badleys traptester geotechlong
TRANSCRIPT
Badleys / TrapTester
Badley Geoscience Ltd, Lincolnshire, UK
www.badleys.co.uk
Application of structural geological methods to E&P problems
Consultancy: Global consultancy presence in all major oil and gas producing areas Expertise in all aspects of structural geology, seismic interpretation, structural basin analysis, fault/fault seal problems, fracture analysis/prediction, 3D model building, impact of structure on reservoir models
Training: In-house and public courses Practical Structural Geology Fault Seal Analysis Software training
Software: TrapTester (and related products) Stretch, FlexDecomp 3rd Party Add-ons e.g. FSA for LGC
TT software clients
• Anadarko • Apache• BP• British Gas• BGIndia • Caltex • ChevronTexaco• ConocoPhillips • CNSPC (Khartoum)• CPC (Taiwan)• ENI/Agip• Encana• GNPC• GUPCO• Norsk Hydro• NIOC
• Dagang GRI• BGP• CNODC• CNOOC• CCSPC• JiangSu• SinoPec
• ONGC• Origin• PDVSA• PEMEX• Petrodar• Pertamina• PetroBras • Petrocanada• QatarPetroleum• Rashpetco• Reliance• Statoil• Shell (incl. Sarawak)• Spirit (incl. Balikpapan)• Woodside• WNPOC
Technical and Commercial support centres
Commercial support centres
UK - Headquarters
Abu Dhabi
New Delhi
Lagos
Beijing
Jakarta
Denver
Villa Hermosa
TrapTester
Interp QC
Structural Analysis
Fault Seal Analysis
3D Stress Analysis
ED Modeling
Framework Builder
Property Model & Viz
2D/3D Seismic Interp
OpenWorks GeoFrame Other ...
Well Interp & Processing
Transmissibility Mapping
Data
• fault interpretation – point sets, sticks, tsurfs
• horizon interpretation – point sets, grids, lines, tsurfs
• interpretation can be sourced from OW (binary), GeoFrame (binary), GoCad or other ascii routes
• interpreted can be edited or generated from scratch in TrapTester
• wells from OW, GeoFrame, ascii
• seismic from OW, GeoFrame, SEGY
• cornerpoint grids ECL (SGrids ... soon)
Different tasks have different data requirements, TT catersfor:
Getting data
Seismic and interpretation Wells, picks and log curves
*
Binary access to wells and seismic from OpenWorks and GeoFrame
TT Stores and accesses multiple concurrent 2D and 3D surveys as (a) direct access, (b) cached data or (c) in BGL format
Simple key concepts to applyduring exploration and prospect generation
Displacement continuity along faults
Displacement conservation at fault linkages
Visualize interpretation in 3D
Effects of structural juxtaposition
Prediction of fault rock properties and seal potential
Displacement contours (lines of equal displacement) are approximately elliptical and decrease in regular fashion from a maxima close to the centre of the fault
Key point is that the abrupt irregularities in the horizon polygon geometry or throw pattern represent areas in the data that should be checked
Idealised displacement distribution on an isolated normal fault
Perspective view of an isolated fault surface. (a) The fault is shaded to show the surface topography and the horizon separations are shown as dark polygons on the fault surface. The separations increase smoothly towards the centre of the fault. (b) Contoured throw on the fault surface. The throw increases systematically from low values (blue) to high values (red). The fault continues below the deepest interpreted seismic horizon.
From: Needham et al. 1996
Using displacement patterns to check fault correction
Map of initial fault interpretation showing fault cuts correlated as a single fault (highlighted in blue). (b) The pattern of throw on the fault distinguished by banded contour pattern. Two maxima are developed suggesting that it is really two separate faults. (c) Fault recorrelated to show two separate structures. (From: Needham et al. 1996)
Use fault throw to QC fault correlation in exploration acreage
1500 ms
300
600
0
900
1200
At the reservoir level, decreases eastwards from a high of about 1500ms TWT at the western end of the fault. A significant decrease in throw (from ca 650 to 200ms TWT) is coincident with the eastern boundary of the prospect (arrowed). Fault viewed looking towards south
East West
Simple key concepts to applyduring exploration and prospect generation
Displacement continuity along faults
Displacement conservation at fault linkages
Visualize interpretation in 3D
Effects of structural juxtaposition
Prediction of fault rock properties and seal potential
Horizon data
2D & 3D seismicsections, time slices
Fault segments
Fault picked onsections, time-slicesand horizons
Visualize, interpret multiple 2D, 3D surveys for fault QC and prospect generation
Client example: Niger Delta
• 70 Interpreters
• 10 3D Cubes
• >25 2D Surveys
• 6400 Fault planes
• 5000 Wells
• Huge problem correlating faults and horizons between surveys
• ALL data loaded into a single TT project• Correlation and QC undertaken using 3D viz in TT• TT now adopted as preferred tool for 2D structural interpretation
Client Recommendation
Simple key concepts to applyduring exploration and prospect generation
Displacement continuity along faults
Displacement conservation at fault linkages
Visualize interpretation in 3D
Effects of structural juxtaposition
Prediction of fault rock properties and seal potential
Fault traps and side sealgeometry or property?
Juxtaposition seal only
Mainly juxtaposition seal; minor contribution from gouge between sands
between sands Seal dominated by gouge
cataclastic def’m bands
shale smear
breccia
gouge
Fault zone properties I
Reduction in grain size by fracturing. Reduction in porosity. Localized mechanical mixing of grain fragments.
Fracturing of lithological units. Mechanical mixing of fragmented lthologies.
?
Predicting fault-zone composition II Shale Gouge Ratio (or SGR)
Outcrop data (metre scale)
Well core data (cm scale)
Slipped interval (T)
Throw, T
Vsh5, z5
Vsh4, z4
Vsh3, z3
Vsh2, z2
Vsh1, z1
Sand
ShaleSGR=(Vsh.z) / T
Photo: G Skerlec
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
O bserved proportion of sha ley gouge in fau lt zone
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Ca
lcu
late
d S
hale
Go
uge
Ra
tio
Expected re lationship
Regression R 2=0.71
Shale Gouge Ratio validation: outcrop data
Outcrops provide ‘ground truth’ for assessing the SGR algorithm.
Fault sampling should be at the appropriate scale.
There is a general correlation between observed shale content of the Moab fault zone and the calculated SGR.
235m
155m
275m
65m
0m
K c mJ n
Transects across the Moab FaultS.E. Utah, Foxford et al 1998
65m
0m
155m
235m
275m
Triangle Plots
Throw = 20m
Increasing throw
Sand C
Sand D
Upthrown
Sand B
Sand B
Sand C
Sand D
Sand B
Sand C
Sand D
Sand B
( c )
Sand A
( b )
Throw = 0
Throw = 10m
Sand B
( a )
Sand A
Triangle
• Identify principal fault seal risks and fault seal opportunities quickly using an industry standard technique.
• Improve exploration team productivity and efficiency.
curve data directly fromOpenWorks or
GeoFrame
Displacement continuity along faults
Displacement conservation at fault linkages
Visualize interpretation in 3D
Effects of structural juxtaposition
Prediction of fault rock properties and seal potential
Juxtaposition of reservoirs
Detailed interpretation and model building:Development/Appraisal/Production
Aim is to produce a “water tight” horizon and fault framework model
Pass the model into a geocellular package
Detailed validation of critical structure
Assess the seal potential
Predict HC column heights
Assess relative impact of certain faults on flow BEFORE building the geocellular model
...
...
Simple key concepts to applyduring exploration and prospect
generation
• Displacement continuity along faults• Displacement conservation at fault linkages• Effects of structural juxtaposition • Prediction of fault rock properties and seal potential
Raw interpretation
Water-tight model
Calculated fault properties
Goal of framework building
TT Provides:
• Advanced tools for editing fault and horizon surface data
• Advanced, semi automated tools for generating and editing the framework topology
• QC tools
• Advanced interpretation aids (e.g. fault slicing)
• Tools for building infill stratigraphy
• Fast, effective, modelling procedures (does not take years to become expert!)
TT uses an automated modelling procedure that is determined by the fault network
(Incorrectly) unconnected fault planes lead to a sloppy and ungeological framework model
Initial model build
Faults need to be linked where appropriate. The process of branchline completion between the splay and master fault is automated but tools exist to optimize the interpretation by hand.
Automated splay intersection lines can be adjusted according to seismic when displayed on the fault surfaces. This improves the accuracy of the fault model.
Here the seismic slice from the down thrown side of the fault is interpreted as though it were a row or column.
Completed interpretation of the hanging wall splay fault
Fault-attribute mapping to quality check interpretation
Problem: Discrepancy in fault polygon due to anomaly in horizon interpretation that may be due to mis-picks, absence of other faults, etc
Solution: Interpret new structure or edit horizons on sections or in 3D or edit fault polygons directly on fault surface or edit the modelling patches
Local framework completion. Applies to all layers and all faults in the project.
• Identify trap-bounding faults• Assign risk (leaking or sealing)• Estimate potential column heights • Better understanding of fault zone properties • More cost-effective reservoir management• Increased recovery (= dollar savings)
Assessment of fault Seal potentialSome goals
Some Observations ...
• Paucity of fault-seal studies by oil companies
Less to do with available technology but more with the reluctance toincorporate routine fault-seal analysis techniques into primaryworkflows
• Faults are typically regarded as a special problem
It is not sufficient to treat the faults as a ‘special problem’ but ratheras part of an integrated container framework of faults and horizonswhose ‘seal capacity’ varies over the surface of the container
Observations ...
Behaviour of faults is time-dependent
In exploration we are interested in the capillary entry pressureof the fault zone and its ability to support an economic columnheight (static trapping)
In production we are more interested in the permeability; a faultthat admits flow over geological time may become a barrier overproduction time scales
Fault Seal – Towards Allan Diagrams
Allan diagram – shows areas of juxtaposition seal and areas of potential cross-fault leakage
Downthrown Red juxtaposed against Upthrown Yellow
Yellow reservoir zones self-
juxtaposed across fault
Areas of reservoir non-overlap = juxtaposition seal at fault.Grey = non-reservoir sealing lithology on both sides of fault
plane.
up
down
Downthrown Yellow juxtaposed against Upthrown Green
Detail of the pattern of juxtaposition
The most import contacts here, i.e. good reservoir to good reservoir, are coloured dark blue. In the absence of any knowledge of the fault rock properties, these are the locations where we would expect the fault to leak.
Volume properties – three routes
SGR=(Vsh.z) / T
Stratigraphic Infill
Constant:Constant distance below or above a primary horizon
Scaled value:Fraction of the interval between two primary horizons
Absolute Depth:Horizontal at depth value
Constant above II
I
II
III
IV
I
II
III
IV
Absolute depth
Constant below I
Scaled below III
Scaled above IV
0.25
0.3
0.3
0.25
I – IV = primary seismic horizonsDashed = markers created in Well Editor
3500
Well 2
B
A
2500
20003000
Stratigraphy is defined in wells and distributed by thickness rules
Example showing a detail of an infilled 3D model
VShale attribute within the infill layers
Vsh
VShale direct from wells using CurveMapper
0
-1 +1
RAI scale; IESX
TrapTester seismic scale
1
V-shale
Low acoustic impedance, low V-shale, probably a gas-bearing sand
Use of an RAI cube and Slicer to volume properties
High acoustic impedance indicateshigh V-shale
Allan Diagrams- Extra note
Allan Diagrams so far have been produced on a layer by layer basis.
New ways of generating volume propertites lead to new ways of considering Allan diagrams
The geometry of the juxtaposition concept can be generalized to apply to layers, seismic voxels, curve mapped properties or cell-cell connections
Map of SGR on a fault surface. Red = high SGR, Yellow = intermediate SGR and Green = low SGR.
Badleys has now published the calibrations from many fault seal studies that show that “green” indicates a high probability of fault leakage and “red” indicates a high probability of seal over geological time scales.
Using properties to predict fault zone composition
SGR=(Vsh.z) / T
Predicting seal / leaking behaviour
Fault-seal attributes (e.g. SGR, CSP) are estimates of the relative likelihood of clay gouge or smear being developed at the fault surface.
To use the attributes as estimates of seal capacity, the attributes must be calibrated in datasets where the sealing behaviour is documented from well data. The objective is to derive an empirical relationship that can be used to estimate the ‘strength’ of the fault seal
0
10
20
30
40
50
60
70
80
90
100
15-20% SGR
Seal
LeakRange of SGR values at sand-sand juxtapositions
Oil fields in North Sea
Sha
le G
ouge
Rat
io (
%)
Shale Gouge Ratio (SGR) at sand-on-sand reservoir juxtaposition. Green (low SGR) are potential leak points, Red (high SGR) are sealing
Relationship between SGR and pressure data
Lines represent maximum across-fault pressure that can be supported at a specific gouge ratio value (seal failure envelopes).
Plotting all data onto one diagram permits a general trend of increasing SGR value supporting increasing across-fault pressure difference (AFPD) to be established.
Empirical equation defining seal-failure envelopes:
AFPD (bars) = 10 ((SGR/27) – C)
C is 0.5 for burial depths < ca. 10,000 ft
C is 0.25 for burial depths 10,000-11,500 ft
C is 0 for burial depths > ca. 11,500 ftAcross-fault pressure difference (AFPD) is taken to be equivalent to fault-zone threshold pressure
Data collected from a variety of basins worldwide (inc. North Sea, mid-Norway, Grand Banks, Gulf of Mexico, Columbus Basin, Niger Delta, Vietnam, Gulf of Thailand).
All basins are typically mixed clastic environments.
Faults dominated by extensional faulting. No strike-slip or reverse fault data.
Fault-gouge samples: composition control on capillary entry pressure
Using calibrated Shale Gouge Ratio to estimate hydrocarbon column heights
Workflow:
• Derive the fault-rock distribution (e.g. from SGR) from subsurface data.
• Convert SGR to fault-zone capillary threshold pressure. Apply seal-failure envelopes derived from calibration with in-situ pressure data or from lab-derived empirical equations to estimate threshold pressure Pc.
• Incorporate density data for water, oil or gas phases at reservoir conditions to predict column height using the equation:
w = pore water density (kg/m3); h = hydrocarbon density (kg/m3)
g = acceleration due to gravity (9.81 ms-2 or approx. 10ms-2)
Pc = threshold pressure in Pascals (105 Pa = 1 bar)
In a simplistic approach, traps are often assumed to be filled down to the shallowest structural spill point. The fault is considered to be sealing (and able to support the column) over the entire fault surface.
Leakage of hydrocarbons through a [membrane] fault seal takes place when the buoyancy pressure exceeds the pressure required for hydrocarbons to enter and pass through the largest interconnected pore throat in the seal (capillary entry pressure).
Establishing where the buoyancy pressure equals the fault-zone entry pressure provides a method for predicting the column height supported by an SGR value.
Predicting hydrocarbon column heights
Buoyancy Pressure
Depth
Pressure difference between hydrocarbons and water
Pressure
B
Water pressure trend
A
Depth
Hydrocarbon pressure trend
Buoyancy pressure / depth profile:Pore pressure / depth profile:
2-D fault section
What is TrapTester predicting?
Buoyancy pressure = capillary threshold pressure
Depth
Fault-zone composition
Depth
Low SGR High SGR
Buoyancy pressure(HC pressure – water pressure)
Capillary threshold pressure (function of SGR)
Low
Depth
High
Predicted leak
point on fault
Maximum column supported by this
SGR value
• TrapTester predicts the column height that is supported by an SGR value
B o u y a n c y P r e s s u r eP r e s s u r e
D e p t h D e p t h S u p p o r t a b l e p r e s s u r ef r o m S G R v a l u e s
Footwall (Oil)
Hangingwall (Water)
= Maximum column supported by the fault
Estimating column heights: 3D case
The critical part of the reservoir overlap is the point which exhibits the shallowest base of hydrocarbon column (red arrow)
This point defines a potential spill point on the fault surface
SGR-derived spill point may be shallower then structural spill(fill-to-spill) or contacts derived from pore pressure gradients
= Shallowest column derived from SGR
Juxtaposed reservoir sands
How can TrapTester predictions help us?
• TrapTester predictions provide an alternative method for estimating column heights (ie, why some traps are under-filled)
OWC based on structure filled down to spill point at fault tip
Purple: OWC; black: depth contours
Fault-plane diagram:
Solid = footwall sand; Dashed = hangingwall sandYellow = sand-on-sand
Alternative OWC based on shortest column predicted from SGR
Column heights predicted from SGR
Red line = column support by SGR
Framework model vs. cellular modela way to optimize cellular geometries?
(EarthGrid)
Simplified fault geometrySimplified fault connectivitySimplified horizon geometry
Integrated modelGullfaks Public Data Release, courtesy of the licencees of PL050/PL050B and the Norwegian
Petroleum Directorate.
Framework model of faults and surfaces based on seismic
interpretation
Well
Horizon
Structural differences
End of geocellular fault before true end of fault.
Fault clay content too low in geocellular fault representation
Geocellular Fault
Framework Fault
0SGR (%)
100
•The geocellular fault is too transmissive towards its tip, and is too short in the model.
•The geocellular fault throw is forced to be 0 at its lateral boundary, where the true displacement is greater.
End of model
N
• The number of faults included in the model usually depends on criteria imposed by limitations imposed by the model building process (Y-faults, fault size, intersection geometry etc.).
• Shouldn’t we decide on what faults to include based on their likely effect on fluid communication between reservoir layers…
• …and their effect on fluid flow?
Structural omissions
Horizon dip azimuth
Calculation of SGR – towards transmissibility(TXMmapper)
Vsh4, z4Vsh3, z3Vsh2, z2Vsh1, z1
SGR = (Vsh.z) / t x100%
Throw
0
1Vshale
0
100SGR (%)
Vshale
Reservoir simulators usually incorporate fault properties implicitly as transmissibility multipliers - the ratio by which the slab of fault-zone material degrades the transmissibility between juxtaposed cells.
The multiplier depends on the size and permeability of the juxtaposed cells as well as the thickness and permeability of the fault zone.
The transmissibility multiplier is model-dependent
Fault properties in reservoir simulation: transmissibility multipliers
k1 k2
L
t, kfz
A
t = Fault-zone thickness
Kfz = Fault-zone permeability
K1 K2 = Cell permeability
L = Distance between cell centers
A = Area of connection between cells
TM =
Fault-gouge samples: control on permeability
General decrease in permeability with increasing phyllosilicates in gouge
• Clay-smear samples show very low permeability.
• Gouges generated from clean sands have very variable properties that depend on their geological history (depth at time of faulting, maximum burial depth – greater depths give lower permeabilities).
Comparison of permeability measurements on core samples (symbols) and predicted permeability (solid lines).
Zf = Initial burial depth (during faulting)Zmax = Maximum burial depthVf = Clay fraction of fault rock
From Sperrevik et al 2002
Fault-gouge samples: control on permeability
kf = 80000.exp-[19.4Vf + 0.00403zmax + (0.0055zf - 12.5)(1 - Vf)7]
Pf = 31.84.kf-0.3848 where Pf = Hg/air threshold pressure
Examples of transmissibility multiplier calculations
• Gullfaks reservoir model contains >50 faults, 25 zones, 38 rows and 87 columns.
• Our examples come from the Northern part of the model.
Gullfaks reservoir model
An example: the “big fault”
• Big fault tips out to the south (left)
• Towards the centre of the fault the maximum throw corresponds to the juxtaposition of Tarbert Sands against Etive and Rannoch sands.
• Shale Gouge Ratio in this area is higher because of the intervening more shaley Ness Fm.
• The fault should be more transmissive towards the tip where self juxtaposition of of the Tarbert Etive and Rannoch Fms occurs.
0
1Vshale
0
100SGR (%)
0
150
ThrowThrow (m)
Hangingwall
Footwall
SGR
Vshale
NS
Transmissibility
0.0002
0.02
Transmissibility(mDm)
0
1
TMX Transmissibility Multiplier
Faulted
Unfaulted
Transmissibility
•Unfaulted transmissibility is the flow potential across a fault that has no fault rock (such as clay smear, cataclasis or diagenetic alteration).
•Faulted transmissibility implicitly incorporates permeability, using SGR, and a thickness of fault rock products into the calculation of potential flow.
•Transmissibility multiplier (TM) is the ratio between unfaulted and faulted transmissibility.
•Low transmissibilty multipliers occur where the fault impedes flow.
NS
Recent examples of reservoir simulation using geologically-driven transmissibility multipliers:
• Heidrun Field
Fault-zone permeability derived from core analysis & applied to juxtaposition diagrams. Accurate prediction of water breakthrough.
• Snorre Field
Transmissibility multipliers derived from SGR analysis. Excellent history match achieved, compared to using non-geological ‘default’ multipliers.
• Scott Field
Transmissibility multipliers derived from SGR analysis. Excellent history match achievable after 1 day instead of 3 months.
Transmissibility Multipliers: Examples
PRODUCTION DATA
ALL FAULTS CLOSED
SELF JUXTAPOSED OPEN
MODIFIED OPEN (3 months)
SGR METHOD (<1 day)
Time
Simulation Results: Water Production vs.
Time1994 1995 1997 1998 19991996
Cumulative WaterProduction (STB*106)
0
4.0
8.0
12.0
20.0
24.0 Scott Field - Block Ib
ALL FAULTS CLOSED (all Tm = 0)
SELF JUXTAPOSED OPEN (Tm = 1; non-juxtaposed Tm = 0)
MODIFIED OPEN (manual input to modify cell-by-cell multipliers)
TMX key benefits
Fault reactivation risk
likelyhood of faults being active under present day stress conditions
Main geomechanical relationships in TrapTester
Ratio of shear to normal stress
Risk of slip increases as the ratio approaches the coeff. friction (~0.6)
Slip Tendency Fracture Stability
P(P)
Pore pressure increase required to induce failure
Assumes fault rock has mechanical strength
Fracture stability Slip tendency
Fault-plane diagrams: geomechanical attributes
Green: small increase in pore pressure required to induce failure (~ high risk of reactivation)
Blue: large increase in pore pressure (~ low risk of reactivation)
Red: high slip tendencyYellow: low slip tendency
Regions on the fault with low fracture stability (green) coincide with high slip tendency (red)
Fracture prediction - FaultED
Introduction to Elastic Dislocation methodology
A ‘fault panel’ is a rectangular dislocation with uniform slip, embedded in an elastic medium. Using the equations of Okada (1992), the resulting displacement and strain tensor can be computed at any observation point in the medium.
The corresponding stress tensor and failure mode (if any) at the observation point can then be computed using appropriate material properties.
Mapped faults in the subsurface can be approximated by an array of rectangular fault panels, each of uniform slip.
Define ‘observation grid’ upon which strains, displacements & stresses are calculated.
FaultED modeling: Workflow
3: Include regional strain estimates. Forward model deformation to match
observed horizon geometry
1: Build faulted framework model
4: View model properties
2: Panel faults (rectangular fault panels)
1. Thrust fault with displacement pattern (blue: 0, red:700m slip).
2. Observation grid set up around thrust fault (node spacing 200m).
3. ED forward model deforms the observation grid to mimic the hangingwall anticline.
Example: Thrust anticline
Comparison of example reservoir horizon and modelled observation grid
modelhorizon
Model does not include regional tilting, effects of other faults, or local diapirism
Elastic strain tensor
Axes of strain tensor
A ‘pseudo-stress’ tensor is calculated from the strain tensor (incorporating material properties)
Volumetric strain
FaultED modelling workflow: Predictions
Maximum Coulomb Shear Stress (MCSS)
Most-favourable fracture orientations
Stress tensor is used to predict the mode and orientation of likely failure planes at all nodes on the grid
Normal Strike slip
Reverse
small faults mapped on seismic
(Sand 2)
Small faults concentrated in
high-MCSS region at east end of main
thrust
Maximum Coulomb Shear Stress (proximity-to-failure indicator),
map view
main thrust fault
normal faults
reverse faults
strike-slip faults
main thrust fault
How do observed and predicted fracture orientations
compare in detail?…..
Predicted fracture orientations, map view
reverse faults on back limb
tear fault at E end of fold
0o
(perfect) 90o (bad)
angular misfit
Mapped fault-trace azimuth (o)
ED Predicted fault-trace azimuth (o)
Mean angular misfit = 23.6o.Cf 45o for random set,
and 62o for faults // main thrust
Coloured squares show local angular misfit between mapped and predicted fault strikes
Simple key concepts to applyat all stages
Displacement continuity along faults
Displacement conservation at fault linkages
Visualize interpretation in 3D
Effects of structural juxtaposition
Prediction of fault rock properties and seal potential
Leverage all the above to build defensible 3D Models
To risk effectively seal / leak behaviour
To assess risk of reactivation
To predict fracture densities and orientations
To build TXM maps for fault-flow behaviour during production