Xin Q. Diesel Engine System Design

Подождите немного. Документ загружается.

912 Diesel engine system design

to regulate the EGR valve and the VGT vane opening. But even with the

most sophisticated map-based gain settings, it is still difcult to achieve

a prescribed emissions trade-off during speed and load transients. Model-

based control is promising in shaping the transient emissions proles as a

function of time.

14.1.4 Analysis of hardware design for engine transients

Turbocharged diesel engines cannot respond to a sudden change in speed

or load as fast as naturally aspirated diesel engines because the change in

compressor air ow lags behind the change in the fueling rate. The reasons

for turbocharger lag include: (1) because the manifold has certain volumes,

it takes time (usually several engine cycles) to gradually build up the gas

pressure in the exhaust manifold and the intake manifold; (2) during fast

acceleration, after the EGR valve is closed it takes time to purge EGR out

of the intake manifold; and (3) because the turbocharger rotor assembly

has a certain moment of inertia, it takes time for the turbine to gradually

accelerate the compressor to a higher speed under the difference between

the turbine power and the compressor power. For smoke control, fueling

rate and engine power are restricted during the fast acceleration transients

according to the available intake boost pressure. Reducing the volume of the

intake manifold and the exhaust manifold can reduce turbocharger lag. For

example, pulse turbocharging with a smaller manifold volume has a better

transient response than constant-pressure turbocharging. Matching the turbine

size with a small ow area for low speed–load conditions can help build up

the turbine power faster during transients and reduce turbocharger lag. Note

that at high speeds and high loads, a wastegated turbine or VGT needs to be

used to prevent over-boosting. Another way to reduce turbocharger lag is to

use low turbo inertia, such as: (1) reducing the turbine size; (2) using two

smaller turbochargers to replace one big unit; (3) using a low-inertia small

high-pressure-stage turbocharger in a two-stage turbocharger system; and (4)

using a ceramic turbine rotor. Other methods that may reduce turbocharger

lag and improve the smoke limit and transient response include: (1) placing

the EGR valve close to the intake manifold to minimize the EGR purging

time; (2) reducing the heat losses in the cylinder and the exhaust manifold;

(3) using a small valve overlap; (4) improving the transient combustion

efciency; (5) retarding fuel injection timing to increase the turbine inlet

exhaust temperature; and (6) using boost-assisting devices such as injecting

extra air or mechanical supercharging during fast acceleration.

One important consideration for turbocharger selection is controlling

compressor surge during transients. On the compressor map, the operating

trace of a fast speed-increase transient is located at the right side of the

steady-state point. The operating trace of the transient of a fast load-increase

Diesel-Xin-14.indd 912 5/5/11 12:05:47 PM



913Diesel engine transient performance and electronic controls

or fast speed-decrease is located on the left side of the steady-state point,

possibly resulting in compressor surge. The compressor surge during fast

transients happens when the compressor ow rate, which responds quickly

to engine speed change, decreases much faster than the lagged compressor

boost pressure. The boost pressure is affected by turbocharger shaft speed

and turbo inertia. There is a need to simulate the fast deceleration to check

compressor surge in turbocharger matching.

In addition to the turbocharger, other engine hardware design evaluations

for transient acceleration performance usually include the effects of engine

inertia, charge air cooler or inter-stage cooler volume and cooling medium

temperature, manifold volume and piping, EGR circuit volume and EGR

purging time, and the location of the aftertreatment components.

Engine electronic controls are reviewed by Stobart et al. (2001). Reviews

on diesel engine controls are provided by Shigemori (1988), Gant and Alves

(1990), Anderson (1991), Winterbone and Jai-In (1991), Hirschlieb et al. (1995),

Guzzella and Amstutz (1998), Guzzella and Onder (2004), and Grondin et al.

(2004). Overviews on engine model-based controls are provided by Hafner

(2001), Lehner et al. (2001), Smith et al. (2007), Turin et al. (2007, 2008),

Šika et al. (2008), Stobart et al. (1998), Atkinson et al. (2009), and Guzzella

(2010). Engine control process development is decribed by Kämmer et al.

(2003), Baumann et al. (2004), and Erkkinen and Breiner (2007).

14.2 Turbocharged diesel engine transient

performance

Turbocharged diesel engine transient performance has been extensively

researched experimentally and analytically since the 1970s. The transient

performance of variable-geometry turbochargers was investigated by

Lundstrom and Gall (1986), Pilley et al. (1989), Brace et al. (1999), and

Filipi et al. (2001). Dynamic optimization and the differences between the

steady-state and transient conditions were researched by Wijetunge et al.

(1999). They addressed the subsystem coordination and interaction during

the dynamic transients. The transient performance of the EGR circuit in a

diesel engine was evaluated by Serrano et al. (2005). Diesel engine transient

operation was systematically summarized in a book authored by Rakopoulos

and Giakoumis (2009).

Computer simulations on turbocharged diesel engine transient performance

were conducted by Ledger and Walmsley (1971), Watson and Marzouk

(1977), Winterbone et al. (1977), Marzouk and Watson (1978), Watson

(1981), Ma and Gu (1990), Qiao et al. (1992), Ma and Agnew (1994) and

a group of researchers at the Universidad Politecnica de Valencia in Spain

(Payri et al., 1999, 2002; Benajes et al., 2000, 2002), as well as a group of

researcher at National Technical University of Athens in Greece (Rakopoulos

Diesel-Xin-14.indd 913 5/5/11 12:05:47 PM



914 Diesel engine system design

et al., 1997b, 2004, 2005, 2007; Rakopoulos and Giakoumis, 2006, 2007,

2009; Theotokatos and Kyrtatos, 2001).

Combustion heat release rate models were explored by Watson et al.

(1980), Felsch et al. (2009), and Serrano et al. (2009a, 2009b). Unlike

previous research that focused on the impact of engine hardware on transient

performance, Watson (1984) proposed a modeling methodology that linked

the relationship between transient performance and engine controls. The

needs of engine performance simulation with electronic controls have started

to emerge ever since.

14.3 Mean-value models in model-based controls

Mean-value real-time transient models are the primary simulation approach

used in gasoline and diesel engine controls. In order to reduce computing

time, the mean-value model uses engine maps such as the volumetric

efciency map and the exhaust manifold temperature map as a function of

other dependent parameters such as engine speed, load, and boost pressure.

The model does not have the resolution at a crank-angle level. The engine

thermodynamic and ow states are represented by their respective single mean

values over the engine cycle. Important progress in this area are detailed as

follows. Real-time modeling of diesel engines was started in the 1980s by

Shamsi (1980), Hendricks (1989), and Jensen et al. (1991). The real-time

mean-value models on gasoline engines are summarized by Hendricks et al.

(1996) on intake manifold dynamics, Chevalier et al. (2000) on the validity

of the mean-value models, and Buckland et al. (2000) on the application

for direct injection gasoline engines. Important research on the turbocharged

diesel engine modeling for nonlinear engine controls was performed by

Kao and Moskwa (1995). More advanced modeling of mean-value models

after the 1990s was investigated by Moraal and Kolmanovsky (1999),

Allmendinger et al. (2001), Eriksson (2002), Jung et al. (2002), Chung et

al. (2005), Fiorani et al. (2006), Eriksson (2007), Pettiti et al. (2007), Chen

(2008), and Olin (2008). A lot of the above-mentioned modeling work was

conducted using MATLAB/Simulink or other programming languages.

GT-POWER is a leading software tool used for engine system simulation,

commercially available from Gamma Technologies. Mean-value modeling in

GT-POWER including controller simulations has been presented by Silvestri

et al. (2000), Papadimitriou et al. (2005), He (2005), He et al. (2006), and

He and Lin (2007).

The above discussion is for the engine air system. On the fuel system side,

a real-time model of fuel injection dynamics used for hardware-in-the-loop

(HIL) test is presented in Woermann et al. (1999).

Diesel-Xin-14.indd 914 5/5/11 12:05:47 PM



915Diesel engine transient performance and electronic controls

14.4 Crank-angle-resolution real-time models in

model-based controls

The high-delity crank-angle-resolution real-time model is more advanced

than the mean-value model. Instead of using the prescribed cycle-averaged

‘mean value’ maps, it can predict the in-cylinder cycle process details at a

crank-angle level without losing the real-time capability. This type of model

is the direction for future development. The real-time crank-angle-resolution

model without manifold gas wave dynamics is presented in Schulze et al.

(2007). The real-time crank-angle-resolution model including the manifold

gas wave dynamics features is developed by Pacitti et al. (2008). Other

related work has been reported by Wurzenberger et al. (2009).

14.5 Air path model-based controls

14.5.1 Lookup table approach in air system controls for

engine transients

In the traditional lookup table controls, the set point is usually either an actuator

position such as the EGR valve duty cycle or a performance parameter such

as MAF (mass air ow), MAP (manifold air pressure), lambda (equivalent

air–fuel ratio) or air–fuel ratio, and intake manifold oxygen concentration.

During transients, the air system component actuator such as the EGR valve

or VGT vane can be driven to a pre-determined position, which is obtained

from the steady-state calibration at a given engine speed and load and then

superpositioned with the transient PID gains and the transient algorithms with

a feedback PID control. The moving part within the actuator has transient

response characteristics against time. The transient delay produces a non-

instant response to the demand from the engine control unit, and the delay is

due to a characteristic time constant. In the MAF control, the actuator such

as the EGR valve is modulated to achieve a preset air–fuel ratio through a PI

or PID controller fed by the feedback difference between the required MAF

and the sensor signal. The sensor signal can be either the actual measured

signal or a calculated signal from a virtual sensor.

The lookup table approach is not exible enough to handle real-world

engine variations. One example of the limitation of the lookup table is the

aftertreatment dynamic dosage control, e.g., hydrocarbon dosing in active

deNO

catalyst or urea dosing in SCR. The lookup table approach cannot

compensate for the interferences from factors such as production variation,

component aging, and engine acceleration or deceleration transients. During

transient turbocharger lag or valve lag, the dosage rate determined from the

steady-state calibration lookup tables may not be suitable and could result

in an unacceptable slip amount.

Diesel-Xin-14.indd 915 5/5/11 12:05:47 PM



916 Diesel engine system design

14.5.2 Model-based controls for engine transients

Traditional engine controls used lookup tables. The calibration complexity

increases exponentially as new functions and the number of associated control

tables or maps increase in the development of modern turbocharged EGR

engines. Electronic controls have been evolving toward the mathematical

model-based controls, either open loop or closed loop, which have become an

important part of system design and diagnosis. The success of online model-

based controls relies largely on the accuracy of the thermodynamic cycle

performance models. The models can be built for various operating conditions

such as new and aged engines, normal and extreme climates. Model-based air

system controls use the sensors and the actuators to sense and control the gas

 ows inside and outside the cylinder, such as using engine valve actuation,

EGR valve, intake throttle valve, exhaust back-pressure valve, and turbine

vane actuator. Model-based cooling system controls use the sensors and

actuators for coolant  ows. In order to provide  exible cooling as needed to

minimize driving power consumption and enhance engine performance, the

desirable cooling system in the future is to use more electronic controllers

in addition to the current pump, fan, and thermostat. The gas-side data and

coolant-side data may be linked by a model of heat rejection and other

engine performance parameters. In the coordinated model-based control of

EGR valve and turbine actuator, according to the theory in Chapter 4, the

valve opening and the actuator setting may be calculated by specifying two

performance targets among the following three parameters: fresh air  ow

rate, EGR rate, and intake manifold boost pressure. The engine performance

parameters used in model-based controls can be either the measured signals

from real sensors, or the simulated data from virtual sensors.

The opening position of valves (e.g., EGR valve, intake throttle valve,

exhaust back-pressure valve) can be calculated based on target performance

parameters. For example, the gas mass  ow rate through an EGR valve ori ce

can be modeled by using the equation of isentropic compressible  ow of the

ideal gas:



EGREEG

EGR

ups

trea

gas

ups

trea

mC= mC

– 1

–

ups

trea

downs

trea

owns

downsd

oownso

dodownsdod

trea

ups

trea

14.1

where A

EGR

is the theoretical effective  ow area of the EGR valve ori ce

at a given opening and C

EGR

is a variable correction coef cient. A

EGR

obtained from the correlation between valve lift and valve  ow area. C

EGR

Diesel-Xin-14.indd 916 5/5/11 12:05:48 PM



917Diesel engine transient performance and electronic controls

is calibrated by measurement data at various engine ow conditions and is

used to correct any inaccuracy in the theoretical valve ow area. To meet the

valve actuator control demand, the required valve ow area can be calculated

rst from the desired EGR mass ow rate at any given engine speed and load

by rearranging equation 14.1. Then, the valve area can be converted to the

valve opening position or lift. In model-based controls, in order to achieve a

desirable transient EGR rate (not necessarily the minimum deviation from the

steady-state EGR rate), a target transient rate prole may be exibly dened

to a certain shape as a function of the rate of transient speed–load change.

For example, the target transient EGR rate can be imposed by a xed EGR

percentage multiplied by the measured total engine gas mass ow rate. It

is worth noting that in model-based control the dynamic behavior of the

component can also be modeled.

To illustrate the effect of different engine control methods on transient

performance, a step increase in the fueling rate followed by a step decrease

is shown in Fig. 14.1. As a result, engine speed and brake torque change

through three steady-state modes, named A, B, and C. Such an event is a

typical representation of real-world driving cycles. Due to turbocharger lag,

there are inevitable delays in the exhaust manifold pressure and the intake

manifold boost pressure during the fast transient of fueling rate change. The

instant fueling increase results in a fast decrease in the air–fuel ratio at the

beginning of the transient event, possibly reaching the smoke limit. The level

of EGR rate commanded by the engine controls has a direct impact on the

air–fuel ratio, NO

and soot emissions during the transient. For example, in

the method of EGR valve position control, the EGR valve opening undergoes

sharp changes from the steady-state calibration opening of mode A to those

of modes B and C. The resulting EGR ow reduces the air–fuel ratio and

increases the transient soot while keeping the transient NO

low. On the

other hand, in the method of lambda (i.e., equivalent air–fuel ratio) control

or MAF control, in order to maintain the steady-state setting of air–fuel ratio

during transients, the EGR valve is commanded to close when the fueling

rate suddenly increases, and to open more when the fueling rate suddenly

decreases. As a result, during a step load increase, less or even zero EGR

ow is obtained. Consequently, a higher air–fuel ratio and lower transient

soot are achieved, but the transient NO

and pumping loss (due to a higher

engine delta P) may be higher compared with the position control.

The transient soot spike is an inevitable phenomenon during fast accelerations

due to turbocharger lag. The task of the EGR control is to regulate the EGR

ow to minimize the transient emission spikes or to achieve the best NO

–soot

trade-off with the best compromise on drivability. Meanwhile, EGR control

needs to be coordinated with turbocharger control. The intake manifold boost

pressure can then be properly controlled. Changes in intake and exhaust

restrictions, production variations, and changes in ambient conditions can

Diesel-Xin-14.indd 917 5/5/11 12:05:48 PM



Transient responses of position control method

Transient responses of lambda control (solid) and model-based control (dashed lines)

Turbo lag

Mode C

Soot

Time

Use pre-set position

of valve opening

Steady state mode B

Air–fuel

ratio

Boost

pressure

Engine

delta P

EGR rate

Fuel rate

EGR rate

Reach smoke limit

High soot due to EGR on

High pumping loss (lambda control)

Reduced pumping loss

Transient EGR rate target in

model-based control

Reach zero to match pre-set A/F ratio in

lambda control

Low NO

due

to EGR on

Air–fuel

ratio

Soot

Above smoke limit

Transient soot in

model-based control

Lower soot due

to higher A/F

ratio

Low transient NO

in model-based

control

Transient A/F ratio

target in model-

based control

High No

due to EGR

off in lambda control

14.1 Illustration of

transient engine

controls and air

system performance.

Diesel-Xin-14.indd 918 5/5/11 12:05:48 PM



919Diesel engine transient performance and electronic controls

be promptly responded to. It is difcult for the position-based control to

accomplish these tasks. The MAF or MAP control can only partially fulll

the requirements. A promising model-based control may nally compute the

actuator position such as the EGR valve opening or the VGT vane position

accurately with a calibrated model based on the preset desirable engine

performance parameters. The preset parameters may include the desirable

transient EGR ow rate prole or the dynamic transient fuel injection control

parameters, or the desirable NO

or soot limit if the real-time NO

or soot

model is available. Another advantage of model-based control is that its

steady-state engine calibration set points and transient calibration gains are

much less hardware-dependent than in the position-based controls, because

the set points are more fundamental engine performance parameters, such

as air–fuel ratio and EGR rate, rather than the valve opening or the vane

opening of a particular EGR valve or VGT. Figure 14.1 illustrates the concept

of model-based EGR rate control or air–fuel ratio control in order to reduce

transient NO

and pumping loss. It also demonstrates how to optimize the

trade-off between NO

and soot during load increase and decrease.

The VGT effective area opening or turbine wastegate opening can be

modeled similarly with an orice ow equation (4.57) and by using the

turbine power equation. Model-based turbocharger control may reduce

turbocharger lag and prevent transient compressor surge. A detailed theory

of model-based VGT and EGR nonlinear controls was outlined by Ammann

et al. (2003). In summary, hardware design needs to match electronic control

strategies so that the inherent transient difculties (e.g., turbocharger lag or

high transient pumping loss) can be best alleviated.

Air path control strategies for turbocharged diesel engines usually include

the research topics of coordinated EGR–VGT control algorithms, controls by

MAP, MAF or exhaust manifold pressure, and controller designs. These topics

were extensively researched by Watson and Banisoleiman (1988), Winterbone

and Jai-In (1988), Gissinger et al. (1990), Duffy et al. (1999), Shirawaka et

al. (2001), Wijetunge et al. (2004), Nieuwstadt et al. (2000), Osborne and

Morris (2002), Nieuwstadt (2003), Ammann et al. (2003), Kolmanovsky

and Stefanopoulou (2000, 2001), Kolmanovsky et al. (1999), Yokomura et

al. (2004), Mueller et al. (2005), Kobayashi et al. (2005), Darlington et al.

(2006), Black et al. (2007), Luján et al. (2007), Das and Dhinagar (2008),

Plianos and Stobart (2008), and Moulin et al. (2009).

14.6 Fuel path control and diesel engine governors

Fuel path control is another important area in diesel engine controls. The

modeling work includes three main areas: (1) engine speed and governor

controls; (2) real-time modeling of fuel system hydraulic dynamics for HIL;

and (3) fuel delivery unevenness detection and misre detection, and their

Diesel-Xin-14.indd 919 5/5/11 12:05:48 PM



920 Diesel engine system design

model-based controls. The fuel path dynamics of a diesel engine is very

different from that in a port-injection gasoline engine because the diesel

engine does not have such issues as wall-wetting and mixing/evaporation

during the fuel transporting process.

Mechanical governors and modern electric governors have always been

used in diesel engines to control engine speed (SAE J830, 1999). Diesel

fuel control and governor designs are part of the fuel system development.

Diesel engine governors and fuel path control were investigated by Gant

(1984), Okazaki et al. (1990), Bazari (1990), Rakopoulos et al. (1997a),

Mruthunjaya and Dhariwal (2000), Stefanopoulou and Smith (2000), Larisch

and Sobieszczanski (2001), Makartchouk (2002), Chatlatanagulchai et al.

(2009), and Deng et al. (2010).

Fuel delivery unevenness detection and misre detection are important for

fuel injection quantity control, especially real-time correction of fuel injection

failures. The unevenness refers to the difference in fuel quantity between

the cylinders or cycles. Macián et al. (2005, 2006a, and 2006b) explored

this area of fault diagnostics and proposed improved control algorithms

and controller designs. The conventional diagnosis techniques use the

instantaneous crankshaft speed during an engine cycle as input information.

Although these techniques are able to detect misre at low engine speeds,

they are not very effective at high speeds, especially at low loads (Macián et

al., 2006a). Other engine system performance parameters were explored as

better alternatives to replace or supplement the crankshaft speed signal for

the purpose of fault diagnosis and control of the injection and combustion

processes in order to better detect fuel delivery unevenness and misre at

high engine speeds. These parameters explored included exhaust manifold

pressure, instantaneous turbocharger speed, and the mean temperature in

the exhaust port. Macián et al. (2006b) reported that using instantaneous

turbocharger speed as input signal was effective for fuel quantity control

in order to achieve satisfactory correction on fuel injection unevenness.

A system-level fault diagnosis model can be developed in engine system

design by combining the thermodynamic performance and engine dynamics

submodels to facilitate fuel system control development.

14.7 Torque-based controls

Torque-based controls are widely used in both gasoline and diesel engines

to facilitate the overall powertrain control demands. Engine system design

models can play a key role in developing the more accurate torque-based

control models because the diesel engine system design engineers analyze

engine torques in their daily work and have a thorough understanding of

the engine torque behavior during steady-state and transient conditions.

System engineers can complement control engineers very well in this area.

Diesel-Xin-14.indd 920 5/5/11 12:05:48 PM



921Diesel engine transient performance and electronic controls

For example, the friction torque and pumping loss torque calculations/

estimates used in the torque-based controls can be re ned and enhanced by

the system design. Moreover, more advanced parametric dependency of the

engine indicated torque and brake torque can be developed by the engine

system design engineers for the torque-based control models in order to

accurately calculate the engine torques under various operating or aging

conditions. Torque-based controls are discussed by Ginoux and Champoussin

(1997), Müller and Schneider (2000), Greff and Günther (2001), Heintz

et al. (2001), Lee et al. (2001), Park and Sunwoo (2003), Wang and Chu

(2005), Katsumata, et al. (2007), and Livshiz et al. (2004, 2008) for gasoline

engines; and Maloney (2004), Li et al. (2002), Grünbacher et al. (2003),

Chauvin et al. (2004), Brahma et al. (2008), Tian et al. (2008), and Oh et

al. (2009) for diesel engines.

14.8 Powertrain dynamics and transient controls

14.8.1 Transient performance simulation

Transient engine testing is much more dif cult and expensive than steady-

state testing. Using engine cycle performance simulation to study transients

is a very effective approach, especially for real-time transient modeling. The

engine crankshaft lumped model has the governing equation 5.20. In transient

engine torque simulation, the ‘ lling-and-emptying’ method is usually used

to model the zero-dimensional manifold gas dynamics in order to reduce

the computing time. There are several key issues related to model accuracy

in transient cycle simulation. They are listed below.

1. During the fast acceleration transient, a sudden increase in fueling

causes a rapid decrease in the air–fuel ratio that may result in incomplete

combustion. Combustion ef ciency as a function of air–fuel ratio was

usually assumed as input in the thermodynamic cycle transient simulation.

Engine testing or sophisticated combustion simulation that tries to quantify

such ef ciency change remains challenging.

2. Another dif culty is related to the turbocharger. The instantaneous

turbocharger speed N

(revolution per second) is given by

–

dNd

– WW –





14.2

where I

is the moment of inertia of the turbocharger,



is the turbine

power,



is the compressor power, and



is the moment of inertia of the turbocharger,



is the moment of inertia of the turbocharger,

is the turbocharger

bearing friction power. The turbine instantaneous temperature, pressure

and ef ciency affect N

and turbocharger lag. Transient turbocharging

performance modeling is usually approximated by a quasi-steady-state

approach, which computes the instantaneous varying turbocharger

Diesel-Xin-14.indd 921 5/5/11 12:05:49 PM

