UNIVERSITA' DEGLI STUDI DI GENOVA - ICEG

15 downloads 201 Views 9MB Size Report
Tuttavia, a mano a mano che avanza, il guerriero si rende conto che esistono difficoltà di cui non aveva ... Manuale del guerriero della luce – Paulo Coelho.
UNIVERSITA’ DEGLI STUDI DI GENOVA

FACOLTA’ DI INGEGNERIA Dottorato in Ingegneria Meccanica – Indirizzo Ingegneria delle Macchine a Fluido - XX ciclo

Experimental study on turbocharging systems for automotive engines

Tutor Prof. Ing. Massimo Capobianco Student Silvia Marelli

Un sentito ringraziamento a tutte le persone che mi hanno accompagnato in questo percorso, sostenendomi e standomi accanto. Vi porterò per sempre nel cuore.

“Un guerriero della luce studia con molta attenzione la posizione che intende conquistare. Per quanto il suo obiettivo sia difficile, esiste sempre una maniera di superare gli ostacoli. Egli verifica i cammini alternativi, affila la sua spada, e cerca di colmare il proprio cuore con la perseveranza necessaria per affrontare la sfida. Tuttavia, a mano a mano che avanza, il guerriero si rende conto che esistono difficoltà di cui non aveva tenuto conto. Se rimane ad aspettare il momento ideale, non uscirà mai da quel luogo; è necessario un pizzico di follia per compiere il passo successivo. E così il guerriero utilizza un briciolo di pazzia. Perché, in guerra e in amore, non è possibile prevedere tutto.” Manuale del guerriero della luce – Paulo Coelho

“La più bella sensazione è il lato misterioso della vita. È il sentimento profondo che si trova sempre nella culla dell’arte e della scienza pura. Chi non è più in grado di provare né stupore né sorpresa è, per così dire, morto; i suoi occhi sono spenti.” Albert Einstein

In memoria di mia nonna, che ho perso durante questo percorso, ma che mi ha accompagnato lungo il tragitto. Nel cuore e nello spirito.

Abstract

The activity developed during the PhD work has regarded different aspects related to turbocharging systems in order to improve the knowledge of turbochargers behaviour both under steady and unsteady flow conditions. In a first step the component test facility operating at UNIGE-ICEG and the relevant measuring system have been improved, referring to the capability to investigate components under pulsating flow. In particular a VVA cylinder head, driven by an electric motor, was installed in order to generate pulsating flow upstream the turbine. The investigation was focused on different aspects, such as the extended definition of compressor and turbine steady flow performance, measurement of instantaneous pressure levels at the turbine inlet and outlet, evaluation of instantaneous turbine mass flow rate, the assessment of turbine unsteady flow performance, the development of correlation criteria with steady flow results and the effect of waste-gate valve on turbine behaviour both under steady and unsteady flow conditions. Finally an enhanced approach to model the turbocharger turbine working under unsteady operating conditions, implemented in the GASDYN 1D thermo-fluid dynamic code developed at Politecnico di Milano, was validated against the experimental results.

I

Sommario

Il lavoro svolto nell’ambito del corso di Dottorato di Ricerca ha avuto come oggetto lo studio del comportamento di unità di sovralimentazione per MCI automobilistici in condizioni di flusso di alimentazione stazionario e non stazionario. L’attività ha riguardato, in una prima fase, il potenziamento del banco per prove su componenti del sistema di aspirazione e scarico di motori automobilistici operante presso il DIMSET, al fine di rendere possibili studi a livello di sottosistema mediante il progetto e la realizzazione di un nuovo dispositivo generatore di flusso pulsante basato su una testa motore che permette di riprodurre fedelmente il circuito di scarico del motore sovralimentato. Sono state inoltre messe a punto apposite metodologie di misura da utilizzare nell’ambito di indagini sperimentali in condizioni stazionarie e non stazionarie su sistemi di sovralimentazione per MCI automobilistici. L’attività di ricerca ha riguardato l’approfondimento di diversi aspetti correlati a tale tematica, quali la definizione estesa delle curve caratteristiche del compressore e della turbina di sovralimentazione in condizioni di flusso stazionario, il rilievo delle pressioni e delle portate tempovarianti nel caso di alimentazione della turbina in regime non stazionario, la definizione di idonei criteri per la valutazione delle prestazioni non stazionarie di una macchina motrice, l’impiego di metodologie di correlazione delle prestazioni stazionarie e non stazionarie della turbina e di approcci quasi-stazionari, nonché lo studio dell’effetto della valvola waste-gate sul comportamento della turbina di sovralimentazione con la definizione della ripartizione della portata evolvente attraverso il rotore e la valvola di by-pass e l’analisi della propagazione dell’instazionarietà del flusso attraverso la stessa. Alle estese campagne di prova si è affiancata un’attività teorica relativa alla simulazione non stazionaria del circuito di sovralimentazione utilizzando il codice di calcolo GASDYN, sviluppato presso il Politecnico di Milano.

II

Contents 1. Introduction

1

2. Exhaust turbochargers testing

6

2.1 Turbocharger test facility options 2.2 The UNIGE-ICEG test rig 2.2.1 Pulse generator devices

3. Instrumentation and measurement techniques 3.1 Measuring system: general layout 3.2 Measuring stations 3.3 Acquisition procedures and data processing

6 10 12

18 18 19 22

4. Investigation programme

24

5. Turbocharger performance characteristics

30

5.1 Definition of turbocharger steady flow maps 5.1.1 General remarks 5.1.2 Non-dimensional representation of compressor and turbine performance characteristics 5.1.3 Performance maps 5.2 The radial flow compressor 5.2.1 Compressor characteristics and flow range 5.2.2 Instability phenomena 5.2.3 Choking 5.2.4 Flow range and variable geometry 5.3 The radial flow turbine 5.3.1 Introduction 5.3.2 The volute or inlet casing 5.3.2 Turbine characteristics and flow range 5.3.4 Turbine mass flow regulating devices 5.4 Definition of turbocharger steady flow maps at UNIGE-ICEG

6. Unsteady flow turbocharger turbine behaviour 6.1 Introduction 6.2 Turbine inlet and outlet pressure diagrams

30 30 32 34 36 36 37 41 42 43 43 44 47 49 51

54 54 54

III

6.3 Wave propagation phenomena in the exhaust circuit 6.4 Instantaneous turbine mass flow rate 6.5 Assessment of turbine unsteady flow performance 6.5.1 Effect of pulse parameters on turbine inlet energy 6.5.2 Turbine unsteady flow efficiency

59 63 66 66 72

7. Unsteady turbine performance predicting methods and modelling

76

7.1 Correlation between steady and unsteady turbine performance 7.2 Quasi steady flow (QSF) predicting methods 7.3 Unsteady flow turbine modelling

76 78 81

8. Effect of waste-gate valve opening on turbine performance 8.1 Investigation goal 8.2 Mass flow sensitivity to WG valve setting 8.3 Turbine and waste-gate steady mass flow contributions 8.4 Evaluation of turbine efficiency in the opened WG valve region 8.5 Effect of waste-gate opening on wave propagation phenomena 8.6 Analysis of turbine instantaneous mass flow rate

92 92 92 94 98 100 106

9. Conclusions and further steps

109

Acknowledgments

114

Nomenclature

115

References

118

IV

1. Introduction

The reciprocating Internal Combustion Engine (ICE) will probably remain the dominant propulsion system for automotive applications at least in a medium-term scenario. In the meanwhile, the request of lower fuel consumption and exhaust emissions will become even more severe, moving from the consideration that road transport has been identified as a main source of environmental impact both in terms of chemical pollutants and greenhouse gases. In addition, the increasing cost of fuel will push towards the development of high efficiency powertrain concepts [1, 2]. In order to achieve the goal of near-zero pollutant emissions with low engine fuel consumption, being cost effective, a big effort is required to develop powertrain systems that are able to face the medium-term challenge and represent a bridge between the present solutions and potential long-term (2020 or later) alternatives. Within this frame, it is clear the potential of available technologies integration which can be managed, at least on a subsystems level, by proper control strategies to optimise engine operation and govern the interactions between installed components. Recently the automotive diesel engine was substantially improved, due to the introduction of several technological innovations, first of all electronically controlled fuel injection systems [3] and improved turbochargers fitted with advanced regulating devices (variable geometry turbines) [4]. As a result, modern diesel engines for car application have achieved specific power levels comparable to those typical of gasoline engines, with the advantage of a lower fuel consumption. Consequently, their market share has significantly increased during the last few years, and the request of high specific performance will continue to be a driving factor of success for this engine type. On the other side, the spark ignition engine was the first propulsion system to take advantage of electronic fuel injection systems and is now able to attain, in the case of stoichiometric homogenous combustion, the lowest exhaust pollutant levels by using the three-way catalytic converter. However, gasoline passenger car engines have to significantly reduce fuel consumption to accomplish CO2 emission targets.

1

The automotive spark ignition engine development should therefore address to new solutions which allow a substantial increase of energetic efficiency, particularly at part load operation, while maintaining current specific power and exhaust emissions levels together with an excellent car drivability. Among proposed solutions, the introduction of gasoline direct injection (GDI) alone has to face the problem of simultaneous reduction of fuel consumption and exhaust emissions [5]. A scheme on which an increasing interest of European car manufacturers has been confirmed is based on the combination of homogenous stoichiometric GDI [6] with other technologies, such as fully flexible valve control systems (VVA) [7, 8] and charge boosting, thus allowing for a substantial reduction of engine displacement at constant rated power (engine downsizing) [9]. In any case, the trend to a more complex configuration of the intake and exhaust engine circuit is apparent, with the adoption of specific devices and the need to develop suitable control strategies. As above mentioned, turbocharging is today widely used in automotive diesel engines and the introduction of advanced regulating systems allowed for a significant progress to solve some critical aspects related to the automotive application of turbocharging technique, such as transient response and torque curve configuration. However, the use of turbocharging in spark ignition engines has to face different problems related both to technological aspects (for example those associated to the different exhaust gas temperature level) and to functional characteristics [10, 11, 12]. It is therefore apparent the interest for dedicated investigations on the behaviour both of single intake and exhaust components and of significant subsystems (such as the one made up of exhaust valves, manifolds and turbocharger turbine) in order to get a better understanding of their performance also in unsteady flow conditions and to optimise the relevant control strategies. To this purpose, measurements performed on specific test facilities can supply a lot of information to be used in the development of simulation models and to assess correlation criteria between components behaviour in steady and transient operation [13]. A specialised components test rig has been operating since several years at ICE Laboratory of the University of Genoa (UNIGE-ICEG) [14]. The experimental facility allows to perform investigations on single devices and subassemblies of the intake and exhaust circuit of automotive ICE. The test rig allows a continuous operation by feeding the tested device with compressed air, both under steady flow conditions and with a pulsating flow which reproduces the exhaust gas flow from multi-cylinder engines. The test bench is particularly

2

suited to perform investigations on turbocharging systems, due to the availability of two independent circuits by which the turbocharger compressor and turbine can be fed with controlled flow parameters. During the PhD work several tests on three different turbocharging units for gasoline engine application were developed using the above mentioned experimental facility. The investigation was mainly focused on the turbocharger turbine while the turbocharger compressor was used as a dynamometer. By controlling the compressor supply pressure through a dedicated feeding circuit it was possible to considerably extend the investigated operating range and explore the turbine characteristics in a wider range than that usually considered in the maps supplied by the turbocharger manufacturer, thus limiting extrapolations of performance curves when they are used within theoretical simulation models which take into account instantaneous transient operating conditions. One of the main features of the UNIGE-ICEG facility is the possibility to perform tests under unsteady flow conditions. Referring to this aspect, substantial improvements of the test rig and of the relevant measuring system were carried out during the PhD work in order to allow unsteady flow tests on exhaust turbochargers with different circuit configurations. The previous pulsating flow generating system, based on rotating valves, was maintained in order to allow parametric studies on the turbocharger response to the main pulse parameters. An alternative turbine feeding circuit was designed and installed during the PhD work by which it is possible to supply the turbocharger turbine with the unsteady flow generated by a cylinder head (driven by an electric motor with controlled rotational speed), using also a typical engine exhaust manifold. The new system was equipped with an advanced variable valve actuation (VVA) system (UNIAIR, developed by Centro Ricerche Fiat), which allows to set the valve opening law in a wide range, considering also the case of cylinder deactivation. Therefore, it is possible to analyse the behaviour of the whole subsystem valves-manifoldturbine and the interactions between these components in order to optimise the system geometry and the relevant control strategies, referring to the characteristics of available regulating devices. The circuit was also fitted with a specific device in order to investigate the turbocharger behaviour when a typical engine mass flow transient occurs. The test circuit enhancement was accompanied by the improvement of the relevant measuring system: to this purpose, particular attention was dedicated to the evaluation of the main turbocharger parameters in transient operation, such as the instantaneous turbine mass flow

3

rate and the pressure diagrams upstream and downstream the turbocharger. This required the installation of high response transducers fitted with adequate signal conditioning and data sampling systems (high speed acquisition cards). New data acquisition procedures were also developed in LabVIEW environment to evaluate transient turbocharger performance. The relevant virtual instruments allow to acquire data and calculate the main turbocharger operating parameters. The investigation on the subsystem valves-exhaust manifold-turbocharger were particularly focused on the effect of turbine upstream circuit geometry on the characteristics of unsteady flow supplying the turbocharger, comparing pressure diagrams measured with various line configurations. The influence of the main pulsating flow parameters on the amount of available energy at the turbine inlet was also analysed and different calculation procedures to evaluate turbine unsteady efficiency were considered. Another important aspect which was deepened during the PhD work was the effect of the waste-gate valve on turbine performance both under steady and unsteady flow conditions. With reference to this topic, it is well known that vaneless nozzle radial-flow turbines, which usually fitted turbocharging systems for automotive applications, work within a large range of inlet mass flow rate. Therefore, a dedicated turbine mass flow regulating system is required. In the case of gasoline engines, variable geometry systems can’t be used, due to the harsh thermal exhaust environment. A simply by-pass device (waste-gate valve) is usually adopted as a turbocharger control system due to its effectiveness, low cost and ability to work at high exhaust gas temperatures. However, to optimise engine-turbocharger matching, turbine performance when the waste-gate valve is partially or totally opened should be known, both under steady state and in the typically unsteady flow conditions occurring in automotive engines. Unfortunately, turbocharger manufacturers usually provide very little information about steady flow turbine characteristics in the opened waste-gate field, though this is a fundamental input for theoretical simulation models. To analyse the behaviour of a turbocharger turbine for downsized SI engines when the wastegate valve is opened, an experimental study was performed under steady flow conditions, measuring the mass flow contribution through the by-pass valve and the turbine impeller at different waste-gate openings and comparing these levels to the swallowing capacity of both the rotor and the by-pass device when working alone.

4

The analysis was then extended to unsteady flow turbocharger operation, focusing on the magnitude of wave propagation phenomena in the turbine circuit when the waste-gate valve is opened, also considering turbine outlet pipe geometry. Finally, the effect of waste-gate opening on instantaneous turbine mass flow rate was examined and the relative results were compared with calculated levels assuming quasi-steady flow turbine behaviour. Within the study, the UNIGE-ICEG test facility operating under unsteady flow conditions was also simulated by using the GASDYN 1D thermo-fluid dynamic code developed at Politecnico di Milano [15, 16, 17]. An enhanced model of the turbocharger turbine was implemented and the relevant results were validated against experimental unsteady flow data. The main results of the PhD work can be summarized as follows: 

updated test facility layout;



new experimental system for the generation and control of unsteady flow supplying the turbocharger, fitted with a fully flexible VVA system;



enhanced measuring system with reference both to dedicated instruments to evaluate instantaneous levels of significant turbocharger operating parameters and to data acquisition and performance calculation procedures;



experimental data base on performance characteristics of different turbocharging systems for automotive SI engines, measured in a wide operating range, taking into account the position of fitted regulating device;



experimental results under pulsating flow conditions of the different turbocharging systems, referring to o

measurements of turbine inlet and outlet pressure diagrams

o

wave propagation phenomena in the engine exhaust circuit

o

turbine instantaneous mass flow rate

o

assessment of turbine unsteady flow performance



effect of waste-gate valve opening on turbocharger performance



correlation between steady and unsteady turbine performance



unsteady flow turbine modelling using GASDYN 1D thermo-fluid dynamic code

5

2. Exhaust turbochargers testing 2.1

Turbocharger test facility options

Turbochargers are generally tested separately from the engine on a gas stand test facility which simulates most conditions of engine environment in order to get free from the engine operating condition. A modern turbocharger test facility is a complex interrelated system of architectural, structural, mechanical, electrical and software components. The main objectives for construction of a turbocharger test facility are to support all types of turbocharger testing and to maximise measurement accuracy and test repeatability. The better way to run the units for test is to supply compressed air from an external source, take it through a combustion chamber and then feed it into the turbine. A central compressor station is an obvious economic choice. Reciprocating, centrifugal and lubricated rotary screw compressors can be considered for this application. The reciprocating type offers excellent efficiency, long service life and good controllability but has high initial cost. The centrifugal type is less efficient in this size range, has less controllable range and is also more costly than the rotary screw type compressors. Lubricated screw type compressors are very efficient, are easily controlled and offer lowest purchase cost.

Fig. 2.1 – Gas stand schematic at Schwitzer U.S. Inc. in Indianapolis [18]

6

Fig. 2.2 – Hot gas generator at Schwitzer U.S. Inc. in Indianapolis [18] In Figure 2.1 is shown the layout of the test facility adopted at Schwitzer U.S. Inc. in Indianapolis [18]. Four rotary screw compressors were selected for this specific application; this compression station is able to simultaneously serve four different test benches. A combustion air heating (Fig. 2.2) for each test stand was chosen; both cold and hot air are distributed to each test stand, with final mixing at the test stand to achieve the desired temperature. In this specific application natural gas was selected on the basis of several consideration, although diesel fuel has been used in such test facilities. Natural gas combustion is clean and complete free from airborne particles that could foul turbine parts. The airborne particles are particularly troublesome when testing with the laser. Furthermore, natural gas generates more homogeneous combustion products. Conversely, diesel fuel is prone to uneven injection and greater temperature stratification leading to hot spots that can damage the burner liner and housing. Natural gas is disadvantageous compared to liquid fuel in that it is more difficult to handle safely. The safety drawbacks can be alleviated by careful design of flame safety, leak detection and fuel shutoff systems. It is possible to omit the combustion chamber from the installation, if the inlet air temperature to the turbine can be kept high enough to avoid icing at the turbine exit. This can be achieved by inserting an electrical heater in the supply line: this has the additional benefit that produces clean hot air without any carbon particles, a necessity for any laser anemometry work on the turbine stage. The Table 2.1 mentioned in [18] shows a comparison of the major factor associated with considered air heating system, using natural gas as a reference.

7

Tab. 2.1 – Comparison between different fuel [18] Cost factor Initial investment Fuel operating cost Maintenance costs

Diesel fuel 0.74 1.06 13.50

Natural Gas 1.00 1.00 1.00

Electric Heating 1.35 2.81 1.67

Another important issue to take into account is the procedure to absorb turbine power. Test rig investigations are usually performed using turbocharger compressor as a dynamometer. Since the compressor operating range is limited by chocked flow and surge, specific experimental techniques are required to widen the definition of turbine characteristics, especially as regards tests performed at low inlet air temperatures. Alternatively, high speed dynamometers (Fig. 2.3) can be used to absorb turbine power [19, 20]. These devices can considerably extend the measurement of turbine characteristics and in some cases [20] also allow the actual torque produced by the turbocharger bearings to be evaluated. The main drawbacks of turbine dynamometers are complexity of design, restricted rotational speed range (usually below 80.000 rpm) and difficulties in coupling them to different turbines. Additionally, torque values are extremely low, thus requiring very sensitive measuring equipment. The low level of torque means that any small frictional effects produce errors: to avoid them, the dynamometer used by Winterbone at al. [20] (Fig. 2.3) had the absorbing element mounted on air-bearings to minimise friction. The device shown in Figure 2.3 is based on the turbine housing and rotor of an existing turbocharger, and hence it was necessary to mount turbines with a similar rotor. It would obviously be preferable to fit the turbine wheel to the dynamometer by a more universal attachment device, and this was done by Dale and Watson [19].

Fig. 2.3 – Block diagram representation of a turbine dynamometer [20]

8

The design of a totally flexible turbocharger test rig also has to consider that the turbine usually operates under unsteady flow conditions and sometimes with partial admission. This adds further complexity to the test facility, requiring the introduction of a pulse generator system upstream of the turbine and the use of high speed measuring systems. The possibility of performing tests in unsteady flow conditions is an essential pre-requisite for investigating the effect of the main pulsating flow parameters on turbine performance, developing suitable comparison criteria with steady flow data and analysing the results obtained from theoretical models and quasi-steady flow prediction procedures [20]. The performance of a turbine under unsteady flow conditions can be measured on a rig similar to that for measuring steady flow characteristics, but with the addition of a pulse generator upstream the turbine, as shown in Figure 2.4. The pulse generator can take a number of forms. The one used by Winterbone et al. [20] consists of a cylindrical rotor with two rectangular passages cut through it, which rotates in a block with the appropriate inflow and outflow passages. The rotor is made of two parts which can be moved relative to each other and hence the pulses from the pulse generator can be out of phase for each entry of a twin entry turbine.

Fig. 2.4 – Rig for measuring the steady and unsteady flow characteristics of a turbine [20]

9

Fig. 2.5 – Non-steady flow rig [22] The pulse generators used by other researchers, such as Dale and Watson [19] and Arcoumanis et al. [21] used a disc rotor to generate the pulses. The advantage of the latter type of pulse generator is that the slots in the rotor can be shaped to provide pulses of different profile; this is more difficult with cylindrical rotor. Benson and Scrimshaw [22] used a cylinder pulse generator (Fig. 2.5). In this scheme each cylinder is supplied by a high pressure and a low pressure air supply through two separately timed valves controlled by a camshaft. A second camshaft controlled the exhaust valve timing. With this configuration it is also possible to study turbine fitted with a two entries. Also Osnaghi et al. [23] used a cylinder head to generate pulsating flow upstream the turbine.

2.2

The UNIGE-ICEG test rig

The test rig operating at UNIGE-ICEG is a continuous flow apparatus which allows tests on single components and subassemblies of automotive engines intake and exhaust circuit. The facility is particularly suitable to develop investigations on exhaust turbochargers due to the availability of two independent supply lines. A schematic of the experimental apparatus is illustrated in Figure 2.6. Two different air compression stations are available, one of which consists of three electrically driven screw compressors providing a total mass flow rate of about 0.6 kg/s at a maximum pressure of 8 bar. As an alternative, a single stage centrifugal compressor with a delivery of up to 2.2 kg/s and a maximum compression ratio equal to 2.1 can be used. A

10

DC

SC

SC

SC

AR AF PC

AF

LM

AF

PC

AH FM

Pulse genera tor system

C

T LC

AF AH AR C DC FM

Air Filter Air Heater Air Receiver Compressor Dynamic Compressor Flow Meter

LM LC PC SC T

Laminar flow Meter Lubricating circuit Pressure Control Screw Compressor Turbine

Fig. 2.6 – Experimental test facility proper regulating system makes it possible to supply both the turbine and the compressor supply circuit with air at controlled pressure levels. The compressor is used as a dynamometer in order to investigate turbine characteristics over an extended range. The turbine feeding line is fitted with a 54 kW electrical heater to moderately raise the air temperature (up to 400 K) in order to avoid any condensation and freezing problem during the expansion. During the PhD work, the test rig upgrade related to a new pulse generator system required a substantial improvement of the plant lubricating circuit. Therefore this system was upgraded in order to allow turbocharger and cylinder head lubrication with controlled oil temperature and pressure levels at each component entry. The restructured layout of the test rig lubricating circuit (represented in Figure 2.7) is based on two independent lines respectively dedicated to the turbocharger and the cylinder head. The oil mass flow rate supplied to the turbocharger is measured in order to estimate the relevant mechanical losses.

11

Alarm oil tank

PC

M

Turbocharger line Cylinder head line

°C

°C

°C

PA PA

°C °C

Turbocharger

Cylinder head

°C

°C

M

°C

Thermocouple

Oil heater

PCi

Regulating pressure system

Filter

°C

Platinum resistance thermometer

Heat exchanger

PA

Pressure alarm

M

Electric motor

Manometer

One-way valve

Oil flow meter

Motor driven valve

Valve

Pump

Fig. 2.7 – Test rig lubricating circuit layout 2.2.1 Pulse generator devices Turbine performance can also be investigated under unsteady flow conditions by using different pulse generator systems. Two turbine feeding line arrangements fitted with different pulse generators are used to simulate engine operation. The first layout (Arrangement A) was designed to perform parametric studies on the effect of the main unsteady flow parameters on turbine performance. The second outline (Arrangement B) was planned to more precisely replicate turbocharger unsteady flow operation when matched with an automotive engine and to extend experimental investigations to a subsystems level.

12

Rotating valves pulse generator system In the first circuit arrangement (Fig. 2.8) pulsating flow is generated by dedicated rotating valves with a diametral slot [24, 25, 26, 27]. This element consists of a cylindrical rotor with a diametral slot revolving inside a stator fitted with inlet and outlet ports. Different flow area diagrams can be generated by replacing the rotor and the stator ports. This configuration makes it possible to control the main pulsating flow characteristics, such as mean value, amplitude, shape and frequency of pressure oscillations at the turbine inlet. Tests on single and two-entry turbines can be developed, with thermodynamic parameters controlled independently at each entry. For each turbine entry two separate branches are provided, in one of which a pulsating flow is generated by a dedicated rotating valve. Valve rotational speed can be adjusted to allow investigations in the typical pulse frequency range of the exhaust system of high-speed multi-cylinder engines (10250 Hz). The main pressure pulse parameters (amplitude and mean value) at each turbine entry can be controlled by correctly mixing two flow components (steady and pulsating) in a Y-junction and adjusting upstream plenum pressure. For two-entry devices, unequal admission and out-of-phase pulses can be reproduced [26]. Cylinder head pulse generator system In the second arrangement, designed and developed during the PhD work, a cylinder head is

Turbine air supply

Plenum

Rotating valves

Compressor air supply Turbocharger

T

C

Fig. 2.8 – Pulse generator system with rotating valves

13

used to generate the turbine inlet pulses with a view to more realistically replicating turbocharger unsteady flow operation when working on the engine (Fig. 2.9). The cylinder head is mounted on a flow distributor (Fig. 2.10) which reproduces the reference engine cylinder block. This layout is particularly suited for investigating the effect of exhaust subsystem features on turbine performance, i.e.: different valve opening profiles or exhaust manifold geometries. A variable rotational speed electrical motor is used in order to adjust pulse frequency over the typical range of automotive engine intake and exhaust systems (10200 Hz). In a first step an Alfa Romeo cylinder head for two litres engine (model M646 IDB13) was mounted on the flow distributor. Since the intake side was used as the exhaust one, a cylinder head arrangement provided with intake ducts characterised by reduced equivalent diameter (23 mm) (Fig. 2.11) was selected. Within the study, the system was upgraded with the introduction of a cylinder head fitted with a Variable Valve Actuation (VVA) system (UNIAIR, developed by Centro Ricerche Fiat). The VVA control system required the design and manufacturing of dedicated wheels for phase and rotational speed sensors, the introduction of a dedicated power supply system for solenoid valves and the improvement of measuring system with the introduction of oil temperature sensors. Particular attention was dedicated to the relative position between the phase and the speed wheel and to the location of the relevant sensors in order to comply with

Plenum

Turbine air supply Flow distributor

Engine head

Compressor air supply

Manifold T

C Turbocharger

Fig. 2.9 – Pulse generator system with cylinder head

14

Fig. 2.10 – Flow distributor (cylinder block) ECU requirements. The UNIAIR Control Unit, managed by EDS software, uses crankshaft and camshaft signals in order to operate the solenoid valves; EDS allows to set valve opening and closing timing independently each other. The experimental facility was further upgraded in order to reproduce engine load transients (referring to mass flow rate). The new device was based on throttle valves acting in the flow distributor (Fig. 2.12). This system allows: 

simultaneous valves operation;



independent valve position setting;



small pressure losses when fully opened.

The throttle valves driving mechanism is based on an external rack sliding on two guides. Adjustable screws allows to set relative valve phasing.

Fig. 2.11 – Intake side of cylinder head (model M646 IDB13) (on the left) and particular of intake duct (on the right)

15

Fig. 2.12 – Particular of the throttle valves regulating device Through the VVA system any valve opening profile can be reproduced, thus modifying unsteady flow characteristics in the exhaust circuit, the geometry of which can be easily changed by installing different manifolds. Within the investigation a first real engine exhaust manifold was installed downstream of the pulse generator. The most applied configuration for modern 4-cylinder automotive engines (Fig. 2.13) was chosen, made up of four small diameter pipes (equal to 24.8 mm) of the same length, joining in a reduced volume mixing section (36 cm3). During the study described in the section 6.3, aimed at investigating the effect of different circuit arrangements on flow unsteadiness, some modifications of the manifold were carried out by reducing its volume, thus removing the mixing volume. With the substantial upgrade of the test facility performed during the PhD work, experimental investigations on turbocharging systems developed at UNIGE can be addressed to: 

investigate the behaviour of specific turbocharger configurations in pulsating flow conditions, highlighting the influence of the main flow parameters on the components performance;

Figure 2.13 Exhaust manifold

16



study the transient response of the turbocharger and of the whole exhaust subsystems in order to optimize the geometry of the circuit and the relevant control strategies.

17

3. Instrumentation and measurement techniques 3.1

Measuring system: general layout

The test rig is fitted with a measuring system which allows to perform investigations on intake and exhaust internal combustion engine components both under steady and unsteady flow conditions. Dedicated measuring stations were provided in different sections in order to acquire thermodynamic parameters such as pressure, temperature and frequency through dedicated transducers (§ 3.2). Figure 3.1 shows a schematic of the measuring system used within the investigations performed during the PhD work. Measurements were performed by a PC-controlled data acquisition system using interactive procedures in LabVIEW® environment (§ 3.3). In steady flow conditions average static pressures are evaluated by capacitive transducers, while temperatures are measured by platinum resistance thermometers. Turbine mass flow rate is estimated by a laminar flow

Digital volmeter

Transducers signals

Digital counter HP-5316A Dig. stor. oscill. Gould OS-4020

Instrument Bus (IEEE 488)

Data acquis. unit HP-3497A

AD converter IOTECH ADC 488 GP-IB interface NI PCI-6110 NI PCI-6250M

Personal Computer

Fig. 3.1 – Test rig measuring system

18

meter, while a sharp edged orifice is fitted on the compressor supply line to assess the relevant mass flow rate. Turbocharger rotational speed and pulse frequency are measured by inductive probes. Turbine regulating system, such as a waste-gate valve or a variable nozzle, is also measured by a variable resistance transducer generally referring to the angular position or to the linear displacement of the relative driving shaft. Under pulsating flow operation, instantaneous pressures and turbine mass flow rate are also evaluated. High frequency response straingauge pressure transducers are mounted near the duct walls in order to eliminate signal modifications caused by the connecting lines. Instantaneous mass flow rate is estimated through a constant temperature hot-wire anemometric system providing a measuring frequency of up to 10 kHz. A triggering signal provided by a magnetic sensor makes it possible to start data acquisition at a definite position of the rotating valve pulse generator. For the pulse generator system fitted with a cylinder head, the triggering signal is provided by a Hall sensor mounted on the driving cam shaft, phased 40 crank angle degrees after TDC related to cylinder no.1 overlap. Data acquisition is governed by a PC which controls an HP-3497A unit (via an IEEE 488-1975 interface). Transducers signals are detected by a digital voltmeter or a frequency meter and transmitted to the computer. In order to simultaneously acquire different pressure signals in transient conditions two high frequency data acquisition cards (NI PCI-6110, allowing measurements on 4 channels with a measuring frequency up to 5 MS/s for each channel and NI PCI-6250M, enabling measurements on 16 channels with a measuring frequency up to 1,25 MS/s for single channel and 1 MS/s in the case of multichannel acquisition) are used.

3.2

Measuring stations

When testing a turbocharger unit, dedicated measuring stations are used; their location is shown in Figure 3.2: 

Station1 : compressor inlet



Station 2: compressor outlet



Station 3: turbine inlet



Station 4: turbine exit



Station 5: turbine mass flow rate meter



Station 6: compressor mass flow rate meter



Station 7: measuring stations on manifold branches

19

DC

SC

SC

SC

AR AF PC AF LM

AF PC

AH

sect. 5

PULSE GENERATOR SYSTEMS AR

AR

RV RV

CB CH

sect. 6

FM

sect. 7

sect. 3

sect. 1

C

T sect. 4

sect. 2

LC

AF air filter AH air heater AR air receiver C compressor CB cylinder block CH cylinder head (VVA) DC dynamic compressor

FM flow meter LM laminar flow meter LC lubricating circuit PC pressure control RV rotating valve SC screw compressor T turbine

Fig. 3.2 – Schematic of UNIGE components test facility Stations 1  4 These stations are used to evaluate thermodynamic parameters upstream and downstream of turbocharger turbine and compressor and are fundamental in order to characterise turbocharger unit. Each measuring station is fitted with temperature and pressure taps. Wall static pressure is evaluated through purpose designed taps. The total pressure level is then calculated starting from volumetric flow rate assuming a uniform velocity distribution in the measuring plane. In the case of steady flow measurements, several wall pressure measurements are usually performed through small diameter taps (Fig.3.3) and the relevant lines are connected in parallel. In this specific application, pressure transducer is kept at a distance from the tap.

20

Fig. 3.3 – Pressure tap (steady flow condition) Under pulsating flow conditions taps geometry is very important in order to avoid signal distortion related to the natural frequency of the connecting volume between the duct wall and the transducer membrane. An optimised geometry is therefore used (Fig.3.4), the natural frequency of which was estimated about 800 Hz. [28]. Station 5 Average turbine mass flow rate is measured in this plane through a laminar flow meter. Two devices are available, characterised by different capacity (100 and 200 l/min), which can work independently or in parallel. Volumetric turbine flow rate is calculated starting from pressure drop across the laminar matrix. In order to evaluate mass flow rate, the air density is calculate starting from temperature and pressure measurements downstream the meter. Station 6 This measuring station is dedicated to the evaluation of the compressor mass flow rate

Fig. 3.4 – Pressure taps (unsteady flow condition)

21

through a sharp edged orifice. Inlet section pressure and temperature and pressure drop across the orifice are measured. Compressor mass flow rate is then evaluated. Station 7 In the case of unsteady flow investigations performed using the cylinder head pulse generator (§ 2.2.1) this measuring station is dedicated to the evaluation of pressure signals in the manifold branches and/or in the mixing volume (§ 6.3).

3.3

Acquisition procedures and data processing

A dedicated programme in LabVIEW environment was developed in order to control the data acquisition process and to calculate performance parameters both under steady and pulsating flow operation. LabVIEW is a graphical programming language that uses icons instead of lines of text to create applications. In the case of investigations under pulsating flow conditions, the sampling rate is chosen according to the pulsating flow frequency, in order to define signal at least with 150 points over the period. Since periodic oscillations in the intake and exhaust circuit of automotive engines are characterised by a fundamental frequency varying between 10200 Hz, therefore the acquisition system must be characterised by a high frequency response (about 1020 kHz). During the PhD work, an acquisition card (NI PC 6250M) was introduced in the measuring system. This card allows measurements on 16 channels with a measuring frequency up to 1,25 MHz for single channel and 1 MHz for multichannel acquisition. Multichannel acquisition is not synchronous, since a single AD converter is provided. The time delay between two following channels is td 

1 f s N CH

where fs = sampling frequency NCH = number of acquisition channels For this reason, modifications to acquisition procedures have been necessary in order to phase acquired signals. Instantaneous signals are then processed using various filtering techniques. During the PhD

22

work, a specific analysis tool was set up to allow the post-processing of instantaneous signals. Within this tool, a Fast Fourier Transform algorithm is used in a preliminary step in order to detect harmonic components due to mechanical or electric noise and allow signals prefiltering. A better definition of instantaneous signals is then achieved by suitable averaging techniques applied to measured signals. All signals were acquired for several complete cycles and appropriate averaging techniques were then applied to the measured data, such as the definition of a mean cycle or the adoption of a mobile mean method (centred on three or five points). In Figure 3.5 an example of post-processing Virtual Instruments developed in LabVIEW environment is shown.

Fig. 3.5 – Example of LabVIEW Front Panel

23

4. Investigation programme

The experimental investigations developed during the PhD work have been carried out on the test rig operating at University of Genoa. Studies on different turbochargers were in particular worked out both under steady and unsteady flow conditions. A detailed description of turbocharging units used during investigations performed is reported below. The activity developed has regarded in particular: 

UNIGE-ICEG test rig upgrade to allow steady and fully flexible dynamic investigations on components (turbochargers) and engine subsystems (§ 2)



Methodology for testing automotive turbochargers (§ 3-5)



Steady flow characterization of different TC units (§ 5)



TC turbine behaviour in pulsating flow conditions o

Measurement of turbine inlet and outlet pressure diagrams (§ 6.2)

o

Wave propagation phenomena in the engine exhaust circuit (§ 6.3)

o

Turbine instantaneous mass flow rate (§ 6.4)

o

Assessment of turbine unsteady flow performance (§ 6.5)



Unsteady turbine performance predicting methods (§ 7.1-7.2)



Unsteady flow turbine modelling (§ 7.3)



Effect of waste-gate valve opening on TC performance (§ 8)

As above mentioned different turbochargers were considered during the PhD work; in particular the investigations were developed on the following units: 

Garrett GT2052 ELS (referred in the work as TC1)



Borg Warner KP39 (referred in the work as TC2)



IHI RHF3 (referred in the work as TC3)

The main features of tested turbocharging units are here reported.

24

Garrett GT2052 ELS (TC1) Turbocharger Garrett GT2052 ELS (Fig.4.1) is a current production TC unit matched for application on a 2-liter gasoline engine. Single entry nozzleless radial flow turbine is characterised by 47 mm rotor diameter and by TRIMt =(D22a/D21a)t equal to 0.72 (Fig. 4.2). The turbocharger was installed on the UNIGE-ICEG test rig and connected to the feeding circuit through dedicated adaptors (Fig.4.3). A waste-gate valve (WG) is used as turbine mass flow control device.

Fig. 4.1 – Garrett GT2052 ELS

s

a D1

Dt b D2

D2a

Fig. 4.2 – Rotor turbine

25

Fig. 4.3 – GT2052 ELS installation on test rig Borg Warner KP39 (TC2) This TC unit (Fig.4.4), matched for downsized gasoline engines application, has the typical architecture of a small - medium size turbocharger: 

radial inflow turbine (centripetal)



radial outflow compressor (centrifugal)



carter for bearing

Regulating system (necessary to control over-speed, over-boost and engine torque) consists in a waste-gate valve. Turbine, characterized by TRIMt =(D22a/D21a)t equal to 0.69 and by a rotor diameter of 38.40 mm, is a single entry nozzleless radial flow. Dedicated adaptors fitted with optimised measuring stations were designed and manufactured. Figure 4.5 shows the installation of KP39 at UNIGE-ICEG test facility.

Fig. 4.4 – Borg Warner KP39

26

Fig. 4.5 – Installation of KP39 at UNIGE-ICEG test facility IHI RHF3 (TC3) IHI turbocharger (model RHF3) is matched to a downsized SI automotive engine with a capacity of approximately 1.4 litres. The turbocharger (Fig.4.6) is fitted with a single entry

Fig. 4.6 – Installation of IHI RHF3 at UNIGE-ICEG test facility (on the right) and particular of turbocharger unit (on the left)

27

nozzleless radial flow turbine and uses a waste-gate valve as a turbine mass flow control device. The assembly was installed on the UNIGE-ICEG test rig and connected to the feeding circuit through dedicated adaptors. Turbine is characterised by a TRIMt =(D22a/D21a)t equal to 0.81 and by a rotor diameter of 32.40 mm. Also in this case different adaptors were manufactured, providing in particular different turbine downstream circuit arrangements. Three different turbine exhaust pipe configurations were designed and made up with rapid prototyping techniques (Fig. 4.7) in order to study the effect of circuit geometry on turbine behaviour, with special reference to the interactions between the flow passing through the

a

b

c

Fig. 4.7 – Turbine exhaust pipe configurations fitted on IHI turbocharger

28

turbine rotor and the one through the waste-gate valve (§ 8). A reference exhaust pipe arrangement (Fig. 4.7-a) is based on a convergent flow duct without any physical separation between the flow from the waste-gate and that from the rotor. In a second arrangement (Fig. 4.7-b) a short dividing wall separates the two flow components in order to avoid

an

immediate mixing of them at the rotor exit. A third solution (Fig. 4.7-c) is based on a dividing wall extended along a considerable length of the exhaust pipe to guide the two flow components to a proper downstream mixing section.

29

5. Turbocharger performance characteristics 5.1

Definition of turbocharger steady flow maps

5.1.1 General remarks As regards steady flow operation, there are a number of difficulties connected with measuring turbocharger characteristics in a broad range. This is a basic requirement, especially for the turbine that usually operates under unsteady flow conditions and instantaneously experiences expansion ratio levels varying in a substantially wider range than that considered in the maps provided by the turbocharger manufacturer. The ICEG test facility allows an extended definition of components steady flow characteristics to be made. This aspect is particularly important in the case of turbocharger compressors and turbines which are generally simulated within engine global models using steady flow curves. In Figure 5.1 turbocharger turbine steady flow characteristics (referred to mass flow rate) measured on the test rig operating at the University of Genoa are represented taking into account different rotational speed levels and waste-gate openings referring to TC1 turbocharger (§ 4). It is worth to notice the extension of measured curves which was achieved through suitable experimental techniques: this is apparent in Figure 5.1 where the operating range considered in the maps supplied by the TC manufacturer is represented by the red portion of plotted curves. Besides, the test rig investigation allowed to define turbine performance also in the case of partially opened waste-

0,105 GARRETT GT2052 ELS

Corrected Mass Flow Rate [Kg/s]

0,095 A = 20%

0,085 A = 0%

0,075

0,065

TURBINE 0,055

Rot. s peed factor [rpm/K 0.5] 3000 4000 5000 6000

0,045

Correct ed rot. speed [rpm] 50900 67900 84850 101800

0,035 1

1,2

1,4

1, 6

1,8

2

2,2

2,4

2,6

Pressure Ratio (Total/Static)

Fig. 5.1 – Turbocharger turbine mass flow characteristic

30

gate operation (A = 20 %), while the TC manufacturers don’t provide this information. Extrapolation of turbine curves can generate significant errors, particularly in the field of the lowest expansion ratios. Consequently, if a quasi-steady flow assumption is used to predict pulsating performance, a detailed definition of turbine steady flow curves is essential for improving calculation accuracy. To this end, the possibility to control the compressor supply pressure in a broad range, thus modulating its power absorption, allows the experimental definition of turbine curves to be considerably extended. If further expansion of measured characteristics for very low levels of the expansion ratio is required, specific devices can be used. As an example, an original impulsive system acting on the compressor side of the turbocharger is presented in [24]. This allowed the measurement of turbine characteristics near zero mass flow conditions to be significantly extended without modifying the turbocharger rotary assembly. In this specific application the compressor housing was replaced by a new dedicated casing, fitted with three injector nozzles, fed with compressed air. No modifications of turbocharger rotary assembly were necessary. The nozzle position was set in order to direct the air impulse to the compressor rotor blades, keeping the usual direction of rotation. The positive torque contribution on the turbine shaft can be modulated by adjusting the nozzle supply pressure, and therefore the air flow rate through the injection system. By this impulsive device it was possible to importantly extend the measurement of

Fig. 5.2 – Turbine steady flow characteristics extended in the low range [24]

31

the turbine characteristics near zero mass flow conditions (Fig. 5.2). Besides, experimental information on the effect of regulating system setting (waste-gate valve or variable geometry device) on turbine characteristics is a fundamental input for theoretical simulation models and a pre-requisite for developing effective turbocharger control strategies. 5.1.2 Non-dimensional representation of compressor and turbine performance characteristics When studying the performance characteristics of turbomachinery it is a great help if results from different sources are directly comparable, even for machines of somewhat different size. It is advantageous that the behaviour of turbomachines can be illustrated using dimensionless parameters involving all relevant variables. For example the mass flow rate, the efficiency and the temperature rise of a compressor can be expressed as functions of the independent variables: M , , T0  f ( p01 , p02 , T01 , n, D, R, k ,  ) where n = rotational speed D = diameter R = gas constant k

cp cv

= specific heat ratio

μ = dynamic viscosity 0 = stagnation condition 1 = inlet section 2 = outlet section These can be reduced to a set of non-dimensional groups: .

M ( RT01 ) T p nD M , , 0  f ( , 02 , , k) 2 p01D T01 ( RT01 ) p01  D Since turbochargers operate on a specific gas (air for the compressor, exhaust gas for the turbine), the values of R and k are specified. Hence the non-dimensional groups become

32

.

M ( RT01 ) T p nD M , , 0  f ( , 02 , ) 2 p01D T01 ( RT01 ) p01  D .

 M  Fortunately the Reynolds number of the gas   has little effect on the performance of the  D 

machine and can be usually ignored. Hence .

M ( RT01 ) T p nD , , 0  f ( , 02 ) 2 p01D T01 ( RT01 ) p01 A relationship between the efficiency, the temperature rise and the compression ratio can be expressed through the following equation k 1

 p02  k   1 p01    T0 T01

cTT

For a particular machine the diameter is constant and therefore may be ignored. Thus the complete performance of the compressor may be represented by the relationships (for a specific fluid, such as air) .

M T01 n p02 and   f ( , ) p01 T01 p01 or .

M T01 T0 n p02 and  f( , ) p01 T01 T01 p01 By plotting compression ratio against the mass flow parameter for a series of values of the speed factor, the complete performance of the machine is represented. Lines of constant efficiency may be superimposed on this map. A typical compressor map is shown in Figure 5.3. A detailed explanation is given in § 5.1.3. A similar performance map may be drawn for a turbine, although naturally the curves will appear very different. A disadvantage of removing the length term (or diameter) and gas constant is that the mass flow and speed parameters are no longer dimensionless. The major advantage of the presentation is that it illustrates the

33

performance independently of the inlet conditions (pressure and temperature) and can therefore be used for the compressor at any time and place. 5.1.3 Performance maps It is conventional to plot compressor characteristics in terms of pressure ratio against mass flow parameter for lines of constant speed parameter as in Figure 5.3. In addition, contours of constant isentropic efficiency are superimposed. There are essentially three areas on a compressor map. The central area is the stable operating zone. This area is separated from the unstable area on its left by the surge line. A detailed explanation of the causes of surge has yet to be fully accepted, but it is clear that when the mass flow rate through a compressor is reduced while maintaining a constant pressure ratio, a point arises at which local flow reversal occurs in the boundary layers. This should result in low efficiency but not necessarily in instability. If the flow rate is further reduced, complete reversal occurs. This will relieve the adverse pressure gradient until a new flow regime at a lower pressure ratio is established. The flow will then build up again to the initial condition and thus flow instability will continue at a fixed frequency. Surge is actually a more complex phenomenon than that described, but it is sufficient at this stage to realise that the compressor

Fig. 5.3 – Compressor characteristic [29]

34

must not be asked to work in the area to the left of the surge line. The area to the right of the compressor map is associated with very high gas velocity. It is the result of choking of the limiting flow area in the machine. Extra mass flow through the compressor can only be gained by higher speeds. This additional mass flow will certainly be limited by the ability of the diffuser area to accept the flow. When diffuser chocking occurs, compressor speed may rise substantially with little increase in mass flow rate. The area of maximum efficiency naturally falls in the central stable operating zone. In practice it tends to lie in an area roughly parallel to the surge line with vaneless-type diffusers and very close to the surge line in the case of vaned-type diffusers. In the case of radial flow turbine, due to the centrifugal field created by the rotor, there is a noticeable spread of constant speed parameters for chocked flow conditions, as shown in Figure 5.4. Since the operational area of the turbine occupies such a restricted area on the pressure ratio/mass flow parameter map, it become simpler to present efficiency on a separate diagram. It is conventional to plot efficiency against velocity ratio (blade speed ratio). This is the wheel tip speed in the case of the radial flow turbine, divided by the velocity equivalent of the isentropic enthalpy drop across the turbine stage. Lines of constant pressure ratio (total to total or total to static) or sometimes constant speed parameter are drown on the map. This method of representation is important in matching the compressor and the turbine wheel size to ensure operation of the turbine at optimum efficiency (at constant pressure operation). It is

Fig. 5.4 – Radial flow turbine characteristic [29]

35

usual to include the turbocharger mechanical (bearing) losses in the turbine efficiency since it is difficult in practice to separate them.

5.2

The radial flow compressor

5.2.1 Compressor characteristics and flow range As mentioned above, the important operating area on the compressor map was presented as the stable operating zone separated from the unstable area on the left by the surge line and on the right by the choke conditions. In the specific application of the compressor to a turbocharger, in particular for truck or automotive-type engines requiring a wide range of operation, the achievement of a wide flow range between surge and choke is as important as high efficiency. Compressor flow range is commonly defined as the difference between the surge and choke line at a given speed, expressed as the percentage of the maximum flow rate (choke) (Fig. 5.3). If a compressor with a radial vaned impeller is run at constant speed and the mass flow rate through it is controlled by means of a delivery valve, the shape of the constant speed line on the pressure ratio-mass flow rate curve might be expected to be that shown in Figure 5.5. The specific energy transfer (E) will remain roughly constant as mass flow rate increases or reduces, since it is a function of tip speed and slip factor (if pre-whirl or backsweep are not used); it is: .



W E  U 2  C 2   U 22 M where (referring to Figure 5.6) 

W = energy transfer in the impeller M = mass flow rate . U 2 = impeller tip speed

C 2 = tangential component of velocity .

C 2 = slip factor (in a radial vaned impeller) U2 Frictional losses in the impeller and diffuser passages must increase with mass flow rate due



to the increased velocities. The other principal loss will be due to incidence effects as the air flow angle at the impeller eye and inlet deviates from the blade angle. Incidence loss will be

36

Fig. 5.5 – Compressor constant speed line shape [29]

least at the design point, but will increase quite rapidly as the mass flow rate (and hence flow angle) increases or decreases. If the incidence and frictional losses are both subtracted from the energy input, the curve shown in Figure 5.5 is obtained. The point at which the energy transfer is greatest will occur at a mass flow rate somewhat less than might be expected from the incidence analysis above. Point A represents a steady condition for zero mass flow rate at a pressure ratio generated by centrifugal force. 5.2.2 Instability phenomena In a diffusing flow there is the possibility that the flow at the wall (the boundary layer) is retarded so severely that it can no longer follow the wall surface. In other words, the kinetic

Fig. 5.6 – Relative flow through radial vaned impeller [29]

37

energy in the boundary layer is not sufficient to overcome the adverse pressure gradient leading to flow separation at the wall, described as a local stall. Similarly the effect of secondary flows, for example in bend ducts, can cause flow separation (stall) leading to flow instability interfering with the diffusion process. The instability occurring in the centrifugal compressor can be identified as: 

component stall;



stage instability or stage stall;



system instability (described as surge).

When the mass flow rate through the compressor is gradually reduced at constant speed, a breakdown in the flow process occurs leading to instability. At the inducer inlet, as the mass flow decreases, the axial component of the absolute velocity decreases, thus increasing the incidence angle of the air approaching the leading edge of the inducer. Beyond a critical incidence angle the flow can no longer adhere to the suction side of the inducer blade (Fig. 5.7). Flow separation from the surface creates a stall condition subsequently encouraging reversal of flow. Inducer stall may contribute to surge, particularly at high pressure ratios, but can exist at a nominally stable operating condition. If there is non-uniformity in the approach flow to the inducer or in the geometry of channels between the inducer blades, breakdown in the flow in one channel (for example “b”, Fig. 5.8) causes air to be deflected in such a way that channel “a” receives fluid at an increased incidence angle. Channel “a” then stalls, causing reduction of incidence angle to channel “b” restoring normal flow. The consequence is that the stall passes from one channel to the next channel. This rotating stall may or may not lead to stage instability, but can introduce aerodynamically induced vibrations resulting in increased noise level. The flow separation occurring on the suction side of the blade in the radial portion of the impeller (Fig. 5.9) leading to the formation of a wake, can be described as impeller stall. Use of high blade

Fig. 5.7 – Flow separation in the inducer [29]

38

Fig. 5.8 – Rotating stall schematic [29] backsweep can reduce the wake region and provide more stable operation of the diffuser. The most common separation of flow in the vaneless diffuser is caused by local reversal of flow normally occurring on the shroud side of the diffuser wall and is much influenced by the flow leaving the impeller. To conserve angular momentum, the flow in a parallel walled vaneless diffuser tries to maintain a constant flow angle. However, due to compressibility and viscosity effects, the streamlines close to the wall have less kinetic energy and follow a path of much reduced spiral angle until eventually they are swept back towards the impeller. This backflow effect is a diffuser stall, which can be avoided by choosing larger impeller exit flow range. It becomes more difficult to obtain a wide flow range at high pressure ratios since the impeller tip speed increases. At compressor pressure ratios above 4:1, the flow leaving the impeller becomes supersonic, losses can increase and the flow range narrows substantially. The vaneless diffuser can maintain shock-free diffusion with a supersonic inlet velocity. However, there comes a point, when operating at constant speed with ever-reduced mass flow rates, that causes a critical component or a combination of a number of components to stall, introducing a strong reversal of the flow. The stage cannot continue to operate stably and the phenomenon is described as stage stall. When the disturbance (stage stall) becomes periodic and grows to a large magnitude, the system can become self-exciting. This phenomenon is then a system instability described as surge. During surging a noisy and violent flow process can occur causing cyclic periods of backflow through the whole compressor. Operation in surge not only drastically reduces the performance of the compressor but can damage it and its installation. The surge frequency and amplitude are dictated by the size of the volumes of the installation before and after the compressor, their flow characteristics and compressor speed. Surge tends to occurs where the constant speed parameter becomes horizontal as indicated by point “C” in Figure 5.5:

39

.

d  cTT 0 dM .

p02 is the pressure ratio. p01 Consider a compressor connected between two constant pressure reservoirs with a delivery where  cTT 

valve and operating at point D on the negative slope of the speed line (Fig. 5.6). If a sudden closure of the delivery valve temporarily reduces the flow, an increase in the delivery pressure from the compressor and a reduction in compressor flow occurs. The former encourages a larger mass flow rate through the delivery valve, reducing compressor delivery pressure and increasing compressor flow. This is therefore self-compensating and an inherently stable system. However, if the compressor were operating at point C, a reduction in mass flow rate would result in reduced compressor delivery pressure reducing flow through the delivery valve, moving the operating point further and further to the left. The mass flow reduction will be so great that the pressure upstream of the delivery (for example, inlet manifold of an engine) falls below the compressor delivery pressure (whose minimum is governed by the centrifugal pressure rise at zero flow). Mass flow will then increase until the system is drawn back to operating point “C”, and the whole cycle will repeat. This analysis implies that surge will occur when the slope of the constant speed parameter curve is positive (that is, to the left of point “C”), but it is not necessarily so. If the flow characteristic of the throttling system has a slope which is greater than the positive slope of the compressor speed line (Fig. 5.10) then as mass flow rate reduced, the fall-off in compressor delivery pressure is less than that of the system and hence the flow system stabilises immediately. Surge will occur only at a point when the positive slope of the speed line exceeds that of the delivery system. Usually the flow

5.9 – Flow separation: meridional plane (on the left) and blade-to-blade (on the right)[29]

40

Fig. 5.10 – Effect of throttle characteristic slope [30] characteristic of the inlet manifold system has only a small positive slope, hence surge occurs when the slope of the compressor speed characteristics is zero or slightly positive (with vaneless diffuser) or slightly negative (with vaned diffusers), due to the smaller flow range. When the turbocharger compressor is connected to an engine inlet manifold, the volume of the manifold is often not sufficient to damp out the pressure fluctuations arising from periodic suction strokes of the pistons. When two or more turbochargers are connected to a common inlet manifold there is a danger of one compressor pushing the other into surge due to variation in energy supplied to the turbine or small differences in their compressor speed characteristics. For this reason the inlet manifolds on engines with multi-turbocharger installations are sometimes kept separate. 5.2.3

Choking

The other limit to the flow range is due to choking, when the flow reaches the velocity of sound at some cross section. In centrifugal compressors, choking normally occurs at the throat of the impeller eye (inducer) or at the entry to a vaneless diffuser or at the throat of a vaned diffuser. When choking occurs at the impeller eye it is the relative velocity which equal the speed of sound. Assuming one-dimensional isentropic flow, Dixon [31] derived the relationship for choking mass flow rate for a swirl-free impeller as k 1

2  2  (k  1) U12 / a01  2( k 1) M  A1  01  a01    k 1  

where 01 = p01/(RT01) is the inlet stagnation density a01 = (kRT01)1/2 is the inlet stagnation sonic velocity A1= effective inducer inlet throat area

41

U1= tangential velocity By approximating the thermodynamic conditions at the impeller eye to the thermodynamic conditions at the inlet machine the relation becomes as

M  T01 p01

k 1

 A1

2  2( k 1) k  2  (k  1) U12 / a01   R  k 1 

The choking mass flow rate at the diffuser increases less rapidly with rotational speed than the choking mass flow rate of the inducer. In order to choke the vaneless diffuser the radial component of the absolute velocity entering the diffuser must reach sonic velocity. If the flow stagnation conditions are known at the inlet of the vaneless diffuser, then simple onedimensional flow calculations can predict the choked vaneless diffuser mass flow rate. 5.2.4 Flow range and variable geometry An important factor that effects flow range of the centrifugal compressor is the choice of diffuser, vaneless or vaned, and the operating pressure level. The flow range of a vaned diffuser is narrower than that of a vaneless diffuser as shown in Figure 5.11. This is mainly attributable to diffuser vane stalling when the air entry angle departs significantly from its design value. Since the choke line of the vaneless diffuser is limited by the impeller tip exit area, the only remaining way of extending flow range is to move the surge line to the left on the compressor map (toward smaller mass flow rates). Rapid reduction of vaneless diffuser width increases the radial component of the absolute velocity, delaying stall. A more uniform flow distribution at the impeller exit can also improve stable operation of the diffuser. Blade backsweep may delay separation of the flow on the critical suction side of the impeller blade

Fig. 5.11 – Vaneless and vaned diffuser compressor characteristics [32]

42

passage and hence raise compressor efficiency and produce a larger compressor mass flow rate. Variable geometry is another way of increasing the flow range of the centrifugal compressor. Pre-whirl vanes or diffuser vanes, whose angle can be altered while the compressor is running, can both be used to increase flow range. Use a variable geometry does introduce a complication in compressor design and manufacture since in addition to the movable compressor components themselves a control mechanism is required.

5.3

The radial flow turbine

5.3.1 Introduction There are two basic type of turbine suitable and used at present in turbochargers, the radial flow and the axial flow. The radial flow is mainly used for small automotive or truck turbochargers; the axial type is commonly used for the large turbochargers applied to medium-speed stationary and railway traction engines and large marine engines. The radial flow is similar in appearance to the centrifugal compressor, but with the flow in the inward direction and nozzle vanes replacing the diffuser vanes. Its great advantage is that it maintains a relatively high efficiency when reduced to very small size. The turbocharger turbine is required to accept quite unfavourable inlet conditions if the pulse turbocharging system is used. The inlet might be divided into two or three separate sectors, each containing highly pulsating flow. It is indeed fortunate that the turbine is able to accept these conditions without a complete deterioration on its performance. The radial inflow turbine consists of a scroll or inlet casing (Fig. 5.12), a set of inlet nozzles (sometimes omitted) followed by a short vaneless gap and the turbine wheel itself. Most small turbocharger turbines use a nozzleless casing to improve flow range at some penalty in peak

Fig. 5.12 – Components of a radial flow turbine [29]

43

efficiency, but also reducing cost. However, considering the more conventional type with nozzles, the function of the inlet casing is purely to deliver a uniform flow of inlet gas top the nozzle entries. The nozzle accelerate the flow, reducing pressure and increasing kinetic energy. Energy transfer occurs solely in the impeller, which should be designed for minimum kinetic energy at the exit. A detailed description on the volute turbine is reported below due to its significance related to the regulating device. 5.3.2 The volute or inlet casing The function of the volute will depend on whether the turbine has inlet nozzle vanes. If it has, then the volute must simply deliver a uniform gas flow to the nozzles. A single entry volute casing is shown in Figure 5.13. In addition is desirable that the flow angle does not depart far from the entry angle of the vanes, but since the velocity is low, losses will not be large if it does. A spiral volute casing should be beneficial. This can be designed using incompressible flow theory with constant angular momentum (and hence uniform pressure). Thus, from Figure 5.13 rC  constant=K For incompressible flow, the mass flow rate at an azimuth angle  is

M   A C For uniform mass flow distribution

M  M 

 2

where M=total mass flow rate entering the volute. Hence

Fig. 5.13 – Volute type turbine scroll [29]

44

A

M    C 2

Eliminating C by using rC  constant=K gives

A

M  r    2 K

that is, the cross-sectional area of the volute reduces with the azimuth angle and mean radius. The mean radius will vary as the volute curves in round the circumference, but the reduction is not large when expressed as a fraction of radius. In consequence, the area is often reduced linearly with azimuth angle. The above analysis has been developed for the case of a singleentry casing with the flow area reducing to zero. An identical analysis can be applied to each half of a double-entry volute. Under the pulse turbocharging system a double or even tripleentry casing may be required to isolate the gas flow from each separate group of manifolds. This isolation holds throughout the inlet casing and the nozzle ring, and hence must be achieved by means of separate angular sectors (Fig. 5.14). It is a consequence that one entry passage will be longer than the other if a common flange is used at the exhaust manifold connection. Naturally the nozzles must be arranged to preserve the isolation between sectors as shown in Figure 5.14. It is common, whether the casing is single or double entry, to allow some reduction in flow area between the exhaust manifold flange and the point at which the scroll first admits to the turbine (cross-sectional area A1, Fig. 5.14). This results in some flow acceleration and a compromise must be reached between the frictional losses that result from high velocities and the benefit of reduced turbocharger size and weight due to the smaller

Fig. 5.14 – Cross- section of a turbine with a double entry [29]

45

casing. If the inlet nozzles are eliminated, the casing or volute must take over the function of flow guidance at entry to the rotor in addition to flow distribution and acceleration. The geometry of the volute must determine the gas angle at the volute exit or rotor tip (4). From Figure 5.13

cot  

C r C

Denoting the effective volute inlet as station 1 (the point at which the scroll first admits to the turbine, Fig. 5.13), with r1 being the radius of the centroid of section A1, r1C1  r4C 4 From conservation of mass M  1 A 1 C1  4 2 r4b4Cr 4 where b4 is the rotor inlet width. Thus

cot  4 

1 A1C 1 1 A1r4 A  1   1 1 4 2 r4b4C 4  4 2 r4b4 r1 r1  4 2 b4

For incompressible flow

cot  4 

A1 1 r1 2 b4

For a specified rotor tip width (b4), the volute exit gas angle is defined by the ratio A1/r1. This value is often used to define casing geometry since it directly relates to the vane angle of an equivalent vaned inlet casing. It’s important to note that flow angle is independent of mass flow rate, hence the vaneless volute is only slightly better able to adjust to varying flow rates than a vaned housing. If the cross-sectional area is reduced linearly with azimuth angle  (or even the area divided by the mean radius), quite large variations in flow angle (4) result. For the purpose of matching the turbocharger for a particular engine duty, a method of adjusting the flow angle at the rotor tip is required. If inlet nozzles are used it is simple to replace the nozzle ring by another in which the vanes are set at different angle. If a nozzleless casing is used, the flow angle can be altered by increasing or reducing the cross-sectional area of the volute (that is, area A1). The cross-sectional area of the scroll will be altered in proportion, all the way round the rotor. Thus a range of nozzleless inlet casings having different effective areas (or A1/r1) or a range of nozzle rings set at different blade angles are required for final matching of the flow characteristics to those of the engine.

46

5.3.3 Turbine characteristics and flow range The performance of a radial flow turbine is best defined by curves showing the relationship between expansion ratio, isentropic turbine efficiency and pseudo-non-dimensional mass flow rate and speed. A typical turbine characteristic, showing constant speed parameter plotted against the non-dimensional mass flow rate and expansion ratio is given in Figure 5.4. The speed dependence principally arises from the effect of the centrifugal field created by the speed of the rotor. At low rotational speeds a certain inlet pressure is required for mass to flow through the turbine in the correct direction. At higher rotor speeds a greater inlet pressure will be required, since a higher centrifugal pressure field will oppose the gas motion. A turbocharger turbine is required to handle a very wide mass flow range in automotive application where the engine operates over a broad speed and load range. The flow range of the radial flow turbine is limited by choking at high pressure ratios. No exact low flow limits exists, since in the ultimate situation the radial turbine can acts as a compressor. The maximum flow through the turbine is limited by the choking conditions in the nozzle throat (or throat of the nozzleless casing) or rotor exit. The critical pressure ratio between the inlet and the nozzle throat, when flow at the latter is sonic, is a function of the gas specific heat capacity ratio (k=Cp/Cv). Since turbine flow characteristics become asymptotic to the choking conditions, and the rotor exducer may also choke, the turbine is almost choked at a considerably lower pressure ratio. The effective area of the vaned nozzles depends on the vane angle and hence the choked mass flow will be altered by a change in nozzle angle. If a nozzleless inlet casing is used, the effective throat area will be dependent on the gas angle and exducer area. Gas angle control is achieved by varying the cross-sectional area of scroll, since this increases or decreases the tangential component of velocity. Thus the choked mass flow rate will be a function of the effective area of the scroll and can be varied by changing scrolls. As important as choked flow rate is need for high efficiency over a wide flow range. Although a turbine with inlet nozzles usually achieves a higher peak efficiency than one with a nozzleless casing (due to better and more uniform flow distribution and guidance around the rotor tip), the latter is better equipped to cope with off design conditions. Regardless of mass flow rate, the use of inlet nozzles will constrain the flow angle at entry to the rotor. As the flow rate moves away from design conditions, the inlet velocity will change in magnitude, not

47

direction. Thus the relative velocity approaching the impeller tip will depart from the radial direction, creating an incidence loss. If a nozzleless casing is used, the flow angle can adept to the flow rate. An incidence loss will still result, but the incidence angle (and hence the loss) will be less. Thus the nozzleless turbine has a wider efficient flow range and is commonly used on turbochargers for engines working over wide speed and load ranges. Not only does the choice of nozzle angle or scroll affect the choked mass flow rate, but, since it changes the inlet velocity triangle, it affects the efficiency map. Changes in nozzle angle or casing will move the area of optimum efficiency across the working map of the turbine. If the changes in area or angle are large, the peak value of efficiency will probably also drop, since incidence losses will increase at what was previously the design point. However, changes of this type do provide an extremely useful method of tailoring the performance of a turbocharger to the requirements of a particular engine. From the economic point of view it is desirable to produce a single turbocharger unit that can cover a wide range of engine sizes and hence of mean mass flow rates. To adapt a basic design of the turbine such that it can provide high efficiency over a very large mass flow range involves changing the effective choking area. There are three principal ways in which this can be done: 

varying the angle of nozzle blades (see Figure 5.15a) or varying the turbine inlet casing cross-sectional area for nozzleless turbine;



cropping the trailing edge of the nozzle blades to increase effective throat area (Fig.

Fig. 5.15 – Variation of turbine area [29]

48

5.15b) 

axial cropping of the nozzle ring and rotor together with the provision of suitable sized turbine casings (Fig. 5.15c)

The third method has the most drastic effect on throat and rotor exducer area and hence mass flow rate, with the least loss in turbine efficiency. For these reasons a manufacturer may offer a basic design of a turbocharger with two or three rotors and corresponding turbine inlet casings to cover a large range of engine sizes. This can provide up to 50 per cent change in mean mass flow rate. To obtain optimum efficiency in between these rotors options 1 or 2 may be used. Option 2 is the simplest one, but it increases losses associated with the change in trailing edge thickness and nozzle gas outlet angle, and introduces a larger gap between nozzle and rotor tip. Changes in nozzles or nozzleless casings, and rotors, are commonly called “trim” changes within a certain “frame size” (turbine casing, centre casing and bearing) of the turbocharger. Together with similar trim change on the compressor this provides a very versatile and adaptable turbocharger unit not only to move the area of highest efficiency to the desired engine operating conditions but also to adapt the turbocharger unit to a reasonably wide range of engine sizes. 5.3.4 Turbine mass flow regulating devices Although the flow range of a particular turbine trim is large, operation away from the design point results in a loss of efficiency. It is also important the fact that the expansion ratio of the turbine will vary greatly with mass flow rate (Fig. 5.4). Since compressor pressure ratio and turbine expansion ratio are directly related, boost will be low at low mass flow rates and high at high mass flow rates. This has serious consequences for an engine designed to operate over a large speed range. The engine may have insufficient boost at low speeds and be over-boosted at higher speeds. Since a change in nozzle angle modifies the throat area of the nozzles (Fig. 5.15a), and therefore controls the expansion ratio for a certain mass flow rate, it appears attractive to use adjustable nozzle vanes. The performance of a turbine with variable nozzle vane angles can be illustrated by superimposing characteristics measured at various fitted settings (Fig. 5.16). Clearly major benefits are obtainable in terms of increased pressure ratio at low mass flow but, in practice, variable geometry has to face some difficulties essentially mechanical, such as obtaining

49

Fig. 5.16 – Variable nozzle angle: mass flow characteristics [29] reliable working for long periods of time while exposed to the action of high temperature and corrosive exhaust gas. This is the reason why this regulating system is used mainly on diesel engine, which is characterised by lower exhaust gas temperature than in the case of SI engine, with the exception of rare applications. For SI engine application a waste-gate valve (Fig. 5.17) is usually fitted as a turbocharger regulating system due to its effectiveness, low cost and ability to work at high exhaust gas temperature. It consists in a by-pass valve of the exhaust gas which is not transferred to turbine in order to decrease turbine available energy. Control system can be designed using the boost pressure as a pneumatic controller. An experimental study on this subject has been developed during the PhD work and the relevant

Fig. 5.17 – Waste-gate valve regulating system

50

Fig. 5.18 – Variable Area Turbine (on the left) and Variable Nozzle Turbine (on the right) results are presented in chapter 8. Figure 5.18 shows two different variable geometry turbine; Variable Area Turbine is a regulating system which properly control the volute cross-sectional area A1 (mentioned above) while in the case of a Variable Nozzle Turbine nozzle vane angles can be regulated.

5.4

Definition of turbocharger steady flow maps at UNIGE-ICEG

Three different turbochargers (see chapter 4) were tested on UNIGE test rig during the PhD work. The experimental definition of compressor and turbine steady flow maps in an extended range was performed, taking also into account different waste-gate valve settings. Figure 5.19 shows a typical measured compressor map referred to TC1 turbocharger (§ 4). Figure 5.20 shows turbine steady flow characteristics of TC2 unit (§ 4). Turbine constant speed characteristics are reported in terms of mass flow factor and overall efficiency (defined as turbine total-to-static isentropic efficiency multiplied by turbocharger mechanical efficiency) vs. total-to-static expansion ratio. The investigation was extended to several turbine speed factor levels, ranging from 2000 to 7000 rpm/√K. Different waste-gate openings (A) were considered for each turbine speed, being this parameter related to the linear displacement of the by-pass valve push rod and defined as a percentage of total rod displacement, ranging from 0 to 100 percent when varying the waste-gate flow area from zero to maximum. Four different waste-gate settings were selected (corresponding to A equal to 0, 5, 20 and 100 percent) on the basis of a preliminary investigation on the effect of the by-pass valve opening on turbine mass flow rate (§ 8.2).

51

COMPRESSOR

2.00

Corrected rot. speed [rpm] 1.95

60000 80000 100000 110000 120000 130000 140000

1.90

1.85

1.80

1.75

1.70

=0.65

Pressure Ratio (Total/Total)

1.65

GARRETT GT2052 ELS

=0.75

1.60

=0.70 1.55

1.50

= 0.74

=0.72

1.45

1.40

=0.70 =0.60 = 0.72

1.30

1.25

=0.50

=0.65

1.35

=0.60 = 0.55

=0.45 =0.40

1.20

=0.35 1.15

1.10

=0.30

=0.25 =0.20

1.05

=0.05

=0.15 =0.10

1.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.20

Corrected Mass Flow Rate [Kg/s]

Fig. 5.19 – Compressor map measured at ICEG laboratory The extension of measured characteristics achieved by controlling the compressor supply pressure while working at almost constant turbine inlet air temperatures (about 400 K) can be clearly seen. The operating range explored proved to be considerably wider than that considered in the maps provided by the turbocharger manufacturer and allowed a satisfactory definition of steady flow curves to be achieved for application within simulation models. A substantial increase in turbine mass flow rate (approximately 50-60 percent) was observed when the waste-gate valve is fully opened. However, a detailed analysis developed at constant turbine rotational speed (n/√TT3 = 4000 rpm/√K) and expansion ratio (εtTS = 1.32) confirmed [25] a substantially higher sensitivity (defined as the ratio of change induced in the output, i.e., mass flow rate, to the related input change, i.e., push rod position) of this turbocharger regulating system in the field of lower by-pass settings. When opening the waste-gate valve, overall turbine efficiency (defined with reference to the ideal work related to an isentropic expansion of the whole mass flow from stagnation conditions at the housing entry to outlet static pressure) decreased definitely (Fig. 5.20). This result is mainly related to thermodynamic losses connected with the wasted enthalpy of the

52

1.7

Turbine mass flow rate factor

1.5 1.3 1.1 0.9

A [%]

0.7

0 5 20 100

0.5

n TT3

 rpm     K  2000 3000 4000 5500 7000

0.3 1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

2.6

2.8

Expansion ratio (total to static) 0.7

Turbine overall efficiency

0.6

0.5

0.4

0.3

0.2

1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

2.6

2.8

Expansion ratio (total to static)

Fig. 5.20 – Turbine steady flow characteristics [33]

by-passed working fluid, even though a contribution to the efficiency drop may be due to the flow pattern change in the turbine rotor induced by the by-pass opening, affecting the relative energy losses [34].

53

6. Unsteady flow turbocharger turbine behaviour 6.1

Introduction

In automotive applications, the turbocharger turbine usually operates under unsteady flow conditions and it is extremely difficult to evaluate pulsating performance since several parameters can affect measured data. The possibility of performing tests in unsteady flow conditions is an essential pre-requisite for investigating the effect of the main pulsating flow parameters on turbine performance, developing suitable comparison criteria with steady flow data and analysing the results obtained from theoretical models and quasi-steady flow prediction procedures [20]. The measurement of turbine performance under unsteady flow conditions is much more complex than in the steady flow case. The intake and exhaust circuit of automotive engines is subjected to periodic oscillations characterised by a fundamental frequency varying between 10 and over 200 Hz. It’s therefore apparent that the acquisition system must be characterised by a high frequency response (with a sampling rate of about 10  20 kHz for an accurate definition of the signal profiles) and with a poor sensitivity to noise. All transient parameters should be measured using fast response transducers and integrated in a high speed data acquisition system. It’s also important to take into account the geometry of measuring taps (especially for pressure evaluation) that should be optimised in order to avoid signal distortions due to the natural frequency of the sampling system [35].

6.2

Turbine inlet and outlet pressure diagrams

In this section a detailed analysis of pressure signals evaluated through the two test rig configurations available at ICEG will be presented, including the effect of waste-gate valve opening. Figure 6.1 shows measured pressure diagrams at the turbine inlet and outlet when using the two available pulse generator systems, referring to TC1 unit (§ 4). The same average operating parameters are considered, referring to a typical low load and speed engine condition (n/√TT3 = 3000 rpm/√K, f = 66.67 Hz, p3m=1.38 bar). The non-dimensional time scale is in both cases referred to the period of the main harmonic component of the generated

54

1.6

1.4

1.2

f = 66.67 Hz

1.6

A=0

1.2

1.0

Turbine inlet

1.0

Arrangement A

0.0

0.5

1.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

0.0

1.1

1.1

1.1

1.1 pressure [bar]

pressure [bar]

p3m = 1.38 bar

1.4

Turbine inlet

1.0

1.0

Arrangement B

0.5

1.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

1.0

1.0 Turbine exit

Turbine exit

Arrangement A

0.9

n/ eTT3 = 3000 rpm/ eK

1.8 pressure [bar]

pressure [bar]

1.8

0

0.5

1

Arrangement B

1.5 2 2.5 time/period

3

3.5

4

0.9 0.0

0.5

1.0

Fig. 6.1 – Measured turbine inlet and outlet pressure diagrams with different test circuit configurations [36] pulsating flow. The waste-gate valve was kept completely closed (A = 0). Measurements with the rotating valve pulse generator system (arrangement A) were performed without any steady flow contribution in order to attain the maximum pulse amplitude. It is apparent that inlet pulse amplitude resulted substantially lower in the case of the arrangement B (cylinder head pulse generator system). Even if the different flow area diagram of the two pulse generators (rotating valve and cylinder head) shouldn’t be neglected, the main reason of this result seems related to the geometry of the turbine feeding circuit. With reference to this aspect, the small pulse amplitude with the arrangement B is not connected to the damping effect of the volume interposed between the pulse generator device and the turbine entry. In fact, the related volume with the arrangement A resulted considerably higher (over 3 liters) than that with the arrangement B (about 1 liter). On the contrary, the exhaust manifold geometry seems to play a fundamental role to understand this result. In Figure 6.2 pressure signals simultaneously measured in the different manifold branches are represented, referring to a higher engine load condition (n/√TT3 = 4000 rpm/√K, p3m=1.6 bar). The same pulsating flow frequency of Figure 6.1 was selected (f = 66,67 Hz, corresponding to the same engine

55

1.9

1.8

1.8 pressure [bar]

pressure [bar]

1.9

1.7

1.6

1.5

1.7

1.6

1.5 Manifold no.1

0.5

1.0

1.5 2.0 2.5 time/period

Manifold no.2

3.0

3.5

1.4 0.0

4.0

1.9

1.9

1.8

1.8 pressure [bar]

pressure [bar]

1.4 0.0

1.7

1.6

1.5

1.0

1.5 2.0 2.5 time/period

n/ eTT3 = 4000 rpm/ eK f = 66.67 Hz

3.0

3.5

4.0

p3m = 1.6 bar A=0

1.7

1.6

1.5 Manifold no.3

1.4

0.5

0

0.5

1

Manifold no.4

1.5 2 2.5 time/period

3

3.5

4

1.4 0.0

0.5

1.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

Fig. 6.2 – Experimental pressure signals in the exhaust manifolds [36] rotational speed) in order to better understand the limited pulse amplitude at the turbine inlet obtained with the arrangement B. Recorded waveforms proved to be similar: in each diagram the pressure pulse associated to the related cylinder valve opening is detectable. However, pressure oscillations at higher frequency than the main harmonic component of the pulsating flow are present, with amplitudes often comparable with the main pulse. In order to highlight the propagation phenomena in the engine exhaust circuit, pressure diagrams in different sections of the upstream turbine line were measured with the arrangement B. In Figure 6.3 signals respectively measured in one manifold branch (Fig. 6.3a, corresponding to cylinder no.1), in the exhaust manifold mixing volume (Fig. 6.3b) and at the turbine inlet (Fig. 6.3c) are represented, for the same operating conditions of Figure 6.2. By comparing Figures 6.3a and 6.3b the above mentioned effect of the manifold geometry is clear. As widely reported in the open literature [29, 37], the junction of the four manifold pipes in one mixing volume causes a substantial reduction of downstream flow unsteadiness, with a conversion of potential energy contained in the pressure pulses into kinetic energy, partially reconverted into pressure before the turbine inlet (as confirmed by the slightly higher average level of the relevant pressure diagram, see Figure 6.3c). In the meanwhile, reflected waves are

56

n/ eTT3 = 4000 rpm/ eK

p3m = 1.6 bar

f = 66.67 Hz

A=0

pressure [bar]

1.8 1.7 1.6 1.5 1.4 0.0

a)

0.5

1.0

pressure [bar]

1.8

1.5 2.0 2.5 time/period

3.0

3.5

4.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

1.7

1.6

1.5 b)

1.4 0.0

0.5

1.0

pressure [bar]

1.8

1.7 1.6

1.5 c)

1.4 0.0

0.5

1.0

Fig. 6.3 – Measured pressure diagrams in the upstream turbine circuit [36] present in each manifold branch (Figs. 6.2 and 6.3a), generated at the open end boundary of each pipe. This aspect should be carefully taken into account in the design of the engine exhaust manifold, since both the specific energy at the turbine inlet and the turbine conversion efficiency are affected by flow unsteadiness, usually related to pressure pulse amplitude [38]. To this purpose, different manifold geometries were considered in a stage of the investigation (§ 6.3). The waste-gate opening effect on pressure diagrams measured upstream and downstream of the turbocharger turbine was also preliminarily investigated. Figures 6.4 and

57

2.4

Turbine inlet f = 50 Hz A=0

2.4

p3m = 1.56 bar

2.0 1.8 1.6

1.8 1.6 1.4

1.2

1.2 0.5

1.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

0.0

1.3

pressure [bar]

1.2

1.1

1.0

0.9

0.8

0.5

1.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

3.0

3.5

4.0

1.3 Turbine exit f = 50 Hz A=0

1.2 pressure [bar]

2.0

1.4

0.0

Turbine inlet f = 50 Hz A = 20 %

2.2 pressure [bar]

pressure [bar]

2.2

n/ eTT3 = 4000 rpm/ eK

Turbine exit f = 50 Hz A = 20 %

1.1

1.0

0.9

0

0.5

1

1.5 2 2.5 time/period

3

3.5

4

0.8 0.0

0.5

1.0

1.5 2.0 2.5 time/period

Fig. 6.4 – Effect of the waste-gate opening on measured inlet and outlet pressure diagrams (f = 50 Hz) [36] 6.5 show the relevant results referred to two different pulse frequency levels (50 and 100 Hz) in the case of the circuit arrangement A. Recorded signals with the by-pass valve completely closed (A = 0) are compared with those measured at partial waste-gate opening (A = 20%). The signal shape proved to be affected by the pulse frequency. The wave action in the circuit pipes significantly influences the inlet waveform at the lower frequency level, reducing the pulse amplitude at constant mean value. The effect of the waste-gate opening on signal shape seems negligible, as it can be observed by comparing the corresponding diagrams for the two by-pass system conditions. The same considerations can be extended to the analysis of turbine downstream waves. In any case it is apparent that the pulse amplitude at the turbine exit resulted significant, especially when the reflection phenomena don’t affect the pulse shape. This effect may be due to the small volume associated to the turbocharger turbine which moderates the typical damping effect of this component. The assumption of considering a negligible flow unsteadiness at the turbine outlet, often used within quasi-steady flow calculations of turbine performance, doesn’t seem acceptable in the case of a small turbocharging unit for automotive application, resulting in significant errors in the evaluation of instantaneous turbine expansion ratio.

58

2.6

2.6 Turbine inlet f = 100 Hz A=0

2.4

2.4 pressure [bar]

p3m = 1.56 bar

2.2 pressure [bar]

n/ eTT3 = 4000 rpm/ eK

2.0 1.8 1.6 1.4

2.0 1.8 1.6

1.2

0.0

0.5

1.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

0.0

1.4

0.5

1.0

1.5 2.0 2.5 time/period

3.0

3.5

4.0

3.0

3.5

4.0

1.4 Turbine exit f = 100 Hz A=0

1.3 pressure [bar]

1.3 pressure [bar]

2.2

1.4

1.2

1.2 1.1 1.0 0.9 0.8

Turbine inlet f = 100 Hz A = 20 %

Turbine exit f = 100 Hz A = 20 %

1.2 1.1 1.0 0.9

0

0.5

1

1.5 2 2.5 time/period

3

3.5

4

0.8 0.0

0.5

1.0

1.5 2.0 2.5 time/period

Fig. 6.5 – Effect of the waste-gate opening on measured inlet and outlet pressure diagrams (f = 100 Hz) [36] As regards the effect of the waste-gate opening on measured pressure diagrams, an increased pulse amplitude was detected with opened by-pass device, as a consequence of higher total mass flowing through the system. The relatively higher raise of downstream signal amplitude is probably related to the significant transmission of flow unsteadiness through the by-pass orifice. A more detailed analysis on the effect of the waste-gate valve opening on turbine performance will be presented in § 8.

6.3

Wave propagation phenomena in the exhaust circuit

The effect of turbine feeding circuit geometry in unsteady flow conditions was investigated referring to a typical arrangement of exhaust line for small automotive engines. Tests were performed using arrangement B. A real engine exhaust manifold (Fig. 2.13, § 2.2.1) was installed downstream of the pulse generator, coupled to TC2 turbocharger (§ 4). The exhaust manifold configuration was the most applied in modern 4-cylinder automotive engines, made up of four small diameter pipes (equal to 24.8 mm) of the same length, joining in a reduced volume mixing section (36 cm3). In the baseline arrangement of the exhaust

59

circuit (configuration 0), an adaptor (length 103 mm, volume 166 cm3) and a small volume connecting pipe (length 81 mm, volume 76 cm3), where the turbine inlet measurement plane was located, were interposed between the manifold and the turbine. Pressure diagrams were simultaneously measured in different sections of the upstream and downstream turbine circuit, referring to typical operating conditions corresponding to intermediate engine speed and load. As regards the upstream line portion, static pressure taps were located in each manifold pipe (close to the cylinder head), in the manifold mixing volume and at the turbine entry. Downstream pressure was evaluated in the fitted outlet pipe, close to the turbine exit. The study was aimed at investigating the effect of different circuit arrangements on flow unsteadiness, with particular reference to the pulse characteristics at the turbine entry. To this purpose, subsequent modifications of the baseline circuit were carried out: as a first step, the flow pipes finishing was improved (configuration no.1) and the interposed adaptor was optimised by reducing its volume of about 60 per cent (configuration no.2). Both the modifications had no significant impact on measured pressure diagrams. This is probably related to the acceptable finishing of the original manifold flow pipes and to the small overall reduction of turbine upstream circuit volume (from 1252 cm3 to 1158 cm3). In a later stage of the investigation, the effect of a flow dividing wall included in the circuit portion between the manifold branches and turbine inlet was analysed, thus avoiding an immediate mixing of pulsating flow components from the engine in order to extend the wave propagation close to turbine entry. Different lengths of the dividing wall were tested, extending the flow separation up to the manifold exit (configuration no.3) or until the turbine volute casing inlet. In this case, taking into account the engine firing order, both the situations of equally spaced pulses in each divided duct (configuration no.4) and of partially overlapped pulses (configuration no.5) were investigated. Figure 6.6 shows a typical pressure diagram measured in one of the manifold branches (cylinder no.1). The pressure pulse due to the relevant valve opening is evident, but appreciable oscillations at higher frequency were also detected. These are the result of waves reflection and transmission in manifold pipes. It is interesting to note that each subsequent valve opening (referred as EO in Figure 6.6) causes a noticeable modification of pressure profile which is detected in the measurement plane of manifold no.1 with an average delay (about 12 crank angle degrees) corresponding to the time required for wave propagation.

60

2.0

cyl3EC cyl2

EO

pressure [bar]

cyl1

EO

1.9

cyl4

EO

EO

cyl3 EC

EC

EC

1.8

1.7

1.6 0

90

180

270

360

450

540

630

720

crank angle [deg]

Fig. 6.6 – Measured pressure diagram in one of manifold pipes (cylinder no.1) [39] Manifold pressure diagrams proved to be significantly influenced by the inclusion of a dividing wall in the downstream line as regards the portion of pressure profiles more affected by disturbances from other branches (Fig. 6.7), due to the change of wave paths. The relative position of the dividing wall in the duct also affected measured pressure profile, as it is apparent by comparing the results referred to circuit configurations no.4 and 5. The junction of four manifold pipes in one mixing volume caused a substantial decrease of flow unsteadiness, resulting in a reduced pulse amplitude at the turbine inlet [36]. Figure 6.8 shows how the presence of a dividing wall extended up to the turbine entry allowed to increase the pulse amplitude in this section only if equally spaced pulses in each duct are

pressure (referred to mean value) [bar]

determined (configuration no.4), while in the case of partially overlapped pulses 0.10

0.05

0.00

-0.05

Configuration 2

-0.10

Configuration 4 Configuration 5

-0.15 0

90

180

270

360

crank angle [deg]

450

540

Fig. 6.7 – Effect of circuit configuration on manifold pipe pressure diagrams [39]

61

1.9

1.7

1.6

0

180

360

540

720

1.9 cylinders 1-4

cylinders 1-2

cylinders 2-3

cylinders 3-4

1.8

pressure [bar]

1.8

pressure [bar]

pressure [bar]

1.9

1.7

1.6

0

180

crank angle [deg]

360

540

crank angle [deg]

1.8

1.7

720 1.6 0

180

360

540

720

crank angle [deg]

Fig. 6.8 – Effect of exhaust circuit geometry on turbine inlet pressure diagrams [39] (configuration no.5) the resulting amplitude proved to be similar to the baseline configuration. In any case, the inclusion of a dividing wall produced an unequal mean pressure level in the two sectors. The increase of measured pulse amplitude both in the manifold pipes and at the turbine inlet when using a dividing wall producing equally spaced pulses in each duct is also highlighted in Figure 6.9. It is interesting to note that the same effect was achieved by extending the dividing wall only up to the manifold exit (configuration no.3), thus confirming the prevailing effect of the manifold mixing portion on flow unsteadiness reduction [36]. As a consequence, the optimisation of manifold design can play a fundamental role to determine the pulsating flow characteristics at the turbine entry, also affecting the available inlet energy and the turbine efficiency. 0.25

Manifold pipe

f = 50 Hz

Inlet pulse amplitude [bar]

Turbine inlet 0.20

0.15

0.10

0.05

0.00

0

1

2

3

4

5

Exhaust circuit configuration

Fig. 6.9 – Effect of exhaust circuit geometry on measured pulse amplitude [39]

62

6.4

Instantaneous turbine mass flow rate

The study was extended to the analysis of instantaneous turbine mass flow rate in unsteady flow conditions. To this end, several measurements were performed on TC3 turbocharger (§ 4), which was installed on the UNIGE-ICEG test rig. The relative turbine mass flow rate and efficiency steady state curves were previously defined over an extended range for different rotational speeds and waste-gate opening levels [40]. During the unsteady flow tests, turbine inlet and outlet pressure diagrams and inlet instantaneous mass flow rate were measured simultaneously using high frequency response straingauge transducers and a hot-wire probe respectively. The anemometric system calibration curve was formerly defined in steady flow conditions with the probe installed in the measurement plane, assuming as a reference the mass flow level supplied by the laminar flow meter. The analysis was extended to different turbine expansion ratio levels, whilst keeping average rotational speed factor (n/√T3=4000 rpm/√K) and pulse frequency (f=66.67 Hz) constant. The waste-gate valve was kept closed during this investigation. The assumption of instantaneous steady flow behaviour of the turbine operating under pulsating flow conditions, often used within simulation models to predict unsteady flow performance, was also considered. Using this approach, instantaneous QSF turbine performance was calculated starting from inlet and outlet pressure diagrams and referring to steady flow characteristics, according to the procedure first proposed in [41]. Figure 6.10 shows instantaneous measured and QSF calculated mass flow rate and inlet pressure diagrams over the pulse period for three different average levels of turbine expansion ratio (εm). All experimental signals were acquired for several complete cycles (generally 50) and appropriate averaging techniques were then applied to the measured data. Experimental mass flow rate profiles proved to be affected by higher frequency components when compared with pressure and calculated mass flow diagrams. A Fast Fourier Transform (FFT) analysis highlighted significant harmonic components up to a frequency of approximately 600 Hz, confirming the results of a preliminary investigation developed at a different pulse frequency level [40]. Higher frequency oscillations are probably related to flow disturbances in the measuring section. On the contrary, QSF mass flow rate profiles proved to be strictly related to measured pressure diagrams, due to the computation procedure used. Experimental and calculated diagrams were almost in phase at a lower turbine expansion ratio, while a noticeable phase shift (about 1/10 of the oscillation period) was observed when increasing the

63

0.10

1.6

α=0

Mass flow rate [kg/s]

0.08

M t calc. p3 meas.

εm = 1.24

1.5

1.4

0.06 1.3

pressure [bar]

n/ T3 = 4000 rpm/ K f = 66.67 Hz

M t meas.

0.04 1.2

0.02

0.0

1.1 1.0

0.5

time/pulse period 0.10

1.6

ε m = 1.32 1.5

1.4 0.06 1.3

pressure [bar]

Mass flow rate [kg/s]

0.08

0.04 1.2

0.02

0.0

0.5

1.1 1.0

time/pulse period 0.10

1.8

εm = 1.48

1.7

1.6 0.06 1.5

pressure [bar]

Mass flow rate [kg/s]

0.08

0.04 1.4

0.02

0.0

0.5

1.3 1.0

time/pulse period

Fig. 6.10 – Instantaneous pressure and mass flow rate profiles [42] expansion ratio and consequently average mass flow rate level, probably due to more substantial filling and emptying effects in the turbine volute casing. Similar behaviour was observed, at a constant turbine overall expansion ratio, when opening the waste-gate valve, thus increasing the mass flow level and the magnitude of unsteady flow phenomena [40] (§ 8). Mean measured and QSF mass flow rate values proved to be similar in these operating

64

conditions; calculated levels slightly underestimated experimental ones, with deviations generally below 5%. This result confirms the conclusions of previous investigations performed on different turbocharger turbines [24]. Different behaviour, instead, was observed as regards wave amplitude. The linear correlation between inlet pressure oscillation amplitude and turbine mean expansion ratio was confirmed, but experimental mass flow rate oscillation Mass flow parameter [kg√K/s bar]

0.85

n/ T3 = 4000 rpm/ K f = 66.67 Hz α=0

0.75

0.65

0.55

0.45 66.67 Hz

0.35

Steady state

εm =1.24 0.25 1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

Expansion ratio (static to static)

Mass flow parameter [kg√K/s bar]

0.85

0.75

0.65

0.55

0.45 66.67 Hz

0.35

Steady state

εm =1.32 0.25 1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

Expansion ratio (static to static)

Mass flow parameter [kg√K/s bar]

0.85

0.75

0.65

0.55

0.45 66.67 Hz

0.35

Steady state

εm =1.48 0.25 1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

Expansion ratio (static to static)

Fig. 6.11 – Turbine instantaneous mass flow parameter vs. pressure ratio [42]

65

amplitude also proved to increase significantly at a higher average pressure ratio, due to the higher mean mass flow level and the relative inertial effects. On the contrary, calculated mass flow rate profiles highlighted an almost constant oscillation amplitude when changing the turbine expansion ratio: this result confirms that a QSF approach is not adequate to describe system behaviour when unsteady effects become prominent [43]. The deviation of unsteady flow turbine performance from steady state behaviour is apparent if instantaneous measured mass flow parameter is plotted against the turbine expansion ratio, as in the case of steady flow characteristics (Fig. 6.11). The filling and emptying of the volute is highlighted by the hysteresis loop surrounding the steady state curve, the area of which was found to increase at a higher turbine expansion ratio, where flow unsteadiness is more significant (Fig. 6.10). It is confirmed that, at typical pulsating flow frequencies occurring in the exhaust system of automotive engines, the pulse is so rapid that mass flow does not have enough time to incrementally fill the volute volume with pressure, hence causing the hysteresis observed between measured mass flow rate and pressure. Besides, the steady state curve is not totally encapsulated by the unsteady mass flow loop. As reported by other authors [43], this may be due to the fact that there is not enough time to fill the volute volume before the peak of the pulse is achieved and suggests that the filling and emptying action is influenced not only by gas velocity oscillation but also by the unsteadiness of pressure wave.

6.5

Assessment of turbine unsteady flow performance

In high speed automotive engines, the turbocharger turbine usually operates under unsteady flow conditions and its performance is substantially affected by feeding flow characteristics [37]. It is therefore important to correctly assess the main parameters used to characterise turbine unsteady flow behaviour, i.e.: mass flow rate, available inlet energy and conversion efficiency. Unsteady flow turbine performance was analysed referring to TC1 and TC2 turbochargers (§ 4); all tests were performed using arrangement A (rotating valve pulse generator system). 6.5.1 Effect of pulse parameters on turbine inlet energy Under real turbocharger operation, the available gas energy at the turbine inlet in pulsating flow operation is affected by the exhaust valve opening profile and the manifold geometry. The relative specific level (referred to mass unit) over the pulse cycle (Δhs0

NSF)

can be

66

calculated as the integral of the instantaneous gas energy, referring to an isentropic expansion from turbine upstream conditions (for which reference is usually made to static levels) to downstream ambient pressure (p0), thus avoiding to take into account real turbine outlet pressure (p4), the level of which can be affected by the particular geometry of the downstream circuit or by the operation of specific devices, such as waste-gate valves [40]. If the pulse period is split into a number of short time intervals, according to [38, 44], Δhs0NSF can be calculated as the ratio of the sum of gas energy in each time interval and the accumulated mass of gas flowing through the turbine during the entire pulse period: k 1       p0  k       M ti  t i  c pi  T3i  1    p i 1  3 i         n  M ti  t i  n

hs 0 NSF

(6.1)

i 1

To evaluate available gas energy at the turbine inlet, it is therefore necessary to know the instantaneous turbine mass flow rate, inlet pressure and temperature (assumed to be constant within each time interval). Pressure diagrams can be measured with good degree of accuracy by using high frequency response transducers and suitable pressure taps in the inlet measurement plane, while the experimental evaluation of instantaneous mass flow rate and temperature is more critical. Alternatively, calculated levels of these quantities can be used, referring to 1D simulation models the results of which can be validated by comparing measured and calculated instantaneous pressure and average mass flow rate and temperature [44, 45] (§ 7.3). Within the investigation performed at the University of Genoa, instantaneous temperature at the turbine inlet was not directly measured and was calculated on the basis of the adiabatic process of an ideal gas, using mean measured pressure and temperature as a reference:  p  T3i  T3m   3i   p 3m 

k 1 k

(6.2)

The results provided by this procedure are usually assumed as a reasonable approximation of instantaneous temperature level if this quantity cannot be measured [43, 45].

67

As regards instantaneous mass flow rate, two different approaches were followed: the experimental level of this parameter at the turbine inlet was detected through a hot wire probe. Since this technique is not usually applicable to engine investigations, instantaneous mass flow rate was also evaluated starting from measured pressure diagrams and turbine steady mass flow characteristics, using a quasi-steady flow (QSF) assumption. The effect of the main pulsating flow characteristics on available energy at the turbine inlet was analysed. Since several unsteady flow parameters proved to be simultaneously affected by the change of turbine operating conditions in real operation, the behaviour of isentropic specific gas energy was first analysed referring to theoretical waves. To this end, sine and square pressure waves at the turbine inlet were considered. The available gas energy at the turbine inlet was calculated according to Eq. (6.1), assuming an isentropic correlation for the evaluation of instantaneous inlet temperature T3i (Eq. (6.2)) and a quasi-steady flow (QSF) assumption to assess turbine mass flow rate in each time interval (Mti). Reference was made to a typical turbocharger turbine for automotive gasoline engines, operating at a fixed rotational speed factor level (n/T3 = 4000 rpm/K). In the case of sinusoidal pressure waves, Figure 6.12 shows the effect of pulse amplitude on specific energy available at the turbine entry, for different average pressure levels. A trend towards higher energy values when increasing oscillation amplitude can be seen for each mean inlet pressure, with reduced impact at higher p3m. This behaviour confirms that available energy is mainly related to flow unsteadiness, which is usually associated with the ratio

Turbine inlet energy [kJ/kg]

70

n/ T3 = 4000 rpm/ K

65

T3m = 353 K

60

p3m =1.5 bar p3m =1.7 bar p3m =1.9 bar

55 50 45 40 35

0

0.2

0.4 0.6 Inlet pulse amplitude [bar]

0.8

Fig. 6.12 – Effect of pulse amplitude on turbine inlet energy (sine wave) [42]

68

between pulse amplitude and the relative mean level [38]. The influence of inlet temperature oscillation amplitude on turbine available energy was also investigated, at constant pressure diagram, by correlating instantaneous temperature with the relative mean value through different polytropic relationships. A small impact of temperature oscillation amplitude was found, particularly at moderate mean inlet temperatures (as experienced on the UNIGE-ICEG test rig). This result confirms that the use of an approximate procedure to assess instantaneous turbine inlet temperature when the experimental value is not available does not give rise to substantial errors in the determination of specific inlet energy. Depending on the design of the engine exhaust manifold and of the turbine admission (single or twin-entry), pressure oscillations at the turbine inlet can be extended over a different fraction of the period. It can therefore be useful to introduce a proper pulse duty cycle factor (Φ), defined as the pulse length of the wave as a fraction of the wavelength [46]. Referring to sinusoidal pressure oscillations with the same amplitude and mean level, Figure 6.13 shows the effect of the pulse duty cycle on available specific energy (with reference to the condition Φ=1). It is apparent that a wider pulse extension over the period generates an increase in turbine inlet energy: at constant pressure oscillation amplitude and mean level, the maximum specific energy content is achieved with a pulse extended over the entire engine operating period. This situation is typical of most automotive engine arrangements fitted with singleentry turbocharger turbines. Since experimental tests revealed significant modifications in the pressure oscillation shape when changing pulse frequency (and therefore engine rotational speed) [36], the inlet energy

Turbine inlet energy (ref. to Φ =1)

1.06 n/ T3 = 4000 rpm/

1.04

K

T3m = 353 K p 3m = 1.5 bar

1.02

 p 3 = 0.5 bar

1.00

0.98

0.96

0.94

0.25

0.5 0.75 Pulse duty cycle Φ [%]

1

Fig. 6.13 – Effect of pulse duty cycle on available energy at the turbine inlet [42]

69

Turbine inlet energy (ref. to Δp3=0.2)

1.30 n/ T3 = 4000 rpm/ K

1.25

T3m = 353 K p3m = 1.5 bar

1.20 1.15 1.10 1.05 1.00 0

0.2

0.4

0.6

0.8

Pulse amplitude [bar]

Fig. 6.14 – Comparison of theoretical sine and square pressure waves [42] levels related to sine and square waves with the same mean pressure value were compared. Figure 6.14 shows the relative results plotted against pulse amplitude and referred, for each wave configuration, to the energy level calculated at the minimum pulse amplitude considered (p3=0.2 bar). It is evident that the available energy increase at higher pulse amplitudes is greater in the case of the square pressure wave, for which the instantaneous shift from the average level over the pulse period is more significant. It is therefore apparent that the ratio between pulse amplitude and mean value is not sufficient to characterise flow unsteadiness which should also be related to oscillation shape. As mentioned above, measured pressure diagrams at constant turbine rotational speed and mean inlet pressure were significantly affected by pulsating flow frequency, due to wave action in the turbine feeding circuit [33]. However, a general trend towards higher pressure

1.05 n/ T3 = 4000 rpm/ K

Inlet pulse amplitude [bar]

T3m = 353 K

1.00

p3m = 1.5 bar

0.95

0.90

0.85 30

50

70

90

110

Pulse frequency [Hz]

Fig. 6.15 – Effect of pulse frequency on measured turbine inlet pulse amplitude [42] 70

48

Turbine inlet energy [kJ/kg]

n/ T3 = 4000 rpm/ K

47

T3m = 353 K p3m = 1.5 bar

46

45

44

43 30

50

70

90

110

Pulse frequency [Hz]

Fig. 6.16 – Correlation between turbine inlet energy and pulse frequency [42] oscillation amplitudes was found when increasing the frequency (Fig. 6.15). The specific inlet gas energy calculated proved to be substantially dependent on pulse amplitude, as shown in Figure 6.16, confirming the trend highlighted by theoretical waves analysis. At constant pulse frequency, pressure profile shapes were similar when changing turbine inlet pressure level and the correlation between inlet pulse amplitude and turbine mean inlet pressure proved to be almost linear (Fig. 6.17). The behaviour of mean available gas energy is the outcome of the combined effect of both these parameters (Fig. 6.18). 1.2

Inlet pulse amplitude [bar]

1.1

n/ T3 = 4000 rpm/ K f = 66.67 Hz

1.0 0.9 0.8 0.7 0.6 0.5 1.3

1.4

1.5

1.6

1.7

1.8

Average turbine inlet pressure [bar]

Fig. 6.17 – Effect of mean inlet pressure on pulse amplitude at constant frequency [42]

71

55

Turbine inlet energy [kJ/kg]

n/ T3 = 4000 rpm/ K 50

f = 66.67 Hz

45

40

35

30

25 1.3

1.4

1.5

1.6

1.7

1.8

Average turbine inlet pressure [bar]

Fig. 6.18 – Turbine inlet energy behaviour at constant pulse frequency [42]

6.5.2 Turbine unsteady flow efficiency Turbine efficiency is another important performance parameter to be evaluated in unsteady flow conditions. Various approaches can be used to assess this quantity, depending on the amount of available experimental information. If no by-pass system operates in parallel to the turbine rotor [40], average cycle efficiency (ηt NSF) can be calculated referring to the integral over the pulse period of instantaneous actual (Δhtr) and isentropic (Δhts) turbine specific work: T

 t NSF 

htr NSF hts NSF

 h

tr



0 T

(6.3)

 h

ts

0

Assuming a number of short time intervals over the pulse period, average isentropic work (Δhts NSF) can be expressed by: k 1       p 4i  k       M ti  t i  c pi  T3i  1    p i 1    3i       n

hts NSF 

n

 M i 1

ti

 t i 

(6.4)

72

which is formally similar to equation (6.1) when replacing ambient pressure (p0) with turbine downstream pressure (p4i). In this investigation, Δhts NSF was calculated referring to measured turbine inlet and outlet pressure diagrams and calculated instantaneous inlet temperature (using equation (6.2)). As regards turbine mass flow rate, both measured and QSF calculated instantaneous levels were used to evaluate specific isentropic work and efficiency. It was therefore possible to assess the acceptability within the considered calculation of the QSF approach, the results of which are often assumed to be acceptable [24], even though noticeable deviations of the instantaneous unsteady mass flow rate from quasi-steady behaviour are often observed (see Figures 6.10 and 6.11) [43]. It is not an easy task to determine instantaneous turbine actual work (Δhtr) since it is difficult to directly measure instantaneous torque for high speed rotating components. Some attempts to determine this quantity are presented in open literature [13, 19, 43], based on the measurement of turbine mean torque (from compressor work or using a suitable dynamometer) and the estimation of the fluctuating torque component according to the mass moment of inertia of the rotating assembly and its angular acceleration, calculated from accurate measurements of instantaneous rotational speed. If no information on instantaneous turbocharger rotational speed is available, the cycle average turbine actual work (Δhtr

NSF)

is usually calculated starting from mean measured

levels of turbine power (Ptr NSF) and mass flow rate (Mt NSF): htr NSF 

Ptr NSF M NSF

(6.5)

If a high speed dynamometer is not available on the test rig, turbine actual work is generally evaluated on the basis of measurements made on the compressor side of the turbocharger. As a consequence, an overall efficiency level (turbine isentropic efficiency multiplied by turbocharger mechanical efficiency) is estimated through equation (6.3). Turbine unsteady average efficiency (ηt NSF) calculated according to the above procedure can be compared with steady flow value (ηt

SF)

and with the level provided by a simplified

approach based on the evaluation of reference isentropic work starting from mean measured pressure and temperature values. The resulting apparent turbine unsteady flow efficiency (ηta NSF)

can therefore be expressed as:

73

 ta NSF 

htr NSF   p c pm  T3m  1   4 m   p3 m 

  

k 1 k

(6.6)

    

Finally, if a quasi-steady flow approach is followed, turbine unsteady efficiency can be evaluated on the basis of the reference steady flow curves. To take into account the distribution of turbine mass flow rate over the pulse period, in this investigation a modified calculation method was considered. QSF cycle average efficiency (ηt QSF) was evaluated as a weighted average value, assuming the QSF mass flow rate in each time interval (Mt QSF) as weight:

  n

 t QSF 

i 1

ti QSF

 M

 M ti QSF  t i 

n

i 1

ti QSF

(6.7)

 t i 

Figure 6.19 shows the results obtained from the application of the above approaches to assess turbine unsteady efficiency. The steady state efficiency curve measured on the UNIGE-ICEG test rig using the same measurement inlet and outlet planes and proper data processing is also plotted as a reference. It is apparent that, all test conditions being uniform, steady state 0.75

n/ T3 = 4000 rpm/ K f = 66.67 Hz α=0

0.70

Turbine efficiency

0.65

0.60

ht SF

0.55

hta NSF ht NSF

0.50

ht QSF

h*t NSF 0.45 1.15

1.20

1.25 1.30 1.35 1.40 Expansion ratio (static to static)

1.45

1.50

Fig. 6.19 – Turbine efficiency vs. mean expansion ratio [42]

74

efficiency values were always higher than the unsteady results provided by the different procedures, confirming the fact that pulsating flow deteriorates turbine efficiency [29]. Unsteady efficiency levels provided by the various evaluation methods proved to be different from each other, thus confirming the importance of an appropriate procedure for assessing this parameter. Apparent turbine unsteady efficiency (ηta NSF), calculated on the basis of mean measured pressure and temperature levels, was lower than measured steady state efficiency. The reductions, amounting to approximately 7-8%, confirmed the trend observed by other authors [44] who measured differences of about 13% referring to a twin-entry turbocharger turbine for a heavy-duty diesel engine application. The procedure based on the evaluation of instantaneous inlet and outlet parameters provided the most reliable turbine unsteady efficiency levels. In this case, the calculated efficiency values were about 12-13% less than the steady state efficiency levels measured at the same average expansion ratio. No significant differences were found between unsteady efficiency levels (ηt NSF and η*t NSF respectively) evaluated using measured or calculated instantaneous mass flow rate to estimate average turbine isentropic work. This result may be related to comparable average levels of experimental and QSF mass flow rate. The shifts observed between mass flow profiles (Fig. 6.19) probably affect instantaneous efficiency level, though to evaluate this, turbine actual work must also be assessed. The modified QSF procedure, based on the calculation of a weighted average efficiency level, did not provide satisfactory results in the selected operating conditions. The relative levels proved to be lower than steady state values but higher than those provided by the other procedures for evaluating turbine average unsteady efficiency. This result seems related to the prevailing influence of steady flow efficiency levels in the calculation procedure.

75

7. Unsteady turbine performance predicting methods and modelling 7.1

Correlation between steady and unsteady turbine performance

As above mentioned, in automotive applications the turbocharger turbine generally operates under unsteady flow conditions and it is extremely difficult to evaluate pulsating performance since several parameters can affect measured data. The analysis of average turbine performance under unsteady flow conditions requires the definition of suitable methods of comparison that can highlight the effect of the main pulsating flow parameters and the shifts from steady flow results. To this end, several approaches have been proposed and discussed by various authors [13, 20, 38, 41, 47], often conditioned by available experimental information. In any case, performance comparison developed at the same turbine average expansion ratio and rotational speed is an easy and immediate solution since its application requires just inlet and outlet pressure diagrams to be measured in order to calculate the turbine average expansion ratio. This procedure was used at ICEG during previous investigations [24, 25] referring to the K ratios between mean pulsating turbine performance (mass flow, torque or power) and the corresponding steady flow values, expressed as: K

average pulsating performance steady flow performance

Using this approach, KM and KP factors were calculated for different operating conditions in order to highlight the effect of the main pulsating flow parameters on average turbine performance, referring to TC1 turbocharger (§ 4). These results make reference to circuit arrangement A using rotating valve as pulse generator system. Measured unsteady mass flow and power were generally different from steady flow values at the same average expansion ratio and rotational speed, with absolute shifts of up to 20 percent. These differences were lower when the waste-gate valve was opened, thus suggesting that flow unsteadiness was reduced on the by-passed fluid portion. As in the case of previous investigations performed on fixed and variable geometry turbines [24, 25], both unsteady mass flow and power were found to depend on pulse frequency though the actual magnitude of the effect was different at each frequency level. This result is

76

probably related to the significant modifications in inlet pulse shape and amplitude which were observed when changing the pulsating flow frequency (§ 6.2), due to the wave action in the turbine feeding circuit [24, 25, 37]. At constant pulse frequency, a general trend to lower KM values and higher KP levels was found when the average expansion ratio was increased. Figure 7.1 shows the results referred to a pulse frequency of 66.67 Hz, for which both average unsteady mass flow rate and power were lower than the corresponding steady flow values in the explored range. Even if the absolute values of KM and KP proved to be affected by the selected pulse frequency level, the trend shown in Figure 7.1 generally confirmed and seems related to the corresponding change in inlet pulse amplitude when varying the average expansion ratio (Fig. 7.2). The resulting link between inlet pulse amplitude and average turbine performance confirmed, at least as regards mass flow rate, the conclusions of previous investigations [24, 25] and the hypothesis of turbine quasi-steady flow behaviour in unsteady conditions, related to the variable slope of constant speed steady flow curves [24]. Figure 7.1 shows that, compared with a slight reduction in the turbine mass flow factor, power parameter substantially grows when the expansion ratio is increased (and consequently the inlet pulse amplitude). This result suggests a remarkable influence of flow unsteadiness, usually related to the pulse amplitude or to the ratio between this parameter and the average pressure level [38], on turbine specific work. In order to highlight this effect, turbine power

1.00

Mass flow ratio

0.98

n rpm  4000 T3 K

0.96

f  66.67 Hz

A= 0

0.94 0.92 0.90 0.88 1.3

1.4

1.5

1.6

1.7

1.8

1.7

1.8

Average expansion ratio 1.00

Power ratio

0.96 0.92 0.88 0.84 0.80 1.3

1.4

1.5

1.6

Average expansion ratio

Fig. 7.1 – Comparison between average pulsating and steady turbine performance [33]

77

Inlet pulse amplitude [bar]

1.3

1.1

0.9

0.7

n rpm  4000 T3 K

0.5 1.3

1.4

f  66.67 Hz

1.5

1.6

A= 0

1.7

1.8

Average expansion ratio

Fig. 7.2 – Correlation between inlet pulse amplitude and average expansion ratio [33]

Specific work ratio

1.4 1.3 1.2 1.1 1.0

n rpm  4000 T3 K

0.9

f  66.67 Hz

A= 0

0.8 0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

Inlet pulse amplitude [bar]

Fig. 7.3 – Effect of inlet pulse amplitude on turbine [33] levels under steady and pulsating flow conditions were compared for the same average mass flow rate. Figure 7.3 shows the relative result with reference to the operating conditions of Figure 7.1 and 7.2: in this case, the ratio K'P between mean pulsating and steady turbine power was plotted against inlet pulse amplitude. Assuming that the coefficient K'P may be representative of the ratio between turbine specific work in the two flow situations, its behaviour is the outcome of the combined effect of the inlet energy increase for higher flow unsteadiness (i.e., pulse amplitude) (§ 6.5.1) [38] and the deterioration of turbine energy conversion efficiency in unsteady flow conditions [29]. In the specific operating situation shown in Figure 7.3, the reduction of turbine efficiency seems to prevail over the inlet energy rise for the lower considered pulse amplitudes; on the contrary, for higher flow unsteadiness levels, K'P was always above unit, suggesting a prevalent effect of the available energy at the turbine inlet.

7.2

Quasi Steady Flow (QSF) predicting methods

The assumption of instantaneous steady flow behaviour of the turbine operating under pulsating flow conditions is often used to predict unsteady flow performance, particularly in the case of small automotive turbocharging units. Using this approach, instantaneous quasi-

78

steady flow (QSF) turbine performance was calculated starting from inlet and outlet pressure diagrams and referring to steady flow characteristics, according to the procedure described in [41], referring to TC1 turbocharger (§ 4). Starting form pressure diagrams measured upstream (p3) and downstream (p4) the turbine, the pulse period was divided into N equal intervals and the instantaneous expansion ratio εti (static to static) for each of them was determined. The static inlet temperature T3i was assumed related to the mean measured value through the isentropic relationship, following the Eq. (6.2) [41]. The speed parameter n/√T3i was evaluated assuming the turbine rotational speed measured level as a reliable mean value over the pulse period. Instantaneous levels of turbine mass flow and power were determined referring to the steady flow characteristics. As regards power, it was assumed that the bearing losses were unchanged in pulsating flow operation for the same rotational speed, oil pressure and temperature at the turbocharger inlet [34]. Average calculated mass flow (MQSF) and power (PQSF) were compared with the mean measured values (MNSF e PNSF) using the relevant influence factors IM and IP:

IM 

M NSF M QSF

(7.1)

IP 

PNSF PQSF

(7.2)

Three different weighting techniques of instantaneous QSF levels [24] were applied: in addition to the mean arithmetic values (M'QSF and P'QSF) [41], average turbine performance was calculated as weighted average levels over the pulse period, assuming the instantaneous expansion ratio as weight (M''QSF and P''QSF) or on the basis of a suitable “mean equivalent expansion ratio”, related to the pulse characteristics through the instantaneous mass flow level (M'''QSF and P'''QSF) [24]. In Figure 7.4, the estimated influence factors IM and IP are plotted against average turbine expansion ratio for the same operating condition considered in previous comparisons. The results provided by the different calculation procedures proved to be almost unaffected by the expansion ratio level (and then by the inlet pulse amplitude) as regards turbine mass flow rate,

79

Influence factor

1.4

I 'M

1.2

I '' M I ''' M

1.0 0.8 0.6

n rpm  4000 T3 K

0.4 0.2

f  66.67 Hz

A= 0

0.0 1.3

1.4

1.5

1.6

1.7

1.8

Average expansion ratio

Influence factor

1.4

I 'P

1.2

I '' P

1.0

I ''' P

0.8 0.6 0.4 0.2 0.0 1.3

1.4

1.5

1.6

1.7

1.8

Average expansion ratio

Fig. 7.4 – Comparison of measured and QSF calculated turbine performance [33] while a noticeable influence of this parameter was found on IP factors, with absolute values increasing with the expansion ratio. The best results were achieved using the traditional averaging procedure (I') and the weighted-average method (I''), while the process based on the “mean equivalent expansion ratio” produced the greatest deviations from mean experimental performance, overestimating measured values. In the case of mass flow rate, the best procedures gave satisfactory results (with shifts generally lower than 10 percent) suggesting a quasi-steady flow behaviour of small turbocharger turbines operating under pulsating flow conditions referring to this parameter [19, 24]. The approximation provided by QSF calculations as regards turbine power was less satisfactory: deviations from the average experimental levels resulted higher (even more than 20 percent), depending on the mean expansion ratio. The calculated influence factors also proved to be affected by the pulse frequency but no functional trend with respect to this parameter was found, probably due to the observed modifications of wave shapes.

80

7.3

Unsteady flow turbine modelling

In this section an approach to model the engine turbocharger working under unsteady operating conditions, implemented in the GASDYN 1D thermo-fluid dynamic code (developed at Politecnico di Milano) is described. The experimental measurements are then compared with the predictions of the simulations. In order to generate pulsating flow upstream the turbine, the pulse generator system fitted with the rotating valve was used (see Figure 7.5, where the considered elements and measurements sections are indicated). The geometric characteristics of the circuit feeding turbine are reported in Table 7.1. Although the pressure ratio-mass flow characteristics of turbines are strongly time dependent, it is common

Plenum

Rotating Valves Sect. RV a

Sect. Y

c

Sect. 1 e d

C

b

Sect. 2

T

Silencer Sect. 3

Sect. 4

Fig. 7.5 – Pulse generator system with rotating valve

Tab. 7.1 – Geometric characteristics of turbine upstream circuit Element

Description

a

Y-junction, span angle of 29°, length 578 mm, F 42 mm Length 759 mm, F 42 mm, with a ball valve Bend, mean radius 100 mm, length 160 mm, F 42 mm Inlet F 40 mm, outlet 44.2x33 mm, length 110 mm, with ball valve Length 98 mm, 44.2x33 mm

b c d e

81

practice to use steady flow turbine characteristics despite of calculations of pulsating flows are performed. This approach may sometimes lead to poor results, in particular when the effects of the pulsating flow become relevant. No methods to directly relate unsteady flow phenomena as equivalent flow data have been currently found by researchers; however, the accuracy of the prediction of the turbine operating conditions under pulsating flows may be improved as equivalent lengths and volumes are used, together with the steady flow characteristics, to model the turbine. In this study the turbine has been treated as a 1D boundary condition: steady pressure ratiomass flow-speed characteristics have been used together with a zero-dimensional volume boundary conditions, representing the flow storage occurring in the volute and in the diffuser of the radial flow turbine (Fig. 7.6). The volume of the volute and of the rotor have been connected by a duct, whose length must account for the wave propagation time from a nominal entry located around the case and the rotor itself. The behaviour of the rotor has been assumed as quasi-steady, because of the short flow path crossing the turbine. The analysis of the turbine operating under unsteady conditions implies to account for the passage of the waves through the turbine. In most of the engine arrangements, the pulsation downstream of the turbine may be significant; so, a detailed schematization of the pipe system downstream of the turbine is required; the pressure ratio across the device is given by the ratio of the instantaneous pressure on either sides of it. If the downstream back pressure is assumed as constant (equal to the mean back pressure), the pressure ratio across the turbine, and hence the instant power given by the device, cannot be correctly calculated. The use of the steady-state characteristics of the turbine to model the rotor introduces an approximation, since they implicitly account for the pressure variation of the gas flowing into

Fig. 7.6 – Schematic layout of the turbocharger for the 1D simulation

82

pp p0 pp

p1

p1

Fig. 7.7 – Schematic of a volume connected with two pipe the volute and exiting by the diffuser. Hence, no pressure drop was considered in the volume boundary condition, since this information was already included in the maps. Both the turbine volute and the diffuser (whenever it is mounted downstream of the turbine) have been modelled by means of a 0D volume, in order to account for the mass flow storage. Referring to Figure 7.7, it is assumed that the gas in the volume is under stagnation conditions, denoted by p0 and a0; p1 , u1 and a1 denote the flow condition at the volume port, pp, up and ap are the flow condition in the pipe. Two basic flow situations are now considered. The first relates the flow from the pipe to the volume: it represents the flow entering into volute and the flow going from the pipe (representing the equivalent length of the diffuser) to the volume downstream of the rotor boundary condition (see Figure 7.7). The second relates the flow from a volume (representing the volute and the turbine diffuser) to a pipe (representing the volute equivalent length and the exhaust pipe downstream the turbine respectively). In the boundary condition considered, the gas is supposed to expand isentropically through the volume port from the pipe (section p) to the port (denoted by the suffix 1). The pressure p1 in the port is equal to the volume pressure p0 and the gas velocity at the port is dissipated at constant pressure with an increase of entropy, to the condition 0; the incoming gas, then, mixes with the volume contents resulting in a further change in entropy. The continuity and the energy equations are used. Referring to Figure 7.7, the gas leaves the volume with stagnation pressure p0 and speed of sound a0 and expands isentropically as it passes through passage between the volume itself and the port. The isentropic expansion ceases when the gas stream reaches its minimum crosssectional area (denoted by the index 1). The gas is then considered to expand adiabatically, but irreversibly at constant pressure to fill the pipe cross sectional area. Hence, for subsonic flows, the pressure at the volume port is assumed equal to the pressure at the pipe end p1 = pp.

83

The momentum equation is therefore not required and only the energy and the continuity equation are used. In order to correctly account for the wave propagation time from the volute to the rotor, an equivalent length (represented by a duct) must be considered. The pipe length has been assumed equal to the distance between the flange and the rotor inlet. The conservation equations for one-dimensional, unsteady, compressible, reacting flows are solved through the internal nodes of the pipes. In this work, symmetric finite difference techniques with second order accuracy (such as the MacCormack predictor-corrector or the two-step Lax-Wendroff method), with the addition of flux-limiting techniques (FCT, TVD) were used. The source term in the energy equation calculated for each node accounts for the heat exchange between the gas and the volute wall [45]. The pipe representing the equivalent length of the volute connects the 0D volume to the boundary condition containing the information given by the steady pressure ratio-mass flowspeed characteristics. Since the flow path crossing the turbine rotor is very short, the behaviour of the rotor has been assumed as quasi-steady. In order to correctly evaluate the operating conditions of a turbocharger, the pressure downstream of the turbine cannot be considered as a constant, if correct predictions of the instant pressure ratio must be achieved. Hence, the wave equations in the inlet and the outlet pipes must be coupled with the turbine flow characteristics. The boundary condition developed for the turbine rotor is an extension of the boundary condition proposed by Benson [37, 48] and it solves the governing equations by means of the method of the characteristics (MOC). In particular, the wave equations are

Turbine m ass flow rate factor [(kg/s) √K/bar]

0.90

0.60

2000 rpm/√K 3000 rpm/√K 4000 rpm/√K 5500 rpm/√K 7000 rpm/√K

0.30 1.00

1.20

1.40

1.60

1.80

2.00

2.20

2.40

Expansion ratio (total to static)

Fig. 7.8 – Turbine steady state characteristics [45]

84

combined with the turbine flow characteristics (see Figure 7.8). The algorithm is used from zero flow rate up to the maximum flow rate parameter. Since the model has been devised to work mainly on engine applications, the hypothesis that no reverse flow may occur in the turbocharger turbine during the normal engine working operating conditions has been considered; the turbine is not supposed to allow the flow for being transmitted from the outlet end towards the inlet end. As the reverse flow occurs, the turbine works as a closed end and a pressure wave (having equal magnitude of the incident one) is reflected from the turbine outlet toward the exhaust end. Hence, the steady characteristics, representing the turbine performance data, are processed from zero flow rate up to the mass flow rate corresponding to the condition of chocked flow. The system of the governing equations is solved iteratively, until a value M1 that allows for the convergence of the incident Riemann variable (Eq.(7.3))

    in ,c  in,n   in ,c out ,c  2  

(7.3)

is found. In order to obtain the whole boundary curve, it is necessary to extrapolate the measured constant speed lines back to zero mass flow. The boundary curves deriving from the steady characteristics relates λin,1 and λout,2. The upper bound of the area is defined by the chocked flow and the lower bound by no flow: the model assumes that no reverse flow may occur during the normal engine working conditions. The test stand configuration and operation of the ICE Laboratory of the University of Genoa has been simulated by using the new turbine model described above (Fig. 7.9), which has been implemented in the 1D thermo-fluid dynamic numerical code GASDYN [15, 16, 17],

Fig. 7.9 – One-dimensional schematic layout of the UNIGE test rig

85

developed by the researchers at Politecnico di Milano. The experimental investigation was performed on TC3 turbocharger (§ 4). The single-entry nozzleless radial flow turbine was equipped with a waste-gate regulating system: the by-pass valve was turned off for this specific application. Steady flow characteristics of both the compressor and the turbine were measured (Fig. 7.8) and they were used in the 1D turbine rotor boundary condition, previously described. All the measurements were performed by keeping the mean speed parameter as constant (n/√TT3 = 4000 rpm/√K), while the average expansion ratio was varied. Two pulse frequencies were selected: (40 Hz and 66.67 Hz), corresponding to 1200 and 2000 rpm respectively for a 4 cylinder engine. Instantaneous static pressure pulses were measured at different locations, as shown in Figure 7.5: near the rotating valve (Sect. RV), near the Yjunction (Sect. Y), upstream and downstream of the turbine (Sect. 3 and 4 respectively). In order to reproduce the pressure profiles recorded at the measuring point close to the rotating valve (Sect. RV of Figure 7.5), the measured pressure pulse was imposed at the inlet end of the system modelled. A non-reflecting boundary condition sets the value of the Riemann variables at the inlet section of the system, on the basis of the incident pressure wave, that is assumed equal to the measured pressure pulse: k 1

out

 p  pin  2 k  1  2 amb   p ref  

(7.4)

in  1

(7.5)

AA  1

(7.6)

where λout is the incident Riemann variable, pin the relative incident pressure, λin is the reflected Riemann variable, AA is the entropy level, pamb is the ambient pressure. In the simulations, the working fluid was modelled as a perfect gas, whose temperature was set accordingly to the experimental measurements of mean static temperature. At the lower pulsating frequency of the signal (f = 40 Hz), a condition corresponding to a turbine average expansion ratio equal to 1.36 was considered; at the highest frequency (f = 66 Hz), three different mean turbine expansion ratios ε (1.16, 1.24 and 1.43) were tested. In this

86

study, as above mentioned, it is described an attempt to model the radial flow turbine by considering the case volume and by locating the effective position of the boundary halfway around the casing, by the introduction in the schematic layout of an equivalent length. The results reported in Figure 7.10 show a comparison between experimental and calculated pressure waves at the Y-junction (Sect. Y), upstream (Sect. 3) and downstream of the turbine (Sect. 4). In Sect. Y and Sect. 3 the agreement between simulations and experiments looks satisfying at both frequencies, although some discrepancies are apparent. This may be

Fig. 7.10 – Pressure traces upstream of the turbine (see Fig.7.5): comparison between simulations and experiments [45]

87

attributed to several factors. At first, although the unsteady behaviour of the turbocharger has been simulated by introducing case volumes and equivalent lengths, the turbine rotor boundary condition still makes use of steady-state maps. Moreover, it is very difficult to obtain steady state characteristics over the full potential operating range of the turbine: the loading of the turbine using a compressor restricts the maximum speed achievable and the definition of the operating points at low mass flows, corresponding to the compressor surge; this implies that the turbine maps sometimes must be manually extrapolated. The same holds for the turbocharger compressor model: in order to obtain the full boundary curve it is necessary to extend the constant speed lines back to zero mass flow through the area where the compressor operates under surge conditions. Finally, the exact value of the polar moment of inertia of the whole rotor assembly in the turbocharger is another critical issue. Although a reasonable value has been assumed as input data for the code, it will differ from the exact value; it is then apparent that this discrepancy leads to an error in the calculation of an operating point. Even if the amplitude of the pressure trace at the turbine outlet end is related on the frequency of the pulses entering into the turbine, the length of the downstream pipe looks having a strong influence on the pressure profile. Figure 7.10 confirms that the agreement between simulations and experiments in Sect. 4 is acceptable both at 40 Hz and at 66 Hz. High lengths of the exhaust pipes downstream of the turbine correspond to low natural frequencies; at lowmedium pulse frequencies, this is the main factor influencing the wave motion downstream of the turbine. Since the turbine performance depends on the pressure-ratio (that should be calculated as the ratio of the instantaneous pressures on either side of the device), the use of a variable outlet pressure model of the turbine is very important. No experimental information on instant temperature T3i was available upstream of the turbine; so, within the experimental investigation, the instant temperature was calculated by assuming the gas as ideal and the process across the turbine as adiabatic, following the Eq.(6.2). Figure 7.11 shows the comparison between the instant temperature T3i calculated and the instant temperature predicted by the numerical code upstream of the turbine at 66 Hz, for the three different values of expansion ratio considered. A good agreement between the results provided by the two methods was observed [43].

88

Fig. 7.11 – Temperature profiles upstream of turbine [45] A comparison between experimental and calculated instantaneous turbine mass flow rate over the pulse period was performed at the turbine entry (Sect. 3 of Figure 7.5) by a hot wire anemometer. Figure 7.12 shows different mass flow traces, referring to the higher flow forcing frequency (f=f1=66.67 Hz) and the three considered average levels of the expansion ratio ε. The 1D model seems to be able to provide a satisfying prediction of the instant mass flow rate; of course a 1D model cannot capture the high-frequency components of the signal evidenced by the experimental analysis, since those components may be attributed to waves propagating in the transverse direction; hence, those effects may be captured only by a multi-dimensional approach. A Fast Fourier Transform algorithm has been applied on to the measured and simulated signals, in order to study the magnitude of the different harmonic components (ΔMn).

Fig. 7.12 – Mass flow rate profiles and related FFT analysis [45]

89

As shown in Figure 7.12, the model satisfactorily estimates the components of the signals at low and medium frequencies (fn/f1 < 10); on the other hand, the model cannot reproduce harmonic components of the signal at the highest frequencies (fn/f1~15, corresponding to a frequency of about 1 kHz). If the experimental and the simulated mass flow are plotted against the correspondent expansion ratio (Fig. 7.13), they exhibit a significant hysteresis. The hysteresis is strictly related to the unsteadiness of the flow: it is the consequence of the tendency of the gas to flow and accumulate into the volume of the turbine casing and to build up the pressure inside it, as the pulsating flow enters into the system by the rotating valve. Similarly, when the mass flow in the pipes and in the volute exceeds the pulsating flow entering through the pulse generator, the mass in the system decreases. Hence, the mass flow at the turbine entry is not exactly the

Fig. 7.13 – Instantaneous mass flow vs. expansion ratio of the turbine [45]

90

same of the mass flow near the turbine rotor, since there is a mass-storage contribution due to the volute volume. Figure 7.13 shows a comparison between the experimental data and simulation results. As expected, the classical approach (based on the turbine steady characteristics) is not able to predict any hysteresis loop, since it does not take into account of any volume blockage. Conversely, the new approach looks able to take into account of the unsteady effects of the turbine, although the prediction cannot still be considered as satisfying.

91

8. Effect of waste-gate valve opening on turbine performance 8.1

Investigation goal

The successful application of exhaust turbocharging to downsized SI engines must overcome various difficulties, related both to the specific operating environment (exhaust gas temperature level) and to engine performance, focusing on low-end torque and transient response [10, 11]. Variable geometry turbines (VGT) are widely used in diesel turbocharging but several problems, mainly related to the harsh thermal exhaust environment, must be solved before they can be applied to SI engines. Advanced SI engine turbocharging technologies are currently being developed. These include electrically driven waste-gate valves, twin-entry turbine housings and variable geometry mechanisms especially developed for engine thermal environment and comprising just a few moving parts [12, 49]. At this stage, a waste-gate valve is usually fitted as a turbocharger control system due to its effectiveness, low cost and ability to work at high exhaust gas temperatures. In order to optimise engine-turbocharger matching, turbine performance when the waste-gate valve is partially or totally opened should be known, both in steady state and in the typically unsteady flow conditions occurring in automotive engines. Unfortunately, turbocharger manufacturers usually provide very little information about steady flow turbine characteristics in the opened waste-gate field, though this is a fundamental input for theoretical simulation models. To analyse the behaviour of a turbocharger turbine for downsized SI engines when the wastegate valve is opened, an experimental study was performed on the ICEG test rig, extended both to steady and unsteady flow conditions.

8.2

Mass flow sensitivity to WG valve setting

A substantial increase in turbine mass flow rate was generally observed when opening the waste-gate valve (see Figure 8.1).

92

Turbine mass flow rate factor [kg √K/s bar]

2.00

1.75

1.50

1.25

1.00

0.75

0.50

Turbine overall efficiency

0.25 1.00

1.25

1.50

1.75

2.00 2.25 2.50 Expansion ratio (total to static)

0.70

0.60

αWG [deg] 0

0.50

8 20 60

0.40

n  rpm   TT3TT 3K  n

2000 3000 4000 5500 7000

0.30

0.20

0.10 1.00

1.25

1.50

1.75

2.00 2.25 2.50 Expansion ratio (total to static)

Fig. 8.1 – Turbine steady flow characteristics [40] 93

Referring to TC2 turbocharger (§ 4), a detailed analysis developed at constant turbine rotational speed (n/√TT3=4000 rpm/√K) and expansion ratio (εtTS=1.32) confirmed a substantially higher sensitivity (defined as the ratio of change induced in the output, i.e., mass flow rate, to the related input change, i.e., push rod position) of such TC regulating system in the field of lower by-pass settings. This aspect is highlighted in Figure 8.2, where the mass flow rate increase determined by the waste-gate opening (ΔMWG) is plotted as a function of its opening degree (A), being this parameter related to the linear displacement of the by-pass valve push rod and defined as a percentage of total rod displacement, ranging from 0 to 100 percent when varying the waste-gate flow area from zero to maximum. It is clear that over 95 percent of the turbine swallowing rise was achieved at a waste-gate setting of 50 percent; it is also interesting to note that a by-pass opening of 5 percent (corresponding to a rod displacement of 1.5 mm for the specific considered device) determined a mass flow increase of approximately 40 percent of the total range. This behaviour is not unexpected and is mainly related to the geometric characteristics of the regulating device and the relationship between the connecting rod displacement and the related change of the by-pass flow area. All the same, this aspect should be carefully considered when designing the relative actuator system and when defining correct control strategies, also taking into account the inaccuracies induced by mechanical clearances and hysteresis phenomena.

8.3

Turbine and waste-gate steady mass flow contributions

Turbine mass flow rate and efficiency steady flow curves were previously defined for different rotational speed and waste-gate opening levels, being the last expressed in terms of the rotation angle (αwg) of the relative driving shaft, referring to TC3 turbocharger (§ 4). Figure 8.1 shows turbine steady flow characteristics measured on the unige test rig, referring to different rotational speed factor levels and waste-gate openings. The extension of measured curves achieved by suitable experimental techniques while working at moderate turbine inlet air temperature (about 400 K) proved to be satisfactory for application on simulation models. It was found to be considerably wider than that considered in the maps provided by the turbocharger manufacturer, which only refer to closed waste-gate conditions. A substantial increase in turbine mass flow rate was observed when the by-pass port was opened wide (between 100 and 140 percent at αwg = 60°, at which a over 95 percent of swallowing rise was measured).When opening the waste-gate valve, overall turbine efficiency

94

Waste-gate mass flow rate increase [%]

100

80

60

40

n rpm  4000 TT 3 K

20

 tTS  1.32

0 0

20

40

60

Waste-gate opening degree [%]

80

100

Fig. 8.2 – Waste-gate mass flow sensitivity [33] η’t (defined as turbine total-to-static isentropic efficiency multiplied by turbocharger mechanical efficiency) decreased definitely (Fig. 8.1) if this quantity is referred to the isentropic work related to the expansion of the whole mass flowing through the system (turbine rotor and waste-gate), according to the following equation:

 ' t   t TS   m TC 

Pt Pc   m TC  M tot  hst M tot  hst

(8.1)

Within turbocharged engine simulation models, turbine steady mass flow curves with the bypass port fully closed are usually implemented (the only ones generally available) and the waste-gate flow contribution is estimated through a proper orifice working in parallel to the turbine under the same overall expansion ratio. This assumption does not appear to be completely acceptable for small automotive turbochargers in which there is a substantial bypass flow contribution, often greater than the mass flowing through the turbine rotor. Significant interactions between the two flow components probably take place within the turbine volute casing and lead to considerable modifications in mass flow rate and efficiency. To study this aspect further, a dedicated investigation was performed in order to analyse the swallowing capacity of both the turbine and the waste-gate valve and the splitting of total mass flow rate between the two flow ducts. To allow the by-pass component to be measured, a proper configuration of the turbine outlet circuit was set up, using a suitable outlet pipe fitted with a dividing wall extended to a length of about 150 mm before a mixing section (Fig. 8.3). The waste-gate mass flow rate was estimated using a hot wire probe placed in the relative outlet pipe portion. Selected turbine operating conditions were considered by working

95

Fig. 8.3 – Turbocharger turbine outlet pipe at constant rotational speed factor and overall expansion ratio and varying the waste-gate opening angle (αWG). Measurements were extended to different flow configurations, with the working fluid passing only through the turbine rotor (closed waste-gate condition), only through the waste-gate port (by closing the rotor passage) or through both flow ducts. Even if the position of the hot wire probe remained unchanged for the entire experimental campaign, calibration curves were defined for each waste-gate setting considered, due to possible modifications of the flow field in the measuring section induced by the different position of the valve plate. Figure 8.4 shows the mass flow rate results, reported in terms of equivalent isentropic flow area Aeq, at a constant turbine rotational speed factor and overall expansion ratio (based on measurements performed at the inlet and outlet measuring stations). Generally speaking, it was difficult to obtain good repeatability of measured levels due to the typical clearances of the by-pass mechanism resulting in slight differences of the valve plate position at the same driving shaft opening angle. Even if this aspect often gave rise to a poor correlation of measured mass flow values which prevented an exact definition of measured levels, general trends were clear and suggested some interesting points for further analysis. The by-pass mass flow rate increased substantially when the valve was opened (Fig. 8.4). However, the waste-gate swallowing capacity proved to be lower in the case of combined flow (rotor + by-pass port) (Aeq WG) than flow through the valve alone (A*eq WG). A similar trend was observed for the turbine rotor, since its equivalent flow area (Aeq R) (evaluated on the basis of measured total mass flow rate and waste-gate contribution) decreased when the by-pass valve was opened.

96

‫‏‬Equivalent isentropic flow area

11 10 9

*

n/ T3 = 3000 rpm/ K

A eq tot

p T3 /p4 =1.2

8

A eq tot

7

*

A eqWG

6

A eqWG

5 4

A eqR

3 2 1 0 0

10

20

30

40

50

60

70

Waste-gate opening [deg]

Fig. 8.4 – Effect of waste-gate opening on turbine mass flow components [40] As mentioned above, in engine simulation models turbine mass flow rate in the opened wastegate region is usually evaluated by assuming the impeller and the by-pass valve to be two nozzles working in parallel under the same overall expansion ratio. The relative flow curves (either measured or calculated) are therefore used to assess the overall swallowing capacity of the system. Following this procedure, the total equivalent flow area was calculated (A*eq tot) and compared with the overall measured level (Aeq

tot).

As shown in Figure 8.4, calculated

mass flow levels always proved to be higher than measured ones, at the same waste-gate opening and turbine operating parameters. Differences ranged between 10 and 25 percent and the error was generally higher for wider waste-gate openings. This result partially confirms the conclusions of a previous investigation performed at UNIGE-ICEG [34]. Since both the rotor and the by-pass mass flow rate proved to be lower than their reference value (single flow condition), a reduced level of the effective expansion ratio across the impeller and the by-pass port can be assumed when the waste-gate valve is opened, probably due to a higher pressure drop between the external inlet measuring plane and the real entry section of both the turbine stage and the waste-gate orifice, due to the higher velocity in the volute casing where total mass flow occurs. To analyse this aspect further, a dedicated investigation into pressure distribution along the turbine volute casing will be developed by ICEG researchers in a subsequent stage with a view to correlating the overall turbine

97

expansion ratio with the real pressure ratio experienced by both the turbine stage and the waste-gate valve when changing the by-pass opening. Also taking into account the significant interactions between the two flow components probably occurring in the turbine outlet section (highly dependent on the geometry of the downstream mixing volume), the assumption that the total swallowing capacity of the system can be calculated by summing turbine and waste-gate flow contributions was confirmed to be inadequate, resulting in substantial errors within engine-turbocharger matching calculations.

8.4

Evaluation of turbine efficiency in the opened waste-gate valve region

As previously mentioned, overall turbine efficiency η’t, evaluated on the basis of the isentropic expansion of the entire mass flowing through the system (turbine rotor and wastegate) significantly decreased when opening the waste-gate valve. Figure 8.5 shows this result (referring to the level corresponding to the waste-gate when fully closed) for one of turbine operating conditions considered within the investigation. As can be seen, efficiency is strictly related to the increase of total mass flow rate which proved to be substantial for small wastegate openings and less important for large by-pass settings. Since the experimental study made it possible to estimate the mass flow passing through the turbine impeller in opened waste-gate conditions, a different definition of overall turbine efficiency (η’tR) was also considered, referring to the fluid portion effectively working within

Turbine efficiency (referred to closed WG condition)

the turbine stage (MR): 1.2

' ηt

n/ T3 = 3000 rpm/ K p T3 /p4 =1.2

' η tR

1

0.8

0.6

0.4

0.2

0 0

10

20

30

40

50

60

70

Waste-gate opening [deg]

Fig. 8.5 – Turbine efficiency curves referred to total and rotor mass flow rate [40]

98

 't R 

Pt Pc   m TC  M R  hst M R  hst

(8.2)

A different behaviour of this quantity was noticed (Fig. 8.5), with a minimum at intermediate waste-gate settings (αWG between 20° and 30°), followed by an increase for larger openings. Considering that, for each turbocharger operating condition, measurements were taken at constant overall turbine expansion ratio and inlet temperature, the term Δhst proved to be almost constant. Consequently, the η’tR curve represents the trend of specific turbine work (per unit mass) and is the outcome of the relative changes of rotor mass flow rate (MR) and turbine power (measured on the compressor side of the turbocharger, Pc). Both these quantities are shown in Figure 8.6, referring to a different turbine operating condition. While MR continuously decreases when opening the by-pass port, compressor power shows a significant drop for small openings, followed by a reduced slope section resulting in an increase of η’tR. Even if further investigation into this aspect seems desirable, the turbine efficiency trend observed when this quantity is referred to the fluid portion effectively working within the impeller may be explained by considering the fluid dynamic phenomena occurring in the turbine stage when opening the waste-gate port. For small valve settings, the flow perturbations relating to the interaction between the rotor and the by-pass components in the

Parameter (referred to closed WG condition)

dividing section probably prevail, resulting in a substantial reduction of specific turbine work. 1.05

n/ T3 = 4000 rpm/ K

1.00

p T3 /p4 =1.35

0.95

0.90

0.85

0.80

η't R

0.75

PC MR

0.70

0.65 0

10

20

30

40

50

60

70

Waste-gate opening [deg]

Fig. 8.6 – Effect of waste-gate opening on factors affecting turbine efficiency [40]

99

However, since the rotor mass flow rate significantly decreases for larger waste-gate openings (Fig. 8.4), this effect is partly offset by a reduction in fluid friction losses in the stator and rotor passages, producing the observed change in efficiency.

8.5

Effect of waste-gate opening on propagation phenomena

A broad unsteady flow investigation was developed in order to assess the effect of waste-gate opening on wave propagation phenomena along the turbine feed line and through the turbine itself. First of all, the turbine casing exhaust section was slightly modified by completing the dividing wall between the impeller and the waste-gate flow. Different downstream circuit geometries were then considered (§ 4): in a baseline arrangement (configuration A), a proper mixing pipe was directly coupled to the turbine outlet flange. Two modified arrangements were considered, including an additional dividing wall at the beginning of the mixing pipe. This was extended for 30 mm (configuration B) or 150 mm (configuration C) downstream of the turbine connection plane. Measurements were performed on TC3 (§ 4) for selected mean turbine rotational speed factor (n/√T3=4000 rpm/√K) and inlet pressure (p3m=1.4 bar) constants values, considering different by-pass valve settings (αWG=0, 8, 15, 20, 25, 35, 50 and 60°). The test rig pulse generator system fitted with rotating valves was used and two pulsating flow frequency levels, equal to 40 and 66.67 Hz respectively, corresponding to low engine rotational speed values for which unsteady flow effects in manifolds are usually significant, were considered. Pressure diagrams were measured in different sections of the upstream turbine circuit in order to highlight wave propagation and reflection aspects. As regards the turbocharger turbine, instantaneous wall static pressure was evaluated (by means of suitable pressure taps [33]) at the inlet measuring plane and at the outlet of both the rotor and the waste-gate valve. All signals were acquired for several complete cycles and appropriate averaging techniques were then applied to the measured data. Pulse frequency proved to significantly affect pressure diagrams evaluated along the turbine supply circuit, including the inlet pressure profiles. Substantial modifications of both pulse shape and amplitude, due to the wave action in pipes, were observed [37]. For example, Figure 8.7 shows measured pressure diagrams at an intermediate turbine feed line plane (downstream rotating valves) for two different pulse frequency level. Pressure oscillation profiles, instead, were similar when working at constant pulse frequency and

100

n/ T3 = 4000 rpm/

K

Configuration A

α WG = 20° p 3m = 1.4 bar

1.9

Pressure [bar]

1.8

f = 40 Hz

1.7 1.6 1.5 1.4 1.3 1.2 1.1

0

1

2

3 4 time/pulse period

1.9

f = 66.67 Hz

Pressure [bar]

1.8 1.7 1.6 1.5 1.4 1.3 1.2 1.1

0

1

2

3 4 time/pulse period

Fig. 8.7 – Measured pressure diagrams at different pulsating flow frequencies (section 2) [40] changing the waste-gate position. A clear increase in turbine inlet pulse amplitude at wider by-pass openings was observed as a consequence of the higher total mass flowing through the system (Fig. 8.8). As regards wave propagation phenomena through the turbocharger turbine, attention focused on the influence of the waste-gate setting, also taking the exhaust pipe geometry into account. Figure 8.9 shows measured pressure diagrams at the outlet of both the turbine rotor and the waste-gate valve, for a pulse frequency of 40 Hz and different by-pass settings. Even though the effect of pulse frequency on signal shape proved not to be insignificant, some general considerations can be drawn from analysis of downstream waves. In any case, pulse amplitude at the turbine outlet was lower than the inlet one (see Figure 8.8). However, with larger waste-gate openings, flow at the turbine outlet became substantially unsteady, confirming that the hypothesis of constant downstream pressure for automotive turbocharger turbines (often used within simulation models) is unacceptable and gives rise to significant errors in the evaluation of the instantaneous turbine expansion ratio. Pressure

101

f = 40 Hz

1.9

α WG = 0

1.8 1.7 1.6 1.5 1.4 1.3

Configuration A Turbine inlet pressure [bar]

Turbine inlet pressure [bar]

n/ T3 = 4000 rpm/ K

1.2 0

1

2

α WG =8°

1.8 1.7 1.6 1.5 1.4 1.3

α WG =20°

1.8

1.1

3 4 time/pulse period

1.9

1.7 1.6 1.5 1.4 1.3 1.2 1.1

1.9

1.2

Turbine inlet pressure [bar]

Turbine inlet pressure [bar]

1.1

p3m =1.4 bar

0

1

2

3 4 time/pulse period

1.9

α WG = 60°

1.8 1.7 1.6 1.5 1.4 1.3 1.2

0

1

2

3 4 time/pulse period

1.1

0

1

2

3 4 time/pulse period

Fig. 8.8 – Effect of waste-gate opening on measured turbine inlet pressure diagrams [40] signals at the waste-gate outlet proved to be affected by higher frequency harmonic components, especially at wide by-pass openings. A discussion on this aspect will be developed below. Considering the fact that measurements were taken at a constant average inlet pressure and that an increase in inlet pulse amplitude was also observed when opening the waste-gate valve (Fig. 8.8), pressure signal transmission through the turbine can be effectively analysed by referring to the ratios between outlet and inlet mean pressure level (p4m/p3m) and pulse amplitude (Δp4/Δp3). Figure 8.10 represents these quantities for both the impeller and the bypass outlet as a function of the waste-gate setting, referring to a specific exhaust geometry (configuration B) and a pulse frequency of 66.67 Hz. It can be seen that both average pressure and pulse amplitude ratio rise when opening the by-pass valve. In the range of small wastegate settings (αWG≤15°), pressure signals at the impeller and the by-pass outlet proved to be very similar (see also Fig. 8.9), while with intermediate openings (αWG between 20° and 35°)

102

f = 40 Hz

Configuration B

1.22 1.18

Outlet pressure [bar]

Outlet pressure [bar]

n/ T3 = 4000 rpm/ K

p4R

α WG = 0

p4WG

1.14 1.10 1.06 1.02 0.98

p3m =1.4 bar

1.22

α WG =8°

1.18 1.14 1.10 1.06 1.02

0

1

2

3

0.98

4

0

1

2

1.22 1.18

α WG =20°

1.14 1.10 1.06 1.02 0.98

1

2

3

4

1.22 1.18 1.14 1.10 1.06 1.02

0

3

time/pulse period Outlet pressure [bar]

Outlet pressure [bar]

time/pulse period

4

0.98

α WG = 60° 0

1

2

time/pulse period

3

4

time/pulse period

Fig. 8.9 – Measured pressure diagrams at the rotor and waste-gate outlet [40] the average pressure ratio measured at the waste-gate discharge was lower than that measured at the rotor outlet, though pulse amplitude ratios were still comparable.

0.82 p4mR /p 3m

0.80

p4mWG /p3m

0.78 0.76

f =66.67 Hz

Configuration B Pulse amplitude ratio

Mean pressure ratio

n/ T3 = 4000 rpm/ K

0.74 0.72

p3m =1.4 bar

0.35 0.30

Dp 4R /Dp3

0.25

Dp 4WG /Dp3

0.20 0.15 0.10

0

10

20

30 40 50 60 waste-gate opening [deg]

0.05

0

10

20

30 40 50 60 waste-gate opening [deg]

Fig. 8.10 – Effect of the waste-gate opening on turbine mean pressure and amplitude ratios [40]

103

On the contrary, with wide by-pass openings, both average pressure level and amplitude at the valve outlet clearly exceeded values measured at the rotor outlet, highlighting a significant transmission of flow unsteadiness through the by-pass orifice. Similar behaviour was observed for the different outlet pipe geometries (configurations A and C), regardless of pulse frequency. This denotes that wave propagation phenomena are mainly affected by the transmission and damping characteristics of the relevant flow passages (turbine rotor and bypass port). As mentioned above, pressure diagrams at the waste-gate outlet were often affected by high frequency disturbances. This aspect can be highlighted if a Fourier analysis of measured signals is performed. Figure 8.11 shows pressure waves detected both at the impeller and at the waste-gate outlet, at the same unsteady flow forcing frequency (f1= 66.67 Hz). Pressure diagrams with the by-pass port fully closed (αWG=0) are compared with those measured at an intermediate waste-gate setting (αWG=20°), taking into account different arrangements of the downstream connecting pipe. Fast Fourier Transform (FFT) analysis results are also represented in terms of magnitude of the various harmonics (Δpn). It is important to underline that all acquired signals were pre-processed using the same averaging and filtering techniques. By comparing the different pressure profiles reported in Figure 8.11, the signal measured at the rotor outlet is clearly characterized just by low and intermediate frequency harmonics (fn/f1 less than 10, with significant amplitudes of the first three components), regardless of the outlet pipe configuration. Only in the case of an extended downstream dividing wall (configuration C), are intermediate harmonic components (fn/f1 around 7-8) not completely negligible. On the contrary, waste-gate outlet signals exhibit noticeable noise at higher frequencies (fn/f1 around 40, corresponding to a frequency of about 2.7 kHz in the operating condition considered) when the by-pass port is significantly opened. The noise proved to be more remarkable when no dividing wall was provided in the turbine outlet pipe (configuration A). This confirms the conclusions of a previous investigation performed on a different automotive turbocharger turbine [34] and seems mainly related to significant interference between the flow components from the impeller and the waste-gate valve at the turbine outlet. This hypothesis is confirmed by the reduced amplitude of high frequency harmonics observed when working with a connecting pipe fitted with a dividing wall between the two flow components (configuration B and C), even though the extension of the dividing wall does not seem to play an important role in reducing high frequency noise.

104

n/ T3 = 4000 rpm/ K

p4WG

1.10

1.06

1.06

1.02

1.02

1.02

time/pulse period

0.98

1

Δpn [mbar]

0.5

time/pulse period

50

40

p4R 0

0.5

time/pulse period

0.98

1

50

40

20

20

20

20

10

10

10

10

20

30

40

0

50

1

10

20

30

40

1

1.14

1.10

p4WG

40

1.02

1.02

1.02

0.98

Δpn [mbar]

50

40

0

0.5

time/pulse period

0.98

1

p4R 0

0.5

time/pulse period

20

20

10

10

10

10

0

50

0

1

10

20

30

40

fn/f1

30

40

1.06

1.02

1.02

time/pulse period 1

0.98

0.98 0.5

1 Δpn [mbar]

40

time/pulse period

40

0

0.5

time/pulse period 1

50

0.98

40

20

20

20

10

10

10

10

40

50

fn/f1

0

1

10

20

30

40

50

fn/f1

0

1

10

20

30

40

50

fn/f1

0.5

1

time/pulse period

40

20

30

0

50

30

20

50

p4WG

1.02

30

10

40

1.10

30

1

30

1.14

30

0

20

1.06

p4R

1.02

0

10

αWG = 20°

1.10

1.06

50 Δpn [mbar]

50

1

fn/f1

pressure [bar]

p4WG

0

50

1.14

1.10

1

time/pulse period

fn/f1

Configuration C

1.14

1.06

0.5

20

Δpn [mbar]

p4R

0

10

pressure [bar]

pressure [bar]

1.14

0.98

1

fn/f1

αWG = 0 1.10

50

0.5

40

20

40

0

50

40

40

0.98

1

50

50

50

p4WG

1.02

20

30

40

1.06

30

20

30

1.10

30

10

20

1.14

30

1

10

αWG = 20°

30

0

1

fn/f1

Δpn [mbar]

time/pulse period 1

0

50

1.10

1.06

0.5

30

1.14

1.06

0

20

Configuration B

1.06

0.98

10

1

time/pulse period

fn/f1

pressure [bar]

pressure [bar]

p4R

1.10

0

50

fn/f1

pressure [bar]

10

0.5

40

30

1

0

50

30

1.14

Δpn [mbar]

1.02

30

αWG = 0 pressure [bar]

1.06

30

fn/f1

pressure [bar]

0.98

1

p4WG

1.10

Δpn [mbar]

Δpn [mbar]

40

0

1.14

Δpn [mbar]

0.5

Δpn [mbar]

0

50

0

Δpn [mbar]

1.10

1.06

0.98

αWG = 20°

1.14

pressure [bar]

pressure [bar]

1.14

pressure [bar]

pressure [bar]

p4R

1.10

p3m =1.4 bar

Configuration A

αWG = 0 1.14

f =66.67 Hz

0

1

10

20

30

40

50

fn/f1

Fig. 8.11 – Signals measured at the rotor and waste-gate outlet and related Fourier analysis [40] 105

8.6

Analysis of turbine instantaneous mass flow rate

The experimental study was extended to analysis of instantaneous turbine mass flow rate. In this stage of the investigation, flow unsteadiness was generated using the rotating valve pulse generator and measurements were restricted to one turbine operating condition (n/√T3=4000 rpm/√K, εm=1.3 bar, f=40 Hz), with the by-pass valve completely closed (αWG=0) and partially opened (αWG=20°). Turbine inlet and outlet pressure diagrams and inlet instantaneous mass flow rate were simultaneously measured using high frequency response straingauge transducers and a hot-wire probe respectively. Since no experimental information was available on instantaneous temperature at the turbine inlet under pulsating flow conditions, this quantity was estimated, referring to the adiabatic process of an ideal gas, m =1.3

f = 40 Hz

0.143

1.6

α WG = 0

Mt measured 0.125

Mt QSF

1.5

p3 measured

0.107

pressure [bar]

Mass flow rate [kg/s]

n/ T3 = 4000 rpm/ K

1.4 0.089 1.3 0.071

1.2

0.053

0.035

0.0

1.1 1.0

0.5

0.143

1.6

α WG =20 °

0.125

1.5

pressure [bar]

Mass flow rate [kg/s]

time/pulse period

0.107 1.4 0.089 1.3 0.071

1.2

0.053

0.035

0.0

1.1 1.0

0.5

time/pulse period

Fig. 8.12 – Pressure and mass flow rate profiles at different waste-gate settings [40]

106

using mean pressure and temperature measurements (see Eq.(6.2)). Quasi-steady mass flow calculations were also performed, referring to steady flow curves, in order to assess the accuracy of this prediction procedure. Figure 8.12 shows instantaneous inlet pressure and mass flow rate (both measured and QSF calculated) diagrams vs. nondimensional time referred to the period of the main harmonic component of the pulsating flow. Measured mass flow rate profiles proved to be affected by higher frequency components when compared with pressure and calculated mass flow diagrams. Through FFT analysis, significant harmonic components were detected up to a frequency of approximately 600 Hz, probably related to flow disturbances in the measurement section. QSF mass flow rate profiles

Mass flow parameter [kg√K/s bar]

n/ T3 = 4000 rpm/ K

m =1.3

f = 40 Hz

0.85 0.80

40 Hz

α WG = 0

Steady state

0.75 0.70 0.65 0.60 0.55 0.50

Mass flow parameter [kg√K/s bar]

0.45 1.1

1.2

1.3

1.4

1.5

1.6

pressure ratio

1.80 1.70 1.60

40 Hz

α WG = 20 °

Steady state

1.50 1.40 1.30 1.20 1.10 1.00 0.90 1.1

1.2

1.3

1.4

1.5

1.6

pressure ratio

Fig. 8.13 – Instantaneous mass flow parameter vs. pressure ratio for different waste-gate openings [40]

107

were confirmed to be strictly related to measured pressure diagrams, highlighting oscillation amplitudes comparable to the experimental levels, at least for αWG=0. However, average calculated values slightly underestimated mean measured levels for both operating conditions. A noticeable phase shift (about 1/6 of the oscillation period) was also observed when comparing measured and calculated mass flow profiles, due to the filling and emptying effect in the turbine volute casing. This effect is apparent if instantaneous measured mass flow parameter is plotted against the turbine expansion ratio, as in the case of steady flow performance (Fig. 8.13): the filling and emptying of the volute is highlighted by the hysteresis loop surrounding the steady state curve. As reported by other authors [43], at typical automotive engine pulsating flow frequencies the pulse is so rapid that the volute volume is not filled incrementally with pressure, resulting in the observed hysteresis between measured mass flow rate and pressure. Since the waste-gate opening affects the average mass flow level and the magnitude of unsteady flow phenomena occurring in the turbine circuit (Fig. 8.8), the hysteresis loop surrounding the steady state curve exhibits a deviation when the by-pass port is opened (Fig. 8.13).

108

9. Conclusions and further steps

Turbocharging is becoming a key technology for the development of automotive downsized SI engines working at high brake mean effective pressure (bmep). The application of charge boosting, together with the use of alternative fuels (especially CNG and LPG), can help to attain a substantial reduction in SI engine fuel consumption and CO2 emission, while maintaining the advantages of consolidated aftertreatment devices, high power output and excellent vehicle driveability. It is therefore apparent the interest for dedicated investigations on the turbocharger alone and on significant intake and exhaust engine subsystems in order to get a better understanding of their performance also in unsteady flow conditions and to optimise the relevant control strategies. To this purpose, measurements performed on specific test facilities can supply a lot of information to be used in the development of simulation models and to assess correlation criteria between components behaviour under steady and transient operation. The PhD work moved from this scenario in order to achieve better knowledge on turbocharger behaviour both under steady and unsteady flow conditions. An updated arrangement of the components test facility operating at UNIGE-ICEG and of the relevant measuring system and data analysis procedures was carried out. The pre-existing circuit layout was maintained, through which experimental studies in unsteady flow conditions can be addressed to the analysis of components response when changing the main characteristics of a controlled pulsating flow (i.e., pulse frequency, amplitude, mean value, etc). In addition, a new feeding line, based on a motor driven cylinder head, allowed to extend experimental investigations to a subsystem level taking also account of the circuit geometry. An extensive investigation was developed on three typical automotive turbochargers for gasoline engines, fitted with a waste-gate valve as a regulating system. Compressor and turbine steady flow performance was first measured by using suitable experimental techniques which allowed to considerably extend the explored operating range within a “cold air” experimental investigation. The attention was then focused on unsteady flow performance of the turbocharger turbine: to this purpose, a broad test programme was performed, aimed at highlighting different aspects.

109

In a first step the performance of the two available pulse generator systems was compared for the same turbine operating conditions: measured pressure diagrams were analysed referring to the circuit geometry and to the pulse generator device characteristics. Flow unsteadiness at the turbine inlet resulted quite different when using the two feeding circuits, with considerably smaller amplitudes in the case of the cylinder head system. The exhaust manifold geometry seems to play a fundamental role to determine this result, especially as regards the mixing zone before the turbine inlet. On the contrary, the damping effect of the volume interposed between the pulse generator and the turbocharger seems less important. These aspects should be taken into account in the design of the engine exhaust circuit in order to optimise the energy transfer from the engine to the turbine when a pulse turbocharging scheme is used. The effect of pulsating flow frequency on pressure signals at the turbine inlet and outlet was also analysed: wave action in the turbine feeding line can significantly affect both the pulse shape and its amplitude, while the damping effect of the turbocharger turbine is reduced in the case of small volume units, resulting in significant unsteady phenomena also downstream of the turbine. This is to be considered when the instantaneous turbine expansion ratio is evaluated to be used within provisional models based on a quasi-steady flow approach. The influence of engine exhaust circuit geometry on flow unsteadiness was also experimentally analysed, focusing on the pressure diagrams measured in different sections. In the case of a ‘4 into 1’ exhaust manifold design, flow unsteadiness in each pipe proved to be substantial, even if higher frequency pulses due to transmitted and reflected waves were detected. On the contrary, the junction of four manifold pipes in one mixing volume caused a substantial decrease of flow unsteadiness, resulting in a reduced pulse amplitude at the turbine inlet. Several configurations of the exhaust line were taken into account, both improving its finishing and considering the insertion of a dividing wall to better propagate pressure pulses to the turbine. A noticeable increase of turbine inlet pulse amplitude was found when using a dividing wall generating two ducts with equally spaced pulses, due to the pulse separation in the manifold mixing volume. Further work on this subject seems desirable, in order to optimise the manifold design with respect to the turbine inlet energy level in unsteady operation, also using calculations performed through wave action models. The effect of unsteady flow characteristics on the available energy at the turbine entry was therefore deepened within the investigation, referring both to theoretical inlet pulses and to the real pressure profiles measured in different operating conditions. The isentropic specific

110

inlet gas energy along the pulse cycle was calculated starting from the evaluation of instantaneous gas energy, with some assumptions related to available experimental information. The energy level in unsteady flow conditions proved to be related to the pulse amplitude at the turbine entry, which was affected, at constant turbine rotational speed, both by the pulse frequency and the mean pressure level. The calculation of available inlet gas energy allowed a proper estimation of turbine overall efficiency under pulsating flow operation: the relevant results were compared with the efficiency values calculated on the basis of mean measured levels and with the steady flow values measured on the test rig. The effect of unsteady flow on turbine efficiency and the error induced by evaluating the pulsating flow efficiency on the basis of mean inlet parameters were highlighted. A wide investigations was finally performed on a small turbocharger for downsized SI automotive engines in order to highlight turbine behaviour when the waste-gate valve is opened, both under steady and unsteady flow operating conditions. In a first step the study was aimed at evaluating the swallowing capacities of both the turbine and the waste-gate valve and the total mass flow split between the two flow ducts. This investigation suggested several hints for further analysis: 

both the waste-gate and the turbine rotor swallowing capacity decreased with flow occurring on both devices, probably due to the reduced effective expansion ratio across the impeller and the by-pass port;



consequently, the assumption that the total swallowing capacity of the system can be calculated by summing turbine and waste-gate mass flow contributions at the same overall expansion ratio proved to be inadequate, resulting in errors between 10 and 25 percent;



overall turbine efficiency, evaluated on the basis of the isentropic expansion of the entire mass flowing through the system (turbine rotor and waste-gate) significantly decreased when opening the waste-gate valve;



a different behaviour was observed if turbine efficiency is referred on the fluid portion effectively working within the rotor, with a minimum at intermediate waste-gate setting followed by an increase for larger openings.

The analysis was then extended to turbocharger unsteady flow operation, focusing on the effect of the waste-gate opening on wave propagation phenomena, taking into account

111

different turbine outlet pipe geometries. The main results provided by this stage of the investigation can be summarized as follows: 

a noticeable increase in turbine inlet pulse amplitude at wide by-pass openings was observed as a consequence of the higher total mass flowing through the system;



the pulse amplitude at the turbine outlet was lower than the inlet one; however, with significant waste-gate openings, flow at the turbine outlet became substantially unsteady;



both the average pressure and pulse amplitude levels at the rotor and waste-gate outlet increased when opening the by-pass valve;



with wide by-pass openings, mean pressure level and amplitude at the valve outlet clearly exceeded values measured at the rotor outlet, highlighting a preferential transmission of flow unsteadiness through the by-pass orifice;



pressure signals at the waste-gate outlet proved to be affected by high frequency harmonic components when the device was opened, probably due to interactions between the flow from the impeller and the waste-gate valve at the turbine outlet. The use of an outlet pipe fitted with a dividing wall reduced the high frequency noise, regardless of the extension of the dividing diaphragm.

A preliminary analysis into the effect of the waste-gate opening on instantaneous turbine mass flow diagrams showed that: 

measured mass flow rate profiles were affected by higher frequency components when compared with pressure diagrams, probably related to flow disturbances in the measuring section;



instantaneous mass flow rate calculated according to a QSF assumption was found to be strictly related to experimental pressure profiles and measured and calculated mass flow oscillations showed comparable amplitudes;



a noticeable phase shift between measured and calculated mass flow profiles was observed, due to the filling and emptying effect in the turbine volute casing;



the instantaneous experimental mass flow parameter showed a significant hysteresis loop surrounding the steady state curve when plotted against the turbine expansion ratio. The loop configuration proved to be affected by the waste-gate opening.

The experimental investigation performed at UNIGE-ICEG Laboratory was supported by a numerical analysis carried out at Politecnico di Milano. An attempt to describe turbine

112

behavior under unsteady flow conditions was performed, developing a model in GASDYN code that accounts for the case volume and that introduces an equivalent scroll length in the calculation procedure. A validation of proposed model was performed by comparing the simulation results with the experimental measurements, referring to a small turbocharger for downsized SI automotive engines. The numerical investigation has shown that the pulsations downstream of the turbine may be significant and that the quality of the results concerning with the passage of the pressure waves through the turbine is related to a correct schematization of the pipes downstream of the turbocharger. The model is able to reproduce with a good agreement pressure signals passing through the turbine; it is also able to give a prediction of the behavior of the turbine under unsteady flow conditions, although the predictions of the above mentioned mass flow hysteresis loop cannot be considered as satisfying. This means that locating the effective position of the turbine boundary halfway around the casing is not enough to capture the unsteady flow phenomena. Future work should be taken into account focused on the introduction of a corrective length (function of the pulsation frequency) to be added to the volute equivalent length in the schematic layout modeled in GASDYN code. Measurements of the instantaneous turbocharger rotational speed should also be carried out, in order to estimate the turbine instantaneous actual work; it will be therefore possible to describe turbine efficiency diagrams over the pulse period, thus improving the predictive capability of turbine behaviour under pulsating flow conditions. In a next step the experimental and numerical investigations will be extended to a wide range of waste-gate opening, in order to highlight propagation phenomena through the regulating system and to consider the effect of the regulating device on turbine performance both under steady and unsteady flow conditions. In author’s opinion another important topic related to turbocharger operation to be investigated in future steps is the interaction between the unsteady flow generated by engine intake valves and the turbocharger compressor with special reference to the surge limit. Theoretical and experimental investigation on this subject is scheduled at UNIGE-ICEG.

113

Acknowledgments

This work was partly developed within the EU Integrated Project “NICE - New Integrated Combustion System for Future Passenger Car Engines” (6th Framework Programme, Priority 6.2, Contract TIP3-CT-2004-506201) and the MIUR-PRIN 2005 Project “Turbocharging systems for downsized automotive ICE”. The author would like to thank the European Commission and the Italian Government for their financial support.

114

Nomenclature

Definitions and acronyms n/√T3 u/cs I K (Mt · √T3) / p3 cyl CNG EC EO ECU FFT GDI LPG NSF QSF SF TC TDC VVA

turbine rotational speed factor blade speed ratio influence factor ratio between mean pulsating and steady flow performance turbine mass flow factor cylinder Compressed Natural Gas Exhaust Valve Closing Exhaust valve opening Electronic Unit Control Fast Fourier Transform Gasoline Direct Injection Liquefied Petroleum Gas Non Steady Flow Quasi Steady Flow Steady Flow Turbocharger Top Dead Center Variable Valve Actuation

Nomenclature Notation a c f h k n p r s t A C D E M N P R

sound velocity specific heat pulse frequency specific gas enthalpy (referred to mass unit) specific heat ratio rotational speed pressure radius blade thickness time waste-gate opening position, flow area, entropy velocity diameter specific energy transfer mass flow rate number of acquisition channels power gas constant

115

T U

temperature tangential velocity

W α β ε η λ µ ρ σ ∆ φ

energy flow rate waste-gate opening angle, nozzle angle pressure ratio expansion ratio efficiency incident Riemann variable dynamic viscosity density slip factor amplitude, difference phase, duty cycle



Subscripts 0 1 2 3 4 a b c d i m n p r s t ϑ ψ amb eq in nd out ref tot R Y CH NSF QSF SF

ambient condition component inlet section, referred to forcing frequency component outlet section turbine entry turbine exit apparent, blade tip blade hub compressor delay referred to time interval mean, mechanical referred to n-th harmonic component, number of time intervals constant pressure, pipe actual, radial referred to isentropic process, referred to sampling rate turbine tangential component azimuth angle ambient equivalent inlet non dimensional outlet reference total referred to turbine rotor Y-junction section acquisition channels non steady flow value quasi-steady flow value steady flow value

116

TS TT WG

total to static total to total referred to waste-gate

117

References

1. Scolari P., Towards a Sustainable Mobility at the Turn of the Century: the Environmental Challenge, 6th EAEC Conference, Cernobbio, 1997 2. Capobianco M., Improvement of Automotive Diesel Engine for the 21st Century Challenge, International Conference on Advances in Vehicle Control and Safety, AVCS, Amiens, 1998 3. Stumpp G., Ricco M., Common Rail – An Attractive Fuel Injection System for Passenger Car DI Diesel Engines, SAE Paper 960870, 1996 4. O’Connor G.F., Smith M.M., Variable Nozzle Turbochargers for Passenger Car Applications, SAE Paper 880121, 1988 5. Fraidl, G., Piock W., Fuerhapter A., Sikinger H., Pinter A., The Potential of Next Generation Gasoline Direct Injection Technologies, International Conference “Spark Ignition Engine: the CO2 Challenge”, paper no.02A5008, Venezia, 2002 6. Menne R.J., Foulkes D., Shamel A., Pingen B., Limbach S., Stratified DI or Homogenous Downsized DI – An Analysis of Competing Gasoline Fuel Economy Concepts, International Conference “Spark Ignition Engine: the CO2 Challenge”, paper no.02A5004, Venezia, 2002 7. Denger D., Kapus P., The CO2 Reduction Potential of Gasoline Engines with Variable Valve Train Systems, International Conference “Spark Ignition Engine: the CO2 Challenge”, paper no.02A5009, Venezia, 2002 8. Richter H., Schwarzenthal D., Spiegel L., Variable Valve Actuation – Key Technology for High Specific Power Output and Low Exhaust Emissions, International Conference “Spark Ignition Engine: the CO2 Challenge”, paper no.02A5010, Venezia, 2002 9. Lecointe B., Monnier G., Downsizing a Gasoline Engine Using Turbocharging with Direct Injection, SAE paper 2003-01-0542, 2003 10. Wirth M., Mayerhofer U.,Piock W.F., Fraidl G.K., Turbocharging the DI Gasoline Engine, SAE paper 2000-01-0251, 2000 11. Lake T., Strokes J., Muphy R., Osborne R., Turbocharging Concepts for Downsized DI Gasoline Engines, SAE paper 2004-01-0036, 2004 12. Petitjean D., Bernardini L., Middlemass C., Shahed S.M., Advanced Gasoline Engine Turbocharging Technology for Fuel Economy Improvements, SAE paper 2004-010988, 2004 13. Winterbone D.E., Nikpour B., Frost H., A Contribution to the Understanding of Turbocharger Turbine Performance in Pulsating Flow, International Conference on Internal Combustion Engine Research, Paper no.C433/011, Instn. Mech. Engrs., London, 1991 14. Acton O., Capobianco M., Il Banco Prova per Gruppi Turbosovralimentatori del Laboratorio di Macchine di Genova, XLI Congresso Nazionale ATI, Napoli, 1986 15. Ferrari G., Onorati A. and Piscaglia F., Fluid Dynamic Simulation of a Six-Cylinder S.I. Engine with Secondary Air Injection in the Exhaust After-Treatment System, ICE2003 – SAE International Conference on Internal Combustion Engine, SAE paper 2003-0116, Capri (Napoli – Italy), September 14-19, 2003

118

16. Montenegro G., Piscaglia F., Onorati A., Catalano G. and Cioffi, P., A 1D Unsteady Thermo-Fluid Dynamic Approach for the Simulation of the Hydrodynamics of Diesel Particulate Filter, SAE Int. Congress & Exp. (Detroit, Michigan) 2006, SAE paper 2006-01-0262, April 3-7, Detroit (MI), 2006. SAE Transactions, Journal of Engines 2007 17. Ferrari G., Piscaglia F. and Onorati A., 1D Modeling of the Hydrodynamics and of the Regeneration mechanism in Continuous Regenerating Traps, ATI-SAE International Session, Perugia (Italy), paper n. 2006-01- 3011, September 14, Perugia (Italy),2006. SAE Transactions, Journal of Engines 2007 18. Young Y. Michael, Penz David A., The design of a new turbocharger test facility, Power Boost, International Congress and Exposition Detroit, Michigan, 1990 19. Dale A., Watson N., Vaneless radial flow turbine performance, 3rd International Conference Turbocharging and Turbochargers, Paper C110/86, Instn Mech Engrs, London, 1986 20. Winterbone D.E., Nikpour B. and Alexander G.I., Measurement of the performance of a radial inflow turbine in steady and unsteady flow, 4th International Conference Turbocharging and Turbochargers, Paper C405/15, Instn Mech Engrs, London, 1990 21. Arcoumanis C., Hakeem I.,Khezzar L., Martinez-Botas R.F., Performance of a Mixed flow turbocharger turbine under pulsating flow conditions, ASME 95-GT-210, 1995 22. Benson R.S., Scrimshaw K.H., An experimental investigation of non-steady flow in radial gas turbine, Paper 23, Instn Mech Engrs, London, 1965 23. Osnaghi C., Giuliani P., Calderoni F., Comportamento della turbina di un sovralimentatore per autotrazione in regime pulsante, XL Congresso Nazionale A.T.I., Trieste, 1985 24. Capobianco M., Gambarotta A., Unsteady flow performance of turbocharger radial turbines, 4th International Conference Turbocharging and Turbochargers, Paper C405/017, Instn Mech Engrs, London, 1990 25. Capobianco M., Gambarotta A., Variable geometry and waste-gated automotive turbochargers: measurements and comparison of turbine performance, ASME Transactions, Journal of Engineering for Gas Turbine and Power, Vol. 114, 1992 26. Capobianco M. and Gambarotta A., Performance of a Twin-Entry Automotive Turbocharger Turbine, ASME Energy-sources Technology Conference and Exhibition, paper 93-ICE-2, Houston, 1993 27. Capobianco M., Gambarotta A., Silvestri P., Effects of volute configuration and flow control system on the performance of automotive turbocharger turbines, ASME Internal Combustion Engine Division Spring Technical Conference, paper 95-ICE-11, Marietta, 1995 28. Capobianco M., Buratti R., Gambarotta A., Considerazioni sulle metodologie di previsione quasi-stazionaria applicate a piccole turbomacchine radiali, XLIII Congresso Nazionale ATI, Ancona, 1988 29. Watson N. and Janota M.S., Turbocharging the Internal Combustion Engine, London, McMillan Press, 1982 30. Wallace F.J., Cave P.R., Matching of high-output diesel engines with associated turbomachinery, Paper 48/73, Instn Mech Engrs, London, 1973 31. Dixon S.L., Fluid mechanics, thermodynamics of turbomachinery, Pergamon, Oxford, 1978

119

32. Miyashita T., Tomita T., Ishihara D., Performance of invar radial flow turbine under un steady flow conditions, IHI Engineering Review, Vol.7, 1974 33. Capobianco M, Marelli S., Turbocharger Turbine Performance Under Steady and Unsteady Flow: Test Bed Analysis and Correlation Criteria, 8th International Conference Turbocharging and Turbochargers, Paper C647, Instn Mech Engrs, London, 2006 34. Capobianco M., Gambarotta A., Cipolla G., Effect of inlet pulsating pressure characteristics on turbine performance of an automotive wastegated turbocharger, SAE Transactions, Journal of Engines, Vol.99, Section 3, Part 1, 1990 35. Capobianco M., Gambarotta A., Nocchi M., Sperimentazione in regime instazionario su componenti dei sistemi di ricambio della carica di motori a c.i.: problematiche di misura e di analisi delle prestazioni, XLVI Congresso Nazionale ATI, Gaeta, 1991 36. Capobianco M., Marelli S., Transient performance of automotive turbochargers: test facility and preliminary experimental analysis, 7th International Conference on Engine for Automobile – ICE2005 – Isola di Capri, SAE Paper 2005-24-066, 2005 37. Winterbone D.E., Pearson R.J., Design Techniques for Engine Manifolds: Wave Action Methods for IC Engines, Wiley, 1999 38. Zinner K., Supercharging of Internal Combustion Engines, Springer – Verlag, 1978 39. Capobianco M., Marelli S., Unsteady flow behaviour of the turbocharging circuit in downsized SI automotive engines, FISITA World Automotive Congress 2006 – F2006P116 – Yokohama, 22-27 Ottobre, 2006 40. Capobianco M., Marelli S., Waste-gate turbocharging control in automotive SI engines: effect on steady and unsteady turbine performance, 14th Asia Pacific Automotive Engineering Conference APAC, Hollywood, 5-8 Agosto – SAE Paper 2007-01-3543, 2007 41. R S Benson, Non steady flow in a turbocharger nozzleless radial gas turbine, SAE paper 740739, 1974 42. Capobianco M., Marelli S., Unsteady flow turbine performance in turbocharged automotive engines, 11th European Automotive Congress EAEC 2007, Budapest, 30 Maggio-1 Giugno, 2007 43. Szymko S., Martinez-Botas R.F., Pullen K.R., Experimental Evaluation of Turbocharger Turbine Performance under Pulsating Flow Conditions, Proceedings of ASME Turbo Expo 2005: Power for Land, Sea and Air, paper GT2005-68878, RenoTahoe, Nevada, USA, 2005 44. Lujan J.M., Galindo J., Serrano J.R., Efficiency Characterization of Centripetal Turbines under Pulsating Flow Conditions, SAE Paper 2001-01-0272, 2001 45. Piscaglia F., Onorati A., Marelli S., Capobianco M., Unsteady behaviour in turbocharger turbines: experimental analysis and numerical simulation, 8th International Conference on Engines for Automobile (ICE 2007) – Isola di Capri, 1610 Settembre 2007 – SAE Paper 2007-24-0081, 2007 46. Costall A., Szymko S., Martinez-Botas R.F., Filsinger D., Ninkovic D., Assessment of Unsteady Behaviour in Turbocharger Turbines, Proceedings of ASME Turbo Expo 2006: Power for Land, Sea and Air, paper GT2006-90348, Barcelona, Spain, 2006 47. Kosuge H., Yamanaka N., Ariga I. and Watanabe I., Performance of radial flow turbines under pulsating flow conditions, ASME Transactions, Journal of Engineering for Gas Turbine and Power, 1976

120

48. Benson R. S., The Thermodynamics and Gas Dynamics of Internal Combustion Engines - Vol. 1. Edited by J. H. Horlock, and D. E. Winterbone. Oxford: Clarendon Press, 1982 49. Andersen J., Karlsson E., Gawell A., Variable Turbine Geometry on SI Engines, SAE paper 2006-01-0020, 2006

121