7 tipsandtraps upd

OLGA Tips and Traps

Pipeline discretisation

• For each pipeline GEOMETRY we must specify the number of calculation segments (sections) per pipe

• The number of sections is a compromise: accuracy vs computation time

• The number of sections is important when temperatures, pressuresand velocities change rapidly

Pipeline discretisation cont.

• Some terrain effects are lost if grid is too coarse– due to the staggered grid solution scheme

• Example case:– 3000 SCF/BBL GOR multiphase production– very hilly terrain

Hilly Terrain Multiphase Pipeline Profile

2500270029003100

0 6000 13500 19000Distance (m)

Elev

atio

n (m

)

Effect of pipeline discretisation

Pressure drop varying for varying number of segments

320

360

400

0 2 4 6 8 10 12 14 16 18Multiplication factor for # of segments

Tota

l pre

ssur

e dr

op(p

si)

Pipeline discretisationsummary

• Check sensitivity to discretisation– Finer discretisation reduces errors and increase computing

time• Try to use at least 2 segments for any pipe• Try not to violate the following rule:

Li Li+1

0.5 ≤ Li / Li+1 ≤ 2 for all i

Pipeline discretisation and time step

• The time step HT is limited by the CFL criterion for implicit solutions (default = ON)

CFL: HT < min ( Lsection/Ufluid)

• You may also limit HT by the 2nd derivative of pressure (default = OFF)

• Use DTCONTROL to change defaults

• Moreover, HT is always limited byController action and Slugtracking

SHUT-IN simulations

• In some cases OLGA gives un-physical pressure rises in a pipeline that has been shut-in for a while, where the temperature is falling.

• Ways of getting around it:– Reduce the time step and/or reduce the ratio of

MAXDT/ MINDT to limit the changes in time-step.– Re-section the pipeline with somewhat shorter

sections where there are liquid interfaces.• The problem is mainly due to limitations in the

Fluid Table approach and the best solution is to use either Compositional Tracking or the Black Oil model.– Black oil parameters may have to be tuned with PVTsim

to work properly

Shut-in continued

• The main source of the problem is flashing of liquid to gas when the gas-liquid interface crosses section boundaries due to movements caused by the numerics. There is usually no equivalent condensation, and the result is a slight increase in pressure over time.

• Most frequent problem cases:– several gas/liquid interfaces and relatively little

gas (pressure increases quite rapidly when extra gas is created),

– the liquid density decreases with pressure – there are sections filled with liquid

Steady state - an example

FLOWLINE

Elev

atio

n [m

]

0

-1

-2-2.5

Horizontal length [m]

150010005000

Elevation profile 2

PIPE ID = 0.5 m

total # of sections: 20

Steady state example cont.

Total liquid inventory in % of pipe volume

0

5

10

15

20

25

30

35

0 5 10 15 20 25Flowrate (kg/s)

(%)

Steady state preprocessorDynamic steady state

Steady state example cont.

Total liquid inventory in % of pipe volume

0

4

8

12

16

12 14 16Flowrate (kg/s)

(%)

Steady state preprocessorDynamic steady state

General observation

• At lower flowrates and a pipe with dips and humps (gravity dominated flow) an obtained steady state solution may be meaningless since the flow in reality may exhibit continuous fluctuations.

Steady state stratified flow • All steady state stratified flow models solve the same type of

equation by iterating on liquid holdup (OLGA uses time stepping to “iterate” to a dynamic steady state solution)

• The equation has normally only one root, but it may havemultiple roots for:– upward inclination angles– low liquid velocities– low/medium gas velocities

• Stability analysis show that only one root is stable• The existence of multiple roots is now demonstrated by

experiments• OLGA does not always converge to the minimum liquid holdup

solution (maximum gas volume fraction)

Stratified flow - possible steady state solutions

α = gas volume fraction ≤ 1

1.0

0.5

0.0

-0.5

0.6 0.7 0.8 0.9

F(α)

the solution to F (α) =0 is found by iteration, starting at α =1

liquid holdup = 1- α

UsL = 0.001 m/s

ϕ = 0.2 °

Usg = 0.8 m/s

0.75

0.70.65

Stratified flow - conclusion

• Be concerned whether OLGA finds the right liquid inventory for high GOR flow in hilly terrain

• OLGA should find the right liquid inventory after pigging or after a rate change from a high to a lower rate

Initial conditions

WELL

Pressure boundary

reservoir pressure

The steady state pre-processor may have problems finding a solution at t = 0

Use INITIALCONDITIONS and specify

- a high gas volume fraction (void fraction) everywhere

- pressure gradient with correct direction

Network

• OLGA was designed to handle converging networks

Split nodes and closed loops are also possible

Loop line design case Schematic

Gas Plant

Cluster 1

Cluster 2

Cluster 3

Node Node

Pressure boundary

Pressure boundary

Source1 Source 2 Source 3

Loop line design case- single branch OLGA model

P plant

Loop line design case closed loop OLGA model

Source1

Source2 Source3

Terminal Node Pressure boundary

Merging Node

splitting Node

Terminal Node Closed boundary

A typical network

to e.g. flare

SD Valve

Blow Down Valve

Possible OLGA model

Source

BDV back pressureBDV

SDV

Separator pressureNode:

INTERNAL

ClosedNODE

Pressure NODE

Pressure NODE

BDV

SDVQ

Pb

PsFlow direction

Blow down simulation

Normal operation

Source flow = Q

BDV opening = 0

SDV opening = 1

Blow down

Source flow = 0

BDV opening = 1

SDV opening = 0

Bubble / Dew points

• OLGA does NOT check against dew points– In the gas region - outside the 2-phase region, the

predicted gas mass fraction can be different from 1 due to linear interpolation across the dew point line

• Above bubble point – no mass transfer– i.e. gas is not condensed/dissolved

• When actual P,T is close to critical then fluid properties could be dubious

Phase envelope

0

1000

2000

3000

4000

5000

6000

-100 0 100 200 300 400 500 600 700 800 900

Temperature/°F

Crit P

TEST 4 BHS OIL C10+

Table Points Gas mass fraction could be < than 1 due to linear interpolation

Water vapour and free water

• No separate mass equation for water vapour in basic OLGA

• Assumption:– The gas phase is always saturated with water vapour

Water

Liquid water

Water vapour

FlashingOK

No check on available

wat.vapour

Mass transfer of water

• Water vapour mass fraction from fluid table:

• Change in water vapour mass fraction due to change in pressure and temperature:

TT

Rswpp

RswRsw Δ∂∂

+Δ∂

∂=Δ

gas

OvapourHsw m

mR 2=

Saturated and undersaturated gas

If there is a change in Rsw due to changes of P,Twater may condense even if no water vapour is left in the gas

• Be careful simulating undersaturated gases with standard OLGA

• Possible to turn off water flashing

Standard volume flow in sources and wells

• You may specify e.g. PHASE = OIL in SOURCEQoil , GOR (or GLR) and Water Cut. (Equivalent for gas)

• All the variables are assumed to be at standard conditions• Standard conditions (1 atm. and 60 F) must be a point

within the fluid table P&T boundaries • If you do not specify GOR the fluid table value is used• On terminal nodes you specify GOR (or GLR) and WC

Equivalent mass rate is calculated :

– calculates volume flow of gas using GOR – calculates volume flow of water using WC– converts to mass flows using std. densities– adds up mass flow for each phase

wc: watercut fraction [0, 1]

)1

( STDw

STDg

STDo

STDotot wc

wcGORQm ρρρ−

+⋅+=&

Standard conditions – tips

• Your specified GOR should be ≈ to the GOR of the fluid table

• Use a dense grid in the PVT table around std. cond.– fluid densities are taken from interpolation in PVT table

• OLGA expects that GOR is from a 0. stage flash