QC at GEOMODELING

Geo modeling QC

1.GOOD QC

• •Visualize your data
• •Check before moving to next stage in workflow
• •Work in multi disciplinary team – ask!
• •Make 2D maps
• •Look at statistics - averages, SD, min, max
• •Make cross plots
• •Compare against simple deterministic and other calculations

2.WHAT IS MODEL BEING USED FOR?

• •Determine goals for the project first as this will determine level of accuracy and QC required

• •Volumetrics model may require
• high level of QC

3.LOADING DATA

• •Generally few problems
• •View data

1. –Raw logs
2. –Structure maps in 3D
3. –Fault sticks etc

• •Ideally fix data at source (LANDMARK), not in IRAP

4.BUILDING GRID

• •QC grid

1. –look through slices
2. –Run QC job in IRAP
3. –Simulation grid export to Sensor and discuss with RE first

• •Lots of options, so decide what is grid being used for first!

5.BLOCKING WELL DATA

• •First step in upscaling process
• •Look at statistics before and after upscaling, for all wells, and individual wells
• •Can use bias to correct for any sampling bias of facies
• •Block Porosity using facies weighted arithmetic average
• •Block permeability using arithmetic, geometric or harmonic – it depends!
• •Run single well models with RE to check blocking and grid size is working ok

6.FACIES MODELS – (NTG)

• •Look at blocked and raw log facies proportions to determine facies%
• •Use body logs for stacked facies
• •Must consider impact of well sampling and trends being used in model
• •Are models slow to condition to well data?
• •Look at body thickness wrt to well data, and grid size
• •Look at vertical distribution of facies
• •Be careful when combining/merging facies models – are proportions maintained correctly

7.Sample Bias and use of trends?

• •Well data has 66% sand facies
• •What facies proportion should you apply in model?

8.Example Distribution of facies, and stacked facies

• •Is there a vertical trend?
• •Are reservoir bodies stacked?

9.PETROPHYSICS

• •Fraught with dangers!
• •Can apply lots of complex transformations
• •ALWAYS CHECK DISTRIBUTION OF RAW LOG, BLOCK WELL, AND MODEL DATA AND COMPARE, may also compare core data
• •Make some maps = average porosity, net pay
• •Can use pseudo wells to compare against actual wells

10.UPSCALING

• •Fraught with dangers!
• •Set unresolved cells to -999 to view where problems
• •Use layered resampling when possible
• •Look at statistics of fine grid and upscaled
• •Compare volumetrics – should be <2%>
• •Run flow simulation of fine scale sector, and upscaled sector and compare first
• •Use net volume weighted arithmetic average for porosity
• •Use tensor or other methods for permeability
• •Use volume arithmetic average for NTG

11.Final comments

• •Don’t assume because the computer has calculated something it is done what you think – CHECK
• •More complex the model more chances of error
• •Check before going too far – can save a lot of wasted time
• •Visualize, make maps, and look at statistics

By

A. Anriansyah, dkk
(Dari coba2 & beberapa handsout)