NEW APPROACH FOR ENGINE

CONTROL SYSTEM SOFTWARE

DEVELOPMENT
6th CIMAC CASCADES 2015
Borja Pintos

AVL Software and Functions

AGENDA

1. Engine control software

development: Concept
2. Engine control software
development over time
3. New approach: Model based
development
a. Model based development:
phases

b. Model based development:

advantages

c. Model based development:

example

4. Conclusions

| | 27 February 2015 | 2
ENGINE CONTROL SOFTWARE DEVELOPMENT
-CONCEPT
 Software to control the complete powertrain system
Includes: sensors and actuators needed for control strategies
 Typical control systems for large engines:
 Rotational engine speed control  Monitoring and diagnosis
 Rail pressure control  Emissions optimization
 Boost pressure control  Consumption optimization

CLOSED LOOP CONTROL

| | 27 February 2015 | 3
ENGINE CONTROL SOFTWARE DEVELOPMENT
-CONCEPT

Typical example for a large engine: rotational engine speed control

 A lever sets an engine speed set point
 The Engine Control Unit (ECU) functionality shall keep the engine
speed constant to the demand value

Input ECU Control Functionality Output

Visualization
Engine speed
set point
Speed
Controller Fuel mass

Engine speed

Engine

| | 27 February 2015 | 4
ENGINE CONTROL SOFTWARE DEVELOPMENT
OVER TIME
Mayor software improvements over time allowing for: shorter
development time, reduce costs and increased quality

1 2

+ +
No test of requirements Open loop test

Manual code generation Manual code generation

• Very time consuming • Open loop test
• Poor overview • Graphical model overview
• Need software experts • Need software experts
~70%

0% Time consuming 100% 0% Time consuming 100%

| | 27 February 2015 | 5
ENGINE CONTROL SOFTWARE DEVELOPMENT
OVER TIME

3 4
+

+
Open loop test

Auto code generation

Auto code generation

• Open loop test • Closed loop test

• Graphical overview • Graphical overview
• Fast code generation. Code • Fast code generation. Code
generated without errors generated without errors
~50% ~30%

0% Time consuming 100% 0% Time consuming 100%

| | 27 February 2015 | 6
NEW APPROACH: MODEL BASED DEVELOPMENT
Working flow:
1. Development of software functions using a graphical programming
language (e.g. Simulink) according to the customer requirements
2. Test the software functions in Simulink using a virtual engine
model (closed loop test)
3. Generate C code and flash it into the ECU
4. Verification of software functions at engine test bench (closed loop test)
Input ECS Control Functionality Output
Visualization
Engine speed
set point Speed
Controller Fuel mass

Stimuli
Engine speed Engine

| | 27 February 2015 | 7
NEW APPROACH: MODEL BASED DEVELOPMENT

 Detailed engine model:

Crank angle resolution for
cylinder and gas path

 Innovative, numerically-
optimized approach and
advanced solver technology
allows real-time capability

 Multiple applications of real-

time engine models in the
software development
process

| | 27 February 2015 | 8
MODEL BASED DEVELOPMENT: PHASES

Model-in-the-Loop (MiL)
 ECU concept
 Functions and ECU development
Simulink

Hardware-in-the-Loop (HiL) ECU

 ECU development HiL

 Pre-calibration, Monitoring, OBD
Simulink
 Hardware-Tests

ECU
Engine test bench
 Verification
 Refinement Simulink

| | 27 February 2015 | 9
MODEL BASED DEVELOPMENT: PHASES

Model-in-the-Loop (MiL)
 ECU concept
 Functions and ECU development
Simulink

 Entire Simulink environment

 Development of functions
 Virtual engine model interface
available for Simulink
 Closed loop test using virtual
engine models

Motor model
State of the Art:
No model or very simple model (mean model, map)
AVL-Proposal:
Physical Model (Cylinder and air path in CRK angle
resolution)
| | 27 February 2015 | 10
MODEL BASED DEVELOPMENT: PHASES

Hardware-in-the-Loop (HiL) ECU

 ECU development HiL

 Pre-calibration, Monitoring, OBD
Simulink
 Hardware-Tests

Real time PC ~
Computer for
Virtual engine
data analysis

ECU
Actuators

Wiring harness

Lever

| | 27 February 2015 | 11
MODEL BASED DEVELOPMENT: ADVANTAGES

• Higher maturity of engine

control achieved earlier in
the development
• No risk of engine-
component damage in MiL
& HiL environment
• Pre-calibration and
validation of control
functionality are developed
in MiL & HiL environment
• Model based software
development saves time
and money and improves
quality of the software

| | 27 February 2015 | 12
MODEL BASED DEVELOPMENT: EXAMPLE

 Motor type (~ 1 MW per Cyl.)

Bore > 300 mm

 Dual Fuel (Diesel, Gas)

 One stage turbo with

wastegate

 Motor bypass and compressor

bypass valve developed for fine
-control

 Variable valve timing

 Generator operation, ship

propulsion

| | 27 February 2015 | 13
MODEL BASED DEVELOPMENT: EXAMPLE

Building the model Software development Closed loop test (HiL)

Fuel supply
(Gas & Diesel)

Brake,
Charging & external load
Intercooling

Cylinder & Motor (Wall heat

transfer, combustion
Air path (incl. characteristics, valve lift, flow
Wastegate, characteristics, moment of
Compressor-Bypass inertia…)
Variable valve
and Motor-Bypass) timing

| | 27 February 2015 | 14
MODEL BASED DEVELOPMENT: EXAMPLE

Building the model Software development Closed loop test (HiL)

ECS Control Functionality Plant Model

Engine speed
set point Speed Fuel mass
Engine
Controller Engine speed

Lambda

Lambda Boost pressure Waste gate

set point set point Boost
PBoost
Pressure Bypass valve Boost pressure
Demand
Controller

Operating point

| | 27 February 2015 | 15
MODEL BASED DEVELOPMENT: EXAMPLE

Building the model Software development Closed loop test (HiL)

Load isLast
constant
ist konstant(100%)
(100%).
[rpm]

750

700
[Upm]

650
speed
Motordrehzahl

600

550
Engine

Fuel quantity [g/stk]

500 1.4

450 1.2

Einspritzmenge [g/Hub]
1.0

0.8

0.6

0.4

0.2

0.0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Zeit [s]

| | 27 February 2015 | 16
MODEL BASED DEVELOPMENT: EXAMPLE

Building the model Software development Closed loop test (HiL)

Load isLast
constant
ist konstant(100%)
(100%).
750
-Istwert
Actual
700 3.5
Motordrehzahl [Upm]

Lambda [-] [-]

650 -Sollwert
Setpoint 3.0

Lambda
600 2.5
Ladedruck [bar] [bar]

550 2.0

5.5
500 1.5
Ladedruck-Istwert
Boost pressure Boost pressure
Ladedruck-Sollwert
5.0
450 actual value setpoint 1.0
Boost pressure

4.5
4.0

[%] [%]
3.5

3.0 100

position
2.5 80

Wategate-Stellung
60

40
20

Wg
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Zeit [s]

| | 27 February 2015 | 17
CONCLUSIONS

 Model based software development saves time and money and

improves software quality
 Innovative, physical real-time engine models (simulation in crank
angle resolution, 0D gas dynamics) allow the observation of various
motor phenomena that can not be considered with mean value models

| | 27 February 2015 | 18
THANK YOU

www.avl.com