Transition from History Matching to Prediction

[besucher]

Hey all,

I am working on a simulation case in ECLIPSE and after I matched the History, I am going now for the prediction part... To get a reliable prediction I want to have a smooth transition between my HM and my Prediction in the well rates...
Unfortunately this is not the case... When my prediction starts, it makes a jump upwards...
I used WPIMULT (PI Multiplier for that well) to get rid of this jump and to smoothen my rate between HM and Prediction out... Actually I thought, when I decrease my WPIMULT the rate should also decrease, but the opposite is the case (Is it normal !? It seems strange to me)
So: WPIMULT higher than 1.0 decreases my rate and goes in the right direction...

But even if i multiply it with high values, the sudden jump in the rate won't completely disappear.

Why do I have that jump in the rate ? How can i get rid of it ? Other Keywords I can use?!

Thanks for any ideas in advance!

[vinomarky]

You are on the right track - the key being to get the well PI's correct. Plot your WBHP's through history match to forecast - I'll bet that in history, your WBHP's were substantially higher than in prediction.

As to why decreasing the WPIMULT increases your rate - that does indeed seem strange on face value, we'd probably need to know more of the specifics to give better advice.

How are you controlling rates in forecast? By BHP? If so, (assuming that the WBHP you are using in forecast is appropriate) use this same WBHP value as a last history match point for your wells

[besucher]

hi vinmoarky,

thanks for the reply!

First of all: I found out, that I did a mistake in my prediction, thats why I got that jump in the rate. I solved it so far, but this also leads me to the next question:

I don't have pressure data (just initial pressure), so I didnt match the pressure data... Thats why I am not sure how will my pressure be in the future. I created so far 3 scenarios with different Aquifer, leading to different pressure in the here&now and also for the prediction. Thats also why my control mode for the forecast is the liquidrate (like during the HM) with the last L-Rate of the HM... I just gave the bubble point as a limit in the keyword WCONPROD... Was this approach fine so far or should I leave the pressure completely out of my prediction !? I am not sure about this point, because I get two different forecasts of the future production if I use BHP limit or not, and I dont know which one is the right/better approach...

[vinomarky]

You should be able to (gu)estimate reasonable flowing BHP's. If you have recorded THP's (on natural flow) you can back calculate. If you are pumping, what are your pump parameters? For example you should know roughly how much (in case of ESP/PCP etc) fluid above the pump the operators aim to keep, and from that infer what the operating FHP's generally are around.... There might even be records of sonalogs the operators run... generally you can find SOME sort of information - however fuzzy - from which you can infer BHP's..... as a rule of thumb, your operators are generally pretty good at keeping pumps close to optimum, so if you history match shows that you are only needing draw down the wells a little, then there is probably something wrong with your PI and/or local perms

[besucher]

I didn't understand the last part:

"so if you history match shows that you are only needing draw down the wells a little, then there is probably something wrong with your PI and/or local perms"

Could you please explain it further or differently ?! :> thx

[vinomarky]

Let's say your 'ideal' producing scenario is to have 200ft of fluid above the pump - assuming it is water, this would equate to ~100psi of BHFP. If your reservoir pressure is ~2000psi, this equates to a drawdown of 1900psi available. If on the other hand your 'matched' model only needs 500 psi to achieve the observed rates, then either you have been operating the wells sub-optimally or your PI's are too high

[itag]

FOlks - the advice to watch for smooth transition of WBHP as you move from HM to forecast is good advice. I just want to suggest 2 points:
a) plotting WBP9 ( or WBP4 or just WBP) and WBHP during the HM & forecast provides visual information in regards PI. ( These are Eclipse terms)
b) If the wells are naturally flowing - using flow tables for producers may be one option for forecasting ( the PI calcibration recommendation still apply). This option allows you to better model wells dying at high watercut - even with lift. Cosntant WBHP may not be usueful as a constraint under these conditions. ( If the wells are "pumped off" this advice may not apply.

itag