Potential of 2-stage Turbocharging

E. Codana,1 and T. Hubera,2

a ABB Turbo Systems Ltd. Bruggerstrasse 71a, CH-5401, Baden, Switzerland

Abstract The first applications of two stage turbocharging on large four stroke en-gines are in operation [1], [2]. The main drivers for the introduction of this technology are power density, engine efficiency and low emissions. This paper discusses certain aspects and requirements of engine and tur-bocharging system design. Starting from a very basic analysis of the ther-modynamic cycles, the potential of two stage turbocharging is studied for different configurations of the turbocharging system. The results of de-tailed studies for the different engine segments are also presented. Engine size, operational envelope and performance requirements influence both the design parameters of the system and its development potential. Emis-sions represent a further challenge for large engines; the contributions of the turbocharging system are discussed. Key Words: Turbocharging, Thermodynamic Processes, Miller

1 E-mail: ennio.codan@ch.abb.com, URL: http://www.abb.com/turbocharging 2 E-mail: thomas.huber@ch.abb.com, URL: http://www.abb.com/turbocharging

1 Introduction

High pressure turbocharging in combination with the Miller process has recently been intensively investigated. Several studies have been per-formed by means of advanced simulation tools and confirmed by engine tests [3] [4] [5] [6]. These studies show that there is substantial im-provement potential for engine performance by applying two stage turbo-charging in combination with Miller timing working at a boost pressure in the range of 8 bar and higher. This type of process relies on the reduction of the temperatures as per the Miller process. This reduction in tempera-ture has a very positive effect on the engine’s efficiency and its raw NOx-emissions. At the same time, the high turbocharging efficiency achievable with two stage turbocharging allows the delivery of the required air pres-sure in an efficient way and additional power due to the improvement in gas exchange work. The above mentioned studies can be seen as a conventional approach to applying two stage turbocharging to further developed engines. The intro-duction of new technologies like two stage turbocharging and the Miller process in the field of large engines takes the engineer in a new, widely unexplored landscape. The interactions between turbocharging system and engine are much more important than before. It is then important to have a look at the basics of the turbocharged engine and see how the new boundary conditions change and enhance our basic knowledge. After a theoretical review, the presented relations are applied to different engine configurations. Finally the influence of engine size and type are also considered, showing that the optimisation potential is similar for all engines, but the boundary conditions of different engines can lead to different targets. Low speed two stroke engines are not considered here; application of two stage turbocharging on two stroke engines is subject to other boundary conditions and would require a different approach.

2 Thermodynamics of the Turbocharged Engine

The analysis and optimisation of the complete thermodynamic process of a two stage turbocharged engine is complex. It is useful to rely on refer-ence processes which help to break down the complexity. For this scope, the following processes are considered: the engine process, divided into closed cycle, gas exchange and Miller loss and the turbocharging process, divided into expansion and compression. Additionally, some considerations are necessary regarding scavenging the combustion space.

2.1 The closed cycle

The closed cycle is defined here as a process starting and ending at BDC at the points start of compres-sion (ac) and end of expansion (ex) (figure 1). These points are defined in the p–V diagram as extrapola-tions to BDC by means of an isen-tropic expansion of the respective points where the inlet valve closes (IVC) and the exhaust valve opens (EVO). In the case of Miller with IVC before BDC, the point ac is the effective start of compression. Taking the closed cycle as the most important part of the process, one obviously must consider parame-ters like the compression ratio , firing pressure pmax and its ratio to compression pressure pc. The latter plays an important role in the trade-off between efficiency and NOx-formation. Experience has shown that increasing pmax/pc leads to better efficiency, albeit with higher NOx emissions. In a conventional cycle the points ac and ex are only connecting points to the gas exchange cycle. The point ac defines the available charge air per cycle; it is dictated by the required air/fuel ratio. The point ex depends on the engine energy balance; the pressure difference pex – pac correlates di-rectly with the difference between fuel energy and indicated work. With the introduction of the Miller cycle, the point ac acquires an additional de-gree of freedom. Theoretically the temperature tac could be reduced con-siderably, but there are limitations imposed by the increasing ener-getic losses of extreme Miller tim-ing and by the issues of what we call here “cold combustion” - it has been observed that below certain pressure and temperature values, especially at part load, diesel com-bustion assumes a knocking char-acter with low efficiency and high NOx-emissions. It has already been shown [3] that lowering the tem-perature at point ac reduces the temperature level of the whole closed cycle and that this has a very positive impact on the cycle efficiency: at lower temperatures

TDC BDC

exac

EVO

V/VD

pp

V/VD(IVC)TDC BDC

exac

EVO

V/VD

pp

V/VD(IVC)

Figure 1: Pressure – volume dia-grams of the closed cycle

0.48

0.49

0.50

0.51

0.52

250 300 350 400Tac [K]

HD

Lambda 2.2Lambda 2.0Lambda 1.8Lambda 1.8 - trec -20Linear (Lambda 2.0)

Figure 2: Closed cycle efficiency over Tac

the specific heat of the charge air is reduced, i.e. less fuel energy is re-quired for producing the same pressure increase in the cylinder. At the same time the lower combustion temperature reduces NOx formation. The temperature at ac also has another consequence: a lower pressure pac is required for preserving the air/fuel ratio and the whole compression curve is lower, giving room for increasing the ratio pmax/pc, even at high and , without going beyond the design limit for firing pressure. The curves in figure 2 show the evolution of the closed cycle efficiency over temperature tac. They have been calculated for a model engine in which IVC is varied under the following boundary conditions: the air/fuel ratio C is maintained constant by varying charge air pressure pRec, firing pressure has been kept constant by adjusting the start of injection, while the turbocharging efficiency of a two stage turbocharging system is ap-proximately constant. The curves show some variability with different val-ues of C, but the general trend can be expressed as a slope of 1% of closed cycle efficiency improvement for 15 °C tac reduction. This result will be used in section 2.5 for evaluating the induction process. The closed cycle is characterised by its indicated mean effective pressure pmi,HD and its thermal efficiency HD:

QVp DHDmi

HD, Eq. 1

respective its variation against a reference cycle:

ReferenceHDHDHD , Eq. 2

2.2 The Gas Exchange Process

Looking at the gas exchange work is an important issue because it con-tributes in a positive way in case of a highly turbocharged engine. There-fore we could speak of gas exchange work instead of “pumping work”, which resembles the parasitic losses by natural aspirated engines. The mean effective pressure of the gas exchange work is:

HDmimiLWmi ppp ,, Eq. 3

2.2.1 The Theoretical gas exchange

The work of the gas exchange correlates strongly with the pressure differ-ence delivered by the turbocharging system; therefore it is convenient to define the theoretical work and efficiency produced by the turbocharging system:

TIcLWthmi ppp Re, Eq. 4

QVp DLWthmi

thLW,

, Eq. 5

2.2.2 The Miller Loss

Extreme Miller timing has a big impact on the engine process, as shown in figure 3.

pmi,HD

pmi,LWth

pex

precpTI

pac=

TDC BDCV/VD

p pmi,HD

pmi,LWth

pex

precpTI

pac=

TDC BDCV/VD

p

a) Conventional cycle

pmi,HD

pmi,LWth

12

pex

precpTI

pac

TDC BDCV/VD

p pmi,HD

pmi,LWth

12

pex

precpTI

pac

TDC BDCV/VD

p

b) Miller cycle

Figure 3: Miller loss

For a conventional filling optimised theoretical cycle, the closed cycle and the gas exchange cycle are not overlapping. In a cycle with extreme Miller timing the gas exchange loop is raised well above pac. Doing so, some positive work is lost: The area 1 is cut from the closed cycle; the area 2 is cut from the gas exchange cycle. It would be rather complicated to correct the definitions of both parts of the cycle. It is much more convenient to leave them unchanged and to introduce the Miller loss, which is the sum of both areas and can be calculated as follows:

1

111

11 Rec

ac

Rec

acRecMillermi p

ppppp , Eq. 6

QVp DMillermi

Miller, Eq. 7

2.2.3 The Valve Losses

The real process deviates from the ideal one depicted in figure 3 due to pressure losses through the valves. These losses can be calculated using the terms defined so far:

MillermiLWmiLWthmiFlowmi pppp ,,,, Eq. 8

QVp DFlowmi

Flow, Eq. 9

The valve losses can range between 1 and 4 points of engine efficiency depending on the engines speed and cam design (maximum acceleration). But for a given design at constant speed they are practically constant.

2.3 The Complete Engine Process

The real complete engine process can be split into the four part defined above (figure 4).

2.4 The Exhaust Process

The exhaust process involves the discharge of gas from the cylinder and the expansion of this gas outside the cylinder. The ideal process starts from point ex. According to present engine cycle simulations for a highly turbocharged 4-stroke diesel engine with Miller timing and an indicated mean effective pressure pmi = 30 bar, values of pex = 11 bar and Tex = 1000 K are defining the thermodynamic characteristics of the cycle. For the sake of simplicity, scavenging has not been taken into consideration.

Closed cycleTheoreticalgas exchange work

Miller loss

Flow lossespex

prec

pTI

pac

TDC BDC0 0.2 0.4 0.6 0.8 1 1.2

V/VD

p

Closed cycleTheoreticalgas exchange work

Miller loss

Flow lossespex

prec

pTI

pac

TDC BDC0 0.2 0.4 0.6 0.8 1 1.2

V/VD

p

Figure 4: Complete process analysis

The task is now to establish the maximum power that can be obtained from the expansion of this exhaust gas. The work of the exhaust process has been idealised as shown in figure 5 [7]. To make the following discus-sion more understandable, two curves have been plotted in the p-V dia-gram representing:

a) an isentropic expansion of the gas in the cylinder to ambient pres-sure starting from the point ex at BDC

b) a curve derived from a) assuming that the gas is collected in a manifold and expansion work outside the cylinder is dissipated.

The exhaust gas temperature for the curves a and b could be calculated under the assumption of a constant heat capacity ratio as follows:

1

ex

TIexaTI p

pTT , Eq. 10

ex

TIexbTI p

pTT 11, Eq. 11

With the help of the curves a and b, different exhaust processes, the ideal pulse and the ideal constant pressure process can be analysed. The ideal pulse process relies on using the blow-down energy. The gas expands to ambient pressure with the piston at BDC. This proc-ess would require putting the tur-bine in the position of the exhaust valve. The available energy is the sum of areas 1 and 2. The constant pressure process re-lies on filling a volume at an inter-mediate pressure pTI. The blow-down energy is not converted into turbine expansion work and in-creases gas enthalpy to the turbine: the isentropic expansion in the dia-gram starts from the intersection of the pTI line with curve b. The piston has to provide work during the exhaust stroke. The available energy for the turbine is given by the sum of the areas 2, 3 and 4, whereby area 4 is provided by the piston, i.e. available engine power is reduced by this amount. A real pulse process is something between the two ideal cases. A fraction of the blow-down energy (area 1) can be utilised; the remaining fraction is converted into increased enthalpy for the steady flow expansion (area 3). Calling x the conversion rate of the blow-down energy, the available tur-

0

1

2

3

4

5

6

7

8

9

10

11

12

0 2 4 6 8 V/Vd

p

4

1

2 3

pTI

pex

BDC

TDC

[bar]

b

a

0

1

2

3

4

5

6

7

8

9

10

11

12

0 2 4 6 8 V/Vd

p

4

1

2 3

pTI

pex

BDC

TDC

[bar]

b

a

Figure 5: Pressure – volume dia-grams of the exhaust process

bine energy is given by the sum of the areas 2 and 4 plus x times area 1 and (1-x) times area 3. The evolution of the available power with the different ideal exhaust proc-esses, pulse and constant pressure, starting with pex = 11 bar and tex = 1000 K, is represented versus the expansion ratio T (fig.6a). The turbine isentropic power with pulse (a) starts from a high value at expansion ratio 1 and increases in linear progression with the expansion ratio. The same curve (c) for constant pressure starts from zero and reaches the line for pulse at the highest expansion ratio. The middle lines (b and d, respec-tively) represent the net isentropic power, i.e. turbine power minus the piston power needed to expel the gas from the cylinder. The pulse line shows a slight negative slope; and the constant pressure curve again shows an increase from zero to the same final value. The line e is the net effective power of the exhaust process, applying constant pressure and constant overall turbine efficiency. The curve now shows a very flat maxi-mum at an expansion ratio of 7. The value of the curve is practically con-stant in the range from 6 to 8, which indicates the optimum range of the exhaust process. A sensitivity analysis has been per-formed for the curve of effective power e with a constant pressure exhaust system (figure 6b). The influences are listed as follows:

± 1% turbine efficiency ± 1.85% net turbine power

± 1% pex ± 1.6% net tur-bine power, ± 1% mass flow rate

1 3 5 7 9T Expansion ratio

Turb

ine

wor

k pe

r cyc

le

CP Isentropic turbine workPulse Isentropic turbine workCP Net isentropic workPulse Net isentropic workCP Net effective workPulse Net effective work

a

b

c

d

e

50%

eng

ine

outp

ut

f

1 3 5 7 9T Expansion ratio

Turb

ine

wor

k pe

r cyc

le

CP Isentropic turbine workPulse Isentropic turbine workCP Net isentropic workPulse Net isentropic workCP Net effective workPulse Net effective work

a

b

c

d

e

50%

eng

ine

outp

ut

f

a) Turbine power versus T

2 4 6 8 10T Expansion ratio

Net

effe

ctiv

e tu

rbin

e w

ork

±0.05 turbine efficiency ±1 bar pex

±100 K Tex

2 4 6 8 10T Expansion ratio

Net

effe

ctiv

e tu

rbin

e w

ork

±0.05 turbine efficiency ±1 bar pex

±100 K Tex

b) Sensitivity of turbine power

Figure 6: Energy balance of the exhaust process.

2

4

6

8

10

12

14

16

0 1 2 3 4V/VD

Pressure[bar]

CylinderExhaust manifoldIsentropic expansion

ex

EVO

pRec

pTI,mean

Expansion loss

BDC2

4

6

8

10

12

14

16

0 1 2 3 4V/VD

Pressure[bar]

CylinderExhaust manifoldIsentropic expansion

ex

EVO

pRec

pTI,mean

Expansion loss

BDC

Figure 7: Blow-down energy conver-sion

± 1% Tex ± 0.1% net turbine power, 1% mass flow rate The curve of the effective power for the pulse system (f) is lower than curve (e) in the upper range of expansion ratio. It is difficult to define an efficiency for the conversion of the blow-down energy, because the proc-ess is highly unsteady. A value of 25% coupled with 2 points reduction for the steady state expansion has been obtained by simulations. The reason for the low conversion efficiency is that high expansion losses always occur at the beginning of the blow-down phase (figure 7). On the other hand, the power of the pulse system at full load is only marginally lower and the possibility for using pulse turbocharging for part load opti-misation will be discussed in the following section.

2.5 The Charging Process

After fixing the maximum power that can be extracted from the exhaust gas, the next step is to find out which is the most efficient way to use this power. The proc-ess is described in figure 8. Area 1 represents the isentropic com-pression work for the air entering the cylinder. The intermediate step in the compression curve is the effect of the intercooler be-tween the compressor stages. Area 2 is the work that would be gained from piston movement un-der receiver pressure. Area 3 is

4 6 8 10 12Pressure ratio C

Com

pres

sion

pro

cess

wor

k

Isentropic compressor work

Compressor work

Compressor - piston work

Compressor - piston work – closed cycle gain

a

b

c

d

4 6 8 10 12Pressure ratio C

Com

pres

sion

pro

cess

wor

k

Isentropic compressor work

Compressor work

Compressor - piston work

Compressor - piston work – closed cycle gain

a

b

c

d

Fig. 9: Compressor work versus C

0

1

2

3

4

5

6

7

8

9

0 1 2 3 4V/VD

p[bar]

1

2

3

prec

intercooling

pac

TDC BDC

Compressors‘work

1Pistonwork

2

Millerloss

3

0

1

2

3

4

5

6

7

8

9

0 1 2 3 4V/VD

p[bar]

1

2

3

prec

intercooling

pac

TDC BDC

Compressors‘work

1Pistonwork

2

Millerloss

3

Figure 8: p –V diagram of the charging process

the loss of piston work due to the early IVC typical of Miller. Starting from this idealised process, in figure 9 the different work contributions have been plotted versus the pressure ratio. The first curve (a) represents isen-tropic compressor work; the second (b) has been derived by applying suitable compressor maps and adapted intercooler temperatures accord-ing to water condensation. It can be noted that due to intercooling, the two curves are approximately parallel, i.e. the efficiency of the equivalent single stage compressor increases with the overall pressure ratio and the additional power required for a pressure increase is rather low in the up-per range. In order to calculate the net power required for the charge air, the gain in piston work (c) must still be taken into account (d), in addition to the gain in closed cycle work due to the reduction of tac. It can be noted that the curve d is very flat at its start. The plotted curve, obtained by means of an analytical approach, shows a flat minimum at a pressure ratio below 6. An approach by means of engine cycle simulation shows a monotone course for curve d, i.e. the minimum occurs at pressure ratio 4, as defined by the

C requirement.

2.6 The Scavenging Process

So far only the exhaust and charging processes for a non-scavenged en-gine have been considered. Consequently the exhaust gas temperatures are very high. In reality, some scavenging is always present in large en-gines. The scavenging process requires a positive pressure difference over the cylinder and leads to some efficiency reduction, but also has several advantages. A closer look into this process is thus necessary. Even with-out valve overlap there is some additional trapped air in the cylinder thanks to the positive pressure difference. This additional air is due to the

-20 0 20 40 60 80 100Valve overlap [°CA]

Mas

s flo

w

Mass flow compressorsMass flow turbines

Sca

veng

edga

s

Sca

veng

edai

rG

as fr

omdi

spla

cem

ent

Trap

ped

airc

l. vo

lum

eTr

appe

dai

rdis

plac

emen

t

-20 0 20 40 60 80 100Valve overlap [°CA]

Mas

s flo

w

Mass flow compressorsMass flow turbines

Sca

veng

edga

s

Sca

veng

edai

rG

as fr

omdi

spla

cem

ent

Trap

ped

airc

l. vo

lum

eTr

appe

dai

rdis

plac

emen

t

a) Mass balance

-20 0 20 40 60 80 100Valve overlap [°CA]

Pow

er

200

300

400

500

600

700

800

tTI

[°C]

Engine powerTurbines' powerCompressors' powerExhaust gas temperature

-20 0 20 40 60 80 100Valve overlap [°CA]

Pow

er

200

300

400

500

600

700

800

tTI

[°C]

Engine powerTurbines' powerCompressors' powerExhaust gas temperature

b) Power balance and tTI

Figure 10: Scavenging process

fact that the residual gas in the clearance volume is compressed from pTI to pRec, leaving some volume free for fresh air, which could be considered as some kind of scavenging air. Starting from this mixture of residual gas and air, the influence of increasing valve overlap is analysed. A set of simulations has been performed with constant pressures before and after the cylinder and with increasing valve overlap. The first point is simulated for reference without pressure difference over the cylinders; the other points have a fixed pressure difference of 2 bar, without considering turbocharging energy balance. A zero overlap is usually connected to high flow losses, due to the reduction in valve flow area before and after TDC. To avoid this, the points of maximum valve lift have been fixed and the overlap has been changed, with corresponding variation of the opening and closing ramps. The scavenging mass balance is shown in figure 10a. After the step with zero overlap, caused by the reduction of pressure in the exhaust manifold, the mass of air introduced to the clearance volume is progressively increased. In the simulated cases up to about 40°CA the corresponding volume of residual gas is scavenged, but no significant mass of air should be present on the exhaust side. The global scavenging factor, defined as the ratio of the total air flow to the amount of trapped air is always 1, but with the overlap of 40°CA, almost 50% of the residual gas is already scavenged. Further increasing valve overlap produces bet-ter scavenging but the optimum range for efficiency seems to be between 50 and 60°CA. This is also shown by the curves in figure 10b, where en-gine and turbine power are slightly increased in parallel with the valve overlap. The compressor power curve has a comparable slope up to about 50°CA, after which compressor power increases more steeply, impairing system efficiency. Under the chosen conditions the exhaust gas tempera-ture reaches 500°C at an overlap of 66°CA. Going to even lower tempera-tures is not beneficial for system efficiency.

3 Applications

3.1 Medium-speed en-gine

A conventional medium-speed en-gine must be able to burn heavy fuel (HFO). This requires maintain-ing a certain level of air to fuel ra-tio, scavenging and exhaust gas temperature. Taking as a reference a single-stage turbocharged model engine fulfilling IMO Tier I, a pres-sure ratio variation has been per-formed applying two stage turbo-

-0.08

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.06

4 6 8 10 12Pressure ratio C*

[-]

Th. gas exchange work LW

Gain closed cycle HD

Flow losses Flow

Miller losses Miller

Engine efficiency Eff

Reference engine-0.08

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.06

4 6 8 10 12Pressure ratio C*

[-]

Th. gas exchange work LW

Gain closed cycle HD

Flow losses Flow

Miller losses Miller

Engine efficiency Eff

Reference engine

Figure 11: Efficiency contributions for the reference engine

charging. In these simulations following conditions have been applied: IVC has been adapted to for a constant air fuel ratio The start of injection has been adjusted for constant firing pressure Valve overlap has been adapted for constant exhaust gas tempera-

ture The temperatures of the intercooler and aftercooler have been

adapted for a constant margin against water condensation Turbocharging efficiency remains approximately constant

The efficiency variations as defined in 2.1-2.2 are plotted in figure 11. Un-der the given boundary conditions the curve for the possible gain in en-gine efficiency shows an optimum at pressure ratio of around 7. This is valid for the full load operation point with respect to achieving the maxi-mum possible engine efficiency. Additionally, it has been observed that this configuration gives a substantial reduction in the NOx emissions, while exhaust gas temperature remains moderate. Engine efficiency is about 6% (equal to 3 points efficiency in figure 11) better than the refer-ence engine.

3.2 Pulse turbocharging

The simulations above have been performed with a compact undivided exhaust manifold, which is slightly better than a constant pressure mani-fold. For a potential evaluation a 3-pulse turbocharged engine model has been built with very small manifold volumes and without any penalty in the turbine performance, which should be expected with the unequal ad-mission typical of pulse operation. Taking the matching for pressure ratio 8 a first set of simulations was performed on the propeller curve with constant valve timing. The re-sults (figure 12) show that the pulse turbocharged engine per-forms slightly better at part load, but for both cases, pulse and con-stant pressure, the air to fuel ratio goes to unacceptable values. Ap-plying variable Miller the air to fuel ratio is much better, but the con-tribution given by the pulse turbo-charging is very small. The conclusions of these evalua-tions are:

Pulse turbocharging seems to be not a valuable alternative to variable valve timing to solve the part load weakness of an engine with extreme Miller tim-ing.

1.5

2

2.5

3

3.5

4

4.5

0 20 40 60 80 100bmep [bar]

lam

bdaV

mea

n [-]

300

350

400

450

500

550

600

TTI [

°C]

90

100

110

120

0 20 40 60 80 100bmep [bar]

bsfc

[%]

Brake spec. fuel consumption Brake spec. fuel consumption

Brake spec. fuel consumption Brake spec. fuel consumption

Turbine inlet gas temperature

Air / fuel ratio

Specific fuel consumption

C

tTI[°C]

Engine load [%]

Const. Pressure w/o VVT

Const. Pressure with VVT

Pulse w/o VVT

Pulse with VVT

be[%]

1.5

2

2.5

3

3.5

4

4.5

0 20 40 60 80 100bmep [bar]

lam

bdaV

mea

n [-]

300

350

400

450

500

550

600

TTI [

°C]

90

100

110

120

0 20 40 60 80 100bmep [bar]

bsfc

[%]

Brake spec. fuel consumption Brake spec. fuel consumption

Brake spec. fuel consumption Brake spec. fuel consumption

Turbine inlet gas temperature

Air / fuel ratio

Specific fuel consumption

C

tTI[°C]

Engine load [%]

Const. Pressure w/o VVT

Const. Pressure with VVT

Pulse w/o VVT

Pulse with VVT

be[%]

Figure 12: Propeller operation with and without variable valve timing

When variable valve timing is available, the engine performance is excellent and the improvement with pulse turbocharging becomes only marginal.

No benefit could be identified, which would justify the effort to de-velop a pulse turbine for two stage turbocharging.

3.3 Turbocompounding

The diagrams in the figures 6 and 9 show a maximum for the turbines’ power at expansion ratio around 7 and a minimum for the compres-sors’ power at much lower pressure ratio. This would result in the pos-sibility to build a turbocompound machine where the turbine makes more than 20% of the system power. In order to verify the potential of such a machine complete cycle simulations have been performed on the model engine described in 3.1. The results are plotted as sys-tem specific fuel consumption over pressure ratio in figure 13, in compari-son with the engine alone. The upper bound of the area is valid for a con-figuration which preserves a positive pressure difference over the cylinder; the lower bound is referred to the maximum efficiency without this con-straint. The fuel input has been kept constant for every pressure ratio. With this condition the engine power is reduced but the system power is finally increased. Since the turbine power occurs at very high speed, for practical use either a gear box or a high-speed generator are required. This has been taken into account with a transmission efficiency of 90%. The efficiency curves for the turbocompound engine show an optimum range at lower pressure than for the engine alone. Increasing the pressure ratio from 4 to 7 the additional turbine power is reduced from 20 to 10%, and the optimum exhaust gas temperature from 620 to 570°C. A turbocompound engine would have an interesting potential but also many drawbacks:

Additional costs, complexity, technical risk. Higher exhaust gas temperatures. Lower Miller effect, with negative consequences on NOx-emissions,

knock margin by gas engines, thermal load.

Specific fuel consumption

88

90

92

94

96

98

100

102

4 5 6 7 8 9 10

Charge air pressure ratio

be [%] 1-stage reference

2-stage turbocharging

2-stage turbocharging+Power turbine / PTO

Specific fuel consumption

88

90

92

94

96

98

100

102

4 5 6 7 8 9 10

Charge air pressure ratio

be [%] 1-stage reference

2-stage turbocharging

2-stage turbocharging+Power turbine / PTO

Figure 13: Potential of turbocom-pounding.

3.4 High speed engine

It is well known that engines with extreme Miller timings require some kind of variability in terms of IVC. A solution is offered by ABB under the name VCM (Valve Con-trol Management) [10]. Medium-speed engines are today mainly used in constant speed applications or in a narrow speed range, e.g. baseload power generation and controllable (CPP) or fixed pitch propeller (FPP) operation. In all cases the operating points with maximum power and torque are met at nominal and maximum rated speeds. For these applica-tions variable valve timing is nec-essary for starting and loading the engine, for improving part load be-haviour in FPP applications and en-hancing the application range. High speed engines cover a wider speed range and the point with maxi-mum torque is not necessarily that with the highest speed. The potential of two stage turbocharging for this kind of application has been studied by means of simulations. As the engine operating curve an FPP curve with 10% torque rise from pme = 22.5 bar at 1900 rpm to pme = 25 bar at 1500 rpm has been considered (figure 14). With single stage turbocharg-ing (boost pressure 4.7 bar) this operating curve is extremely challenging without additional control possibilities. At part load, the maximum exhaust gas temperature is about 720°C with a very low air/fuel ratio ( C = 1.3). With two stage turbocharging (boost pressure 9.2 bar), Miller timing, and variable valve timing, this very demanding curve can be operated with a much lower exhaust gas temperature and with a higher air/fuel ratio. In addition, fuel consumption is considerably reduced and transient behav-iour is highly likely to be better.

4 Requirements on the Turbocharging System

Looking at the design of the two stage turbocharging system for 4-stroke engines under the aspect of fuel efficiency only, the target pressure ratio might be around 8. If we take into account the trade-off between effi-ciency and emissions and the requirement of improving power density and altitude capability, a single figure compression ratio would not be the right

05

1015202530

80

90

100

110

120

130

140

800 1000 1200 1400 1600 1800 2000bmep [bar]

bsfc

[%]

-100

-50

0

50

100

150

200

NO

x [%

]

bsfc42.7

bsfc42.7_2st

1

1.5

2

2.5

3

3.5

4

400

450

500

550

600

650

700

Mean effective pressure

pme[bar]

Turbine inlet gas temperature

tTI[°C]

CAir equivalence ratio

NOx-Emission

NOx[%]

Engine speed [rpm]1-stage no Miller2-stage – Miller – Variable valve timing

Specific fuel consumption

bsfc[%]

05

1015202530

80

90

100

110

120

130

140

800 1000 1200 1400 1600 1800 2000bmep [bar]

bsfc

[%]

-100

-50

0

50

100

150

200

NO

x [%

]

bsfc42.7

bsfc42.7_2st

1

1.5

2

2.5

3

3.5

4

400

450

500

550

600

650

700

Mean effective pressure

pme[bar]

Turbine inlet gas temperature

tTI[°C]

CAir equivalence ratio

NOx-Emission

NOx[%]

Engine speed [rpm]1-stage no Miller2-stage – Miller – Variable valve timing

Specific fuel consumption

bsfc[%]

Figure 14: Improved torque capacity with two stage turbocharging

solution. There will be a wide range of pressure ratios in the future, even with values above 10.

4.1 Pressure Ratio Dis-tribution

Even if only total pressure ratio has been mentioned up to now, this still must be broken down into sin-gle pressure ratios for each stage. It is well known that turbocharging efficiency defined according to [11] is a function of the pressure distribu-tion between the low pressure stage C,LP and the high pressure stage C,HP The ratio C,LP/ C,HP which gives the maximum efficiency is a function of the intercooling temperature. It is convenient for reason of size to choose a ratio close to two. In this case, the loss in global efficiency is marginal but the LP turbine area is smaller (figure 15). By reducing engine load the pressure distribution is changed. The above mentioned diagram is valid for full load operation but the choice of the pressure distribution at full load also has an influence on its evolution over the engine load profile. It has been observed that the pressure ratio of the

0 1 2 3 4C,LP/ C,HP

T SeffT

Eta-TurbochargingSeff-High Pressure turbineSeff-Low Pressure turbine

Figure 15: Pressure repartition

0

0.5

1

1.5

2

2.5

1 2 3 4 5 6Expansion ratio T

Red

uced

mas

s flo

w

2-stage_4.4-1.72-stage_3.4-2.2

LP turbine mred=f( T,LP) LP turbine mred=f( T,overall)

HP turbine mred=f( T,HP) HP turbine mred=f( T,overall)

TI

TITIred p

Tmm

C,LP / C,HP = 2.6

C,LP / C,HP = 1.55 kPaK

skg

0

0.5

1

1.5

2

2.5

1 2 3 4 5 6Expansion ratio T

Red

uced

mas

s flo

w

2-stage_4.4-1.72-stage_3.4-2.2

LP turbine mred=f( T,LP) LP turbine mred=f( T,overall)

HP turbine mred=f( T,HP) HP turbine mred=f( T,overall)

TI

TITIred p

Tmm

C,LP / C,HP = 2.6

C,LP / C,HP = 1.55 kPaK

skg

Figure 16: Flow characteristics

HP turbocharger remains fairly constant in the upper load range. Within this range, the change of overall pressure ratio is determined by the low pressure stage only and the HP stage acts as a constant multiplier. Ac-cording to simulations, for part load operation it is convenient to extend this constant multiplier range as far as possible. This can be achieved by increasing the ratio C,LP/ C,HP. An explanation for this behaviour can be found in figure 16. With a con-stant overall pressure ratio, this diagram shows the behaviour of the re-duced turbine mass flow from both stages, plotted over the stage expan-sion ratio and the overall expansion ratio. The curves are similar to the flow characteristic of a nozzle. The reduced mass flow increases with the expansion ratio up to a maximum defined by the critical expansion ratio, and then remains approximately constant. In this case the critical expan-sion ratio is defined with reference to 98% of the maximum flow. The crit-ical expansion ratio for the whole system is about 2.8. When the LP stage has also reached its critical pressure ratio, the expansion ratio of the HP turbine can no longer be changed, even though it is well below its critical value. This condition occurs in figure 16 at an expansion ratio of about 2.25 for the low pressure stage, which corresponds to an overall expan-sion ratio between 3 and 4, depending on pressure distribution at full load. The higher the ratio C,LP/ C,HP, the earlier the low pressure stage attains its critical expansion ratio and the longer the C,HP remains constant, lead-ing to higher efficiency at part load.

4.2 Turbocharging Effi-ciency

Turbocharging efficiency has al-ways been important for achieving high pressure ratios at convenient exhaust gas temperatures. Two-stage turbocharging with intercool-ing ensures a substantial improve-ment, but the achievable efficiency is even more important than be-fore. This is illustrated in fig. 17, where the achievable pressure dif-ference over the cylinders is plot-ted versus the total pressure ratio for different turbocharging efficien-cies and exhaust gas temperatures. This pressure difference, which is responsible for the theoretical gas exchange work, represents the most important contribution to engine effi-ciency (see figure 11). The line of the Miller losses is plotted in the same diagram. An engine optimised for efficiency only should be matched well below the point where the curve of the Miller losses intersects the p

0.0

1.0

2.0

3.0

4.0

3 4 5 6 7 8 9 10 11 12

Pressure ratio C

pRec - pTI

[bar]

tTI = 500 °C, eta = 75%

tTI = 500 °C, eta = 70%

T = 75%

T = 70%tTI = 600 °C

tTI = 550 °C

tTI = 500 °C

pmi, Miller

0.0

1.0

2.0

3.0

4.0

3 4 5 6 7 8 9 10 11 12

Pressure ratio C

pRec - pTI

[bar]

tTI = 500 °C, eta = 75%

tTI = 500 °C, eta = 70%

T = 75%

T = 70%tTI = 600 °C

tTI = 550 °C

tTI = 500 °C

pmi, Miller

Figure 17: Pressure difference over cylinder and Miller loss.

curve. Taking the pressure loss over the cylinder and the Miller losses into account at the same time the maximum gain from the gas exchange proc-ess is located on the left of the Miller losses’ curve. To achieve high pres-sure ratios it is in any case mandatory to have high turbocharger efficien-cies and exhaust gas temperatures as high as possible within the limita-tion given by the engine design and operation.

4.3 Compressor development

So far only the pressure ratio and efficiency of two stage turbocharg-ing have been discussed. Flow ca-pacity plays an important role in the physical dimensions of the two stage turbocharging system. The benefits of two stage turbocharging can be accessed in a profitable way by using components with high specific capacity. Fig. 18 shows the evolution of ABB/BBC compressor stages for a constant wheel diame-ter in terms of pressure ratio and flow capacity. It can be seen that the HP compressor stage is designed for a high capacity at moderate pressure ratio as well as ensuring a wide compressor map (figure 19).

4.4 Availability

First applications of the new ABB product known as Power2 two stage tur-

Progress in Compressor Performance at Full LoadAluminium Compressor - Exchange interval 50'000 h

1

2

3

4

5

6

V298 [m3/s]

C2003

19961996

2003

2004

1992

1989

1978

1970

1946

1954-1964

1924

20092009

1983

2010

1-stage compressors

LP-compressor

HP-compressor

Figure 18: Compressor development

1.0

2.0

3.0

C tot/tot

*sV

.V298 [m3

/s]

1.0

2.0

3.0

.V298 [m3

/s]

C tot/tot

*sV

5

6

7

8

9

1000 2000 3000 4000 5000 6000 7000 8000

Turbocharged power per Power2 unit

Pre

ssur

e ra

tio

Size 1 Size 3Size 2

Figure 19: HP Compressor map vs. standard compressor map (green ar-

eas: same efficiency)

Figure 20: Power 2 applica-tion ranges

bocharging have been documented in [1], [2]. Two sizes of the first gen-eration Power2 are available; a third size covering the power range from 2500 to 4000 kW will be launched according to market needs (fig. 20).

5 Application and Potential of Power2 in Differ-ent Engine Segments

Most simulations have been performed for a representative engine, and the results have been applied to different engine sizes. In order to achieve a better understanding of the potential of two stage turbocharging, an ex-haustive series of simulations have been performed for different diesel engine models:

Model A – Large medium-speed engine Model B – Small medium-speed engine Model C – High-speed engine

The results are summarized in the trade-off diagram in figure 21. All engine models have been normalised to constant values of mean effec-tive pressure, compression pressure and firing pressure. The cam profiles are also normalised, but scaled for providing reasonable acceleration at the different engine speeds. The reference points are valid for single stage turbocharging. The most important differences derive from the cylinder dimensions (bore and stroke) and rotational speed. As can be expected, fuel consumption increases from A to C due to size effects (combustion, heat losses, mechanical losses, turbocharging efficiency). The turbocharg-ing systems and valve timings have been matched for air/fuel ratios and exhaust gas temperatures that are typical for medium-speed engines in HFO applications (A and B) and for high-speed engine running on distillate oil (C). Additional remarks:

Model A. Engine and turbocharging efficiency are high with moder-ate scavenging.

Model B. Due to the lower efficiency a higher scavenging ratio is re-quired for attaining the same exhaust gas tem-perature as in A. This engine model has a lar-ger stroke to bore ratio compared to the others. The comparatively lower speed has a positive ef-fect on the efficiency but increases NOx-emissions.

0.18

0.19

0.20

0.21

0.22

0 2 4 6 8 10 12 14

NOx in g/kWh

bsfc

in k

g/kW

h

Model A

Model B

Model C

IMO Tier II limits

IMO

Tie

r III

EC

A li

mits

NOx emissions

Brak

e sp

ecifi

cfu

elco

nsum

ptio

n

0.18

0.19

0.20

0.21

0.22

0 2 4 6 8 10 12 14

NOx in g/kWh

bsfc

in k

g/kW

h

Model A

Model B

Model C

IMO Tier II limits

IMO

Tie

r III

EC

A li

mits

NOx emissions

Brak

e sp

ecifi

cfu

elco

nsum

ptio

n

Figure 21: Trade-off efficiency emissions

Model C. This model has a very small scavenging ratio and high ex-haust gas temperatures. It has the highest specific fuel consumption and the lowest NOx-emissions.

The shaded areas represent the results obtained by applying two stage turbocharging and Miller timing under the following boundary conditions:

Pressure ratios ranging from 6 to 12.5 Constant air/fuel ratio and exhaust gas temperatures by adjusting

Miller valve timing and valve overlap Constant compression and firing pressure by adjusting injection tim-

ing and compression ratio Variation in turbocharging efficiency.

For further evaluation a slightly higher exhaust gas temperature has also been considered for models A and B. The diagram shows that all the engines have a potential for a simultane-ous reduction of specific fuel consumption and NOx-emissions. The differ-ence between the 2 NOx-emission levels is approximately constant for all models, while model C can profit from the largest relative reduction. Under the given boundary conditions, the pressure ratio related to highest efficiency is about 7 for the medium-speed engines, whereas for the high-speed engine it amounts to values above 8. This is due to the higher ex-haust gas temperature in accordance with figure 17. The condition of constant compression and firing pressures keeps the ar-eas in a relatively narrow range. Increasing the ratio of firing pressure to compression pressure (not shown in the diagram) could lead to lower fuel consumption within the IMO Tier II NOx limits. Reducing the ratio by means of retarded injection (additional lines in figure 21) could be one possibility for achieving the lowest possible NOx-emissions. These lines give an indication of the possibility of reaching the low emission limits an-nounced in IMO tier III and EPA Tier 4 by using only Miller timing and late injection. According to the applied simulation models, it seems that for the medium-speed engines and especially for engine B, the limits could only be approached in connection with a very high penalty in fuel consumption. With engine C the simulations show that very low emissions levels could be achieved. These considerations are limited to the nominal operation points at full engine load. At part load, the effect of Miller on emissions is substantially reduced, calling for additional emission reduction technolo-gies. But the application of the proposed technology allows for substantial reductions in raw engine emissions and, consequently, reductions in the capacity required from the additional technologies. Emission reduction technologies, typically selective catalytic reduction (SCR) and exhaust gas recirculation (EGR) are considered and have inter-actions with the turbocharging system. These interactions have been stud-ied [8] and first experiments with the application are on-going [9].

6 Conclusions

A theoretical investigation on the turbocharged engine process has indi-cated that two stage turbocharging, extreme Miller timing, and variable inlet valve closure represents a very effective solution for engines with high power density. It is a proven measure for satisfying the requirements of high fuel efficiency in connection with low NOx emissions and high op-erational flexibility. With Power2 and VCM, ABB offers two of the core components for achiev-ing these aims. Power2 is the product name of ABB’s two stage turbo-charging concept, consisting at its minimum of an LP- and an HP-stage. The LP-stage is specifically designed for operation with high efficiency and high specific flow capacity. The HP-stage is specifically designed for opera-tion at moderate pressure ratios, high volume flow and high absolute pressure with enhanced map width. VCM (Valve Control Management) is offered as ABB’s solution for variable valve timing. It is an electro-hydraulic system offering wide flexibility for control of valve timing as well as valve lift. Additionally, the potential of this engine concept has been studied for dif-ferent sizes of 4-stroke diesel engines. It has been shown that a range of engines can be positioned in a global trade-off diagram. Large medium-speed engine offer very good values of specific fuel consumption but high NOx-emissions whereas high-speed engines give higher fuel consumption and lower emissions. After a discussion of the specific characteristics of various engines, the feasibility of the concept for application on a high-speed engine featuring high torque requirements at variable speed has been evaluated.

NOMENCLATURE

p Pressure [Pa, bar] pc Compression pressure [bar] pmax Firing, maximum pressure [bar] pme Brake mean effective pressure [bar] pmi Indicated mean effective pressure [bar] Q Fuel energy input [J/cycle] T, t Temperature [K, °C] V Volume [m3] Vd Displacement [m3] V298 Reduced volume flow [m3/s] Compression ratio Efficiency Ratio of specific heatsC Mass ratio of trapped to stoichiometric air Pressure ratio

Subscripts ac Start of compression ex End of expansion HD Closed (high pressure) cycle LWth theoretical gas exchange Rec charge air receiver red Reduced TI Turbine inlet Abbreviations BDC Bottom dead centre C Compressor CB Constant boundary CP Constant pressure EVO Exhaust valve open HP High pressure IVC Inlet valve close LP Low pressure T Turbine TDC Top dead centre VOL Volume

References

[1] Raikio, T., B. Hallbäck & A. Hjort, 2010, Design and first application of a 2-stage turbocharging system for a medium-speed diesel en-gine, 26th CIMAC World Congress in Bergen (N)

[2] Haidn, M., J. Klausner, J. Lang & Ch. Trapp, 2010, Zweistufige Hoch-druck-Turboaufladung für Gasmotoren mit hohem Wirkungsgrad, 15. Aufladetechnische Konferenz, Dresden (D).

[3] Codan, E. & Mathey, Ch., 2007, Emissions – a new challenge for tur-bocharging, 25th CIMAC World Congress in Vienna, Austria

[4] Codan, E., Mathey, Ch. & Vögeli, S., 2009, Applications and Poten-tials of 2-stage Turbocharging, 14. Aufladetechnische Konferenz, Dresden (D).

[5] Codan, E., Mathey, Ch. & Rettig, A., 2010, 2-Stage Turbocharging – Flexibility for Engine Optimisation, 26th CIMAC World Congress in Bergen (N)

[6] Millo, F., Gianoglio, M. & Delneri, D., 2010, Combining dual stage turbocharging with extreme Miller timings to achieve NOx emissions reductions in marine diesel engines, 26th CIMAC World Congress in Bergen (N)

[7] Watson, N. & Janota, M.S., 1982, Turbocharging the internal com-bustion engine, MacMillan Press Ltd.

[8] Codan, E., S. Bernasconi, & H. Born, 2010, IMO III Emission Regula-tion: Impact on the Turbocharging System, 26th CIMAC World Con-gress in Bergen (N)

[9] Ruschmeyer, K, Rickert, C. & Schlemmer-Kelling, U., 2011, Potential des Caterpillar MaK 6 M32 C mit zweistufiger Abgasturboaufladung, 16. Aufladetechnische Konferenz, Dresden (D).

[10] Mathey, Ch., 2010, Variable Valve Timing – A necessity for future large diesel and gas engines, 26th CIMAC World Congress in Bergen (N)

[11] CIMAC, 2007, Turbocharging Efficiencies – Definitions and guidelines for measurement and calculation, Recommendation Nr. 27, Conseil International des Machines à Combustion, Frankfurt am Mein (D), (www.cimac.com)

[12] Codan, E. & Huber, T., 2012, Application of two stage turbocharging systems on large engines, 10th International Conference on Turbo-chargers and Turbocharging, London (UK)