Energy 238 (2022) 121698

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Exhaust waste heat recovery from a heavy-duty truck engine:

Experiments and simulations
Jelmer Rijpkema a, *, Olof Erlandsson b, Sven B. Andersson a, Karin Munch a
a
Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg, Sweden
b €lvesborg, Sweden
TitanX Engine Cooling AB, So

a r t i c l e i n f o a b s t r a c t

Article history: Waste heat recovery using an (organic) Rankine cycle is an important and promising technology for
Received 4 March 2021 improving engine efficiency and thereby reducing the CO2 emissions due to heavy-duty transport. Ex-
Received in revised form periments were performed using a Rankine cycle with water for waste heat recovery from the exhaust
22 July 2021
gases of a heavy-duty Diesel engine. The experimental results were used to calibrate and validate steady-
Accepted 4 August 2021
Available online 14 August 2021
state models of the main components in the cycle: the pump, pump bypass valve, evaporator, expander,
and condenser. Simulations were performed to evaluate the cycle performance over a wide range of
engine operating conditions using three working fluids: water, cyclopentane, and ethanol. Additionally,
Keywords:
Experiments
cycle simulations were performed for these working fluids over a typical long haul truck driving cycle.
Exhaust gases The predicted net power output with water as the working fluid varied between 0.5 and 5.7 kW, where
Heavy-duty the optimal expander speed was dependent on the engine operating point. The net power output for
Internal combustion engine simulations with cyclopentane was between 1.8 and 9.6 kW and that for ethanol was between 1.0 and
Organic rankine cycle (ORC) 7.8 kW. Over the driving cycle, the total recovered energy was 11.2, 8.2, and 5.2 MJ for cyclopentane,
Reciprocating piston expander ethanol, and water, respectively. These values correspond to energy recoveries of 3.4, 2.5, and 1.6%,
Long haul truck respectively, relative to the total energy requirement of the engine. The main contribution of this paper is
Waste heat recovery
the presentation of experimental data on a complete Rankine cycle-based WHR system coupled to a
heavy-duty engine. These results were used to validate component models for simulations, allowing for a
realistic estimation of the steady-state performance under a wide range of operating conditions for this
type of system.
© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

1. Introduction control, hybridization, reduction of friction and other losses,

renewable fuels, hydrogen, fuel cells, and waste heat recovery
The continuous increase of anthropogenic greenhouse gas (WHR) [4,5]. WHR systems generate power from the waste heat of
(GHG) emissions is causing severe adverse effects on the climate. the heat sources in a HD engine, namely the charge air cooler (CAC),
Consequently, the global energy system must rapidly reduce its the exhaust recirculation (EGR) cooler, the engine coolant, or the
emissions [1]. Because heavy-duty (HD) trucks and buses are exhaust gases [6]. Many different WHR concepts and technologies
responsible for over 5% of the total GHG emissions in Europe [2], exist, including turbocompounding [7], thermoacoustic convertors
ways to reduce emissions from HD vehicles are needed. As a result, [8], thermoelectric generators [9], and technologies based on
the European Union has imposed CO2 emission standards for HD thermodynamic cycles such as the Brayton [10], Stirling [11],
vehicles that require emission reductions of 15% from 2025 on- Rankine [12], or various flash cycles [6]. Systems based on the
wards and 30% from 2030 onwards, relative to the 2019 baseline (organic) Rankine cycle (ORC) have been found to achieve good
[3]. Several technologies and powertrain concepts have been pro- performance and flexibility, although the added weight,
posed to help meet these requirements, including improvements in complexity, and payback time remain obstacles to their commercial
combustion and air management efficiency, predictive powertrain implementation [13].
In an ORC-based WHR system, waste heat is used to evaporate a
working fluid at elevated pressure. The high pressure, high tem-
* Corresponding author.
perature fluid is then expanded, converting the heat energy into
E-mail address: jelmer.rijpkema@chalmers.se (J. Rijpkema). power, after which the fluid is condensed before entering the

https://doi.org/10.1016/j.energy.2021.121698
0360-5442/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

pump. The goal of the WHR system is to maximize the net power For the CRF engine, a system with R245fa and a turbine was used,
output, while maintaining superheated vapor conditions at the giving a maximum power output of 2.5 kW. Latz et al. [45] used a
inlet of the expander [14]. The performance of such systems in similar setup as in this paper featuring the same engine, but
heavy-duty Diesel (HDD) engines has been the topic of many recovered heat from the EGR cooler using a reciprocating piston
publications (mostly simulation studies), with reported fuel savings expander and water. The maximum recovered power in this case
of over 5% [13,15]. Important issues addressed in these studies was 2.7 kW. Simulations were performed to identify important
include heat source selection [16], the choice of working fluid parameters for performance improvement of the EGR evaporator
[16e18], heat exchanger modeling [19,20], expander selection and and piston expander. Shu et al. [46] compared the working fluids
performance [21], cycle configuration [22], and techno-economic R123 and R245fa in a WHR system with an intermediate oil loop
performance optimization [23]. Due to the transient operation of connected to the exhaust of an HDD engine. No expander was
an internal combustion engine, much recent research has focused installed in the system, but, by using estimated efficiencies, a
on developing and validating dynamic models and controls for ORC maximum power output of 9.7 kW for R123 was predicted. In
systems in HD engines [14,24e27]. In addition to the academic another study, Yu et al. [47] replaced the oil loop with another
interest, ORC systems for HDD engines have attracted considerable Rankine cycle with water. This improved the estimated power
interest from major automotive industry firms in the last decade, output to 12.7 kW or 5.6% in relative terms. More recently, Shi et al.
including AVL [28], BMW [29,30], Bosch [31], Cummins [32,33], [48] used the same engine for WHR from the exhaust and engine
Daimler [34], Honda [35], Mahle [36], Scania [37], Volkswagen [38], coolant with four configurations of a CO2 transcritical Rankine cy-
Volvo Car Corporation [39,40], and Volvo Group [24,41]. cle. Using estimated efficiencies, the maximum power output was
A number of notable recent publications concerning experi- predicted to be 3.5 kW. Guillaume et al. [49] simulated the exhaust
mental research on WHR from HDD engines are briefly discussed conditions of a HDD engine using a boiler with thermal oil and
below and listed in Table 1. Seher et al. [31] reported an experi- concluded that R1233zd(E) performed better than R245fa in their
mental study on a Rankine cycle with water for a 12 L HDD engine. experiments, providing a maximum power output of 2.8 kW with a
A maximum power output of 14 kW was obtained with a piston turbine expander. Alshammari et al. [50] recovered the exhaust
expander, corresponding to 4.3% of the engine power, while 9 kW heat using an ORC with R1233zd(E), a turbine, and an intermediate
(2.8%) was achieved with a turbine expander. Based on simulations thermal oil loop, giving a maximum recovered power of 6.3 kW.
it was concluded that water or ethanol with a piston or ethanol Their results were complemented by CFD simulations and evalua-
with a turbine were the preferred solutions, giving a maximum tions of the radial inflow turbine performance. In a more recent
relative power of 5.3%. Furukawa et al. [42] tested two ORC systems study [51], the same group subsequently tested the same engine
for WHR from a downsized HDD engine. In their first system, heat with a WHR system featuring a thermal loop, a turbine, and
was recovered from the engine coolant and EGR gases, reducing Novec649 as the working fluid. The maximum power output in this
fuel consumption by 3.8%. In their second system, the coolant case was 9.1 kW (11.2%). Additionally, they showed that increasing
temperature was increased from 86 to 105
C, the exhaust gas was the cooling water temperature and superheating temperature
included as a heat source, and a recuperator was added, improving reduced the performance of the turbine.
the fuel consumption reduction to 7.5%. Yang et al. [43] used an ORC Although there have been many publications regarding the use
with R245fa and a screw expander to recover up to 28.6 kW (10.2%) of ORC for WHR in HDD engines, there is still a lack of publicly
from a HDD engine. Zhang et al. [44] also used a screw expander in available experimental data. This is partly because publications
an ORC with R123 for exhaust heat recovery from a HDD engine and from industry often report performance improvements without
achieved a maximum power of 10.4 kW. Bettoja et al. [41] per- providing much detail on the cycle components. Additionally, many
formed experiments on two systems for two different engines: a publications reporting results of dynamic models often present
Volvo US10 and a CRF Cursor 11. For the Volvo engine, heat was experimental data that was used to validate the model. These re-
recovered from the exhaust and EGR with a water/ethanol mixture. sults only specify the controlled parameters (e.g. mass flow rate or
An orifice was used instead of an expander and the system was evaporator outlet temperature) without offering insight into the
predicted to achieve a relative power recovery between 1.5 and 3%. cycle or component performance. In this paper, an experimental

Table 1
Recent experimental studies on WHR from HDD engines using ORCs.

Reference Year Engine Heat source Fluids(s) Expander _ max

W _ max
W
W_ eng

- - L/kW - - - kW %

Seher et al. [31] 2012 12.0/326 Exhaust Water Piston, Turbine 14 4.3
Furukawa et al. [42] 2014 9.0/- Exhaust, EGR, HFE Turbine e 7.5
Coolant
Yang et al. [43] 2014 9.7/280 Exhaust R245fa Screw 28.6 10.2
Zhang et al. [44] 2014 -/250 Exhaust R123 Screw 10.4 4.2
Bettoja et al. [41] 2016 11.1/353 Exhaust R245fa Turbine 2.5 e
12.7/317 Exhaust, EGR Water/Ethanol e e 3.0
Latz et al. [45] 2016 12.8/373 EGR Water Piston 2.7
Shu et al. [52] 2016 8.4/243 Exhaust R123, R245fa e 9.7 4.0
Yu et al. [47] 2016 8.4/243 Exhaust Water, R123 e 12.7 5.6
Guillaume et al. [49] 2017 -/ - Exhaust R245fa, Turbine 2.8 e
R1233zd(E)
Shi et al. [48] 2017 8.4/243 Exhaust, CO2 e 3.5 e
Coolant
Alshammari et al. [50] 2018 7.3/206 Exhaust R1233zd(E) Turbine 6.3 7.6
Alshammari et al. [51] 2019 7.3/206 Exhaust Novec649 Turbine 9.1 11.2

2
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

setup consisting of a Rankine cycle with water connected to the

exhaust of a HDD engine is evaluated and used to calibrate and
validate models of the relevant cycle components. The components
are then used to develop a realistic Rankine cycle model to deter-
mine the optimum performance in a driving cycle for a variety of
working fluids. Simulation of the driving cycle allows for an eval-
uation of the performance of the WHR system over the full oper-
ational range of the engine. However, the corresponding transient
effects are considered outside the scope of this paper, and are,
therefore, not taken into account. The goal of this publication is to
provide detailed experimental data on a full Rankine system con-
nected to a HDD engine, and to develop models that give an ac-
curate prediction of the WHR system performance under realistic
operating conditions.

2. Experimental setup

The experimental setup is shown in Fig. 1 and a schematic Fig. 2. Schematic depiction of the experimental setup.
overview is presented in Fig. 2. The setup consists of a heavy-duty
truck engine whose exhaust gases are used as the heat input for a
Table 2
WHR system based on a Rankine cycle using water as the working
Engine specifications.
fluid. Both the engine and WHR system are placed in an engine test
cell in which the temperature and pressure can be regulated. Only Type Volvo D13 US 2010
Configuration 4 Stroke/6 Cylinder inline/EGR
steady-state measurements are available from the engine due to
Peak power 373 kW (500 hp)
limitations on the engine brake. The setup is monitored and Peak torque 2373 Nm
controlled from the adjacent control room using multiple modules Compression ratio 16.0:1
installed in two National Instruments CompactRIO 9074 controllers Bore x Stroke 131  158 mm
coupled to a Labview interface. Several cameras and a connecting Displacement 12.8 L
Aspiration Turbocharged
window allow for visual observations while running the setup.
Sensor data was measured at a sampling frequency of 10 Hz which
was written to disk every second. For each measuring point, three
minutes of data were collected and averaged. at the minimum pump speed would otherwise be too large to
The engine is a turbocharged 12.8 L Volvo Diesel engine with permit full evaporation under low load engine operating condi-
charge air cooling (CAC) and exhaust gas recirculation (EGR); its tions. The evaporator uses the exhaust gases downstream the en-
specifications are shown in Table 2. A Schenck D900-1e water brake gine turbocharger to evaporate and superheat the water from the
is used to control the engine speed. The engine torque is controlled pump. It is specifically manufactured for this experimental setup by
by regulating the fuel flow through manual operation of the gas TitanX and is shown in Fig. 3. Under start-up and shut-down con-
pedal. Fuel is provided from a Diesel tank located in a separate fuel ditions, the fluid leaving the evaporator is not always fully evapo-
storage. rated. Therefore, the expander inlet and outlet valves are closed and
Fig. 2 shows the main components of the WHR system; their the controllable expander bypass valve (BPV) open so the flow
specifications are listed in Table 3 together with the corresponding bypasses the expander, preventing liquid from entering. The
controller where applicable. The WHR system is a typical Rankine expander is a reciprocating piston type with two cylinders using a
cycle with water as the working fluid. The suction side of the pump separate crankcase with its own oil circuit. Water enters the
is connected to the buffer tank, which is open to the atmosphere. A expander as superheated steam and exits as a two-phase mixture at
controllable pump bypass valve (BPV) is installed because the flow low pressure. During expansion, the expander converts some of the
energy in the steam into power via an electric motor, which con-
trols the expander speed. Since some of the steam enters the
crankcase of the expander, the oil is heated to 140
C, causing this
water to evaporate and be expelled to the environment. To prevent
hot oil from entering the oil pump, the oil is subsequently cooled. In
the cycle, the oil is separated from the water and the low-pressure
two-phase mixture enters the condenser, where it is condensed
and subcooled to around 15
C using process water from the test
cell. From the condenser, the subcooled water enters the buffer
tank. To avoid overpressures in the system, safety valves (SV) are
installed on the high and low pressure sides of the cycle.
Fig. 2 also shows the locations of the different sensors in the
system; the details and accuracies are shown in Table 4. The engine
speed and torque were taken from the Schenck D900-1e mea-
surements and the fuel flow using an AVL 730 fuel balance. The
engine inlet air flow was measured using the pressure drop over a
calibrated venturi tube, sufficiently upstream the turbocharger to
avoid flow pulsations. Pressures in the system were measured using
WIKA A-10 pressure transmitters with different ranges depending
Fig. 1. Experimental setup.

3
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Table 3
Specifications of cycle components.

Component Brand Type Controller

Condenser Modine Plate, counter-current flow

Evaporator TitanX Plate, cross-counter flow
Expander Voith Reciprocating piston, 2-cylinder
Expander bypass valve Swagelok SS-18RS8 Integral-bonnet needle Hanbay MCL-000 AF
Expander electric motor David McClure LTD 400 V, 3-phase, 37 kW Parker DC590þ Integrator 2
Pump Danfoss PAH2 Axial piston
Pump bypass valve Swagelok SS-1RS4 Integral-bonnet needle Hanbay MCL-000 AF
Pump electric motor Hoyer HMA2 90L-4 230 V, 3-phase, 1.5 kW IMO iDrive EDX-220-21-E

Fig. 3. Exhaust evaporator.

Table 4 3.1. Engine

Measurement devices accuracy.

Input Type Range Accuracy Unit The experimental measurements of the exhaust mass flow and
outlet temperature are shown in Figs. 4 and 5. These results are
Engine speed Schenck D900-1e 0e6500 ±2 rpm
Engine torque Schenck D900-1e 4000e4000 ±8 Nm
averages based on the measurements acquired in experimental set
Expander speed HBM T40B 0e20,000 ±10 rpm 5 (see Table 5). The engine operating points are named in accor-
Expander torque HBM T40B 500e500 ±0.25 Nm dance with the conventions of the European Stationary Cycle (ESC):
Fuel flow AVL 730 0e150 ±0.9 kg/h the letters A, B, and C indicate different engine speeds, and the
Mass flow Micro Motion F025S 0e100 ±0.2 g/s
numbers 25, 50, 75, and 100 indicate the load percentages at the
Cycle high pressure WIKA A-10 0e60 ±0.6 bar(g)
Cycle low pressure WIKA A-10 0e6 ±0.06 bar(g) corresponding speed [54]. An additional highway (HW) point was
Exhaust pressure WIKA A-10 0e2.5 ±0.03 bar(g) tested, which represents typical engine conditions during highway
Pressure drop Yokogawa EJA110E 0e5000 ±2.75 Pa driving. These measurements were used to define the heat input


Temperature RS Pro Type K 75e1100 ±1.5 C
conditions used in the cycle model. The original measured mass
flows were somewhat higher than those obtained in previous
experimental studies on similar engines, suggesting that the mass
on the location. Temperature measurements were taken with 3 mm flow values measured in the test cell were systematically biased
diameter RS Pro Type K thermocouples. In the Rankine cycle, the upward. Consequently, the original mass flow values were multi-
mass flow was measured with a coriolis mass flow meter and the plied by an error factor of 0.75 to obtain more realistic values. The
expander speed and torque were both measured with an universal mass flows presented in Fig. 4 have been corrected in this manner.
digital torque transducer.
3.2. Pump

The mass flow from the pump is determined from its inlet
3. Experimental results and component calibration density (rin) and volume flow (V_ pmp ) using Eq. (1). The specifica-
tions of the pump are shown in Table 7.
In this section the experimental results and modeling relations
of the main components in the WHR system are combined in _ pmp ¼ rin V_ pmp
m (1)
separate sections: Engine, Pump, Pump Bypass Valve, Evaporator,
Expander, and Condenser. If the results are used to calibrate and/or Although the axial piston pump is relatively insensitive to
validate the component, it is also shown. The last section shows the pressure changes, the actual volume flow (V_ pmp ) is calculated by
comparison of the experiments and simulations for the full cycle applying a correction to the theoretical flow (V_ ):
th
performance. A total of six different experimental sets were used, as
ppmp; out
listed and described in Table 5, which shows the corresponding V_ pmp ¼ V_ th  V_ corr (2)
number of experimental points and a description for each experi- pmax
mental set.
The theoretical flow in L/min is defined as:
Multiple quantities derived from the performed measurements
are presented in the subsequent sections. The standard deviations
V_ th ¼ Vpmp Npmp (3)
for the measured (i.e. non-derived) data are represented by error
bars in the figures. Table 6 shows the maximum measurement error The flow correction in L/min depends on the pump outlet
based on the standard rules for error propagation [53]. pressure (ppmp, out) in bar and the pump speed (Npmp) in rpm. This is
4
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Table 5
Numbering, quantity of experimental points (Qty.) and description of the experimental sets.

Set Qty. Description

1. 7 Cold system, no expander, low pressure, pump validation

2. 6 Cold system, no expander, high pressure, pump validation
3. 13 Cold system, no expander, pump bypass valve calibration
4. 16 Cold system, no expander, pump bypass valve validation
5. 28 Hot system, no expander, engine results, evaporator calibration, condenser results
6. 41 Hot system, expander calibration, cycle validation

Table 6
Measurement error for the derived quantities.

Quantity Symbol Max. Error

Engine mass flow _ eng

m ±5.9%
Pump power _ pmp
W ±9.4%
Bypass valve effective area Abpv ±4.5%
Evaporator heat transfer rate Q_ evap ±6.8%
Expander filling factor 4f, is ±6.0%
Expander efficiency hexp ±8.9%
Expander power W _ exp ±8.2%
Condenser heat transfer rate Q_ cond ±7.9%

Fig. 5. Engine speed-torque map showing the measured engine exhaust outlet
temperature.

Table 7
Pump specifications [55].

Maximum pressure pmax 100 bar

Displacement volume Vpmp 0.002 L

3.3. Pump bypass valve

The pump bypass valve is modeled as an incompressible flow

Fig. 4. Engine speed-torque map showing the measured engine exhaust mass flow. valve, as shown in Eq. (7). The discharge coefficient (Cd) and valve
The measurements were corrected by a factor 0.75 based on an estimated error of the area (A) can be combined into an effective area (Abpv).
measured values.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
_ bpv ¼ Cd A
m 2rin ðpin  pout Þ ¼ Abpv 2rin ðpin  pout Þ (6)
approximated in Eq. (4) using technical data from the manufacturer Although the pump bypass mass flow (m _ bpv ) is not known, the
[55]. _ evap ) was measured and can be subtracted
evaporator mass flow (m
from the pump mass flow (m _ pmp ), available from the pump model:
V_ corr ¼ 0:0001Npmp þ 1:2 (4)
m _ pmp  m
_ bpv ¼ m _ evap (7)
The pump power can be calculated with an estimated efficiency:
The effective area is a function of the valve position (xbpv) and is
. calibrated using the data from experimental set 3. The calibrated
_ pmp ¼ m
W _ pmp ðhpmp; out  hpmp; in Þ hpmp (5) model consists of a combination of two linear functions of which
the coefficients are shown in Table 8.
To validate the pump model, measurements were conducted at The resulting effective area as a function of the valve position is
a low (set 1) and a high (set 2) pressure. In both cases, no heat was shown on the left on Fig. 7, together with the corresponding
added to the system, the pump bypass valve was fully closed, the experimental data. During these experiments no heat was added to
expander was not running, and the expander bypass valve was used the system, the pump bypass valve position was controlled, the
to control the pressure. The results of the experiments and simu- expander was not running, and the expander bypass valve was held
lations are shown in Fig. 6, indicating that a good agreement was at a fixed position. The corresponding results for the mass flow is
achieved. shown on the right of Fig. 7, along with results obtained under
5
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Fig. 8. Schematic depiction of the heat transfer area of the exhaust evaporator.

Table 9
Heat exchanger dimensions.

Heat exchanger length L 144 mm

Heat exchanger width W 241 mm
Heat exchanger height H 247 mm
Fig. 6. The mass flow as a function of the pump speed and pump outlet pressure. Channel height b 3.00 mm
Plate thickness t 0.40 mm

Fluid-specific wf exh
Table 8
Pump bypass valve coefficients. Number of channels nch 35 36 e
Number of passes npass 3 1 e
xbpv % 0 1 2 7 100 Number of fins in x-dir. nf, x 96.0 45.4 e
Abpv mm2 0 0.25 0.35 1 2.65 Number of fins in y-dir. nf, y 77.8 164.7 e
Number of fins nf 7468 7468 e
Total flow area Ac 4309 22,808 mm2

different conditions (set 4) that were used for model validation.

Good agreement with the experiments is observed for both for the
calibration and validation data sets.
Aflow ¼ sf hf (9)

.
3.4. Evaporator Abase ¼ 2ðsf Lf þ tf ðsf  tf Þ 2Þ (10)

The exhaust evaporator is a cross-counter flow plate heat

exchanger with fins; its heat transfer area is depicted schematically Afin ¼ 2ðhf Lf þ hf tf Þ (11)
in Fig. 8. The exhaust flow side has one pass and 36 channels, while
the water side has three passes and 35 channels. The dimensions of For each pass and channel the number of fins for the working
the heat exchanger are listed in Tables 9 and 10. The model only fluid and exhaust can be calculated:
accounts for heat transfer; the pressure drop over the evaporator is
ignored. L H
Heat transfer is modeled using fin-specific equations. The geo- nf; x; wf ¼ ; nf ; y; wf ¼ (12)
pf Lf
metric parameters of the fins are computed using the values pre-
sented in Table 10 and the following equations [56]:
L H
nf; x; exh ¼ ; nf; y; exh ¼ (13)
4sf bLf Lf pf
Dh ¼ (8)
2ðsf Lf þ bLf þ tf bÞ þ tf sf
As a result, the total cross-sectional area (or total flow area) for
the flow can be calculated:

Fig. 7. Pump bypass valve effective area (left) and mass flow (right) as a function of the bypass valve position.

6
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Table 10 To solve the heat transfer equations, the heat exchanger is dis-
Fin geometry. cretized into a set of elements as shown in Fig. 9. A low resolution
Spacing sf 1.35 mm model is compared to a high resolution TitanX model whose set-
Thickness tf 0.15 mm tings are shown in Table 11. The heat transfer surface (As) is the
Strip flow length Lf 3.175 mm combined base (Abase) and fin (Afin) surface:
Pitch pf 1.50 mm
Effective channel height hf 2.85 mm
Hydraulic diameter Dh 1.791 mm As; el ¼ Abase; el þ Afin; el (23)
Flow area Aflow 3.848 mm2
Base area Abase 8.753 mm2 The heat transfer is calculated for all channels based on the total
Fin area Afin 18.95 mm2 surface area for each element:

As; el; tot ¼ nch; wf As; el (24)

nf; x; wf For each element the heat transfer can be calculated:
Ac; wf ¼ nch; wf A (14)
npass; wf flow
Q_ el ¼ Uel As; el; tot ðTexh; el  Twf ; el Þ (25)
nf; y; exh
Ac; exh ¼ nch; exh A (15) The overall heat transfer coefficient (U) consists of the sum of
npass; exh flow
the separate contributions:
The mass flux (G) for the exhaust and water sides is calculated
by dividing the mass flow by the corresponding cross-sectional 1 1 tw 1 1 1
_ c ). With these definitions and the associated ther-
area (G ¼ m=A ¼ þ þ z þ (26)
U awf lw aexh awf aexh
modynamic and transport properties, the Reynolds number (Re),
the Prandtl number (Pr), and the heat transfer coefficient (a) can be The experimental and simulation results of the evaporator heat
determined using the Nusselt (Nu) number. For single-phase heat transfer rate and outlet temperature are shown in Fig. 10, based on
transfer, the following expression for the Nusselt number [56] is experimental set 5 from Table 5. Experiments were performed at
used: every engine operating point shown in Figs. 4 and 5 (A25-C100). In
these experiments, the pump and pump bypass valve position were
Nu ¼ jRePr1=3 (16) controlled to maintain a constant mass flow at each operating
point. The expander was not running and the expander bypass
Where: valve was used to control the pressure. Both the low resolution
!0:154 !0:150 !0:068 model and the TitanX model exhibit good agreement with the
sf tf tf experiments. However, it should be noted that in most cases the
j ¼ 0:652Re0:540
hf Lf sf available heat from the exhaust gases was so large that the water
was superheated to a temperature close to the exhaust gas inlet
2 !0:504 !0:456 !1:06 30:1 temperature. Increasing the mass flow would reduce the evapo-
sf tf tf
,41 þ 5:269,10 5 1:34
Re 5 rator outlet temperature of the water, but the magnitude was
hf Lf sf
difficult to control experimentally. Because of the relatively small
(17) mass flows and high latent heat of water, small deviations in mass
flow caused large deviations in the evaporator outlet temperature.
The two-phase heat transfer coefficient is the sum of the Another possible source of error is that the temperatures were
nucleate boiling (anb) and convective (acv) components [56]: measured at a single location, slightly downstream of the evapo-
rator outlet. This could cause variations between the experimental
a ¼ anb þ acv (18) values due to heat loss and local effects, leading to deviations from
_
The nucleate boiling component is a function of the heat flux (q) the model values. However, since the heat transfer to the working
for each element, the molecular weight (Mw), and the reduced fluid is the most important for the prediction of the cycle perfor-
pressure (pcrit), and is defined as: mance, the model can still be used in the cycle simulations.

 0:225    0:55
p p
anb ¼ 55q_ 2=3 M1=2
w  log10 (19)
pcrit pcrit
The convective component is obtained from the saturated liquid
heat transfer coefficient (al), which is computed using Eq. (16) with
saturated liquid properties. This coefficient is then multiplied by a
factor (F) that depends on the steam quality (x) and the saturated
liquid and vapor densities (rl, rv) and viscosities (ml, mv):

acv ¼ F al (20)

 0:372
28
F¼ 1þ (21)
Xtt

 0:9  0:5  0:1

1x rv mv
Xtt ¼ (22)
x rl ml
Fig. 9. Discretization of the exhaust evaporator.

7
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Table 11 Table 12
Heat exchanger geometry for each element. Expander specifications.

Model TitanX Supply cut-off fa 0.16 e

Exhaust cut-off fp 0.78 e
Number of els. in x-dir. nel, x 6 9 e
Displaced volume Vexp 0.8 L
Number of els. in y-dir. nel, y 3 10 e
Number of fins nf, el 415 83 e
Base area Abase, el 3631 726 mm2
Fin area Afin, el 7864 1573 mm2
Heat transfer area As, el 11,495 2299 mm2

3.5. Expander

The expander is an uniflow reciprocating piston expander with

two cylinders of which the relevant geometrical specifications are
shown in Table 12.
The expander model is based on a semi-empirical model for
volumetric expanders [57] which is schematically shown in Fig. 11.
The semi-empirical model consists of thermodynamic equations
with tuning parameters that are determined by calibrating them Fig. 11. Schematic depiction of the semi-empirical model used for the expander [58].
against experimental results. Using these parameters, the de-
viations from ideal expander performance caused by pressure
drops, leakage, heat losses, and mechanical losses can be deter- which is used to calculate the expander outlet enthalpy:
mined. More details on the modeling, results, and validation of the
expander were presented in a previous publication [58]. hexp; out ¼ hexp; in  εis ðhexp; in  hexp; out; is Þ (29)
The operation of the expander is characterized by three main
The resulting model outputs and the corresponding experi-
performance factors: the isentropic filling factor (4f, is), the
mental values are shown on the left of Figs. 12 and 13, taken from
expander efficiency (hexp), and the isentropic effectiveness (εis). All
experimental set 6. Experiments were performed at four engine
three factors depend on the process conditions and expander
operating points (A25, HW, A50, and B25). In the experiments, the
operation. The isentropic filling factor (4f, is) is used to predict how
pump and pump bypass valve position were controlled to provide a
much the expander mass flow will deviate from ideal conditions
constant mass flow at each engine operating point, the expander
under isentropic outlet conditions. For a piston expander, the
speed was varied, and the expander bypass valve was closed. The
theoretical flow is equal to product of the expander speed (Nexp)
results show that the expander mass flow is well captured by the
and the difference between the available mass at inlet valve closing
isentropic filling factor. However, the shaft power is overpredicted
(rexp, infaVexp) and the trapped mass at exhaust valve closing (rexp,
for low pressure ratios (and corresponding low power outputs) and
outfpVexp). In all tested conditions, the expansion ended in the two-
underpredicted at higher pressure ratios (and corresponding high
phase region. Therefore, the outlet conditions are based on the
power outputs). Since the expander efficiency is not only depen-
isentropic conditions (rexp, out, isfpVexp). This leads to the following
dent on the pressure ratio, variations between experimental values
expression:
for similar pressure ratios occur. Other important physical quanti-
  ties include the expander speed, the cycle mass flow, and the
N
_ exp ¼ rexp; in fa  rexp; out; is fp 4f; is Vexp exp
m (27) expander inlet temperature. Deviations between model and
60 experimental values are mainly attributed to the expansion in the
_ exp ), the
To predict the shaft power output of the expander (W two-phase region, high leakage rate in the expander, and the
expander efficiency is used: change in lubrication properties over time. These topics are dis-
cussed in more detail in a separate publication [58].
_ exp ¼ h m
W exp _ exp ðhexp; in  hexp; out; is Þ (28)
3.6. Condenser
For the expander considered in this paper, not only the shaft
power output, but also the leakage and heat loss effects were sig- The condenser is modeled as a heat sink only because neither
nificant. Therefore, the isentropic effectiveness (εis) is introduced, detailed information on its geometry nor flow measurements on

Fig. 10. Evaporator heat transfer rate (left) and outlet temperature (right) as a function of the mass flow.

8
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Fig. 12. The isentropic filling factor (left) and mass flow (right) of the expander as functions of the expander speed. Symbols indicate experimental results and lines indicate model
outputs.

Fig. 13. The efficiency (left) and shaft power (right) of the expander as functions of the expander speed. Symbols indicate experimental results and lines indicate model outputs.

the process water used for cooling were available. Assuming no 3.7. Rankine cycle
leakage of the working fluid, the modeled condenser flow was
considered to be equal to the evaporator flow. This gives the Simulations using the full cycle model incorporating all of the
following relations for the condenser: calibrated component models discussed in the preceding sections
were performed in MATLAB [59] using fluid maps generated from
the CoolProP [60] database. Fig. 15 shows the comparison of these
m _ evap
_ cond ¼ m (30)
simulation results to data from experiment set 6 in Table 5. Ex-
periments were performed at four engine operating points (A25,
HW, A50, and B25). The pump speed and pump bypass valve po-
Q_ cond ¼ m
_ cond ðhcond; out  hcond; in Þ (31) sition were controlled to provide a constant mass flow, the
expander speed was varied, and the expander bypass valve was
Fig. 14 shows the experimental results of the condenser heat
closed.
transfer rate from experimental set 5 in Table 5 together with the
The comparison of the model output to the experimental values
corresponding outlet temperatures. Experiments were performed
shows that the pump outlet pressure is well captured. Deviations in
at all of the engine operating points shown in Figs. 4 and 5 (A25-
the mass flow are mostly due to the sensitivity of the pump bypass
C100). In these experiments, the pump and pump bypass valve
valve model; small pressure changes cause large changes in the
position were controlled to provide a constant mass flow at each
predicted flow through the bypass valve, leading to poor agreement
engine operating point. The expander was not running and the
between the model and experiment. Because the evaporator heat
expander bypass valve was used to control the pressure.

Fig. 14. Condenser heat transfer rate (left) and outlet temperature (right) as functions of the mass flow.

9
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Fig. 15. Comparison of full cycle model outputs and experimental results.

transfer rate is proportional to the mass flow, the same effect is speeds outside the range specified by the manufacturer.
visible there. Also visible is the overprediction of the evaporator
outlet temperature. Since no heat loss was considered in the 4.1. Working fluids
models, the evaporator outlet temperatures of the working fluid
almost reached the exhaust gas temperature. In reality, heat losses To evaluate the performance of the WHR system, simulations
would lead to slightly lower temperature, explaining the deviations were performed with two additional working fluids: cyclopentane
between simulations and experiments. Finally, the expander and ethanol. These fluids were selected based on their promising
output power is significantly overpredicted at low power outputs thermodynamic performance in heavy-duty engine applications
and underpredicted at higher power outputs; the reasons for this [17,61]. Previous studies by the authors [6,16,62], where different
are discussed in more detail in a separate publication [58]. heat sources from the engine were evaluated for many different
The main source of error between the experimental and simu- working fluids, also showed good thermodynamic performance for
lated values of the mass flow is the bypass valve model. A smaller these two fluids. Additionally, they are environmentally friendly,
pump in the experimental setup would possibly eliminate the need relatively non-toxic, and non-corrosive, although flammability is a
for a bypass valve. Alternatively, a more sophisticated model or concern for both of them. Table 13 lists a number of important
calibration method for the bypass valve could improve the fit be- properties for the three working fluids.
tween experiments and simulations. Using a different working fluid
could also help improve the fit. A fluid with a smaller latent heat
4.2. Driving cycle
would mean higher mass flows to extract the available heat.
Simultaneously, the calibration and validation for the heat
The model was calibrated against experimental data obtained in
exchanger and the expander models could be improved. Another
an engine test cell. However, temperatures under driving condi-
improvement would be the addition of pressure drop and heat loss
tions are usually lower due thermal inertia and heat loss in the
correlations to the models. Although the figures show that the
aftertreatment systems and exhaust piping. Therefore, the input
deviations between simulation and experimental results can be
conditions for the simulations were taken from a representative
significant, the simulation results are based on physical models and
driving cycle for a 40 tonne EU6 Scania long haul truck driving on a
the general trend is well-captured. This means that these models
European road. The vehicle speeds and road gradients for this
are a valid tool to compare the performance of different working
driving cycle are shown in Fig. 16.
fluids for operational range of the engine, which is done in the
The exhaust outlet conditions (temperature and mass flow)
subsequent sections.
during the driving cycle are divided into a four-by-four grid, as

4. Simulation setup Table 13

Properties of the selected working fluid.
The validated cycle model allows for predicting the performance
Fluid MW pcrit Tcrit GWP ODP Type
of the WHR system under conditions outside the experimentally


- kg/kmol bar C -
tested range (e.g. over a driving cycle) and with different working
fluids. For this purpose, the bypass valve is removed from the Cyclopentane 70.1 45.7 239 0 0 isentropic
original model, giving the cycle schematically depicted in Fig. 13. To Ethanol 46.1 62.7 240 0 0 wet
Water 18.0 220 374 0 0 wet
obtain the desired mass flow, the pump is allowed to operate at
10
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

shown in Fig. 18. The percentages in the grid represent the relative
duration of these conditions during the driving cycle. The green
values indicate the duration for a positive torque on the engine and
the red values indicate negative torque. The center points for all 16
elements were used as inputs for the steady-state simulations.
Transient effects during the driving cycle were not taken into
account.

4.3. Cycle performance

To run the simulations, a number of inputs and constraints were

specified, as summarized in Table 14. The exhaust mass flow,
temperature, and pressure were taken from the 16 previously Fig. 17. Representative driving cycle for a long haul truck on an European road.
defined operating conditions. Depending on whether the engine
torque was positive or negative, the expander was coupled directly
to the engine via a mechanical coupling or to an electrical gener-
ator. A subcooling temperature difference of 5 K was set to prevent
vapor entering the pump. The pump mechanical efficiency, elec-
trical generator efficiency, and the efficiency of the mechanical
coupling between expander and engine were taken to be 0.50, 0.85,
and 0.98, respectively. Only subcritical conditions were taken into
account. The evaporator outlet temperature was limited to avoid
thermal instability of the working fluid and overheating of sus-
pended oil. To allow for temperature control, a minimum and
maximum superheating temperature difference were set. The
range of expander speeds was based on the specifications from the
manufacturer [63]. The pump speed was not limited by these
specifications; instead it was set to give the highest possible mass
flow. No pressure drops in the components were considered and
component heat losses other than the expander heat loss were
ignored. A golden section search was performed to find the
expander speed providing the maximum power output for the
stated inputs and constraints.
Cycle performance is evaluated based on the net (shaft) power Fig. 18. Relative time distributions of the exhaust mass flow and temperature over the
and thermodynamic efficiency: driving cycle. Green values indicate positive torque and red negative torque. The total
drive cycle duration (Dtdc) is 2748 s. (For interpretation of the references to colour in
_ exp  W
_ net ¼ W _ pmp this figure legend, the reader is referred to the Web version of this article.)
W (32)

_
W The performance in the driving cycle is estimated using the
hth ¼ _ net (33) definitions expressed in Eqs. (34) and (35). At positive engine tor-
Q evap
que (teng), the expander power is directly provided to the engine,
while at negative engine torque the expander power is converted
into electrical power.

_ pmp
W
_
W pmp; el ¼ (34)
hel
(
_ _
_ exp ¼ W exp; mech ¼ hmech W exp ; if teng > 0
W _ _ exp ; (35)
W exp; el ¼ h el W if teng  0

5. Results and discussion

5.1. Steady-state performance

Steady-state simulations using the cycle model with water as

the working fluid were performed for 16 engine operating points
with exhaust mass flows (m _ exh ) ranging from 150 to 450 g/s and
exhaust outlet temperatures (Texh, out) between 260 and 320
C, as
previously presented in Fig. 18 and Table 14. In the following dis-
cussion, these will be designated with an M for mass flow and a T
for temperature. Thus, M150T300 corresponds to an exhaust mass
Fig. 16. Schematic of full cycle model. flow of 150 g/s and an outlet temperature of 300
C. The pump
11
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Table 14
Cycle inputs and constraints.

Inputs

Exhaust gas mass flow m_ exh 150e450 g/s

Exhaust gas inlet temperature Texh, in 260e320 C
Exhaust gas inlet pressure pexh, in 1.03e1.06 bar
Pump inlet subcooling temperature DTsub 5 K
Pump mechanical efficiency hpmp 0.50 e
Electrical generator efficiency hel 0.85 e
Mechanical coupling efficiency hmech 0.98 e

Constraints

Pump outlet pressure ppmp, out 10 e pcrit bar

Evaporator outlet temperature Tevap, out n/a e 260 C
Evaporator outlet superheating temp. DTsup 10e30 K
Expander speed Nexp 500e3500 rpm
Pump speed Npmp 150e6000 rpm

outlet pressures and mass flows at four engine operating points for Simulations were performed to obtain the maximum power
a range of expander speeds are shown in Fig. 19. In these cases, the output for the 16 engine operating points with water as the
evaporator superheating temperature difference was kept within working fluid. The expander speed was varied at each operating
the constraints specified in Table 14. At each engine operating point to obtain the maximum net power output, which is shown on
point, the exhaust temperature was kept constant while varying the left of Fig. 21. The values shown at the edges of this figure were
the mass flow, effectively changing the heat available for recovery. obtained by linear extrapolation. Depending on the exhaust mass
As expected, the pump outlet pressure decreased as the flow and temperature, the recoverable net power ranges from 0 to
expander speed increased and increased as the available exhaust 8 kW. Another important aspect for automotive applications is the
heat increased (represented by the engine operating points). An amount of heat that must be rejected to allow condensation of the
increase in the available exhaust heat allows for higher cycle mass working fluid, which is shown on the right of Fig. 21. When using
flows, as shown on the right of Fig. 19. The same effect was exhaust gases as a heat source, this heat must be either transferred
observed when increasing the expander speed. Increasing the to the coolant and rejected in the coolant radiator or rejected
expander speed (and thus reducing the evaporating pressure of directly via a separate radiator. The results of the simulations show
water in the cycle, as explained further on; see Fig. 22) made it that the heat transfer rate in the condenser can be as high as 60 kW.
possible to recover more heat from the exhaust gases. The resulting Simulations using the same cycle components were also per-
net power, shown on the left of Fig. 20, depends on the required formed with cyclopentane and ethanol as the working fluid, and
pump power, the amount of heat transferred from the heat source, the results all selected working fluids are shown in Table 15. The
and the expander power obtained. The pump power is determined condensation pressure was set at 1.1 bar for all fluids, resulting in a
by the pump efficiency, the pressure difference over the pump, and different condensation temperature for each fluid. A lower
the mass flow. The amount of heat transferred from the source to condensation temperature means a smaller temperature difference
the cycle is a function of the exhaust mass flow and temperature, as between the working fluid and the ambient temperature, making it
well as the cycle temperature, pressure, and mass flow. The more difficult to reject excess heat. In a practical system, this could
expander power depends on how effectively the recovered heat is lead to increased power consumption by the cooling fan, which
transformed into power, which is shown on the right of Fig. 20. would reduce the net power output of the system. This effect is not
Because of the interaction between the pump power, recovered taken into account here.
heat, and expander power, there is no single optimal expander The results from Table 15 show that the highest power output
speed that maximizes the power output for all engine operating was obtained with cyclopentane, then ethanol, and finally water.
points. For the lower exhaust mass flows, the maximum net power These differences in the performance can be explained by consid-
is around 1.1 kW and is achieved at a relatively low expander speed ering the M150T300 engine operating point, for which the results
of around 900 rpm. As the exhaust mass flow increases, both the are shown in Table 16.
power output and the optimal expander speed increase, with The power outputs for cyclopentane and ethanol were higher
maxima of 4.2 kW and 2800 rpm, respectively. than for water because of higher mass flows and expander

Fig. 19. Pump outlet pressure (left) and mass flow rate (right) for water as functions of the expander speed at different engine operating points.

12
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Fig. 20. Net shaft power (left) and expander efficiency (right) for water as functions of the expander speed at different engine operating points.

Fig. 21. Net shaft power (left) and condenser heat transfer rate (right) for water as a function of the exhaust mass flow and temperature. The values at the edges were linearly
extrapolated.

Fig. 22. QT-diagrams of cyclopentane, ethanol, and water for the M150T300 engine operating point.

Table 15
Range of cycle conditions for the 16 engine operating points.

Fluid _ pmp
m pevap Texp, in Nexp pcond Tcond Q_ cond _ net
W hth



- g/s bar C rpm bar C kW kW %

Cyclopent. 38.6e142 19.4e30.9 194e225 1285e3500 1.1 52 17.4e75.0 1.8e9.6 7.7e11

Ethanol 18.6e74.4 20.9e30.1 193e232 861e2774 1.1 80 14.6e66.3 1.0e7.8 5.9e9.8
Water 5.16e22.6 15.1e26.5 210e255 700e3139 1.1 102 10.2e45.8 0.5e5.7 3.7e10

efficiencies. The higher mass flows of cyclopentane and ethanol are temperature of the working fluid and the maximum heat can be
due to the lower latent heat of these fluids and to a better thermal extracted for this specific heat exchanger geometry. For ethanol and
match between the heat source and the cycle. This is visualized in especially for water, less heat can be extracted and a lower power
the heat transfer-temperature (QT) diagrams for the different output is achieved.
working fluids at the M150T300 engine operating point, which are The corresponding temperature-entropy (Ts) diagrams are
presented in Fig. 22. For cyclopentane, the temperature slope dur- shown in Fig. 23. This shows that the expansion ends in the two-
ing preheating matches the temperature profile of the heat source, phase region in the cases of ethanol and water whereas for cyclo-
meaning that the heat transfer is not limited by the evaporating pentane it ends in the superheated region. For most expanders,

13
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

Table 16
Cycle conditions for the M150T300 engine operating point.

Fluid _ pmp
m pevap Texp, in Nexp pcond Tcond Q_ cond _ net
W hth



- g/s bar C rpm bar C kW kW %

Cyclopentane 52.9 21.6 199 1546 1.1 52 27.7 2.75 8.7

Ethanol 25.2 23.4 214 1023 1.1 80 20.9 1.81 6.9
Water 7.45 18.7 228 1023 1.1 102 14.7 1.09 6.0

operation in the superheated region is preferable because it pre-

vents droplets from damaging the expander. Additionally, when
using cyclopentane, a recuperator could be added to improve the
power output or reduce the condenser load. Although the Ts-
diagrams suggest that the expander is operating at isentropic effi-
ciency, this is not the case. During the expansion, not all of the
available energy is converted to power; a significant portion is lost
as heat, which is why two definitions for the expander power and
energy change are necessary, as defined in Eq. (28) and Eq. (29).

5.2. Driving cycle performance

The exhaust flow and temperature over the driving cycle pre- Fig. 24. Net shaft power for multiple fluids during the driving cycle.
sented in Fig. 17 were used as inputs for the steady-state simula-
tions of the three selected working fluids. The resulting net power
outputs over the driving cycle of these fluids are shown in Fig. 24. Table 17
In accordance with the steady-state simulations, the best per- Driving cycle performance with a total engine work requirement of 333 MJ.

formance was obtained with cyclopentane, followed by ethanol and Fluid Qcond Wpmp, el Wexp, mech Wexp, el W *net
then water. To estimate performance over the whole driving cycle,
MJ MJ (%) MJ (%) MJ (%) MJ (%)
the results were numerically integrated using a timestep (Dt) of 1 s:
Cyclopentane 100 1.62 (0.49) 11.7 (3.50) 1.17 (0.35) 11.2 (3.37)
X
n Ethanol 86.1 0.90 (0.27) 8.42 (2.53) 0.69 (0.21) 8.21 (2.46)
W¼ _ Dt
W (36) Water 57.1 0.20 (0.06) 5.02 (1.51) 0.37 (0.11) 5.19 (1.56)
i
i¼1 *Wnet ¼ Wexp, mech þ Wexp, el  Wpmp, el.

The integrated results for Eqs. (31), (34) and (35) are shown in
Table 17, both in absolute and relative terms. The relative perfor-
Even though a driving cycle was used to evaluate the perfor-
mance is obtained by dividing the absolute result by the total work
mance of the working fluids over the operational range of the en-
done by the engine during the driving cycle (Wdc ¼ 333 MJ). The
gine during actual driving conditions, transient effects were
results show that the WHR system can recover a significant amount
ignored in this paper. The cycle components are assumed to react
of energy, corresponding to as much as 3.37% of the total engine
instantaneously to changes in the exhaust flow and temperature. In
energy requirement. This relative recovery can be roughly trans-
reality, the performance of the components will be affected by
lated into fuel consumption reduction, assuming that the increased
inertia during transient operation, with the thermal inertia in the
backpressure due to the exhaust evaporator does not affect the
heat exchangers being dominant [64]. This has important impli-
engine efficiency. The results also show that the recovered elec-
cations for the control of the system, as superheated conditions at
trical work is comparable to the electrical work done by the pump,
the inlet of the expander should be ensured [14]. Another study
although it must be noted that the efficiencies of storing and
[65] showed in a comparison between a steady state and transient
extracting power from the battery are not included. Additionally,
model that a transient model will lead to lower fuel savings,
the table shows that the pump work is much higher for cyclo-
although the resulting ranking remains the same. The expander
pentane than for the other fluids because of the higher mass flow in
coupling is another point of discussion. If the expander is me-
the cycle. However, this is more than offset by the increase in
chanically coupled to the engine, its speed will be determined by
expander power.

Fig. 23. Ts-diagrams of cyclopentane, ethanol, and water at the M150T300 engine operating point.

14
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

the engine speed and a predetermined gear ratio. However, the C The steady-state results for the three working fluids were
driving cycle results presented here were obtained under the used to simulate the performance over a typical driving cycle
assumption that the expander could operate at its optimal speed for of a long haul truck. Although transient effects were not
which the corresponding power output were determined in the taken into account and the expander speed was not
earlier steady-state simulations. The results in this section allow for controlled by the engine speed (as would be the case in a real
comparison of different working fluids on the basis of their ther- system due to the mechanical coupling), the results still
modynamic performance. In practice, however, the selection of the allow for a comparison between the thermodynamic per-
best working fluid is also subject to other constraints such as costs, formance of the systems with the different working fluids.
component sizing, and environmental impact. The total recovered energy during the driving cycle was 11.2,
8.2, and 5.2 MJ for cyclopentane, ethanol, and water,
respectively, corresponding to recoveries of 3.4, 2.5, and 1.6%
6. Conclusions relative to the total energy requirement of the engine.

Experimental investigations were performed to evaluate the Declaration of competing interest

performance of a Rankine system with water for WHR from the
exhaust gases of a heavy-duty engine. The results of these experi- The authors declare that they have no known competing
ments constitute one of the main contributions of this paper. financial interests or personal relationships that could have
Additionally, models of the relevant cycle components were appeared to influence the work reported in this paper.
developed and then calibrated and validated against the experi-
mental data. These models provided more detailed overview of the Acknowledgments
physical processes occurring within each component. This allowed
for predictions of the performance of these components under This research was made possible by funding provided by the
conditions outside the experimental range and when using Strategic Vehicle Research and Innovation Programme (FFI) of the
different working fluids. The component models were combined to Swedish Energy Agency. The authors would like to thank the
create model of the full cycle, allowing the performance of the partners in the WHR project: Gnutti Carlo, IAV, Lund University,
Rankine system to be simulated over a typical long haul truck Scania, TitanX, Volvo Cars, and Volvo Group.
driving cycle. The main results and conclusions obtained during
this work were: Nomenclature

C Experimental results were obtained for a wide range of en- A area (m2)
gine operating conditions. The experiments were divided Cd discharge coefficient ()
into six distinct experimental sets and their results were h specific enthalpy (J/kg)
used to calibrate and validate models of the main compo- MW molecular weight (kg/kmol)
nents of the Rankine cycle, i.e. the pump, pump bypass valve, m_ mass flow rate (kg/s)
evaporator, expander, and condenser. Experimental mea- N rotational speed (rpm)
surements of the expander shaft power were performed at p pressure (Pa)
four different engine operating points (A25, HW, A50, B25).
Q_ heat transfer rate (W)
The expander power ranged from 0.2 to 3 kW, corresponding
s entropy (J/kg/K)
to 0.2e2.5% of the engine power.
t time (s)
C Steady-state simulations of the Rankine cycle with water as
T temperature (K)
the working fluid exhibited good agreement with the
V volume (m3)
experimentally determined mass flow and evaporator heat
V_ volume flow rate (m3/s)
transfer, but the expander power was overpredicted at low
_
W power (W)
expander powers and underpredicted at high powers. Sim-
ulations performed at 16 engine operating points gave net
power outputs between 0.5 and 5.7 kW, and the optimal Greek symbols
expander speed was found to be dependent on the engine h efficiency ()
operating point. The added heat needing to be rejected in the ε effectiveness ()
condenser was between 10 and 46 kW. These values can be 4f filling factor ()
extrapolated to obtain results for the full range of operating r density (kg/m3)
conditions, yielding net power outputs between 0 and 8 kW t torque (Nm)
and condenser heat transfer rates between 0 and 60 kW.
C To evaluate the performance of different working fluids in Subscripts
the studied WHR system, simulations were also performed amb ambient
with cyclopentane and ethanol as the working fluids. The bpv bypass valve
results indicated that the evaporating pressures and cond condenser
expander inlet temperatures for these fluids were similar to cool coolant
those of water, but that they had higher mass flows in the corr correction
cycle. The increased mass flows were a result of the lower crit critical
latent heat and a better thermal match with the heat source,
allowing for more heat transfer between source and cycle. Subscripts (continued)
Because of the higher flows and expander efficiencies, el electrical
cyclopentane and ethanol outperformed water, providing eng engine
net power outputs between 1.8 and 9.6 kW and 1.0 and evap evaporator
7.8 kW, respectively. exh exhaust
15
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

exp expander €lsch B, Pischinger S, Bardow A. Integrated design of

[18] Schilling J, Eichler K, Ko
working fluid and organic Rankine cycle utilizing transient exhaust gases of
is isentropic
heavy-duty vehicles. Appl Energy 2019;255.
mech mechanical [19] Li X, Song J, Yu G, Liang Y, Tian H, Shu G, Markides CN. Organic Rankine cycle
pmp pump systems for engine waste-heat recovery: heat exchanger design in space-
sh shaft constrained applications. Energy Convers Manag 2019;199.
[20] B. Xu, D. Rathod, A. Yebi, S. Onori, Z. Filipi, and M. Hoffman, “A comparative
sub subcooled analysis of dynamic evaporator models for organic Rankine cycle waste heat
sup superheated recovery systems,” Appl Therm Eng, vol. 165, 2020.
th theoretical/thermodynamic [21] Alshammari F, Karvountzis-Kontakiotis A, Pesyridis A, Usman M. Expander
technologies for automotive engine organic rankine cycle applications. En-
ergies 2018;11.
[22] Guillaume L, Lemort V. Comparison of different ORC typologies for heavy-duty
Abbreviations
trucks by means of a thermo-economic optimization. Energy 2019;182.
CAC charge air cooler [23] Imran M, Haglind F, Lemort V, Meroni A. Optimization of organic rankine cycle
BPV bypass valve power systems for waste heat recovery on heavy-duty vehicles considering
EGR exhaust gas recirculation the performance, cost, mass and volume of the system. Energy 2019;180.
[24] Galuppo F, Nadri M, Dufour P, Reiche T, Lemort V. Evaluation of a coupled
EGRC exhaust gas recirculation cooler organic rankine cycle mild hybrid architecture for long-haul heavy-duty truck.
ESC European stationary cycle IFAC-PapersOnLine 2019;52.
HD heavy-duty [25] Huster WR, Vaupel Y, Mhamdi A, Mitsos A. Validated dynamic model of an
organic Rankine cycle (ORC) for waste heat recovery in a diesel truck. Energy
HDD heavy-duty Diesel 2018;151.
HW highway [26] Koppauer H, Kemmetmüller W, Kugi A. Modeling and optimal steady-state
GHG greenhouse gas operating points of an ORC waste heat recovery system for diesel engines.
Appl Energy 2017;206.
GWP global warming potential [27] Xu B, Rathod D, Kulkarni S, Yebi A, Filipi Z, Onori S, Hoffman M. Transient
ODP ozone depletion potential dynamic modeling and validation of an organic Rankine cycle waste heat
ORC organic Rankine cycle recovery system for heavy duty diesel engine applications. Appl Energy
2017;205.
SV safety valve
[28] Neunteufl K, Stevenson PM, Hülser H, Theissl H. Better fuel consumption by
WHR waste heat recovery waste heat recovery. MTZ worldwide 2012;73(12).
[29] Freymann R, Strobl W, Obieglo A. The turbosteamer: a system introducing the
principle of cogeneration in automotive applications. MTZ worldwide
References 2008;69(5).
[30] Freymann R, Ringler J, Seifert M, Horst T. The second generation turbosteamer.
[1] International Energy Agency. Energy technology Perspectives 2020. 2020. MTZ worldwide feb 2012;73.
[2] European Environment Agency (EEA). National emissions reported to the [31] Seher D, Lengenfelder T, Gerhardt J, Eisenmenger N, Hackner M, Krinn I.
UNFCCC and to the EU greenhouse gas monitoring mechanism. 2017. https:// “Waste heat recovery for commercial vehicles with a rankine process,” 21 st
www.eea.europa.eu/ds_resolveuid/a6e1bc85fbed4989b0fd6739c443739a. Aachen Colloquium Automobile and engine technology. 2012.
[Accessed 15 December 2020]. [32] Koeberlein D. Cummins SuperTruck Program. Technology and system Level
[3] European Parliament, Council of the European Union. Regulation (EU) 2019/ Demonstration of highly efficient and Clean, Diesel powered Class 8 trucks.
1242 - Setting CO2 emission performance standards for new heavy-duty ve- Department of Energy Annual Merit Review; 2014.
hicles. 2019. [33] Dickson J, Damon K. Cummins/peterbilt Supertruck II. Department of Energy
[4] Hofer F, Gruber W, Raser B, Theißl H. Technology scenarios for fulfilling future Annual Merit Review; 2019.
EU CO2 targets for commercial vehicle fleets. ATZheavy duty worldwide [34] Allain M, Atherton D, Gruden I, Singh S, Sisken K. Daimler's super truck
2020;13(1). program; 50 % brake thermal efficiency. In: 2012 Directions in engine-
[5] Pischinger S, Schaub J, Aubeck F, van der Put D. Powertrain concepts for efficiency and emissions research (DEER) Conference; 2012.
heavy-duty applications to meet 2030 CO2 regulations. ATZheavy duty [35] Endo T, Kawajiri S, Kojima Y, Takahashi K, Baba T, Ibaraki S, Takahashi T,
worldwide 2020;13(3). Shinohara M. Study on maximizing exergy in automotive engines. SAE Trans:
[6] Rijpkema J, Munch K, Andersson SB. Thermodynamic potential of twelve Journal of Engines 2007;116.
working fluids in Rankine and flash cycles for waste heat recovery in heavy [36] Marlok H, Bucher M, Ferrand N. Further Development of exhaust waste heat
duty diesel engines. Energy 2018;160. recovery, vol. 13. ATZheavy duty worldwide; 2020. 3.
[7] Aghaali H, Ångstro € m H-E. A review of turbocompounding as a waste heat [37] Thantla S, Fridh J, Erlandsson A, Aspfors J. Performance Analysis of volumetric
recovery system for internal combustion engines. Renew Sustain Energy Rev expanders in heavy-duty truck waste heat recovery. SAE Technical Paper
2015;49. 2019-01-2266; 2019.
[8] W. Bou Nader, J. Chamoun, and C. Dumand, “Thermoacoustic engine as waste [38] Carstensen A, Horn A, Klammer J, Gockel J. Waste heat recovery in Passenger
heat recovery system on extended range hybrid electric vehicles,” Energy Cars and trucks, vol. 80. MTZ worldwide; 2019. 4.
Convers Manag, vol. 215, 2020. [39] Ekstro€ m F. A mild hybrid SIDI turbo passenger car engine with Rankine waste
[9] Lan S, Smith A, Stobart R, Chen R. Feasibility study on a vehicular thermo- heat recovery. SAE Technical Paper 2019-24-0194; 2019.
electric generator for both waste heat recovery and engine oil warm-up. Appl [40] Rijpkema J, Ekstro €m F, Munch K, Andersson SB. Experimental results of a
Energy 2019;242. waste heat recovery system with ethanol using the exhaust gases of a light-
[10] Di Battista D, Fatigati F, Carapellucci R, Cipollone R. Inverted Brayton Cycle for duty engine. In: Proceedings of the 5th International Seminar on ORC po-
waste heat recovery in reciprocating internal combustion engines. Appl En- wer systems; 2019.
ergy 2019;253. [41] Bettoja F, Perosino A, Lemort V, Guillaume L, Reiche T, Wagner T. NoWaste:
[11] Güven M, Bedir H, Anlas G. Optimization and application of Stirling engine for waste heat Re-use for greener truck. Transportation Research Procedia
waste heat recovery from a heavy-duty truck engine. Energy Convers Manag 2016;14.
2019;180. [42] Furukawa T, Nakamura M, Machida K, Shimokawa K. A study of the rankine
[12] Trabucchi S, De Servi C, Casella F, Colonna P. Design, modelling, and control of cycle generating system for heavy duty HV trucks. apr 2014.
a waste heat recovery unit for heavy-duty truck engines. Energy Procedia [43] Yang K, Zhang H, Song S, Zhang J, Wu Y, Zhang Y, Wang H, Chang Y, Bei C.
2017;129. Performance analysis of the vehicle diesel engine-ORC combined system
[13] Xu B, Rathod D, Yebi A, Filipi Z, Onori S, Hoffman M. A comprehensive review based on a screw expander. Energies 2014;7.
of organic rankine cycle waste heat recovery systems in heavy-duty diesel [44] Zhang Y-Q, Wu Y-T, Xia G-D, Ma C-F, Ji W-N, Liu S-W, Yang K, Yang F-B.
engine applications. Renew Sustain Energy Rev 2019;107. Development and experimental study on organic Rankine cycle system with
[14] Feru E, Willems F, de Jager B, Steinbuch M. Modeling and control of a parallel single-screw expander for waste heat recovery from exhaust of diesel engine.
waste heat recovery system for euro-VI heavy-duty diesel engines. Energies Energy 2014;77.
2014;7(10). [45] Latz G, Erlandsson O, Skåre T, Contet A, Andersson S, Munch K. Performance
[15] Mahmoudi A, Fazli M, Morad MR. A recent review of waste heat recovery by analysis of a reciprocating piston expander and a plate type exhaust gas
Organic Rankine Cycle. Appl Therm Eng 2018;143. recirculation boiler in a water-based Rankine cycle for heat recovery from a
[16] Rijpkema J, Andersson S, Munch K. Thermodynamic cycle and working fluid heavy duty diesel engine. Energies 2016;9(7).
selection for waste heat recovery in a heavy duty diesel engine. SAE Technical [46] Shu G, Yu G, Tian H, Wei H, Liang X, Huang Z. “Multi-approach evaluations of a
Paper 2018-01-1371 2018. cascade-Organic Rankine Cycle (C-ORC) system driven by diesel engine waste
[17] Preißinger M, Schwo € bel JA, Klamt A, Brüggemann D. Multi-criteria evaluation heat: Part A e thermodynamic evaluations. Energy Convers Manag 2016;108.
of several million working fluids for waste heat recovery by means of Organic [47] Yu G, Shu G, Tian H, Huo Y, Zhu W. Experimental investigations on a cascaded
Rankine Cycle in passenger cars and heavy-duty trucks. Applied Energy; 2017. steam-/organic-Rankine-cycle (RC/ORC) system for waste heat recovery

16
J. Rijpkema, O. Erlandsson, S.B. Andersson et al. Energy 238 (2022) 121698

(WHR) from diesel engine. Energy Convers Manag 2016;129. [57] Lemort V. Contribution to the Characterization of Scroll Machines in
[48] Shi L, Shu G, Tian H, Huang G, Chen T, Li X, Li D. Experimental comparison Compressor and expander Modes. PhD thesis; 2008.
between four CO 2 -based transcritical Rankine cycle (CTRC) systems for en- [58] Rijpkema J, Andersson S, Munch K, Thantla S, Fridh J. Experimental investi-
gine waste heat recovery. Energy Convers Manag 2017;150. gation and modeling of a reciprocating piston expander for waste heat re-
[49] Guillaume L, Legros A, Desideri A, Lemort V. Performance of a radial-inflow covery from a truck engine. ” Applied Thermal Engineering; 2020.
turbine integrated in an ORC system and designed for a WHR on truck [59] MathWorks. MATLAB, Version R2019a. 2019.
application: an experimental comparison between R245fa and R1233zd. Appl [60] Bell IH, Wronski J, Quoilin S, Lemort V. Pure and pseudo-pure fluid thermo-
Energy 2017;186. physical property evaluation and the open-source thermophysical property
[50] Alshammari F, Pesyridis A, Karvountzis-Kontakiotis A, Franchetti B, library CoolProp. Ind Eng Chem Res 2014;53(6).
Pesmazoglou Y. Experimental study of a small scale organic Rankine cycle [61] Galuppo F, Reiche T, Lemort V, Dufour P, Nadri M. Organic Rankine Cycle
waste heat recovery system for a heavy duty diesel engine with focus on the based waste heat recovery modeling and control of the low pressure side
radial inflow turbine expander performance. Appl Energy 2018;215. using direct condensation and dedicated fans. Energy; 2020.
[51] Alshammari F, Pesyridis A. Experimental study of organic Rankine cycle sys- [62] Rijpkema J, Munch K, Andersson SB. Combining low- and high-temperature
tem and expander performance for heavy-duty diesel engine. Energy Convers heat sources in a heavy duty diesel engine for maximum waste heat recov-
Manag 2019;199. ery using rankine and flash cycles. In: Junior C, Dingel O, editors. “. Energy and
[52] Shu G, Zhao M, Tian H, Huo Y, Zhu W. Experimental comparison of R123 and thermal management, air-Conditioning, and waste heat Utilization. Springer
R245fa as working fluids for waste heat recovery from heavy-duty diesel International Publishing; 2019.
engine. Energy 2016;115. [63] Bredel E, Nickl J, Bartosch S. Waste heat recovery in drive systems of Today
[53] Taylor J. An introduction to error Analysis: the study of Uncertainties in and Tomorrow, vol. 72. MTZ worldwide; 2011.
physical measurements. University Science Books; 1997. [64] S. Quoilin, R. Aumann, A. Grill, A. Schuster, V. Lemort, and H. Spliethoff,
[54] “DieselNet: Emission test cycles - European Stationary Cycle (ESC).” https:// “Dynamic modeling and optimal control strategy of waste heat recovery
www.dieselnet.com/standards/cycles/esc.php. Accessed: 2020-07-02. Organic Rankine Cycles,” Appl Energy, vol. 88, 2011.
[55] Danfoss. Nessie high pressure pumps for technical water, type PAH. Data [65] Grelet V, Reiche T, Lemort V, Nadri M, Dufour P. Transient performance
Sheet; 2007. evaluation of waste heat recovery rankine cycle based system for heavy duty
[56] Webb R, Nae-Hyun K. Principles of Enhanced heat transfer. CRC Press; 2005. trucks. Appl Energy 2016;165.

17