This Content Downloaded From 202.3.77.169 On Thu, 26 Nov 2020 09:25:05 UTC
This Content Downloaded From 202.3.77.169 On Thu, 26 Nov 2020 09:25:05 UTC
This Content Downloaded From 202.3.77.169 On Thu, 26 Nov 2020 09:25:05 UTC
Author(s): Nicolo Cavina, Andrea Borelli, Lucio Calogero, Ruggero Cevolani and Luca Poggio
Source: SAE International Journal of Engines , Vol. 8, No. 5 (November 2015), pp. 2120-2132
Published by: SAE International
Stable URL: https://www.jstor.org/stable/10.2307/26278107
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/10.2307/26278107?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
SAE International is collaborating with JSTOR to digitize, preserve and extend access to SAE
International Journal of Engines
ABSTRACT
The paper presents possible solutions for developing fast and reliable turbocharger models, to be used mainly for control applications.
This issue is of particular interest today for SI engines since, due to the search for consistent CO2 reduction, extreme downsizing
concepts require highly boosted air charge solutions to compensate for power and torque de-rating. For engines presenting at least four
in-line cylinders, twin-entry turbines offer the ability of maximizing the overall energy conversion efficiency, and therefore such
solutions are actually widely adopted.
This work presents a critical review of the most promising (and recent) modeling approaches for automotive turbochargers,
highlighting the main open issues especially in the field of turbine models, and proposing possible improvements. The main original
contribution is then on solving specific issues related to the twin-entry turbine, to develop a control-oriented model able to predict the
machine behavior under all possible admission conditions.
The results of this study have been applied to a V8 high-performance GDI engine with twin-entry turbochargers. Experimental data are
shown throughout the paper, to demonstrate the benefits of the proposed approach.
CITATION: Cavina, N., Borelli, A., Calogero, L., Cevolani, R. et al., "Turbocharger Control-Oriented Modeling: Twin-Entry Turbine
Issues and Possible Solutions," SAE Int. J. Engines 8(5):2015, doi:10.4271/2015-24-2427.
INTRODUCTION consequence is that the instantaneous mass flow rate and efficiency
values (i.e., at the actual rotational speed and pressure ratio) are
The physics governing the turbocharger operation is particularly
obtained from steady-state experimental data [1].
complex, since the flow is viscous, unsteady and three-dimensional,
and a significant amount of heat is exchanged between turbine and
This work focuses on the analysis and control-oriented modeling of a
compressor and with the surrounding environment [1]. For these
turbocharging system for a high-performance V8 engine. For this
reasons, a detailed and “white” box model (typically 3-D) of the
reason, each turbocharger is equipped with a Waste-Gate (WG)
turbocharger fluid dynamics is needed during the components design,
by-pass system, and the turbines are of twin-entry type. During this
development and optimization phases. Some authors have recently
project, different turbocharger models, produced by the same
demonstrated that a 1-D model of the turbine still allows representing
supplier, have been modeled and tested. The paper initially presents
the turbine behavior under unsteady operating conditions quite
experimental data, control-oriented models, critical issues and
accurately [2, 3]. Due to the complexity of this kind of approach, and
possible solutions for single-entry, or “standard”, turbines. The
to the need for expensive and time consuming experimental tests
second part of the paper is instead entirely focused on twin-entry
([4]), a zero-dimensional, map-based (or “grey/black”) representation
turbine modeling. The main original contribution is the development
of the turbocharger performance is still preferred for studying the
of a map-based, zero-dimensional twin-entry turbine control-oriented
interactions between the internal combustion engine and the
model, able to predict its behavior under all admission conditions.
turbocharger itself (so called turbo-matching), or for developing and
testing engine control strategies [5, 6, 7, 8, 9, 10, 11]. This modeling
approach considers the flow quasi-stationary, and the main
2120
Figure 1. Hot gas test bench with compressor closed loop control. Source [13].
To further extend the turbine maps at the lowest ERs, run-away speed
(1) measurements are performed by replacing the compressor rotor with
a bladeless one of the same mass. By neglecting the aerodynamics
friction on the bladeless disc, it is possible to determine the minimum
ER at which the turbine is still able to provide mechanical work, and
(2)
therefore to identify the zero turbine efficiency condition for each
rotational speed level (Figure 2).
(3)
(4)
proposed the use of a cold-gas test-bench [13, 14]. For what concerns The interpolation and extrapolation of experimental compressor data
friction losses determination, specific test benches may be developed has been carried out following the well-known approach proposed by
using electric machines for measuring the mechanical efficiency. A [15], and based on the non-dimensional parameters introduced by
simplified approach consists of including such losses in the turbine [16] (head parameter Ψ and non-dimensional flow rate ϕ):
efficiency map [1]. Finally, it should be observed that flow-bench
tests are typically performed with the waste-gate in closed position
(sometimes the waste gate is mechanically sealed to prevent leakage),
and this makes it very complex to model the turbine performance
under all other operating conditions. (5)
The coefficients Ki,j have been identified using Matlab-based custom Figure 4 shows that the interpolation and extrapolation of
genetic algorithms, converging over minimum Root Mean Squared experimental data is now much more robust. In particular,
Error (RMSE) between experimental and modeled data. A key issue extrapolation at low PRs is now consistent with the compressor
is the algorithm initialization, and the genes of the starting population experimental choking behavior. Still, the extrapolation towards
(the initial values of the coefficients Ki) have been selected by forcing smaller mass flow rates is strictly monotone, while experimental data
the corresponding homographic function to pass through three denote a PR reduction when approaching the surge line (mainly due
specifically selected experimental values, for each speed line (max to incidence losses [1]), especially at high speeds. Two further
PR, max CMF, and intermediate conditions). considerations are needed to completely clarify this issue: Since the
extrapolated maps will be used in a filling-emptying 0-D model,
In [15] it is also stated that the proposed methodology is not providing mass flow rate in response to PR inputs, the curves need to
particularly accurate for the highest compressor rotational speeds, be monotone and a more mathematically fit function would introduce
when M becomes greater than 1.1 and the flow characteristics change uncertainty. Secondly, such operating conditions (highest speeds and
significantly. The accuracy loss is clearly visible in Figure 3. To PRs) are typically not reached during engine-turbocharger interaction.
overcome this limit, some authors have proposed solutions purely
based on mathematical fitting techniques (such as spline functions or It should finally be noted how the extrapolation at even higher
neural network algorithms [7, 10]), while others have attempted to rotational speeds (in this case higher than 175500 rpm) is not reliable,
maintain the parametric description introduced by [15, 16], evolving and this is due to the type of function used for data interpolation/
it according to what proposed by [11, 14]. In particular, in [11] it is extrapolation. A homographic function as the one defined by
suggested that each constant-speed group of experimental data should Equation (8) presents a discontinuity as K3 → ϕ, and it provides
be treated separately, and that a set of Ki parameters should be reliable information only if the left branch of the function is involved.
identified for each constant-speed dataset. Then quadratic functions
are used to fit the Ki coefficients with respect to turbocharger speed, For describing the compressor efficiency characteristics, the
thus allowing to extrapolate data also at turbocharger speeds that methodology proposed by [11, 14] has been adopted, which is based
were not experimentally tested. By following this approach, a much on the following definition of isentropic efficiency:
more robust description of the compressor behavior has been
achieved, as shown in Figure 4.
(11)
Where
(12)
(13)
(19)
(14)
(20)
(24)
(25)
(23)
Figure 11. Twin-scroll asymmetric turbine. Possible inlet flow conditions (and
testing configurations). Figure 13. RMF vs ER, twin-entry turbine under equal and single admission
- SCROLL 1.
Due to asymmetry effects, different RMFs are to be expected for the
Figures 13 and 14 show the very different behavior of the same scroll
two scrolls when testing single admission conditions (i.e., conditions
under different admission conditions, demonstrating a high sensitivity
corresponding to center and left pictures of Figure 11), as
to the adjacent scroll instantaneous mass flow, and therefore the need
demonstrated by Figure 12. This poses additional challenges onto
to experimentally characterize the turbine at least under two
interpolation/extrapolation algorithms, but even more critical is the
“extreme” conditions for each scroll: single and equal admission. The
question of how to deal with such wide variety of experimental (or,
same type of experimental facilities as the ones presented in the
better, interpolated/extrapolated) data during engine simulation runs.
Introduction section may be used for such purposes, if minor
In the next figures, ER represents each scroll pressure ratio, while
modifications are made to the test bench to alternatively stop each
ERavg is used for average pressure ratio across the turbine.
scroll’s flow.
Figure 14. Efficiency vs ER, twin-entry turbine under equal and single Figure 17. Twin-entry turbine efficiency characteristic under equal admission
admission - SCROLL 1. conditions - SCROLL 1.
(27)
Figure 20 presents the results of such interpolation/extrapolation Figure 21. Schematic of the interpolation algorithm for estimating twin-entry
algorithm. It should be recalled that equal admission conditions turbine performance.
correspond to PRscroll=1, and in that case the operating condition is
uniquely identified by the average turbine ER (and the interpolating/ To better support this interpolation phase, experimental tests with
extrapolating function on the interaction map is simply given by intermediate admission conditions may result particularly useful, but
PRscroll=1). Relatively large errors are noticeable at high ERs, they require a much more complex hot-gas test-bench, as explained
showing that this simplified approach is not able to completely model in [24]. As an example, Figure 22 shows the extended interaction
the turbine behavior. If a finer interpolation/extrapolation should be map of the same twin-entry turbine of Figure 20, with additional
needed, speed effects may be directly taken into account by experimental and interpolated/extrapolated admission conditions,
identifying different x and y parameters for each iso-speed curve, but corresponding to intermediate MFRs. In particular, purple color is
it should also be observed that such error won't have significant used for a MFR condition equal to 0.30 (greater than 0.00 and smaller
effects during the simulation, since large average ERs are normally than 0.47), while dark red color represents an MFR condition equal to
not reached during engine-turbine interaction. 0.70 (greater than 0.47 and smaller than 1.00).
CONCLUSIONS
The paper presents possible solutions for developing fast and reliable
turbocharger models, to be used mainly for control applications. A
critical review of the most promising (and recent) modeling
approaches for automotive turbochargers is presented in the first part
of the paper, highlighting the main open issues especially in the field
of turbine models, and proposing possible improvements.
12. Serrano, J., Pla, B., Gozalbo, R., and Ospina, D., “Estimation of the RMSE - Root Mean Square Error.
Extended Turbine Maps for a Radial Inflow Turbine,” SAE Technical
Paper 2010-01-1234, 2010, doi:10.4271/2010-01-1234. T - Temperature.
13. De Bellis, V., Bozza, F., Schernus, C., and Uhlmann, T., “Advanced U - Blade tip speed.
Numerical and Experimental Techniques for the Extension of a Turbine
Mapping”, SAE Int. J. Engines 6(3):1771-1785, 2013, doi:10.4271/2013-
24-0119. GREEK SYMBOLS
14. Martin, G., “Modèlisation 0D - 1D de la chaîne d'air des moteurs á
combustion interne dédiée au contrôle”, P.h.D. Thesis, Universite γ - Specific heats ratio.
d'Orleans, 2010. Δ - Variation.
15. Jensen, J., Kristensen, A., Sorenson, S., Houbak, N. et al., “Mean Value
Modeling of a Small Turbocharged Diesel Engine,” SAE Technical η - Efficiency.
Paper 910070, 1991, doi:10.4271/910070.
16. Winkler, G., “Steady state and dynamic modeling of engine
ρ - Density.
turbomachinery systems”, PhD Thesis, 1977 Φ - Non-dimensional flow rate.
17. CANOVA, M., FIORANI, P., GAMBAROTTA, A., and TONETTI,
M., “A real-time model of a small turbocharged Multijet Diesel engine: Ψ - Head parameter.
application and validation.,” SAE Technical Paper 2005-24-065, 2005,
doi:10.4271/2005-24-065.
SUBSCRIPTS
18. Canova, M., Chiara, F., Rizzoni, G., and Wang, Y., “Design and
Validation of a Control-Oriented Model of a Diesel Engine with 0 - Stagnation conditions.
Two-Stage Turbocharger”, SAE Int. J. Fuels Lubr. 2(2):387-397, 2010,
doi:10.4271/2009-24-0122. 1 - Condition at compressor inlet.
19. Hiett, G. f. Johnston, I. H., “Experiments Concerning the Aerodynamic 2 - Condition at compressor outlet.
Performance of Inward Flow Radial Turbines”, Proc. Inst. Mech. Engrs,
178, Pt 31 (II), 1963-64. 3 - Condition at turbine inlet.
20. GT-SUITE Flow Theory Manual, version 7.1, 2010.
4 - Condition at turbine outlet.
21. Eriksson, L., “Modeling and Control of Turbocharged SI and DI
Engines”, Oil & Gas Science and Technology - Rev. IFP, Vol.62 (2007), app - Apparent.
No. 4, pp. 523-538, Copyright c 2007, Institut français du pétrole,
doi:10.2516/ogst:2007042. avg - Average.
22. Watson, N., “Transient Performance Simulation and Analysis of
Turbocharged Diesel Engines,” SAE Technical Paper 810338, 1981,
C - Compressor.
doi:10.4271/810338. corr - Corrected.
23. Fredriksson, C. F., Qiu, X., Baines, N. C., Müller, M. et al., “Meanline
Modeling of Radial Inflow Turbine With Twin-Entry Scroll”, Paper No. ext - External
GT2012-69018, ASME Turbo Expo 2012: Turbine Technical Conference
and Exposition, 2012, doi:10.1115/GT2012-69018. Fl1 - Flow (scroll) 1.
24. Schorn, N., “The Radial Turbine for Small Turbocharger Fl2 - Flow (scroll) 2.
Applications: Evolution and Analytical Methods for Twin-Entry
Turbine Turbochargers”, SAE Int. J. Engines 7(3):1422-1442, 2014, int - Internal.
doi:10.4271/2014-01-1647.
is - Isentropic.
DEFINITIONS/ABBREVIATIONS MAX - Maximum.
Aeff - Effective nozzle flow area. norm - Normalized.
BSR - Blade Speed Ratio. red - Reduced.
CMF - Corrected Mass Flow rate. ref - Reference.
cp - Specific heat at constant pressure. scroll - Scroll (entry).
cs - Velocity achieved by a gas expanding in an adiabatic nozzle. TC - Turbo Charger.
D - Diameter. TT - Total-to-Total.
ER - Expansion Ratio.
h - Enthalpy.
- Mass flow rate.
n - Revolutions per minute.
p - Pressure.
PR - Pressure Ratio.
r - Rotor radius.
R - Gas constant.
RMF - Reduced Mass Flow rate.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or
otherwise, without the prior written permission of SAE International.
Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE International. The author is solely responsible for the content of the paper.