Experimental determination of turbocharger heat flows for validation of a power separation approach [original]

Content from this work may be used under the terms of theCreativeCommonsAttribution 3.0 licence. Any further distribution

of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Published under licence by IOP Publishing Ltd

ISROMAC18

Journal of Physics: Conference Series 1909 (2021) 012061

IOP Publishing

doi:10.1088/1742-6596/1909/1/012061

Experimental determination of turbocharger heat

flows for validation of a power separation approach

Philipp Brodbeck

Research Assistant, Department of Powertrain Technologies, Technische Universit¨at Berlin,

Carnotstr. 1A, 10587 Berlin, Germany

E-mail: [email protected]

Abstract. This work presents a new method to measure the heat flow from a turbocharger

to the ambient air. The experimental setup, performed on a hot gas test bench, includes an

insulated box with two openings, encapsulating the turbocharger. These openings enable a

controlled air flow through the box, which is measured at the inlet duct of the box. With

temperature measurements at the inlet and outlet of the box, the quantitative heat flow

can be calculated. With this information, an approach to estimate those heat flows from

standard turbocharger test bench measurements can be validated and refined. The performed

measurements at different turbine inlet temperatures and turbine mass flows show a good

conformity with expected heat flows. Due to convective and radiative effects, the heat flows

increase with both turbine mass flow and turbine inlet temperature. However, the setup with

the insulated box has an effect on the overall thermal conditions, since the convection around

the turbocharger decreases if the specimen is not as exposed as in common test bench setups.

1. Introduction

The precise prediction of turbocharged engine performance is a central element of automotive

powertrain design. Modelling of turbochargers in process simulation usually relies on maps

of stationary operating points, which are measured on hot gas test benches with standardized

conditions. But the thermodynamic description of the turbocharger behavior is sensitive to

the thermal conditions, which distort the isentropic efficiencies significantly on the turbine side.

This is caused by heat flows between turbine and compressor gases, oil, cooling liquid, and the

surrounding air, see Figure 1, which cannot be prevented by the duct insulation commonly used

at test benches. At low rotor speed and pressure ratios, this can result in calculated isentropic

turbine efficiencies greater than one, which is a violation of the first law of thermodynamics.

The process of expansion within the turbine at low rotor speed is shown in Figure 3. The

temperature under diabatic conditions T4,dia is the one measured at the test bench, but the

temperature, which would be obtained under adiabatic conditions T4,adi is the one of interest

and the relevant value to calculate the correct isentropic turbine efficiency. Hence, the turbine

efficiency is usually defined as the ratio of compressor power and isentropic turbine power [1]

to avoid the temperature measurement at the turbine outlet. Although thermal distortions are

mostly averted this way, turbine efficiency and the mechanical efficiency are inseparably merged

in one coefficient, which has a negative influence on the results calculated with map based

turbocharger models in process simulation. Turbocharger rotation speed and turbine outflow

ISROMAC18

Journal of Physics: Conference Series 1909 (2021) 012061

IOP Publishing

doi:10.1088/1742-6596/1909/1/012061

Figure 1. Scheme of turbocharger test

bench and occurring heat flows within

the measurement plane boundaries.

Figure 2. Turbine power curves under hot and

adiabatic conditions (schematic).

temperature prediction are particularly difficult with conventional maps, if the turbine inlet

temperature is different from the 600◦C as prescribed in the standard test bench code.

Methods to correct the thermodynamic performance of turbine and compressor have been

developed in previous works, for example by L¨uddecke [2] et. al. However, the described method

requires significantly increased experimental effort, which is hardly reasonable for non-research

users. A potentially more accessible method to separate the aerodynamic power of turbine

and compressor as well as the bearing friction losses from the occurring heat flows has been

developed by Baar [3] et al. This approach can be conducted with any turbocharger test bench

data, only a reliable method to measure the turbine outflow temperature is mandatory. At

the chair of powertrain technologies, this is done with a static flow mixer placed in the turbine

outlet pipe between the turbine and the respective measurement plane to create a uniform

temperature distribution across the turbine outlet duct. The turbine outlet temperature is

required to calculate the diabatic turbine power, which can be correlated with the isentropic

compressor power as shown in Figure 2. Previous works from Savic [4] et al. have shown,

that the interpolated line through the respective compressor power maxima of each speed line

intersects the turbine power axis at a value, which corresponds to the total amount of heat

transferred from the exhaust gas to other turbocharger parts and the ambient air. This theory

is supported by turbocharger test bench data, which has been recorded with close to adiabatic

conditions. The turbine inlet temperature and the mean oil temperature have been matched with

the compressor outlet temperature for those measurements to minimize the heat losses, especially

on the turbine side. As expected, the line interpolated through the isentropic compressor power

maxima intersects the y-axis close to zero since there are no significant heat losses, which

is schematically shown in Figure 2. The ’power-based approach’ has been developed on the

assumption, that the isentropic compressor power and the bearing friction are not dependent on

the turbine inlet temperature. The extra turbine power apparently needed under hot conditions

for the same compressor power output, can be traced back solely to the temperature T4, which is

distorted by heat losses. Hence, the diabatic turbine power measured under hot conditions can

be corrected by shifting the data downwards to match the adiabatic measurements and obtain

the pure aerodynamic turbine power using the equation

PT,adi = ˙mT·cp·(T3−T4)−˙

QT urbine (1)

where ˙mTis the turbine mass flow, cpis the specific heat capacity of exhaust gas and the

ISROMAC18

Journal of Physics: Conference Series 1909 (2021) 012061

IOP Publishing

doi:10.1088/1742-6596/1909/1/012061

Figure 3. Schematic thermodynamics of

turbine expansion process (total to static).

Figure 4. Schematic thermodynamics of

compression process (total to total).

temperature at the turbine inlet T3is the total temperature. The measurements under adiabatic

conditions are not mandatory to use the power-based approach. However, for validation

purposes, several adiabatic measurements on differently sized turbochargers have been conducted

prior to this investigation. The results showed a similar correlation between the diabatic and

adiabatic turbine power curves as displayed in Figure 2.

The power-based approach can also be applied on the diabatic compressor power to separate

it from any heat flow disturbances, which are generally small compared to the turbine side.

Previous investigations have shown, that the compression process is mainly influenced by added

heat from the exhaust gas side [5], which is schematically shown in Figure 4, and almost constant

across the compressor operation range. Higher temperatures at the compressor outlet cause an

underestimation of the isentropic compressor efficiency, hence the compressor power can be

corrected using

PC,adi = ˙mC·cp·(T2−T1)−˙

QCompressor (2)

where ˙mCis the compressor mass flow, cpis the specific heat capacity of air and T1and T2are

total temperatures. The heat flow ˙

QCompressor is obtained analogously by plotting the diabatic

compressor power correlated to the isentropic compressor power and shifting the data by the

occurring offset. The corrected values for turbine and compressor power can now be used to

calculate corrected efficiencies and the friction power as shown in the Equations 3 and 4. Using

corrected maps and an external input for the bearing friction in process simulation makes the

model independent of the overall thermal conditions. However, turbine and compressor perfor-

mance are still characterized thermodynamically, so the heat flows between the components and

the surrounding air have to be added as a sub model to receive the best results. This is done

with a network of thermal nodes with connections of convective, conductive and radiative heat

transfer to the respective parts.

ηT,is =PT,adi

PT,is

(3)

PF riction =PT,adi −PC,adi (4)

While the total heat losses of the turbocharger can be estimated with this approach, it cannot be

distinguished, how the individual heat flows are allocated between the components and ambient

air. Investigations with conjugate heat transfer (CHT) simulations have shown, that the latter

is a relatively large part of the total heat flow [5]. The heat flow box has been developed to

determine the exact amount of heat transferred from the turbocharger to the surrounding air

during stationary turbocharger operation. This information will help to clarify, if the offset

ISROMAC18

Journal of Physics: Conference Series 1909 (2021) 012061

IOP Publishing

doi:10.1088/1742-6596/1909/1/012061

between the turbine power charts in Figure 2 directly correlates to the external heat flows

measured. The dependency of external heat flow on turbine inlet temperature, turbine mass

flow and compressor load are also subjects of the investigation.

2. Experimental Setup

The test bench, which the turbocharger and heat flow box have been investigated on, complies

with international test bench standard codes such as the norm SAE J1826. Compressor and

turbine flow are separate, the turbine side is supplied with air by a screw compressor and

mixed with diesel fuel in a combustion chamber. The compressor load is controlled by two

valves arranged in parallel, a complete scheme of the test bench can be found in the appendix

Figure A1. The object of investigation is a turbocharger for a 2.0-liter engine of series production

with a variable geometry turbine, which is fully open for all tests. It is not water-cooled and the

oil is conditioned at 3 bars relative and 90◦C. As an addition, the turbocharger parts have been

equipped with a total of 21 thermocouples to record the outer surface temperature distribution

on the housing parts of turbine, bearings and compressor. The 1 mm thermocouple tips have

been glued or welded (depending on the expected temperature) into 1.5 mm drill holes in the

housing. This is not mandatory for the heat flow box measurement as a standalone system, but

gives an impression of the convective conditions between turbocharger and air. Housing wall

temperatures have also been recorded during the reference measurement, which was a standard

600◦C turbine inlet temperature map with the turbocharger exposed in the air-conditioned test

cell and insulated ducts within the measurement plane boundaries.

Computational fluid dynamics (CFD) has been used prior to the construction to optimize

the shape of the box. The air stream has to surround the whole turbocharger without creating

any recirculation. The box consists of two angular-shaped funnels with a cuboid in the middle,

which contains the turbocharger and its peripheral as seen in Figure 5. Natural convection,

caused by the hot turbocharger parts, creates a uniform air flow through the box from bottom

to top. The amount of air passing through the box is measured by a vortex flowmeter integrated

into the inlet pipe. The respective temperatures at the inlet and outlet are measured with

five thermocouples each, evenly distributed across the ducts. The rate of heat flow from the

Figure 5. Complete experimental setup,

concept and implementation.

Figure 6. Setup of turbocharger inside the box,

view from above.

ISROMAC18

Journal of Physics: Conference Series 1909 (2021) 012061

IOP Publishing

doi:10.1088/1742-6596/1909/1/012061

turbocharger to ambient air can be obtained using the equation

QT C = ˙mHF B ·cp·(Tout −Tin) (5)

with the measured values as displayed in Figure 5. With a measurement accuracy of 1.5◦C

of the thermocouples and an accuracy of 1.5% of the vortex flowmeter, the mean accumulated

uncertainty of the calculated heat flow rate is ±6.5%, which is sufficient to prove the feasibility

of this novel measurement concept.

Each wall of the box consists of two layers of aluminum plating with an insulation layer in

between, to create a system which is close to adiabatic. The three layers are assembled with

threaded rivets and screws and built into an aluminum framework. The temperature on the

outer surface of the box is measured at ten evenly distributed points, to estimate the amount of

heat leaking through the system. The box has an orifice for each of the four gas ducts leading

to and away from turbine and compressor, two orifices for the oil system and one for electric

wiring, the whole assembly is showcased in Figure 5. The insulation on the ducts has been added

according to the reference measurement without the box to recreate the thermal conditions. The

same has been done with the insulation on the duct parts inside the box as seen in Figure 6,

which is a view on the turbocharger from the top without the lid on. The smoothed aluminum

walls on the inner side of the box serve as a reflector for the heat radiated mainly from the

turbine parts. All connected edges of the box are sealed with high-temperature silicone.

3. Results and Discussion

Measurement results are shown for three different turbine inlet temperatures of 400, 600 and

800 degrees Celsius. The operating points have been distributed across the standard compressor

map with three points on four different speed lines reaching from choke line to surge line. The

additional measurements include the air flow through the box and the temperature difference

between the inlet and the outlet of the box. The temperature difference is plotted in Figure 7.

It is visible, that the heat flow to ambient air seems to rise significantly with the turbine mass

flow. This complies with the turbine housing temperatures, which can be seen in Figure 8 for

a turbine inlet temperature of 600 degrees. The reason for that increase is a larger convective

heat flow from exhaust gas to housing at higher pressure and velocity. The gradients of the

100 200 300 400 500 600 700 800

100

150

˙mT urbine[kg ·h−1]

∆THF B[K]

600◦C

400◦C

800◦C

Figure 7. Temperature difference related to

turbine total mass flow for different turbine

inlet temperatures.

100 200 300 400 500 600 700 800

460

480

500

520

540

˙mT urbine[kg ·h−1]

TT H [K]

in HFB

Reference

Figure 8. Turbine housing temperature TT H ,

T3= 600◦C.

Loading more pages...