CwftS'LAAS— 953 (o 6 CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE

LABORATOIRE DANALYSE ET ©’ARCHITECTURE DBS SYSTEMES

S.L ENGINE IDLE CONTROL IMPROVEMENT BY USING AUTOMOBILE REVERSIBLE "

L. KOUADIO, P. BID AN, M. VALENTIN, J.P. BERRY

LAAS REPORT 95268

JUNE 1995 DISTRIBUTION OF THIS DOCUMENT IS INJURED FOREIGN SALS PROHIBITS) ftn 6& LIMITED DISTRIBUTION NOTICE This report has been submitted for publication outside of CNRS It has been issued as a Research Report for early peer distribution DISCLAIMER

Portions of this document may be illegible in electronic image products. Images are produced from the best available original document S.I. ENGINE IDLE CONTROL IMPROVEMENT BY USING AUTOMOBILE REVERSIBLE “ALTERNATOR”

L.K. Kouadio, P. Bidan, M. Valentin, J.P. Berry

L.A.A.S./C.N.R.S., 7 Avenue du Colonel ROCHE, 31077 Toulouse Cedex FRAACF. Phone: (33) 61.33-64-17; Fax: (33) 61.55.35.77; e-mail: [email protected]

Abstract. This paper describes an original method for engine idle improve­ ment. It is well known that inappropriate idle-speed control increases fuel consumption, pollutant emissions and also reduces the idle quality. To prevent engine stalling and improve its performance during idling it is proposed a strategy based on two control loops such as that of the tra­ ditional air-flow ratio and an other providing an external supplementary torque via the automobile’s reversible “alternator”. Hence engine stalling is prevented due to the faster torque response and secondly Fuel/Air ratio is easily controllable due to the slow variation of the air-flow.

Keywords. Spark-Ignition Engine, Idle-speed control, Electrical assistance, Synchronous machine, Fuel Control, Automotive emissions, Multi-input system

1. INTRODUCTION others) to reduce pollutant emissions and fuel consump­ tion. Furthermore driveability improvement depends Spark ignition engine is nowadays the mostly used en­ on the smallness of torque fluctuation during idling. Un­ gine in automotive design because of its better power to fortunately, a lower speed causes engine stability deteri­ weight ratio despite its relatively low efficiency around oration and increases the engine sensitivity to external 30% combined with the high lower heating value of the torque disturbance (Nishimura and Ishii, 1986). To solve fuel. Nevertheless, all the politics of energy savings and the above mentioned problems this paper proposes to environment protection initiated since the last twenty include an external torque control loop within the auto­ years call for many interrogations. Thus we have to think mobile reversible “alternator”. The reasons behind this about the necessary improvements in the engine opera­ approach is to avoid large opening gradient and tion. While “cruising speed” phases are considered to be thereby to reduce the air intake nonlinearities effects on more or less clean, the transient are difficult to control. the control. In this way, an accurate F/A ratio is easily In particular the optimal Fuel/Air (F/A) ratio control attainable to allow the three-way catalyst converters to is more difficult to realize in transient due to the in­ keep to the bearable ±0.3% excursions from the stoichio­ take subsystems nonlinearities and dynamics (Aquino, metric conditions (Amstutz, 1994). The main novelty of 1981; Bidan et al. 1993; Bidan et at., 1995). As a con­ the system resides in the fact that the electrical machine sequence. fuel consumption and pollutant emissions in­ based the torque generation is the automobile “alterna­ creases (Braun el al., 1988). Concerning the idling it tor” . This machine is a Wound-Rotor Synchronous Ma­ is known that lower idle-speed is one method (amongst chine (S.M.) and is controlled to be able to operate in drive motor phase while improving its traditional use 3. S I. ENGINE MODELING as an alternator. Fig.l shows the overall system block diagram. Various models of S.I. engine have been presented in the past twenty years (Dobner, 1980). To achieve a control- oriented model the system was considered through the mean values of the state variables and divided in three

Tracking basic subsystems as shown in Fig.2. Controller Spark advance

Throttle opening Air flowrate Torque r.h Mechanical Speed Generation N r Injection time Fuel/Air ratio Engine T orque Dynamics Combustion)

Engine Synchronous Regenerative Battery Machine AC-DC Fig. 2. S.I. engine model description Converter The study of this model clearly shows two fundamental Fig. 1. Functional block diagram of the Idle-speed con­ dynamics in the intake manifold. The first one concerns trol system the air filling effect due to the intake manifold volume and the second one depends on the wall wetting by the fuel. In this paper it is considered that the electronic injection device compensate the fuel film dynamic to 2. ENGINE IDLING PROBLEMS maintain the Fuel/Air ratio (r*) to its nominal point during transients. Spark advance (a) is also supposed Engine idling has always revealed itself in terms of en­ to its nominal value. Thus, the throttle opening (<£) is ergy savings, reduction of pollutant emissions and also the only engine parameter available for the idle-speed in terms of idling quality which supposes a constant av­ control considered here. erage idle-speed and a minimum oscillation. However, when oscillations increase with weaker idle-speed, stabil­ ity degradation may occur following such uses as air con­ ditioning and etc, which in worst cases, 3.1 Air intake subsystem model will cause the engine to stall. Hence the crucial question about how to reduce engine idle-speed and at same time The air intake subsystem is governed by nonlinear equa­ improve its robustness with respect to disturbances. Sev­ tion where the manifold pressure (Pm) is the state vari­ eral works have been concerned with automotive en­ able. Taking Tm to be the manifold temperature, Pa at­ gine idle-speed control. HITOSHI (Inouie and Washino, mospheric pressure, R the gas constant of the mixture, 1990) proposes an idle-speed control system based on Mq the air molar mass, 7 the air specific heat ratio, compensating the variation of the alternator’s current Mjvc the air mass that can be absorbed by each cylin­ seen as a load disturbance. This control induces throttle der in normal conditions, ma the air mass really entering opening and thereby anticipates idle-speed drop. Also into a per , Vbdc the cylinder volume at air-how ratio and spark-advance are more thoroughly bottom dead center and n the number of cylinder respec­ studied by other authors (Aquino, 1981; William and tively, we consider the following nonlinear equations de­ Citron. 1984; Francis and Fruechte, 1983) to perform en­ scribing the air intake subsystem (Dobner, 1983; Moska gine idle-speed control. The solution proposed here has and J.K.Hedricks, 1992; Chaumerliac et al, 1994): t he dist inctive feature of combining the traditional air­ flow ratio control (via an electric throttle actuator) si­ multaneously with the S.M. operating as a synchronous Or(*)FWr,,) motor, to provide a supplementary torque. This control 120 approach leads to a multivariable system with, on the -[Pm-fr(A)] one hand two input variables defined by the throttle opening and the synchronous machine R.M.S. current (1) and on the other hand one output represented by the engine speed. The purpose of this configuration is to whe take advantage of the S.M. faster torque response to / 27 9(rp) = - r„ impose a smooth gradient of the throttle opening. 7 - 11 l2QRTm f' __ A/pV rr, 3.3 Linear idle modeling of the engine rcrit ' -y+1 ■ Before establishing the overall idle model of the engine, Pm >f ^ Fcrit the relation governing the rotational dynamics is here Pa modeled with a constant inertia J and given by the well r crit 7 ^ l*crit known mechanics law:

^(30^^^ " AF;& - AFfood (5) and Cr($) is the aeraulic resistance of the throttle, ra is the engine air intake ratio (normalized value). Pr is a Hence, all the equations established so far lead us to the partial pressure depending on acoustic phenomenon. linear engine idling model below (see Fig.3), deriving In this paper given that our purpose consist of control­ from Dobner’s idling linear small signal model (Francis ling the idle-speed of a four-cylinder engine at the vicin­ and Fruechte, 1983). ity of a nominal speed the linearisation of system (1), for small perturbations, yields ( —A Pm = — (A"$A

(A) denotes a small fluctuation around a nominal point.

3.2 Engine Torque modeling Fig. 3. Linear small signal model of the engine idling

Modeling the average of the engine torque consist in a re­ lationship between the engine control parameters (such as the spark advance a, the mixture F/A ratio r,, the air-mass ratio ra into the cylinder and the engine speed 4. ELECTRICAL ISSUES .V) to yield the torque at any speed. For this, assum­ ing an optimal spark advance the following relation has As explained previously, the principle of the proposed been experimentally identified: Idle-speed Control System (I.C.S.) is based on the ca­ pability of the automobile’s Synchronous Machine to F,/! = a0[l - ki(ri{t - To) - rim)2]ra{t - To) operate in motor phase and thus sustain the S I. en­ —o,i — an A (t) — a3N 2(t) (3) gine in idling phase. In this way to perform the control system some experiments and tests have been made on where rim is the Fuel/Air ratio which yields the maxi­ 1.5KW automobile “alternator”. The main test consists mum torque, F^ is the engine torque. a0, a\, a2, <2.3 and of recording the torque this machine is able to produce ki are constants. when it operates in motor phase (see Fig.6). But in first time the measurements of the “static torque”, which As shown above in 3, the torque modelreveals a time de­ consists of feeding two phases of the synchronous ma­ lay To between the change in the air-flow ratio (through chine windings with a Direct Current (/c) and record­ the inlet valve) and the corresponding change in the ing the torque versus the angle between phase 1 of the torque. This time lag becomes more and more prevailing stator and the rotor, leads to us a practical result in­ at low speed and increases the engine stalling risks if a volving a relation proportionality between torque and significant torque disturbance occurs. The idling control current (see Fig.4). Considering the back e.m.f. to be a system proposed in this paper should be able to prevent sinusoidal form the current necessary to produce a con­ the consequences of this delay by producing the external stant torque have to be also sinusoidal form (Marchand torque mentioned early. Assuming that r; is maintained and Razek , 1993). Thus the theoretical torque is given constant the simplified small signal linear model of the by: engine torque is expressed as: Fats = 3p\$vIejjcostf + Ld ^ LqI2f;sm(2^)j, (6)

AF th. = flo(l — k 1 (r, — rimY)Ar a(t — To) — f A.\ where p is the number of pole pair, <£„ the flux in the = knAra(t - To) — fA.\ (4) rotor and ip the angle between the current and the back where / — a-> + 2a3 .V0 emf, respectively. Ld and L„ are d,q axis inductances. If it is assumed the equality between La and Lq (smooth To estimate the theoretical maximum torque that the rotor machine) the equation above is reduce to S.M. is able to produce at any speed some simulations based on equation (8) has been made. These simula­ r.v/s = 3 p<$„ Ie// cost'. (7) tions takes account of the estimated values of the electri­ cal parameters of the synchronous machine Ls and Rs. In accordance with the generated torque of S.I. engines during idling (a few N.m) the simulations results allows us to determine the D.C. voltage necessary to permit the S.M. to produce a significant torque in order to sus­ tain effectively the S.I. engine. For instance according the mechanical output torque curve displayed by Fig.7 the S.M. is able to produce 13N.m to sustain the engine for a nominal idle-speed equal to 650 rpm and a gear ratio Nms/Nidi e equal to 2. In fact, given the automo­ bile environment the D.C. voltage E must not involves a radical change in on-board electrical equipments nor to too much supplement of weight. All the above consider­ ations suggest to us a D.C. voltage of 24V at the input of

Direct current Ic (A) the inverter. This inverter has been designed in our lab­ oratory especially for the idle control studied here and Fig. 4. Maximum “static torque ” versus the current Zc controlled by an hysteresis current controller (Lajoie- Mazenc et al., 1985). The Fig.6 shows a comparison of theoretical electromagnetic torque and the experimen­ tal others torque (experimental electromagnetic torque INVERTER and mechanical output torque) for approximative^ the Absolute encoder same current R.M.S at any speed. Obviously this voltage On-Board Battery level can appears to be much higher than the classical 12V in the automobile. However considering the growth of electrical on-board equipments in new generation of

Hysteresis it seems necessary to be take into account a fu­ current control ture innovations in automotive industry to accept this voltage level. reference generator

Comparison of theoretical and experimental torques

Fig. 5. Block diagram of the electrical part of the system According equation (7) three parameters are available to control the torque of the synchronous machine which are /e//,

600 800 1000 1200 1400 1600 1800 2000 Rotational: ed of the S.M. (rpm)

Fig. 6. Comparison between (l):theoretical torque, (8) (2)measured electromagnetic torque, (3): mechan­ where Vej / is the R.M.S voltage applied to one phase. Rs ical output torque a stator phase resistance, Ls one stator phase inductance and k = Ajp a speed coefficient respectively. 5. IDLE-SPEED CLOSED LOOP CONTROL 5.1 Simulation and results

This section is focussed on simulations for the proposed At time to of the idling a disturbance corresponding to idle control. The aim of these simulations is to show, a torque of 8N.m is applied on the engine shaft. Then on the one hand, the effect of a torque disturbance on we observe the idle-speed evolution in various cases such the idle-speed and on the other hand the improvements as: (1) no control is applied to cancel the disturbance ef­ that could be brought. To implement the control system fects, (2) a P.I controller is applied on the throttle open­ both the S.M. and the inverter are simply modeled as ing to modulate the air-flow ratio (3) both the throttle a pure gain depending on <$„ and i[>. The Idle Control opening control and the S.M. control are applied to can­ System (ICS) proposed here has to impose a short tran­ cel the disturbance effect. As concerns this last case, at sient on the engine speed when disturbance torque is the onset of the torque disturbance, the S.M. action pre­ directly applied on it. That supposes an energy provid­ vails while the air-flow control progressively takes over. ing from the battery. Conversely when the disturbance The results of these simulations given below show sig­ is due to the variation of on-board electrical energy con­ nificant improvements in robustness of the engine’s dy­ sumption the same ICS operates as a low-pass filter to namics simply by utilization of the serial “alternator” of allow a smooth application of the corresponding electri­ the automobile as a torque generator. cal torque on the engine. It implies that the supplement of energy consumption is provided by the battery in transient. Hence, in all of the cases described above the battery is requested to supply energy in transient. For that matter we opted for a control which combines the two control loops mentioned previously: (1) the air flow control loop and (2) the Synchronous Machine R.M.S. current control loop. In this configuration the current loop is designed without an integral action in order to avoid a continuous drive-motor functioning of the syn­ chronous machine. Also, since there are noises affect­ ing the engine speed we do not utilize derivative action. Only a proportional action is implemented to generate the R.M.S. current reference of the synchronous motor.

However, an integral action that allows the cancelling of Time (S) the idle-speed error is necessary in the air flow control loop. The Fig.7 shows the simplified diagram of the con­ Fig. 8. Comparison of three cases of idle-speed control trol strategy proposed here. In this diagram Ich is the in the presence of disturbances on-board load current and Nr represents the reference of the engine speed. 1: Disturbance torque

2: S.l. engine torque va

Engine Idling Linear model 3: Torque produced by the S.M ['see Fig.3) Controller

Time (S) Filter

Fig. 9. Different torque variation in the case of the com­ Fig. 7. The proposed idle control block diagram bined control in the presence of disturbance

Comparing curves 2 and 3 of Fig.8 we observe that the combined control strategy provides a better dynamics 120 Bidan. P.. S. Boverie and J.C. Marpinard (1993). State feedback linearizing control: Application to an en­ gine car. 12th World congress International Feder­ ation of Automatic Control. Bidan, P., S. Boverie and V. Chaumerliac (1995). Non linear control of a spark-ignition engine. IEEE Transactions on control systems technology. Braun, H.S., G. Kramer and M. Theissen (1988). Dy­ namic response-a new goal in engine-control appli­ 3 40 - cation. SAE paper n° 881155. Chaumerliac, V., P. Bidan and S. Boverie (1994). = 20 - Control-oriented spark engine model. Control En­ gineering Practice, Vol2, n°3, p. 381-387. Dobner, D.J. (1980). A mathematical engine model for Time (S) development of dynamic engine control. SAE paper Fig. 10. Throttle angle and the manifold pressure vari­ Dobner, D.J. (1983). Dynamic engine models for control ation in the presence of disturbances development-part 1: Non linear and linear model for­ for the mechanical system and a less idle-speed droop mulation. Ini. J. of Vehicle Design, Technological with a good damping than when the control is generated advances in Vehicle design series, SP4, Application by the throttle alone. These results are the consequences of control theory in the automotive industry. of the torque produced by the synchronous motor. Fig.9 Francis, E.C. and R.D. Fruechte (1983). Dynamic engine shown the disturbance and the S.M. torque evolution models for control development- part.2: Application while Fig. 10 shows that the throttle angle and the man­ to idle-speed control. Int. Journal of vehicle design. ifold pressure alter slowly. This control strategy leads to Special publication SP4- a smooth air-flow dynamics and ensures a better F/A Inouie, H. and S. Washino (1990). A improvement in ratio regulation. idle-speed control system with feed-forward com­ pensation for the alternator load current. S.A.E. International congress and exposition. 6. CONCLUSION Lajoie-Mazenc, M., C. Villanueva and J. Hector (1985). Study and implementation of hysteresis controlled This paper has proposed a novel method for the idle- inverter on a permanent magnet synchronous ma­ speed regulation. The simulations results show that the chine. ieee transactions on industry applications, combined method that has been proposed is interest­ ing in many respects. First it leads to a small idle-speed March and, C. and A. Razek (1993). Optimal torque drop when disturbance occurs reducing the risk of engine operation of digitally controlled permanent mag­ stalling. Secondly the ICS proposed here avoid a rough net synchronous motor drives. IEE Proceedings-B, control on engine parameters to maintain the nominal W.TM n°.3. idle speed in the presence of disturbances. Moreover Moska, J.J. and J.K.Hedricks (1992). Modelling and val­ it also helps improve both the utilization of the syn­ idation of automotive engines for control algorithm chronous machine and the battery charging manage­ development. ASME Journal of Dynamic Systems, ment. Theoretically this ICS could improves the car Measurement and control, p. 278-285 Volll4- driveadility. Also the strategy developed in this paper Nishimura, Y. and K. Tshii (1986). Engine idle stability could be extended to control transients when engine analysis and control. SAE paper n° 860412. speed varies up to 2000rpm. William, P. and S.J. Citron (1984). An adaptive idle mode control system. SAE paper n° 840443■

7. REFERENCES Amstutz, A. (1994). Model-based air-fuel ratio control in s.i. engines with a switch-type ego sensor. SAE paper n1 940972. Aquino. C.F. (1981). Transient a/f control of character­ istics of the 5 liter central engine. SAE Papers n° 810494.