I have a series of 6 pumps that can be pumping or not in any order at anytime of the day or night. The pumps are used to pump water in to a reservoir. I have numerous PI tags at my disposal the main ones relative to this case are

Pump electrical breaker (digital)
Pump output isolation valve (digital)
Pump power (analogue)
Reservoir level (analogue)

When the reservoir gets low then the pumps will be started and top it up. The pump breaker can be closed without the pump actually pumping.
My project is to make a spread sheet that will be run once a week and will record the start and end times of any pumping (thus it doesn't matter if it is one or all pumps running) along with the reservoir level for each start and stop. In almost all cases there will be one pumping session per day (overnight) but in some rare cases there may be a short pumping session during the day. At the moment this is done manually and though I have knocked up some long winded auto sheets they wade though all the date in excel and try and spot the times etc but this is very slow and cumbersome.
Ok my first plan is to find the minimum reservoir level between two time and date inputs. I can then sequentially go through the reservoir data in minute steps and find when the level increases and this would then be my pump start time. Similarly the Max value will be the pump end time.
Problem is this will do for almost all cases but won’t cover the sneaky day time pump runs where the level may not get that low or the software will pick up the pump start during the day but not see the pump end until the next pump run and give completely wrong results. So my second thought is I've got to condition the data somehow. I could look at pump power over all 6 pumps per 24 hours and try and spot start and stop pairs but this would be a fair bit of data crunching over the week.
Would anyone have any suggestions?
Thanks.

OSIsoft vCampus is a subscription-based, online offering that consists of providing everything people need to develop applications on the PI System.
We invite you to take a "tour" of the OSIsoft Virtual Campus - also feel free to consult the FAQ  or contact OSIsoft vCampus for more details.

Thanks for the reply.
You are quite correct about a baseline load signifying pump or not but I'm afraid the rest of your reply went several inches (I hope not yards) over my head. I had thought the best plan would be to get the PI server to do as much as possible as the delay would probably be much greater in transferring data then crunching it in excel. I can query PI and get the power per minute and even filter it to only give me the data where the power is over base line but I'm not sure how to do this for all six at once or even how standard deviation will give me a start time. I'm trying to get the company I work for to stump up the dosh to get me on a course. They sent me on one maybe 8 or 9 years ago but it was only the basic use of process book. Do you do courses by chance? I'm not that far down the coast from you. I can see your point about midnight to midnight period and do a simple check to see if the start time of the after midnight pump is the same as the end of the before midnight pump to see if it's continuous. Thanks for your help you've given me a few things to think about. I find that finding stuff on the net about PI is pretty difficult if not I'd go off and teach myself these performance equations.

Perfo...sure we do courses, it would seem a tailored course would suit to get to the points that you need. We can talk about that when you are ready...

Let me sit down and have a think about the rest of your comments.