Engine Calibration Process for Evaluation across the Torque-Speed Map Brian Froelich Tara Hemami Manish Meshram Udaysinh Patil November 3, 2014

Cummins Confidential

Outline :

– Background – Objective – Calibration process for torque speed map – Steady state calibration results – Transient simulation using Fast Running Model – Summary

13-10-2014 GT Conference 2014 2

Cummins Confidential

Background Information :

Optimization

Engine Speed

To

rqu

e

DoE Design

Test Cell

IV ~1000+ pts Testing Process

~200 to 400 pts

2500 pts Simulation Approach

Steady State Transient New controls strategies

Simulation

Cummins Confidential

Background Information :

Test engine at limited points. Calibrate engine at limited operating points GT-Power model to predict engine performance across entire torque-

speed map.

Torq

ue (N

-m)

Speed (RPM)

All test points Selected test points

13-10-2014 GT Conference 2014 4

Cummins Confidential

Overall Process :

13-10-2014

Combine Data for MLR Modeling

CyberCal Process

Test Plan to collect data for model calibration and

validation

Engine Pre-Test work complete; Targets, Constraints and Levers

identified; Parent GT Power Model identified

Model ok to use for Cal

Development?

NO YES

Run Engine

Modify GT-Power model

Calibrate GT-Power model

Validate GT-Power model

Run full space fill for GT-Power

models

Correct for “bias” in model predictions

~ 200 to 400 Points

GT Conference 2014 5

Combine data for optimization

Model ok for further studies?

Optimization Studies

Cummins Confidential

Test Cell Data Detailed GT-Power model with the correct geometry

Calibrate standalone Subsystems

Combine the Subsystems into the whole engine model

Check the model with test cell data

Check the model with Validation data

Calibration Data

Validation Data

Calibrated GT-Power Model

Subsystem models Intake system Compressor Charge Air Cooler Intake manifold + Cylinder + Exhaust manifold EGR Loop Turbine Exhaust system

Process development workflow : ~ 200 to 400 Points

~ half of selected Points

Goal •Engine performance parameters within error bars. (torque, flow, pressures, temperatures)

13-10-2014 GT Conference 2014 6

~ half of selected Points

Intake restriction subsystem with imposed boundary conditions

Develop a correlation of intake restriction diameter and air flow

Predicted performance within +/-3 % error band

Model calibration : (Standalone Subsystem level calibration)

Example: Calibration of standalone intake restriction model

Air flow rate In

take

rest

rictio

n

Experimental

Sim

ulat

ion

Calibrate standalone system models like turbochargers, after cooler etc. Calibrate within error limits

13-10-2014 GT Conference 2014 7

Intake system

Turbocharger

Example :Integrating intake system and turbocharger model The independently calibrated intake system and turbocharger coupled together.

Model calibration (Integrating Subsystem models together)

Experimental

Sim

ulat

ion

Experimental Si

mul

atio

n

Comp out Pressure Turbo speed

Integrate the well calibrated subsystem models

Cummins Confidential 13-10-2014

Observations –Experimental vs Simulation : Torque Air flow rate

GIMEP NIMEP

Predictive GT-Power combustion model used for predicting performance across entire torque speed map

GT Conference 2014 9

Cummins Confidential 13-10-2014

Peak Cylinder Pressure Intake Manifold Temp

Turbine Inlet Temp Turbocharger speed

Observations –Experimental vs Simulation

Predictive GT-Power combustion model used for predicting performance across entire torque speed map

GT Conference 2014 10

Cummins Confidential

Observations –Experimental vs Simulation :

13-10-2014 GT Conference 2014 11

GT-Power predictions for 1000+ operating points.

Cummins Confidential

Transient Cycle Calibration using GT-Power

13-10-2014 GT Conference 2014 12

Cummins Confidential

Transient Requirement: GT-Power model is the engine and Simulink harness will act as

interface to Electronic Control Module (ECM) Quantify effect on engine performance using levers like Start of

Injection, Rail Pressure, Fueling quantity, EGR Valve position, Turbocharger Rack Position.

Need a predictive combustion model to capture effect of Start of Injection , EGR % and Rail Pressure

GT Model should run close to real time

Transient inputs [ 1200 sec cycle ] in which controls engineer is interested

13-10-2014 GT Conference 2014 13

Cummins Confidential

Engine Model Capability : DI-Jet GT-

Power model DI-Pulse GT-Power model

Fast Running GT-Power

model

Reduced GT-Power model

Predictive Combustion Model

Old New New No

Air Handling Detailed Detailed Simplified Simplified

Pressure pulsations Yes

Yes

Yes

No

Crank angle resolved data output

Yes

Yes

Yes

No

Simulation Time [ wrt real time ]

Very slow Moderate Close to real time

Very fast

New DI-Pulse combustion model from GT

13-10-2014 GT Conference 2014 14

Cummins Confidential

Transient Cycle simulation work flow :

13-10-2014

• Model Calibrated with process developed

GT-Power model for

Calibration

• Baseline Transient Simulation results

GT-Power model Transient

Simulation • Improves model simulation speed

Fast Running Model

conversion

• Check for Fast Running Model results

Fast Running Model Transient

simulation • Fast Running Model with Simulink Harness

Fast Running model for controls

Detailed GT-Power Model Fast Running Model with

Simulink Harness

GT Conference 2014 15

Cummins Confidential

Fast Running Model transient simulation results :

FRM S-Function in Matlab / Simulink environment

FRM Transient simulation results in Simulink environment

Improvement in transient simulation time using FRM

13-10-2014 GT Conference 2014 16

Cummins Confidential

Summary :

Reduced calibration time for large simulation points. Confidence in GT-Power prediction across the entire torque speed

map. FRM developed following the calibration process which gives good

predictions for transient simulation FRM with predictive combustion [ DI-Pulse ] model runs close to real

time simulation

13-10-2014 GT Conference 2014 17

Cummins Confidential

Thank You

13-10-2014 GT Conference 2014 18