Sunteți pe pagina 1din 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267576414

Vehicle Thermal Management: A Model-Based Approach

Conference Paper · October 2004


DOI: 10.1115/ICEF2004-0975

CITATIONS READS

7 449

2 authors:

Roberto Cipollone Carlo Villante


Università degli Studi dell'Aquila Università degli Studi dell'Aquila
146 PUBLICATIONS   713 CITATIONS    63 PUBLICATIONS   256 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

CO2 capture and sequestration View project

All content following this page was uploaded by Carlo Villante on 05 March 2014.

The user has requested enhancement of the downloaded file.


ICEF04:
ASME Internal Combustion Engine Division 2004 Fall Technical Conference
October 24-27, 2004 – Long Beach, CA, USA

ICEF2004-975

VEHICLE THERMAL MANAGEMENT: A MODEL-BASED APPROACH

Roberto Cipollone and Carlo Villante


Dept. of Energy -University of L’Aquila - ITALY

ABSTRACT block, increasing engine performances (compression ratio,


Cooling system design has a crucial role in defining engine boost pressure, etc.);
performance, operational limits and thermal comfort. Many • a down-sizing of many components (radiators, heat
further improvements with respect to the actual situation can be exchangers, etc.) can be proposed in order to achieve cost
obtained through a more accurate control of an-board thermal reductions;
needs. To this new interest the definition of new technical • cooling system layout may be simplified.
specifications must follow. The technical literature, however,
seems not fully satisfying this need, not focusing on the To reach these goals a traditional thermostat is no longer
influence of these technical specifications on system design, sufficient and more actuation variables must be included and
reliability and costs. controlled by means of an electronic device (Srun, 1999, Da
In this paper the authors present a contribution in this Silva, et al., 2000).
direction, showing the capabilities of an active intelligent The definition of new technical specifications in this three
management of the engine cooling system. This can be frames has implications on different aspects connected with
obtained through different control strategies, strongly circuit layout, component choice, system cost and engine
diversified for their cost-performance ratio. The potentiality of performances, specially in terms of environmental impact and
a model-based controller has been also investigated and noise. Engine manufacturers often choose less innovative
compared with the correspondent closed-loop controller. solutions based on the traditional system (enhanced on peculiar
aspects). On the other end, technical literature usually proposes
INTRODUCTION very appealing and innovative systems which are certainly
Both environmental and energetic considerations are superior in terms of vehicle thermal management but are often
nowadays changing the perspective in cooling system design. too expensive for the manufacturers.
Many are, in fact, the opportunities which can be caught by a To give an engineering approach to the problem, the
more intelligent engine thermal management (Melzer, et al., technical specifications for the cooling system can be divided in
1999; Kanefski, et al., 1999; Ng, et al., 2000, Valaskai, and three subgroups which respectively regard: the metal parts; the
Jouannet, 2000; Forrest, and Bhatti, 2002): lubricant; cabin comfort.
• engine warm-up period may be limited, enhancing the The first two aspects make reference to traditionally
efficiency of the catalytic converter and limiting emissions; foreseen expectations (engine power and maximum torque, fuel
• lubricant temperature may be optimized, increasing engine consumption, emissions, noise, etc.), while the third is strictly
mechanical efficiency and limiting energy losses; connected to vehicle comfort (and consequently to its market
• mean metal temperature may be increased, so determining an success).
increase in the mechanical and thermal efficiency of the This makes it necessary to introduce a sort of
engine; standardization on the measurement of cooling system
• the quality of cabin conditioning may be improved; performances to better permit the comparison of different
• coolant temperature fluctuations during time can be limited solutions and afterwards the choice of the desired system in
producing a more stable behavior of the cooling system; terms of a better performance-cost ratio.
In this paper the authors firstly describe in some detail the
• the temperature difference between the fluid entering and
novel technical specification to be obtained by the cooling
leaving the engine may be reduced increasing engine
system and then propose a set of standard vehicle missions to
resistance to thermal fatigue and reducing thermal
be used for system performance comparison.
distortions;
• a separated engine head and block cooling can be obtained,
lowering mean head temperature and increasing that of the

1 Copyright © #### by ASME


assure passenger comfort. Stable air and coolant flow rates
SYSTEM SPECIFICATIONS are therefore needed during engine transients. Air and coolant
flow rates may anyway be changed to follow correspondent
On engine metal parts variations in the power request by passenger demand: this, on
• Maximum temperature must be limited to preserve materials its turn, may change as a consequence of variations in the
and, in spark ignition engines, to avoid knocking phenomena temperature desired at cabin, or in the outside temperature;
due to the presence of hot points inside the cylinder: the vehicle speed also plays its role, varying the convective heat
thermal inertia of the metal parts and the response time of the transfer between the vehicle and the environment. A low
cooling system play a key role to delay engine heating in response time of the cooling system is required to limit cabin
critical conditions. temperature oscillations: this may be certainly reduced by
• Temperature gradients must also be limited because they may increasing fan speed. Fan noise, however, also plays its role
determine dangerous distortions during engine exercise, in the definition of passenger comfort and must be, therefore,
causing mechanical stresses and imperfect coupling of the taken into account.
various moving parts: a suitably designed distribution of • Cabin warm-up (in winter) or cool-down (during summer)
metal masses and coolant paths inside engine head and block period must be also limited. During winter the power request
is crucial at this regard. would increase engine warm-up time. The two specifications
• Temperature oscillations during time must be limited to may be therefore be in contrast.
increase engine resistance to thermal fatigue. It’s convenient • Other specifications have to be imposed in terms of the air
to distinguish between high frequency (in-cycle) and low humidity inside the cabin, but are not the object of this paper.
frequency (cycle-by-cycle, due to load and speed transients)
fluctuations. High frequency temperature fluctuations are These technical specifications are firstly taken into account
limited to a very small part of the engine walls facing the in the design of fluid (coolant and lubricant) paths inside the
burnt gases (usually referred as “penetration thickness”). engine head and block, which should guarantee the slightest
Literature showed that at most 2-3 millimeters of the walls temperature gradients inside the metal parts, together with the
experience high frequency fluctuations: the phenomenon is limitation of metal masses and lubricant content to shorten
dominated by turbulence intensity inside the cylinder, and by warm-up phases.
the thermal conductivity of the metal parts. Lower frequency After this first design activity, the technical specifications
fluctuations, on the other end, are strongly dependent on the regarding the metal parts and the lubricants are translated in
efficiency of the cooling system and the response time of the terms of correspondent specifications to be imposed on the
cooling system plays a fundamental role. coolant temperature. These latter specifications are the
• The warm-up time must be limited to reduce engine objectives to be reached by a choice and correct design of the
emissions (which are high until the catalytic converter has remaining components of the cooling system (radiator, cabin
not reached its operating temperature), and to enhance engine conditioner, electric or mechanical pump, thermostat or
thermal efficiency lowering fuel consumption. This proportional valves, …), and of the system control device.
specification calls for a low inertia of the metal parts and for
low or null coolant flow rates in warm-up conditions: it’s TEST MISSIONS
important to notice that these requests are partially in contrast Normal practice and economy reasons suggest to attempt
with the formers: a low metal inertia, in fact, may lead to too the obtainment of the technical specifications by means of two
high temperatures during high charge and low speed transient set-points to be matched by system controller: the first on
conditions. For this reason, engine head and block are usually coolant temperature; the second one on cabin temperature. A
designed for normal operation, leaving the satisfaction of more accurate control would be possible aiming at a direct
warm-up specifications on charge of the cooling system measurement of the metal temperature in some critical point in
controller. engine head: however many aspects connected with system
wirings and with its maintenance, reliability and cost make this
On the lubricant opportunity still difficult to be applied (unless on high
performances and cost vehicles).
• Maximum lubricant temperature must be limited, as well as
All this considered, at least five different vehicle missions
temperature fluctuations during speed and charge engine
should be performed to sufficiently define the performance of a
transients, to grant stable lubricant properties. A low
cooling system and of its control device. These five missions
response time of the cooling system is required and high
are proposed to put in evidence the performances of various
coolant flow rates at the oil cooler must be assured in most
control schemes, different both in engine thermal behavior and
critical situations.
in cabin interior comfort. Making only reference to a typical
• Lubricant warm-up time must be short to limit the extension EuroIV test driving cycle some differences in system
of insufficiently lubricated operating conditions, enhancing performances could, in fact, remain hidden.
engine mechanical efficiency as well as the life of most of the
metal components life. Low or null coolant and lubricant
M1. Typical driving cycle
flow rates are therefore needed at the oil cooler during warm-
The EuroIV test driving cycle (known as NEUDC) is
up phases.
normally used as representative of a standard succession of
transient operating conditions. The objective of the controller in
Cabin comfort this condition is the limitation of coolant and cabin temperature
• Temperature oscillations inside the cabin must be low to oscillations around predefined set points under stabilized

2 Copyright © 2004 by ASME


transient conditions. The amplitude of these oscillations in the value. Some systems, normally equipping only top-class
coolant and in the cabin temperatures (respectively ∆Tcool,M1 and vehicles can assure cabin comfort in these conditions with an
∆Tcab,M1) may be used as performance parameters. additional electric air heater.
In fact, if the oscillations of coolant temperature around an
imposed set-point are maintained low, benefic effects will be M5. Hot stop in front of a wall
obtained in terms of thermal fatigue of the metal parts (coolant The vehicle should stop with the engine off starting from
temperature variations generate correspondent fluctuations in hot engine conditions. The radiator inlet at the air side should
metal temperature). Moreover the limitation of temperature be partially obstructed (simulating car stop in front of a wall).
oscillations normally permits to raise the set-point enhancing The objective of the controller in this mission is the limitation
both engine thermal and mechanical efficiency (coolant of coolant temperature below a maximum acceptable value.
properties are much more stable), and saving energy. The maximum temperature reached by the coolant in this test
Cabin temperature fluctuations around the set point (imposed (Tcool,max,M5) may be used as a performance index.
by vehicle user), on the other end, are a good index of the cabin In most of the traditional cooling systems, in fact, when the
interior comfort. In this work we will not cope with further vehicle is stopped the energy stored in the metal parts is
specifications regarding the humidity control inside the cabin. transferred to a still coolant (coolant pump stops feeding the
circuits). In these conditions, even with the radiator fan on,
M2. Warm-up transient energy transfer may not be adequate since almost no fluid runs
The vehicle should run steadily at a medium charge steady through the radiator. Coolant may therefore vaporize, so that
condition starting from cold engine conditions. The aim of the very high temperatures are reached at the metal surfaces. The
controller is the limitation of coolant and cabin warm-up times problem is enhanced if the vehicle stops in front a wall, so
(∆tcool,M2 and ∆tcab,M2): those can be defined as the reaching partially obstructing radiator air inlet.
times of a given percentage of the set-point values. Temperature overshoot may be avoided if the pump is kept
This times, in fact, can be chosen as representative of the on by a delay in engine stop (kept in idle for some seconds), or,
engine capacity to reach easily an hot operative condition. A in more advanced systems, by an electric engine equipping the
faster warming of the engine will result in lower emissions and pump and controlled by the ECU.
energy consumption. The choice of a steady operative
conditions makes the results of the tests with different cooling POSSIBLE CONTROL LAYOUTS
system more comparable. The two most widely proposed innovations consist in the
electrical actuation of the coolant pump and in the adoption of
M3. Critical mission one or more electronically controlled proportional valves
The vehicle should run steadily at most critical charge replacing the thermostat. Whatever is the solution, the extended
conditions for the cooling system (normally low vehicle speed number of degrees of freedom makes it necessary to deepen
and high engine charge) starting from cold engine conditions. dynamical behavior of the system, so being able to synthesize
The aim of the controller is the limitation of coolant an efficient controller (Srun, 1999, Da Silva, et al., 2000; Yoo,
temperature below an acceptable value. The maximum et al., 2000; Cortona, and Onder, 2000; Wagner, et al., 2001;
temperature reached by the coolant in this test (Tcool,max,M3) is a Allen, and Lasecki, 2001; Ng, et al., 2002; Wagner, et al.,
good index of the response time of the cooling system. 2002).
In those transient conditions, in fact, coolant temperature Many are, however, the possible obtainable layouts,
can go beyond its maximum steady value, due to the delays obtained by the adoption of the various innovative components.
introduced in the actuation of the control variables. In standard Here, five control layouts are proposed and described in
thermostat-based systems, this phenomenon makes it necessary some detail. Those are here reported in order of complexity
to limit normal coolant temperature far below its possible (and cost) of the required additional components (having as
limits. Faster-response control systems may overcome this reference a traditional thermostat controller)
problem, permitting the system to work much closer to its
limits (therefore enhancing thermal and mechanical efficiency). C1. Traditional controller
Coolant temperature is thermo-mechanically controlled by
M4. Easiest mission the thermostat. Radiator fan must be activated to increase the
The vehicle should run steadily less critical charge thermal efficiency of the radiator in particular conditions: low
conditions for the cooling system (high vehicle speed, low coolant flow rate (idle engine) and air flow rate (vehicle
engine charge) at low external temperature (high cabin stopped). Cabin temperature is electronically controlled by a
requirements) starting from hot engine conditions. The valve intercepting the air flow through the cabin conditioner.
objective of the controller in this mission is the limitation of Many systems are now being developed by the manufacturers,
coolant and cabin temperature above an acceptable value. The which include a mechanical clutch to activate the coolant
minimum coolant and cabin temperatures reached by coolant in pump. When the pump is actuated its speed is proportional to
this test (Tcool,min,M4 and Tcab,min,M4) are performance indexes for engine speed: the clutch may be used at engine start to shorten
the cooling system. the warm-up period.
In low charge conditions, in fact, the power release by the Advantages:
engine can be lower than that required by cabin heating. If this • Cost, consolidated practice and reliability.
happens, engine is cooled back enhancing emissions and energy Drawbacks:
consumption. Moreover, cabin temperature set point may not • Pump must be over-dimensioned to give high flow rates at
be maintained and cabin will be cooled down up to a steady low speed;

3 Copyright © 2004 by ASME


• Radiator must be over-dimensioned to give high heat transfer
at low air flow rates; C4. On-off electric pump
• Thermostat must be passed through by high flow rates (high Thermostat is replaced by one proportional valve on the
inertia - response time); radiator branch. The two set-points (coolant maximum
• Need of an additional on-off valve or a pump clutch to temperature and cabin interior temperature) are obtained by a
shorten warm-up; proper choice of the position of the two valves (one on the
• The low response time of the thermostat makes it hard to coolant and the other on the air flow). The mechanical pump is
satisfy mission M3, forcing the designer to lower the opening substituted by an electric one, leading to a simpler control:
temperature of the thermostat. As a result normal operative pump speed may be fixed at one or two predefined levels (so
temperature (e.g. during mission M1) are much lower than that system behaviour doesn't depend on engine speed) to
those optimal for engine and lubricant working (lowering simplify pump control electronics and limit additional costs.
overall vehicle efficiency and enhancing emissions). Pump may be stopped during warm-up. The by-pass branch
may be eliminated.
C2. With one proportional valve Advantages:
Thermostat is replaced by one proportional valve on the • More efficient control (both the set-points may be varied;
radiator branch. The two set-points (coolant maximum without pump speed disturbances);
temperature and cabin interior temperature) are obtained by a • Energy saving by pump optimization;
proper choice of the position of the two valves (one on the • Pump and radiator can be significantly reduced due to pump
coolant and the other on the air flow). An additional on-off speed increase (higher flow rates at the radiator, especially in
valve may be used to shorten engine warm-up period, critical conditions at low engine speed)
increasing the hydraulic resistance of the system and therefore • No clutch is needed to shorten warm-up
lowering coolant flow rates. Drawbacks:
The cost of the proportional valve is much higher than that • Cost
of the thermostat. The increase in system cost is justified only • More space needed under-hood.
by a correspondent relevant increase in system performances.
Advantages:
• Efficient electronic control; C5. Electric pump with speed control
• Coolant temperature set-point may be varied continuously by Thermostat is replaced by one proportional valve on the
the ECU; radiator branch. The mechanical pump is substituted by an
• Mean coolant temperature levels are significantly increased electric one. The air valve is eliminated since the control action
due to the lower response time of the controller. is performed by an additional pump speed controller. The by-
Drawbacks: pass branch may be eliminated. The radiator fan may be
• Need of an additional on-off valve or a pump clutch to probably eliminated: when engine is idle and vehicle is
shorten warm-up. stopped, in fact, the efficiency of the radiator may be increased
by increasing pump speed (and consequently coolant flow rate).
C3. With two proportional valve Advantages:
Thermostat is replaced by one proportional valve on the • More efficient control (both the set-points may be varied;
radiator branch. The air valve is replaced by another without pump speed disturbances)
proportional valve on the cabin conditioner branch. The two • Radiator fan elimination (cost and space saving, noise
set-points are obtained by a proper choice of the position of the reduction).
two valves. The two valves may be closed to shorten engine • No clutch, No by-pass branch, No air valve (space saving
warm-up period, increasing the hydraulic resistance of the under-hood)
system and therefore lowering coolant flow rates. The by-pass Drawbacks:
branch may be eliminated. • A non-standard cabin heat exchanger should be used since
Advantages: those normally implied are not strongly dependent on coolant
• Efficient electronic control; flow rate. This means that coolant flow rate (and therefore
• Coolant temperature set-point may be varied continuously by the position of a proportional valve) cannot be efficiently
the ECU; used to control cabin temperature. If a standard cabin is used
• Space saved under-hood since the air valve is not included. an air valve should be included.
Drawbacks: • Higher cost
• Cost
• Energy losses during warm-up (pump continuously running); CONTROL ORIENTED MODELING
• A non-standard cabin heat exchanger should be used since To compare all the possible layouts it is necessary to set-up
those normally implied are not strongly dependent on coolant an intense theoretical activity aimed to the derivation of a
flow rate. This means that coolant flow rate (and therefore complete control-oriented model of the overall system.
the position of a proportional valve) cannot be efficiently The model must be able flexible and fully modular in order
used to control cabin temperature. to take into account all the presented options and to compare
those in terms of their performance on the previously discussed
vehicle missions.
It is necessary to keep the models of the components

4 Copyright © 2004 by ASME


involved in the control action (thermostat, proportional valves, cc01, Dutycc and V are the input values for the
speed controllers, air valves, …) substantially distinguished by “Aerodynamics” sub-model: this simulates the air-side
the models of those components involved in heat transfer behavior of the cabin conditioner and of the radiator and
(engine, radiator, cabin heater, …). evaluates the air flow rates passing through those components
The first set of models would be used, in fact, in the (mrad and mcac).
synthesis of an overall model of the controller. On the other The remaining two sub-models (respectively regarding the
end, the second set of models would be properly used to hydraulic and thermodynamic behavior of the components) are
represent the various possible system layouts, building what we described in more detail in the following. These two sub-
called a “virtual cooling system”. models, in fact, have been afterwards inverted in order to derive
The coupling of the “virtual engine cooling system” with the model-based controller.
the controller model permits an off-line evaluation of the
efficiency of new control strategies, model-based too. The aerodynamic sub-models
It is also necessary to keep distinguished the models of the Equations (1) describe the aero-dynamical behavior of the
sensors and of the actuators, so avoiding confusions among the radiator: it is convenient to proceed by steps, modeling firstly
intrinsic response-time of the cooling system and those the behavior of the radiator in two basic conditions (fan on and
introduced by the controller (sensors and actuators). vehicle stopped: State 1; fan off and vehicle in motion: State 2)
and afterwards representing the generic state of the radiator (fan
THE VIRTUAL COOLING SYSTEM On and vehicle in motion) as a linear combination of its basic
The authors developed an integrated model of an engine states.
cooling system (Cipollone, et al., 2003; Cipollone, and For as regards State 1, the model balances the pressure
Villante, 2003) which is characterized by a complete drop defined by the characteristic curves of the fan (as a
modularity, so permitting to simulate all the possible design function of air mass flow rate and fan duty cycle (Drad%), being
configurations The modeling approach was to keep physical Arad, Brad and Crad functions of Drad%:) and that lost in the
consistency as high as possible, in order to allow generality. radiator (evaluated by component characteristic curve) so
However, the simulation is complex and requires the utilization evaluating the air flow rate through the radiator.
of experimental data coming from conventional tests. For as regards State 2, a suitable experimental correlation
The description of the heat transfer dynamics between the was used, giving air flow rate as a function of vehicle speed.
various fluids (air, gas, coolant, lubricant) and the metal parts Lastly, an interference coefficient (αrad) was used, which
makes it necessary to define the properties of the components defines the interaction between the two basic behaviors: the
form various different point of views: hydraulics, thermo- actual air flow rate through the radiator may be therefore
dynamics an aerodynamics are therefore involved. determined as a linear combination of those evaluated in the
All the components must be modeled in a way which two basic states. αrad has to be experimentally calibrated.
makes it easy to interface them with the models of the control
devices (thermostat, proportional valves, sensors, actuators,
…). Fan ON - Stopped
The hydraulic and thermal behavior of the system are ∆p Rad = ARad (D Rad % ) ⋅ m Rad 2
+ ...
usually inter-dependent since some components of the  Fan

controller modify the hydraulic layout (by the thermostat or ... + B Rad (D Rad % ) ⋅ m Rad Fan + C Rad (D Rad % )
prop. valves) depending on coolant temperature. 
∆p Rad = k1Rad ⋅ m Rad Fan + k 2 Rad ⋅ m Rad Fan
2

Moreover, the aerodynamic behavior of the cabin heater is


also sometimes dependent on coolant temperature since the air Fan OFF - Vehicle in motion (1)
valve position may be modified by the controller.
An engine model is also needed to define its heat m RadV = c1 Rad ⋅ V 2 + c 2 Rad ⋅ V
production in the various operative conditions.
Generic condition
Model derivation m Rad = m Rad Fan + α ⋅ m RadV
The derivation of the model in full detail is reported in 
α = 1 if the fan is OFF, < 1 elsewhere
(Cipollone, et al., 2003): it is fully modular and it’s
characterized by various independent sub-models. Vehicle
Equations (2) describe the aero-dynamical behavior of the
mission is defined as a function of time in terms of vehicle
cabin conditioner: the model used is analogue to that proposed
speed (V), gear (g) and load (Ch). It is also necessary to
for the radiator in State 1: the air mass flow rate is calculated
describe the disturbance introduced within the system by the
balancing the pressure drop given by the fan (as a function of
user maneuvering cabin exchanger equipment (by means of cc01
its duty cycle) and that given by a pressure losses correlation
and Dutycc variables, defining if the conditioner is on or off,
for the cabin conditioner. The air flow to the cabin conditioner
and the duty cycle of the fan).
V, g and Ch are the input values for the “Engine” sub- may be intercepted by a valve, whose opening degree is φAir
model: this simulates engine behavior through an inverse quasi-
steady approach determining pump speed (Npump), thermal ∆p Cac = ACac (DCac % )⋅ mCac2
+ ...

... + BCac (DCac % )⋅ mCac + C Cac (DCac % )
power transmitted to the coolant (Pt) and the temperature and (2)
flow rate of the exhaust gas which is re-circulated (Tegr and
megr).

( )
∆p Cac = k1Cac ⋅ mCac + k 2Cac ⋅ mCac φ Air
2 2

5 Copyright © 2004 by ASME


considering their inertial properties (MMet⋅cMet): PGas is
The hydraulic sub-models calculated by the Engine sub-model through steady-state maps;
In the simulation of the hydraulic behavior of the system, PCool is correlated to coolant flow rate and to its entering and
the hypothesis is made that each component has a quadratic leaving temperatures, PAir is calculated by an overall heat
behavior for as regards the pressure losses at coolant side transfer coefficient (considering conductive and convective
(eq.3). The pump characteristic curves, on the other end, are phenomena) from the metal parts to the external air. TOut in eq.
quadratic in the coolant volumetric flow rate for any given (10) can be evaluated by eq. (11).
pump speed (eq. 4). Pressure drop at valves is modeled as an
inverse quadratic function of their opening degree φ (eq. 5).  PCool = Q ⋅ cCool ⋅ (TOut − TIn )
P = S ⋅ hMet Air ⋅ (TMet − T Air )
 Air Met Air
∆pi = K i ⋅ Qi2 (3) (10)
∆pP = AP (N P ) ⋅ Q + BP (N P ) ⋅ Q + CP ( N P )
2
(4) dTMet PGas − PAir − PCool
=
dt M Met ⋅ c Met
∆p = K ⋅ Q 2 φ 2 (5)
Q ⋅ cCool ⋅ (TOut − TIn ) = S Met Cool ⋅ hMet Cool ⋅ (TMet − T )
The quadratic approach for the pressure drops at each  (11)
T = (TIn + TOut ) 2
component makes it possible to consider the complete cooling
circuit fed by the pump as an equivalent quadratic pressure
drop deriving from series and parallel connection of the various A similar model has also been tested for the radiator to describe
branches. The mathematical rules describing those series and the thermal inertia of the metal parts, properly substituting eq. 8
parallel links are respectively reported in eq. 6 and 7. with a dynamical equation like eq. 9. For the tested systems,
however, (see Cipollone, et al., 2003) the description of the
∆p = K eq ⋅ Q 2 , ∆p = ∆p1 + ∆p2 , Q = Q1 = Q2 only metal inertia of the engine metal parts proved to be
 (6) sufficient to describe the overall behaviour of the system.
 K eq = K1 + K 2
∆p = K eq ⋅ Q 2 , ∆p = ∆p1 = ∆p2 , Q = Q1 + Q2 CONTROL SCHEMES
Each one of the non conventional electronic controllers can

(
 K eq = (K 1 ⋅ K 2 ) K 1 + K 2 + 2 K 1 ⋅ K 2 ) (7)
be synthesized following one of the 3 following approaches:

Considering such links, system layout can be represented Closed-loop


as “seen from the pump”: an equivalent circuit hydraulic The system will be fully auto-adaptative. The quality of the
resistance can so be evaluated and matched with the pump control action depends on the correct determination of the
characteristic curves. proportional, derivative and integrative constants of the PID
controller: depending on the chosen control layout the actuation
The thermodynamic sub-models variables may vary (proportional valve and air valve openings,
The energy conservation equation is applied to each node pump actuation and speed, etc.). The definition of the PID
where two or more fluid flows converge (eq. 8). The constants should also consider possible control instability.
temperature of the fluid coming out from the node (TOut) results
by a mean of the temperatures of the entering fluids (TIn1 and Open-loop (Model-based)
TIn2) weighted by their volumetric flow rates (Q1 and Q2). The system will not be auto-adaptative. Steady state errors
The model used for the radiator is reported in eq. (9). The can be found if mathematical model used to set the control
term [A’Rad(Q,mrad)/Q] substantially represents a dimensionless action is not exactly reproducing the real system. These errors
heat transfer coefficient between coolant and external air: the are not visible off-line since the controller is tested on a virtual
proposed approach matches the normal off-board experimental cooling system.
characterization of the radiators. A similar approach is Some parameters can be anyway varied in the virtual
proposed for the cabin conditioner and for the EGR exchanger, system to perform a sensitivity analysis.
which involve an heat transfer process between the coolant and The actuated variables are set at each time instant to the
an air or burnt gas flow. value they should have at the steady-state to guarantee the
obtainment of the set points.
(
TOut = Q1 ⋅ TIn1 + Q2 ⋅ TIn2 ) (Q + Q )
1 2 (8) The model-Based approach should normally result in a
lower response time and cannot lead to system instability.
QRad ⋅ (TIn − TOut ) = A Rad (QRad , mRad ) ⋅ (TIn − TAir )
'

 ' (9)
 A Rad (QRad , mRad ) QRad = k1rad (mrad ) ⋅ QRad k 2 rad (mrad ) Mixed approach
The system should take the best from the two approaches: a
first tentative solution is given by a model-based controller
The mass of the coolant store at the coolant tank introduces
granting low response time.
an inertial effect, which substantially introduces a first order
The tentative solution is afterwards corrected by a closed-loop
dependency of the temperature of the exiting flow on that of the
controller to reach the needed precision and to gain an auto-
entering flow coming back from the circuit.
adaptative behavior.
Eq. (10) and (11) are relative to the mathematical model of
engine metal parts. Eq. (10) describes energy conservation,

6 Copyright © 2004 by ASME


Some thermostats also include, at the by-pass branch outlet
SIMULATED LAYUOTS an additional pressure-dependent valve which lets the coolant
flow towards the by-pass only if the pressure drop through the
C1. Traditional controller thermostat is over a certain activation level. This additional
The thermostat has been modeled using an non- valve can normally be modeled by a further pressure-dependent
dimensional approach by the definition of two variables (T% opening degree φPr to multiply the φbypass value evaluated by
and xTh %) which correlate the state of the thermostat eq.14. Moreover, the dynamical behaviour of the valve is
(temperature TTh and needle position xTh) with opening start and usually not negligible and can be taken into account by a first
end conditions (TTh start and TTh end): order model:

( )(
T% = TTh − TThStart TThEnd − TThStart )  P% = (∆PPr − ∆PPr Start ) (∆PPr End − ∆PPr Start )

( )(
 xTh % = xTh − xThStart xThEnd − xThStart ) (12) 
( )(
 x Pr % = x Pr − x PrStart x PrEnd − x PrStart )
x Pr % = APr P%5 + B Pr P%4 + C Pr P%3 + D Pr P%2 + E Pr P%
x% is expressed as a function of T% by a fifth order (16)
dx Pr % x Pr % x Pr %
polynomial experimental correlation: + =
dt τ Pr τ Pr
xTh % = ATh T + BTh T + C Th T + DTh T + ETh T% φ Pr = (1 − e ) (1 − e )
5 4 3 2 − k Pr ⋅ x Pr % − k Pr
% % % % (13)

Anyway the thermostat normally shows a strongly where P% and xPr % are two non dimensional variables
unsteady behavior due to an hysteresis cycle to which the wax introduced to correlate the state of the valve (pressure ∆PPr and
is subjected. Quasi-steady opening and closure tests for the needle position xPr) with opening start and end conditions (∆PPr
thermostat, in fact, show that two different laws (both of eq. 13
start and ∆PPr end):
type) are followed depending whether temperature is raising or
decreasing. In addition to this, needle movement has its
mechanical inertia due to its mass and to the spring preloads, as
C2. With one proportional valve
well as to wax inertial effects. The controller must be able to evaluate the values of the
To describe the overall behavior of the component, two opening degrees of the proportional valve on the radiator
distinct opening laws were used depending on the sign of
temperature variations and a first order model was introduce to branch (φrad) and of that on the air flow to the air conditioner
describe valve dynamics. (φAir) basing on the imposed values for the two set points: the
coolant temperature leaving the engine (Tcool) and the
temperature desired inside the cabin (Tcab).
 ATh '
T%5 + BTh '
T%4 + C Th
'
T%3 + DTh
'
T%2 + ETh
'
T% ,
 Here the synthesis of a feed-forward controller is briefly
if dT reported, which is based on the inversion of some parts of the
%
≥0
 dt mathematical model of the cooling system.
xTh % =  For as regards the identification of φrad two separate steps
 ATh T% + BTh T% + C Th T% + DTh T% + ETh T% ,
'' 5 '' 4 '' 3 '' 2 ''

 (14) must be followed: firstly, the coolant flow rate needed at the
 dT%
if <0 radiator is calculated basing on thermal considerations;
 dt secondly, φrad is evaluated basing on hydraulic considerations
dxTh % xTh % xTh % Some assumptions were made to permit the inversion of
+ = the thermal model of the cooling system:
dt τ Th τ Th
• the system has to be considered at its steady state;
• the power dissipated by the engine towards the environment
This complex dynamical behavior normally results in high (PAir) is negligible with respect to that dissipated through the
response times of the C1 controller. This makes it necessary to coolant.
decrease the opening temperature of the thermostat Under these assumptions, the thermal power dissipated by
(anticipating the coolant flow towards the radiator) not to the radiator must be equal to PGas+PEGR-PCab: PGas and PEGR are
experience to high coolant temperature during severe engine evaluated by the ECU by look-up tables; PCab is instead
transients (like M3 mission). This decreases mean engine calculated by ECU maps basing on external temperature and on
operating temperatures, so lowering thermal and mechanical cabin interior requirements.
efficiency. Rearranging eq.10, the mass flow rate at the radiator which
Once needle position is known, the opening degrees of the allows for the required thermal exchange is calculated as:
three thermostat outlets can be normally expressed as
exponential functions of x%: 1
 Ptot  k 2 rad (mrad )
( )(
φ Rad = 1 − e − kOpened ⋅ xTh % 1 − e − k Opened

) Qrad = 
 ccool ⋅ ρ cool ⋅ k1rad (mrad ) ⋅ (TCool − TExt )
(17)

((k
φ bypass = 1 − e Closed Th %
⋅x −1)
)(
1 − e − kClosed ) (15)
 φrad must be now evaluated, being known Qrad. The
φ cab = 1 hydraulic model of the system (eq. 4-8) has therefore been

7 Copyright © 2004 by ASME


inverted: the procedure to be used during the inversion depends
on the peculiar geometry of the considered layout. Any system
however can be inverted properly considering the parallel and
series branches and applying eq. 4 to 8 to evaluate the needed
opening degree for the proportional valve.
Once φrad is known, eq. 4-8 are used to evaluate the
coolant flow rate at the cabin conditioner which is used in the
evaluation of φAir.
Also for the identification of φAir a two step procedure must
be followed: the air flow rate needed at cabin conditioner is
calculated basing on thermal considerations (inverting eq.9-12);
afterwards, φAir is evaluated basing on aerodynamic
considerations (inverting eq.3).

SIMULATION RESULTS
Model validation required a great amount of experimental
data regarding not only vehicle mission (in terms of actual
vehicle speed, gear and load condition) and some out variable
(temperature and flow rates at various point of the system), but
especially the experimental calibration of all the components of
the cooling system (EGR and cabin conditioning heat
exchangers, radiator, thermostat, pipes and connections, etc.).
All of them had to be tested separately and subjected to
experimental campaigns aimed to the definition of suitable
parameters and experimental correlations to be mathematically
represented.
The validation of the integrated model has been carried out
basing on the data relative to a driving test on the road which
was realized on a dedicated and properly instrumented vehicle
Fig.1
(a Citroen C3 1.4L TDI). This validation activity is fully
described in (Cipollone et al, 2003).

Further model validation


A dedicated experimental activity was carried out to newly
validate the hydraulic behavior of the system in the traditional
C1 layout.
Measured values for the coolant temperature at the engine
outlet and for the engine speed has been used as inputs for the
model of the hydraulic behavior of the system, and are reported
in Figure 1.
The two horizontal lines define the operational limits of the
thermostat (respectively the temperature values at which it
starts and ends its opening movement towards radiator branch
(contemporaneously determining a closure of the by-pass
branch).
Upper part of Figure 2 reports the evaluated pressure drop
at the by-pass orifice: the two horizontal lines define the
operational limits of the pressure dependent valve (opening
start and end conditions).
These data, in combination with those plotted in the upper
part of Figure 1, determine the opening ratios of the orifices
towards the radiator and the by-pass (lower part of Figure 2.
Those latter values completely define the pressure drops of all
the branches of the system and permit the evaluation of any
possible flow rate or pressure drop.

Fig.2

8 Copyright © 2004 by ASME


Measurements were taken for radiator and cabin
conditioner flow rates, as well as for radiator and cabin
conditioner inlet pressure drops. A comparison between the
measured values and the correspondent model evaluations is
reported in Figures 3 and 4. The agreement is quite satisfactory,
especially considering the difficulty of the on-board
measurements and the complexity of the model: all the major
trends are correctly predicted. The time constants of the two
valves proved therefore to be equal to the imposed values of 30
and 15 seconds, respectively for the temperature and the
pressure dependent orifices (τTh and τPr). As a final conclusion,
it can be observed that the hypotheses made to represent the
behaviour of the thermostat appear quite congruent, as well as
the physical consistency of the model.

Simulation and comparison of some control layouts


The behavior of a vehicle equipped by a C1 and C2
cooling system have been simulated. The C2 controller has
been simulated for all three possible control schemes: closed-
loop (C2A), feed-forward (C2B), and mixed (C2C).
Here only the results relative to a M1 mission (EURO4 test
driving cycle) are reported since no substantial differences
between the systems are found in the other missions.
The only insertion of a proportional valve in place of the
thermostat, in fact, cannot cope for example with the
performance of the system during warm-up. It is anyway
expected to substantially influence the mean operating
temperature of the engine during normal unsteady functioning
Fig.3 (like that represented by the EURO4 test cycle of M1 mission).
Figure 5 reports the comparison of the performances of the
four simulated layouts in terms of the two controlled variables:
respectively coolant temperature and cabin interior temperature.
The reported oscillations are those evaluated in stabilized
transient conditions, simulating subsequent M1 missions.
The results are plotted as a percentage of the optimal
desired values: 95 °C coolant temperature; 20 °C cabin
temperature. External temperature is 5 °C. The optimal coolant
temperature was chosen to optimize engine thermal and
mechanical efficiency.
Table 1 reports the evaluated values for the mean square
root error around the desired set-points (σ) of the four
simulated layouts both for coolant and for cabin temperature
values.
The first thing which is noteworthy to be underlined is that
the C1 controller maintains mean coolant temperature much
below the optimal value. This is due to the anticipated opening
of the thermostat, which has to start opening at 80°C not to
reach to loose coolant temperature control under more severe
transients (due to its high response time).
On the other end, Figure 5 clearly demonstrates that all the
C2 controllers succeed in taking coolant temperature up to the
desired value and to maintain it very near to the optimal value.
Table 1 highlights that the best performance is that of the
C2C system, as expected. This kind of controller in fact is able
to join the advantages of the feed-forward and closed-loop
controller obtaining a low response time and a limited steady-
state error.

Fig.4

9 Copyright © 2004 by ASME


fuel consumption;
• maximum lubricant temperature must be limited, as well as
its temperature fluctuations stabilize lubricant properties;
• lubricant warm-up time must be shortened to limit the
extension of insufficiently lubricated conditions;
• temperature oscillations inside the cabin must be lowered to
assure passenger comfort.
• cabin warm-up (in winter) or cool-down (during summer)
period must be also limited.
• cabin air humidity has also to be controlled.

These specifications are reachable with new control


schemes which could produce a substantial increase in vehicle
overall performance. Nevertheless an engineering approach to
the matter should also permit to take into account further costs
which have to be balanced by correspondent performance
increments.
Here a layout including an electronically controlled
proportional valve in substitution of the thermostat is proposed.
The proportional valve intercepts the coolant directed towards
the radiator. Both closed-loop and feed-forward (model-based)
control strategies were tested. A mixed actuation was also
Cabin temperature proposed where a model-based action is corrected by a closed-
103 loop controller.
The feed-forward controller was derived making use of a
physically consistent model which was previously developed
% of the optimal value

102
by the same authors. The model was inverted to derive a
model-based controller.
101 The three control schemes (closed-loop, model-based and
mixed) were tested off-board and compared with a traditional
100 “thermostat-based” controller.
As expected, the traditional controller proved no to be able
to take coolant temperature up to the optimal value (chosen in
99 order to optimise engine thermal and mechanical efficiency)
due to the anticipated opening of the thermostat (usually
98 required for safety reasons).
0 200 400 600 800 1000 1200 The proposed electronic controllers, instead, succeed in
Time [s] taking coolant temperature up to the desired value and to
Fig.5 maintain it very near to the optimal value. The best
performance is that of the system equipped by the mixed
C1 C2A C2B C2C closed-loop & feed-forward controller, so justifying a model-
based approach.
σTcool [K] 12.1 3.7 3.3 3.1
σTcabin [K] 0.28 0.20 0.17 0.19
ACKNOWLEDGMENTS
Table 1 This work has been funded by Cluster 20 – Piano
tecnologie eco-compatibili – MURST”, by the Progetto di
CONCLUSIONS ricerca di interesse nazionale 1999-2002 – MURST: “Procedure
Traditional cooling systems (which make use of a teorico sperimentale per lo sviluppo di sistemi innovativi di
thermostat, eventually electrically assisted) are not capable to iniezione nei MCI”, and by MARK-IV Systèmes Moteurs -
match the new specifications that are nowadays required in France.
terms of the thermal state of the engine, as well as of cabin
comfort: REFERENCES
• maximum temperature of the engine metal parts must be Alzer, F., Hesse, U., Rocklagge, G., Schmitt, M. 1999,
limited to preserve materials and to avoid knocking Thermo-management, SAE paper 1999-01-0238.
phenomena; Kanefski, P., Nelson, V., Ranger, M. 1999, A Systems
• temperature gradients must also be limited inside metal parts Engineering Approach to Engine Cooling Design, SAE paper
not to determine distortions during engine exercise; 1999-01-3780.
• temperature oscillations during time must be limited to Ng, E. Y., Johnson, P. W., Watkins, S., Grant L. 2000,
increase engine resistance to thermal fatigue; Wind-tunnel tests of vehicle cooling system performance at
• metal parts warm-up time must be limited to reduce engine high blockage, SAE paper 2000-01-0351.
emissions and to enhance engine thermal efficiency lowering Valaszkai, L., Jouannet, B. 2000, Cooling system

10 Copyright © 2004 by ASME


optimization for Euro4 – EPA/02 heavy duty trucks, SAE paper Cortona, E., Onder, C. H., Guzzella, L., 2002, Model based
2000-01-0964. temperature control for improved fuel economy of SI engines,
Forrest, W. O., Bhatti, M. S. 2002, Energy efficient IFAC paper.
automotive air conditioning system, SAE paper 2002-01-0229. Ng, E. Y., Watkins S., Johnson, P., Mole, L. 2002, Use of a
Srun, N. A. 1999, A simple engine cooling system pressure-based technique for evaluating the aerodynamics of
simulation model, SAE paper 1999-01-0237. vehicle cooling systems, SAE paper 2002-01-0712.
Da Silva, K., Lebrun, M., Samuel, S. 2000, Modeling and Wagner, J. R., Ghone, M. C., Dawson, D. W., Marotta, E.
simulation of a cooling system using multi-port approach, SAE E. 2002, Coolant flow control strategies for automotive thermal
paper 2000-01-0292. management systems, SAE paper 2002-01-0713.
Yoo, I. K., Simpson, K., Bell, M., Majikowski, S. 2000, An Cipollone, R., Villante, C., Komurian, R., Peuch, J. 2003,
engine coolant temperature model and application for cooling A system approach to mathematical modeling of cooling system
system diagnosis, SAE paper 2000-01-0939. dynamics, 4th Int. Conf. On control and Diagnostics in
Cortona, E., Onder, C. H. 2000, Engine thermal Automotive Applications (CDAuto03) – Genova – ITALY, July
management with electric cooling pump, SAE paper 2000-01- 2003.
0965. Cipollone, R., Villante, C. 2003, A system approach to
Wagner, J. R., Marotta, E. E., Paridis, I. 2001, Thermal mathematical modeling of cooling system dynamics, Int. Conf.
modeling of engine components for temperature predictions “Transport Means 2003” – Kaunas – LITUANIA, October
and fluid flow regulation, SAE paper 2001-01-1014. 2003.
Allen, D. J., Lasecki, M. P. 2001, Thermal management
evolution and controlled coolant flow, SAE paper 2001-01-
1732.

11 Copyright © 2004 by ASME

View publication stats

S-ar putea să vă placă și