M th d F I i Methods For Improving Turbocharger Turbocharger Simulation Accuracy in GT PGT-PowerOwen RyderGanesan Subramanian

ContentsContents

Basic Errors – Pressure Drops

Turbocharger Shaft PropertiesTurbocharger Shaft Properties

Adjacent Objects To Turbo Maps

T bi MTurbine Maps

Compressor Map Efficiency Extrapolation

2

Basic Errors

⎞⎛⎞⎛ ×=vol PND

m&η

From the definition of volumetric efficiency:

⎟⎟⎠

⎞⎜⎜⎝

⎛×

×⎟⎠⎞

⎜⎝⎛ ×

ref

refTR

PNRPC

ND

Tref = inlet manifold temperatureTref inlet manifold temperaturePref = inlet manifold pressure

= ((Pamb – ∆Pfilter) × PR) – (∆Pcooler + ∆Pthrottle )

[ ]throttlecoolervol

PPDN

IMTNRPCRm

PRΔ+Δ+⎥

⎦

⎤⎢⎣

⎡××

×××η

&

Where:m = mass flow (g/s)

)( filteramb

vol

PPDN

PRΔ−⎦⎣=

η

P R f (P l )

3

m mass flow (g/s)D = Engine swept volume (m³)N = Engine speed in (rev/sec)NRPC = Number of Revs Per Cycle

Pref = Reference pressure (Pascals)Tref = Reference temperature (Kelvin)R = Gas constant for air (= 0.287 J/g.K)

Example Mistakes

Inlet manifold temperature 108.0

111.0

4.0

4.5

Corp. Speed(rps/√K)

prediction was incorrect (50K).98.2

3.5Engine

Test DataW Pressure drop

errors after the compressor have a similar effect

88.43.0

re R

atio

(T-T

) Wrong Filter ∆P

a similar effect.

Pressure drop errors before the

68.7

78.6

2.0

2.5

Pres

su Simulation result

errors before the compressor move running points up the map.

49.1

58.9

1.5

4

the map.

1.00 20 40 60 80 100 120

m√T/p Corp. (kg/sec.√K/MPa)

Back Pressure EffectsOn Turbine Performance

Turbine Size Required To Maintain Lambda With Increasing Exhaust Back Pressure

100Increased exhaust back pressure

80

90e

Swal

low

ing

quire

d (%

)

back pressure reduces turbine power, lowering turbo speed and boost pressure

60

70

ativ

e Tu

rbin

eC

apac

ity R

eqboost pressure.

Back pressure is frequently a source

40

50

0 100 200 300 400 500 600

Rel

a Cfrequently a source of turbine specification errors.

5

Exhaust System Back Pressure (mbar)

Turbocharger Shaft Properties

Bearing Losses at 90°C Oil Temperature80 100% Bearing loss varies

Shaft Properties

60

70

96%

98%

(%)

gwith speed and thrust load, which changes with operating condition

40

50

er (k

W)

92%

94%

Effic

ienc

y operating condition.

20

30Pow

e

88%

90%

Mec

hani

cal

0

10

84%

86%

M

Min. bearing power loss

Max bearing power loss

6

0 20 40 60 80 100 120

Turbo Speed (rpm x 103)

Max. bearing power loss

Compressor Power

Turbocharger Shaft Properties

Bearing Losses at 90°C Oil Temperature80 100% Bearing loss varies

Shaft Properties

60

70

80

96%

98%

00%

(%)

gwith speed and thrust load, which changes with operating condition

7% error

40

50

er (k

W)

92%

94%

Effic

ienc

y operating condition.At 40,000rpm, assuming no frictional losses

20

30Pow

e

88%

90%

Mec

hani

cal

would give around a 7% error in shaft power to the compressor!

0

10

84%

86%

M

Worst Mechanical EfficiencyBest Mechanical EfficiencyTypical Mechanical Efficiency

compressor!

7

0 20 40 60 80 100 120

Turbo Speed (rpm x 103)

Min. bearing power lossMax. bearing power lossCompressor Power

Bearing LossesLossesAs a minimum, have frictional mechanical efficiency dependent on turbo speed.The actual efficiency values will XYTablevary with turbo bearing system and

ti l l il

XYTable

8

particularly oil temperature.

Turbo Rotor InertiaFixed Inertia Value– Low inertia reduces time for turbo speed to reach correct value.

Excessive reduction may mean unrealistic values read from maps

Inertia Multiplier1000

– Excessive reduction may mean unrealistic values read from maps.– High inertia slows the approach to the final value.

100

1000

acto

r

High inertia for initial cycles until model is reasonably stable.

Inertia Profile– Correct inertia and

multiplier profile gives f

10

ultip

licat

ion

Fa

Low inertia to allow turbo to quickly reach a suitable speed.

Ramped return so turbo speed converges correctly.

fast convergence and correct turbo speed.

– Use a monitor to b b d

0.1

1Mu p g y

observe turbo speed convergence. Adjust multiplier profile to get best results for your

9

0 2 4 6 8 10 12

Simulation Cycles0.05 to 0.1 is an adequate minimum.If too small, unrealistic speeds will result.

best results for your particular model.

Sensitivity to Rotor InertiaSensitivity to Rotor Inertia“Correct” values were derived with convergence to 10rpm turbo speed.The following inertia scenarios were simulated: Correct inertia wrongThe following inertia scenarios were simulated: Correct inertia, wrong wheel profiles, next turbo frame size up, inertia ÷1000 and inertia ×1000.

Transient Response120000

Error In Steady State Simulations5%

80000

100000

(rpm

)

1%

2%

3%

4%

40000

60000

Turb

o sp

eed

(Correct InertiaWrong Profile-2%

-1%

0%

1%

Correctinertia

Wheelprofile

Frame Size 1000 × less 1000 × moreErro

r (%

)

Turbo speed (rpm) Air Flow (kg/s)

0

20000

0 1 2 3 4 5 6

gWrong Frame Size1000 × less1000 × more

-5%

-4%

-3%

Turbo speed (rpm) Air Flow (kg/s)

Comp Pressure Ratio Comp eff. (%)

Turb Pressure Ratio Turb eff (%)

AFR EGR

10

The inertia value should be appropriate for steady state and must be accurate for transient.

Time (seconds)

Inertia Effects On Maps

Correct inertia

Inertia 1000 times too small

The difference in inertia is noticeable on the compressor map running points.

11

Maps to right show running point movement during the simulation.

Map Object ConnectionsMap Object ConnectionsThe parts adjoining the turbo maps have an influence on the points plotted on the maps. The adjacent pipe diameter is used to calculate the velocity, and thus dynamic head, defining the difference between static and total pressure.

2⎡ ⎤P =

8 * R * T * m * P * ddynamic

s 2

2 s 4&

π⎡

⎣⎢

⎤

⎦⎥

Inlet dia Outlet dia PR PRmm mm (tot-tot) (stat-stat)

Engine data 61 43.5 2.491 2.525Simulation (correct) 61 43 5 2 495 2 507

Be aware that engine pressure sensors may not be at the turbo so

Simulation (correct) 61 43.5 2.495 2.507Simulation (incorrect) 61 75 2.497 2.569

12

comparing engine data directly to GT-Power comp maps is not comparing like with like.

Missing Turbo I t l V lInternal Volumes

Turbine volute – a missing volume!

Internal volumes of compressor and turbine housings are not usually included.Add as loss-less pipes. Including these volumes affects steady state simulations by 1%.steady state simulations by 1%.Transient simulations are slowed by 4-5% with volumes included.

O fOnly transient events and detailed exhaust manifold studies need volumes adding.Only compressor outlet and turbine inlet volumes are needed.

13

Generic values for each turbo frame size are acceptable.

Turbine Maps

Cummins Turbo Technologies maps turbines on a dynamometer.Reduces need for SAE format file to input for GT-Power. We have test cell data to populate almost the full TRB file with minimal extrapolation. SAE format from Cummins Turbo Technologies, use “PRfitted” option.

SAE Turbine Mapp

Cummins Turbo Technologies turbine dynamometer map

14

dynamometer map.

Compressor Map Issues

Accurate low speed map data for both compressors/turbines is increasingly being demanded by OEMs

Demand

increasingly being demanded by OEMs

Problems involvedHeat transfer between hot turbine end to cold compressor end leads pto inconsistency in test dataMeasurement repeatability, accuracy and the cost involved.M th ti l t l ti d t t th h i f flMathematical extrapolation does not capture the physics of flow losses which is more important at low flows

Experimental figures

+/ 3 5% pts efficiency25 K40k rpm+/- 1.5% pts efficiency60 K60k rpm

Effect of +/- 1 K temp error (inlet / outlet)Approx dTSpeed

15

+/- 10% pts efficiency7 K20k rpm+/- 3.5% pts efficiency25 K40k rpm

Compressor Map Issues p p

80

85

65

70

75

cy (T

-T)

50

55

60

Effic

ienc

40

45

50 Black : TIT coldRed : TIT hot

Flow PR repeat reasonably well

0 10 20 30 40 50 60 70 80 90 100

m (T^0.5)/P (kg/sec. (K^0.5)/MPa)

16

Flow, PR repeat reasonably well.TIT has certainly an effect on compressor efficiency measurement.

Low Speed Extrapolation p pA method based on flow similarity is developed at CTT and being investigated for low flow predictions of compressor/turbine maps.Initial results are encouraging. Simulation results clearly show the importance low speed data accuracy.83

68

73

78

T-T)

20000 rpm40000 rpm

53

58

63

68

Effic

ienc

y (T

43

48

53

0 5 10 15 20 25 30 35 40

Black : TIT cold, Red : TIT hotBlue : Prediction by similarity

17

0 5 10 15 20 25 30 35 40

m (T^0.5)/p Corp. (kg/sec.(K^0.5)/MPa)

Flow and PR Prediction2.4

Corp. Speed(rps/√K)Red : Run hot

Black : Run cold

68.7

2.0

2.2Black : Run cold

Blue : Prediction

58.9

1.8

ress

ure

Rat

io (T

-T)

Flow/PR predictions are reasonably good.

1.4

1.6Pr

39.3

19.6

1.2

18

1.00 10 20 30 40 50 60 70 80 90 100

m√T/p Corp. (kg/sec.√K/MPa)

Engine Simulation MatrixEngine Simulation MatrixThree different map variants were provided for engine performance study:study:– Map – 1, which uses standard mathematical extrapolation.– Map – 2, which contains flow/PR from test data and assumes same

efficiency levels as those at high speeds (60 000 rpm)efficiency levels as those at high speeds (60,000 rpm)– Map – 3 is predicted based on similarity approach.

E i i diti h t b ithi th t l t dEngine running conditions are chosen to be within the extrapolated region of the map. Details are as below.

Engine model 6 Cylinder 12LEngine model 6 Cylinder, 12LEngine speed (rpm) 1100 and 1300Load 20%, 10%EGR f ti 0%

19

EGR fraction 0%

Engine Performance

Compressor PR and Efficiency Vs CA

Engine Performance

1100 rpm, 10% load

MAP-1 Standard mathematical extrapolationMAP-2 PR/MFP Test data with no efficiency lossMAP-3 Similarity concept

1.23

1.25

1.27

70

80

90

10% load MAP-3 Similarity concept

Standard extrapolation predicts low PR and

1.17

1.19

1.21su

re ra

tio

40

50

60

70predicts low PR and low efficiency than other map variants.The map predictions

1.11

1.13

1.15

Pres

s

20

30

40MAP: 1,10%LOADMAP: 2,10%LOADMAP: 3,10%LOAD

The map predictions by similarity is close to test data within 1-2% accuracy.

1.07

1.09

40 80 120 160 200 240 280 320 360 400

Crank angle (deg)

0

10y

20

g ( g)

Sample CMP FilepCTT map conversion GT-Power conversion

Compressor maps are now being generated at CTT using raw test data from gasCompressor maps are now being generated at CTT using raw test data from gas stand in order to avoid intermediate manipulations and subsequent errors. At low speed/flow regions similarity approach is applied instead of a linear extrapolation method.

21

After validation, maps will be available directly in GT-Power format (.cmp and .trb).

The EndThe End

Thank you for your attention!Thank you for your attention!

Owen RyderOwen Ryder Ganesan Subramanian

Cummins Turbo TechnologiesCummins Turbo TechnologiesSt Andrew’s RoadHuddersfieldW t Y k hiWest YorkshireHD1 6RAEngland

/

22

www.cummins.com/turbos