the sensors are reading with the ones controling the ac system.
What the data set I have seems to tell me that the process parameters need to be moved or the tolerance adjusted. But thats a a client issue.Anyhow

 

[GregLocock]

+/- 2.5 is way outside of acceptability for comfort, in my limited experience

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[IRstuff]

18ºC is 64ºF, which is indeed COLD. +/-2.5ºC is nearly +/- 5ºF, so that's a pretty large excursion.

Large excursions are usually a sign that the control loop is incorrectly tuned for the application. Your EE should be looking at adjusting his control parameters.

TTFN

FAQ731-376

 

[SWComposites]

You should run the test both with and without people in the carriage, as it could make a big difference.

 

[dhyperman]

I am inclined to agre with IRSTuff, asif you look within say a period of 20 seconds the the sensors record a number of rises and falls in 5 second intervals. Thats looking at the local picture, you look at the bigger picture then its starting out rises above the zone and then drops below the zone, climbs back into the zone for about twenty minutes then drops out for 12minutes and starts to climb outside again. The test results then stop.
I dont have much experience in this field but I would think the test should be run for a period of say 3 hours at least in something that is used all day long.
Is there something I should be doing with the sensors? The time period of dedection is every five seconds and the temperature goes up and down alot in the five seconds any where from 0.2 degrees to 0.7 degrees.
Lets just say it ocillates in the local picture and the overall picture also.
Any good books or articles I should read that may help me make sense of this. I am more into stress and strain stuff but due to having done some work in HVAC b4 I was landed inn this area... but my hvac was in buildings and just envolved the installation of consultants .
Its driving me mad?