Design and experiments of two-stage intercooled electrically assisted [626953]

Design and experiments of two-stage intercooled electrically assisted
turbocharger
Aki Grönman⇑, Petri Sallinen, Juha Honkatukia, Jari Backman, Antti Uusitalo
Lappeenranta University of Technology, School of Energy Systems, P.O. Box 20, 53851 Lappeenranta, Finland
article info
Article history:
Received 5 October 2015
Accepted 23 December 2015
Available online 5 January 2016
Keywords:
TurbochargerElectrically assistedDiesel engineabstract
The demand for continually improving the efficiency of large diesel engines requires increasing the pres-
sure ratio and efficiency of turbochargers. One answer to these requirements is to use two-stage electri-cally assisted turbochargers, which will enable higher pressure ratios than before but also generate
electricity in sufficient operating conditions. This study shows the design and experimental performance
of the novel two-stage single-shaft electrically assisted turbocharger. It was predicted that an increase of2.5 percentage points in overall efficiency can be achieved by using a two-stage single-shaft intercooled
electrically assisted turbocharger in a 1200 kW diesel engine instead of a conventional single-stage tur-
bocharger. It is experimentally shown that the studied turbocharger has potential for improving theengine efficiency although the performance was below the expected. Additionally, the critical points
for the design improvements are assessed to guide future development work.
/C2112015 Elsevier Ltd. All rights reserved.
1. Introduction
More energy efficient and less polluting processes are required
in the energy generation sector. Diesel engines are widely used in
generating electricity in land, marine and transportation. There-
fore, improvements in the diesel engine energy efficiency will have
a direct influence on the variety of applications. The constant
increase of mean effective pressure (MEP) within large two-
stroke diesel engines requires higher turbocharger pressure ratios
but also improved efficiency [1]. Simultaneous NO xemission
reductions are also needed to fulfill the TIER III goals [2], where
the use of Miller timing offers one solution but requires higherboost pressure from the turbocharger. On the other hand, the
reduction of the CO
2emissions is of major concern in every energy
generation sector. To meet these constraints, the design of future
engines strives for better utilization of the engine waste heat, but
also substantial improvements in the turbocharger performance
are required.
Several studies have shown that marked improvements can be
expected on the overall engine fuel consumption and efficiency
while using a separate power cycle to recover the engine waste
heat. A good approximation for the recoverable potential was given
by Baldi and Gabrielii [3], who suggested that 5–15% fuel savings
are realistic in the studied ship. Multiple methods for convertingwaste heat into electricity have been shown to have potential. Sev-
eral authors have shown the Organic Rankine Cycle (ORC) to be a
good alternative in converting engine waste heat into electricity
[4–7] . Additionally, the potential of the Kalina cycle [7], the con-
ventional Rankine cycle [7–9] or the combination of the conven-
tional Rankine cycle and ORC [10] have been shown. In addition
to waste heat recovery by means of Rankine cycles, the use of ther-
moelectric generators that are capable of converting heat directly
into electricity by using the Seebeck effect have been identified
as another potential technology for recovering waste heat from
engines [11–13] .
To increase the pressure ratio of the turbocharger, a two-stage
compressor has been recommended if the pressure ratio exceeds
4.5[1]in large marine diesel engines and approximately 3.5 in
road use [14]. In currently available designs, this is done by using
separate low pressure (LP) and high pressure (HP) turbochargers.
Cui et al. [15] have carried out a theoretical optimization of a
two-stage turbocharger for a marine diesel engine with Miller-
timing for reducing NO xemissions. They concluded that two-
stage turbocharging should be applied in this type of engine, since
single-stage turbocharging cannot reach the required pressure and
efficiency level to avoid the power reduction of the engine caused
by Miller timing. Wik et al. [16] highlight that the state-of-the-art
simulation models can be used for predicting the engine
performance with different types of turbocharging configurations,
but experimental work has to be carried out especially when
evaluating the NO xemissions in engines adopting two-stage
http://dx.doi.org/10.1016/j.enconman.2015.12.055
0196-8904/ /C2112015 Elsevier Ltd. All rights reserved.⇑Corresponding author.
E-mail address: gronman@lut.fi (A. Grönman).Energy Conversion and Management 111 (2016) 115–124
Contents lists available at ScienceDirect
Energy Conversion and Management
journal homepage: www.elsevier.com/locate/enconman

turbocharging. Codan and Huber [17] have concluded that the use
of two-stage turbocharging, together with extreme Miller timing
represents an effective way to reach high fuel efficiency, low NO x
emissions, high operational flexibility, and high power density.Serrano et al. [18] have carried out an experimental study for a
low NO
xemission two-stage turbocharged heavy-duty engine.
Their results show that two-stage turbocharging can significantly
improve drivability and reduce NO xemissions especially in tran-
sient operation when compared to single-stage turbochargers.
Galindo et al. [19] have studied the impact of different two-stage
turbocharging architectures, especially on pumping losses in auto-
motive engines. Their study generated equations for describing the
relation between the total compression and expansion ratio of a
two-stage turbocharger as a function of engine parameters. They
compared the use of a single-stage and two-stage turbocharger
for different pressure levels and their results showed that signifi-
cant efficiency improvements can be achieved with a two-stage
configuration, especially when the pressure ratio of the turbo-
charger is high. The two spool design, however, requires more
space than a design where LP and HP compressors and turbines
are on the same shaft. Space limitation is an important design con-
straint especially for diesel engines in marine use.
Electrically assisted turbochargers have an electric motor/gen-
erator connected to a turbocharger to assist the turbocharger orconvert excess energy to usable form. The general idea is to speed
up the turbocharger acceleration during startup and to produce
electricity when the engine runs at higher loads. Ono et al. [20]
reported successful operation of an electrically assisted large tur-
bocharger, which was suggested to be the world’s first practical
application of a marine hybrid turbocharger. Terdich et al. [21]
showed the applicability of electric turbocharger assist on a
200 kW diesel engine. Terdich and Martinez-Botas [22] also pro-
vided an experimental characterization of the turbine and electric
motor/generator of an electrically assisted turbocharger designed
to be applied to non-road medium duty diesel engines. A study
by Millo et al. [23] demonstrated 1–6% reductions in fuel consump-
tion while employing an electrically assisted turbocharger on a
heavy-duty diesel engine.
From this background, it can be concluded that very little
information is available in public literature about the use of electri-
cally assisted turbochargers, and there are no reported applications
where the two-stage compression would be made in a single-shaftturbocharger. There is also a clear lack of detailed design informa-
tion about the complex system and discussion on the practical
design constraints of the electrically assisted turbocharger. This
study attempts to fill this gap of information as it presents the
design and experimental performance of a novel two-stage
single-shaft electrically assisted turbocharger that is designed
and tested with a 1200 kW diesel engine in real operating condi-
tions. From the turbocharger design point of view, special attention
is paid to the mechanical design, which is usually the main limita-
tion of the whole process from idea to prototype.
The preliminary design of a two-stage turbocharger is pre-
sented first to show the advantages of the current approach in
comparison with the conventional single-stage turbocharger. This
is followed by a section discussing the design of the turbomachin-
ery and the electric machine. Also the mechanical design is pre-
sented in separate chapters including stress, bearing, and rotor
dynamic analyses. Finally, the experimental setup and the results
are presented and analyzed.
2. Preliminary simulations on two-stage turbocharger
The simulations were carried out for a 1200 kW impulse
charged diesel engine with a conventional single-stage tur-
bocharger and a two-stage intercooled electrically assisted tur-bocharger (more information about the diesel engine in
Section 5). The schematic diagrams of the two studied cases are
shown in Figs. 1 and2. The conventional single-stage turbocharger
does not contain an electric machine, and the turbine runs the
compressor that is directly coupled on the same shaft. In the
two-stage intercooled electrically assisted turbocharger, the tur-
bine, the LP and HP compressor, as well as the electric machine
are directly coupled on the same shaft, and the electric machine
is connected to the electric network via a frequency converter.
The idea is to use the electric machine as a motor to speed up
the turbocharger acceleration during the diesel engine startup
and as a generator to produce electricity when the diesel engine
runs at higher loads.
First, the turbine as well as the LP and HP compressor stages
were preliminarily designed, including the off-design calculations,
to model the performance of the turbocharger. Then, simulations
were conducted for comparing the performance of the diesel
engine–turbocharger system when the conventional single-stageNomenclature
b blade height (mm)
D diameter (mm)
Dh enthalpy change (J/kg)
M Mach number (–)
N number of blades (–)
N rotational speed (rpm, 1/s)
Ns specific speed (–)
P power (kW)
qv volume flow (m3/s)
R degree of reaction (–)
R2coefficient of determination (–)
p pressure ratio (–)
w blade loading coefficient (–)
r von Mises stress (MPa)
Subscripts
amb ambientg gauge
h hubin inlet
out outlet
s isentropic
t tipt-s total-to-statict-t total-to-total
1 compressor inlet
2 compressor rotor/turbine stator outlet
Abbreviations
BMEP brake mean effective pressure
BSFC brake specific fuel consumptionCFD computational fluid dynamicsHP high pressure
LP low pressure
MEP mean effective pressureORC organic Rankine cyclePM permanent magnet116 A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124

turbocharger was replaced by a new two-stage intercooled electri-
cally assisted turbocharger. The simulations were based on 1D
modeling of the diesel engine–turbocharger system by using
GT-Power together with Matlab/Simulink. The preliminary results
are presented in Table 1 and they were used as starting-point val-
ues for the final design of the new turbocharger. Compared to the
conventional turbocharger, the pressure ratio is significantly
higher in the new turbocharger, but the aim was to keep the engine
shaft power unchanged by means of a Miller timing. The brake
mean effective pressure (BMEP) is also unchanged in both cases.
Compared to the conventional turbocharger, the fuel power is1.4% higher with the new turbocharger producing an additional
electric power of c. 89 kW, and thus, the overall efficiency
increases from 41.7% (=1200.9/2882.9 kW) to 44.1% (=(1200.8
+ 89.2) kW/2923.5 kW) if the mechanical power and electrical
power are equally valued. Based on this calculation, the overall
efficiency would increase by c. 2.5 percentage points and relatively
by c. 6%. In the new turbocharger, the inlet air temperature of the
LP compressor is slightly higher than the inlet air temperature of
the compressor in the original turbocharger because in the new
turbocharger, the inlet air of the LP compressor is used for the cool-
ing of the electric machine as presented in Fig. 2 .
From the preliminary simulations, it can be concluded that the
design criteria for the new turbocharger were to design the tur-
bocharger components to reach the performance presented in
Table 1 . The main criterion for the new two-stage compressor
was to produce the targeted pressure ratio of 6.04, which, com-
bined with Miller timing, would reduce NO
xemissions. The turbine
should also produce enough power close to the nominal conditionsto allow the generator to produce electricity. Another design crite-
rion was that the electric machine should accelerate rapidly during
the startup.
3. Design of turbomachinery and electric generator
3.1. Aerodynamic design of the two-stage centrifugal compressor
The challenge with the centrifugal compressor design was the
high hub/tip ratio of the two compressor wheels as they were to
be assembled to the middle of the shaft of the turbocharger. The
design procedure needed several cycles developing from the initial
vaneless diffuser to a vaned diffuser in the first stage. The design of
the compressors was based on the in-house design codes regarding
the geometry design and one-dimensional efficiency prediction.
Additionally, numerical simulations were carried out for the first
stage with Navier–Stokes solver Finflo. The design of the volutes
was based on a variable cross-sectional area and a moderate coni-
cal diffuser connecting the volute and piping. The main design
parameters of the compressors are presented in Table 2 , which
shows a good agreement between the 1D and CFD simulations.
The specific speed Nsis defined as
Ns¼2pNffiffiffiffiffiffi ffiqv1p
Dh0:75
s: ð1Ț
Overall, it can be concluded that the predicted performances of
the compressors are very near or equal to the targeted efficiencies
of 81.1 and 79.7 per cent for the LP and HP compressors,
Fig. 1. Schematic diagram of the main flow of the diesel engine–turbocharger system provided with a conventional single-stage turbocharger.
Fig. 2. Schematic diagram of the main flow of the diesel engine–turbocharger system provided with a two-stage intercooled electrically assisted turbocharg er.A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124 117

respectively. The manufactured compressor impellers are shown
assembled to the shaft in Fig. 3 . The intercooler (shown in Fig. 5 )
is a crossflow type finned tube heat exchanger in which the cooling
water is on the tube side and the air to be cooled flows outside the
tubes.
3.2. Aerodynamic design of the turbine
Dividing the expansion into a two-stage turbine would have
allowed to keep the flows subsonic and also avoid losses associated
with the stator trailing edge shocks. A two-stage turbine would
have made the shaft and the whole turbocharger axially longer
and the rotor dynamics would have become more challenging in
terms of rotor stiffness. The degree of reaction was chosen to be
R= 0.15 in order to have the axial force relatively low for the axial
bearings. The absolute flow velocity after the stator was supersonic
due to a high pressure ratio with M2= 1.4, but the relative rotorinlet flow remained subsonic (see Table 2 and rotor in Fig. 3 ).
The latter makes the rotor design straightforward and also
improves the off-design performance in comparison with turbines
having supersonic rotor inlet conditions. With impulse charged
diesel engines, the importance of having a good part load perfor-
mance in the turbocharger turbine is especially highlighted due
to highly pulsating turbine inlet flow [24–26] . The CFD simulation
in design conditions predicted an isentropic total-to-total effi-
ciency of 81.2%, while also reasonably good performance in the
off-design was expected [27]. A preliminary design estimation by
using the loss correlation of Kacker and Okapuu [28] predicted
gs,t-t= 77.9 ± 1.5% for the turbine at design conditions, which is rea-
sonably close to the CFD prediction. Based on the predicted turbine
performance it can be expected that the targeted 79.4% turbine
efficiency can be reached in the design operating conditions.
3.3. Electric machine design
The electric machine of the high-speed turbocharger has a nom-
inal speed of n= 30,000 rpm and a shaft power of P= 130 kW. The
turbocharger integrated electric machine can operate as a motor or
generator depending on the power balance of the turbocharger
unit. The rotor has a carbon-fiber sleeve (see Fig. 3 ) as a retaining
material and an aluminum cage for shielding the four pole magnets
from eddy currents.
Because of the high power density of the generator, a thermal
analysis was also conducted. This analysis was essential since per-
manent magnets (PM) have a maximum temperature level which
cannot be exceeded. Failure in keeping the temperature below
the limit will make the permanent magnets lose their magnetic
properties. Additionally, a carbon-fiber sleeve has also poor ther-
mal conductivity, which makes the rotor cooling in the generatorair gap more demanding. The cooling was arranged by leading
the air through the electric machine before entering into the LP
compressor. A detailed description of the results of this analysis
is given in [29]. Also, more detailed stress calculations of the
carbon-fiber sleeve, air friction and cooling losses are discussed
in[30].
4. Mechanical design
The mechanical design included the stress analyses of turboma-
chinery impellers, bearing design and rotor dynamic analysis. It is
also worth mentioning, that the carbon fiber sleeve, compressor
and turbine wheels were connected with shrink fit to the shaft.
FEM analysis predicted the wheel of the LP compressor made of
aluminum to have maximum von Mises stress
r= 351 MPa at a
nominal 31,500 rpm speed, which was clearly below the yield limit
of 535 MPa. For the smaller HP compressor wheel made of alu-
minum, the predicted maximum stress was 234 MPa. The FEM
analysis of the turbine wheel predicted a maximum stress of
850 MPa, which was much lower than the 1600 MPa yield limit
of the used tempered steel. These results give a clear indication
that the compressor and turbine wheels can handle the associated
stresses.
4.1. Bearing design
Oil lubricated tilting pad journal bearings were chosen to be
used as radial bearings, since they have good stability characteris-
tics at high rotation speeds. The tilting pad bearing differs from the
fixed profile bearing in that each pad rotates about a pivot,
enabling each pad to have higher degrees of freedom correspond-
ing to movement about the pivot point. The pad tilts such that its
center of curvature moves to create a converging pad filmTable 1
Preliminary results of simulations for comparing the performance of the original andthe new diesel engine–turbocharger system.
Originalsystemwithconventionalsingle-stageturbochargerNew system with
two-stageintercooledelectricallyassistedturbocharger
Diesel engine
Diesel engine BMEP, bar 27.3 27.3
Diesel engine brake power, kW 1200.9 1200.8
Fuel flow rate, kg/h 243.1 246.5Fuel power, kW 2882.9 2923.5
Turbocharger
Rotational speed, rpm 47,736 29,998Frequency converter electric power
output, kW– 89.2
Compressor or compressors (
a)
Inlet pressure, bar 1.00 0.99Outlet pressure, bar 4.59 5.95Pressure ratio, – 4.60 6.04Inlet temperature, /C176C 21.4 23.0
Outlet temperature, /C176C 224.7 142.4
Mass flow rate, kg/s 2.51 2.41
Power, kW 525.5 547.0
LP compressor
Inlet pressure, bar – 0.99Outlet pressure, bar – 2.94Pressure ratio, – – 2.99Inlet temperature, /C176C – 23.0
Outlet temperature, /C176C – 154.9
Mass flow rate, kg/s – 2.41Power, kW – 322.3
HP compressor
Inlet pressure, bar – 2.91Outlet pressure, bar – 5.95
Pressure ratio, – – 2.04
Inlet temperature, /C176C – 51.6
Outlet temperature, /C176C – 142.4
Mass flow rate, kg/s – 2.41Power, kW – 224.7
Turbine
Inlet pressure, bar 3.95 5.66
Outlet pressure, bar 1.05 1.04
Pressure ratio, – 3.78 5.45Inlet temperature, /C176C 513.0 552.8
Outlet temperature, /C176C 338.1 323.0
Mass flow rate, kg/s 2.57 2.48Power, kW 539.7 658.3
aFor the original system: single-stage compressor only. For the new system: LP
and HP compressors in total.118 A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124

thickness. These pads are able to follow the shaft motion resulting
in low cross-coupled stiffness and damping [31]. They are widely
used to stabilize machines that have sub synchronous vibrations.
The drawback of the tilting pad bearings is that they have higher
power losses and higher manufacturing costs when compared to
bearings with a fixed profile.
Also the thrust bearing was a tilting pad type and was inte-
grated with a journal safety bearing at the electric machine end.
This design of thrust bearing was chosen due to its high loading
capacity.
4.2. Rotor dynamic analysis
The finite element method was used to simulate the rotor
dynamic performance. Rotor geometry was modeled as a beam-
like structure, which is the most typical way to perform rotor
dynamic analysis because it usually yields to results that are accu-
rate enough for most practical purposes while simple enough to
allow relatively straightforward simulations to be performed.
Approaches by Chen and Gunter [32] as well as by Genta [33] were
used for simulations to increase the reliability of the analysis. The
first approach [32] used viscosity as a function of temperature and
the second [33] as a function of rotational speed. Also a constant
viscosity was used for the simplest analysis.The conducted simulations included critical speed, unbalance
response and damped eigenvalue analyses for the rotor. The anal-
ysis was carried out with the Timoshenko beam [33] and included
the gyroscopic effect, shear deformation and centrifugal stiffening
of the shaft. The compressor and turbine wheels were assembled
with line-to-line contact, shrink fitted. At a high speed, the com-pressors, turbine, magnets, aluminum cage and carbon-fiber sleeve
have certain shaft stiffening effects due to centrifugal growth.
Therefore, to be conservative, the stiffness of the system was pro-
vided by the steel shaft alone in the model. The stiffness and damp-
ing of the bearings were also included in the model to have a
damped solution. Additionally, the unbalance response and desta-
bilizing aerodynamic type forces acting at the compressor and tur-
bine wheels were modeled. Fig. 4 presents an example of the
beam-like rotor model including the electric machine, turboma-
chinery and bearings.
A comparison between the predictions between the two pro-
grams is given in Table 3 . Two approaches are presented, first with
viscosity as a function of temperature and the second with viscos-
ity as a function of rotational speed. It is noticeable that the first
and second rigid modes are well predicted with both methods.
The location of the more important first forward bending flexural
speed has, however, slight differences between the predictions.
For comparison, the constant oil viscosity approach predicted the
first forward bending flexural speed to be 530 Hz. As a result of thisTable 2
Turbomachinery design parameters and predicted design point performance.
Nin Nout pt-s D2,m m b2,m m Dh/Dt Ns gs,t-t(CFD) gs,t-t(1D)
LP compressor 9 18 3.0 283 12.7 0.59 0.75 80.7 81.1
HP compressor 9 18 2.1 231 9.6 0.65 0.60 – 79.7
Nstator Nrotor pt-s R w M2(abs) M2(rel) gs,t-t(CFD) gs,t-t(1D)
Turbine 20 35 5.6 0.15 1.7 1.4 0.6 81.2 77.9
Fig. 3. Rotor of the two-stage turbocharger.
Fig. 4. Turbocharger model used in rotor dynamic analysis.A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124 119

analysis, it can be concluded that the design turbine rotational
speed 525 Hz is predicted to be between the first and the second
forward bending flexural speeds with both methods. However,
the compressor design speed 500 Hz is close to the predicted first
forward bending flexural speeds, which does not allow the tur-
bocharger to be operated close to 500 Hz. This means that a tur-
bocharger speed-up sequence has to be designed so that the
predicted frequencies are not used for continuous operation.
5. Experimental setup
The experiments were performed with the turbocharger con-
nected to a 1200 kW impulse charged diesel engine. The volumet-
ric displacement of the test engine was 52.8 l, the nominal rotationspeed 1000 rpm and the compression ratio 15. During the mea-
surements, the engine speed was kept constant at 1000 rpm while
the engine load was varied with a water brake. Fig. 5 shows an
overview of the measurement setup.
The engine speed, torque and fuel consumption were measured
during the tests. Also the power input/output of the electric
machine was recorded. The turbocharger instrumentation included
static pressure and temperature measurements at the inlet and
outlet of the LP and HP compressors and the turbine. Static pres-
sures were measured with pressure tappings and temperatures
with PT100. Engine temperatures were also measured by a
K-type thermocouple and PT100 as part of facility diagnostics.
The LP compressor inlet air flow was measured with an ISA 1932
nozzle. The estimated uncertainty of the measurements is pre-
sented in Table 4 together with the measurement range.
Vibration monitoring of the turbocharger included two orthog-
onally (mounted 90 degrees apart) installed proximity sensors,
which were used to provide motion tracking of the shaft end inboth xand ydirections, see Fig. 6 (a). The described non-contact
vibration measurement system was assembled on the electric
machine and the turbine ends of the shaft. An example of the mea-
sured shaft orbit at 28,500 rpm is presented in Fig. 6 (b) for x- and
y-directions in the electric machine end. The maximum vibration
amplitude was about 0.1 mm and was mainly due to local imbal-
ance at the electric machine end of the shaft, which was caused
by a key slot. For comparison, an amplitude of 0.04 mm was found
at the turbine end. These results indicate that the shaft was still
operated at a sub-critical frequency, i.e. below the modeled 490–
504 Hz (29,400–30,240 rpm), which confirms the FEM results.
5.1. Startup and shutdown of the turbocharger
At the beginning, the turbocharger rotational speed was electri-
cally raised to 9000 rpm. In this condition, the compressor was in
surge because the diesel engine was not operating. After the tur-
bocharger had achieved a 9000 rpm speed, the diesel engine was
started in idle mode. When the engine started to take air in, the
turbocharger operated normally and the rotational speed was
increased with the electric motor to 12,000 rpm. Then the system
was operated at 12,000 rpm for a warm-up period before the actual
power runs. During the power runs, the rotational speed of the tur-
bocharger was electrically stepwise increased and also load was
given for the engine. During the startup and power run, the com-
pressor blow off valve was kept closed.
After the power runs, the diesel engine load was decreased to
idle mode and was operated there until the exhaust temperature
had decreased to about 250 /C176C after the turbine. During this cooling
period, the turbocharger rotational speed was kept constant at
18,000 rpm. After the cooling period, the turbocharger speed was
decreased in three steps from 18,000 rpm to 12,000 rpm and finally
to 9000 rpm when the engine was shut down. After the engine shut
down, the compressor blow off valve was opened. Then the rota-
tional speed of the turbocharger was raised again to 12,000 rpmand it ran there until the bearing body temperature had decreased
enough (15 …30 min) and the turbocharger was shut down.
6. Experimental results
6.1. Turbocharger and engine performance
Three sets of measurements were taken: the first set of
measurements with a regular camshaft and the second as well as
the third with Miller camshafts (Miller Camshaft 1 and Miller
Camshaft 2). Due to an electric power measurement malfunction
at the end of the campaign, the electric power is calculated with
a Miller camshaft 2 test from the turbomachine measurements.
Fig. 7 shows the required electric input as a function turbocharger
rotational speed for all the measured setups. It illustrates, that the
required electric power has a decreasing trend. However, due to an
increase of the engine outlet temperature close to the maximum
allowable values and since the modeled critical speed was closeTable 3
Predicted rigid modes and predicted forward and backward whirl critical speeds (Hz)with viscosity as a function of temperature or rotational speed.
Direction Viscosity 1st
rigid2nd
rigid1st
flexural2nd
flexural
Forward whirl f( T) 96 204 504 1283
Backward whirl f( T) 95 199 439 868
Forward whirl f( N) 96 200 490
Backward whirl f( N) 95 191 417
Fig. 5. Overview of the measurement setup.Table 4
Estimated measurement uncertainty and measurement range.
Uncertainty Range
pamb ±50 Pa 90–110 kPa
pg ±0.5% 0–400 kPa
T ±0.5 K 293–818 K
qm ±1% 0.9–2.0 kg/s
N ±0.4 Hz 0–500 Hz
Electric power ±1% 0–130 kWEngine load ±0.5% 0–1400 kpShaft proximity ±0.006 mm 0–1 mm120 A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124

to 30,000 rpm, the maximum speed was limited to 28,500 rpm and
the targeted 31,500 rpm was not achieved during the experiments.
To give an estimate for the performance at higher speeds, a
rough linear least squares estimation shown in Fig. 7 suggests that
with a Miller camshaft the turbocharger would produce electricity
at speeds above 33,000 rpm, which is still achievable with the tur-
bocharger. The differences in the required electric power with the
lowest rotational speed (18,000 rpm) are explained by the different
engine load, where the highest value 74 kW is recorded without
engine load as shown in Fig. 8 . When the rotational speed is
increased from 18,000 rpm to 21,000 rpm, the performance of the
turbocharger is not markedly improved and the input electric
power increases. This is due to that both compressors operate out-
side of the design conditions, thus having poor efficiency (see com-
pressor operating maps in Fig. 10 ). By increasing the rotational
speed more, the required power decreases due to better tur-
bocharger performance while moving further away from the surge
limit with the LP compressor and toward higher efficiency areas
with the HP compressor. As the rotational speed is increased more,the turbocharger operation moves toward the design point and less
electricity is used to boost the system.From Fig. 9 , it is noticeable that the introduction of the two-
stage turbocharger and Miller timing has a minor influence on
the brake specific fuel consumption (BSFC) compared with the
baseline engine utilizing a single-stage turbocharger with a regular
camshaft if the additional electricity used by the electric machine
is not included (filled markers). The higher BSFC with a regular
camshaft is explained by the compressor operating at the low effi-
ciency region. All in all, the shown trend indicates that the engine
performance improves while the engine power increases with the
studied two-stage turbocharger. The performance improvement is
even more drastic when the additional electricity is included in
BSFC (total values in Fig. 9 ).
6.2. Compressor performance
The improving performance of the two-stage compressor is pre-
sented in Figs. 10 and 11 , showing an increase of the pressure ratio
as a function of rotational speed. It is worth mentioning, that the
lower constant speed lines in Fig. 10 are based on rough estima-
tions and the LP compressor is therefore operating at smaller flows
than the surge line is limiting. A second degree polynomial inter-
polation (with R
2=1 )i n Fig. 11 predicts that the design pressure
Fig. 6. Vibration monitoring with two proximity sensors (a) and example of the shaft orbit at the electric machine end at 28,500 rpm (b).
Fig. 7. Electric power consumption of the turbocharger as a function of tur-
bocharger rotational speed with a regular camshaft and Miller camshafts. Thedashed line denotes a linear approximation for electric powerconsumption/generation.
Fig. 8. Electric power consumption of the turbocharger as a function of engine load
with a regular camshaft and Miller camshafts.A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124 121

ratio (6.04) is achieved at 33,000 rpm. This prediction is also in line
with the results shown in the LP and HP compressor operating
maps. Overall, the LP compressor reaches the targeted pressure
ratio closer to the predicted rotational speed than the HP compres-
sor. For example the measured pressure ratio of the LP compressor
at 26,000 rpm corresponds with the value predicted at 27,000 rpm,
but for the HP compressor over 28,000 rpm is required to produce
the pressure ratio corresponding to the predicted value at
26,000 rpm. The pressure ratio of the HP compressor is lower than
expected mainly because the efficiency is lower than predicted.
The isentropic efficiency of the HP compressor was 67% at
28,500 rpm, which is much lower than the design value 79.7%.For comparison, the LP compressor efficiency was 77% at the same
operating point, which is much closer to the design value 81.1%.
One reason for this observed behavior is the challenge of matching
the two compressors. Therefore, a redesign of the HP compressor
could lead to better matching of the two stages and also improve
the overall performance.
6.3. Pressure losses
Fig. 12 shows a comparison between modeled and measured
pressure losses in the intercooler, and the combined losses for
the aftercooler, engine and turbine inlet piping are reported. There
seems to be a good agreement between the design and measured
value of the pressure loss in the intercooler. However, the lossesbetween the HP compressor outlet and the turbine inlet are clearly
under predicted in the modeling. This also decreases the turbine
power and transfers the power balance into higher rotational
speeds. When the causes for the increased losses were studied
more, it was found that the main reason was in the connection
between the turbine inlet part and the exhaust manifold. To con-
nect these two parts, an additional contraction was made, and this
was estimated to cause the difference between the modeled and
measured pressure loss. The influence that the additional pressure
loss has on the turbine power is presented in Fig. 13 for the Miller
Camshaft 2 test. It suggests that about 8–11 kW additional losses
were caused by the decreased mass flow through the turbine dur-
ing the test. The changed mass flow has been calculated based on
the turbine ellipse law. Extrapolated power losses at the design
pressure ratio and temperature are estimated to be approximately
23 kW, which also moves the power balance and electricity pro-
duction toward higher rotational speeds. Since the turbine limits
the mass flow of the turbocharger, it also moves the compressor
Fig. 9. Effect of a two-stage turbocharger on the BSFC (filled marker is calculated
without external electricity).
Fig. 10. Predicted operating map of LP (a) and HP compressors (b) and operating
points during different test runs.
Fig. 11. Compressor pressure ratio as a function of turbocharger rotational speed
with a regular camshaft and Miller camshafts.122 A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124

operation toward the surge line. This shift is beneficial for the HP
compressor but detrimental for the LP compressor and contributes
to the mismatching of the two stages.
7. Conclusions
A two-stage intercooled electrically assisted turbocharger con-
nected to a 1200 kW impulse charged diesel engine was computa-
tionally and experimentally studied. The preliminary simulations
predicted that there is a potential for increasing the overall effi-
ciency of a 1200 kW diesel engine by 2.5 percentage points in
the design operating conditions by using a two-stage single-shaft
intercooled electrically assisted turbocharger instead of a conven-
tional single-stage turbocharger. The design of the turbocharger
was presented in detail with special attention to mechanical design
and rotor dynamic analysis, which included a presentation of a
novel measurement system for the shaft orbit. The performance
of the prototype was measured with the turbocharger connected
to a diesel engine. Due to the excessive exhaust gas temperature
and the risk of reaching the rotor critical speed, the turbocharger
was not operated at the nominal speed. However, based on theconducted measurements, the achievable energy production capa-
bility of the system was estimated. The results indicated that the
turbocharger would be in power balance slightly above its nominal
operating point and the electricity production would be possible
above the nominal speed. This shift was found to be partly due
to additional pressure losses due to poor exhaust manifold-
turbine connection design, which influences the turbine power
through decreased mass flow. Some part of the shift can also be
explained by the underestimation of mechanical losses. However,
this path remains uncertain due to a lack of measured oil temper-
ature data at the high rotational speeds. The remaining differences
are attributed to the mismatching between the compressor stages,
and to the poor operation of the HP compressor. Therefore, a rede-
sign of the HP compressor could lead to marked improvement of
the turbocharger performance.
It was also found that the engine performance was affected by
introducing the two-stage turbocharger. The identified trend indi-
cated a decreasing BSFC while the engine load was increased and
can lead to values below the baseline engine with high engine
loads where the turbocharger was designed to be operated.
However, since the studied single spool design causes the HP
compressor to have a specific speed below its optimal value
0.7–0.85, it is recommended that a two spool design could be more
feasible at the studied pressure ratio and especially if higher
pressure ratios are targeted. In such a design, the compressor of
both LP- and HP-turbochargers could be designed to have more
optimal specific speeds, and thus also improved performance is
to be expected compared to single spool design where the com-
pressor designs are made for a single rotational speed. Also in
the two spool design, the turbine can be operated at subsonic pres-
sure ratios where the efficiency is also most likely higher since the
shock losses can be avoided. Based on the given arguments, an
electrically assisted two spool turbocharger is also suggested for
future development together with the proposed modifications to
the current single spool design.
Acknowledgements
The study was funded by Tekes – the Finnish Funding Agency
for Innovation in the GENSET project. The guidance of Emeritus
Professor Jaakko Larjola is also greatly acknowledged.
References
[1] Heim K. Existing and future demands on the turbocharging of modern large
two-stroke diesel engines. In: 8th Supercharging conference, 1–2 October2002, Dresden, Germany.
[2] IMO, Nitrogen Oxides (NOx) – Regulation 13. London: International Maritime
Organization.
[3]
Baldi F, Gabrielii C. A feasibility analysis of waste heat recovery systems for
marine applications. Energy 2015;80:654–65 .
[4]Uusitalo A, Honkatukia J, Turunen-Saaresti T, Larjola J. A thermodynamic
analysis of waste heat recovery from reciprocating engine power plants by
means of organic rankine cycles. Appl Therm Eng 2014;70:33–41 .
[5]Yang F, Dong X, Zhang H, Wang Z, Yang K, Zhang J, et al. Performance analysis
of waste heat recovery with a dual loop Organic Rankine Cycle (ORC) system
for diesel engine under various operating conditions. Energy Convers Manage
2014;80:243–55 .
[6]Zhang J, Zhang H, Yang K, Yang F, Wang Z, Zhao G, et al. Performance analysis
of Regenerative Organic Rankine Cycle (RORC) using the pure working fluid
and the zeotropic mixture over the whole operating range of a diesel engine.
Energy Convers Manage 2014;84:282–94 .
[7]Larsen U, Sigthorsson O, Haglind F. A comparison of advanced heat recoverypower cycles in a combined cycle for large ships. Energy 2014;74:260–8
.
[8]Livanos GA, Theotokatos G, Pagonis D-N. Techno-economic investigation of
alternative propulsion plants for Ferries and RoRo ships. Energy Convers
Manage 2014;79:640–51 .
[9]Hossain SN, Bari S. Waste heat recovery from the exhaust of a diesel generatorusing Rankine cycle. Energy Convers Manage 2013;75:141–51
.
[10] Choi BC, Kim YM. Thermodynamic analysis of a dual loop heat recovery system
with trilateral cycle applied to exhaust gases of internal combustion engine for
propulsion of the 6800 TEU container ship. Energy 2013;58:404–16 .
Fig. 12. Comparison between the measured and modeled pressure losses at the
intercooler and from HP compressor outlet to the turbine inlet (Compressor-Turbine) during Miller Camshaft 2 test.
Fig. 13. Power loss due to additional pressure losses during Miller Camshaft 2 test.A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124 123

[11] Saidur R, Rezaei M, Muzammil W, Hassan M, Paria S, Hasanuzzaman M.
Technologies to recover exhaust heat from internal combustion engines.
Renew Sustain Energy Rev 2012;16:5649–59 .
[12] Riffat S, Ma X. Thermoelectrics: a review of present and potential applications.
Appl Therm Eng 2003;23:913–35 .
[13] Rowe DM. Thermoelectrics, an environmentally-friendly source of electricalpower. Renewable Energy 1999;16:1251–6
.
[14] Knecht W. Diesel engine development in view of reduced emission standards.
Energy 2008;33:264–71 .
[15] Cui Y, Hu Z, Deng K, Wang Q. Miller-cycle regulatable, two-stage
turbocharging system design for marine diesel engines. J Eng Gas TurbinesPower 2014;136:022201
.
[16] Wik C, Salminen H, Hoyer K, Mathey C, Vögeli S, Kyrtatos P. 2-Stage turbo
charging on medium speed engines – future supercharging on the new LERF-
test facility. In: 16th Supercharging conference. 29–30 September 2011,
Dresden, Germany.
[17] Codan E, Huber T. Application of two stage turbocharging systems on large
engines. In: 10th International conference on turbochargers andturbocharging. 15–16 May 2012, London, UK.
[18]
Serrano JR, Arnau FJ, Dolz V, Tiseira A, Lejeune M, Auffret N. Analysis of the
capabilities of a two-stage turbocharging system to fulfil the US2007 anti-
pollution directive for heavy duty diesel engines. Int J Automotive Technol
2008;9:277–88 .
[19] Galindo J, Serrano JR, Climent H, Varnier O. Impact of two-stage turbocharging
architectures on pumping losses of automotive engines based on an analytical
model. Energy Convers Manage 2010;51:1958–69 .
[20] Ono Y, Shiraishi K, Yamashita Y. Application of a large hybrid turbocharger for
marine electric-power generation. Mitsubishi Heavy Ind Tech Rev 2012;49(1):29–33
.
[21] Terdich N, Martinez-Botas RF, Romagnoli A, Pesiridis A. Mild hybridization via
electrification of the air system: electrically assisted and variable geometry
turbocharging impact on an off-road diesel engine. J Eng Gas Turbines Power
2014;136:031703 .
[22] Terdich N, Martinez-Botas RF. Experimental efficiency characterization of an
electrically assisted turbocharger. SAE technical paper 2013-24-0122; 2013.http://dx.doi.org/10.4271/2013-24-0122
.[23] Millo F, Mallamo F, Pautasso E, Ganio Mego G. The potential of electric exhaust
gas turbocharging for HD diesel engines. In: SAE 2006 world congress and
exhibition. 3–6 April 2006, Detroit, Michigan, USA. SAE technical paper 2006-
01-0437.
[24] Padzillah MH, Rajoo S, Martinez-Botas RF. Influence of speed and frequency
towards the automotive turbocharger turbine performance under pulsatingflow conditions. Energy Convers Manage 2014;80:416–28
.
[25] Chiong MS, Rajoo S, Romagnoli A, Costall AW, Martinez-Botas RF. Non-
adiabatic pressure loss boundary condition for modelling turbocharger turbine
pulsating flow. Energy Convers Manage 2015;93:267–81 .
[26] Chiong MS, Rajoo S, Romagnoli A, Costall AW, Martinez-Botas RF. Integrationof meanline and one-dimensional methods for prediction of pulsating
performance of a turbocharger turbine. Energy Convers Manage
2014;81:270–81
.
[27] Grönman A, Turunen-Saaresti T, Röyttä P, Jaatinen A, Backman J. Performance
and flow fields of a supersonic axial turbine at off-design conditions. Proc
IMechE Part A: J Power Energy 2013;227:285–94 .
[28] Kacker SC, Okapuu U. A mean line prediction method for axial flow turbineefficiency. J Turbomach 1982;104:111–9
.
[29] Kolondzovski Z, Sallinen P, Arkkio A. Thermal analysis of a high-speed PM
machine using numerical and thermal-network method. In: International
conference of electrical machines, ICEM, 6–8 September, 2010, Roma, Italy.
[30] Kolondzovski Z, Arkkio A, Larjola J, Sallinen P. Power limits of high-speedpermanent-magnet electrical machines for compressor applications. IEEE
Trans Energy Convers 2011;26:73–82
.
[31] Allaire PE, Nicholas JC, Barret LE, Gunter EJ. Bearing characteristics and
stability for finite bearings – a manual for use with the computer program
FINBRG1. Report, School of Engineering and Applied Science, University ofVirginia, Charlottesville, Virginia, August 1975.
[32]
Chen WJ, Gunter EM. Introduction to dynamics of rotor-bearing
systems. Trafford Publishing Ltd.; 2005. ISBN: 1-4120-5190-8 .
[33] Genta G. Dynamics of rotating systems. Springer; 2005. ISBN: 0-387-20936-0 .124 A. Grönman et al. / Energy Conversion and Management 111 (2016) 115–124