Electric Turbocharging for Energy Regeneration and Increased Efﬁciency at Real Driving Conditions

applied sciences

Article Electric Turbocharging for Energy Regeneration and Increased Efﬁciency at Real Driving Conditions

Pavlos Dimitriou 1,*, Richard Burke 1, Qingning Zhang 1, Colin Copeland 1 and Harald Stoffels 2 1 Department of Mechanical Engineering, University of Bath, Bath BA2 7AY, UK; [email protected] (R.B.); [email protected] (Q.Z.); [email protected] (C.C.) 2 Powertrain Research & Advanced, Ford-Werke GmbH, Köln-Merkenich, Cologne D-50725, Germany; [email protected] * Correspondence: [email protected]; Tel.: +44-1225-38-3313

Academic Editor: Jose Ramon Serrano Received: 23 January 2017; Accepted: 14 March 2017; Published: 1 April 2017

Abstract: Modern downsized internal combustion engines benefit from high-efficiency turbocharging systems for increasing their volumetric efficiency. However, despite the efficiency increase, turbochargers often lack fast transient response due to the nature of the energy exchange with the engine, which deteriorates the vehicle’s drivability. An electrically-assisted turbocharger can be used for improving the transient response without any parasitic losses to the engine while providing energy recovery for increasing overall system efficiency. The present study provides a detailed numerical investigation on the potential of e-turbocharging to control load and if possible replace the wastegate valve. A parametric study of the optimum compressor/turbine sizing and wastegate area was performed for maximum torque, fast response time and energy regeneration across the real driving conditions speed/load area of the engine. The results showed that the implementation of a motor-generator could contribute to reducing the response time of the engine by up to 90% while improving its thermal efficiency and generating up to 6.6 kWh of energy. Suppressing the wastegate can only be achieved when a larger turbine is implemented, which as a result deteriorates the engine’s response and leads to energy provision demands at low engine speeds.

Keywords: turbocharger; e-turbo; boosting; electrically-assisted; turbo-compound; energy regeneration; internal combustion engines; 1D simulation

1. Introduction

The demand for low fuel consumption and CO2 generation vehicles over the last few years has popularly increased the necessity of downsizing and increasing the overall thermal efficiency of Internal Combustion (IC) engines. Downsizing is the process of reducing the volumetric capacity of an engine for reduced throttling and friction losses while its boosting capabilities need to be increased for higher specific heat. This can be achieved by the implementation of a boosting device (turbocharger or supercharger) for increased air pressure at the intake of the engine and therefore higher volumetric efficiency. A turbocharger is a device that recovers the waste energy from the engine’s exhaust gasses and uses it to compress the air at the engine’s intake. The level of compression is directly linked to the amount of air passing through the turbine, and it can be controlled by either bypassing part of the flow through a Wastegate (WG) or by changing the nozzle position of the turbine (VGT, Variable-geometry turbocharger). The proportion of the waste-gated flow can be up to 50% for high speed and load conditions, which imply a vast amount of unexploited energy. The drawback of this device is that due to the nature of the energy exchange between the engine and the turbocharger (filling of the intake and exhaust manifolds and low exhaust energy at low speeds/loads), the transient performance during

Appl. Sci. 2017, 7, 350; doi:10.3390/app7040350 www.mdpi.com/journal/applsci Appl. Sci. 2017, 7, 350 2 of 25 engine load increase is relatively poor, which deteriorates the drivability of the vehicle [1]. On the other hand, a supercharger is a device mechanically driven by the engine to increase its volumetric efficiency. It has a very fast response time in transient conditions, but the power required for its operation is a parasitic loss for the engine; therefore, it is not widely used in recent technologies. The parallel use of a turbocharger and a supercharger could potentially improve the transient performance of the system during load increase while reducing the engine losses compared to a solely supercharger boosted system. However, this will increase the complexity, as a two-stage boosting system is required [2,3]. An electrically-assisted turbocharger of a larger size and with higher efficiency could be used in case a simpler one-stage boosting system is needed. The motor can provide the electricity required at the periods of load increase for a faster response, while the turbocharger works as a conventional system during steady-state and tip-out conditions. Katrasnik et al. [4] investigated the influence of an electric motor attached to the turbo shaft on the transient response of a diesel engine. It was found that the time required to perform transient power increase was reduced from 3.9 s for the original conventional turbocharger down to 1.7 s. Ibaraki et al. [5] tested a hybrid turbo developed by Mitsubishi Heavy Industries under transient operating conditions. An improved engine peak torque and enhanced transient response compared to a conventional system was demonstrated. Torque was enhanced by 18% at low engine speeds, while the response time was reduced by 70% when 2 kW of turbo motor assistance was provided. Millo et al. [6] investigated the potential of an electrically-assisted turbocharger for a heavy-duty diesel engine to evaluate the turbo-lag reductions and the fuel consumption savings that could be obtained in an urban bus at different operating conditions. The system allowed fuel consumption reductions of 6% to 1%, depending on the driving cycle, with lower values corresponding to congested traffic conditions. Burke [7] applied various electric boosting systems to a gasoline and a diesel engine and evaluated their steady state and transient performance from the perspective of the air path. The author compared the performance characteristics of an electrically-assisted turbocharger with those of a two-stage system with an electrically-driven compressor placed before or after the main waste-gated turbocharger. He found that under steady-state operating conditions, there was significant system efficiency to installing the electric compressor downstream of the turbocharger’s compressor. The author also concluded that an electrically-assisted turbocharger is not ideal as a replacement for the second compressor in a two-stage system, as it will push the compressor into surge. However, it was mentioned that this could be overcome through re-matching of the compressor. Bumby et al. [8,9] investigated the technical problems in selecting an appropriate machine to use with an electrically-assisted turbocharger. The authors also demonstrated that the required time for accelerating an electrically-assisted turbocharger from 40 to 110 krpm could be around 50% less than a conventional turbocharger. However, albeit an electrically-assisted turbocharged engine requires less energy for its operation than a solemnly supercharged engine, the power required is still a parasitic loss for the engine, which results in an increased fuel consumption. Divekar et al. [10] proposed an electrical supercharging and a turbo-generation system integrated in a diesel engine for overcoming the issue of parasitic losses. The proposed system showed distinct transient response improvement benefits over a conventional turbocharged system and 7% improvement in fuel consumption over the Federal Urban Driving Schedule (FUDS) cycle. Furthermore, during transients and high load operation, the proposed system did not build up exhaust backup pressure in order to accelerate the supercharger, and as such, no additional pumping losses were incurred. However, the authors commented that the electrical energy required by the supercharger can be only partially obtained by the turbo-generation system, which still leads to additional energy losses. Panting et al. [11] was one of the first research groups to implement the idea of a motor-generator electrical turbocharger for a 5.2 L truck diesel engine in a theoretical study. Adding a directly-coupled motor-generator offers tremendous advantages to the operation of the turbocharger. It abolishes the requirement of the turbine and compressor power to be matched under steady-state conditions while it assists the acceleration and deceleration of the shaft during transients with no engine energy needs. Appl. Sci. 2017, 7, 350 3 of 25

Over the last few years, keen interest has been shown in the numerical and experimental investigation of the motor-generator technology application in high-duty engines by several research groups. Terdich and Martinez-Botas [12] experimentally characterized a variable geometry turbocharger with a motor-generator technology. The authors found that the motor-generator is capable of delivering a maximum shaft power of 3.5 kW in motoring mode and 5.4 kW in generating mode. The peak electrical efficiency was more than 90% in both modes and occurred at 120,000 revs/min. Airse et al. [13] developed a comprehensive powertrain model to evaluate the benefits of an electric turbo-compound, working in both generator and motor mode, in reducing CO2 emissions from small diesel passenger cars. The simulations showed a reduction in CO2 and fuel consumption of over 4% for the New European Driving Cycle (NEDC). Algrain [14] developed an advanced control system for a motor-generator fitted in a heavy-duty diesel engine for improving the overall fuel efficiency. The simulation results showed that at the rated power, the fuel consumption of a Class-8 on-highway truck engine could be reduced by almost 10%, while the overall reduction in fuel consumption was estimated to be around 5%. Pasini et al. [15] focused on the evaluation of the benefits resulting from the application of an Electric Turbo Compound (ETC) to a small-sized twin-cylinder Spark-Ignition (SI) engine and to a four-cylinder Compression-Ignition (CI) engine with the same power rating. They found that by absorbing electrical energy from the battery, the ETC can lead to significant Brake-Specific Fuel Consumption (BSFC) reductions of up to 4% at the highest engine speeds and loads for the SI engine and up to 6% for the CI engine at 4000 rpm half load. Furthermore, calculations have shown that in the case of the CI engine, the maximum electric power that can be recovered by the ETC is around 4 kW at 4000 rpm full load, while in the case of the SI engine, the maximum power is 1.5 kW. Tavcar et al. [16] presented a comprehensive study on engine performance improvement attributable to the application of different electrically-assisted turbocharger topologies, including a single-stage turbocharger, an electrically-assisted turbocharger, a turbocharger with an additional electrically-driven compressor and an electrically-split turbocharger (supercharger and turbo-generator). The results revealed that all of the electrically-assisted turbocharger topologies improve the transient response of the engine and, thus, the drivability of the vehicle. However, no electrically-assisted turbocharger topology could clearly be favoured in general. In this paper, a detailed investigation of the potential of e-turbocharging to control load while providing energy recovery for increasing the overall system efficiency and if possible replace the wastegate boost control is provided. The current approach of e-turbocharging requires larger turbine systems that do not build up backup pressure and provide electrical assistance at low engine speeds. However, with this configuration, the energy regeneration occurs only at high speed and load areas of the engine, as shown in Figure1a. The proposed study focusses on shifting the energy regeneration towards the low-speed area, which represents more realistic driving conditions, as shown in Figure1b. Energy assistance is provided to the engine when the maximum power characteristics of the engine need to be met. The study is performed on a 2.0 L turbocharged SI engine under steady-state and transient driving conditions. The research work focuses on the availability of energy and the effects of component sizing, the transient behaviour for eliminating turbo lag and the overall system energy balance during various driving conditions. After a comprehensive introduction in this section, Section2 outlines the engine model used for the study and any relatively small modifications occurring for the purpose of electrifying the turbocharger. A detailed model validation against experimental data is also presented in Model Validation. Section3 describes the methodology followed for all of the simulation studies conducted under steady-state and transient conditions. The simulation results for all of the studies performed are analysed and discussed in Section4. Finally, Section5 summarizes the main findings of this work and proposes future work in the area of the electrification of turbocharging systems. Appl. Sci. 2017, 7, 350 4 of 25 Appl. Sci. 2017,, 7,, 350350 4 of 25

FigureFigure 1.1. Energy regenerationregeneration areasareas overover thethethe load/speedload/speedload/speed mapmap ofof an an engine: (( a)) current current approach;approach; (((bb))) desireddesireddesired operation.operation.operation.

2.2. Computational ModelModel TheThe presentpresent numericalnumerical studystudy waswas performedperformed onon aa 1D1D gasgas dynamicdynamic environmentenvironment usingusing thethe GT-PowerGT-Power software software tool tool (v7.5.0, (v7.5.0, Gamma Gamma Technologies, Technologies, LLC., LLC., Westmont, Westmont, IL, USA, IL, USA, 2014). 2014). The modelThe model used forused this for study, this study, shown shown in Figure in 2Figure, represents 2, represents a 2.0 L spark-ignitiona 2.0 L spark-ignition turbocharged turboc engine.harged Theengine. model, The previouslymodel, previously used in used [17], in was [17], provided was provided by the engineby thethe supplier,engineengine supplier,supplier, and it has andand been itit hashas validated beenbeen validatedvalidated against experimentalagainst experimental data in data both in steady-state both steady-state and transients andand transientstransients by POWERTECH byby POWERTECHPOWERTECH Engineering EngineeringEngineering (see Model (see(see Validation).Section 2.1).

Figure 2. 1D1D model model of of the the baseline baseline 2.0 2.0 L L turbocharg turbochargedturbocharged engine usedused inin thethe presentpresent study.study. Appl. Sci. 2017, 7, 350 5 of 25 Appl. Sci. 2017, 7, 350 5 of 25

The baseline 2.0 L engine model was used only for the Phase 1 study of the steady-state simulations,The baseline which 2.0 included L engine a wastegate model was enthalpy used loss only study for the for Phasequantifying 1 study theof energy the steady-state availability simulations,across the speed/load which included map of athe wastegate engine. enthalpy loss study for quantifying the energy availability acrossFor the the speed/load Phase 2 and map Phase of the 3 of engine. the steady-state simulations, as well as the transient simulations includedFor the in Phasethis paper, 2 and Phasethe model 3 of the was steady-state modified simulations, by implementing as well asa themotor/generator transient simulations (M/G) includedcomponent in thismodel. paper, The the motor/generator model was modiﬁed was connected by implementing directly ato motor/generator the turbocharger’s (M/G) shaft, component as shown model.in Figure The 3. motor/generator was connected directly to the turbocharger’s shaft, as shown in Figure3.

FigureFigure 3.3. Motor/generatorMotor/generator component model used for co convertingnverting the model to an e-turbo engine.

The originaloriginal model model was was fitted fitted with with a wastegate/boost a wastegate/boost pressure pressure and a throttle/brakeand a throttle/brake mean effective mean pressureeffective pressure proportional-integral-derivative proportional-integral-derivative (BMEP PID)(BMEP controllers PID) controllers for achieving for achieving the targeted the targeted boost pressureboost pressure and BMEP. and TheBMEP. target The values target for values different for different engine speeds engine and speeds loads and were loads provided were toprovided the model to usingthe model look-up using tables. look-up For thetables. electrically-assisted For the electrically-assisted model, twonew model, PID two controllers new PID were controllers introduced were or replacedintroduced the or original replaced controllers the original as necessary, controllers depending as necessary, on the type depending of study. on These the are:type of study. These are: • A motor/generator (M/G)/boost pressure controller for achieving the desired intake pressure • A motor/generator (M/G)/boost pressure controller for achieving the desired intake pressure • A WG/pre-turbine pressure controller for achieving the desired pre-turbine pressure • A WG/pre-turbine pressure controller for achieving the desired pre-turbine pressure The compressor and turbine mass flow multipliers were modified and spanned in the range of 0.7 to 1.3The for compressor the purpose and of undertaking turbine mass parametric flow multiplier studiess were with modified different componentand spanned sizing, in the as range shown of in0.7 Equation to 1.3 for (1). the By purpose restricting of undertaking or enhancing parametric the mass flowstudies rate with within different a component, component the sizesizing, of the as componentshown in Equation can be simulated. (1). By restricting or enhancing the mass flow rate within a component, the size of the component can be simulated. R · ∫ṁdm dt AverageAverage mass mass flow flow rate rate= =Multiplier Multiplier× R (1)(1) ∫ddt wherewhere mṁ˙ isis thethe massmass flowflow raterate throughthrough thethe partpart (in(in thethe casecase ofof thethe turbine,turbine, thethe wastegatewastegate mass flowflow isis excluded).excluded). Although thisthis approachapproach wouldwould not not promise promise a a high high level level of of accuracy, accuracy, it isit is deemed deemed to to be be reliable reliable to predictto predict the trendthe trend of the of overall the overall system’s system’s behaviour behaviour when thewhen turbine the orturbine compressor or compressor sizing is increasedsizing is orincreased decreased or compareddecreased tocompared the original to the components original components based on a highly-calibrated based on a highly-calibrated engine. Furthermore, engine. theFurthermore, wastegate areathe wastegate was modified area fromwas modified completely from closed completely to diameters closed larger to diameters than the baselinelarger than engine. the However,baseline engine. due to However, the change due of to the the size change of critical of the components, size of critical the componen operationts, of the the operation engine and of the its mainengine parameters and its main needs parameters to be closely needs monitored to be closel to ensurey monitored realistic to performance.ensure realistic For performance. this reason, theFor parametersthis reason, the limits parameters shown in limits Table shown1 were in set Table and 1 established, were set and which established, were not which violated were not during violated the parametricduring the studies.parametric studies.

Table 1.1. MaximumMaximum limitslimits setset forfor thethe parametricparametric studies.studies.

Parameter Unit Value Parameter Unit Value Engine torque (Nm) 304 EngineTurbine torque inlet temperature (Nm) (°C) 1000 304 Turbine inlet temperature (◦C) 1000 Turbocharger speed (krpm) 200 Turbocharger speed (krpm) 200 Pre-turbine pressure (bar) 3 Pre-turbine pressure (bar) 3 Post-compressorPost-compressor temperature temperature (◦ (°C)C) 200 200 Appl. Sci. 2017, 7, 350 6 of 25 Appl. Sci. 2017, 7, 350 6 of 25

ModelModel ValidationValidation AA high fidelity ﬁdelity engine engine model model validated validated against against ex experimentalperimental data data for for a conventional a conventional layout, layout, in inboth both steady-state steady-state and and transients, transients, has has been been applied applied in in this this study. study. The The calibration calibration of thethe combustioncombustion modelmodel (SI(SI Wiebe) Wiebe) was was performed performed by isolatingby isolating Cylinder Cylinder 1 of the1 of engine the engine and performing and performing a Three Pressurea Three AnalysisPressure (TPA).Analysis Figure (TPA).4 represents Figure 4 represents the comparison the comparison between measurements between meas andurements simulation and resultssimulation for allresults engine for cylindersall engine at cylinders low loads. at low loads.

(a) (b)

Figure 4. Simulated cylinder pressure from Three Pressure Analysis (TPA) and measured pressure at Figure 4. Simulated cylinder pressure from Three Pressure Analysis (TPA) and measured pressure at low loads. (a) Cylinder 1; (b) Cylinder 2; (c) Cylinder 3; (d) Cylinder 4. low loads. (a) Cylinder 1; (b) Cylinder 2; (c) Cylinder 3; (d) Cylinder 4. The small cylinder-to-cylinder variations observed in the measured data were difficult to captureThe in small the 1D cylinder-to-cylinder simulation. These variations variations observed may be ina result the measured of several data factors, were such difficult as 3D to captureair and ingas the flow 1D behaviours simulation. at These the intake variations and mayexhaust be aor result different of several thermal factors, conditions. such as Although 3D air and there gas were flow behavioursslight variations, at the a intake good andagreement exhaust between or different measured thermal and conditions. simulated mass Although air flow there and were averaged slight variations,indicated amean good effective agreement pressure between (IMEP) measured (error and of simulated less than mass 2%) air was flow observed and averaged for indicatedfull load meanconditions, effective as shown pressure in(IMEP) Table 2. (error of less than 2%) was observed for full load conditions, as shown in Table2. Table 2. Measured data and simulation results at full loads. IMEP, indicated mean effective pressure. Table 2. Measured data and simulation results at full loads. IMEP, indicated mean effective pressure. Attribute Value Cylinder 1 Cylinder 2 Cylinder 3 Cylinder 4 Measured totalAttribute mass Value air flow (kg/h) Cylinder 1 Cylinder 2 258.69 Cylinder 3 Cylinder 4 MeasuredSimulated total total mass mass air flow air (kg/h)flow (kg/h) 258.69 263.65 SimulatedError in totaltotal mass mass air air flow flow (kg/h) (%) 263.65 1.92 ErrorMeasured in total massnet IMEP air flow (bar) (%) 20.71 21.19 1.92 20.75 18.29 MeasuredSimulated net net IMEP IMEP (bar) (bar) 20.71 20.44 21.19 20.76 20.75 20.30 18.02 18.29 Simulated net IMEP (bar) 20.44 20.76 20.30 18.02 AbsoluteAbsolute error error in net in IMEPnet IMEP prediction prediction (bar) (bar) 0.39 0.39 0.43 0.43 0.45 0.45 0.27 0.27 ErrorError in netin net IMEP IMEP prediction prediction (%) (%) 1.87 1.87 2.05 2.05 2.15 2.15 1.47 1.47 AveragedAveraged absolute absolute error error in net in IMEP net IMEP (bar) (bar) 0.36 0.36 AveragedAveraged error error in net in IMEPnet IMEP (%) (%) 1.89 1.89

ForFor part-loadpart-load cases, cases, the the discrepancy discrepancy between between measured measured and simulatedand simulated results results was slightly was slightly larger, aslarger, can beas foundcan be in found Table 3in. However,Table 3. However, despite the despite 9.37% the error 9.37% in the error averaged in the IMEPaveraged for theIMEP low-load for the case,low-load the averagedcase, the averaged absolute absolute error in neterror IMEP in net was IMEP below was 0.3 below bar. 0.3 Therefore, bar. Therefore, the modelling the modelling of the scavengingof the scavenging system system and combustion and combustion system sy atstem low at engine low engine load is load satisfactory. is satisfactory.

Appl.Appl. Sci. 2017,, 7,, 350350 77 ofof 2525 Appl. Sci. 2017, 7, 350 7 of 25 Table 3. Summary of combustion model validation at 2000 rpm. Table 3. Summary of combustion model validation at 2000 rpm. Table 3. Summary of combustion model validation at 2000 rpm. Attribute Value Low-Load Medium-Load Full-Load Error in total mass airAttributeAttribute flow prediction Value Value (%) Low-Load Low-Load 0.69 Medium-Load Medium-Load 3.26 Full-Load Full-Load 1.92 ErrorAveragedError in total error mass mass in air netair flow flowIMEP prediction prediction prediction (%) (%) (bar) (Cylinder 1) 0.690.69 0.19 3.263.26 0.39 1.921.92 0.39 AveragedAveraged errorerror in in net net IMEP IMEP prediction prediction (bar) (bar)(%) (Cylinder (Cylinder (Cylinder 1) 1) 1) 0.19 0.197.39 0.39 0.392.56 0.39 0.391.87 Averaged error in net IMEP prediction (%) (Cylinder 1) 7.39 2.56 1.87 Averaged error in net IMEP prediction (%) (Cylinder 1) 7.39 2.56 1.87 The overall shape of the port pressures at three operating points was captured. Figure 5 outlines the measuredThe overall and shape simulated of the port instantaneous pressures atpressure three operating at the exhaust points wasport captured.of Cylinder Figure 1 for5 5 outlines mediumoutlines theengine measured loads. As and can simulated be seen instantaneousfrom the figure, pressure the magnitude at the exhaust of the portpressure of Cylinder wave reflections 1 for medium was enginewell predicted. loads. As As can can be seen from the figure, figure, thethe magnitude of the pressure wave reflectionsreflections was well predicted.

Figure 5. Instantaneous pressure at exhaust port of Cylinder 1, 2000 rpm medium-load. Figure 5. InstantaneousInstantaneous pressure at exhaust port of Cylinder 1, 2000 rpm medium-load.medium-load. The validation of the transient performance of the model was also performed against The validation of the transient performance of the model was also performed against experimentalThe validation data ofcollected the transient from performancetransient tests of theperformed model was with also various performed wastegate against positions. experimental The experimental data collected from transient tests performed with various wastegate positions. The datacalibration collected was from performed transient for tests three performed different boost with variouspressures wastegate and the comparison positions. The between calibration measured was calibration was performed for three different boost pressures and the comparison between measured performedand simulated for threemass differentair flow is boost presented pressures in Figure and the 6. comparisonThe experimental between and measured simulation and results simulated show and simulated mass air flow is presented in Figure 6. The experimental and simulation results show massa good air agreement ﬂow is presented with small in Figure variations6. The (5%) experimental that could and be simulationa result of the results scavenging show a goodmodel agreement in the 1D a good agreement with small variations (5%) that could be a result of the scavenging model in the 1D withengine small simulation. variations (5%) that could be a result of the scavenging model in the 1D engine simulation. engine simulation.

Figure 6.6. Comparison between measurement and simulationsimulation results of transienttransient performanceperformance for differentFigure 6. boost boostComparison targetstargets with withbetween imposedimpose measurementd turbochargerturbocharger and rotationalro simulatationaltion speed.speed. results of transient performance for different boost targets with imposed turbocharger rotational speed. 3. Methodology 3. Methodology The simulations performed in this paper are divided into two main subsections of steady-state and transientThe simulations analysis. performed Each of the in subsectionsthis paper are is furtherdivided divided into two into main smaller subsections segments of steady-stateto represent and transient analysis. Each of the subsections is further divided into smaller segments to represent Appl. Sci. 2017, 7, 350 8 of 25

3. Methodology

Appl.Appl. Sci.TheSci. 20172017 simulations,, 77,, 350350 performed in this paper are divided into two main subsections of steady-state88 ofof 2525 and transient analysis. Each of the subsections is further divided into smaller segments to represent differentdifferent typestypes ofof studies.studies. A A detailed detailed flowchart ﬂowchartflowchart of of all all ofof thethe simulationssimulations performedperformed in in thisthis paperpaper isis shownshown inin FigureFigure7 7.7..

FigureFigure 7.7. FlowchartFlowchart ofof thethe simulationsimulation process.process. WG,WG, wastegate.wastegate.

Finally,Finally, it itit needs needsneeds to toto be bebe highlighted highlightedhighlighted that thatthat all of allall the ofof studies thethe studiesstudies presented presentedpresented in this paperinin thisthis show paperpaper the showshow potential thethe availablepotentialpotential energyavailableavailable in theenergyenergy system. inin thethe Any system.system. electrical AnyAny losses electrelectr suchicalical aslosseslosses alternator, suchsuch asas converter alternator,alternator, and converterconverter battery losses andand havebatterybattery not losseslosses been have consideredhave notnot beenbeen at thisconsideredconsidered point. atat thisthis point.point.

3.1.3.1. Steady-State Steady-State Simulations Simulations TheThe steadysteady statestate analysisanalysis ofof thethe modelmodel waswas performedperfperformedormed in in the the area area ofof threethree axes,axes, compressorcompressor andand turbineturbine sizesizesize and andand WG WGWG area, area,area, as asas shown shownshown in Figureinin FigureFigure8. The 8.8. The purposeThe purposepurpose of the ofof study thethe studystudy was to waswas understand toto understandunderstand the effects thethe ofeffectseffects the compressor ofof thethe compressorcompressor and turbine andand sizing turbineturbine on the sizingsizing amount onon of the energythe amountamount that needs ofof energyenergy to be provided/harvested thatthat needsneeds toto bebe byprovided/harvestedprovided/harvested the motor/generator byby thethe and motor/generatormotor/generator to identify any andand regions toto identifyidentify where anyany the e-turboregionsregions wherecanwhere replace thethe e-turboe-turbo the wastegate cancan replacereplace for loadthethe wastegatewastegate control. forfor loadload control.control.

FigureFigure 8.8. AreaArea ofof investigationinvestigation forfor thethe steady-statesteady-state simulations.simulations. Figure 8. Area of investigation for the steady-state simulations. TheThe steady-statesteady-state simulationssimulations areare divideddivided intointo threthreee phasesphases basedbased onon thethe typetype ofof investigation.investigation. •• ThePhasePhase steady-state 1:1: AssessmentAssessment simulations ofof thethe amountamount are divided ofof enthalpyenthalpy into three lolossss phasesforfor thethe baseline basedbaseline on engineengine the type acrossacross of investigation. thethe speed/loadspeed/load mapmap andand howhow thisthis isis affectedaffected byby thethe compressorcompressor andand turbine’sturbine’s size.size. •• PhasePhase 2:2: InvestigationInvestigation ofof thethe potentialpotential ofof suppressuppressingsing WGWG andand usingusing e-turboe-turbo toto controlcontrol boosting.boosting. •• PhasePhase 3:3: ReinstatementReinstatement ofof thethe WGWG toto controlcontrol exhaustexhaust manifoldmanifold pressurepressure andand exploreexplore thethe trade-offtrade-off betweenbetween WGWG andand turbineturbine size.size. Appl. Sci. 2017, 7, 350 9 of 25

• Phase 1: Assessment of the amount of enthalpy loss for the baseline engine across the speed/load map and how this is affected by the compressor and turbine’s size. • Phase 2: Investigation of the potential of suppressing WG and using e-turbo to control boosting. • Phase 3: Reinstatement of the WG to control exhaust manifold pressure and explore the trade-off between WG and turbine size.

3.1.1. Phase 1: WG Enthalpy Loss Study The first phase of the study was performed to calculate the amount of energy that is lost through the wastegate of a modern 2.0 L turbocharged gasoline engine. The amount of power loss was calculated using the equation: Q = m × cp × ∆T, (2) where Q is the amount of heat to the system (power), m is the mass flow rate through the wastegate, cp is the specific heat of the gas and ∆T is the temperature difference. A Design of Experiment (DoE) analysis was performed to evaluate the effects of the compressor and turbine’s size on the amount of enthalpy loss, combustion limiting parameters and maximum engine power. The nine cases studied are described in Table4.

Table 4. Phase 1: Design of Experiment (DoE) analysis for different compressor and turbine sizes.

Case Turbine Size Multiplier Compressor Size Multiplier 1 0.7 0.7 2 0.7 0.85 3 0.7 1 4 0.85 0.7 5 0.85 0.85 6 0.85 1 7 1 0.7 8 1 0.85 9 1 1

3.1.2. Phase 2: Suppressing WG and Using e-Turbo to Control Boosting Phase 2 of the steady-state simulations section involves the application of a motor/generator, which is directly linked to the shaft of the turbocharger. The purpose of the motor/generator is to provide or harvest energy, as needed, to/by the compressor for achieving the boosting demands of the engine. For this phase, the WG has been completely suppressed and the model run once with the original size of the compressor and turbine. However due, to the shut off of the WG valve, extremely high pre-turbine pressures violating the limits at high loads were noticed. A sweep study was performed for ﬁve different turbine sizes (10% to 50% larger) to evaluate the optimum turbine size for the model without a WG valve. Then, the optimum turbine was tested for different compressor sizes (20% smaller to 20% larger) to evaluate its effect on the energy harvesting/provision requirements.

3.1.3. Phase 3: Reinstating WG to Control Exhaust Manifold Pressure Phase 3 of the steady-state simulations involves the reinstatement of the WG valve for reducing the required size of the turbine and enhancing the energy performance across the low speeds area of the engine’s speed/load map. This section is divided into four studies, as shown below: • Turbine size: WG area balance study • Smaller turbine: increased WG area study • Smaller turbine: increased pre-turbine pressure study (smaller WG area) • Smaller turbine: variable pre-turbine pressure study (smaller and larger WG area) Appl. Sci. 2017, 7, 350 10 of 25

3.2. Transient Simulations The findings in the steady-state simulations section (see the Results Section) demonstrated that a model fitted with a motor-generator and a turbine 10% smaller than the original could provide energy harvesting and thermal efficiency gain across the speed/load map of the engine. The transient behaviour of the model with this configuration, an original size compressor and increased pre-turbine pressures was tested by performing three different types of simulations, as shown below:

• Load step transient • Fixed gear vehicle speed transient • Energy balance study for various driving cycles and real driving conditions

The purpose of the transient study was to reveal the response time improvement under different levels of energy provision to the compressor, as well as to perform an energy balance review. This is required in order to identify whether the e-turbocharger can operate without energy consumption from the engine and if any excess on the total energy generated.

3.2.1. Load Step Transient Simulations The load tip-in study was performed for two engine speeds of interest, 1600 and 1100 rpm, close to the surge limit of the compressor. The load tip-in step requested was from 2 bar BMEP to full load BMEP (19 bar). However, due to the engine speeds at which the tests were performed, the requested BMEP could not be achieved. The study was initially performed for the model with all of the PID controllers. However, for ensuring that the results are not affected by the tuning of the controllers, the study was repeated in an open-loop environment (no PID controllers). The PID controllers in the model were removed and replaced by look-up tables for controlling the boost pressure and throttle position of the engine, as shown in Table5. This allows a hardware capability investigation to be performed by eliminating any effects of the PID controllers. The energy provided to the compressor was divided into four sections of 0.25 seconds, and different levels of power (1, 3 and 5 kW) were provided in each section.

Table 5. Control of basic process parameters for the model with proportional-integral-derivative (PID) controllers and the open-loop model. BMEP, brake mean effective pressure; M/G, motor/generator.

Process Parameter Model with PIDs (Control Variables) Open-Loop Model M/G Power - Provided power proﬁles Boost pressure M/G power Speed/load look-up table Throttle position Engine speed/BMEP Speed/load look-up table Pre-turbine pressure WG position WG position (PID)

3.2.2. Fixed Gear Vehicle Speed Transient Simulations The purpose of this study was to represent the vehicle’s driving conditions and test the model for a ﬁxed gear (third) vehicle speed transient. For this reason, a driving resistance load of 30 Nm was applied to represent the rolling resistance along with the ﬂywheel’s inertia given by the following formula: Ieq = Ieq,eng + Ieq,prop + Ieq,axle + Ieq,veh, (3) 2 where Ieq is the total inertia (kg·m ), Ieq,eng is the engine’s inertia, Ieq,prop is the prop shaft inertia, Ieq,axle is the axle inertia and Ieq,veh is the vehicle mass equivalent inertia, as shown in Figure9. Appl. Sci. 2017, 7, 350 10 of 25

• Load step transient • Fixed gear vehicle speed transient • Energy balance study for various driving cycles and real driving conditions The purpose of the transient study was to reveal the response time improvement under different levels of energy provision to the compressor, as well as to perform an energy balance review. This is required in order to identify whether the e-turbocharger can operate without energy consumption from the engine and if any excess on the total energy generated.

3.2.1. Load Step Transient Simulations The load tip-in study was performed for two engine speeds of interest, 1600 and 1100 rpm, close to the surge limit of the compressor. The load tip-in step requested was from 2 bar BMEP to full load BMEP (19 bar). However, due to the engine speeds at which the tests were performed, the requested BMEP could not be achieved. The study was initially performed for the model with all of the PID controllers. However, for ensuring that the results are not affected by the tuning of the controllers, the study was repeated in an open-loop environment (no PID controllers). The PID controllers in the model were removed and replaced by look-up tables for controlling the boost pressure and throttle position of the engine, as shown in Table 5. This allows a hardware capability investigation to be performed by eliminating any effects of the PID controllers. The energy provided to the compressor was divided into four sections of 0.25 seconds, and different levels of power (1, 3 and 5 kW) were provided in each section.

Table 5. Control of basic process parameters for the model with proportional-integral-derivative (PID) controllers and the open-loop model. BMEP, brake mean effective pressure; M/G, motor/generator.

Process Parameter Model with PIDs (Control Variables) Open-Loop Model M/G Power - Provided power profiles Boost pressure M/G power Speed/load look-up table Throttle position Engine speed/BMEP Speed/load look-up table Pre-turbine pressure WG position WG position (PID)

3.2.2. Fixed Gear Vehicle Speed Transient Simulations The purpose of this study was to represent the vehicle’s driving conditions and test the model for a fixed gear (third) vehicle speed transient. For this reason, a driving resistance load of 30 Nm was applied to represent the rolling resistance along with the flywheel’s inertia given by the following formula:

Ieq = Ieq,eng + Ieq,prop + Ieq,axle + Ieq,veh, (3)

Appl.where Sci. 2017 Ieq, 7 is, 350 the total inertia (kg·m2), Ieq,eng is the engine’s inertia, Ieq,prop is the prop shaft inertia, Ieq,axle11 of 25 is the axle inertia and Ieq,veh is the vehicle mass equivalent inertia, as shown in Figure 9.

Figure 9. Schematic of total inertia applied to the engine. Appl. Sci. 2017, 7, 350 Figure 9. Schematic of total inertia applied to the engine. 11 of 25

The study was performed for for four four different different levels levels of of targeted targeted BMEP, BMEP, as shown in in Figure Figure 10,10, representing the aggressiveness with which a vehicle’svehicle’s pedal can be pressed. The tests were repeated twice for different pre-turbine pressures, which were controlled by implementingimplementing a pre-turbine pressure/WG PID PID controller. controller.

Figure 10. Fixed gear vehicle speed BMEP tip-ins.

3.2.3. Energy Balance Simulations The studies performed in the previous sections highlight the signiﬁcantsignificant improvement on the transient response time of the engine during load and speed tip-in conditions conditions (see (see Results Results Section). Section). The work performed in this section investigates whether the power harvested by the generator is enough to cover the motor demands. The energy balance study for a specific speciﬁc vehicle was conducted for thethe followingfollowing driving driving cycles cycles with with the the pre-turbine pre-turb pressureine pressure target target set at 15%set at higher 15% thanhigher the than baseline the baselineengine for engine the low for tothe medium low to medium loads and loads 5% for and the 5% full for load the full conditions: load conditions:

1. NewNew European DrivingDriving Cycle Cycle (NEDC): (NEDC): designed designed to representto represent the typicalthe typica usagel usage of a car of in a Europe, car in Europe,but often but criticized often criticized for delivering for deli unrealisticvering unrealistic economy economy ﬁgures. figures. 2. Worldwide harmonized Light vehicles Test Cycle (WLTC): a harmonized driving cycle 2. Worldwide harmonized Light vehicles Test Cycle (WLTC): a harmonized driving cycle representing realistic driving conditions data in different regions around the world, combined representing realistic driving conditions data in different regions around the world, combined with suitable weighting factors. with suitable weighting factors. 3. US06: representing aggressive, high-speed and/or high acceleration driving behaviour, rapid 3. US06: representing aggressive, high-speed and/or high acceleration driving behaviour, rapid speed fluctuations and driving behaviour following startup. speed ﬂuctuations and driving behaviour following startup. 4. Combined driving conditions: including equal amounts of various realistic driving conditions, 4. Combined driving conditions: including equal amounts of various realistic driving conditions, such as low speed, start-stop, highway, motorway, uphill, downhill and dangerous such as low speed, start-stop, highway, motorway, uphill, downhill and dangerous overtaking conditions. overtaking conditions. The engine speed and BMEP demands for the four driving cycles are shown in Figure 11. The engine speed and BMEP demands for the four driving cycles are shown in Figure 11.

Figure 11. Driving cycles conditions: (a) engine speed; (b) engine BMEP. NEDC, New European Driving Cycle; WLTC, Worldwide harmonized Light vehicles Test Cycle. Appl. Sci. 2017, 7, 350 11 of 25

The study was performed for four different levels of targeted BMEP, as shown in Figure 10, representing the aggressiveness with which a vehicle’s pedal can be pressed. The tests were repeated twice for different pre-turbine pressures, which were controlled by implementing a pre-turbine pressure/WG PID controller.

Figure 10. Fixed gear vehicle speed BMEP tip-ins.

3.2.3. Energy Balance Simulations The studies performed in the previous sections highlight the significant improvement on the transient response time of the engine during load and speed tip-in conditions (see Results Section). The work performed in this section investigates whether the power harvested by the generator is enough to cover the motor demands. The energy balance study for a specific vehicle was conducted for the following driving cycles with the pre-turbine pressure target set at 15% higher than the baseline engine for the low to medium loads and 5% for the full load conditions: 1. New European Driving Cycle (NEDC): designed to represent the typical usage of a car in Europe, but often criticized for delivering unrealistic economy figures. 2. Worldwide harmonized Light vehicles Test Cycle (WLTC): a harmonized driving cycle representing realistic driving conditions data in different regions around the world, combined with suitable weighting factors. 3. US06: representing aggressive, high-speed and/or high acceleration driving behaviour, rapid speed fluctuations and driving behaviour following startup. 4. Combined driving conditions: including equal amounts of various realistic driving conditions, such as low speed, start-stop, highway, motorway, uphill, downhill and dangerous overtaking conditions. Appl. Sci. 2017, 7, 350 12 of 25 The engine speed and BMEP demands for the four driving cycles are shown in Figure 11.

Figure 11.FigureDriving 11. Driving cycles cycles conditions: conditions: (a ()a) engine engine speed; speed; (b) ( bengine) engine BMEP. BMEP. NEDC, NEDC, New European New European Driving Cycle; WLTC, Worldwide harmonized Light vehicles Test Cycle. Driving Cycle; WLTC, Worldwide harmonized Light vehicles Test Cycle.

4. Results and Discussion Appl. Sci. 2017, 7, 350 12 of 25 The results of the simulation studies performed in this paper are categorized in a similar manner 4. Results and Discussion to that of the Methodology Section. The results of the simulation studies performed in this paper are categorized in a similar manner 4.1. Steady-Stateto that of Analysis the Methodology Section.

4.1. Steady-State Analysis 4.1.1. Phase 1: WG Enthalpy Loss Study (Baseline Engine) The results4.1.1. Phase in Figure 1: WG Enthalpy 12 show Loss that Study the (Baseline energy Engine) loss through the wastegate for the model with the original size ofThe compressor results in Figure and 12 turbine show that (Case the energy 9) can loss reach through the the amount wastegate of for 5 kW the model at high with engine the speeds and loads.original The amount size of compressor of the potential and turbine energy (Case dismissed 9) can reach atthe medium amount of load 5 kW and at high engine engine speed speeds conditions and loads. The amount of the potential energy dismissed at medium load and engine speed is of lowerconditions magnitude, is of lower up to magnitude, 2 kW. At up low to 2 loads kW. At and low loads speeds, and speeds, there is there no is wasted no wasted energy, energy, as as the WG is completelythe closed. WG is completely closed.

Figure 12.FigurePhase 12. 1: Phase DoE 1: analysis DoE analysis ofthe of the waste-gated waste-gated en enthalpythalpy loss loss for the for different the different compressor compressor and and turbine sizes; the blue line represents the maximum torque line of the baseline engine’s model as turbine sizes; the blue line represents the maximum torque line of the baseline engine’s model as provided by the manufacturer (λ < 1); the ○ symbols indicate data points; the ⨂ symbol indicates a N provided bylimit the violation manufacturer at specific points. (λ <1); the symbols indicate data points; the symbol indicates a limit violation at speciﬁc points. # Figure 12 highlights the effect of a smaller turbine and compressor on the amount of waste-gated flow and the maximum power of the engine. As can be seen, a smaller compressor (Cases 7 and 8) cannot provide the boosting requirements needed to achieve maximum engine’s power at medium and high engine speeds. On the other hand, a smaller turbine (Cases 3 and 6) can slightly increase the amount of waste-gated flow at medium engine speed and load conditions. However, a smaller turbine leads to increased pre-turbine pressure, in some cases violating the maximum limit set, and reduces the maximum torque of the engine. Appl. Sci. 2017, 7, 350 13 of 25

Figure 12 highlights the effect of a smaller turbine and compressor on the amount of waste-gated ﬂow and the maximum power of the engine. As can be seen, a smaller compressor (Cases 7 and 8) cannot provide the boosting requirements needed to achieve maximum engine’s power at medium and high engine speeds. On the other hand, a smaller turbine (Cases 3 and 6) can slightly increase the amount of waste-gated ﬂow at medium engine speed and load conditions. However, a smaller turbine leads to increased pre-turbine pressure, in some cases violating the maximum limit set, and reduces theAppl. maximum Sci. 2017, 7, 350 torque of the engine. 13 of 25

4.1.2. Phase 2: Suppressing WG and Using e-Turbo toto ControlControl BoostingBoosting Suppressing the wastegate of an engine leads to extremely high pre-turbine pressures at medium to high loads.loads. Figure Figure 1313 presents the pre-turbin pre-turbinee pressures and peak engine torques achieved for various turbine sizes and the wastegate valve closed.closed. As shown from the figure, figure, a turbine 30% larger than the original size can provide or even increase the maximum power targets of the baseline engine without violating thethe pre-turbine pressurepressure limit.limit. AA torquetorque deficitdeficit occursoccurs atat 1000 rpm, but this is mainly due toto aa poor poor model model convergence convergence at this at specificthis specific engine engine speed speed near the near surge the line surge and notline the and incapacity not the ofincapacity the engine of the to achieve engine theto achieve targeted the torque. targeted torque.

Figure 13. Engine’s performance results for various turb turbinesines sizes and the wastegate completely completely shut: shut: (a) pre-turbinepre-turbine pressure;pressure; ((bb)) engineengine torque.torque.

The results in Figure 14 show that with this configuration, extremely high amounts of energy, The results in Figure 14 show that with this conﬁguration, extremely high amounts of energy, more than 15 kW, can be harvested at high engine speeds and loads. The system can harvest energy more than 15 kW, can be harvested at high engine speeds and loads. The system can harvest energy up to 10 kW at medium to high engine speeds. However, as can be seen, the larger turbine results in up to 10 kW at medium to high engine speeds. However, as can be seen, the larger turbine results considerable amounts of energy that need to be provided by the motor for meeting, if possible, the in considerable amounts of energy that need to be provided by the motor for meeting, if possible, targeted maximum torque at low engine speed conditions. This is happening due to the lower the targeted maximum torque at low engine speed conditions. This is happening due to the lower pre-turbine pressures and increased turbocharger’s inertia at a non-boosting area of the engine’s map. pre-turbine pressures and increased turbocharger’s inertia at a non-boosting area of the engine’s map. The next stage in this phase was to investigate the potential of reducing the amount of energy needs to be provided by changing the compressor’s size. For this reason, ﬁve different compressors (multipliers of 0.8 to 1.2) were tested, and the results are presented in Figure 15. It is clearly shown in Figure 15 that a small compressor could reduce the power level needs to be provided by the motor from 0.6 down to 0.2 kW at 1600 rpm with no effect on the engine’s maximum torque. However, a smaller compressor would also reduce the amount of energy that can be harvested (from 18.5 down to 17 kW) at high engine speeds; but foremost, it would struggle to meet the maximum engine’s torque demands.

Figure 14. Phase 2: Motor-generator average energy for the case with original compressor, 30% larger turbine and the WG valve completely shut; negative values indicate energy harvesting; positive values indicate the need for energy provision; the blue line represents the maximum torque line of the baseline engine’s model as provided by the manufacturer (λ < 1); the ⨂ symbol indicates limit violation; ⌼ indicates energy provision area.

The next stage in this phase was to investigate the potential of reducing the amount of energy needs to be provided by changing the compressor’s size. For this reason, five different compressors (multipliers of 0.8 to 1.2) were tested, and the results are presented in Figure 15. Appl. Sci. 2017, 7, 350 13 of 25

4.1.2. Phase 2: Suppressing WG and Using e-Turbo to Control Boosting Suppressing the wastegate of an engine leads to extremely high pre-turbine pressures at medium to high loads. Figure 13 presents the pre-turbine pressures and peak engine torquesAppl. Sci.achieved 2017, 7, 350 for 13 of 25 various turbine sizes and the wastegate valve closed. As shown from the figure, a turbine4.1.2. Phase 30% larger2: Suppressing WG and Using e-Turbo to Control Boosting than the original size can provide or even increase the maximum power targets of the baseline engine Suppressing the wastegate of an engine leads to extremely high pre-turbine pressures at medium without violating the pre-turbine pressure limit. A torque deficit occurs at 1000 rpm, but this is mainly to high loads. Figure 13 presents the pre-turbine pressures and peak engine torques achieved for due to a poor model convergence at this specific engine speed near the surge linevarious and turbine not the sizes and the wastegate valve closed. As shown from the figure, a turbine 30% larger incapacity of the engine to achieve the targeted torque. than the original size can provide or even increase the maximum power targets of the baseline engine without violating the pre-turbine pressure limit. A torque deficit occurs at 1000 rpm, but this is mainly due to a poor model convergence at this specific engine speed near the surge line and not the incapacity of the engine to achieve the targeted torque.

Figure 13. Engine’s performance results for various turbines sizes and the wastegate completely shut: (a) pre-turbine pressure; (b) engine torque. Figure 13. Engine’s performance results for various turbines sizes and the wastegate completely shut: (a) pre-turbine pressure; (b) engine torque. The results in Figure 14 show that with this configuration, extremely high amounts of energy, more than 15 kW, can be harvested at high engine speeds and loads. The system can harvestThe results energy in Figure 14 show that with this configuration, extremely high amounts of energy, up to 10 kW at medium to high engine speeds. However, as can be seen, the larger turbinemore than results 15 kW, in can be harvested at high engine speeds and loads. The system can harvest energy considerable amounts of energy that need to be provided by the motor for meeting,up if to possible, 10 kW at themedium to high engine speeds. However, as can be seen, the larger turbine results in targeted maximum torque at low engine speed conditions. This is happening dueconsiderable to the lower amounts of energy that need to be provided by the motor for meeting, if possible, the Appl.pre-turbine Sci. 2017, 7, pressures 350 and increased turbocharger’s inertia at a non-boosting area of thetargeted engine’s maximum map.14 of 25torque at low engine speed conditions. This is happening due to the lower pre-turbine pressures and increased turbocharger’s inertia at a non-boosting area of the engine’s map.

Figure 14. Phase 2: Motor-generator average energy for the case with original compressor, 30% larger FigureFigure 14. 14.Phase Phase 2: 2: Motor-generator Motor-generator averageaverage energy for for th thee case case with with original original compressor, compressor, 30%turbine 30% larger larger and the WG valve completely shut; negative values indicate energy harvesting; positive turbineturbine and and the the WG WG valve valve completely completely shut; shut; negative negati valuesve values indicate indicate energy energy harvesting; harvesting; positive valuespositive values indicate the need for energy provision; the blue line represents the maximum torque line of indicatevalues the indicate need forthe energyneed for provision; energy provision; the blue line the representsblue line represents the maximum the maximum torque line torque of thethe line baselinebaseline of engine’s model as provided by the manufacturer (λ < 1); the ⨂ symbol indicates limit N violation; ⌼ indicates energy provision area. engine’sthe baseline model engine’s as provided model by as theprovided manufacturer by the manufacturer (λ < 1); the (λ symbol< 1); the indicates ⨂ symbol limit indicates violation; limit indicates energy⌼ provision area. Appl. Sci.violation; 2017, 7, 350 indicates energy provision area. The next14 stageof 25 in this phase was to investigate the potential of reducing the amount of energy needs to be provided by changing the compressor’s size. For this reason, five different compressors The next stage in this phase was to investigate the potential of reducing the amount(multipliers of energy of 0.8 to 1.2) were tested, and the results are presented in Figure 15. needs to be provided by changing the compressor’s size. For this reason, five different compressors (multipliers of 0.8 to 1.2) were tested, and the results are presented in Figure 15.

FigureFigure 15. 15.Effect Effect of of compressor’s compressor’ssize size onon averageaverage power and and torque: torque: (a (a) )1600 1600 rpm; rpm; (b ()b 5800) 5800 rpm. rpm.

It is clearly shown in Figure 15 that a small compressor could reduce the power level needs to 4.1.3. Phase 3: Reinstating WG to Control Exhaust Manifold Pressure be provided by the motor from 0.6 down to 0.2 kW at 1600 rpm with no effect on the engine’s Turbinemaximum Size: torque. WG Area However, Balance a Studysmaller compressor would also reduce the amount of energy that can be harvested (from 18.5 down to 17 kW) at high engine speeds; but foremost, it would struggle to meetThe the ﬁrst maximum part of thisengine’s phase torque includes demands. a DoE study for investigating the beneﬁts on the engine’s maximum power and energy recovery for a smaller turbine (multipliers less than 1.3) and the WG valve4.1.3. open Phase at different3: Reinstating positions WG to (smaller Control than Exhaust the originalManifold model), Pressure as shown in Table6.

Turbine Size: WGTable Area 6. BalancePhase 3: Study DoE analysis for different turbine sizes and WG areas. The first part of this phase includes a DoE study for investigating the benefits on the engine’s Turbine Size Compared to the WG Area Compared to the Case maximum power and energy recovery forOriginal a smaller Model turbine (multipliersOriginal less Model than 1.3) and the WG valve open at differentTurb×1.1–WG/2 positions (smaller than 10% larger the original model), as shown 50% smaller in Table 6. Turb×1.1–WG/3 10% larger 67% smaller TurbTable×1.2–WG/2 6. Phase 3: DoE analysis 20% for larger different turbine sizes and 50% WG smaller areas. Turb×1.2–WG/3 20% larger 67% smaller Case Turbine Size Compared to the Original Model WG Area Compared to the Original Model Turb×1.1–WG/2 10% larger 50% smaller TheTurb results×1.1–WG/3 in Figure 16 show that10% withlarger a 20% larger turbine (rather than 67% smaller 30%) and the WG valve open atTurb half×1.2–WG/2 the size as the baseline20% engine, larger maximum engine torque can 50% be smaller achieved. The smaller turbineTurb and×1.2–WG/3 the open WG reduced the20% larger maximum amount of energy harvested 67% smaller at high speeds from 18 down to 12 kW. However, the beneﬁt on the low-speed side of the map was relatively low, as the amountThe of energyresults in that Figure needs 16 to show be provided that with bya 20% the larger motor turbine at 1600 (rather rpm went than from 30%) 0.6and down the WG to 0.45valve kW. open at half the size as the baseline engine, maximum engine torque can be achieved. The smaller turbine and the open WG reduced the maximum amount of energy harvested at high speeds from 18 down to 12 kW. However, the benefit on the low-speed side of the map was relatively low, as the amount of energy that needs to be provided by the motor at 1600 rpm went from 0.6 down to 0.45 kW.

Figure 16. Balance study for different turbine sizes and wastegate areas for meeting the baseline engine’s maximum torque. Appl. Sci. 2017, 7, 350 14 of 25

Figure 15. Effect of compressor’s size on average power and torque: (a) 1600 rpm; (b) 5800 rpm.

It is clearly shown in Figure 15 that a small compressor could reduce the power level needs to be provided by the motor from 0.6 down to 0.2 kW at 1600 rpm with no effect on the engine’s maximum torque. However, a smaller compressor would also reduce the amount of energy that can be harvested (from 18.5 down to 17 kW) at high engine speeds; but foremost, it would struggle to meet the maximum engine’s torque demands.

4.1.3. Phase 3: Reinstating WG to Control Exhaust Manifold Pressure

Turbine Size: WG Area Balance Study The first part of this phase includes a DoE study for investigating the benefits on the engine’s maximum power and energy recovery for a smaller turbine (multipliers less than 1.3) and the WG valve open at different positions (smaller than the original model), as shown in Table 6.

Table 6. Phase 3: DoE analysis for different turbine sizes and WG areas.

Case Turbine Size Compared to the Original Model WG Area Compared to the Original Model Turb×1.1–WG/2 10% larger 50% smaller Turb×1.1–WG/3 10% larger 67% smaller Turb×1.2–WG/2 20% larger 50% smaller Turb×1.2–WG/3 20% larger 67% smaller

The results in Figure 16 show that with a 20% larger turbine (rather than 30%) and the WG valve open at half the size as the baseline engine, maximum engine torque can be achieved. The smaller turbine and the open WG reduced the maximum amount of energy harvested at high speeds from 18 down to 12 kW. However, the benefit on the low-speed side of the map was relatively low, as the Appl. Sci. 2017amount, 7, 350 of energy that needs to be provided by the motor at 1600 rpm went from 0.6 down to 15 of 25 0.45 kW.

Appl. Sci. 2017, 7, 350 13 of 25 Figure 16. Balance study for different turbine sizes and wastegate areas for meeting the baseline Figure 16.engine’sBalance maximum study torque. for different turbine sizes and wastegate areas for meeting the baseline engine’s maximum torque. 4.1.2. Phase 2: Suppressing WG and Using e-Turbo to Control Boosting Suppressing the wastegate of an engine leads to extremely high pre-turbine pressures at medium Smaller Turbine: Increased WG Area Study to high loads. Figure 13 presents the pre-turbine pressures and peak engine torques achieved for various turbine sizes and the wastegate valve closed. As shown from the figure, a turbine 30% larger

DespiteAppl. the Sci. fact2017, 7, that 350 a large turbine allowsthan energy the original harvesting size can within provide the or even pre-turbine increase15 of 25 the pressure maximum power targets of the baseline engine limit at the high speed and load conditions, it alsowithout leads violating to low the speeds pre-turbine and loads’ pressure poor limit. performance. A torque deficit occurs at 1000 rpm, but this is mainly This could beSmaller theoretically Turbine: Increased resolved WG byArea implementing Study due to aa smallpoor model turbine convergence and controlling at this thespecific pre-turbine engine speed near the surge line and not the incapacity of the engine to achieve the targeted torque. pressure limitsDespite by increasing the fact that the a wastegatelarge turbine area.allows Theenergy following harvesting DoE within study the pre-turbine results show pressure the effects of limit at the high speed and load conditions, it also leads to low speeds and loads’ poor performance. three compressorsThis could smaller be theoretically than theresolved original by implemen (multipliersting a small of0.7 turbine to 0.9) and andcontrolling the WG the pre-turbine area open at values 10% to 50%pressure larger limits (multipliers by increasing of 1.1 the towastegate 1.5) than area. in The the following baseline DoE engine. study results show the effects of The resultsthree compressors in Figure 17 smaller illustrate than the that original a small (multipliers turbine of can 0.7 to lead 0.9) and to lower the WG energy area open demands at values and power generation at10% low to 50% speed/load larger (multipliers conditions. of 1.1 to However,1.5) than in the the baseline smaller engine. the turbine, the higher the pre-turbine The results in Figure 17 illustrate that a small turbine can lead to lower energy demands and pressures, whichpower generation deteriorate at low the speed/load engine’s conditions. power output However, at fullthe smaller load conditions. the turbine, the By higher increasing the the WG area, the pre-turbinepre-turbine pressures, pressure which drops, deteriorate and therefore, the engine the’s fullpower load output performance at full load ofconditions. the engine By increases. However, evenincreasing a 10% the increase WG area, inthe thepre-turbine WG area pressure leads dr toops, high and levelstherefore, of the energy full load requirements performance of for meeting the engine increases. However, even a 10% increase in the WG area leads to high levels of energy the baseline engine’s power characteristics. requirements for meeting the baseline engine’s power characteristics.

Figure 13. Engine’s performance results for various turbines sizes and the wastegate completely shut: (a) pre-turbine pressure; (b) engine torque.

The results in Figure 14 show that with this configuration, extremely high amounts of energy, more than 15 kW, can be harvested at high engine speeds and loads. The system can harvest energy up to 10 kW at medium to high engine speeds. However, as can be seen, the larger turbine results in considerable amounts of energy that need to be provided by the motor for meeting, if possible, the targeted maximum torque at low engine speed conditions. This is happening due to the lower pre-turbine pressures and increased turbocharger’s inertia at a non-boosting area of the engine’s map.

Figure 17. Phase 3: DoE analysis of the motor-generatoFigurer average 14. energyPhase for2: Motor-generator different turbine sizes average and energy for the case with original compressor, 30% larger Figure 17. Phase 3: DoE analysis of the motor-generator average energy for different turbine sizes WG areas; negative values indicate energy harvesting;turbine positive and thevalues WG in dicatevalve needcompletely for energy shut; negative values indicate energy harvesting; positive and WG areas;provision; negative blue line values represents indicate the maximum energy torque harvesting; valuesline of theindicate positive baseline the engine’s valuesneed for model indicate energy as provided provision; need for th energye blue line represents the maximum torque line of provision; blueby the line manufacturer represents (λ < the 1); ⨂ maximum symbol indicates torque limit linethe violation; baseline of the ⌼ baseline indicatesengine’s engine’senergymodel provision as modelprovided area. as by provided the manufacturer by (λ < 1); the ⨂ symbol indicates limit the manufacturer (λ < 1); N symbol indicates limit violation;violation; ⌼ indicatesindicates energy energy provision provision area. area. Smaller Turbine: Increased Pre-Turbine Pressure Study The next study in phase 3 focuses on the effectThes of anext smaller stage turbine in this with phase a reduced was to WGinvestigate area the potential of reducing the amount of energy compared to the baseline engine for benefiting needsfrom a togood be providedenergy balance by changing across the the low compressor’s and high size. For this reason, five different compressors load areas of the engine’s speed/load map. For(multipliers this study, theof 0.8WG to area 1.2) iswere controlled tested, indirectly and the resultsby are presented in Figure 15. setting up a PID controller between the WG valve and the pre-turbine pressure. The targeted value Appl. Sci. 2017, 7, 350 16 of 25

Smaller Turbine: Increased Pre-Turbine Pressure Study

The next study in phase 3 focuses on the effects of a smallerAppl. turbine Sci. 2017 with, 7, 350a reduced WG area 13 of 25 compared to the baseline engine for beneﬁting from a good energy balance across the low and high load areas of the engine’s speed/load map. For this study, the WG4.1.2. area Phase is controlled2: Suppressing indirectly WG and Using by e-Turbo to Control Boosting setting up a PID controller between the WG valve and the pre-turbineSuppressing pressure. Thethe wastegate targeted of value an engine leads to extremely high pre-turbine pressures at medium Appl. Sci. 2017, 7, 350 16 of 25 is the pre-turbine pressure, which is set to increased values of 5%,to 10% high and loads. 15% Figure (multipliers 13 presents of the 1.05 pre-turbine pressures and peak engine torques achieved for various turbine sizes and the wastegate valve closed. As shown from the figure, a turbine 30% larger to 1.15)is the compared pre-turbine to thepressure, baseline which engine, is set butto increased always withinvalues of the 5%, set 10% pre-turbine and 15% (multipliers pressure limit. of 1.05 than the original size can provide or even increase the maximum power targets of the baseline engine Figureto 1.15) 18compared shows to that the baseline for a 10% engine, smaller but al turbineways within (multiplier the set pre-turbine of 0.9) than pressure the baseline limit. engine, without violating the pre-turbine pressure limit. A torque deficit occurs at 1000 rpm, but this is mainly the e-turboFigure can harvest18 shows energy that for (upa 10% to smaller 4 kW) atturbine most (multiplier points of of the 0.9)due map, than to excepta thepoor baseline model the 1000 engine,convergence rpm the speed, at this specific engine speed near the surge line and not the wheree-turbo the results can harvest are not energy highly (up trusted to 4 kW) due at tomost non-convergence points of the map, of except theincapacity model. the 1000 of Although, the rpm engine speed, to the achieve where increased the targeted torque. pre-turbinethe results pressure are not leads highly to trusted increased due energyto non-conv harvestingergence of levels, the model. it also, Although, as expected, the increased reduces the maximumpre-turbine power pressure output leads of the to engine.increased Aenergy 5% pre-turbine harvesting levels, pressure it also, increase as expected, leads toreduces a 5% the penalty maximum power output of the engine. A 5% pre-turbine pressure increase leads to a 5% penalty on on the engine’s maximum torque, while for the case with the pre-turbine pressure set at 15% higher, the engine’s maximum torque, while for the case with the pre-turbine pressure set at 15% higher, the the torque penalty is around 10%. torque penalty is around 10%.

Figure 13. Engine’s performance results for various turbines sizes and the wastegate completely shut: (a) pre-turbine pressure; (b) engine torque.

Figure 18. Phase 3: DoE analysis of the motor-generator average energy for differentFigure 14. turbine Phase 2:sizes Motor-generator and average energy for the case with original compressor, 30% larger Figure 18. Phase 3: DoE analysis of the motor-generator average energy for different turbine sizes and pre-turbine pressures; negative values indicate energy harvesting; positiveturbine values andindicate the WGthe needvalve completely shut; negative values indicate energy harvesting; positive pre-turbine pressures; negative values indicate energy harvesting; positive values indicate the need for energy provision; the blue line represents the maximum torque line of thevalues baseline indicate engine’s the needmodel for energy provision; the blue line represents the maximum torque line of the baseline engine’s model as provided by the manufacturer (λ < 1); the ⨂ symbol indicates limit for energyas provided provision; by the the manufacturer blue line represents (λ < 1); the the ⨂ symbol maximum indicates torque limit line violation; of the baseline ⌼ indicates engine’s energy model N violation; ⌼ indicates energy provision area. as providedprovision by area. the manufacturer (λ < 1); the symbol indicates limit violation; indicates energy provision area. The next stage in this phase was to investigate the potential of reducing the amount of energy Smaller Turbine: Variable Pre-Turbine Pressure Study needs to be provided by changing the compressor’s size. For this reason, five different compressors Smaller Turbine:The previous Variable study Pre-Turbine showed that Pressure with the Studyright component sizing,(multipliers energy ofharvesting 0.8 to 1.2) couldwere tested, be and the results are presented in Figure 15. Theachieved previous across study most of showed the speed/load that with map the area right of the component engine. However, sizing, this energy leads to harvesting torque sacrifice could be at full load conditions. This penalty can be reduced or even eliminated by adjusting the targeted achieved across most of the speed/load map area of the engine. However, this leads to torque sacriﬁce pre-turbine pressure of the engine when running at full load conditions. The present study shows the effects of various pre-turbine targeted values, larger and smaller than the baseline engine, to the maximum power output of the engine and the average power harvested or that needs to be provided by the motor-generator. Appl. Sci. 2017, 7, 350 17 of 25 at full load conditions. This penalty can be reduced or even eliminated by adjusting the targeted pre-turbine pressure of the engine when running at full load conditions. The present study shows the effects of various pre-turbine targeted values, larger and smaller than the baseline engine, to the maximumAppl. Sci. 2017 power, 7, 350 output of the engine and the average power harvested or that needs to be provided17 of 25 by the motor-generator. As is shownshown inin FigureFigure 1919,, atat fullfull loadload conditions,conditions, thethe maximummaximum powerpower outputoutput of thethe baselinebaseline engine can be met by running the engine at the or originaliginal pre-turbine pressures or or slightly lower. lower. However, thisthis leads leads to to energy energy provision provision demands demands by theby e-turbothe e-turbo of up of to 2up kW. to Higher2 kW. powerHigher outputs, power whichoutputs, often which work often as selling work points as selling for automotive points for manufacturers, automotive manufacturers, can be achieved can by runningbe achieved at lower by pressuresrunning at and lower providing pressures signiﬁcant and providing amounts sign ofificant power amounts to the compressor. of power to the compressor.

Figure 19.19. Effect of pre-turbine pressure %% changechange on:on: ((aa)) engine’sengine’s torquetorque atat fullfull load;load; (b(b)) M/GM/G performance atat fullfull load;load; negativenegative valuesvalues indicateindicate energyenergy harvesting.harvesting.

4.1.4. Overall Engine EfficiencyEfficiency The overall efficiencyefficiency ofof the baseline engine, whichwhich is calculated using the brake power achieved over thethe totaltotal fuelfuel power, power, can can reach reach levels levels higher higher than than 35% 35% at mediumat medium to highto high load load conditions. conditions. On theOn otherthe other hand, hand, the overallthe overall efficiency efficiency of the of electrically-assisted the electrically-assisted turbocharged turbocharged engine engine is given is given by the by brake the powerbrake power achieved achieved plus the plus energy the energy harvested harvested or provided or provided by the e-turboby the e-turbo divided divided by the totalby the fuel total power. fuel Figurepower. 20 Figure shows 20 the shows comparison the comparison charts between charts betw the overalleen the efficiency overall efficiency of the baseline of the enginebaseline and engine four casesand four studied cases in studied Phase 2in and Phase Phase 2 and 3 of Phase the steady-state 3 of the steady-state simulations. simulations. ItIt is evident in Figure 20 that the the implementati implementationon of a motor-generator motor-generator linked to the turbocharger leadsleads toto anan overalloverall engineengine efficiencyefficiency increaseincrease atat mediummedium and high loads and engine speeds for all of thethe cases.cases. The cases with a reducedreduced turbineturbine size (multiplier(multiplier of 0.9) benefitbenefit fromfrom anan increasedincreased overalloverall efficiencyefficiency acrossacross thethe whole whole map. map. Finally, Finally, the the high high pre-turbine pre-turbine pressures pressures show show a significant a significant effect effect on theon overallthe overall efficiency efficiency gain gain of the of model.the model. By assessing the work presented in thethe threethree phasesphases ofof thethe steady-statesteady-state simulations,simulations, the idealideal engine layout cancan bebe summarizedsummarized asas shownshown inin FigureFigure 2121..

Appl. Sci. 2017, 7, 350 17 of 25

As is shown in Figure 19, at full load conditions, the maximum power output of the baseline engine can be met by running the engine at the original pre-turbine pressures or slightly lower. However, this leads to energy provision demands by the e-turbo of up to 2 kW. Higher power outputs, which often work as selling points for automotive manufacturers, can be achieved by running at lower pressures and providing significant amounts of power to the compressor.

Figure 19. Effect of pre-turbine pressure % change on: (a) engine’s torque at full load; (b) M/G performance at full load; negative values indicate energy harvesting.

4.1.4. Overall Engine Efficiency The overall efficiency of the baseline engine, which is calculated using the brake power achieved over the total fuel power, can reach levels higher than 35% at medium to high load conditions. On the other hand, the overall efficiency of the electrically-assisted turbocharged engine is given by the brake power achieved plus the energy harvested or provided by the e-turbo divided by the total fuel power. Figure 20 shows the comparison charts between the overall efficiency of the baseline engine and four cases studied in Phase 2 and Phase 3 of the steady-state simulations. It is evident in Figure 20 that the implementation of a motor-generator linked to the turbocharger leads to an overall engine efficiency increase at medium and high loads and engine speeds for all of the cases. The cases with a reduced turbine size (multiplier of 0.9) benefit from an increased overall efficiency across the whole map. Finally, the high pre-turbine pressures show a significant effect on the overall efficiency gain of the model. Appl. Sci. 2017By, 7assessing, 350 the work presented in the three phases of the steady-state simulations, the ideal18 of 25 engine layout can be summarized as shown in Figure 21.

Appl. Sci. 2017, 7, 350 18 of 25 Appl. Sci. 2017, 7, 350 18 of 25

FigureFigure 20. Absolute 20. Absolute difference difference in overallin overall efficiency efficiency between between four different different e-turbo e-turbo configurations configurations and and Figure 20. Absolute difference in overall efficiency between four different e-turbo configurations and the baselinethe baseline engine; engine; negative negative values values indicate indicate efficiency efficiency loss;loss; positive values values indicate indicate efficiency efficiency gain; gain; the baseline engine; negative values indicate efficiency loss; positive values indicate efficiency gain; the areathe in area red in circle red circle indicates indicates non-convergence non-convergence of of the the model: model: (a) model model with with 30% 30% larger larger turbine turbine and and theWG area completely in red circle shut; indicates (b) model non-convergence with 20% larger of the turbine model: and (a) modelWG half with opened 30% larger compared turbine to and the WG completely shut; (b) model with 20% larger turbine and WG half opened compared to the baseline WGbaseline completely engine; shut;(c) model (b) modelwith 10% with smaller 20% turbinelarger turbine and 5% andincreased WG half pre-turbine opened pressurecompared compared to the engine; (c) model with 10% smaller turbine and 5% increased pre-turbine pressure compared to the baselineto the baseline engine; (engine;c) model (d with) model 10% with smaller 10% turbine smaller and turbine 5% increased and 15% pre-turbine increased pre-turbinepressure compared pressure baseline engine; (d) model with 10% smaller turbine and 15% increased pre-turbine pressure compared tocompared the baseline to the engine; baseline (d) engine.model with 10% smaller turbine and 15% increased pre-turbine pressure to thecompared baseline to engine. the baseline engine.

Figure 21. Ideal component sizing for maximum power and energy harvesting. FigureFigure 21. 21.Ideal Ideal component component sizing sizingfor for maximummaximum power power and and energy energy harvesting. harvesting. 4.2. Transient Analysis 4.2. Transient Analysis 4.2. TransientThe Analysis results in the steady-state simulations section demonstrated that a model fitted with a The results in the steady-state simulations section demonstrated that a model fitted with a Themotor-generator results in the and steady-state a turbine 10% simulations smaller than section the original demonstrated could provide that energy a model harvesting fitted and with a motor-generatorthermal efficiency and gain a turbine across the 10% speed/load smaller than map th ofe theoriginal engine. could The providetransient energy behaviour harvesting of the model and motor-generator and a turbine 10% smaller than the original could provide energy harvesting and thermalwith this efficiency configuration, gain across an original the speed/load size compressor map of the and engine. increased The pre-turbinetransient behaviour pressures of were the model tested thermalwithby efficiencyperforming this configuration, gain three across different an theoriginal speed/loadtypes size of simulations.compressor map of and the engine.increased The pre-turbine transient pressures behaviour were of tested the model with thisby performing configuration, three different an original types size of compressorsimulations. and increased pre-turbine pressures were tested by performing4.2.1. Load three Step differentTransient types of simulations. 4.2.1. Load Step Transient Model with PID Controllers Model with PID Controllers The load step transient study was repeated five times for 1600 and 1100 engine speeds by limitingThe theload amount step transient of power study (from was 1 to 5repeated kW) provided five times as electrical for 1600 assistance and 1100 to engine the compressor, speeds by as limitingshown inthe Figure amount 22. of power (from 1 to 5 kW) provided as electrical assistance to the compressor, as shown in Figure 22. Appl. Sci. 2017, 7, 350 19 of 25

4.2.1. Load Step Transient

Model with PID Controllers The load step transient study was repeated ﬁve times for 1600 and 1100 engine speeds by limiting the amount of power (from 1 to 5 kW) provided as electrical assistance to the compressor, as shown in FigureAppl. 22 Sci.. 2017, 7, 350 19 of 25

Appl. Sci. 2017, 7, 350 19 of 25

FigureFigure 22. Limits 22. Limits and and level level of of power power provided provided toto the compressor compressor during during tip-in tip-in transient: transient: (a) 1600 (a) 1600rpm; rpm; (b) 1100(b) 1100 rpm. rpm.

Figure 22. Limits and level of power provided to the compressor during tip-in transient: (a) 1600 rpm; It is clear that the maximum power limit set for the 1100 rpm case exceeds the model (b) 1100 rpm. Itrequirement, is clear that based the maximum on the current power PID limittuning, set for for achieving the 1100 the rpm fastest case possible exceeds response. the model requirement, based on theThe currentresults in PID Figure tuning, 23 reveal for achieving that an electrical the fastest assistance possible of response. 1 kW can reduce the transient It is clear that the maximum power limit set for the 1100 rpm case exceeds the model Theresponse results time in of Figurethe engine 23 byreveal 70% while that anhigher electrical levels of assistance power can oflead 1 kWto a reduction can reduce of more the transientthan requirement, based on the current PID tuning, for achieving the fastest possible response. response85%. timeAs shown of the in engine Figure by23a, 70% power while levels higher of more levels than of 3 powerkW have can almost lead toa negligible a reduction effect of on more the than The results in Figure 23 reveal that an electrical assistance of 1 kW can reduce the transient response time of the engine. It is obvious by looking at Figure 23b that the improvement in response 85%.response As shown time in of Figure the engine 23a, by power 70% while levels higher of more leve thanls of power 3 kW havecan lead almost to a reduction a negligible of more effect than on the time is due to the rapid increase of the turbocharger’s shaft speed. The fact that the initial shaft speed response85%. timeAs shown of the in engine. Figure 23a, It is power obvious levels by of looking more than at Figure 3 kW have23b thatalmost the a improvementnegligible effect in on response the is higher, as a result of the smaller turbine, also contributes to a fast transient change. Moreover, time isresponse due to time the rapidof the increaseengine. It ofis theobvious turbocharger’s by looking at shaft Figure speed. 23b that The the fact improvement that the initial in response shaft speed Figure 23c shows that the electrical assistance provided to the turbocharger pushes the operation of is higher,time is as due a resultto the rapid of the increase smaller of the turbine, turbocharger’s also contributes shaft speed. to Th a faste fact transient that the initial change. shaft speed Moreover, the compressor towards the right side of the mass flow/pressure ratio map, which leads to a more Figureis 23higher,c shows as athat result the of electrical the smaller assistance turbine, also provided contributes to the to turbocharger a fast transient pushes change. the Moreover, operation of efficient operation of the compressor. the compressorFigure 23c shows towards that thethe electrical right side assistance of the mass provided ﬂow/pressure to the turbocharger ratio map, pushes which the operation leads to aof more efﬁcientthe operationcompressor of towards the compressor. the right side of the mass flow/pressure ratio map, which leads to a more efficient operation of the compressor.

Figure 23. Cont. Appl. Sci. 2017, 7, 350 20 of 25 Appl. Sci. 2017, 7, 350 20 of 25 Appl. Sci. 2017, 7, 350 20 of 25

Figure 23. Load tip-in results at 1600 rpm engine speed: (a) BMEP response time; (b) turbocharger Figure 23. Load tip-in results at 1600 rpm engine speed: (a) BMEP response time; (b) turbocharger speed response time; (c) compressor performance. speedFigure response 23. Load time; tip-in (c) compressor results at 1600 performance. rpm engine speed: (a) BMEP response time; (b) turbocharger speed response time; (c) compressor performance. On the other hand, the transient improvement at very low engine speeds is of higher magnitude, Onas shown theOn otherthe in other Figure hand, hand, the24a,b. the transient transientThis happens improvement improvement due to the at very veryvery low lowlow engine enginepre-turbine speeds speeds ispressures of is higher of higher occurring magnitude, magnitude, at as shown1100as shown rpm. in Figure Moreover,in Figure 24a,b. as24a,b. illustrated This This happens happens in Figure duedue 24c, to the the electrical very very low lowassistance pre-turbine pre-turbine provided pressures pressures to the turbochargeroccurring occurring at at 1100 rpm.pushes1100 rpm. Moreover, the Moreover, operation as illustrated asof illustratedthe compressor in in Figure Figure towards 24 24c,c, thethe the electrical surge line assistance assistanceof the map. provided provided to the to turbocharger the turbocharger pushespushes the operation the operation of theof the compressor compressor towards towards thethe surgesurge line line of of the the map. map.

Figure 24. Load tip-in results at 1100 rpm engine speed; the two peaks highlighted in the red circle FigureFigureare 24. a resultLoad 24. Load of tip-in the tip-in mass results results flow at rate at 1100 1100fluc rpmtuation rpm engine engine when speed; the throttle the the two twovalve peaks peaks opens: highlighted highlighted (a) BMEP in response the in red the circle redtime; circle are a(are resultb) turbochargera result of the of massthe speed mass ﬂow responseflow rate rate ﬂuctuation time;fluctuation (c) compressor when when the performance. throttle valve valve opens: opens: (a) (BMEPa) BMEP response response time; time; (b) turbocharger(b) turbocharger speed speed response response time; time; (c ()c compressor) compressor performance.performance. Appl. Sci. 2017, 7, 350 21 of 25

Open-LoopAppl. Study Sci. 2017 (No, 7, 350 PID Controllers) 21 of 25

TheAppl. open-loopOpen-Loop Sci. 2017, 7, 350 Study study (No was PID performedControllers) for investigating the hardware capabilities without21 of 25 any effects from theThe tuning open-loop of thestudy PID was controllers. performed for Figure investig 25ating reveals the hardware the importance capabilities ofwithout the initial any level of power, asOpen-Loop welleffects as from theStudy the average (Notuning PID powerof Controllers)the PID provided controllers. to Figure the compressor. 25 reveals the Casesimportance with of the the maximuminitial level of amount of power providedpower,The open-loop as during well as the study average ﬁrst was section powerperformed provided beneﬁt for investigto from the compressor. theating fastest the hardwareCases response with capabilitiesthe and maximum highest without amount BMEP any levels. of power provided during the first section benefit from the fastest response and highest BMEP levels. On theeffects other hand,from the Figure tuning 25 ofb the shows PID controllers. that the power Figure distribution25 reveals the isimportance equally important of the initial as level the of average On the other hand, Figure 25b shows that the power distribution is equally important as the average power, as well as the average power provided to the compressor. Cases with the maximum amount power providedpower provided to the compressor to the compressor for afor fast a fast transient transient response time. time. of power provided during the first section benefit from the fastest response and highest BMEP levels. On the other hand, Figure 25b shows that the power distribution is equally important as the average power provided to the compressor for a fast transient response time.

Figure 25. Open-loop load tip-in results at 1600 rpm engine speed: (a) BMEP vs. time; (b) average Figure 25. Open-loop load tip-in results at 1600 rpm engine speed: (a) BMEP vs. time; (b) average power vs. time to torque; colour and symbol categorization indicates the amount of power provided power vs. time to torque; colour and symbol categorization indicates the amount of power provided during the first section. during the first section. 4.2.2.Figure Fixed 25. GearOpen-loop Vehicle load Speed tip-in Transient results at 1600 rpm engine speed: (a) BMEP vs. time; (b) average power vs. time to torque; colour and symbol categorization indicates the amount of power provided 4.2.2. Fixed GearThe Vehicleresults in Speed Figure Transient26 illustrate the required time for increasing the vehicle’s speed from during the first section. 1100 to 3000 rpm. As is expected, an aggressive tip-in leads to a fast vehicle speed transient. The pre- The resultsturbinein pressure Figure has 26 almost illustrate a negligible the required effect on timethe required for increasing time. The comparison the vehicle’s between speed the from e- 1100 to 4.2.2. Fixed Gear Vehicle Speed Transient 3000 rpm.turbo As is and expected, the baseline an engine aggressive for thetip-in 2–16 BMEP leads shows to a fastthe improvement vehicle speed in transient transient. response The for pre-turbine pressure hastheThe electrically-assisted almost results a in negligible Figure model. 26 effectillustrate On the on other the the requiredhand, required the comparisontime time. for increasing The for the comparison 2–4 the BMEP vehicle’s tip-in between showsspeed no thefrom e-turbo difference at all due to operating at the non-boosted area of the engine. Figure 26b demonstrates the and the1100 baseline to 3000 enginerpm. As foris expected, the 2–16 an BMEPaggressive shows tip-in the leads improvement to a fast vehicle in speed transient transient. response The pre- for the turbineamount pressure of energy has providedalmost a negligibleor harvested effect by the on motothe requiredr during time.the transient The comparison period of interest.between It theis e- electrically-assistedturboclear and that the during baseline model. the firstengine On phase the for of otherthe the 2–16 tip-in, hand, BMEP the themotor shows comparison provides the improvement a high for amount the in 2–4 oftransient power BMEP (limited response tip-in to showsfor no 5 kW) to the compressor for meeting the BMEP demands. Once the targeted boost pressure is differencethe electrically-assisted at all due to operating model. at On the the non-boosted other hand, the area comparison of the engine. for the 2–4 Figure BMEP 26 tip-inb demonstrates shows no the achieved, the e-turbo starts harvesting energy for most of the cases. The overall average energy amountdifference of energy at all provided due to operating or harvested at the non-booste by the motord area duringof the engine. the transient Figure 26b period demonstrates of interest. the It is demand is negative (energy generation) for most of the cases, except the most aggressive one, as amount of energy provided or harvested by the motor during the transient period of interest. It is clear that duringshown in the Figure first 26b, phase where of the the BMEP tip-in, is increasi the motorng for provides the whole aduration high amount of the transient of power as (limited to 5 kW)clear topresented that the during compressor in Figure the first 26c. forphase The meeting ofBrake-Specific the tip-in, the BMEPthe Fuel motor Consumption demands. provides a Once(BSFC) high amount theof the targeted engineof power is boost (limitedslightly pressure to is achieved,5 kW)deteriorated the to e-turbo the compressor when starts higher harvesting for pre-turbine meeting energy pressuresthe BMEP for th ande most mands.the baseline of theOnce model cases. the targetedare The applied, overall boost as shownpressure average in is energy achieved, the e-turbo starts harvesting energy for most of the cases. The overall average energy demand isFigure negative 26d. (energy generation) for most of the cases, except the most aggressive one, as shown demand is negative (energy generation) for most of the cases, except the most aggressive one, as in Figureshown 26b, in where Figure the 26b, BMEP where is the increasing BMEP is forincreasi theng whole for the duration whole duration of the transient of the transient as presented as in Figure 26presentedc. The Brake-Specificin Figure 26c. The Fuel Brake-Specific Consumption Fuel (BSFC) Consumption of the engine(BSFC) isof slightlythe engine deteriorated is slightly when higher pre-turbinedeteriorated when pressures higher than pre-turbine the baseline pressures model than are the applied,baseline model as shown are applied, in Figure as 26shownd. in Figure 26d.

Figure 26. Cont. Appl. Sci. 2017, 7, 350 22 of 25 Appl. Sci. 2017, 7, 350 22 of 25 Appl. Sci. 2017, 7, 350 22 of 25

Figure 26.FigureFixed 26. Fixed gear gear vehicle vehicle speed speed transient transientresults; results; % % values values represent represent pre-turbine pre-turbine pressure pressure increase increase comparedFigurecompared to 26. the Fixed to baseline the gear baseline vehicle engine: engine: speed (a ) (transienta engine) engine speedresults; speed vs.vs.% values time; time; (representb ()b motor-generator) motor-generator pre-turbine power pressure power vs. engine increase vs. engine speed; (c) BMEP vs. time; (d) Brake-Specific Fuel Consumption (BSFC) vs. engine speed. speed;compared (c) BMEP to vs.the time; baseline (d) Brake-Speciﬁcengine: (a) engine Fuel speed Consumption vs. time; (b) (BSFC) motor-generator vs. engine power speed. vs. engine speed; (c) BMEP vs. time; (d) Brake-Specific Fuel Consumption (BSFC) vs. engine speed. 4.2.3. Energy Balance Study 4.2.3. Energy Balance Study 4.2.3. EnergyThe energy Balance balance Study study was performed for the NEDC, WLTC, US06 cycles, as well as real Thedriving energyThe energyconditions balance balance cycles study study shown was was in performed the performed Methodology forfor the Section. NEDC, NEDC, WLTC, WLTC, US06 US06 cycles, cycles, as well as well as real as real drivingdriving conditionsFigure conditions 27 cycles reveals cycles shown that shown all inof in thethe the driving Methodology Methodology cycles tested Section.Section. provide a negative average power, which indicates energy harvesting. Energy is provided to the compressor during transient load and speed FigureFigure 27 reveals 27 reveals that that all all of of the the driving driving cycles cycles testedtested provide provide a anegative negative average average power, power, which which conditions. The motor-generator is almost inactive when running at low to medium load steady-state indicates energy harvesting. Energy is provided to the compressor during transient load and speed indicatesconditions. energy harvesting.Finally, at extra-urban Energy isdriving provided conditions, to the the compressor generators can during harvest transient a high amount load and of speed conditions. The motor-generator is almost inactive when running at low to medium load steady-state conditions.energy. The The motor-generator average amount of is energy almost harvested inactive is when highly running dependent at on low th toe driving medium behaviour. load steady-state The conditions. Finally, at extra-urban driving conditions, the generators can harvest a high amount of conditions.combined Finally, driving at conditions extra-urban and driving the US06 conditions, cycle, which therepresents generators aggressive can harvestdriving conditions, a high amount energy. The average amount of energy harvested is highly dependent on the driving behaviour. The of energy.demonstrated The average the maximum amount amounts of energy of energy harvested gain (6.6 is highly and 5.9 dependentkWh, respectively). on the The driving NEDC behaviour.and combined driving conditions and the US06 cycle, which represents aggressive driving conditions, The combinedWLTC cycles driving that conditions represent andmild thedriving US06 beha cycle,viour which provided represents energy aggressive gains between driving 0.4 conditions,and demonstrated0.9 kW. the maximum amounts of energy gain (6.6 and 5.9 kWh, respectively). The NEDC and demonstratedWLTC cycles the maximumthat represent amounts mild ofdriving energy beha gainviour (6.6 provided and 5.9 kWh, energy respectively). gains between The 0.4 NEDC and and WLTC0.9 cycles kW. that represent mild driving behaviour provided energy gains between 0.4 and 0.9 kW.

Figure 27. Average power provided/harvested by the motor-generator; negative values indicate energy harvesting. Figure 27. Average power provided/harvested by the motor-generator; negative values indicate FigureThe 27. fuelAverage consumption power provided/harvested of the model in transient by the mode motor-generator; cannot be highly negative trusted, values as it may indicate be affectedenergy by harvesting. the tuning of the different PID controllers used for the baseline and the electrically- energy harvesting. assisted models, as shown in Figure 28. The fuel consumption of the model in transient mode cannot be highly trusted, as it may be Theaffected fuel by consumption the tuning of of the the different model PID in transient controllers mode used cannotfor the bebaseline highly and trusted, the electrically- as it may be affectedassisted by the models, tuning as of shown the different in Figure PID 28. controllers used for the baseline and the electrically-assisted models, as shown in Figure 28. Appl. Sci. 2017, 7, 350 23 of 25 Appl. Sci. 2017, 7, 350 23 of 25

Figure 28. Fuel flow mass flow rate comparison between e-turbo and baseline model for the WLTC Figure 28. Fuel ﬂow mass ﬂow rate comparison between e-turbo and baseline model for the WLTC cycle; highlighted area indicates the disturbancedisturbance of PIDPID tuningtuning onon thethe accuracyaccuracy ofof thethe results.results.

The obtained results for the transient cycles show that with the implementation of the The obtained results for the transient cycles show that with the implementation of the e-turbocharger, the fuel economy of the engine is deteriorated by up to 1.8% (Table 7) due to the e-turbocharger, the fuel economy of the engine is deteriorated by up to 1.8% (Table7) due to the increased pre-turbine pressures. However, this fuel consumption increase is compensated by the increased pre-turbine pressures. However, this fuel consumption increase is compensated by the generated power of the e-turbocharger. The combined driving cycle demonstrates the maximum net generated power of the e-turbocharger. The combined driving cycle demonstrates the maximum net gain of 5.5 kWh, followed by the US06 cycle with 3.8 kWh. The WLTC is the only cycle that provides gain of 5.5 kWh, followed by the US06 cycle with 3.8 kWh. The WLTC is the only cycle that provides a a marginally negative energy gain when considering the increased fuel consumption. marginally negative energy gain when considering the increased fuel consumption.

Table 7.7. Fuel consumption difference between baseline andand e-turbo models for various cycles; energy gain == average average power power values values− heat − loss heat (Q) dueloss to ( increasedQ) due fuelto consumption,increased fuelQ (kJ)consumption, = Calorific valueQ (kJ) of= Calorific fuel (CV) value (kJ/kg) of fuel× (CV)fuel (kg),(kJ/kg) CV × offuel gasoline (kg), CV = of 47,300 gasoline kJ/kg; = 47,300 * considering kJ/kg; * considering the increased the fuelincreased consumption. fuel consumption. Cycle Difference (g/h) Difference (%) Energy Gain * (kWh) CycleNEDC Difference−3.7 (g/h) Difference−0.12 (%) Energy+0.37 Gain * (kWh) NEDCWLTC −−3.766.4 −1.340.12 −0.03 +0.37 WLTCUS06 −−66.4161.3 −1.801.34 +3.78−0.03 CombinedUS06 driving −161.3−81.8 −1.261.80 +5.52 +3.78 Combined driving −81.8 −1.26 +5.52 5. Conclusions 5. Conclusions The present paper provides a detailed theoretical analysis on the potential of e-turbocharging to controlThe load present while paper providing provides energy a detailed recovery theoretical for increasing analysis the on theoverall potential system of efficiency e-turbocharging and if topossible control replacing load while wastegate providing boost energy control. recovery The initia for increasingl results show the overall that at systemmedium efficiency to high andloads, if possiblesignificant replacing amounts wastegate of enthalpy, boost up control.to 5 kW, Theare lost initial through results the show WG that valve at of medium a 2.0 L toturbocharged high loads, significantspark-ignition amounts engine of at enthalpy, medium up to tohigh 5 kW, loads. are lostA significant through the amount WG valve of this of enthalpy a 2.0 L turbocharged loss can be spark-ignitionrecovered by the engine implementation at medium of to a highmotor-generator loads. A significant directly linked amount to the of thisshaft enthalpy of the turbocharger. loss can be recoveredThe study by the reveals implementation that replacing of a motor-generatorthe wastegate is directlyonly achievable linked to when the shaft a 30% of thelarger turbocharger. turbine is used Thefor non-violating study reveals the that pre-turbine replacing thepressure wastegate limit isofonly the engine. achievable However, when a large 30% largerturbine turbine leads to is usedvery poor for non-violating performance the and pre-turbine energy provision pressure needs limit ofat thethe engine.low loads However, and speeds. a large A turbine10% smaller leads toturbine very poorwhen performance combined with and increased energy provision pre-turbine needs pressures at the lowwas loadsfound and to provide speeds. efficiency A 10% smaller gains turbineand energy when harvesting combined conditions with increased across pre-turbine the whole pressuresarea of the was engine’s found load/speed to provide map. efficiency However, gains andthis energy energy harvesting harvesting and conditions efficiency across gain thecome whole with area a 5% of penalty the engine’s on the load/speed maximum power map. However,output of thisthe energyengine. harvestingIncreased andpower efficiency outputs gain are come still with achi aevable 5% penalty by reducing on the maximum the pre-turbine power outputpressure of the(opening engine. the Increased WG) and power providing outputs extra are energy still achievable to the compressor. by reducing the pre-turbine pressure (opening the WG)The andtransient providing response extra time energy of the to electrically the compressor.-assisted engine showed an improvement between 70% and 90% depended on the engine speed and the power provided to the compressor (~1 to 5 kW). Appl. Sci. 2017, 7, 350 24 of 25

The transient response time of the electrically-assisted engine showed an improvement between 70% and 90% depended on the engine speed and the power provided to the compressor (~1 to 5 kW). Testing the system under various driving cycles showed that the average amount of energy generated exceeds by up to 1 kW on average (6.6 kWh) the energy required by the motor during the transient events. The e-turbocharged engine demonstrated a fuel deterioration of up to 1.8% during the transient due to the increased pre-turbine pressures. However, this was compensated by the additional generated power. The maximum net energy gain of 5.5 kWh when considering the fuel consumption increase was for the combined driving cycle followed by the US06 cycle with a net gain of 3.8 kWh. Finally, it needs to be mentioned again that none of the results presented in this study take into account any electrical losses, such as alternator, converter and battery losses, which will obviously reduce the net amount of harvested energy. Future work will include investigations on gasoline and multi-turbo systems with a comprehensive turbine/compressor matching and the electriﬁcation of VGT turbocharging systems. Finally, the simulation studies can be supported by experimental investigations with actual hardware testing and e-turbocharger emulation studies (using an externally-boosted engine transient facility located at the University of Bath) for a comprehensive in-cylinder combustion analysis.

Acknowledgments: The authors would like to acknowledge Ford Motor Company and Jaguar Land Rover for their funding support for this research work. Author Contributions: Pavlos Dimitriou, Richard Burke and Harald Stoffels conceived of and designed the experiments. Pavlos Dimitriou performed the experiments, analysed the data and wrote the paper. Qingning Zhang and Colin Copeland contributed with their experience on electrically-assisted turbochargers and simulation tools. Conflicts of Interest: The authors declare no conflict of interest. The founding sponsors have agreed to publish the findings of this work.

References

1. Winterbone, D.E.; Benson, R.S.; Mortimer, A.G.; Kenyon, P.; Stotter, A. Transient response of turbocharged diesel engines. SAE Technical Paper 1977.[CrossRef] 2. Turner, J.W.G.; Popplewell, A.; Marshall, D.J.; Johnson, T.R.; Barker, L.; King, J.; Martin, J.; Lewis, A.G.J.; Akehurst, S.; Brace, C.J.; et al. SuperGen on ultraboost: Variable-speed centrifugal supercharging as an enabling technology for extreme engine downsizing. SAE Int. J. Engines 2015, 8, 1602–1615. [CrossRef] 3. Rose, A.J.M.; Akehurst, S.; Brace, C.J. Investigation into the trade-off between the part-load fuel efﬁciency and the transient response for a highly boosted downsized gasoline engine with a supercharger driven through a continuously variable transmission. Proc. Inst. Mech. Eng. Part D 2013, 227, 1674–1686. [CrossRef] 4. Katrašnik, T.; Rodman, S.; Trenc, F.; Hribernik, A.; Medica, V. Improvement of the dynamic characteristic of an automotive engine by a turbocharger assisted by an electric motor. J. Eng. Gas Turbines Power 2003, 125, 590–595. 5. Ibaraki, S.; Yamashita, Y.; Sumida, K.; Ogita, H.; Jinnai, Y. Development of the “hybrid turbo”, an electrically-assisted turbocharger. Mitsubishi Heavy Ind. Tech. Rev. 2006, 43, 1–5. 6. Millo, F.; Mallamo, F.; Pautasso, E.; Ganio Mego, G. The potential of electric exhaust gas turbocharging for HD diesel engines. SAE Tech. Pap. 2006.[CrossRef] 7. Burke, R.D. A numerical study of the beneﬁts of electrically-assisted boosting systems. J. Eng. Gas Turbines Power 2016, 138, 092808. [CrossRef] 8. Bumby, J.R.; Spooner, E.S.; Carter, J.; Tennant, H.; Ganio Mego, G.; Dellora, G.; Gstrein, W.; Sutter, H.; Wagner, J. Electrical machines for use in electrically-assisted turbochargers. In Proceedings of the Second International Conference on Power Electronics, Machines and Drives, Edinburgh, UK, 31 March–2 April 2004; IET: Stevenage, Hertfordshire, UK, 2004; Volume 1, pp. 344–349. 9. Bumby, J.; Crossland, S.; Carter, J. Electrically-assisted turbochargers: Their potential for energy recovery. In Proceedings of the IET The Institution of Engineering and Technology Hybrid Vehicle Conference, Coventry, UK, 12–13 December 2006; IET: Stevenage, Hertfordshire, UK, 2006; pp. 43–52. 10. Divekar, P.S.; Ayalew, B.; Prucka, R. Coordinated electric supercharging and turbo-generation for a diesel engine. SAE Tech. Pap. 2010.[CrossRef] Appl. Sci. 2017, 7, 350 25 of 25

11. Panting, J.; Pullen, K.R.; Martinez-Botas, R.F. Turbocharger motor-generator for improvement of transient performance in an internal combustion engine. Proc. Inst. Mech. Eng. Part D 2001, 215, 369–383. [CrossRef] 12. Terdich, N.; Martinez-Botas, R. Experimental efﬁciency characterization of an electrically-assisted turbocharger. SAE Tech. Pap. 2013.[CrossRef] 13. Arsie, I.; Cricchio, A.; Pianese, C.; De Cesare, M.; Nesci, W. A comprehensive powertrain model to evaluate

the beneﬁts of electric turbo compound (ETC) in reducing CO2 emissions from small diesel passenger cars. SAE Tech. Pap. 2014.[CrossRef] 14. Algrain, M. Controlling an electric turbo compound system for exhaust gas energy recovery in a diesel engine. In Proceedings of the 2005 IEEE International Conference on Electro information Technology, Lincoln, NE, USA, 22–25 May 2005; IEEE: New York, NY, USA, 2005; p. 6. 15. Pasini, G.; Frigo, S.; Marelli, S. Numerical comparison of an electric turbo compound applied to a SI and a CI engine. In Proceedings of the ASME 2015 Internal Combustion Engine Division Fall Technical Conference (ICEF2015), Houston, TX, USA, 8–11 November 2015. 16. Tavˇcar, G.; Bizjan, F.; Katrašnik, T. Methods for improving transient response of diesel engines—Inﬂuences of different electrically-assisted turbocharging topologies. Proc. Inst. Mech. Eng. Part D 2011, 225, 1167–1185. [CrossRef] 17. Tang, H. Application of Variable Geometry Turbine on Gasoline Engine and the Optimisation of Transient Behaviours. Ph.D. Dissertation, University of Bath, Bath, UK, 2015.

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).