Archives of Computational Methods in Engineering

https://doi.org/10.1007/s11831-019-09361-9

ORIGINAL PAPER

Internal Combustion Engine Heat Transfer and Wall Temperature

Modeling: An Overview
Leonardo Fonseca1,3 · Pablo Olmeda2 · Ricardo Novella2 · Ramon Molina Valle1

Received: 30 April 2019 / Accepted: 30 September 2019

© CIMNE, Barcelona, Spain 2019

Abstract
Internal combustion engines are now extremely optimized, in such ways improving their performance is a costly task.
Traditional engine improvement by experimental means is aided by engine thermodynamic models, reducing experimental
and total project costs. For those models, accuracy is mandatory in order to offer good prediction of engine performance.
Modelling of the heat transfer and wall temperature is an important task concerning the accuracy and the predictions of any
engine thermodynamic model, although it is many times an overcome task. In order to perform good prediction of engine
heat transfer and wall temperature, models are required for accomplish heat transfer from hot gases to engine parts, heat
transfer inside each engine part, and also heat transfer to coolant and lubricating oil. This paper presents an overview about
engine heat transfer and wall temperature modelling, with main purpose to aid engine thermodynamic modelling and offer
more accurate predictions of engine performance, consumption and emission parameters. The most important correlation
are reviewed for three engine heat transfer approaches: gas to wall, wall to wall and wall to liquid heat transfer models. In
order to obtain good prediction of wall temperature, those three approaches must be coupled, which may imply convection-
conduction-convection problems, although for some applications in diesel engines, radiation problems must be considered.

Keywords Internal combustion engine · Engine wall temperature modeling · Engine heat transfer modelling · Engine
thermodynamic modelling

1 Introduction of cleaner and more efficient powertrains [1]. After decades

of optimization research and design, Internal Combustion
The global awareness towards the greenhouse gases emis- Engine (ICE) are now extremely optimized, in such a man-
sions has led to a more stringent Internal Combustion Engine ner that improving their performance is really difficult.
(ICE) emissions legislation, thus focusing the automotive Experimental research has always been the main option in
researchers and manufacturers attention on the development engine development, although it is very expensive. Simula-
tion tools that predict the engine cycle are frequently used
in the development of new engines. The accuracy of the
The author Leonardo Fonseca acknowledges CAPES
simulation results depends on the accuracy of the underly-
(Coordination for the Improvement of Higher Education
Personnel) for the scholarship from the program “CAPES - ing models [2, 3].
DEMANDA SOCIAL”, Ph.D. level. In the past few decades, engine CFD 3D simulation has
become an attractive option, specially for aiding experimen-
* Leonardo Fonseca tal methods and reducing its cost. In order to aid experi-
[email protected]
mental methods, accuracy is required from engine CFD 3D
1
PPGMEC/UFMG - Mechanical Engineering Post Graduation modelling. The implementation of different engine thermo-
Program - Federal University of Minas Gerais, 6627 Antonio dynamic models depends on the wall temperature used, in
Carlos Av., Belo Horizonte 31270‑901, Brazil order to offer more accurate results. This way, both research-
2
CMT Motores Termicos, Universitat Politècnica de València, ers and engineers need to model or measure properly wall
Camino de Vera s/n, 46022 Valencia, Spain temperatures if they want accuracy.
3
DETEM/UFSJ - Telecommunications and Mechatronic Research on in-cylinder heat transfer of internal combus-
Engineering Department - Federal University of São João tion engines has been conducted since 1923 [4], in order to
del-Rei, MG 443, km 7, Ouro Branco 36420‑000, Brazil

13
Vol.:(0123456789)
 L. Fonseca et al.

optimize engine design, due to the significant impact of heat combustion engine application, in such a manner that some
transfer on engine’s efficiency, power and emissions [2, 5, 6]. regions of the engine are described separately. The wall to
Prediction of heat transfer plays an important role in engine wall classification contains specifically heat transfer phe-
development as heat losses influence overall engine effi- nomena through solid media inside the engine, and the meth-
ciency, exhaust emissions and component thermal stresses ods to solve this heat transfer problem. Finally, the wall to
[7]. By modeling engine heat transfer, design changes can liquid classification contains heat transfer phenomena asso-
be readily assessed with minimal prototype construction and ciated with coolant liquid and lubricating oil. In Sect. 4, the
testing. This enables new engines to be designed more rap- main topics of this review are summarized.
idly, cost effectively, and with improved thermal signatures
[8].
In the past few decades, several research efforts has 2 ICE Thermodynamic Modelling
shown that engine wall temperatures change considerably for
different operating conditions, specially on what concerns Engine thermodynamic models are valuable tools for engi-
wall heat transfer. The wall heat transfer can contribute to neers either working in the automotive industry or belonging
a significant heat loss in the overall fuel conversion pro- to the research community, with the aim to design engines
cess, and the in cylinder combustion processes and engine that comply with the strict emissions legislation while keep-
operation can also be affected. In addition, the wall heat flux ing high performance [10–12]. High cost experimental tests
determines the distribution of thermal stress in the mechani- are increasingly replaced by cheap system simulations of the
cal components of the engine. engine. With current trends like alternative fuels, downsiz-
The measurement of wall temperatures is an expen- ing, EGR, multiple spark plugs per cylinder, etc. it is indeed
sive alternative, once the most used techniques nowadays no longer possible for an R&D engineer to intuitively grasp
involves drilling operations that usually modify the meas- how these factors will affect the engine operation. The
ured engine, or makes it unable to operate. Other measure- employed engine models are obviously required to reproduce
ment techniques involves state of the art equipment or newly any fuel specific effects on the combustion process [13].
developed optical techniques [9], which are expensive. Both Thermodynamic models are the subject of earlier over-
estimation and calculation of engine heat transfer depends view paper [14]. From the thermodynamic point of view,
on proper wall temperature modelling. There are many dif- several model frameworks are used for the simulation of
ferent models for heat transfer calculation on internal com- the closed part of spark-ignition and compression ignition
bustion engines, although most of those models do not cal- engine cycle, these can be classified as [14]:
culate wall temperature.
There are lots of different ways to measure or model inter- – zero-dimensional
nal combustion engine heat transfer and wall temperatures. – multi-zone
It is the purpose of this paper to review internal combustion – multi-dimensional
engine heat transfer and wall temperature modelling. The
state of the art concerning those subjects is described, and Those three categories of engine thermodynamic model-
a link between engine heat transfer modelling and thermo- ling are the subject of comparison for some authors. Pari-
dynamic modelling is highlighted. This way, the engineer otis et al. [10] compared the results obtained using a zero-
or researcher involved in engine design and optimization dimensional, a multi-zone and a multi-dimensional model,
which needs proper estimation/calculation of engine wall all of them applied to the simulation of a high speed single
temperature values, should take note at this review and chose cylinder diesel engine. The engine operates in motoring con-
the option that best fits its goals and resources. ditions, from 1200 until 3000 rpm. The zero-dimensional
This paper is organized as follows. In Sect. 2, ICE ther- model calculates with reasonable accuracy the in-cylinder
modynamic modelling is briefly described as the main tool pressure, while it rather underestimates the mean gas tem-
for aiding engine development, and the importance of heat perature near the top dead center (TDC) compared to the
transfer models and wall temperature values as boundary other two models, meanwhile it has the minor computational
conditions for the first law of thermodynamics is described. time. The multi-zone model provides information for the
In Sect. 3, ICE Heat transfer and wall temperature model- local in-cylinder temperature distribution and describes
ling are reviewed and classified in three main heat transfer qualitatively correctly the effect of the cylinder on the flow
phenomena: gas to wall(convection and radiation problems), field.
wall to wall(conduction and contact resistance problems) The first two types are classified as phenomenological
and wall to liquid(convection problems) values. The gas models, where the equations constituting the basic struc-
to wall classification contains the heat transfer phenom- ture of the model are based on conservation of mass and
ena most widely studied and published concerning internal energy and are only dependent on time (resulting in ordinary

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

differential equations) [14]. Multidimensional models are model. This model was able to predict well the in-cylinder
also termed fluid mechanic or fluid dynamic models, where temperature trace, but could only estimate in-cylinder tem-
the governing equations are the Navier–Stokes equations in perature, meanwhile it was able to resolve the closed cycle
addition to conservation of mass and energy (the equations in only 1 s using a personal computer.
are also dependent on the spatial coordinates, and so take Rashedul et al. [17] used a single zone model to study
the form of partial differential equations) [14]. convective heat transfer inside a spark ignition engine fueled
Independently from its type, the accuracy of thermody- with bioethanol. The thermodynamic model is a single zone
namic models is affected by both heat transfer models and one, based on the first law of thermodynamics.
wall temperature values, once heat transfer models are part Benajes et al. [18] used two different 0D models to per-
of the solution of the first law of thermodynamics, and wall form global energy balance concerning a 4 cylinder high
temperatures are boundary conditions for the heat transfer speed direct injection turbocharged diesel engine, and evalu-
models. The wall heat transfer can contribute to a significant ate the effect of swirl on the global energy balance of the
heat loss in the overall fuel conversion process, and the in evaluated engine.
cylinder combustion processes and engine operation can also
be affected. In addition, the wall heat flux determines the 2.2 Multi‑zone Models
distribution of thermal stress in the mechanical components
of the engine. Consequently, an accurate calculation of the Multi-zone models are distinguished from zero-dimensional
chamber wall heat transfer is essential to improve the per- models by the inclusion of certain geometrical parameters in
formance of engine simulation, which thereby can be more the basic thermodynamic approach. This usually involves the
helpful in the design of more fuel efficient engines [15]. radius of a thin interface (the flame) separating burned from
Several models have been developed to estimate/calculate unburned gases, resulting in a ‘two-zone’ formulation [14].
the heat fluxes from engine gas flows at intake, exhaust and Pariotis et al. [10] used a quasi-dimensional model as
inside the cylinder to the respective walls. These models are a hybrid one, meaning that it used phenomenological sub-
focused on the detailed modeling of thermal boundary layer models to describe the various processes taking place inside
in the fluid phase near the wall and are mostly based on the combustion chamber. Phenomenological sub-models are
assumption of constant wall temperature. Recent measure- used to describe: heat transfer through the cylinder walls,
ment of engine wall temperatures shows differences from the fuel injection rate, fuel spray penetration, evaporation,
cylinder head hot regions to cylinder head and cylinder liner combustion and pollutant formation. The computational
cold regions as high as 50 K for a single engine operating domain covers the whole cylinder volume and is divided
condition [16]. This way, constant wall temperature is not into computational cells, in order that a series of character-
realistic assumption and can introduce significant errors in istic properties 𝜙 of the flow field is calculated at each cell,
estimating the heat flux on cylinder wall [15] . To achieve by solving conservation equation of each property at each
more accurate calculation of heat fluxes involving different cell. The results for multi-zone model show good agreement
heat transfer phenomena inside the engine, different mod- with CFD 3D results for in-cylinder pressure trace, mean gas
els and approaches have been developed and are the main temperature and heat transfer, along with a good estimation
subject of Sect. 3. of flow field velocities.
Vancoillie et al. [13] published the development and vali-
2.1 Zero‑Dimensional Models dation of a quasi dimensional model for methanol and etha-
nol fueled SI engines. The authors presented the evaluation
Zero-dimensional models (also termed single-zone mod- and comparison of the developed model for different correla-
els) are based on mass and energy balance, applied to the tions of turbulent burning velocity, laminar burning velocity
combustion chamber of a given engine. Those models use and knock prediction sub-model, as those correlations are
a predefined mass burning rate, of which the Wiebe law is implemented in the quasi dimensional power cycle model.
the best known example. This mass burning rate has to be
empirically defined for every engine operating point, on the 2.3 Multi‑dimensional Modelling
basis of prior engine experiments or experience and is not
expressed in terms of physical quantities (e.g. fuel proper- Multidimensional engine combustion simulation has been an
ties, engine geometry, etc.) rendering extrapolation to other active area of research, where advanced physical and chemi-
problematic operating conditions [14]. cal models continue to be developed [15]. Since early 90’s,
Pariotis et al. [10] used a single zone model, based on several papers have been published concerning application
the first law of thermodynamics to resolve the closed cycle of engine CFD 3D modelling to better understand in cylin-
of single cylinder diesel engine. This model is coupled to der phenomena, and propose engine optimization [19–21].
a turbulence k-𝜖 simplified model using the heat transfer Nowadays, different approaches involving engine CFD 3D

13
 L. Fonseca et al.

are used for academic or industry purposes, in order to per- Gas to wall heat transfer is the modelling approach for the
form engine optimization. [3, 22–24] heat exchange between gases and walls in cylinder, intake
di Mare et al. [24] presented the application of LES qual- and exhaust engine regions. Most of the papers in literature
ity criteria to internal combustion engine CFD 3D simula- addressing heat transfer inside internal combustion engines
tions, using a commercial CFD 3D code. refers exclusively to this heat transfer approach. Wall to wall
Zhang [15] implemented into a KIVA-4V code a solid heat transfer approaches are those used to calculate heat
domain heat transfer solver, in order to perform simulta- fluxes inside engine parts (head, piston, valves, block, etc.).
neously in cylinder thermal fluid dynamic calculation and The third modelling approach is dedicated to calculate heat
wall heat transfer. The author proposed a vertical partition flux to coolant and lubricating oil.
scheme for mesh sector, in order to reduce computational The combination of gas to wall, wall to wall and wall to
timing. Other strategy used to reduce computational cost liquid approaches is required to estimate/calculate engine
is to use cylinder symmetry and simulate only in cylinder internal wall surface temperature, inside the cylinder, ports
phenomena, this way only the closed cycle is calculated for and ducts, and specially for valves and piston. For any kind
several cycles, until wall temperature reaches a steady state of engine, using those approaches means working with con-
condition. The equations for thermal fluid dynamics pro- vection-conduction problems, although in diesel engines, it
cesses (mass, momentum and energy conservation, coupled can involve radiation in some cases (very sooting flames).
to several sub-models) are adapted to solve solid phase heat Internal combustion engine heat transfer modeling has
flux at cylinder head, liner and piston. been the subject of earlier review papers [4, 25] and were
deeply discussed in PhD thesis [26].
2.4 Summary of ICE Thermodynamic Modelling Borman and Nishiwaki [4] published a review about
internal combustion engine heat transfer. They described
There are different approaches for engine thermodynamic the state of the art about engine heat transfer modeling at
modeling, being those different models represented in three that time. The authors reviewed the most important correla-
different groups in this paper. The different engine thermo- tion for heat exchange between gases and walls inside the air
dynamic models have in common need for proper heat trans- ducts of an internal combustion engine, on what concerns
fer and wall temperature modelling, as boundary conditions in cylinder heat transfer, prechamber, gas exchange process
for the first law of thermodynamics, in order to offer accurate and intake and exhaust system. Most of the models presented
results for in cylinder phenomena. involves empirical and semi empirical correlations for heat
transfer between gases and walls, although there are some
radiant heat transfer models. There are also some one dimen-
3 ICE Heat Transfer and Wall Temperature sional and multidimensional models described. It was an
Modeling important review at the time, and still nowadays it is used in
the literature [15, 27–31].
The importance of heat transfer for engine design and opti- Although it is used until nowadays, Borman and Nishi-
mization has been highlighted by several authors in litera- waki review addresses specifically gas to wall heat trans-
ture. The reasons are the accurate calculation of performance fer modelling. Several models for this engine heat transfer
parameters, fuel consumption and emissions. Still nowadays approach have been proposed since, but there was also mod-
heat transfer and wall temperature measurements are not els proposed for other engine heat transfer approaches. It is
always an option for internal combustion engine design and the purpose of this section to update the state of the art on
optimization, this way the models for this kind of application the subject of models for heat exchange between gases and
are the only alternative in many situations for estimation/ wall inside internal combustion engines, and also review
calculation of heat transfer and wall temperature. Histori- models for heat transfer between engine parts and from
cally, the first model for engine heat transfer was presented engine parts to lubricating oil and coolant.
by Nusselt (1923), based on experiments in a spherical bomb Most of the correlations that have been developed over
[4]. the past eighty years provide spatially averaged heat-transfer
In this section, the models presented in literature for esti- coefficients [32].
mation/calculation of engine heat transfer and wall tempera- The use of the three heat transfer approaches coupled can
ture are reviewed. These models are organized in three heat be found in some applications, for example in order to per-
transfer modelling approaches as follows: form CHT (conjugate heat transfer) analysis where those
heat transfer approaches are coupled to any engine thermo-
1. Gas to wall heat transfer; dynamic model like the ones described in Sect. 2. Figure 1
2. Wall to wall heat transfer; illustrates the use of three heat transfer approaches coupled
3. Wall to liquid heat transfer. in order to obtain heat transfer and temperature calculations.

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

cylinder are coupled to describe heat transfer at valve stem.

For gas to wall heat transfer at the ports, correlations for
turbulent flow over curved ducts are used.

3.1.1 In Cylinder Heat Transfer Modelling

The engine heat transfer phenomena most widely reported

occurs from hot gases inside combustion chamber towards
engine internal surfaces at cylinder head, cylinder liner, pis-
ton and valve faces. Several authors refers to this heat trans-
fer mode as engine heat transfer only, although in this review
other internal combustion engine heat transfer phenomena
are under discussion. In cylinder heat transfer is among the
most complicated of engineering heat transfer problems,
involving rapid changes in gas temperature, pressure and
velocity, together with local complexities of velocity and
temperature distribution [4]. Most of the common gas to wall
heat transfer models rely on the definition of the wall heat
Fig. 1  Combination of different heat transfer approaches for wall heat
transfer and temperature estimation/calculation. [26] flux q′′ using Newton’s cooling law [36]:
q�� = h(T − Tw ) (1)
with the heat transfer coefficient h, and the difference
Benajes at al. [18] used a lumped conductance model to between gas temperature T and wall temperature Tw [36].
calculate wall temperatures in the chamber and ports along The early models of Nusselt, Eichelberg and Pflaum con-
with the heat rejection repartition to coolant and oil, coupled tain dependencies of h on engine specific parameters like
to a 0D thermodynamic model, in order to perform global cylinder pressure p, cylinder temperature T, and mean piston
energy balance in an 4 cylinder diesel engine. It is also pos- speed vp m [36, 37]. To overcome this weakness, later model
sible to use the three heat transfer approaches coupled to a concepts are based on similarities using non dimensional
thermodynamic model in order to provide boundary condi- properties. The wide spread models of Annand [38] and
tions for engine thermodynamic modelling, specially CFD Woschni [39] are based on the convective part of the heat
3D models. transfer adopted from the boundary layer theory:

3.1 Gas to Wall Heat Transfer Nu = CRea Prb (2)

The Nusselt number Nu characterizes the non dimensional
The engine heat transfer modeling approach most widely heat transfer depending on the Reynolds number Re, which
studied and used is in cylinder gas to wall heat transfer. Nev- covers the flow characteristics, and the physical fluid proper-
ertheless, there are studies for this heat transfer approach ties described by the Prandtl number, Pr [36].
concerning heat transfer between gas fluxes and valves [33], Since Borman and Nishiwaki review, great quantity of
and also concerning heat transfer between gas fluxes and new papers have been published on this subject.
intake and exhaust ports [34, 35]. Once most of the studies Some authors evaluate the most used correlations, and
involves exclusively in cylinder heat transfer, the heat trans- test their accuracy against experimental data on engine heat
fer from the other regions is frequently neglected or even transfer.
forgotten. Although, once the calculation of the ducts sur- Han et al. [40] compared the predictions for instantaneous
faces temperatures is a target for modelling accuracy, those heat transfer coefficient and time averaged heat transfer coef-
regions heat transfer approaches can’t be neglected. ficient calculated using 4 different empirical formulations:
The heat transfer between gas fluxes and walls for each Woschni, Hohemberg, Eichelberg and Yoshida. The authors
surface in contact with in cylinder gases is modeled by the concluded that the time averaged values match the experi-
correlations proposed for in cylinder gas to wall heat trans- mental ones, but the instantaneous values does not present
fer. For other surfaces, like valve stem and seat, correla- good agreement with the experimental ones.
tions for gas flow through different ducts are used. Correla- De Cuyper et al. [41] used a 4 cylinder 1.8 l engine,
tions for heat transfer similar to those for mass flux through equipped with TFG (thin film gauge) sensors to evaluate
valve curtain flow area are used for valve seat, meanwhile empirical heat transfer correlations suggested by Annand
heat transfer for flow over a flat plate and cross flow over a and Woschni. They used only those correlations because

13
 L. Fonseca et al.

those are the most commonly used in commercial software The gas velocity is proportional to a tumble gas velocity
for apparent heat release calculation. The results for instan- dissipation function, fw (𝛼) , which depends on the crank
taneous heat flux show considerable difference between the angle, and a characteristic mean gas velocity v̄ m, respectively
values calculated by heat transfer correlations and the values defined as:
calculated based on the data measured by thin film gauges. 𝛼−𝛼0 m
On the other hand, the values for total heat transfer predicted fw (𝛼) =exp
a( 𝛼
f −𝛼0
)
(6)
by the models are similar to the measured ones.
Other authors propose new correlations, with or without
comparison to the traditional ones. 360𝜂tr ṁ a
v̄ m = (7)
Irimescu et al. [42] developed a correlation for gas to Aeff 𝜌a 𝛥𝛼IVO−IVC
wall heat transfer inside a single cylinder optical engine,
based on thermodynamic and optical measurements. The The characteristic mean gas velocity is calculated based on
correlation is developed considering the effect of the com- the mean air density during intake process, calculated using
bustion process into the flow field after top dead center until a mean gas in cylinder temperature T̄ cyl,int:
half expansion stroke, using the cascade model of kinetic
( ) 𝛾−1
energy transfer. Each eddy is considered half of a jet, emerg- p̄ cyl,int 𝛾
T̄ cyl,int = T̄ int (8)
ing from a tube with diameter L, impinging over a flat plate p̄ int
at a distance of L/2 from the tube’s edge, where L is the
characteristic length taken as the clearance height hc until The model proposed obtained a 70% reduction on the root
half of the expansion stroke. The developed correlation for mean square error, compared to the reference Woschni based
the Nusselt number is: model.
Michl et al. [36] derived and validated a heat transfer cor-
T 0.11
Nu = 0.037Re0.8 Pr0.33 (3) relation for hydrogen engines. The authors have based their
Tw
model in the Nusselt Reynolds relation for turbulent pipe flow
The last term on the right side of the equation is a correc- to calculate the heat transfer coeffient h, and calculated the
tion accounting for the temperature difference of the fluid fluid properties using multizone approach based on a CHEM-
within the boundary layer. Along with this expression, an KIN based stirred reactor analysis of hydrogen combustion
additional term is added to the characteristic velocity, taking reaction.
the combustion process as an addition to the kinetic energy h = C ⋅ Lc ⋅ 𝛺 ⋅ vc ⋅ 𝛯 (9)
of the fluid. The developed model is compared to traditional
where C is a constant and 𝛯 is an explicit term for the aug-
models of Woschni [39] and Annand [38], and all of the
mentation of heat transfer during combustion. In (9) Lc is
models are compared to experimental measurements. The
the characteristic length, 𝛺 is a term to account for property
authors found that the developed model has similar results
characteristics of hydrogen burning mixture, vc is a charac-
to the traditional correlations.
teristic speed using a series of correlations proposed by [36],
Olmeda et al. [1] developed a correlation for estimation
respectively calculated by:
of in cylinder heat transfer coefficient h between gases and
walls, considering the influence of intake air tumble motion Lc =D−0,23 (10)
on this coefficient. The authors proposed the modification
( 𝜌 )0,77
based on a theoretical review about heat transfer coefficient,
𝛺 =k (11)
performed the modification based on CFD 3D simulations, 𝜈
and calibrated the new model based on experimental results
here k is thermal conductivity and 𝜂 is kinematic viscosity,
obtained in a single cylinder 2 stroke diesel engine operating
calculated for hydrogen burning mixture using:
under skip firing condition. The equation proposed by the
authors for the heat transfer coefficient h is: k =yair kair + yH2 kH2 + yb kb (12)

h = CD−0.2 p0.8 Tg−0.55 v0.8 (4)

kair =2, 8610−4 T 0.785 (13)
t

here C is a constant, D is cylinder bore, p and Tg are, respec- kH2 =2, 3210−3 T 0.760 (14)
tively, in cylinder gas pressure and temperature, and vt is a
kb =7, 3610−5 T 1.031 (15)
tumble generated gas velocity, defined as:
Vd TIVC 𝜂 =yair 𝜂air + yH2 𝜂H2 + yb 𝜂b (16)
vt = Ct1 cm + Ct2 v̄ m fw + C2 (p − p0 ) (5)
VIVC PIVC
𝜂air =4, 3910−7 T 0.662 (17)

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

𝜂H2 =2, 2710−7 T 0.646 (18) Nu = 0.023Re0.8 Pr0.4 (21)

On the other hand, for high lift, Nusselt number is calculated
𝜂b =1, 8010−7 T 0.779 (19) using Churchill and Bernstein correlation:
The validation was performed based on experiments on
a single cylinder, 4 valve, direct injection spark ignition 1 1
[ ( )5 ] 5
−4

0.62Re Pr
2 3 Re 8
engine. The authors installed 14 thermocouples in the engine Nu = 0.3 + 1+
[ ( )2 ]4
1
282, 000 (22)
head surface, and 15 thermocouples in the piston surface,
1 + 0.4
3

being every thermocouple 2 μ m thickness. Pr

It has been highlighted in this subsection that the cumu-
lative heat flux predicted by Woschni model, and also other
The last region of the valve modeled for gas to wall heat
traditional models, shows good agreement to experimental val-
transfer is named by the authors stem port region.
ues, in most cases. On the other hand, it has been published by
This region is modeled by the authors only for the open
several authors that instantaneous heat flux values calculated
valve case, once the heat transfer correlation presented takes
by Woschni model, and also other traditional models, do not
into account gas flow condition only.
fit experimental data. It has also been highlighted that some
For this condition, the authors model the valve stem heat
authors propose specific correlations, intending to improve
transfer coefficient h as the sum of that for a flat plate and for
accuracy of traditional model predictions for specific cases,
a cylinder, once the flow through the valve stem usually is
like hydrogen engines [36] and high tumble engines [1].
not either normal or parallel to the valve axis, being gener-
ally a composition of those two extreme conditions:
3.1.2 Valves Heat Transfer Modelling
hstem−port = 𝜖stem−port hcylinder + (1 − 𝜖stem−port )hflatplate (23)
Heat transfer from gas fluxes to valve surfaces is also a com- For hflatplate:
plex phenomena, once different valve surface regions are sub-
mited to different heat transfer phenomena. 1 1
[ ( )5 ] 5
−4

Cerdoun et al. [33] performed a heat transfer analysis for Nu = 0.3 +

0.62Re 2 Pr 3
1+
Re 8

intake and exhaust valves inside internal combustion engines, [ ( )2 ]4

1
282, 000 (24)
considering the heat flow through the valve and the differ-
0.4 3
1 + Pr
ent boundary conditions involved in valves operations. The
authors used the Woschni correlation to estimate the heat
For hcylinder:
transfer coefficient, and calculate the heat flux from the gases
inside the combustion chamber to the valve surface in contact Nu = 0.037Re0.8 Pr0.33 (25)
with the gases inside the cylinder. Heat transfer for valves and from valves must be modeled
For the valve seat heat transfer modelling, the authors have carefully, once exhaust valve has the most severe thermal
considered two main situations: closed valve, which is mod- load and is cooled only by the valve seat. On the other hand,
eled using a thermal resistance model (Sect. 3.2); open valve, intake valve can influence intake charge specific mass.
which is modeled similar to the flow modelling through a
restriction. The mass flow rate through engine valves is mod- 3.1.3 Ports Heat Transfer Modelling
elled as a compressible flow through a restriction, with three
different approaches based on the minimum flow area depend- Shayler et al. [34] have performed heat transfer and tempera-
ing on the valve lift. The Reynolds number for compressible ture measurements in the intake port of an internal combus-
flow through a restriction is calculated using: tion engine, than used the measured values to adjust a heat
( )1 [ [ ( ) 𝛾−1 ]] 2
1 transfer correlation. They compared their adjusted correla-
Cd Dh P0 Pt 𝛾 2𝛾 Pt 𝛾
(20) tion, and verified that it indicates that heat transfer arising
Re = 1−
1
𝜇(RT0 ) 2 p0 𝛾 −1 p0 from gas flow on a firing engine is approximately double that
predicted for steady flow results by Martins and Finlay [43].
If the mass flow rate is modelled this way, the energy flow
Nu = 0.135Re0.713 (26)
rate is modeled similarly, in such a way the authors used
different correlations for Nusselt number depending on the 3.1.4 Wall Functions
valve lift, and consequently on the minimum flow area. At
low and medium valve lift, hydraulic diameter Dh = Permin is
4A
Multidimensional thermodynamic models need an
used, and Nusselt numbers is calculated using Dittus–Boe- approach to model heat transfer from cells close to wall
lter correlation: regions and the walls. This is used where only one cell

13
 L. Fonseca et al.

makes the transition from the main flow to the wall region.

lation for boundary layer, specific to hydrogen burning due to fluid properties term 𝛺
Michl et al. [36] h = C ⋅ Lc ⋅ 𝛺 ⋅ vc ⋅ 𝛯 , where Lc = D−0,23 is the characteristic length, and 𝛺 = k( 𝜌 )0,77 Calculation of heat transfer coefficient h based on non dimensional heat transfer corre-
The equations that represent the boundary layer for this
kind of application are named wall functions.
Fan et al. [27] performed CFD-3D simulations of a
spark ignition engine under motoring conditions, using
detached eddy simulation for turbulence modelling. The
authors used PIV measurements on the center of the com-
bustion chamber of an optical direct injection spark igni-
tion, in order to validate the results of the CFD-3D model.
The results are compared in the center of the cylinder, the
macro scale flow field region, because the measurement in
the region close to the wall is a challenging task due to the
curvature of the optical window.
Ma et al. [32] compared traditional wall functions used
in CFD 3D commercial codes, based on boundary layer
equilibrium conditions, with a wall function developed
using micro particle image velocimetry. The results show
that the equilibrium wall functions perform well the shear
velocity calculation within the viscous sublayer, but show
substantial deficiencies if the matching location is outside
the viscous sublayer.

Observation
Šarić et al. [7] simulated a 90 ◦ edge CFD 3D model
of a disk-shaped spark ignition engine, using the tur-
bulence model k − 𝜁 − f and a hybrid wall heat transfer
model implemented in a commercial code. The authors
compared the hybrid formulations with standard ones, and

−3 0.760
,
𝜈
used measured data available in literature to validate the

takes into account fluid properties k = yair kair + yH2 kH2 + yb kb is air thermal
, kH2 = 2, 3210 T
comparison.

kb = 7, 3610−5 T 1.031 𝜂 = yair 𝜂air + yH2 𝜂H2 + yb 𝜂b is air kinematic viscos-

The heat transfer models used as wall functions are spe-

ity for hydrogen burning, 𝜂air = 4, 3910−7 T 0.662, 𝜂H2 = 2, 2710−7 T 0.646,
cific for gas to wall heat transfer in multidimensional ther-
modynamic models. 0.785

3.1.5 Summary of Gas to Wall Heat Transfer Correlations

conductivity for hydrogen burning, kair = 2, 8610 T −4

Tables 1 and 2 summarize the most recent reviewed cor-

relation related to engine gas to wall heat transfer correla-
tions. Some correlations are proposed based on widely used
Woschni [39] and Annand [38] correlations, for specific
cases of high tumble and hydrogen engines. The correla-
tions used for valve heat transfer, and also port wall heat
Table 1  Gas to wall heat transfer correlations

transfer are summarized.

Along with Tables 1 and 2, it is important to highlight
𝜂b = 1, 8010−7 T 0.779

that the authors in literature who tested traditional gas to

wall heat transfer correlations showed that those correla-
tions do not perform well for instantaneous heat flux values,
Equation

although they perform well for calculation of total heat flux

for combustion analysis.
The group of correlations for gas to wall heat transfer
provides heat transfer calculation from gas fluxes to ducts
In cylinder

surfaces, when evaluating engine behaviour. Only in the case

Author

of intake port the heat transfer tends to occur from duct wall
to gas.

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

Table 2  Gas to wall heat transfer correlations

Olmeda et al. [1] Expression for calculation of heat transfer

t , where vt = Ct1 cm + Ct2 v
V TIVC
h = CD−0.2 p0.8 Tg−0.55 v0.8 ̄ m fw + C2 V d (p − p0 )
𝛼−𝛼0
IVC PIVC
coefficient h, based on Woschni correla-
is a tumble generated gas velocity, f (𝛼) = expa( 𝛼f −𝛼0 ) , v̄ m = tion, specific for engines with increased
m 360𝜂tr ṁ a
w Aeff 𝜌a 𝛥𝛼IVO−IVC
and T̄ cyl,int = T̄ int (
p̄ cyl,int 𝛾−1
) 𝛾 tumble motion
p̄ int

Valve
Cerdoun et al. [33] Re =
Cd Dh P0 P 1
2𝛾
( p t ) 𝛾 [ 𝛾−1 [1 − ( p t )
P 𝛾−1
𝛾 ]] 2
1
1
𝜇(RT0 )2 0 0

Nu = 0.023Re0.8 Pr0.4
1 1
5 −4
0.62Re 2 Pr 3 Re
Nu = 0.3 + 2 1 [1 + ( 282,000 )8 ] 5
[1+( 0.4
Pr
)3 ]4

hstem−port = 𝜖stem−port hcylinder + (1 − 𝜖stem−port )hflatplate

1 1
5 −4
0.62Re 2 Pr 3 Re
Nu = 0.3 + 2 1 [1 + ( 282,000 )8 ] 5
[1+( 0.4
Pr
)3 ]4

Nu = 0.037Re0.8 Pr0.33
Ports
Shayler et al. [34] Nu = 0.135Re0.713
Nu = CRea Prb
Nu = CRea Prb

3.2 Wall to Wall Heat Transfer number of nodes much smaller than multidimensional ones,
thus reducing computational cost. Nevertheless, it provides
The wall to wall heat transfer modelling is the approach for results of temperature distribution for few points at each
modelling heat transfer between engine parts, in the case of part, and makes it more feasible to accomplish Biot number
global and nodal models. It is also used to describe the heat restriction for lumped capacitance assumption.
transfer inside one specific engine part, for nodal and mul-
tidimensional models. This approach deals with conduction 3.2.1 Global Models for Wall to Wall Engine Heat Transfer
and contact resistance problems.
Borman and Nishiwaki didn’t mentioned any of those Concise heat transfer and wall temperature models, simi-
models. Other authors have published reviews about inter- lar to the one in Fig. 2, can be accomplished through the
nal combustion engine heat transfer, like Finol and Robinson assumption that the Biot number is much smaller than unit
[25] and Romero, in his PhD thesis [26], but neither have for the evaluated engine parts. The simplified model can be
mentioned this engine heat transfer modelling approach. So, used to estimate engine part temperature and heat transfer,
this is the first time this approach is reviewed.
Global models resolve the total heat transfer through each
engine part based on the lumped capacitance assumption.
This may be a reasonable assumption if Biot number is much
smaller than unit, which is not true for several applications.
Multidimensional models, on the other hand, resolve heat
transfer and temperature distribution at each engine part
under modelling. The use of finite element method on a
mesh implemented over each part makes it possible to solve
the transient heat diffusion equation for 2 or 3 dimensional
problems. This solution is the most complete one, although
it implies a great increase in computational time.
Nodal models are an intermediate solution for this trade-
off between accuracy of solution and computational cost.
They also use the lumped capacitance assumption, but for
the several nodes inside each engine part, instead of one
Fig. 2  Example of global heat transfer and wall temperature model
complete part. This division makes it possible to solve a
proposed by Torregrosa et al. [44]

13
 L. Fonseca et al.

although there is only one temperature value for each part,

which is the main drawback for global models.
Baker and Assanis [45] developed a thermal resistance
model to be used coupled to a thermodynamic model of a
diesel engine. The authors used a 1D thermal resistance net-
work to obtain boundary conditions to a finite element model
of the cylinder liner and piston. The results of the FEM were
used to calculate a 2D thermal resistance model. The values
calculated for thermal resistances were cycle averaged ones,
and only three nodes were present in this model, so any engine
part would have a single node for temperature calculation.
Torregrosa et al. [44] developed a concise thermal resist-
ance model for wall temperature prediction in diesel engines Fig. 3  Example of nodal division for cylinder liner heat transfer and
wall temperature modelling proposed by Torregrosa et al. [46]
with piston cooling. The authors developed a three nodes
model to calculate the heat flux from the hot gases inside
the cylinder, until the lubricating oil and coolant, modeling considerably reduced. The thermal nodes are connected by
the thermal conductance’s between the three nodes and the a network of heat transfer elements, which can be treated as
boundary conditions. thermal resistances or thermal conductance’s.
The wall to wall heat transfer has been presented by the Shayler et al. [47] published a thermal network analysis of
authors as follows: a single cylinder engine, using a lumped capacitance model
� � −1 for considering a thermal network inside a 1.1 L single cylin-
⎛ D+2.elin ⎞ der engine. The authors proposed a 23 nodes model including
⎜ ln ⎟
D 1 engine head, cylinder liner, cylinder block and piston, solving
Klin−cool =⎜ + ⎟ (27)
⎜ 2.𝜋.S.k Al A .h
lin,ext lin−cool ⎟ the transient discrete form of the diffusion equation for the 23
⎝ ⎠
nodes group. The gases inside combustion chamber, lubricating
oil and coolant was used as boundary conditions for this model.
� � −1 Bohac et al. [8] developed a global model for characteri-
⎛ D ⎞
⎜ ln 0.5D 1 ⎟ zation of thermal behaviour of engine parts inside an internal
Kpis−lin =⎜ + ⎟ (28) combustion engine. The authors used a lumped capacitance,
⎜ 2.𝜋.h .k
pc Al 𝜋.D.h .h
pc pis−lin ⎟
⎝ ⎠ lumped resistance model to describe the heat flux through
a 1.9L 4 cylinder production engine, in transient operation.
Along with thermal capacitance, described for all nodes
( )−1 considering only the mass and specify heat of the node, the
ehead 1
Khead−cool = + ( ) authors described thermal resistances for the following phe-
kAl .Ahead hhead−cool . Ahead + Arun
nomena: axial conduction, radial conduction, convection,
(29) radiation (linearized), in flow and out flow.
For each of the above expressions, the wall to wall heat
transfer is modeled as thermal contact resistance, in carte- L
R= (30)
sian and cylindrical coordinates. kA
The model has been validate against measurements in
two diesel engines with 1,6 and 2,2 L displacement. One of r
ln( r2 )
the engines was equipped with 23 thermocouples K type in R= 1 (31)
2𝜋Hk
the cylinder liner, 16 thermocouples K type in the head. The
other was equiped with 22 thermocouples in cylinder liner
and 18 thermocouples in cilinder head. 1
R=
hAs (32)

3.2.2 Nodal Models for Wall to Wall Engine Heat Transfer

T 2 − T1
The division of engine parts into small number of nodes,
R= (33)
𝜖𝜎F21 As (T24 − T14 )
illustrated in Fig. 3, makes nodal models more complete
than global ones. Along with it, the use of the lumped
1
capacitance assumption makes nodal models more versatile R= (34)
min cp,in
than multidimensional ones, once the computational cost is

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

1 Kli = hli Ali (43)

R=
mout cp,out (35)
Ci = mi cv (44)
The authors described a model that took into account heat
transfer inside the cylinder, intake and exhaust ports, exhaust 3.2.3 Multidimensional Models for Wall to Wall Engine
manifold and radiation shield, cylinder head, liner, piston, Heat Transfer
valves, coolant and oil ducts inside head and block, along
with heat exchangers for coolant and oil. The coolant was The multidimensional analysis for heat transfer inside
a 50/50 mixture of ethylene glycol and water. The authors engine parts is performed by the solution of the heat diffu-
take several conclusions, among which can be highlighted: sion equation:
combustion must be modeled accurately in order to predict
engine heat fluxes and temperatures; nucleate boiling is a ∇(k𝛥T) = 𝜌cp
𝜕T
(45)
very efficient heat transfer mechanism, and is essential if 𝜕t
wall temperatures must be modeled accurately. Different options for the solution of this equations are used
Jarrier et al. [48] performed the comparison between in the literature, with the main objective to calculate tem-
experimental and numerical values for a DI diesel engine perature distribution and heat fluxes through engine parts.
during warm up. For the numerical values, the authors used a Baker and Assanis [45] used an axisymmetric 2D finite-
53 nodes model representing metallic masses, coolant, oil and element model to resolve the heat flux through the combus-
gases. The nodes are connected by thermal conductance’s for tion chamber components of a reciprocating engine. The
conduction, convection, radiation, fluid flow, and each node transient heat conduction equation has been solved for a 2D
may have thermal capacitance. The authors use heat source axisymmetric domain, using a Fortran code, in such ways
from combustion gases and from friction inside the cylinder, each component temperature is resolved separately during
meanwhile the heat sinks are lubricating oil and coolant. each time step. This model was coupled to a 0D thermody-
kA namic model, using heat transfer correlations as boundary
G= (36) conditions for heat transfer between combustion chamber
d
parts and combustion gases, lubricating oil and coolant liq-
uid. The results for the finite element models show piston
G = hA (37)
temperatures around 900 ◦C varying in space and time, cyl-
inder liner temperature difference greater than 100 ◦C from
G = 𝜎𝜖1 𝜖2F1∕2 As (T12 + T22 )(T1 + T2 ) (38)
its top to a distance 10 mm below cylinder head. The results
for heat transfer show peak heat transfer to the piston crown
G = mC
̇ p (39) around 20kW, and peak heat transfer to cylinder liner around
40kW. The authors also discussed the coupling of this 2D
MCp (40)
heat transfer model to a 0D thermodynamic model, and
Torregrosa et al. [46] presented a nodal model for estimation the possible coupling to an engine CFD model, in order to
and calculation of diesel engine wall temperature. The model obtain spatially as well as temporally resolved gas boundary
presented had 95 nodes for cylinder liner, piston and cylinder conditions.
head, along with 5 nodes for fluid boundary conditions. The Jafari and Hannani [49] performed an iterative analysis
thermal model included thermal capacitance of each node, between KIVA II CFD 3D code and a finite element heat
and also thermal resistances connecting neighbour nodes. conduction code, in order to estimate the effect of fuel and
The boundary conditions are the hot gases inside the cyl- engine operating characteristics on engine heat losses. The
inder as a hot sink, and the lubricating oil and coolant as authors used an axisymmetric finite element model in order
heat sinks. to calculate the heat flux from the in cylinder gases to the
coolant. No specific modelling is presented for the heat
i
Tt+𝛥t − Tti ∑
Ci =
j i
Kij (Tt+𝛥t − Tt+𝛥t ) transfer to the oil, or water, just wall temperatures are used
𝛥t j (41) for most of the surfaces boundary conditions.
∑ ∑ Trujillo et al. [50] presented a methodology for estima-
l i
+ Qk→i + (T − Tt+𝛥t )
k l tion of cylinder inner surface temperature, using a finite ele-
ment model. The engine used is a single cylinder, air cooled,
direct injection diesel engine. The finite element model is
kAij applied to the cylinder heat transfer, meanwhile the in cylin-
Kij = (42) der gas temperature and the cooling air temperature and heat
dij
transfer coefficient are the boundary conditions applied. For

13
 L. Fonseca et al.

the heat supplied by the in cylinder gas, the authors used a well the contact conductance coefficient and exhaust valve
heat flux distribution in order to model the effect of the heat temperature. The authors present the results of training the
transfer from the gases to the cylinder walls, thus it wasn’t neural networks, and conclude that exhaust valve tempera-
used any thermodynamic model. ture and contact conductance coefficient can be estimated
Trujillo et al. [51] presented a methodology for estima- well with the neural networks algorithm presented, although
tion of head inner surface temperature, using a finite element no validation procedure has been presented.
model. The methodology is similar to the one presented at
[50], in such a manner that the engine used is the same single
3.2.5 Summarize Wall to Wall Heat Transfer
cylinder, air cooled, direct injection diesel engine. A finite
element model is used to resolve the heat flux through the
Tables 3 and 4 summarize wall to wall heat transfer correla-
head of the engine, with boundary conditions considered
tions, specially those used by nodal models.
as the heat flux from the gases inside the cylinder, from the
The selection of each option for wall to wall heat transfer
intake and exhaust gas fluxes, and also the cooling air flux.
approach depends on the trade off between accuracy and
Once again no thermodynamic model is used to calculate
resolution of the results and computational demand gener-
the heat rejected from the in cylinder gases to the cylinder
ated by the set of equations and boundary conditions. For
head, and the authors use a heat flux distribution to model
applications where fast response is mandatory, global heat
this heat source.
transfer models are the only options, and this way they must
Cerit and Coban [52] published a finite element analysis
be calibrated for each application. On the other hand, for
of temperature and thermal stresses analyses of a ceramic-
applications where high resolution of temperature distribu-
coated aluminum alloy piston used in a diesel engine. The
tion is mandatory, FEM models are the only option, and the
authors used finite element analysis, with heat transfer coef-
computational time for this model must be accounted on the
ficient used as boundary conditions, in order to model heat
planning time.
transfer from gases to piston surface.
It is clear in this section the importance of lumped capac-
Lu et al. [53] presented a finite element analysis of ther-
itance assumption, this way it is important to verify Biot
mal stress analysis on a piston engine, using a commercial
number in the model results, in order to verify model valid-
finite element code. The authors used different heat transfer
ity and accuracy. Among the models shown in this review,
coefficient for great quantity of piston surfaces, and used
nodal models have a reasonable trade-off between resolution
a correction methodology for boundary conditions using
of temperature distribution inside the engine and computa-
experimental measured temperature values.
tional demand.
3.2.4 Other Wall to Wall Heat Transfer Models
3.3 Wall to Liquid Heat Transfer
Goudarzi et al. [54] presented an application of artificial neu-
ral networks in the estimation of exhaust valve temperature The wall to liquid heat transfer is the approach for modelling
and thermal contact conductance in an internal combustion heat transfer between engine parts and cooling liquid. It is
engine. The authors presented a transient set of equations also the approach for modelling heat transfer between engine
to solve the problem of intermittent heat transfer between parts and lubricating oil. This heat transfer approach does
exhaust valve and valve seat, considering the only way to not account for air cooled engines. Borman and Nishiwaki
remove heat from exhaust valve is the valve seat contact. didn’t mentioned this heat transfer modelling approach, on
The set of equations solves a transient heat transfer prob- the other hand, Romero [26], on his PhD thesis, presented
lem, with a transient heat flux equation applied as bound- a review concerning heat transfer to the cooling liquid and
ary conditions, to calculate transient heat flux coefficient lubricating oil. Once this is not the first time this subject is
in valve contact, and also valve temperature. The authors reviewed, in this section the main correlations for both heat
concluded that the artificial neural network can estimate transfer modelling to coolant and oil are presented.

Table 3  Wall to wall heat transfer correlation: global models

Torregrosa et al. [44] ln(

D+2.elin
) 1 Series thermal conductance for cylinder wall and cylinder to coolant heat
Klin−cool = ( D
2.𝜋.S.kAl
+ Alin,ext .hlin−cool
)−1 transfer

ln( D
) 1 Series thermal conductance for piston center to piston external border and
Kpis−lin = ( 2.𝜋.h0.5D.k + )−1
pc Al 𝜋.D.hpc .hpis−lin piston to cylinder liner heat transfer

Khead−cool = ( k
ehead
+ 1
)−1 Series thermal conductance for head wall and head to coolant heat transfer
Al .Ahead hhead−cool .(Ahead +Arun )

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

Table 4  Wall to wall heat transfer correlation: nodal models

Bohac et al. [8] R= L Thermal resistance for conduction nodes in Carte-

sian coordinates
kA

Thermal resistance for conduction nodes in cylin-

r
ln( r2 )
R= 2𝜋Hk
1
drical coordinates
R= 1 Thermal resistance for convection nodes
hAs

R=
T2 −T1 Thermal resistance for radiation nodes
𝜖𝜎F21 As (T24 −T14 )

R= 1 Thermal resistance for intake flow nodes

min cp,in

R= 1 Thermal resistance for out flow nodes

mout cp,out
Jarrier et al. [48] G= kA Thermal conductance for conduction nodes in
d
Cartesian coordinates
G = hA Thermal conductance for convection nodes
G = 𝜎𝜖1 𝜖2F1∕2 As (T12 + T22 )(T1 + T2 ) Thermal conductance for radiation nodes
G = mC
̇ p Thermal conductance for intake and out flow
nodes
Torregrosa et al. [46] C Tt+𝛥t
i
−Tti ∑ ∑ ∑ Thermal balance for each node of the model,
);
j i i
= j Kij (Tt+𝛥t − Tt+𝛥t ) + k Qk→i + il Kil (T l − Tt+𝛥t
i 𝛥t taking into account heat capacity, cartesian and
kAij
where Kij = ; Kli = hli Ali ; Ci = mi cv cylindrical conduction, convection
dij

3.3.1 Wall to Coolant Heat Transfer Torregrosa et al. [44] developed correlations for thermal
conductance between cylinder liner and water (27), and
Wall to coolant heat transfer models generally use Nu-Re cor- also for cylinder head and water (28), using thermal resist-
relations for liquid turbulent flow. There are different configu- ance theory. For the cylinder liner, the authors considered a
rations for the coolant galleries inside engine block and head, series thermal circuit between conduction through cylinder
most of the coolant flow enters the block, flowing through the liner wall and convection from cylinder to coolant, in such a
cylinders and head, and leaving the engine on the head. The manner the convective coefficient has been obtained experi-
heat transfer through the cylinders can be approximated as the mentally. For the cylinder head, the authors considered a
flow through a bank of cylinders, although the heat transfer series thermal circuit between conduction through cylinder
from the engine head to the coolant cannot. Whatever is the head wall and convection from head to coolant, in such a
approach, nucleate boiling must be accounted for in the heat manner the convective coefficient has been adjusted from
transfer coefficient estimation, once it is a very important heat experimental data.
transfer mechanism for both engine head and block. Torregrosa et al. [46] used a modified Dittus–Boelter
Shayler et al. [47] used the method proposed by Finlay et al. correlation for calculation of the heat transfer coefficient
[55] , which employs Chen correlation [56] for nucleate boil- between liner and coolant, and also for heat transfer between
ing and Dittus–Boelter correlation for forced convection. head and coolant.
Bohac et al. [8] calculated a heat transfer coefficient h
Nū D = 0.023Re0.8 (49)
0.4
between coolant gallery walls and coolant, by a sum of a con- D Pr

vective heat trasnfer coefficient hconv using Grimson’s Nusselt The cooling system has been optimized in past decades by,
correlation for flow past a bank of cylinders, and a nucleate among other strategies, subcooled boiling heat transfer strat-
boiling heat transfer coefficient hnuc.boiling inside cooling galler- egies. This optimization lead to numerous benefits includ-
ies in contact with hot surfaces proposed by Chen [56]. ing reduced fuel consumption, lower emissions, increased
power output and improved vehicle cabin comfort [57, 58].
h = hconv + hnuc.boiling (46) The application of subcooled boiling heat transfer concept
to internal combustion engine cooling is the subject of some
1
Nū D = 1.13C1 C2 Rem
D
Pr 3 (47) investigation in literature [57, 59–61], and there is also some
review on this subject [58, 62].
( ) Li et al. [63] proposed a modified model with adjustable
kf0.79 c0.45 𝜌0.49
hnuc.boiling = 0.0012
p,f
(
p,f
) F (48) parameters for subcooled boiling heat transfer applications
𝜎 0.5 𝜇𝜙0.29 𝛥h𝜌0.24 in CFD 3D codes. Several modifications are proposed by the
authors on the basic equations in order to adapt them to CFD

13
 L. Fonseca et al.

3D analysis in geometrically complex cooling galleries of IC Factor F is and enhancement factor related to the ratio of
engines. Once these equations are specific to CFD 3D appli- the two-phase Reynolds number Re2𝛷 and the liquid phase
cations, they won’t be described here for sake of brevity. Reynolds number Rel.
Torregrosa et al. [64] performed experiments on a heated
Re2𝛷 0.8
duct to asses the validity of a Chen-type heat transfer cor- F=( ) (57)
Rel
relation for subcooled flow boiling applications in internal
combustion engines. The correlation proposed by Chen [56] In (50), factor S stands for the suppression factor, which
(50) is used to calculate the total heat flux qw as: accounts for the observed decrease in nucleate boiling activ-
ity when the flow velocity increases, and is evaluated by:
qw = qfc + qnb = 𝛷qfc + Sqnb (50)
Here, qfc stands for the forced convection and qnb for the S=
1
<1 (58)
nucleate boiling heat fluxes, which are calculated, respec- 1 + 2.53x10−6 Re1.17
2𝛷
tively by (51) and (52):
The authors noticed from the results obtained that the refer-
qfc =hfc (Tw − Tb ) (51) ence model were in good agreement with experiment for
wall temperatures below 135 ◦ C, and above this value of wall
q̃ nb =hnb (Tw − Ts ) (52) temperature there was a considerable discrepancy. They pro-
where Tw is the wall temperature, Tb is the bulk fluid temper- posed an adjustment parameter 𝛹 , so the modified equation
ature and Ts is the saturation temperature. The heat transfer for the suppression factor S is:
coefficient for forced convection, qfc , is computed using a
1
Dittus–Boelter equation modified by Robinson et al. S=𝛹 <1 (59)
1 + 2.53x10−6 Re1.17
2𝛷
( )0.14 ( )
hfc = 0.023Re0.8 0.4 𝜇b kl
(53) The adjustment parameter 𝛹 was evaluated as a function of
l Prl 𝜇w Dh Prandtl number, and is proposed as [64]:

where kl is the thermal conductivity of the liquid phase, Dh is 𝛹 = 0.104Prl − 0.4 (60)
the hydraulic diameter, 𝜇b and 𝜇w are the dynamic viscosity The authors concluded that for wall temperatures below
evaluated at the wall and bulk temperatures, respectively, 135 ◦ C, the reference correlation is in good agreement with
and Rel and Prl are the Reynols and Prandtl numbers for the experimental measurements. For temperatures above that
liquid phase: value, they proposed the adjustment parameter 𝛹 for the
𝜇l cp,l suppression factor, as a function of the Prandtl number.
𝜌l ub Dh
Rel = , Prl = (54) Mehdipour et al. [65] presented a CFD 3D simulation
𝜇l kl
of water flow inside a conventional channel, taking into
where ub is the bulk flow velocity and cp,l , 𝜌l and 𝜇l are the account boiling and bubble flow into a mixture flow two
specific heat, density dynamic viscosity of the fluid, respec- phase model, in order to represent flow inside the channels
tively, evaluated at its bulk temperature. of an internal combustion chamber.
The heat transfer coefficient for nucleate boiling, hnb , is Modeling adequately heat transfer from engine walls to
calculated using a correlation proposed by Forster and Zuber coolant is an important task in order to quantify the heat
for pool boiling heat transfer: losses properly. Current trends at engine design are at using
each time more intensively nucleate boiling and subcooled
kl0.79 c0.45
p,l
𝜌0.49
l boiling heat transfer as strategies to reduce heat losses and
hnb = 0.00122 𝛥Ts0.25 𝛥p0.25
s (55) maintain engine temperature under control. This way, wall
𝜎 0.5 𝜇l0.29 llg
0.24 0.24
𝜌g
to coolant heat transfer models must take into account sub-
cooled boiling along with convective heat transfer and low
where llg is the latent heat of vaporization, 𝜎 is the surface
coolant mass flow rate in order to predict adequately coolant
tension, 𝜌g is the density of the vapour phase, 𝛥Ts = Tw − Ts
system performance parameters and wall temperatures.
is the wall superheat and 𝛥p = ps (Tw ) − ps (Ts ) is the cor-
responding saturation pressure difference.
(50) is modified by the parameters 𝛷 and F, where the
3.3.2 Wall to Oil Heat Transfer
first one is a correction factor accounting for the influence
of microconvective effects associated with bubble agitation:
Wall to oil heat transfer is the only cooling option for piston,
( )0.444 along with its lubricating function. This way, the correct
Prl + 1
𝛷= F>1 (56) wall to oil heat transfer modelling is mandatory for piston
2 temperature estimation.

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

Bohac et al. [8] used literature values for the convective heat 4𝜋

4𝜋 ∫0
1
transfer coefficient between oil and piston, in such a manner Nu = [aReb Prc ]d𝛼 (63)
that different piston cooling strategies (splash, forced or jet)
would have different heat transfer coefficient ranges. Once the where Re and Pr are Reynolds and Prandtl number, respec-
coefficient was measured in an specific engine speed, an equa- tively, and a, b and c are constants adjusted for each piston
tion for correction of the convective coefficient concerning position. The values for those constants are available at [66]
engine speed effect is proposed. The authors do not mention for different piston position and oil flow regime.
any correlation for heat transfer between cylinder liner and oil. Liu et al. [67] developed an area-average heat transfer
Torregrosa et al. [44] developed correlations for the thermal correlation to be employed for numerical evaluation of oil-
conductance between piston and oil and liner and oil, adjust- jet cooling of pistons inside internal combustion engines.
ing the proposed correlations based on experimental results in
two diesel engines. For the conductance between piston and ( )0.089 ( )−1.398 ( )0.312
Z D 𝜇w
oil Kpis−oil , the authors concluded it is a function of engine Nuavg = 13.932R0.596
e,O Pr
0.239
d0 d0 𝜇aw
speed, and the proposed correlation takes it into account by (64)
the mean piston speed. The authors conducted a series of experiments on a produc-
tion engine piston, using thermal resistances and insulation
�
Kpis−oil = 𝜋 2 .Dgal .(dgal )m .Cgal .(Sp )m (61) system, associated with an oil jet nozzle and an oil circuit, in
order to represent different engine operation conditions. The
where Dgal is the diameter of the oil gallery built into the pis- authors adjusted a Nu-Re correlation, based on the nozzle
ton, dgal is the internal diameter of the oil gallery, similar to exit diameter, to fit all of the data obtained.
the diameter of the piston, m = 0.73 is the exponent adjusted Peng et al. [68] used a CFD 3D model for the oil gallery,
by experimental values. coupled to a finite element model of the piston, to calculate
On the other hand, the authors modeled the conductance the flow and heat transfer of nanofluids inside the oil gallery
between liner and oil using a standard convective correlation, of a diesel engine piston. The nanofluid used is a mixture of
and adjusted the convective coefficient to fit the experimental oil with copper and diamond particles. The nanofluid flow
data. is simulated using a volume of fluid analysis, to take into
Klin−oil = 𝜋.D.S.hlin−oil (62) account the fraction of oil gallery not filled by oil due to the
where D and S are cylinder bore and stroke, respectively, and mechanism of filling this gallery.
hlin−oil = 870 mW2 K is adjusted by experimental values.
The authors concluded, after fitting the model and adjust- 3.3.3 Summarize Wall to Liquid Heat Transfer
ing its constants, that the conductance between piston and oil
depends on piston speed, all of the other conductance’s evalu- Table 5 summarizes wall to coolant and wall to oil heat
ated has shown to be constant. transfer.
Torregrosa et al. [46] used the same correlations as
described by Torregrosa et al. [44] in order to model heat 3.4 Wall Temperature Calculation
transfer between piston and oil, and also cylinder liner and oil.
Torregrosa et al. [66] proposed a correlation for the film The temperature calculation of surfaces inside internal com-
coefficient related to the convective heat transfer through the bustion engines is the main purpose for using different heat
oil gallery in a real life piston geometry of a Diesel automotive transfer models coupled. Several papers in literature presents
engine. The authors used a test bench composed of two oil cir- wall temperature calculation, some of them present calcula-
cuits: one for oil heated until 300 ◦ C by electrical resistances, tion of wall temperature for different surfaces, meanwhile
used to heat piston surface; another one for oil at tempera- other ones present specific engine parts wall temperature
tures below 100 ◦ C with a system to control temperature, used calculation.
to cool down the piston. Piston temperature is measured by
thermal resistances at 6 points, and the piston is divided in 12 Table 5  Wall to liquid heat transfer
nodes, using the 6 measured points and a lumped capacitance Author Equation Observation
model to calculate the temperature of the nodes which doesn’t
have a thermal resistance. Coolant heat transfer
The results obtained for the Nusselt number showed a Bohac et al. [8] h = hconv + hnuc.boiling
dependence on the piston position during the cycle, and also
1
Nū D = 1.13C1 C2 Rem
D
Pr 3
on the regime of the cooling oil flow. The correlation proposed k0.79 c0.45 𝜌0.49
hnuc.boiling = 0.0012( 𝜎 0.5f 𝜇0.29p,f(𝛥h𝜌p,f0.24 ) )F
for the cycle average Nusselt number Nu is in the form of (63): 𝜙

13
 L. Fonseca et al.

Torregrosa et al. [46] presented a nodal model for wall surface temperature, is also named CHT—Conjugated Heat
temperature calculation of the surfaces inside combustion Transfer.
chamber, at the valves and ports of a single cylinder, diesel Kikusato et al. [70] performed a conjugate heat transfer
engine. The model uses the lumped capacitance assumption analysis between a CFD 3D engine thermodynamic code and
for a series of nodes inside cylinder head, ports, cylinder a finite volume model for heat transfer calculation between
liner and piston, in order to resolve heat transfer equation in cylinder gases and coolant. In this model, a closed cycle
for all of the nodes. The model is validated using a series analysis in a single cylinder diesel engine is performed for
of thermocouples installed in the cylinder and head of the the thermodynamic point of view, using KIVA II CFD 3D
engine, in order to evaluate the temperature distribution cal- code, coupled to a chemical kinetics program to solve com-
culated by the model. bustion and a one dimensional transient heat transfer model.
Payri et al. [69] presented a global energy balance model, For the wall heat transfer, a one dimensional transient
in order to provide real time information about engine wall finite volume heat transfer model is used. The heat transfer
temperature for experimental engine investigations. The between gases and walls is resolved using a law-of-the-wall
model uses experimental data measured in a 4 cylinder DI model, and the heat transfer from walls to coolant is calcu-
diesel engine, 1 valve per cylinder, 1,6 L displaced volume, lated using a modified Dittus–Boelter correlation for the cyl-
in order to measure several quantities and perform an exter- inder and liner, where there is coolant flow. For the piston,
nal global energy balance. For the internal global energy the Woschni correlation is used to calculate the coolant side
balance, the authors used several models for heat transfer heat transfer coefficient, which is a procedure substantially
between gases and walls for cylinder and ports, heat transfer different than used by most authors in literature.
inside each engine part, heat transfer to oil and coolant, heat The authors presented validation of the model, comparing
transfer to turbocharger, and several other heat fluxes that in cylinder pressure, heat release rate, and cylinder head wall
occur during engine operation. temperature surface traces for 1200 rpm, at 25% and 50%
Cerdoun et al. [33] performed the calculation of exhaust load. The pressure, heat release and wall surface tempera-
valve temperature distribution using a group of models for ture traces are in good agreement for the cases presented as
estimate heat transfer coefficients over several valve surface validation.
regions. For valve face, Woschni correlation was used once The authors presented results for distribution of wall sur-
this face is in contact with in cylinder gases. For valve seat, face temperature, heat flux and heat transfer coefficient. The
three heat transfer correlations are used in a similar manner changes in wall temperatures locally affects heat fluxes and
to the one used to describe mass flow through the valve dur- increases heat losses, such as those presented for regions
ing open valve time, and thermal contact resistance is used where wall impingement occurred and squish flow regions.
during closed valve time. The valve temperature distribution As a conclusion, the authors state that wall impingement
is calculated using finite element model. must be avoided, in order to reduce heat losses and increase
Lu et al. [53] performed a finite element analysis for cal- thermal efficiency.
culation of steady state and transient temperature distribu- Martín et al. [71] performed a conjugate heat transfer
tion in a diesel engine piston, along with thermal stresses. analysis, using a 0D thermodynamic model coupled to a
The authors used a great quantity of boundary conditions nodal lumped capacitance heat transfer model, in order to
for heat transfer on different regions of piston surface, from perform thermal analysis of a diesel engine operating with
different positions of piston bowl until valve rings, pass- diesel gasoline dual fuel combustion mode under a reactivity
ing through oil galleries. The calculation of the authors is controlled compression ignition approach.
restricted to piston temperatures, and there is no thermo- Zhang [15] performed a conjugate heat transfer analy-
dynamic model considered, once the thermodynamic pro- sis in a closed cycle simulation of a DI diesel engine. The
cess is represented by the boundary conditions over piston CFD-3D code KIVA-4V was used to solve the equations for
surface. conservation of mass, momentum and energy, coupled to
There are several papers in literature presenting wall tem- several models for turbulence, combustion, spray, etc. The
perature calculation, most of them present the calculation equations of the CFD-3D code was also used to solve the
of the wall temperature without coupling the heat transfer heat flux through the solid phase, concerning cylinder head,
models to a thermodynamic one, should it be a zero dimen- cylinder liner and piston top.
sional, multizone or multidimensional one. Benajes et al. [18] performed a conjugate heat transfer
analysis, using a 0D thermodynamic model coupled to a
3.5 CHT—Conjugate Heat Transfer nodal lumped capacitance heat transfer model, in order to
evaluate the effect of swirl ratio on the global energy balance
The use of coupled thermodynamic and heat transfer mod- of a high speed direct injection diesel engine. The authors
els, in order to predict simultaneously engine behaviour and performed experiments in a 4 cylinder, turbocharged diesel

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

engine, in order to perform external global energy balance, local flash buoyancy occurrence close to combustion cham-
and used the coupled thermodynamic and heat transfer mod- ber both in engine head and block. Also, correlations for
els to evaluate the internal global energy balance. heat transfer between lubricating oil and engine parts, like
Broatch et al. [72] proposed an alternative methodology cylinder liner and piston, are reviewed.
to perform conjugate heat transfer at engine CFD 3D simu- The use of all those correlations for different heat transfer
lations, instead of solving full fluid and solid domains dur- approaches are required for engine wall temperature calcu-
ing several engine cycles. The authors propose to perform lation for any specific application, but specially for CHT
a previous simulation of the combustion process using con- analysis coupled to phenomenological or multidimensional
stant wall temperature as boundary conditions, followed by a thermodynamic models.
simulation of the heat transfer process using the rate of heat
release obtained from the combustion process as boundary Acknowledgements The author Leonardo Fonseca acknowledges
CAPES (Coordination for the Improvement of Higher Education Per-
condition. The authors use CONVERGE CFD to perform sonnel) for the scholarship from the program “CAPES - DEMANDA
both full fluid and solid coupled simulations, and also decou- SOCIAL”, PhD level.
pled combustion and rate of heat release simulation. The
results obtained by the authors using both methodologies Funding This study was partially funded by “CAPES - DEMANDA
are in good agreement with GT Power data for the engine SOCIAL” Ph.D. level scholarship, from CAPES (Coordination for the
Improvement of Higher Education Personnel).
evaluated, and the authors induced that a reduction between
38 and 50% by engine cycle was gained.
Many authors present CHT analysis involving CFD 3D
Compliance with Ethical Standards
engine simulations, although every engine model where Conflict of interest The authors declare that they have no conflict of
thermodynamic and heat transfer are coupled to obtain a interest.
solution that satisfy both conditions is a CHT model.

References
4 Summary
1. Olmeda P, Martín J, Novella R, Carreño R (2015) An adapted
A review about internal combustion engine heat transfer and heat transfer model for engines with tumble motion. Appl Energy
wall temperature modelling is presented in this paper. Three 158:190–202. https​://doi.org/10.1016/j.apene​rgy.2015.08.051
2. Broekaert S, Demuynck J, De Cuyper T, De Paepe M, Verhelst
heat transfer approaches are considered for this review: gas
Sebastian (2016) Heat transfer in premixed spark ignition engines
to wall, wall to wall and wall to liquid heat transfer. part i: identification of the factors influencing heat transfer. Energy
For gas to wall heat transfer approach, the evaluation of 116:380–391. https​://doi.org/10.1016/j.energ​y.2016.08.065
most widely used correlation is presented. The literature 3. Kosmadakis GM, Pariotis EG, Rakopoulos CD (2013) Heat trans-
fer and crevice flow in a hydrogen-fueled spark-ignition engine:
reviewed agree that traditional correlation predictions for
effect on the engine performance and no exhaust emissions. Int J
total heat flux are in good agreement with experiments, but Hydrog Energy 38(18):7477–7489. https:​ //doi.org/10.1016/j.ijhyd​
lacks in accuracy on instantaneous heat fluxes. New cor- ene.2013.03.129
relations are presented for specific cases, like high tumble 4. Borman G, Nishiwaki K (1987) Internal-combustion engine
heat transfer. Prog Energy Combust Sci 13(1):1–46. https​://doi.
engines or specific fuel engines like hydrogen ones. The new
org/10.1016/0360-1285(87)90005​-0
correlations are based on traditional ones, and show better 5. Yamakawa M, Youso T, Fujikawa T, Nishimoto T, Wada Y, Sato
performance when compared to experimental results. K, Yokohata H (2012) Combustion technology development for a
For wall to wall heat transfer approach, which is reviewed high compression ratio SI engine. SAE Int J Fuels Lubr 5(1):98–
105. https​://doi.org/10.4271/2011-01-1871
for the first time, global models, nodal models and finite
6. Deng B, Jianqin F, Zhang D, Yang J, Feng R, Liu J, Li K, Liu X
element models are discussed. The best trade off between (2013) The heat release analysis of bio-butanol/gasoline blends
computational cost and temperature distribution is obtained on a high speed SI (spark ignition) engine. Energy 60:230–241.
by nodal models, which also makes it more profitable to https​://doi.org/10.1016/j.energ​y.2013.07.055
7. Šarić S, Basara B, Žunič Z (2017) Advanced near-wall modeling
attend to Biot number restriction for lumped capacitance
for engine heat transfer. Int J Heat Fluid Flow 63:205–211. https​
assumption. It can be concluded by this review that the most ://doi.org/10.1016/j.ijhea​tflui​dflow​.2016.06.019
indicated models to use accomplished to engine thermody- 8. Bohac SV, Baker DM, Assanis DN (1996) A global model for
namic models are global and nodal models, for applications steady state and transient SI engine heat transfer studies. Technical
report, SAE Technical Paper. https​://doi.org/10.4271/96007​3
which thermodynamic results are the main purpose.
9. Bürkle S, Biondo L, Ding C-P, Honza R, Ebert Volker, Böhm
For wall to liquid heat transfer approach, which is Benjamin, Wagner Steven (2018) In-cylinder temperature meas-
reviewed for the first time in a paper, literature agree that urements in a motored ic engine using tdlas. Flow Turbul Combust
heat transfer from engine to liquid coolant should be calcu- 101(1):139–159. https​://doi.org/10.1007/s1049​4-017-9886-y
lated by both convective and buoyancy correlations, due to

13
 L. Fonseca et al.

10. Kosmadakis GM, Pariotis EG, Rakoupoulos CD (2012) Com- 26. Romero CA (2009) Contribución al conocimiento del compor-
parative analysis of three simulation models applied on a motored tamiento térmico y la gestión térmica de los motores de combus-
internal combustion engine. Energy Convers Manag 60:45–55. tión interna alternativos. PhD thesis, Universitat Politècnica de
https​://doi.org/10.1016/j.encon​man.2011.11.031 València. https​://doi.org/10.4995/Thesi​s/10251​/4923
11. Bernard G, Lebas R, Demoulin F-X (2011) A 0d phenomeno- 27. Fan X, Che Z, Wang T, Zhen L (2018) Numerical investigation
logical model using detailed tabulated chemistry methods to of boundary layer flow and wall heat transfer in a gasoline direct-
predict diesel combustion heat release and pollutant emis- injection engine. Int J Heat Mass Transf 120:1189–1199. https​://
sions. Technical report, SAE Technical Paper. https ​ : //doi. doi.org/10.1016/j.ijhea​tmass​trans​fer.2017.09.089
org/10.4271/2011-01-0847 28. Ngang EA, Abbe CVN (2018) Experimental and numerical analy-
12. Ge H-W, Shi Y, Reitz RD, Wickman DD, Willems Werner sis of the performance of a diesel engine retrofitted to use LPG
(2009) Optimization of a HSDI diesel engine for passenger cars as secondary fuel. Appl Therm Eng 136:462–474. https​://doi.
using a multi-objective genetic algorithm and multi-dimen- org/10.1016/j.applt​herma​leng.2018.03.022
sional modeling. SAE Int J Engines 2(1):691–713. https​://doi. 29. Soloiu V, Moncada JD, Gaubert R, Muiños M, Harp S, Ilie M,
org/10.4271/2009-01-0715 Zdanowicz A, Molina G (2018) LTC (low-temperature combus-
13. Vancoillie J, Sileghem L, Verhelst S (2014) Development and tion) analysis of PCCI (premixed charge compression ignition)
validation of a quasi-dimensional model for methanol and etha- with n-butanol and cotton seed biodiesel versus combustion and
nol fueled si engines. Appl Energy 132:412–425. https​://doi. emissions characteristics of their binary mixtures. Renew Energy
org/10.1016/j.apene​rgy.2014.07.046 123:323–333. https​://doi.org/10.1016/j.renen​e.2018.02.061
14. Verhelst S, Sheppard CGW (2009) Multi-zone thermody- 30. Renaud A, Ding C-P, Jakirlic S, Dreizler A, Böhm B (2018)
namic modelling of spark-ignition engine combustion-an over- Experimental characterization of the velocity boundary layer in
view. Energy Convers Manag 50(5):1326–1335. https​://doi. a motored IC engine. Int J Heat Fluid Flow 71:366–377. https​://
org/10.1016/j.encon​man.2009.01.002 doi.org/10.1016/j.ijhea​tflui​dflow​.2018.04.014
15. Zhang L (2018) Parallel simulation of engine in-cylinder pro- 31. Torregrosa AJ, Broatch A, Olmeda P, Salvador-Iborra J, Warey
cesses with conjugate heat transfer modeling. Appl Ther- A (2017) Experimental study of the influence of exhaust gas
mal Eng 142:232–240. https​: //doi.org/10.1016/j.applt ​h erma​ recirculation on heat transfer in the firedeck of a direct injection
leng.2018.06.084 diesel engine. Energy Convers Manag 153:304–312. https​://doi.
16. Broatch A, Olmeda P, García A, Salvador-Iborra J, Warey A org/10.1016/j.encon​man.2017.10.003
(2017) Impact of swirl on in-cylinder heat transfer in a light-duty 32. Ma PC, Ewan T, Jainski C, Lu L, Dreizler Andreas, Sick Volker,
diesel engine. Energy 119:1010–1023. https​://doi.org/10.1016/j. Ihme Matthias (2017) Development and analysis of wall mod-
energ​y.2016.11.040 els for internal combustion engine simulations using high-speed
17. Rashedul HK, Kalam MA, Masjuki HH, Ashraful AM, Imtenan micro-piv measurements. Flow Turbul Combust 98(1):283–309.
S, Sajjad H, Wee LK (2014) Numerical study on convective heat https​://doi.org/10.1007/s1049​4-016-9734-5
transfer of a spark ignition engine fueled with bioethanol. Int 33. Cerdoun M, Carcasci C, Ghenaiet A (2016) An approach for the
Commun Heat Mass Transf 58:33–39. https​://doi.org/10.1016/j. thermal analysis of internal combustion engines’ exhaust valves.
ichea​tmass​trans​fer.2014.08.019 Appl Ther Eng 102:1095–1108. https​://doi.org/10.1016/j.applt​
18. Benajes J, Olmeda P, Martín J, Blanco-Cavero D, Warey Alok herma​leng.2016.03.105
(2017) Evaluation of swirl effect on the global energy balance 34. Shayler PJ, Colechin MJF, Scarisbrick A (1996) Heat transfer
of a HSDI diesel engine. Energy 122:168–181. https​: //doi. measurements in the intake port of a spark ignition engine. Tech-
org/10.1016/j.energ​y.2017.01.082 nical report, SAE Technical Paper. https​://doi.org/10.4271/96027​
19. Weller HG, Uslu S, Gosman AD, Maly RR, Herweg R, Heel B 3
(1994) Prediction of combustion in homogeneous-charge spark- 35. Luján JM, Climent H, Olmeda P, Jiménez VD (2014) Heat
ignition engines. Int Symp COMODIA 94:163–169 transfer modeling in exhaust systems of high-performance two-
20. Heywood John B (1994) Combustion and its modeling in spark- stroke engines. Appl Therm Eng 69(1–2):96–104. https​://doi.
ignition engines. In: International symposium COMODIA, vol 94, org/10.1016/j.applt​herma​leng.2014.04.045
pp 1–15 36. Michl J, Neumann J, Rottengruber H, Wensing M (2016) Deri-
21. Reuss DL, Kuo T-W, Khalighi B, Haworth D, Rosalik M (1995) vation and validation of a heat transfer model in a hydrogen
Particle image velocimetry measurements in a high-swirl engine combustion engine. Appl Therm Eng 98:502–512. https​://doi.
used for evaluation of computational fluid dynamics calcu- org/10.1016/j.applt​herma​leng.2015.12.062
lations. Technical report, SAE Technical Paper. https​: //doi. 37. Pischinger R, Klell M, Sams T (2009) Thermodynamik der
org/10.4271/95238​1 Verbrennungskraftmaschine. Springer, Wien. https ​ : //doi.
22. Wang Z, Shuai S-J, Wang J-X, Tian G-H (2006) A computational org/10.1007/978-3-211-99277​-7
study of direct injection gasoline hcci engine with secondary 38. Annand WJD (1963) Heat transfer in the cylinders of reciprocat-
injection. Fuel 85(12–13):1831–1841. https​://doi.org/10.1016/j. ing internal combustion engines. Proc Inst Mech Eng 177(1):973–
fuel.2006.02.013 996. https​://doi.org/10.1243/PIME_PROC_1963_177_069_02
23. Millo F, Luisi S, Borean F, Stroppiana A (2014) Numerical and 39. Woschni G (1967) A universally applicable equation for the
experimental investigation on combustion characteristics of a instantaneous heat transfer coefficient in the internal combus-
spark ignition engine with an early intake valve closing load con- tion engine. Technical report, SAE Technical paper. https​://doi.
trol. Fuel 121:298–310. https:​ //doi.org/10.1016/j.fuel.2013.12.047 org/10.4271/67093​1
24. di Mare F, Knappstein R, Baumann M (2014) Application of les- 40. Han SB, Chung YJ, Kwon YJ, Lee S (1997) Empirical for-
quality criteria to internal combustion engine flows. Comput Flu- mula for instantaneous heat transfer coefficient in spark igni-
ids 89:200–213. https​://doi.org/10.1016/j.compf​l uid.2013.11.003 tion engine. Technical report, SAE Technical Paper. https​://doi.
25. Finol CA, Robinson K (2006) Thermal modelling of modern org/10.4271/97299​5
engines: a review of empirical correlations to estimate the in- 41. De Cuyper T, Broekaert S, Chana K, De Paepe M, Verhelst S
cylinder heat transfer coefficient. Proc Inst Mech Eng Part D J (2017) Evaluation of empirical heat transfer models using TFG
Automob Eng 220(12):1765–1781. https:​ //doi.org/10.1243/09544​ heat flux sensors. Appl Therm Eng 118:561–569. https​://doi.
070JA​UTO20​2 org/10.1016/j.applt​herma​leng.2017.02.049

13
Internal Combustion Engine Heat Transfer and Wall Temperature Modeling: An Overview

42. Irimescu A, Merola SS, Tornatore C, Valentino G (2015) Devel- 58. Robinson K, Campbell NAF, Hawley JG, Tilley DG (1999) A
opment of a semi-empirical convective heat transfer correla- review of precision engine cooling. Technical report, SAE Techni-
tion based on thermodynamic and optical measurements in a cal Paper. https​://doi.org/10.4271/1999-01-0578
spark ignition engine. Appl Energy 157:777–788. https​://doi. 59. Kandlikar SG (1998) Heat transfer characteristics in partial boil-
org/10.1016/j.apene​rgy.2015.02.050 ing, fully developed boiling, and significant void flow regions of
43. Martins JJG, Finlay IC (1990) Heat transfer to air-ethanol and subcooled flow boiling. J Heat Transf 120(2):395–401. https:​ //doi.
air-methanol sprays flowing in heated ducts and across heated org/10.1115/1.28242​63
intake valves. Technical report, SAE Technical Paper. https​://doi. 60. Robinson K, Hawley JG, Campbell NAF (2003b) Experimental and
org/10.4271/90058​3 modelling aspects of flow boiling heat transfer for application to inter-
44. Torregrosa AJ, Olmeda P, Degraeuwe B, Reyes M (2006) A con- nal combustion engines. Proc Inst Mech Eng Part D J Autom Eng
cise wall temperature model for DI diesel engines. Appl Therm 217(10):877–889. https​://doi.org/10.1243/09544​07037​69683​289
Eng 26(11–12):1320–1327. https​://doi.org/10.1016/j.applt​herma​ 61. Kandlikar SG, Bulut M (2003) An experimental investigation
leng.2005.10.021 on flow boiling of ethylene-glycol/water mixtures. J Heat Transf
45. Baker DM, Assanis DN (1994) A methodology for coupled 125(2):317–325. https​://doi.org/10.1115/1.15618​16
thermodynamic and heat transfer analysis of a diesel engine. 62. Steiner H, Brenn G, Ramstorfer F, Breitschädel B (2011) Increased
Appl Math Modell 18:590–601. https​://doi.org/10.1016/0307- cooling power with nucleate boiling flow in automotive engine
904X(94)90317​-4 applications. In: Chiaberge M (ed) New trends and developments
46. Torregrosa AJ, Olmeda P, Martín J, Romero C (2011) A tool in automotive system engineering, chapter 13. IntechOpen, Rijeka
for predicting the thermal performance of a diesel engine. Heat 63. Li Z, Huang RH, Wang ZW (2012) Subcooled boiling heat trans-
Transf Eng 32(10):891–904. https ​ : //doi.org/10.1080/01457​ fer modelling for internal combustion engine applications. Proc
632.2011.54863​9 Inst Mech Eng Part D J Autom Eng 226(3):301–311. https​://doi.
47. Shayler PJ, Christian SJ, Ma T (1993) A model for the investiga- org/10.1177/09544​07011​41734​9
tion of temperature, heat flow and friction characteristics during 64. Torregrosa AJ, Broatch A, Olmeda P, Cornejo O (2014) Experi-
engine warm-up. Technical report, SAE Technical Paper. https​:// ments on subcooled flow boiling in ic engine-like conditions at
doi.org/10.4271/93115​3 low flow velocities. Exp Therm Fluid Sci 52:347–354. https:​ //doi.
48. Jarrier L, Champoussin JC, Yu R, Gentile D (2000) Warm-up of org/10.1016/j.expth​ermfl​usci.2013.10.004
a DI diesel engine: experiment and modeling. Technical report, 65. Mehdipour R, Baniamerian Z, Delauré Y (2016) Three dimen-
SAE Technical Paper. https​://doi.org/10.4271/2000-01-0299 sional simulation of nucleate boiling heat and mass transfer in
49. Jafari A, Hannani SK (2006) Effect of fuel and engine operational cooling passages of internal combustion engines. Heat Mass Transf
characteristics on the heat loss from combustion chamber surfaces 52(5):957–968. https​://doi.org/10.1007/s0023​1-015-1611-6
of SI engines. Int Commun Heat Mass Transf 33(1):122–134. 66. Torregrosa AJ, Broatch A, Olmeda P, Martín J (2010) A contribu-
https​://doi.org/10.1016/j.ichea​tmass​trans​fer.2005.08.008 tion to film coefficient estimation in piston cooling galleries. Exp
50. Trujillo EC, Jiménez-Espadafor FJ, Villanueva JAB, García MT Therm Fluid Sci 34(2):142–151. https​://doi.org/10.1016/j.expth​
(2011) Methodology for the estimation of cylinder inner surface ermfl​usci.2009.10.003
temperature in an air-cooled engine. Appl Therm Eng 31:1474– 67. Liu YC, Guessous L, Sangeorzan BP, Alkidas AC (2014) Labora-
1481. https​://doi.org/10.1016/j.applt​herma​leng.2011.01.025 tory experiments on oil-jet cooling of internal combustion engine
51. Trujillo EC, Jiménez-Espadafor FJ, Villanueva JAB, García MT pistons: area-average correlation of oil-jet impingement heat
(2012) Methodology for the estimation of head inner surface tem- transfer. J Energy Eng 141(2):C4014003. https​://doi.org/10.1061/
perature in an air-cooled engine. Appl Therm Eng 35:202–211. (ASCE)EY.1943-7897.00002​27
https​://doi.org/10.1016/j.applt​herma​leng.2011.10.032 68. Peng W, Jizu L, Minli B, Yuyan W, Chengzhi Hu, Liang Zhang
52. Cerit M, Coban M (2014) Temperature and thermal stress analy- (2014) Numerical simulation on the flow and heat transfer pro-
ses of a ceramic-coated aluminum alloy piston used in a diesel cess of nanofluids inside a piston cooling gallery. Numer Heat
engine. Int J Therm Sci 77:11–18. https​://doi.org/10.1016/j.ijthe​ Transf Part A Appl 65(4):378–400. https:​ //doi.org/10.1080/10407​
rmals​ci.2013.10.009 782.2013.83207​1
53. Yaohui L, Zhang X, Xiang P, Dong D (2017) Analysis of thermal 69. Payri F, Olmeda P, Martín J, Carreño R (2014) A new tool to
temperature fields and thermal stress under steady temperature perform global energy balances in di diesel engines. SAE Int J
field of diesel engine piston. Appl Therm Eng 113:796–812. https​ Engines 7(1):43–59. https​://doi.org/10.4271/2014-01-0665
://doi.org/10.1016/j.applt​herma​leng.2016.11.070 70. Kikusato A, Kusaka J, Daisho Y (2015) A numerical study on
54. Goudarzi K, Moosaei A, Gharaati M (2015) Applying artificial predicting combustion chamber wall surface temperature distribu-
neural networks (ANN) to the estimation of thermal contact con- tions in a diesel engine and their effects on combustion, emission
ductance in the exhaust valve of internal combustion engine. Appl and heat loss characteristics by using a 3d-cfd code combined with
Therm Eng 87:688–697. https​://doi.org/10.1016/j.applt ​herma​ a detailed heat transfer model. Technical report, SAE Technical
leng.2015.05.060 Paper. https​://doi.org/10.4271/2015-01-1847
55. Finlay IC, Harris D, Boam DJ, Parks BI (1985) Factors influ- 71. Martín J, Novella R, García A, Carreño R, Heuser Benedikt, Kremer
encing combustion chamber wall temperatures in a liquid- Florian, Pischinger Stefan (2016) Thermal analysis of a light-duty ci
cooled, automotive, spark-ignition engine. Proc Inst Mech Eng engine operating with diesel-gasoline dual-fuel combustion mode.
Part D Transp Eng 199(3):207–214. https​://doi.org/10.1243/ Energy 115:1305–1319. https​://doi.org/10.1016/j.energ​y.2016.09.021
PIME_PROC_1985_199_158_01 72. Broatch A, Olmeda P, Margot X, Escalona J (2019) New approach
56. Chen JC (1966) Correlation for boiling heat transfer to satu- to study the heat transfer in internal combustion engines by 3d
rated fluids in convective flow. Ind Eng Chem Process Des Dev modelling. In J Therm Sci 138:405–415. https:​ //doi.org/10.1016/j.
5(3):322–329. https​://doi.org/10.1021/i2600​19a02​3 ijthe​rmals​ci.2019.01.006
57. Robinson K, Hawley JG, Hammond GP, Owen NJ (2003a) Con-
vective coolant heat transfer in internal combustion engines. Proc Publisher’s Note Springer Nature remains neutral with regard to
Inst Mech Eng Part D J Autom Eng 217(2):133–146. https​://doi. jurisdictional claims in published maps and institutional affiliations.
org/10.1177/09544​07003​21700​207

13