energies
Article
Determination of Heat Losses from the Pipeline in SDHW System
during the Continuous Change of the Supply Temperature
Miroslaw Zukowski






Citation: Zukowski, M.
Determination of Heat Losses from
the Pipeline in SDHW System during
the Continuous Change of the Supply
Temperature. Energies 2021, 14, 90.
https://dx.doi.org/10.3390/en14
010090
Received: 16 November 2020
Accepted: 22 December 2020
Published: 26 December 2020
Publisher’s Note: MDPI stays neu-
tral with regard to jurisdictional claims
in published maps and institutional
affiliations.
Copyright: © 2020 by the author. Li-
censee MDPI, Basel, Switzerland. This
article is an open access article distributed
under the terms and conditions of the
Creative Commons Attribution (CC BY)
license (https://creativecommons.org/
licenses/by/4.0/).
Energies 2021, 14, 90. https://dx.doi.org/10.3390/en14010090
Department of HVAC Engineering, Faculty of Civil Engineering and Environmental Sciences,
Bialystok University of Technology, Wiejska 45E Street, 15-351 Bialystok, Poland; m.zukowski@pb.edu.pl
Abstract: In this article, the research object is the solar domestic hot water (SDHW) heating system
that has been in operation since 2015 and is located on the campus of the Bialystok University of
Technology (Poland). The thermal performance of solar collectors are thoroughly investigated so
far. Therefore, special attention was paid to the issue of the heat loss from pipes. The measurements
showed that the heat transfer in circulation pipes is quite complex due to continuous fluctuations in
water temperature at the supply of this loop. As it turned out, the application of the classical method
of energy balancing and the readings from heat meters gave inaccurate results in this case. The main
aim of this study was to develop a different approach to solving the problem of determination of
heat losses. The method presented in this article is based on computational fluid dynamics (CFD)
and measurement results as the input data. The practical result of this study was the development of
two relationships for calculating the heat loss from pipes. A separate issue, that is discussed in this
paper, concerns the impact of the time intervals used in numerical simulations on the accuracy of
calculation results.
Keywords: heat transfer; heat loss; solar domestic hot water system; computational fluid dynamics
1. Introduction
Solar thermal technologies are widely used in small and medium scale in single-family,
multi-family, hospitals, public, and office buildings. Domestic hot water (DHW) heating
systems supported by flat plate or evacuated tube solar collectors are most commonly used
in Europe and China. The amount of converted solar radiation energy received directly by
the final recipient depends on the efficiency of the designed system and climatic conditions.
It usually covers between 30% and 50% of the energy demand for heating the tap water. The
devices used to acquisition energy emitted by the Sun and transfer it in a planned manner
to the recipients have already achieved very high level of energy conversion. Therefore,
improvements should be sought within another field that offers potential opportunities for
increasing the energy efficiency of the solar installations. One of these areas may be the
reduction of heat loss from pipes connecting the system elements.
Beckman [1] was the first to draw attention to the significant impact of heat loss
of the pipes, connecting the solar collector with the storage tank, on the overall system
performance. In a short technical note he gave a computational example proving his
thesis. The model was simplified and did not take into account the continuous temperature
changes of the medium circulating in the solar circuit.
Experimental studies carried out by Tsuda et al. [2] showed that heat loss of pipes
in the solar collector loop is higher during system operation than it results from design
calculations. The authors of this study suggested designing as short as possible connections
between collectors and storage tanks. Unfortunately, the article did not contain informa-
tion on the measurement methods, measuring equipment, pipe diameter and insulation
thickness.
Stubblefield et al. [3] considered the important issue of contact resistance between
insulation and pipe, which is called the fourth boundary condition. The impact of this air
https://www.mdpi.com/journal/energies
Energies 2021, 14, 90 2 of 21

gap thickness at the insulation-pipe interface on the change in thermal resistance based on
the experimental research was determined in this work. The working medium was not
water but saturated steam. Mineral wool with unusually very high thermal conductivity
coefficient of 0.052 W/(mK) was used as an insulating material.
Marshall [4] developed a steady-state solar collector model that includes heat loss
of pipes in the total energy balance of the system. The number of transfer units (NTU)
method was used for this purpose. In the summary of this analysis, it was found that it is
not possible to separate the pipe heat loss from the collector loss without further measuring
the ambient temperature of the pipes. Moreover the use of high-quality pipe insulation
will improve the thermal performance of the entire system of water heating. The research
was carried out for a very small system consisting of collectors with an area of 1 m2 to 6 m2 ,
and a short pipe length of 4 to 6 m with a diameter of 20 mm.
The heat loss assessment of pipelines located in the field of solar collectors was made
by El-Nashar [5]. The research object included 1064 evacuated tube collectors with absorber
area of 1.75 m2 of each. The solar desalination plant was located in Abu Dhabi (UAE). The
dependence of the pipeline heat loss on x-parameter expressed in m2 ·◦ C/W and the piping
area ratio was the result of measurements. It turned out, that the effect of the heat loss on
the solar collector efficiency was significantly smaller during the midday and higher in the
morning and afternoon. The solar system, that was the subject of El-Nashar’s research,
was very different from that presented in the current article. It was designed to work with
a seawater evaporation system, the diameters of the pipes were different, and the outside
air temperature was much higher.
The effectiveness of 2 mm coatings made of polyvinyl chloride as thermal insulation
of hot-water piping was tested by Deeble [6]. The experimental research and computer
modelling showed that this type of insulation allows a noticeable reduction in heat losses.
This is mainly due to the relatively low emissivity of the outer surface of PVC insulation.
The tests were performed only in laboratory and under steady heat exchange conditions.
The problem of determining the optimal thickness of insulation of pipes transporting
hot water inside the building envelope was investigated by Kecebas [7]. He used the exergy
method and life-cycle cost concept in this analysis. As a conclusion, Kecebas stated that the
optimum thickness of the pipes insulation for the exergy optimization is higher compared
to the result of traditional energy-economic optimization. It was only a theoretical analysis
that assumed a constant temperature of the transported medium and the air surrounding
the pipe.
Kohlenbach et al. [8] conducted a study on the impact of insulation material (mineral
wool soaked with thermal oil) on the heat loss of solar circulation pipes. The heat transport-
ing medium was oil at 175 ◦ C and 250 ◦ C. It was observed that the thermal conductivity
increases 2.5 times when the oil content in mineral wool insulation was 33% and more than
three times when this content exceeded 50%. The experimental tests were carried out in
the laboratory condition, on a short pipe section of 0.7 m, and a relatively large diameter of
80 mm.
The potential improvements in a typical solar hot water system located in a multi-
family building were presented by Springer et al. [9]. Particular attention was paid to heat
loss resulting from circulation of hot water in the solar circuit. The potential for savings
resulting from the proposed modifications was over 30% for Concord location (USA) and
over 20% for Lafayette site (USA). Unfortunately, the diameters of pipelines and their
lengths were not included in the test report.
Yu et al. [10] studied a small SDHW system designed for single-family housing.
As a result of this analysis, it was determined that the heat loss through the pipes is
approximately 28% of the energy obtained from solar collectors. However, the article does
not describe the heat transfer model that was used to obtain this value, so it is difficult to
determine the accuracy and reliability of these estimates.
Thermal performances of the solar collector field of the Integrated Solar Combined
Cycle power plant at Ain Beni Mathar (Morocco) were estimated by Zaaoumi et al. [11]. A
Energies 2021, 14, 90 3 of 21

detailed model that included energy balance equations was developed. The authors used
COMSOL Multiphysics software to solve the matrix of these mathematical relationships.
One of the issues of this work was to determine heat losses through pipes, which were
only a structural element of parabolic trough collectors. The analysis of heat loss through
pipeline in the circulation loop was neglected in this study.
The purpose of this article is to present the results of estimating the heat loss of
pipelines that are part of the solar domestic hot water (SDHW) heating system located on
the campus of the Bialystok University of Technology (Poland). A series of measurements
were carried out under normal operating conditions. Particular attention was paid to the
impact of continuous water supply temperature fluctuations on the variability of heat flux
emitted by the external surface of pipes.
Two methods were used to estimate heat losses of circulation pipelines. The first
was to compare the readings of heat meters installed in each loop. The second was a
classical method, i.e., the heat losses were calculated by multiplying the mass flux by
the specific heat and the temperature difference at the beginning and end of the pipe.
The concept of this research emerged when it turned out that the results obtained from
these two methods differ significantly. It turned out that heat losses were not recorded
when the return temperature was higher than the supply temperature. In contrast, when
energy was balanced in a classic way, heat gains appeared instead of heat losses. In real
operating conditions the heat flux was constantly emitted from the entire surface of the
pipelines insulation because the water temperature was higher than the temperature of the
surrounding air. The causes of these fluctuations and “anomalies” in water temperature
will be explained later in this article. In order to properly estimate the heat losses, the
author used the CFD modeling technique and the temperature measurement results were
applied as the input data. A very extensive literature review assured the author that so far
no one has determined heat losses from pipelines in such a precise way under conditions
of high fluctuations in the temperature of the medium.
Most of the leading computer programs for energy simulation of buildings and HVAC
systems use weather meteorological databases that contain average values of parameters
in hourly intervals. In the case of Poland, such typical meteorological years and average
hourly statistical climate data for over 60 locations can be found on the EnergyPlus™
website [12]. The article also presents the answer to the question: is it correct to use
averaged hourly meteorological parameters when there are rapidly changing the heat
transfer conditions?

2. Description of the Solar Domestic Hot Water Heating System

DHW heating system supported by thermal solar collectors was built as part of the
project financed by the European Union under the Regional Operational Program of the
Podlasie Voivodship (part 5.2.—Improving the energy efficiency of the infrastructure of the
Bialystok University of Technology using renewable energy sources) (Zukowski [13,14]).
It consists of a battery of 35 flat plate collectors (Figure 1) with a total area of approximately
72 m2 (gross) and 65.5 m2 (absorber) and 21 evacuated tube type collectors (Figure 2)
with a total area of over 74 m2 (gross) and approximately 67 m2 (active). The entire
solar installation is located on the roof of the hotel for young researchers of the Bialystok
University of Technology.
Energies 2021, 14, 90 4 of 21
Energies 2021, 14, x FOR PEER REVIEW 4 of 21
Energies 2021, 14, x FOR PEER REVIEW 4 of 21

Figure 1. Flat plate solar collectors connected in series of five panels grouped into seven subarrays.
Figure 1. Flat plate solar collectors connected in series of five panels grouped into seven subarrays.
Figure 1. Flat plate solar collectors connected in series of five panels grouped into seven subarrays.

Figure 2. Evacuated tube solar collectors connected in series of three panels grouped into five subarrays.
Figure 2. Evacuated tube solar collectors connected in series of three panels grouped into five subarrays.
Figure 2. Evacuated tube solar collectors connected in series of three panels grouped into five subarrays.
A simplified connection diagram of the system components is shown in Figure 3. The
entire
Ainstallation
A simplified consists of diagram
simplified connection
connection two identical
diagram of thecircuits
of the systemconnected
system in parallel.
components
components is shown
is shown in
in Figure
Figure 3.
3. The
The
entire installation consists of two identical circuits connected in parallel.
entire installation consists of two identical circuits connected in parallel.
Energies 2021, 14, 90 5 of 21
Energies 2021, 14, x FOR PEER REVIEW 5 of 21
Energies 2021, 14, x FOR PEER REVIEW 5 of 21

Figure3.3. Schematic
Schematic diagram of the SDHW heating system (the letters A, B, C,B,DC,
indicate the meas-
Figure
Figure diagram
3. Schematic diagram of of
thethe SDHW
SDHW heating
heating system
system (the
(the letters A, B, A,
letters D indicate
C, D indicate the
the meas-
uring points).
measuring points).
uring points).

Theobject
The objectofofthis
thisstudy
studywere
werefourfourpipes
pipesconnecting
connectingheat heatexchangers
exchangerslocated
locatedononthe the
The object of this study were four pipes connecting heat exchangers located on the
roof with
roof with storage
with storage
storage tankstanks
tanks in in
in thethe basement.
the basement.
basement. They They
They are are made
are made
made of of light
of light galvanized
light galvanized
galvanized steelsteel pipes
steel pipes
pipes andand
and
roof
havea anominal
have nominaldiameter
diameterofofDN DN40. 40.Their
Theirouter
outerdiameter
diameterisis48.3 48.3mmmmandandthethewall
wallthickness
thickness
have a nominal diameter of DN 40. Their outer diameter is 48.3 mm and the wall thickness
ofof
of
2.9mm.
2.9
2.9
mm.The
mm. The
Theaverage
averagelength
lengthofofeach
average length
eachofofthe
of each of
thefour
fourpipes
the four
pipesisis8686m.
pipes is 86 m. The
m.The
Thepipelines
pipelinesare
pipelines are
areinsulated
insulated
insulated
withPE
with PEfoam,
foam,which
whichwas wasproduced
producedby byextrusion,
extrusion,with
witha athickness
thicknessofof3030mm mm(Figure
(Figure4).
4).
with PE foam, which was produced by extrusion, with a thickness of 30 mm (Figure 4).
Theheat
The heattransfer
transfercoefficient
coefficient ofof
thethe insulating
insulating sleeve
sleeve is is equal
equal 0.038
0.038 W/(m·K)
W/(m · K) atatananaverage
average
The heat transfer coefficient of the insulating sleeve is equal 0.038 W/(m·K) at an average
temperatureofof4040◦ C
temperature °Cand
andandanditsitsdensity
densityisisabout
about3030kg/m
kg/m 33..
temperature of 40 °C and and its density is about 30 kg/m . 3

Figure 4. Thermal insulation on pipes under consideration (photo taken by the author).
Figure 4. Thermal insulation on pipes under consideration (photo taken by the author).
Figure 4. Thermal insulation on pipes under consideration (photo taken by the author).
Energies 2021,
Energies 2021, 14,
14, 90
x FOR PEER REVIEW 6 of 21

Letters
Letters A,A, BB mark
markthethebeginning
beginningandandendendofof
thethe
supply
supply pipe connecting
pipe connecting thethe
heatheat
ex-
changer located on the roof with the storage tanks located in the basement.
exchanger located on the roof with the storage tanks located in the basement. While the While the let-
ters C and
letters C andD correspond
D correspond to the beginning
to the andand
beginning end end
of theof return pipepipe
the return connecting the stor-
connecting the
age tanks
storage to the
tanks substation
to the located
substation on the
located roof.
on the TheThe
roof. FP, ET sub-indexes
,
FP ET
relate
sub-indexes to
relatepipes
to in
pipes the
in
flat plate
the flat andand
plate evacuated tubetube
evacuated collector loops,
collector respectively.
loops, respectively.

3. Data Acquisition System

The operating parameters
parameters of of the entire hydraulic
hydraulic system
system are monitored
monitored withwith a one-
minute time step. The measurement results are read from 14 heat meters supporting each
of the collector batteries, two heat meters measuring the energy energy yield of flat and evacuated
tube collectors
collectors separately,
separately,and andoneonegeneral
generalmeter
meterlocated
located in in
thethe heating
heating substation.
substation. In
In ad-
addition, 42 Pt500 resistance temperature sensors are installed.
dition, 42 Pt500 resistance temperature sensors are installed. The total temperature meas-The total temperature
measurement
urement error error is approximately
is approximately ±0.3In◦ C.
±0.3 °C. In addition,
addition, four meters
four meters register
register electricity
electricity con-
consumption
sumption by twoby two
solarsolar
pumps pumps andandtwotwo circulation
circulation pumps.
pumps. TheThe basic
basic parameters
parameters of
of the
the outside
outside air air
andand thethe intensity
intensity of of solar
solar radiationare
radiation arerecorded
recordedby by the
the specially
specially dedicated
dedicated
weather
weather station
station every
every 55 s.s. It
It is
is possible
possible toto monitor
monitor thethe results
results of of measurements
measurements over over the
the
Internet. Screenshots of the monitoring system are presented
Internet. Screenshots of the monitoring system are presented in Figures 5–8. in Figures 5–8.
Selected results
Selected results ofof temperature
temperature measurement
measurement on on two
two supply
supply pipelines
pipelines registered
registered atat
A
AFP , B
FP, BFP
, A
, AETET
FP , B points
, BET points
ET and two return pipes registered
and two return pipes registered at CFP, DFP at C , D
FP, CETFP , C
, DETET , DET(shown
points points
(shown
in Figures in Figures
5–8) were 5–8) were
used in used
orderin toorder to determine
determine the heatthe
lossheat losspipes.
of the of the pipes.

Figure 5. Screenshot of the monitoring system—loop of evacuated tube solar collectors.

Figure 5. Screenshot of the monitoring system—loop of evacuated tube solar collectors.
Energies 2021, 14, 90 7 of 21
Energies 2021, 14, x FOR PEER REVIEW 7 of 21
Energies 2021, 14, x FOR PEER REVIEW 7 of 21

Figure 6. Screenshot of the monitoring system—loop of flat plate

plate solar
solar collectors.
collectors.
Figure 6. Screenshot of the monitoring system—loop of flat plate solar collectors.

Figure 7. Screenshot of the monitoring system—tanks connected with evacuated tube solar collectors.
Figure 7. Screenshot of the monitoring system—tanks connected with evacuated tube solar collectors.
Figure 7. Screenshot of the monitoring system—tanks connected with evacuated tube solar collectors.
Energies 2021, 14, 90 8 of 21
Energies 2021, 14, x FOR PEER REVIEW 8 of 21

Energies 2021, 14, x FOR PEER REVIEW 8 of 21

Figure
Figure 8. 8. Screenshotofofthe
Screenshot themonitoring
monitoringsystem—tanks
system—tanks connected
connectedwith
withflat
flatplate
platesolar
solarcollectors.
collectors.
Figure 8. Screenshot of the monitoring system—tanks connected with flat plate solar collectors.
4. Results and Discussion
4.4.1.
4.Results and
andDiscussion
Characteristics
Results of Temperature Fluctuation in the Circulation Pipes
Discussion
4.1. Characteristics of
In the two analyzed
4.1. Characteristics ofTemperature
loops, Fluctuation
Temperature in which water
Fluctuation in the
in thecirculates
Circulation
Circulation Pipes the heat exchangers
between
Pipes
InInthe
located thetwo
on the analyzed
tworoof and theloops,
analyzed in
in which
heat storage
loops, which water
tanks circulates
located
water betweenthe
in thebetween
circulates basement, the heat
there
heat aexchangers
isexchangers
continu-
located
ous on
change theinroof and
temperature.the heat
The storage
reason tanks
for located
these in the
fluctuations basement,
are, of there
course,
located on the roof and the heat storage tanks located in the basement, there is a continu- is a
changes continuous
in the
amount
change of heat supplied by solar collectors during the day. During the
ous change in temperature. The reason for these fluctuations are, of course, changes in in
in temperature. The reason for these fluctuations are, of night,
course, the
changestemper-
thethe
ature
amount of circulating
amountofofheat water
heatsupplied
suppliedby stabilizes.
bysolar This
solar collectorscyclic process,
during the
collectors during illustrated
theday.
day.During in
Duringthe Figure 9,
thenight, was
night,the shown
thetemper-
tempera-
for
ture
atureflatof
of collector
circulating batteries
circulatingwater for six days
waterstabilizes.
stabilizes. of the
This
This week.
cyclic
cyclic The nature
process,
process, of temperature
illustrated
illustrated ininFigure
Figure9,variation
9, was
was in
shown
shown
the
for vacuum
forflat
flatcollectorcollector
collectorbatteries loop
batteriesfor was
for six only
six days slightly
of the
days of different
the week.
week.The from
Thenature that shown in
natureofoftemperature
temperatureFigure 9, so
variation
variation it in
in
was
the not
thevacuum included
vacuumcollector in this
collectorloop article.
loopwaswasonly
onlyslightly
slightlydifferent
differentfrom
fromthat
thatshown
shownininFigure
Figure9, 9,soso
it was
it
not
was included in thisinarticle.
not included this article.
50
50
45
- temperature at point A FP
45
40 -- temperature
temperature at
at point
point CA FP
FP
40 - temperature at point C FP
35
35
oC oC

30
Temperature,

30
25
Temperature,

25
20
20
15
15
10
9:07 9:07
12:4512:45
16:2416:24
20:0120:01
23:3823:38
3:15 3:15
6:52 6:52
10:2910:29
14:0914:09
17:4817:48
21:2521:25
1:02 1:02
4:39 4:39
8:16 8:16
11:5311:53
15:3015:30
19:0719:07
22:4422:44
2:22 2:22
5:59 5:59
9:36 9:36
13:1313:13
16:5016:50
20:2720:27
0:04 0:04
3:43 3:43
7:20 7:20
10:5710:57
14:4514:45
19:5219:52
23:2923:29
3:06 3:06
6:43 6:43
10:2410:24
14:0114:01
17:3917:39
21:1621:16

10
Time

Figure 9. Change of temperature at the beginning of the supplyTime pipe (AFP—on the roof) and at the beginning of the return
pipe (CFP—in the basement)—circulation loop for flat collectors.
Figure 9. Change
Figure 9. Changeof of
temperature atat
temperature the
thebeginning
beginningofofthe
thesupply
supply pipe
pipe (AFP —onthe
FP—on theroof)
roof)and
andatatthe
thebeginning
beginningof of
thethe return
return
pipepipe FP—in
(CFP(C—in thethe basement)—circulationloop
basement)—circulation loopfor
forflat
flatcollectors.
collectors.
Energies 2021, 14, 90 9 of 21
Energies 2021, 14, x FOR PEER REVIEW 9 of 21

4.2. Determination of Heat Loss in the Classic Way

4.2. Determination of Heat Loss in the Classic Way
Due to the dynamic nature of the heat exchange occurring in the solar system pipes, it
Due to the dynamic nature of the heat exchange occurring in the solar system pipes,
was decided to study this phenomenon in more detailed way. In the first stage, heat loss
it was decided to study this phenomenon in more detailed way. In the first stage, heat loss
were determined in a classic way, i.e., using the following basic balance equations:
were determined in a classic way, i.e., using the following basic balance equations:
.
QS𝑄==m𝑚 A −−T𝑇
c c p𝑐( T(𝑇 B ))(for
(forsupply
supplypipe ) and
pipe) and (1a)
(1a)
.
QS𝑄==m𝑚
c c p𝑐( T(𝑇
C −−T𝑇
D ))(for
(forreturn
returnpipe ),
pipe), (1b)
where:
.
𝑚
mc —mass flow rate of water
water circulating
circulating in
in the
the pipes,
pipes,
cpp—specific
—specific heat
heat at
at constant
constant pressure,
pressure,
𝑇
TA, ,𝑇 T—temperature
C —temperature atatthe
thebeginning
beginningofofthethepipe
pipe(higher),
(higher),
𝑇
TB,,𝑇TD —temperatureatatthe
—temperature theend
endofofthe
thepipe
pipe(lower).
(lower).
However, as
However, asthetheresults
resultsofofpreliminary
preliminary calculations
calculations have
haveshown,
shown,it isitnot possible
is not to use
possible to
classical
use energy
classical balancing
energy balancingequations (Equation
equations (1)) based
(Equation on determining
(1)) based on determining the temperature
the temper-
difference
ature at theat
difference inlet
the and
inletoutlet of theofsystem.
and outlet Thus,Thus,
the system. the use
theof
useheat meters
of heat to estimate
meters to esti-
energy loss will cause incorrect results, because if the return temperature
mate energy loss will cause incorrect results, because if the return temperature is higher is higher than the
supply temperature, the measuring device will show the value 0. In
than the supply temperature, the measuring device will show the value 0. In fact, the heatfact, the heat flux is
emitted from the entire surface of the pipelines.
flux is emitted from the entire surface of the pipelines.
This phenomenon
This phenomenon results results from
from two
two reasons. One is
reasons. One is the
the thermal
thermal inertia
inertia of of the
the pipes
pipes and
and
to a lesser extent the thermal capacity of the insulation jacket. The second reason is the
to a lesser extent the thermal capacity of the insulation jacket. The second reason is the time
time
of water
of waterflowflowthrough
throughthe thepipeline,
pipeline, which
which was
was moremore
thanthan twice
twice as long
as long as the assampling
the samplingtime
time of the data recording system. The chart in Figure 10 was made
of the data recording system. The chart in Figure 10 was made to illustrate this phenomenon. to illustrate this
phenomenon. The graph shows changes in temperature difference between the beginning
The graph shows changes in temperature difference between the beginning (AFP point) and
(AFP point) and end (BFP point) of the supply pipe over a six-day period. As can be seen in
end (BFP point) of the supply pipe over a six-day period. As can be seen in Figure 10, there
Figure 10, there is a cyclical negative temperature difference suggesting the appearance of
is a cyclical negative temperature difference suggesting the appearance of heat gains from
heat gains from the surroundings. In fact, there are still heat loss because the temperature
the surroundings. In fact, there are still heat loss because the temperature of the water in the
of the water in the pipeline is higher than the temperature of the ambient air.
pipeline is higher than the temperature of the ambient air.

9
8 - temperature difference between the point A FP and B FP
7
6
5
Temperature difference, oC

4
3
2
1
0
-1
-2
-3
-4
-5
9:07
12:45
16:24
20:01
23:38
3:15
6:52
10:29
14:09
17:48
21:25
1:02
4:39
8:16
11:53
15:30
19:07
22:44
2:22
5:59
9:36
13:13
16:50
20:27
0:04
3:43
7:20
10:57
14:45
19:52
23:29
3:06
6:43
10:24
14:01
17:39
21:16

Time

Figure 10. Temperature difference between

Temperature difference between at
at the
the beginning
beginning of
of the
thesupply
supplypipe
pipe(A
(AFP
FP—on the roof) and at the beginning of

the return pipe (C —in the basement)—circulation loop for flat collectors which was registered at
the return pipe (CFP —in the basement)—circulation loop for flat collectors which was registered
FP at the
the same
same time.
time.

It was not
not decided
decidedtotoconduct
conductexperimental
experimentalresearch
researchinvolving direct
involving measurement
direct of
measurement
heat flux density on the pipeline insulation surface due to the following reasons:
of heat flux density on the pipeline insulation surface due to the following reasons:
Energies 2021, 14, 90 10 of 21

• Large error that occurs during these types of measurements (small and variable flux
values).
• Large piping lengths (the need to perform measurements at large number of points
simultaneously).
• Continuous temperature fluctuation inside the pipes.
• High costs of a large number of sensors and ensuring their accuracy.
Thus, a special method has been developed for estimating heat loss from pipes with
variable medium temperature. This is an element of novelty, which is presented in the next
part of this paper.

4.3. The New Approach to Determining the Heat Loss of Pipelines

To estimate the heat loss of pipelines with continuous fluid temperature fluctuations,
a new approach has been proposed based on temperature measurements and numerical
simulations. In the first step, a numerical model of the heat transfer in a pipeline was
developed using algorithms of Computational Fluid Dynamics. Then this model was
validated on the basis of comparison the water temperature values at the end of the
pipeline as the results of calculations and measurements. A positive result of model
verification ensured the correct energy balance obtained from numerical simulations. Thus,
the amount of heat emitted from the outer surface of the pipeline could be determined
with high accuracy based on computer calculations. As a result of this numerical analysis,
two analytical equations were developed.

4.3.1. Governing Equations

The incompressible water flow inside the pipes is described by the equations of mass
and energy conservation. The general form of the continuity equation (mass conservation)
for 2D axisymmetric geometry is as follows [15]:

∂vr ∂v x vr
+ + = 0, (2)
∂r ∂x r
where:
ρ—fluid density,
t—time,
x—axial coordinate,
r—radial coordinate,
vx —axial velocity,
vr —radial velocity.
The energy transport equation in terms of enthalpy h can be written in the following
form:   
∂h →  1 µt
+ ∇· v h = ∇· µ + ∇h , (3)
∂t ρ σt
where:
µ—viscosity,
µ—turbulence viscosity,
σt —constant.
The equations of axial and radial momentum conservation are given by [15]:

∂2 vr ∂2 vr
 
∂vr ∂vr ∂vr 1 ∂p µ 1 ∂vr vr
+ vr + vx =− + + 2 + − 2 , (4)
∂t ∂r ∂x ρ ∂r ρ ∂r2 ∂x r ∂r r

and
∂2 v x ∂2 v x
 
∂v x ∂v x ∂v x 1 ∂p µ 1 ∂v x
+ vr + vx =− + + + , (5)
∂t ∂r ∂x ρ ∂x ρ ∂r2 ∂x2 r ∂r
where p is a pressure.
Energies 2021, 14, 90 11 of 21
Energies 2021, 14, x FOR PEER REVIEW 11 of 21

A standard two-equation turbulence model k-ε was used for numerical simulations of
A standard
transient two-equation
water flow. turbulence
The turbulence energyk-kεand
kineticmodel wasrate
used
of for numerical
dissipation simulations
of energy ε are
of transient
presented water[15].
below flow. The turbulence kinetic energy k and rate of dissipation of energy
ε are presented below [15]. " #

∂k ∂k ∂ µt ∂k
ρ 𝜌+ ρu+i 𝜌𝑢 = = µ𝜇++ − ρε + G , (6)
∂t ∂xi ∂x j σk ∂x j − 𝜌𝜀 + 𝐺 ,k (6)

and
and " #
ε2

∂ε ∂ε ∂ µt ∂ε ε
ρ 𝜌+ ρu+i 𝜌𝑢 = = µ𝜇++ ++𝐶 C1ε 𝐺G−
k−𝐶 C2ε
𝜌ρ , , (7)
(7)
∂t ∂xi ∂x j σε ∂x j k k

where, µ𝜇t =
where, = ρC
𝜌𝐶µ ρ𝜌k2 ,, G
Gk—generation
—generation of of turbulence
turbulence kinetic
kinetic energy.
energy.
ε k
The default
default values
valuesof ofthe
theconstants
constantsinin the above
the equations
above areare
equations equal to: C
equal μ =C
to: 0.09, C1ε =
µ = 0.09,
1.44,
C1ε = C 2e =
1.44, 1.92,
C 2e σ
= =
1.92,
k 1.0,
σ σ
k =
ε = 1.3
1.0, [16].
σ ε = 1.3 [16].

4.3.2. Description of the Numerical Model

Numerical
Numerical model
model of
of the pipeline with
the pipeline with thermal
thermal insulation
insulation was
was developed
developed using
using Ansys
Ansys
Release 19.2 software package. The discretization mesh, a fragment of which
Release 19.2 software package. The discretization mesh, a fragment of which is shown in is shown
in Figure
Figure 11, 11,
waswas
mademade using
using a Ansys
a Ansys Meshing
Meshing program
program and geometric
and geometric model model
of theofcom-
the
computational domain was created in the SpaceClaim program. While the calculations
putational domain was created in the SpaceClaim program. While the calculations and and
processing
processing results
results were
were carried
carried out
out in
in Fluent
Fluent solver
solver (finite-volume
(finite-volume numerical
numerical scheme).
scheme). All
All
these
these programs
programs are
are integral
integral parts
parts of
of the
the Ansys
Ansys package.
package.

Figure 11.
Figure 11. Fragment of the mesh
mesh with
with boundary
boundary conditions.
conditions.

An axisymmetric
axisymmetric geometry
geometrywas wasapplied,
applied,thanks
thankstotowhich
whichthis
thisflow
flow issue
issue waswas reduced
reduced to
to two-dimensional.
two-dimensional. In order
In order to select
to select the optimal
the optimal mesh density,
mesh density, four
four tests tests
were wereout
carried carried
with
out with the following
the following number ofnumber
elements: of860,000,
elements: 860,000,
1,118,000, 1,118,000,
1,419,000, 1,419,000,
1,634,000. 1,634,000. Area-
Area-Weighted Av-
Weighted Average Total Surface Heat Flux
erage Total Surface Heat Flux qS was selected as q was selected as a variable whose
S a variable whose value was compared value was
in each
compared
calculation in eachThe
series. calculation series.
tests lasted The
31,620 testsalasted
s with 31,620
time step s with
of 60 s. Thea change
time stepin qof 60 s. The
S compared
change in q compared
to the least dense
S to the least dense mesh was 0.311%, 0.314% and 0.316%.
mesh was 0.311%, 0.314% and 0.316%. Therefore, for further calculations, a Therefore,
for
meshfurther calculations,
with 1,118,000 a mesh
elements with 1,118,000
and 1,147,959 nodes waselements and
used in this1,147,959
study. The nodes
externalwasdimen-
used
in this study. The external dimensions of the model
sions of the model were 86 m (length x) and 0.054 m (length y). were 86 m (length x) and 0.054 m
(length y).
SIMPLE (Semi-Implicit Method for Pressure-Linked equations) algorithm for solving
SIMPLE and
the pressure (Semi-Implicit Method for
velocity dependent Pressure-Linked
equations equations)
was applied. algorithmnumber
The Reynolds for solving
was
the
higher than 20,000, therefore turbulent flow was assumed and the standard k-ε model was
pressure and velocity dependent equations was applied. The Reynolds number was
higher
used, asthan earlier. turbulent flow was assumed and the standard k-ε model was
20,000, therefore
mentioned
used,The
as mentioned earlier.
equations describing the conditions at the model boundaries are shown in Figure
11. All calculations were made with a time step of 60 s.
Energies 2021, 14, 90 12 of 21

The equations describing the conditions at the model boundaries are shown in
Figure 11. All calculations were made with a time step of 60 s.
In order to simplify the model, it was assumed that the pipeline has no bends. To
increase the accuracy of calculations, the heat transfer coefficient on the outer wall of
insulation was declared as a function of insulation surface temperature TW and ambient
temperature TE . The modification of boundary conditions at the air and wall boundary
using User-Defined Function (UDF) was used for this purpose. UDF was also applied to
Energies 2021, 14, x FOR PEER REVIEW 12 of 21
account for changes in the thermal conductivity of the insulation material depending on its
temperature.
Creating an analytical model that would take into account all these assumptions
In order to simplify the model, it was assumed that the pipeline has no bends. To in-
described
creaseabove is impossible.
the accuracy For
of calculations, the this reasoncoefficient
heat transfer CFD modelling
on the outerwas
wall used to determine
of insula-
heat loss of pipelines. This value was obtained using the Fluent postprocessor
tion was declared as a function of insulation surface temperature TW and ambient tempera- function
ture T . The modification of boundary conditions at the air and wall boundary
(Reports-Surface Integrals), which enabled the generation of heat flux emitted at the
E using User-
externalDefined Function
boundary of(UDF) was used
the model at for
eachthistime
purpose.
step.UDF was also applied to account for
changes in the thermal conductivity of the insulation material depending on its temperature.
Creating an analytical model that would take into account all these assumptions de-
4.3.3. Validation of the Simulation Model
scribed above is impossible. For this reason CFD modelling was used to determine heat
Verification of the
loss of pipelines. Thisaccuracy
value was of the numerical
obtained model
using the Fluent was carried
postprocessor out (Re-
function by comparing
ports-Surface
the water Integrals),
temperature which
at the enabled
end thepipes
of the generation of heat from
obtained flux emitted at the external (T ) and
the measurement M
boundary of the model at each time step.
from the calculations (TCFD ). Four typical cases during the system operation were selected.
The first two
4.3.3. (P1, P2)ofconcerned
Validation the Simulation the supply (Figure 12) and return (Figure 13) pipes in the
Model
vacuum collectors loop.
Verification of theThe next two
accuracy of thecases P4) concerned
(P3, model
numerical was carriedthe
outtemperature
by comparing fluctuation
in the flat collector
the water loop: the
temperature supply
at the end of(Figure 14)
the pipes and the
obtained fromreturn (Figure 15)(T
the measurement pipe.
M) andThe ambient
from theof
temperature calculations
pipes TE(Tin the analyzed measuring series was 20 ± 0.5 ◦ C.
CFD). Four typical cases during the system operation were selected.

The first two (P1, P2) concerned the supply (Figure 12) and return (Figure 13) pipes in the
To determine the level of accuracy of computer simulations, two equations were used
vacuum collectors loop. The next two cases (P3, P4) concerned the temperature fluctuation
to calculate thecollector
in the flat mean relative error e(Figure
loop: the supply MRE (Equation
14) and the (8))
returnand root15)
(Figure mean The am-error eRMSE
pipe.square
(Equation (9)).
bient temperature of pipes TE in the analyzed measuring series was 20 ± 0.5 °C.
 T −simulations,
To determine the level of accuracy of computer T two equations were used
∑in=1  M TMCFD 

to calculate the mean relative error eMRE (Equation (8)) and root mean square error eRMSE
e MRE = ·100%, (8)
(Equation (9)). n
s
∑
𝑒 = ∑in=1 ( TM − TCFD )2
∙ 100%, (8)
e RMSE = , (9)
∑ (
ke )
𝑒 = , (9)
where ke = n − 1 for n < 30 and ke = n for n > 30.
Inwhere ke = ntwo
the first − 1 for n < 30(P1,
cases andP2)
ke = nnfor
wasn > 525,
30. which corresponded to 8.75 h; and in the next
In the first two cases (P1, P2) n was 525, which corresponded to 8.75 h; and in the next
two cases (P3, P4) n was equal to 1195, i.e., the simulation period was 19.9 h.
two cases (P3, P4) n was equal to 1195, i.e., the simulation period was 19.9 h.

Figure 12. Comparison of the measured and calculated temperature at the outlet from the supply (left side) and return
Figure 12. Comparison of the measured and calculated temperature at the outlet from the supply (left side) and return
(right side) pipes of the vacuum collectors.
(right side) pipes of the vacuum collectors.
Energies 2021, 14,
Energies 2021, 14, 90
x FOR PEER REVIEW 13
13 of
of 21
21
Energies 2021, 14, x FOR PEER REVIEW 13 of 21

Figure 13.
13. Comparison of of the measured
measured and calculated
calculated temperature at the outlet
outlet from the
the supply (left
(left side) and
and return
Figure 13.Comparison
Figure Comparison ofthe the measuredand and calculatedtemperature
temperatureatatthe
the outletfrom
from thesupply
supply (leftside)
side) andreturn
return
(right
(right side) pipes of the flat plate collectors.
(rightside)
side)pipes
pipesofofthe
theflat
flatplate
platecollectors.
collectors.

9.3
9.3
9.2
9.2
9.1
9.1
9.0 qqs,l
qS,LS,L-for
T1 TT1 qqs,l qqs,l
S,L -for
T3 TT3 qqs,l
qS,LS,L- -for T
9.0 qs,l
- for
T1 1 qS,LS,L-for
qs,l T2 T2
-for
T2 T2 qqs,l
S,L -for
T3 3
qs,l T4
for
T4 T44
8.9
8.9
S,l/( W/m)

8.8
S,l/( W/m)

8.8
8.7
8.7
qq

8.6
8.6
8.5
8.5
8.4
8.4
8.3
8.3
8.2
8.2
0 10 20 30 40 50 60 70 80
0 10 20 30 40 50 60 70 80
l, m
l, m
Figure 14. Distribution of heat loss along the length of the pipeline (l = 0 m—inlet).
Figure14.
Figure 14.Distribution
Distributionofofheat
heatloss
lossalong
alongthe
thelength
lengthofofthe
thepipeline
pipeline(l(l==00m—inlet).
m—inlet).
Energies 2021, 14,
Energies 2021, 14, 90
x FOR PEER REVIEW 14 of
14 of 21
21

Fragment of temperature changes selected for analysis at the inlet of return pipe in the flat-collector
Figure 15. Fragment flat-collector loop.
loop.

Table 11 contains
Table containsthe
theresults
resultsofofmodel
modelvalidation
validation using
using error
error estimation.
estimation. It turned
It turned out
out that
that the divergence
the divergence between
between temperature
temperature valuesvalues obtained
obtained from measurements
from measurements and
and numer-
numerical
ical calculations
calculations are quiteare quite
small small
and and arethe
are within within the accuracy
accuracy range of temperature
range of temperature sensors.
sensors.
Table 1. Results of model validation.
Table 1. Results of model validation.
error P1 P2 P3 P4
Error P1 P2 P3 P4
eMRE/(%) 0.91 0.24 0.82 0.53
eMRE /(%)
eRMSE/(°C) 0.28 0.91
0.12 0.25 0.24
0.16 0.82 0.53
◦
eRMSE /( C) 0.28 0.12 0.25 0.16

The highest differences appeared in the case of a very rapid change in temperature
at theThe highest
inlet to thedifferences appeared
pipe. The return in the caseobtained
temperature of a veryfromrapidthe change in temperature
calculations show that at
inlet totends
the model the pipe. The return
to suppress thistemperature obtained from the calculations show that the
sharp variations.
model tends to suppress
Therefore, points with thismaximum
sharp variations.
deviations  ∙ 100% were distinguished in
 TM −TCFD 
Therefore, points with maximum deviations
each of these measurement series. They were marked with  TM the ·100% were on
red circles distinguished
Figure 12 and in
each
13. Inof
thethese
casemeasurement series. They
of P1 (Figure 12—left graph)were
thismarked
difference with
in thetwored circles
points on Figures
is about 10% and 12
and 13. In the case of P1 (Figure 12—left graph) this difference in two
8.7%; in the second case P2 (Figure 12—right graph) it is 3.9%; in the third series P3 (Figure points is about 10%
and 8.7%;
13—left in thethe
graph) second
error case P2 (Figure
reaches 4.7% and 12—right graph)
in the last it is(Figure
case P4 3.9%; in the thirdgraph)
13—right P3
seriesthe
(Figure 13—left graph) the error reaches 4.7% and in the last
error value is the lowest and does not exceed 2%. However, these single cases did not case P4 (Figure 13—right
graph)the
affect theaccuracy
error value is the lowest
of modelling andindoes
results longnot exceed
time 2%. However,
intervals. As can be these single cases
seen, fluctuations
in the supply temperature in the return pipes (P2 and P4 series) were definitely be
did not affect the accuracy of modelling results in long time intervals. As can seen,
smaller
fluctuations
compared toin the
the supplythat
changes temperature
occurredin inthe
thereturn
supplypipes (P2This
pipes. andisP4 series)
due to thewere
highdefinitely
thermal
inertia of the storage tanks from which the return pipelines are supplied. This to
smaller compared to the changes that occurred in the supply pipes. This is due thetrans-
also high
lates into value of eRMSE and eMRE, which in the case of flow analysis in return pipesalso
thermal inertia of the storage tanks from which the return pipelines are supplied. This are
translates into
noticeably value of eRMSE and eMRE , which in the case of flow analysis in return pipes
lower.
are noticeably
In summary, lower.
it can be concluded that the proposed model accurately simulates the
In summary, it can be concluded that the proposed model accurately simulates the
heat transfer occurring in an insulated tube and can be used to precisely determine the
heat transfer occurring in an insulated tube and can be used to precisely determine the
pipes heat loss in the second stage of this method.
pipes heat loss in the second stage of this method.
Energies 2021, 14, 90 15 of 21

4.4. Determination of the Pipelines Heat Loss

Most professional computer programs designed to simulate the buildings energy
consumption and operation of HVAC systems apply meteorological databases prepared
with one-hour time steps. So, it was decided to check whether assuming hourly temperature
averaging, which is characterized by high variability, will affect the accuracy of the results
of pipe heat loss calculations. A specific heat loss from the surface of the circulation pipes
were determined using the above-described numerical model in two variants.
First, calculations were made assuming the temperature obtained from measurements
TIN (one-minute time step) as an inlet boundary condition. Then the same calculations
were made using the previous input data but averaging them over one hour.

4.4.1. Determination of the Heat Loss per Unit Length of Pipe with One-Minute Time Step
Calculations of the heat flux QS emitted from the surface of the pipelines were made
for four measurement series (P1 ÷ P4) in transient conditions. In total, 3440 values of
QS corresponding to the supply temperature TIN in the range from 20 ◦ C to 60 ◦ C were
obtained. In engineering practice heat losses per 1 m of pipe length are often determined:

Qs
qS,l = , (10)
l
where l is the length of the pipeline.
The distribution of the heat flux density qS,l on the surface of the pipeline shows a
very variable nature, which can be seen in the examples shown in Figure 14.
Four cases (T1 ÷ T4 ) were selected from the fragment of the inlet temperature distribu-
tion (Figure 15), for which the change in qS,l along the pipeline length was determined. In
the case of slight TIN fluctuations, heat loss slowly decrease as a result of water cooling (T1 ,
T2 ). A sharp change in the supply temperature (T3 ) causes the reverse phenomenon, i.e.,
an increase qS,l along the length of the pipe. After seven minutes of continuous increase of
the inlet temperature (T4 ), the tendency to change the distribution of qS,l again reverses,
showing this time the highest gradient downwards compared to the values in cases for T1
and T2 . So, as one can see, the heat transfer on the surface of the pipeline is very dynamic
and thus difficult to describe it in an analytical form.
Figure 16 summarizes the results of calculations illustrating the dependence of the
temperature at the inlet of the pipeline on the unit heat losses qS,l . The dashed line indicates
the trend function (Equation (11)), that approximates the results of the analysis with high
accuracy, because the coefficient of determination R2 is 0.9935.

qS,l = 0.2624· TI N − 5.2444. (11)

Equation (11) can be used to determine the heat loss of DN40 pipes protected by
30 mm-thick thermal insulation.
It is more practical in engineering applications to determine heat losses as a function
of the difference in water and ambient air temperature using the following equation:

qS,l = 0.2624·( TI N − TE ). (12)

If we want to calculate the amount of energy El lost in the ∆t period, then the following
relationship should be used:
El = qS,l ·l ·∆t. (13)
These equations can be used in the design of SDHW systems, optimization of insula-
tion thickness and economic analyses related to the profitability of this type of installations.
Energies2021,
Energies 2021,14,
14,90
x FOR PEER REVIEW 16 of
16 of 21
21

11

10

6
qS,l /(W/m)

0
20 25 30 35 40 45 50 55 60
TIN , oC
Figure 16.
Figure 16. Dependence
Dependence of
of pipeline
pipeline heat
heat loss
loss on
on inlet
inlet temperature.
temperature.

It is more practical in engineering applications to determine heat losses as a function

4.4.2.
of theDetermination of theand
difference in water Heat Loss with
ambient One-Hour Time
air temperature using Step
the following equation:
As mentioned earlier, one-hour time steps are used in computer simulations due to
𝑞 , = 0.2624 ∙ (𝑇 − 𝑇 ). (12)
the time-spans of meteorological parameters contained in databases. Of course, in software
such Ifaswe want to calculate
EnergyPlus or TRNSYSthe amount
we can of
useenergy
shorterEl time
lost in the Δt period,
intervals, but inthen the follow-
this case linear
ing relationship
interpolation should be
of weather dataused:
is done automatically. Therefore, it was decided to calculate
the heat loss of pipelines qS(1 h) based on the results of the P3 and P4 measurement series,
𝐸 = 𝑞 , ∙ 𝑙 ∙ ∆𝑡. (13)
averaging inlet temperature TIN in hourly intervals. In order to assess the accuracy of
theseThese
calculations,
equationsthecan
databepresented previously
used in the design ofobtained
SDHW for a one-minute
systems, time step
optimization were
of insula-
averaged overand
tion thickness an economic
hour interval
analyses markedtoqthe
and related for proper
profitability
S(1 min) comparison
of this both cases.
type of installations.
DqS parameter was used to reflect the relative error resulting from the use of different
time-spans (1 h and 1ofmin)
4.4.2. Determination the and
HeatisLoss
described by the following
with One-Hour Time Steprelationship:
As mentioned earlier, one-hour time steps are used in computer simulations due to
qS(1min ) − qS(1h)
the time-spans of meteorological =
DqS parameters 100%.
contained (14)
in databases. Of course, in soft-
qS(1min)
ware such as EnergyPlus or TRNSYS we can use shorter time intervals, but in this case
linearFigures
interpolation of simultaneously
17 and 18 weather data is show done the
automatically.
tendency toTherefore,
change theitflux
wasofdecided to
heat loss
calculate as
theaheat loss
of of pipelines qS(1 h) basedqon the results of the P3 and P4 measurement
obtained result “precise” calculations S(1 min) and the parameter DqS for assessing
series, averaging inlet temperature T IN in hourly intervals. In order to assess the accuracy
the error of results obtained with a 1-h time step. Comparing these variables, two main
of these calculations,
conclusions the data presented previously obtained for a one-minute time step
can be drawn:
were averaged over an hour interval and marked qS(1 min) for proper comparison both cases.
• The value of q h) is overestimated in relation to “precise” calculations when heat
DqS parameter wasS(1 used to reflect the relative error resulting from the use of different
loss tend to increase.
time-spans (1 h and 1 min) and is described by the following relationship:
• We can observe reduction of qS(1 h) when heat loss decreases.
( ) ( )
𝐷𝑞 = 100%. (14)
( )

Figures 17 and 18 simultaneously show the tendency to change the flux of heat loss
obtained as a result of “precise” calculations qS(1 min) and the parameter DqS for assessing
 the
the error
error of
of results
results obtained
obtained with
with aa 1-h
1-h time
time step.
step. Comparing
Comparing these
these variables,
variables, two
two main
main
conclusions can be drawn:
conclusions can be drawn:
Energies 2021, 14, 90
•• The
The value
value of
of qqS(1 h) is
S(1 h) is overestimated
overestimated in in relation
relation to
to “precise”
“precise” calculations
calculations when
when heat
heat loss
17 loss
of 21
tend to increase.
tend to increase.
•• We can
We can observe
observe reduction
reduction ofof qqS(1 when heat
h) when
S(1 h) heat loss
loss decreases.
decreases.

7
7
30 6
6
30
5
5
25
25 4
4
3
3
20
20 2
2
/(W/m22))

1
S(1min)/(W/m

DqSS,,%%
1
15
15 0
0

Dq
-1
-1
qs(1min)
qqS(1min)

qs(1min)
10
10 -2
-2
Dqs
Dqs -3
-3
5
5 -4
-4
-5
-5
0
0 -6
-6
1
1 2
2 3
3 4
4 5
5 6
6 7
7 8
8 9
9 10
10 11
11 12
12 13
13 14
14 15
15 16
16 17
17
Time, h
Time, h

Figure 17.
Figure 17. Heat
Heat loss
Heat loss calculated
loss calculated with
calculated with aa 111 min
with a min step
min step and
step and error
and error resulting from
error resulting from the
the use
use of
of aa 11 h
h time
time step—P3
step—P3 measurement
measurement series.
series.

7
7
25
25 6
6
5
5
4
4
20
20 3
3
2
2
1
1
/(W/m22))

15
15 0
0
S(1min)/(W/m

DqSS,,%%
-1
-1
qs(1min) -2
-2

Dq
qs(1min)
10
10 -3
-3
qqS(1min)

Dqs
Dqs -4
-4
-5
-5
5
5 -6
-6
-7
-7
-8
-8
0
0 -9
-9
1
1 2
2 3
3 4
4 5
5 6
6 7
7 8
8 9
9 10
10 11
11 12
12 13
13 14
14 15
15 16
16 17
17
Time,
Time, h
h
Figure
Figure 18.
Figure 18. Heat
18. Heat loss
Heat loss calculated
loss calculated with
calculated with a
with aa 111 min
min step
min step and
step and error
and error resulting
error resulting from
resulting from the
from the use
the use of
use of a
of aa 111 h
h time
h time step—P4
time step—P4 measurement
step—P4 measurement series.
series.
measurement series.

However,
However, these these differences
differences areare not
not too
too high,
high, because
because inin the
the P3
P3 series
series the
the error
error ranges
ranges
from about
from about −5% −5% to
−5%totooverover 6%,
over6%, and
6%,and
andinin the
inthe P4
theP4 series
P4series from
seriesfrom less
fromless than
lessthan −9%
than−9% to about
−9%totoabout 6%.
about6%.
6%.
In the
In the case
case ofof energy
energy simulations
simulations
simulations we we are interested
we are interested
interested in in long
long periods
periods
periods ofof calculation.
calculation.
calculation.
Therefore,
Therefore, the
Therefore, the inaccuracy
the inaccuracyfor
inaccuracy forthe
for thewhole
the wholeseries
whole seriesofof
series of measurements
measurements
measurements waswas
was determined.
determined.
determined. For
ForFor the
P3
thethe
P3
P3 series
series the
the the
series errorerror was
waswas
error 1.8%,
1.8%, andand
1.8%, and
in thein case
in the case
the case of second
of second
of second P4 series,
P4 series,
P4 series, the average
the average
the average error
error error was
was 0.9%.
was
0.9%.
This isThis
0.9%. dueis
This istodue
the to
due the
the reduction
reduction
to of
of negative
of negative
reduction and
and positive
and positive
negative deviations.
deviations.
positive deviations.
To
To sum up, it can be stated that energy simulations that are
sum up, it can be stated that energy simulations that are performed
performed with
with aa long
long
time-spans and
time-spans and sinusoidal
sinusoidal changes
changes in in heat
heat loss
heat loss trends
loss trends (successive
trends (successive increases
(successive increases and
increases and decreases)
and decreases)
decreases)
will be
will relatively precise.
be relatively precise.
Should
Should be noted that
be noted that the
the issues
issues discussed
discussed above
above are
are quite
quite complex
complex andand do not cover
do not cover
the whole scope of this
the whole scope of this problem. problem.

4.5. Determination of Nusselt Number

The Nusselt number for the flow through the pipes is calculated from the following
relationship:
h f Din
Nu = , (15)
kf
Energies 2021, 14, 90 18 of 21

where, hf —convection heat transfer coefficient, Din —inner diameter of the pipe, kf —thermal
conductivity.
The convection heat transfer coefficient was determined based on the calculation of
the heat flux qw−f transferred through the water-pipe wall boundary:

qw− f
hf = , (16)
Aw− f ·( TF − TW )

where, Aw−f —area of heat transfer surface, TF —water temperature, TW —wall temperature.
The intensity of the heat transfer inside the pipe depends primarily on the Reynolds
number value. Thus, the Nusselt number in the conditions occurring in the analyzed
hydraulic system (Re range from 20,000 to 40,000 and average water temperature of 40 ◦ C)
can be written in the following form:

Nu = 0.037Re0.8 , (17)
v f Din
where, Re = ν , vf —average water velocity, ν—kinematic viscosity.

4.6. Pressure Loss Characteristic

The characteristic of the pressure drop caused by the flow of water through the pipe
with a nominal diameter of DN40 are shown in Figure 19.

Figure 19. Hydraulic characteristic curve of the steel pipe DN40.

The pressure loss related to 1 m of the pipeline length can be approximated with high
accuracy (R-squared = 0.9996) by Equation (18):

∆p = 242.26V 1.9842 . (18)

Thus, the pressure loss in the 86 m long pipeline DN40 section at a constant flow
V = 0.63 l/s (2.28 m3 /h) was 8417 Pa.
Energies 2021, 14, 90 19 of 21

5. Summary, Conclusions, and the Future Work

The article presents the results of the analysis, which aimed to determine the heat loss
of pipelines that are part of the SDHW heating system. It is supported by solar collector
array consisting of 35 flat plate and 21 evacuated tube panels. Solar collectors are located
on the roof of the Hotel for young researchers at Bialystok University of Technology in
Bialystok (Poland). Four insulated steel pipes with a diameter of DN 40 mm and a total
length of 344 m were the subject of this study. The pipelines connect heat exchangers
located on the roof with heat storage tanks located in the basement of the building. There
is the continuous change in the temperature circulating in these pipes due to fluctuations
in the amount of energy converted by collectors during the day.
At the beginning of this work, it was decided to determine the heat loss based on
measuring the flow rate and temperature difference at the inlet and outlet of the system
(Equation (1)) i.e., in a classical way. However, it turned out that due to the dynamic nature
of the heat exchange in the solar installation, the use of the classic energy balancing method
is the cause of high inaccuracy. For this reason, a new approach has been proposed for
estimating the heat loss of pipelines with a continuous change in the supply temperature.
The development of a numerical model using CFD algorithms was the first step in solving
this problem. Then the model was validated using the measurement results. Thus, the
amount of the heat emitted from the outer surface of the pipelines in operating conditions
was determined on the basis of computer calculations. The results of the supply tempera-
ture measurements taken from the data acquisition system were used as input data. Based
on the approximation of a large database of computer simulation results, two relationships
were proposed (Equations (11) and (12)). They are the practical result of this work and
can be used in engineering calculations. It allows estimating the heat loss of pipelines in
stationary heat exchange conditions as a function of the difference in water and ambient
air temperature.
Building energy simulation software usually uses meteorological databases prepared
with an one-hour time step. It was also decided to estimate the error resulting from
assuming hourly averaging of SDHW system operation parameters. Based on a comparison
of the results of the heat loss calculations carried out with a 1-h and 1-min step, it can be
concluded that the differences for these time-spans are quite large and range from −9% to
6%. However, when it was balanced the whole day of the system operation, the total error
did not exceed 2% due to the elimination of divergences between the first and second half
of the measurement series.
In the next stage of this work, it is planned to develop the universal equations that
will allow the calculation of heat losses from pipes of different diameters and different
thickness and types of thermal insulation materials.

Funding: This work was performed within the framework of grant of the Bialystok University of
Technology WZ/WBiIS/9/2019 and financed by the Ministry of Science and Higher Education of the
Republic of Poland. Numerical simulations were carried out using ANSYS 19.2 software, which is
provided to the BUT on the basis of an agreement between BUT, ANSYS Inc. (Canonsburg, PA, USA),
and MESco LLC (Tarnowskie Gory, Poland).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Acknowledgments: This work was performed within the Grant of the BUT (WZ/WBiIS/9/2019)
and financed by the Ministry of Science and Higher Education of the Republic of Poland.
Conflicts of Interest: The author declare no conflict of interest.
Energies 2021, 14, 90 20 of 21

Nomenclature
Aw−f area of heat transfer surface (m2 )
cp specific heat at constant pressure (J/kg/K)
Din inner diameter of the pipe (m)
El pipeline energy loss (J or kWh)
eMRE mean relative error (%)
eRMSE root mean square error (◦ C)
hf convection heat transfer coefficient (W/m2 /K)
kf thermal conductivity (W/m/K)
l pipe length (m)
.
mc mass flow rate (kg/s)
n number of measurements
Nu Nusselt number
r radial coordinate (m)
Re Reynolds number
QS pipeline heat loss (W)
qS(1 h) heat loss per square meter with one-hour time step (W/m2 )
qS(1 min) heat loss per square meter with one-minute time step (W/m2 )
qS,l heat loss per unit length (W/m)
T, TC time (s)
TA supply temperature (◦ C)
TCFD calculated temperature (◦ C)
TB, TD return temperature (◦ C)
TE external temperature (◦ C)
TIN inlet temperature (◦ C)
TM measured temperature (◦ C)
TF water temperature (◦ C)
TW wall temperature (◦ C)
V volume flow rate (l/s)
vf average water velocity
vx axial velocity (m/s)
vr radial velocity (m/s)
x axial coordinate (m)
∆t time interval (s or h)
∆p pressure loss (Pa)
µ dynamic viscosity (Ns/m2 )
v kinematic viscosity (m2 /s)
ρ fluid density (kg/m3 )

References
1. Beckman, W.A. Duct and pipe losses in solar energy systems. Sol. Energy 1978, 21, 531–532. [CrossRef]
2. Tsuda, I.; Tanaka, T.; Tani, T. Influence of a piping on heat collecting rate of distributed collector system. In Advance in Solar Energy
Technology; Pergamon Press: Oxford, UK, 1988; pp. 1304–1308. [CrossRef]
3. Stubblefield, M.A.; Pang, S.-S.; Cundy, V.A. Heat loss in insulated pipe the influence of thermal contact resistance: A case study.
Compos. Part B Eng. 1996, 27, 85–93. [CrossRef]
4. Marshall, R. A generalised steady state collector model including pipe losses, heat exchangers, and pump powers. Sol. Energy
1999, 66, 469–477. [CrossRef]
5. El-Nashar, A.M. Heat loss through the piping of a large solar collector field. Energy 2006, 31, 2020–2035. [CrossRef]
6. Deeble, V.C. Effectiveness of PVC coatings as thermal insulation for domestic hot-water piping. Appl. Energy 1994, 48, 51–64.
[CrossRef]
7. Kecebas, A. Determination of insulation thickness by means of exergy analysis in pipe insulation. Energy Convers. Manag. 2012,
58, 76–83. [CrossRef]
8. Kohlenbach, P.; Ackermann, L.; Mörtl, K.; Punzel, G.; Osborne, J. Influence of oil-soaked insulation on the heat loss of thermal oil
piping used in high-temperature solar cooling applications. Energy Procedia 2014, 48, 739–748. [CrossRef]
9. Springer, D.; Seitzler, M.; Backman, C.; Weitzel, E. Using Solar Hot Water to Address Piping Heat Losses in Multifamily Buildings;
Office of Energy Efficiency and Renewable Energy, US Department of Energy: Washington, DC, USA, 2015. [CrossRef]
10. Yu, R.; Yan, D.; Feng, X.; Gao, Y. Investigation and modelling of the centralized solar domestic hot water system in residential
buildings. Procedia Eng. 2016, 146, 424–430. [CrossRef]
Energies 2021, 14, 90 21 of 21

11. Zaaoumi, A.; Asbik, M.; Hafs, H.; Bah, A.; Alaoui, M. Thermal performance simulation analysis of solar field for parabolic trough
collectors assigned for ambient conditions in Morocco. Renew. Energy 2021, 163, 1479–1494. [CrossRef]
12. EnergyPlus™. Weather Data by Region—All Regions—Europe WMO Region 6—Poland. Available online: https://energyplus.
net/weather-region/europe_wmo_region_6/POL%20%20 (accessed on 15 November 2020).
13. Zukowski, M. HVAC systems supported by renewable energy sources—Case studies carried out at the Bialystok University of
Technology. In Advances in Renewable Energy Research; Pawłowska, M., Pawłowski, A., Pawłowski, L., Eds.; CRC Press/Taylor &
Francis: Boca Raton, FL, USA, 2017; pp. 33–44.
14. Zukowski, M. Energy efficiency of a solar domestic hot water system. E3S Web Conf. 2017, 22, 00209. [CrossRef]
15. Ansys Fluent Theory Guide; v. 2019 R2; ANSYS Inc.: Canonsburg, PA, USA, 2019.
16. Launder, B.E.; Spalding, D.B. Lectures in Mathematical Models of Turbulence; Academic Press: London, UK, 1972.