Eclipse step by step

[acang_024]

Wconprod
'well-1' 'open' 'wgra' 5* 300 200 1 1* 10000 5* /

[acang_024]

Wconprod
'well-1' 'open' 'wgra' 5* 300 200 1 1* 10000 5* /
/

[vinomarky]

Hmmm... not sure what to say. This is not the 'expected' behaviour

You have specified a wet gas rate target of 10mmscf/day, with THP limit of 200psi and BHP limit of 300 psi

The correct Eclipse behaviour would then be to try to attain a 10 mmscf/day wet gas rate plateau for as long as possible while not falling below EITHER 200 psi THP or 300 psi BHP. If it falls to either one of the pressure limits, it should then switch to control by that parameter which is about to be violated (in this case THP) and your well will start rate decline. Ultimately your gas rate will dwindle to negligible/zero depending on whether your tubing is loading up

The only logical reason I can see for this not to occur would be if subsequently in your SCHEDULE section you have redefined the WCONPROD for that well..... could this be a possibility? It shouldn't affect things, but are you sure your VLP tables are defined over the entire THP range? Eclipse sometimes does funny things when it has to extrapolate.

If not, there may be some issues with calculation convergence at your wells or something similar - contact your local SLB support rep and send them to deck to see if they can figure it out.

As a quick & dirty workaround, take a look at what rate the well is doing as it passes 200 psi THP, and set that rate as a WECON cutoff

[acang_024]

when THP reach to 200 psi, pressure control change to BHP(300 psi), when it change, gas rate exceed 10mmscf/d because draw down higher and THP increase, so the well will not shut in.

[vinomarky]

Thinking on it a bit more... As another workaround, you may want to try using ACTIONW command to reset control to THP control once THP is detected to be less than 202 psi. Your code would look something like;

ACTIONW
A1 'Well-1' WTHP < 202 /

WCONPROD
'well-1' OPEN THP 5* 300 200 1 1* 10000 5* /
/

ENDACTIO


I'm still not sure why you are seeing the behaviour you are.... this approach should not be required

[persian_patriot_1982]

Hi everybody.
I'm so confused. I have a problem in initializing. Here is the case:
1. It is a dual porosity gas condensate reservoir.
2. The reservoir does not have a SCAL, so no relative permeability/Capillary data.
3. There is one exploration well and it has a log.
4. Using the log, a Sw distribution is generated using rms.

Now I want to use this data for initialization. What should I do?
Should I use SWATINIT or Should I use SWL?

Please Please help me.

[vinomarky]

Without seeing your data, some things to check first;

Crossplot your log Sw data (not populated Sw, as this would mean you trust how the geo/software has distributed the property) against log perm in a log-normal plot - See if you can find a reasonable relationship of Sw as a function of Log (Perm). What about if you colour code the points by height above free water level.... What about if you plot Sw vs SQRT(K/PORO)?

What we are looking for is a way to ditch the RMS distributed Sw's and come up with a relationship that Eclipse can calculate at run-time with OPERATE keyword.

In short, I recommend AGAINST using Geo distributed Sw's to initialize your model with

So in summary;
1. Find what relationship best fits observed (log) Sw's as a function of perm, Sqrt(K/Phi) etc
2. Use endpoint scaling
3. Use this function to initialize SWL's using OPERATE function
4. Calc other endpoints appropriately (ie SWU = 1-SWL, SWCR=SWL etc)

If you are in transition zone with mobile water initially, and need to factor in height above FWL, then it gets a bit more complex to do this at runtime, and you may have to back-calculate 'synthetic' J-Func curve in addition...

Note: if you go with a SQRT(K/PORO) type relationship, ENSURE in talking with your geo, that they have properly paired the distribution of poro and Perm through co-krigging etc, otherwise they may have simply distributed the PORO and PERM independently, meaning your nice SQRT(K/PHI) relationship is rubbish away from well control.... In fact you might even want to consider asking them to created a log property ofSQRT(K/Phi) (treat it like a continuous facie value - similar to FZI or flow zone indicator type analysis) and populate THAT property along with PORO, and then calculate appropriate PERM.... just a thought

[persian_patriot_1982]

Dear vinomarky,

Thank you for your great comments. I have a very important conceptual problem.

The water saturation resulting from the log are the initial water saturation. When you use SWL instead of SWATINIT to import them to eclipse, your implicitly tell eclipse to scale the relative permeability curves. For example if in a cell the Sw of the log is 0.8, the SWL of that cell is 0.8. It means the water won't flow until its saturation becomes greater than 0.8. Isn't it unrealistic? Generally this method yields to very little water production (Since you impose immobility in each cell be setting SWL=Sw) which may not
be the case.

Why don't we use SWATINIT which honors the initial water saturation and changes Pc to match it?

[vinomarky]

PP - It's a good question

You'll note in my post I referred to the fact that if you are in transition zone then it gets a bit more complex. This is what I was referring to... You can indeed do as you suggest, that is, using SWATINIT instead of SWL, but then you also have to appropriately pair this with SWCR's in order to prevent masses of initial mobile water when in reality there is often very little if any

Ask yourself what IS the transition zone? And what IS Swir?

If you take cap pressure tests, until you get up to pressures that usually far exceed any reservoir charge, they often (especially for poorer quality rocks) do not fully asymptote. So would this mean you'd expect initial mobile water in all these reservoirs?.... The reality is that we most often have little if any initial mobile water... this being the case I usually choose to adopt the SWL approach for simplicity

If you indeed have a case - usually low relief and/or poorer rock - where there is initial mobile water expected above the FWL, then you would need to go the SWATINIT/SWCR route in order to appropriately represent this. I have done this in some reservoirs, but it is more difficult to get right

FYI - Below is a Petrel macro I created to do just this - It calculates Sw (which I used to define SWATINIT) from a Lambda function, then additionally calculated SWCR etc... Have removed names/fields/formation references etc..

# Created by xxxxx, August 2007
# Macro to calculate actual and endpoint water & oil saturations

# Actual saturations from fitted secondary drainage curves
# Fitted Lambda function Sw = [a x (h^-lambda)] + B
# a = a1 x log(perm) + a2
# lambda = L1 x log(perm) + L2
# B = b1 x log(perm) + b2
# where a1, a2, L1, L2, b1, b2 are constants, h is height above FWL (in m) and k is in mD
#
# For xxxx (xxx & xx) secondary drainage curves, the constants are;
$a1=-7.75017
$a2=34.17743
$l1=-0.00041
$l2=0.002083
$b1=7.5149
$b2=-33.1969


$FWL = -1039

# Calculate height above FWL, set to zero if below FWL to prevent calculation problems
$h = (Depth()-$FWL)*0.3048
$h=if($h<0,0,$h)

# Calculate Sw, error-trap to be within 0.0 to 1.0 range
$a = ($a1*log(Perm))+$a2
$lambda = ($l1*log(Perm))+$l2
$B = ($b1*log(Perm))+$b2
Sw=if($h=0,1,($a*(Pow($h,-$lambda)))+$B)
Sw=if(Sw<0,0,Sw)
Sw=if(Sw>1,1,Sw)

# Irreducible water saturation, Swir from (1 - Soi)
# From regression of perm vs Soi from capillary pressures and NMR
# Swir = 1-((0.0371*LN(perm))+0.3595) (capped to max of 1-Sw)

SWCR = 1-((0.0371*LN(Perm))+0.3595)
SWCR = if(SWCR>Sw,Sw,SWCR)
SWCR =if(SWCR<0,0,SWCR)


# Residual oil saturations from fitting Land constant equation to
# imbibition capillary curves, with constant = 9
# Sor = 1/(9 + (1 / (1 - Swir) ) )

SOWCR= 1/(9 + (1 / (1 - SWCR) ) )
SOWCR=if(SOWCR>0.3,0.3,SOWCR)
SOWCR=if(SOWCR<0,0,SOWCR)

SWCR=if(SOWCR+SWCR>=1,0.999-SOWCR,SWCR)

# Max water saturation = 1 - residual oil, or 1 if below FWL
SWU=if(Depth()<FWL,1,1-SOWCR)

# Scale rel xplot_perm_quartered end-points
# Use Corey curves to determine actual rel xplot_perm_quartered
# Normal perm, Co = 1.93, Cw = 1.15
# High perm, Co = 1.49, Cw = 0.61

# Scale for rel xplot_perm_quartered to oil at Swi
KRO=(0.0344*LN(Perm))+0.5427
KRO=if(KRO>1,1,if(KRO<=0.0001,0.0001,KRO))

# Scale for rel xplot_perm_quartered to water at Sor
KRW=(0.0109*LN(Perm))+0.0357
KRW=if(KRW>1,1,if(KRW<=0.0001,0.0001,KRW))

[paolomaldini]

thanks

[Mehdi0914]

hi every body
I am in an urgent need of a reservoir model,,for my thesis.
Please some one help me,,,,
this model should run by e300,,less than 5000 grids,,,and without any fracture,,,,and I dont want systhetic cartesian models,,,,
other features are not important,..
I am waiting,,,,,
for Gods sake,,
(mahdi_put@yahoo.com)

[vinomarky]

Dont want synthetic grid, yet it has to be compositional and less than 5000 cells... hmmmm.. good luck with that one

People only go to compositional models when PVT is complex enough to warrant, generally this also means their model (grid) is reasonably complex. For even a relatively low vertical resolution of 10 layers, 5000 cells equates to 22 x 22 grid.... not very complex.....

There are a couple examples in the C:\ecl\200x\e300\data directory that you may want to look at - try the ones with the larger DATA files
Conversely, you could simply fire up Petrel and use SGS distribution of properties and create your own grid with required complexity