A Comparison of the Dynamic Temperature Responses of Two
Different Heat Exchanger Modelling Approaches in Simulink
Simscape for HVAC Applications
Samuel F. Fux, Babak Mohajer and Stefan Mischler
Belimo Automation AG, Brunnenbachstrasse 1, 8640 Hinwil, Switzerland
Keywords: HVAC, Heat Exchanger, Dynamic Response, Dynamic Simulation Models, MATLAB/Simulink/Simscape.
Abstract: In the HVAC industry, the dynamic temperature response of water-to-air heat exchangers is of particular
importance for control system design. In this paper, the dynamic temperature responses of two established
thermal dynamic modelling approaches for heat exchangers, the single-segment modelling using the
effectiveness-NTU method and the multi-segment modelling, are investigated. Both approaches are validated
against experimental data recorded with two different heat exchangers used in HVAC systems. A quasi-static
analysis reveals minor differences between the results of the two models considered. The dynamic analysis is
performed with varying inlet conditions. First results show that the single-segment model may fail to properly
reproduce the water outlet temperature dynamics of a heat exchanger under certain conditions. In the tests
performed in this study, however, the multi-segment model captures the relevant dynamics. The influence of
this difference in the dynamic behaviour of the single-segment model on the model-based development of
control algorithms is subject of future studies.
1 INTRODUCTION
In heating, ventilation, and air conditioning (HVAC)
systems in buildings, typically, water-to-air heat
exchangers are used to condition the temperature of
the supply air. To increase the occupant comfort in
the building and to decrease the energy use of the
HVAC system, optimal operation of the heat
exchanger is crucial. The development and testing of
new control algorithms for the operation of HVAC
heat exchangers can be simplified and accelerated by
using building simulation environments with an
appropriate dynamic model of the heat exchanger
(Zhou and Braun, 2004). An appropriate dynamic
heat exchanger model must be able to accurately
predict the water and air outlet temperatures not only
during steady state operation but also during
transients when the inlet conditions are changing.
Furthermore, it is important that the model is accurate
across the entire operating range and not only at full
load.
In the past decades, authors presented different
approaches for modelling the dynamic temperature
behaviour of HVAC water-to-air heat exchangers for
the purpose of control performance analysis. For
instance, (Underwood, 1990) developed a heat
exchanger model based on single energy balance
equations for the water and the air side, respectively.
The same lumped parameter approach is also
employed by, e.g., (Zajic, Larkowski, Sumislawska,
Burnham, and Hill, 2011) and (Afram and Janabi-
Sharifi, 2015).
In (Zhou and Braun, 2004), a model is presented
where the heat exchanger is divided into a series of
basic elements. Each basic element represents a
cross-flow finned tube. Then, a transient model for
the basic element is introduced, which considers
energy storage in the water and in the tube and fin
material. This approach of discretising the heat
exchanger into multiple smaller elements is adopted
by (Jie and Braun, 2016).
A completely different approach is introduced by
(Anderson, 2001), who proposes a linear model at an
operating point. This model is formed by combining
several first-order transfer functions and time delays.
The goal of this paper is to assess the suitability of
two popular modelling approaches for testing control
algorithms. For that purpose, two different heat
exchanger models are compared with temperature
measurement data from real heat exchangers. Both
Fux, S., Mohajer, B. and Mischler, S.
A Compar ison of the Dynamic Temperature Responses of Two Different Heat Exchanger Modelling Approaches in Simulink Simscape for HVAC Applications.
DOI: 10.5220/0012132300003546
In Proceedings of the 13th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2023), pages 417-424
ISBN: 978-989-758-668-2; ISSN: 2184-2841
Copyright
c
2023 by SCITEPRESS Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)
417
models are implemented in the MATLAB®
Simulink® simulation environment using the
Simscape modelling language (The MathWorks, Inc.,
2023a).
The paper is structured as follows: In Section 2,
an overview on the structure of HVAC heat
exchangers is provided. Section 3 introduces the two
models considered in this study. The comparison of
the two models with measurement data is shown in
Section 4. Finally, the conclusions of this study are
summarized in Section 5.
2 STRUCTURE OF HVAC HEAT
EXCHANGERS
Heat exchangers are usually characterized in one of
the four major construction types: tubular, plate-type,
extended surface, and regenerative heat exchangers
(Shah and Sekulić, 2003).
One of the commonly used water-to-air heat
exchanger types in the HVAC industry are the
extended surface exchangers. To compensate the low
heat transfer coefficient on the air side and to increase
the efficiency, fins are being used to extend the
surface area up to 5 12 times the primary surface
area (Shah and Sekulić, 2003), see Figure 1.
Air Flow
Figure 1: Finned tubular cross flow heat exchangers with
both fluids unmixed.
This paper focuses on the modelling of fin and
tube heat exchangers and their circuiting pattern. The
circuiting pattern of the water tubes is one of the
important aspects which influences the performance
of the heat exchanger. The number of tube rows mark
the depth and the number of tubes in each row mark
the height of the heat exchanger. Furthermore, the
number of circuits in a heat-exchanger is defined by
the number of tubes that are fed by the supply header
(Campbell Sevey, 2023). As an example, a quarter
circuit heat exchanger with two feeds is shown in
Figure 2.
Height: 8 Tubes
Depth: 4 Rows
Supply
Return
Air Flow
Figure 2: Quarter circuit heat-exchanger (adapted from
(Campbell Sevey, 2023)).
Determining the proper number of circuits, the
speed of the fluid in the tubes as well as the resulting
pressure drop can be controlled. This both aspects
have a direct impact on the heat transfer efficiency of
the heat exchanger (Campbell Sevey, 2023).
While talking about the circuiting it is important
to also consider how often the air crosses the inner
tubes perpendicularly (number of passes).
Common flow arrangements of water-to-air heat
exchangers are: (VDI Gesellschaft Verfahrenstechnik
und Chemieingenieurwesen, 2019):
multi-row, single-pass
multi-row, multi-pass
multi-row, two-pass
In Figure 3, a flow scenario with 4 rows and 4 passes
is illustrated. In this flow arrangement the tube rows
are connected in series with alternating flow
directions in each row. The outside air crosses the
tubes 4 times.
Air Flow
Liquid Flow
Figure 3: Crossflow definition for 4 rows and 4 passes
(adapted from (VDI Gesellschaft Verfahrenstechnik und
Chemie-ingenieurwesen, 2019)).
Considering the above-mentioned aspects, it is
crucial to take the circuiting arrangement into account
when modelling a heat exchanger.
SIMULTECH 2023 - 13th International Conference on Simulation and Modeling Methodologies, Technologies and Applications
418
3 HEAT EXCHANGER MODELS
As mentioned in the introduction, there exists a
variety of dynamic models to predict the temperature
behaviour of HVAC water-to-air heat exchangers.
However, these models are usually based on one of
the following basic approaches:
- Simple single-segment models, in which the water
and air sides are entirely lumped into single
thermal capacities each.
- Detailed multi-segment models, in which the heat
exchanger is divided into a fixed number of
smaller segments.
Furthermore, the various models differ in how
detailed they compute the heat transfer between the
water and the air side and if they consider the heat
capacity of the tube and fin material or not.
In this paper, an example of a single-segment
model and an example of a multi-segment model are
evaluated using measurement data. The two models
are presented in the following two subsections.
3.1 Single-Segment Model
The first model considered in this study is the heat
exchanger model provided with the Simscape Fluids
component library (The MathWorks, Inc., 2023b).
Since only one energy conservation equation is
formulated for each of the water and air sides, this
model belongs to the category of single-segment
models.
This model uses the effectiveness-NTU method to
compute the heat transfer between the water and the
air side. It can also consider water condensation on
the air side. However, the thermal capacity of the
tubes and fins are not considered. In addition to the
temperature dynamics, the model also computes the
pressure losses across the heat exchanger. Please refer
to (The MathWorks, Inc., 2023c) for a detailed
description of the model.
Important input parameters required by the model
include the following: flow arrangement and
geometries, length of the tubes, tube outer diameter,
number of rows, number of tubes per row,
longitudinal and transverse tube spacing, fouling
factors, total fin surface area, and constant fin
efficiency. The model does not require the input of
parameter values which need to be identified using
measurement data (e.g., convective heat transfers
coefficients).
3.2 Multi-Segment Model
The second model used in this study is a multi-
segment model developed by the authors.
The development of this model is based on the
assumptions typical for this type of model (e.g.,
(Zhou and Braun 2004) or (Mathisen, Morari, and
Skogestad, 1994)):
all segments have the same size,
ideal mixing of the water in each segment,
ideal mixing of the air in each segment,
the tube and fin material separating water and air
side have a uniform temperature distribution in
each segment,
heat conduction between adjacent segments is
neglected,
heat losses from the heat exchanger to the
surroundings are not considered.
The division of a heat exchanger into smaller
segments is exemplified in Figure 4. The schematic
of one segment of the multi-segment model is then
depicted in Figure 5.
Water flow
Air flow
Figure 4: Example of the division of a heat exchanger into
multiple smaller segments.
Figure 5: Schematic of the i-th segment of the multi-
segment heat exchanger model.
𝑇
,

𝑡
𝑇
,

𝑡
𝑚

,
𝑡
𝑇
,

𝑡
𝑇
,

𝑡
𝑚
,
𝑡
A Comparison of the Dynamic Temperature Responses of Two Different Heat Exchanger Modelling Approaches in Simulink Simscape for
HVAC Applications
419
In this model, energy conservation equations are
formulated for the water side, the air side and for the
metal mass. Obviously, the energy conservation
equations account for the fluid flows through the
segment and the convective heat transfers between
the fluids and the metal surfaces.
The required geometric dimensions of the heat
exchanger are computed in the model as presented in
(Réz, 2004) and (Vetter, 2014). The fin efficiency of
the continuous plate fins is calculated considering an
equivalent fin radius as given in (Perrotin and Clodic,
2003).
Finally, the convective heat transfer coefficient on
the air side is computed using the equivalent diameter
𝐷
and the appropriate empiric correlations for the
Nusselt number (Réz, 2004).
The multi-segment model is implemented in
Simulink® using the Simscape modelling language,
too. Except for the calculation of the convective heat
transfer coefficient on the air side, the
implementation of the model is based on component
models available in the Simscape Foundation library.
Consequently, the model also considers water
condensation.
This model requires similar input parameters as
the single-segment model presented above. However,
it must be mentioned that the fin efficiency is not
required since it is calculated online in the model.
4 TEST RESULTS
The two models presented above are evaluated using
measurement data recorded with one heating and one
cooling heat exchanger. Some properties of the two
heat exchangers are given in Table 1 and Table 2,
respectively.
The heating heat exchanger is installed in a test rig
that is located in an indoor room. Therefore, the inlet
air to the heat exchanger is the same as the room air.
For this reason, the inlet air temperature is
unrealistically high for an HVAC heating application.
However, since the temperature dependency of the air
properties are considered in both models, this is not
deemed to be a problem for the validation of the
models.
On the other hand, the cooling heat exchanger is
installed in a real ventilation system that is used for
the conditioning of a break room in an office building.
Both heat exchangers are equipped with
temperature sensors at the inlet and outlet of the water
side and the air side. For the air temperature sensors,
it must be pointed out that they measure the air stream
temperature only at one location in the air duct.
Therefore, if the air stream does not have a uniform
temperature distribution across the duct cross section,
the accuracy of the temperature measurement must be
challenged.
Table 1: Properties of the heating heat exchanger.
Pro
p
ert
y
Value
Height 0.398
m
Width 0.7
m
Number of rows 2
Number of tubes
p
er row 16
Number of circuits 8
Table 2: Properties of the cooling heat exchanger.
Property Value
Height 0.24
m
Width 0.369
Number of rows 4
Number of tubes
p
er row 6
Number of circuits 1
Furthermore, there are water volume flow rate
meters in the connecting water pipes of both heat
exchangers. Additionally, measurement data of the
air volume flow rate through the cooling heat
exchanger is available. The heating heat exchanger is
not equipped with a sensor measuring the air volume
flow rate. Therefore, the air volume flow rate must be
estimated based on an energy balance across the heat
exchanger.
For the evaluation, various measurement data
sequences are recorded with the heat exchangers. The
recorded measurement data of the inlet conditions,
i.e., water and air inlet temperatures and water and air
volume flow rates, is then used as inlet conditions
during the simulations with the two models. Finally,
the simulated outlet temperatures are compared with
the measured outlet temperatures.
In the so-called quasi-static tests, the water
volume flow rate through the heat exchangers is
slowly ramped up and down to avoid the excitation of
dynamics in the heat exchangers. The other inlet
conditions are kept as constant as possible during the
recording of these measurement sequences. These
quasi-static tests are used to assess the accuracy of the
models across a wide range of operating points. For
assessing the ability of the models to correctly capture
the temperature dynamics, tests with fast varying inlet
conditions are used. These tests are called dynamic
tests in the following.
The comparison of the heat exchanger models
with the measurement data of the quasi-static and the
dynamic tests are shown in the following two
subsections.
SIMULTECH 2023 - 13th International Conference on Simulation and Modeling Methodologies, Technologies and Applications
420
4.1 Quasi-Static Tests
The inlet conditions during the quasi-static test with
the heating heat exchanger are shown in Figure 6.
Figure 6: Inlet conditions for the quasi-static test with the
heating heat exchanger.
The comparison of the two models with the
measurement data of the water and air outlet
temperatures is then plotted in Figure 7.
Figure 7: Comparison of both models with measurement
data from the quasi-static test with the heating heat
exchanger.
In addition, the root mean squared errors
(RMSEs) between the measured and the simulated
water and air outlet temperatures are listed in Table 3.
Table 3: Root mean squared errors of both models for the
quasi-static tests.
Heat
exchanger
Outlet
temperature
Model
Single-
segment
Multi-
segment
Heating
Water 0.68 °C 0.83 °C
Air 0.77 °C 0.78 °C
Cooling
Water 0.25 °C 0.34 °C
Air 0.71 °C 0.53 °C
Figure 8 shows then the inlet conditions for the
quasi-static test with the cooling heat exchanger and
the corresponding comparison of the two models with
the measured outlet temperatures is shown in
Figure 9. The RMSEs of both models for this test are
given in Table 3, too.
Figure 8: Inlet conditions for the quasi-static test with the
cooling heat exchanger.
These results show that both models can
accurately predict the water outlet temperature for
quasi-static inlet conditions across a wide range of
operating points. During most of the time, the
deviation from the measured water outlet temperature
is smaller than 1 °C for both models. Only at low
water volume flow rates, where the water flow is
modelled to be laminar, the deviation increases to
values above 1 °C. In addition, it is assumed that at
low water volume flow rates the heat losses to the
surroundings are not negligible.
A Comparison of the Dynamic Temperature Responses of Two Different Heat Exchanger Modelling Approaches in Simulink Simscape for
HVAC Applications
421
Figure 9: Comparison of both models with measurement
data from the quasi-static test with the cooling heat
exchanger.
For the air outlet temperature, both models seem
to be a bit less accurate. However, it must be
remembered that the accuracy of the air outlet
temperature measurement must be questioned as
explained above.
4.2 Dynamic Tests
To validate the dynamic behaviour of the heat
exchanger models, water volume flow rate step
changes are applied to the heat exchangers. In the
following, the results of these dynamic tests are
presented.
The inlet conditions during a segment of the
dynamic test with the heating heat exchanger and the
corresponding comparison of the two models with the
measured outlet temperatures are shown in Figure 10
and Figure 11, respectively. The RMSEs of the
simulated water and air outlet temperatures for the
complete dynamic test are given in Table 4. Please
note that the dynamics of the temperature sensors are
considered in the simulations.
These results suggest that for this heating heat
exchanger even the simple single-segment model can
capture the relevant dynamics of both the water and
the air outlet temperatures.
Figure 10: Inlet conditions during a segment of the dynamic
test with the heating heat exchanger.
Figure 11: Comparison of both models with measurement
data from the dynamic test with the heating heat exchanger.
Table 4: Root mean squared errors of both models for the
dynamic tests.
Heat
exchanger
Outlet
temperature
Model
Single-
segment
Multi-
segment
Heating
Water 1.12 °C 1.18 °C
Air 0.91 °C 0.85 °C
Cooling
Water 0.34 °C 0.35 °C
Air 0.46 °C 0.47 °C
SIMULTECH 2023 - 13th International Conference on Simulation and Modeling Methodologies, Technologies and Applications
422
Figure 12 shows then the inlet conditions during a
segment of the dynamic test with the cooling heat
exchanger. The corresponding comparison of the two
models with the measured outlet temperatures is
plotted in Figure 13. The RMSEs of both models for
the complete dynamic test are also provided in
Table 4. These numbers let assume that both models
are equally able to capture the relevant dynamics of
the outlet temperatures. However, as can be seen in
Figure 12, the water inlet temperature is subject to
oscillating disturbances during the test because the
water inlet temperature controller failed to keep a
constant temperature. These disturbances reveal that
the simple single-segment heat exchanger model fails
in this case to always reproduce the dynamics of the
water outlet temperature properly. The comparison of
the simulated water outlet temperature of the single-
segment model with the measurement data in
Figure 13 indicates a time shift of the simulated
response with respect to the measured one for certain
inlet conditions. The multi-segment model, on the
other hand, can capture these dynamics.
Figure 12: Inlet conditions during a segment of the dynamic
test with the cooling heat exchanger.
5 CONCLUSION
The subject of this paper is to investigate two
different approaches of modeling an HVAC water-to-
air heat exchanger and to compare their dynamic
temperature responses for variations in inlet
conditions. For this assessment, measurement data
recorded with one heating and one cooling heat
exchanger are used.
Figure 13: Comparison of both models with measurement
data from the dynamic test with the cooling heat exchanger.
The single-segment model delivered by the
Simscape Fluids component library and the detailed
multi-segment model developed by the authors are
showing overall good accuracy for quasi-static inlet
conditions across a wide range of operating points.
For the heating heat exchanger, both models show
comparable dynamic behavior. Hence, in this case,
the effort for developing the very detailed multi-
segment model and its additional computational
burden during a simulation are not justified.
For the cooling heat exchanger, however, a
noticeable time shift between the water outlet
temperature response of the single-segment model
and the measurement data is observed for certain
variations in inlet conditions, while the multi-
segment model remains accurate. Reasons for this
could be that the multi-segment model additionally
considers the thermal heat capacity of the tube and fin
material and that it represents the transport delay in
the tubes more accurately due to the discretization
approach.
It must be emphasized that the analysis to
substantiate this assumption is not yet complete. To
confirm this initial hypothesis, tests with further
variations of the inlet conditions must be performed.
Finally, so far, the models have been evaluated
only with open loop tests. Additionally, closed loop
studies need to be performed to examine the impact
of the time shift of the single-segment model on
control loop analysis.
A Comparison of the Dynamic Temperature Responses of Two Different Heat Exchanger Modelling Approaches in Simulink Simscape for
HVAC Applications
423
REFERENCES
Afram, A, Janabi-Sharifi, F. (2015). Gray-box modeling
and validation of residential HVAC system for control
system design. Applied Energy, 137, 134 – 150.
Anderson, M. L. (2001). MIMO robust control for heating
ventilating and air conditioning (HVAC) systems
[Master's thesis, Colorado State University].
Campbell Sevey (2023, April 1). A simple guide to coil
circuiting. https://www.campbell-sevey.com/a-simple-
guide-to-coil-circuiting/.
Jie, C., Braun, J. E. (2016). Self-learning backlash inverse
control of cooling or heating coil valves having
backlash hysteresis. International Refrigeration and Air
Conditioning Conference, West Lafayette.
Mathisen, K. W., Morari, M., Skogestad, S. (1994),
Dynamic Models for heat exchangers and heat
exchanger networks. Computers & Chemical
Engineering, 18, 459 – 463.
Perrotin, T., Clodic, D. (2003). Fin efficiency calculation in
enhanced fin-and-tube heat exchangers in dry
conditions. 21st
IIR International Congress of
Refrigeration: Serving the Needs of Mankind,
Washington DC.
Réz, I. (2004). Numerische Untersuchung des luftseitigen
Wärmeübergangs und Druckverlustes in Lamellenrohr-
Wärmeübertragern mit verschiedenen Rohrformen
[Doctoral dissertation, Technische Universität
Bergakademie Freiberg].
Shah, R. K., Sekulić, D. P. (2003). Fundamentals of heat
exchanger design, John Wiley & Sons, Inc.
The MathWorks, Inc. (2023a, April 5). Simscape. Model
and simulate multidomain physical systems. https://
ch.mathworks.com/products/simscape.html.
The MathWorks, Inc. (2023b, April 5). Simscape Fluids.
https://ch.mathworks.com/products/simscape-
fluids.html.
The MathWorks, Inc. (2023c, April 5). Heat exchanger (tl-
ma). https://ch.mathworks.com/help/hydro/ref/
heatexchangertlma.html
Underwood, D. (1990). Modeling and nonlinear control of
a hot-water-to-air heat exchanger. Final report. United
States.
Vetter, C. (2014). Thermodynamische Auslegung und
transiente Simulation eines überkritischen Organic
Rankine Cycles für einen leistungsoptimierten Betrieb
[Doctoral dissertation, Karlsruher Institut für
Technologie].
VDI Gesellschaft Verfahrenstechnik und
Chemieingenieurwesen (2019). VDI heat atlas (12th
edition). Springer.
Zajic, I., Larkowski, T., Sumislawska, M., Burnham, K. J.,
Hill, D. (2011). Modelling of an air handling unit: A
Hammerstein-bilinear model identification approach.
21st International Conference on Systems Engineering,
Las Vegas.
Zhou, X., Braun, J. E. (2004). Transient modeling of chilled
water cooling coils, International Refrigeration and Air
Conditioning Conference, West Lafayette.
SIMULTECH 2023 - 13th International Conference on Simulation and Modeling Methodologies, Technologies and Applications
424