gas reservoir with bottom aquifer-help needed

[tinni]

i am trying to match history of a gas reservoir with bottom aquifer. the reservoir has 11 producing wells and i have the production history of last 40 years. i am new with eclipse. firstly i have one question about aquifer connections. i have defined a numerical aquifer. can i connect the aquifer under the entire reservoir with single connection data? i saw a data file where someone has given connection to a bottom aquifer like below:
li ui lj uj lk uk face
6 54 32 128 36 41 I+

can an anticlinal shaped volume be used as aquifer connection? shouldnt the lower and upper k value be the same for a bottom aquifer?

secondly my results show a huge water production at the end of historical years. also water production rates are fluctuating very much.in one month it is 200 bbl then next month it is 60000 bbl. my kv/kh is .1. i am asking for some advice. its driving me crazy.

[karakurt2]

yes, you can. to do so you should define connection like this one

AQUCT
1 688.5 -1 811.7 0.36 5.5e-5 38952 400 /

AQUANCON
1 1 143 1 472 1 73 'I+' /
1 1 143 1 472 1 73 'I-' /
1 1 143 1 472 1 73 'J+' /
1 1 143 1 472 1 73 'J-' /
1 1 143 1 472 1 73 'K+' /
/

[karakurt2]

to define geometry of the aquifer you may use ACTNUM reservoir property.

[darcyoil]

11 wells? 40 years history? Why on earth would you waste your time with Eclipse? Simple p/z or RTA would give you aquifer influx numbers, with which you would be able to cross-check your coning correlations. In addition, if you have water data, then it becomes even easier. At the end of the day, all you're really interested in is figuring out how much you can produce before the water comes up. So, if yo're to outrun your aquifer, assuming all wells in same sand, tweak the rates updip so that your structurally lower wells will give more bang for the buck while acting as early-warning nodes.

[tinni]

11 wells? 40 years history? Why on earth would you waste your time with Eclipse? Simple p/z or RTA would give you aquifer influx numbers, with which you would be able to cross-check your coning correlations. In addition, if you have water data, then it becomes even easier. At the end of the day, all you're really interested in is figuring out how much you can produce before the water comes up. So, if yo're to outrun your aquifer, assuming all wells in same sand, tweak the rates updip so that your structurally lower wells will give more bang for the buck while acting as early-warning nodes.

actually its my project to history match with eclipse.so i have to do it with eclipse. my pressure almost matched for early years.but it falls down at the end( last 5 years). also profile is very fluctuating and at the same time i am getting huge water production with fluctuating rates. so i dont know what parameters i should change to correct the problem. i am planning to reducue my Krw as i dont have any exact SCAL data. dont know whether it will solve my problem