Bibilography

Energy & Buildings 222 (2020) 110070

Contents lists available at ScienceDirect

Energy & Buildings

journal homepage: www.elsevier.com/locate/enbuild

Development and analysis of hourly DHW heat use profiles in nursing

homes in Norway

Dmytro Ivanko a , ∗, Harald Taxt Walnum b , Natasa Nord a

a Department of Energy and Process Technology, Norwegian University of Science and Technology (NTNU), Kolbjørn Hejes vei 1 B, Trondheim 7491, Norway b SINTEF Community, P.O. Box 124 Blindern, 0314 Oslo, Norway

a r t i c l e i n f o

Article history:

Received 26 November 2019

Revised 3 March 2020

Accepted 13 April 2020

Available online 8 May 2020

Keywords:

DHW heat use

Energy efficiency

Hourly profiles

Mathematical modelling

Statistical analysis

a b s t r a c t

Representative profiles for domestic hot water (DHW) heat use are the main instruments for improve-

ment in operation and design of DHW systems in buildings. To improve the existing method for DHW

heat use profiles development and analysis, investigations in the three nursing homes in Norway were

conducted. Statistical methods to assess the similarities of the profiles by days of the week and seasons

were proposed. The analysis allowed us to identify two seasons of DHW heat use: the warm season from

June to October, and the cold season including the rest of the year. In addition, it was investigated that

the DHW heat use in the working days was significantly different from the weekends. According to these

results, unified profiles for the months and days of the week with similar characteristics of the DHW heat

use were developed. After, the method for statistical grouping of the DHW hourly heat use was applied

to recognize the timing of the peak, average, and low heat use. Finally, the profiles for the DHW heat use

obtained for the nursing homes were compared with profiles in the national and international standards.

The drawbacks of the standards were identified.

© 2020 The Authors. Published by Elsevier B.V.

This is an open access article under the CC BY license. ( http://creativecommons.org/licenses/by/4.0/ )

1

d

i

e

(

t

o

a

fi

s

a

s

w

e

d

2

i

e

l

o

w

h

c

t

e

c

s

e

l

s

D

f

u

e

s

h

a

h

0

. Introduction

Nowadays, energy efficiency and decarbonisation are the key

riving forces in the development of European Union (EU) energy

ndustry. Among all sectors, buildings sector is one of the most

nergy-intensive. The Energy Performance of Buildings Directive

EPBD) estimates the share of energy use in building as 40% from

he total energy use in the EU [1] . Considering the huge potential

f energy saving in buildings, European Commission (EC) develop

set of long-term and short-term goals for increasing energy ef-

ciency in buildings [2] . For example, by 2020 all new buildings

hould be constructed in accordance with zero emission standards,

nd at least 3% of the total floor area of governmental buildings

hould be renovated [1] . The energy infrastructure in buildings that

ere built 30–40 years ago needs to be replaced by more energy

fficient [3] . According to Energy roadmap 2050 [3] , the goal to re-

uce CO 2 emission to 80–95%, when compared to 1990 level, by

050 scenarios is set [3] . To achieve this goal, all technical systems

n buildings must be designed and operated in such a way as to

nsure efficient energy use.

∗ Corresponding author. Phone number: ( + 47) 48670338 E-mail address: [email protected] (D. Ivanko).

w

t

b

2

ttps://doi.org/10.1016/j.enbuild.2020.110070

378-7788/© 2020 The Authors. Published by Elsevier B.V. This is an open access article u

Until recently, in many European countries, including Norway, a

ot of effort has been put on the investigation of the performance

f the space heating systems [4] . Meanwhile, the DHW heat use

as considered as a small part of the energy needs required for

eating. Therefore, DHW heat use has obtained little focus, espe-

ially in countries with cold climate [5] . However, with introduc-

ion of passive house technologies and improvement of building

nvelope, the space heating heat use in buildings is constantly de-

reasing. At the same time, reduction of DHW heat use remains in-

ignificant [6] . For example, the experience from design of low en-

rgy buildings in Denmark is shared in [7] . In this study, to achieve

ow heat use, passive building strategies with highly insulated, re-

ource efficient, and airtight solution are used, without focusing on

HW use. The authors in [7] conclude that detailed design values

or the passive building show that energy demand for the DHW

se is almost twice bigger than space heating. The analysis of en-

rgy use in four apartment buildings in Finland with various con-

truction years is performed in [8] . In this study, to assess DHW

eat use, the profiles obtained from measured DHW demand in

partment buildings are used as input in IDA-ICE simulation soft-

are. Simulation shows that in the modern buildings, the domes-

ic hot water is the most significant component in heat use. In two

uildings constructed before 2002, the DHW heat use contributes

4% and 30% to the total energy use in the buildings. However, in

nder the CC BY license. ( http://creativecommons.org/licenses/by/4.0/ )

2 D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070

v

h

t

p

n

e

e

s

a

h

e

t

q

s

p

D

m

b

t

s

h

D

s

p

b

t

p

i

r

e

t

p

m

r

s

o

t

fi

e

i

i

b

p

fi

a

t

a

a

p

i

t

t

D

t

a

B

g

i

p

e

a

h

a

m

two well-insulated buildings, the DHW systems is responsible for

52% and 63% of the total energy use. As we can see, the share of

the DHW energy use tends to increase from approximately 20% in

regular buildings [5] to above 50% in passive houses and well in-

sulated buildings [9] . Consequently, heat use for DHW systems is

becoming the critical component for energy saving, especially in

passive houses and nearly zero energy buildings (NZEB) [10] .

Nowadays, heat losses from the hot water tanks and the cir-

culation systems in houses, schools, and other institutions remain

high [11] . As a result, further energy saving measures in buildings

should shift the focus from improving space heating to improve

DHW systems. To realize potential of energy savings in DHW sys-

tems, the research and innovations in the field of DHW energy per-

formance becoming increasingly relevant and valuable [9] .

It should be noted that the operation of the DHW systems

is associated with sanitary and health safety issues. These issues

for different types of buildings is discussed in [12] . Appearance

of Legionella bacterium in DHW systems is a serious problem.

Legionella bacterium can lead to different forms of pneumonia

and even death. The conditions for Legionella spreading are water

temperatures from 25 °C to 42 °C, nutrients, and stagnating water. Therefore, many countries, including Norway, develop regulations

to minimize the risk of Legionella disease appearance. For exam-

ple, despite of energy ineffectiveness, to prevent risks of the bac-

teria growth, the DHW systems in Norway store and distribute hot

water at temperatures above 60 °C. Among all buildings, special at- tention is paid to the nursing homes, because the elderly, who usu-

ally have respiratory problems and weakened immune system, are

heavily affected by this bacterium. The safety of energy effective

solutions is the key factor in DHW systems.

The share of DHW heat use is varying from country to country

and one type of building to another [5] . For example, specific DHW

heat use in households in different EU countries are significantly

varying as shown in [13] . Comprehensive comparison of DHW en-

ergy use in residential buildings in Denmark, Norway, and Swe-

den is performed in nineties [14] . Even though that study is some-

what outdated, it describes well the general trends in the DHW

use in these countries. Sweden, Norway, and Denmark share a sim-

ilar living standard, comparable patterns of household formation,

and a similar climate. Nevertheless, the DHW heat use in Den-

mark is significantly below those in Sweden and Norway. In ad-

dition, the authors conclude that national average, electricity use

per capita for the DHW heating in Norway has almost not changed

for 15 year, and remains high when compared with other countries

within The Organization for Economic Cooperation and Develop-

ment (OECD). The authors explain this phenomenon by difference

in occupants’ behaviour and the insulation of DHW systems in

different countries. More resent research confirms this statement

[15] and it shows that the average individual DHW use reaches 40

L/person/day in Norway, while in Denmark, the average value is at

20 L/person/day [15] .

For the sake of simplification, many methods propose to con-

sider the DHW use as a constant value for calculations [16] . Practi-

cal experience shows that the commonly used standards are based

on assumptions for the DHW heat use in the buildings, but these

standards do not correspond to the real use [17] . For example, sim-

plified, but meantime common way of DHW system performance

simulations is shown in [18] . Further, DHW system performance

are simulated based on daily water need as a constant value of 90

l/day per bedroom and with 25 K temperature difference between

supply and return in [18] . Such simplifications could lead to over-

sizing of the components for DHW systems and additional financial

and energy losses [19] .

DHW heat use profiles are the primary instrument for estimat-

ing the DHW heat use in the buildings [5] . Analysis of DHW heat

use profiles shows the changes in heat use in different time inter-

als [20] . The profiles of DHW heat use allow us to determine the

ours of peak energy loads and other energy load characteristics of

he building.

Performance of DHW systems is a complex and multidisci-

linary issue. It includes economic, sanitary, behavioral, and tech-

ical areas. DHW heat use profiles is a useful for identifying en-

rgy efficient solutions within all these areas. For example, the

conomic analysis of DHW pricing is performed in [21] . The study

hows that the DHW use positively correlated with income and re-

cts to the changes in water prices. Introduction of new energy or

eat tariffs is a way of reducing the DHW use is buildings. How-

ver, in order to implement advanced and flexible energy or heat

ariffs, the in-depth knowledge about profiles of DHW use is re-

uired. Technical solutions dealing with sanitary problems are con-

idered in [22] . Some of these solutions require knowledge of the

rofiles and timing when DHW water is used. Different types of

HW heating systems are investigated in [23] . This study sum-

arises that DHW energy use can be reduced through using com-

ined systems based on traditional and renewable energy solu-

ions. However, due to unstable behaviour of renewable energy

ources, development of accurate profile and prediction of DHW

eat use becoming crucial for successful operation of combined

HW heating systems. Most of building simulation software tools

uch as IDA ICE, EnergyPlus, TRNSYS, TRANSOL, etc. require DHW

rofiles as the basis for simulation of DHW systems performance in

uildings [5] . For example, it is noted that the variations between

he simulated and the real heat use for DHW are caused by inap-

ropriate profiles [24] . Consequently, the authors in [24] claim that

nput data for DHW volume flow rates used in the standards rep-

esent perhaps one of the more critical points in simulation mod-

ls. Therefore, actual knowledge of DHW usage profiles can capture

he real heat use in buildings, making it possible to size systems

roperly. Effective demand-side management, energy conservation

easures, improvement of legislation and standards require accu-

ate DHW profiles for different types of buildings [25] . As we can

ee, scientific and practical work confirms the need to use profiles

f the DHW heat use to solve important issues in the DHW sys-

ems.

The issue of DHW heat use analyses in buildings based on pro-

les is investigated by researchers in Norway and abroad [5] . How-

ver, due to differences in particular characteristic of each build-

ngs, quality of available data, and calculation requirements, there

s no unique method of performing appropriate analysis. The num-

er of scientific works is dedicated to the issue of DHW energy

rofiles development and analysis. For example, hourly DHW pro-

les for five groups of buildings with 1, 3, 10, 31, and 50 residents

re developed based on data from Finnish apartments in [26] . Fur-

her, the profiles for each group with the closest to mean profile

nd have a similar shape, are selected among measured candidates

s representative. The volumetric flow rates, cold and supply tem-

eratures are measured to characterise the DHW use in 20 build-

ngs of different sizes in [27] . Based on the obtained data, the au-

hors executed several stochastic simulations to get a representa-

ive DHW use profiles for end users [27] . Number of methods for

HW profiles development are based on operating schedules for

he primary DHW energy users (showers, baths, sinks, dishwasher,

nd clothes washer) and occupant activities. As an illustration, the

uilding America House Simulation Protocols document provides

uidance for such analysis in new and existing apartment build-

ngs [28] . Lombardi in [29] shows that domestic water use can be

resented as the result of probabilistic use of domestic appliances,

ach one with its particular characteristics. The research of Good

nd Zhang in [30] share the experience of calculation for DHW

eat use profiles based on occupant activities. The DHW modelling

pproach by the coupling of behavioural activities, energy balance

odels, and stochastic modelling is presented in [31] . Time-use

D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070 3

d

D

h

p

o

a

o

k

c

d

r

t

f

i

o

i

c

i

p

b

s

m

t

d

fi

u

fi

H

c

b

t

h

m

r

o

a

r

s

f

i

a

o

k

i

s

n

a

o

r

v

r

e

a

u

s

s

s

T

i

p

t

w

Fig. 1. Method for the analysis of DHW heat use profiles.

i

e

s

t

a

t

l

s

t

N

d

a

s

F

S

2

m

u

m

f

s

v

u

m

d

t

S

o

t

t

e

s

2

o

t

p

b

t

a

ata of activities in households in Sweden are used for generating

HW profiles in [32] . For DHW energy analysis, the occupant be-

aviour, appliance ownership, demographic conditions, and occu-

ancy rate are considered in neural network model in [33] . Most

f the reviewed research work are dedicated to the apartments

nd households, meaning that required parameters were easier to

btain. However, in non-residential buildings obtaining in-depth

nowledge about occupant activities and equipment operation be-

ome time consuming and expensive task [5] . The available input

ata limits the practical application of these methods.

The problem of validation of DHW simulated profiles in non-

esidential buildings is proposed in [34] . For simulation, the au-

hors use SIMDEUM in [35] , which is based on the design rules

or appliance performance and dominant variables in buildings. It

s assumed that the dominant variable for hotels is the number

f rooms, for offices is the number of employees, and for nurs-

ng homes is the number of beds [36] . The validation procedure

onsists of two steps. In the first step, the outcome of simulation

s compared with measured demand values. In the next step, it is

roposed to check if the assumptions on the standardized building

ased on the design rules are validated with measurements and

urveys [34] . This study shows that it is challenging to find infor-

ation of users and appliances in each functional room to equip

he standardised buildings. However, regular demand pattern for

ominant functional room can be obtained.

The problems of comparing the actual DHW energy use pro-

les with the standards, and their verification, are also not going

nnoticed. For example, the comparison of the actual DHW pro-

les in apartments with profile proposed by American Society of

eating, Refrigeration and Air Conditioning Engineers (ASHRAE) is

onducted in [37] . The research shows that the primary difference

etween the actual and the ASHRAE derived data is that the wa-

er use is less evenly distributed in the actual data, and there are

igher peaks and lower troughs and much less use in the early

orning hours in the actual data. Differences in shapes and pa-

ameters of the actual DHW heat use profiles for particular types

f buildings and profiles represented in publications and standards

re considered in [38] . As a conclusion, in this work, the authors

ecommend to rely on actual profiles obtained from measurement

ystems for the analysis of DHW use in the existing buildings.

The aim of our paper was to improve the existing approaches

or the DHW heat use analysis and gain in-depth knowledge about

t in nursing homes in Norway. Non-residential buildings such

s nursing home, hospitals, hostels, schools, etc. in Norway and

ther European countries are less studied then residential [5] . The

nowledge about actual DHW heat use profiles in nursing homes

n Norwegian is currently incomplete and contain many gaps. The

tudy in [39] shows that the specific heat use in the hospitals and

ursing homes in Norway is approximately 270 kWh/m 2 per year,

nd one of the highest comparing to other types of buildings. Quite

ften, profiles presented in standards for nursing homes cannot

epresent the actual DHW heat use [39] . For this reason, the in-

estigation on the DHW heat use in nursing homes in Norway is

equired. Such a study is the basis for the further introduction of

nergy saving in nursing homes in Norway.

In this article, we presented the methods for developing and

nalysing profiles for DHW heat use. The proposed methods allow

s to assess the similarities of the profiles by days of the week and

easons, and identify the timing of the peak heat use of the DHW

ystem. The methods were tested based on one-year hourly mea-

urements from three nursing homes, located in Eastern Norway.

he unified profiles for the months and days of the week with sim-

lar characteristics of the DHW heat use were identified. For these

rofiles the timing of the peak, average, and low heat use was es-

imated. The profiles obtained from measurements were compared

ith profiles from the national standard SN/TS 3031:2016 [41] and

nternational standard NS-EN 12831-3:2017 [42] . The possible ben-

fits from using more accurate energy profiles, obtained by mea-

urements and statistical analysis, are explained in this study.

The paper was organised as the following. Section 2 introduced

he method for developing profiles, divided by days of the week

nd seasons with similar characteristics of the DHW heat use. In

his section, the method for determining the peak, average, and

ow zones of the DHW heat use from the profiles was also pre-

ented. Section 3 explained the main characteristics of DHW sys-

em for the case study - three nursing homes located in eastern

orway. In Section 4 , the method was implemented on the real

ata. The obtained profiles for the nursing homes were analysed

nd compared with the profiles of DHW heat use given in the

tandards. The main results of this investigation were presented.

inally, the main conclusions of the study were emphasized in

ection 5 .

. Method

The method for the analysis of DHW profiles included the four

ain steps shown in Fig. 1 .

The three following subsections covers the methods that were

sed to solve issues in shown Fig. 1 . Section 2.1 described the

ethod for comparison of the DHW heat use profiles from dif-

erent days of the week and assessing their similarities. In this

tudy, we did not assume, beforehand that the profiles can be di-

ided in a certain way. Student’s t-test and Fisher’s exact test were

sed for solving this issue. By using this method, the data tests

ay be used for samples with standard normal distribution and t-

istribution. This allowed to us to determine the statistically jus-

ified days of the week with similar DHW heat use profiles. In

ection 2.2 , a method for determining the duration and boundaries

f time zones with peak, minimum, and average heat use during

he day was showed. In Section 2.3 , a statistical method for iden-

ifying the number of seasons, as well as the months included in

ach season was described. By using this method, the impact of

easonality on DHW heat use was taken into account.

.1. Comparing similarity of DHW heat use profiles in different days

f the week

To determine the days of the week with similar characteris-

ics of DHW heat use, a method based on test statistics was pro-

osed. The similarity of two DHW heat use profiles is checked

ased on the Student’s t-test and Fisher’s exact test. Appropriate

ests can be used for samples with standard normal distribution

nd t-distribution.

4 D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070

Table 1

The form of the matrix of matches.

Mo. Tu. We. Thu. Fr. Sa. Su.

Mo. n 1.1 − − − − − − Tu. n 2.1 n 2.2 − − − − We. n 3.1 n 3.2 n 3.3 − − − − Th. n 4.1 n 4.2 n 4.3 n 4.4 − − − Fr. n 5.1 n 5.2 n 5.3 n 5.4 n 5.5 − − Sa. n 6.1 n 6.2 n 6.3 n 6.4 n 6.5 n 6.6 − Su. n 7.1 n 7.2 n 7.3 n 7.7 n 7.5 n 7.6 n 7.7

i

p

t

p

n

o

d

t

w

D

p

p

p

w

N

s

o

t

c

i

2

h

t

p

s

h

s

t

s

h

o

m

c

p

s

m

e

e

m

f

z

t

u

By applying the Student’s t-test, it was possible to check if the

mean values of DHW heat use from two days of the week were

equal or not. To achieve this, the DHW heat use within each day

was considered as a statistical sample with 24 elements, which

represented the number of hours in the day. The t-test statistical

value was calculated as follows:

T cal = Ē pro f 1 − Ē pro f 2 √

S 2 pro f 1

n pro f 1 + S

2 pro f 2

n pro f 2

(1)

where Ē pro f 1 , Ē pro f 2 were the mean values of the DHW heat use in

the first and second samples. S prof 1 , S prof 2 were the standard devia-

tions of the DHW heat use profiles in the first and second samples.

n prof 1 , n prof 2 were the number of elements in the first and second

samples. Finally, the equation for the standard deviation for i -th

day was written as:

S pro f i =

√ ∑ ( E pro f i. j − Ē pro f i

)2 n pro f i − 1

(2)

where i was the number of the sample, j was the number of ele-

ment in the sample, E profi.j was the DHW heat use in j-th element

in i -th sample.

The obtained value for t-criteria, T cal , was compared with the

critical value, T cr . T cr may be found in literature for different de-

grees of freedom and significance level k . The comparison may lead

to three possible situations as the following:

- If T cal ≤ T cr ( n pro f 1 + n pro f 2 − 2 , k = 0 . 05 ) , then the mean values of the first and the second samples are similar;

- If T cal ≥ T cr ( n pro f 1 + n pro f 2 − 2 , k = 0 . 01 ) , then the mean values of the first and the second samples have a significant differ-

ence;

- If T cal ≤ T cr ( n pro f 1 + n pro f 2 − 2 , k = 0 . 01 ) and T cal ≥ T cr ( n pro f 1 + n pro f 2 − 2 , k = 0 . 05 ) , then the mean values of the first and the second samples may be considered as similar.

However, the final decision should be done based on the

knowledge of researchers.

Meanwhile, Fisher’s criterion allowed us to estimate the simi-

larity of two samples by variances:

f cal = max

( S 2

prof 1 , S 2

prof 2

) min

( S 2

prof 1 , S 2

prof 2

) (3) The comparison obtained by calculations of the Fisher criterion,

f cal with its critical value, f cr led to the following results:

- If f cal ≤ f cr ( n pro f 1 + n pro f 2 − 2 , k = 0 . 05 ) , then the variances of the first and the second samples are similar;

- If f cal > f cr ( n pro f 1 + n pro f 2 − 2 , k = 0 . 05 ) ,then the variances of the first and the second samples have significant difference.

The two profiles are considered to be similar if both Student’s

t-test and Fisher’s exact test show the same results. If at least one

of two tests shows that the mean values or variances of profiles

in the first and the second samples are not similar, it is possible

to conclude that the profiles are dissimilar and should be analysed

separately.

Splitting the DHW profiles by the days of the week should be

made based on a large dataset, which represents DHW heat use

during the year. Therefore, in this study, it was proposed to di-

vide initial statistical data into separate weeks. Within each week,

all combinations of the daily DHW profiles should be compared

among themselves by Student’s t-test and Fisher exact test. For in-

stance, profiles for Monday and Thursday, Monday and Wednesday,

Saturday and Sunday and so on should be compared. Afterwards,

for all the combinations of days, the number of the weeks can be

dentified, when statistical tests show that profiles in considered

airs of days are similar. For further analysis, for each combina-

ions of days of the week, the number of matches of the DHW

rofiles in percentage can be found as:

i . j = N i . j · 100 / N total (4) The elements in Equation (4) are the following, n i.j is number

f matches in percentage, when the DHW profiles of i-th and j-th

ays were similar. N i.j was the number of the weeks, when sta-

istical tests showed that the i-th and j-th days were similar. N total as the total number of the weeks in the statistical data sample of

HW heat use. i was the day of the week of the first comparable

rofile (from 1 to 7). j was the day of the week of the second com-

arable profile (from 1 to 7). For better clarity, the results could be

resented in the form of matrix of the matches as in Table 1 .

Based on the matrix of matches, the groups of the days of the

eek with similar profiles of DHW heat use could be identified.

amely, the days of the week, which have n i. j ≥ 100 − er ror , have imilar characteristics of DHW heat use and should be placed in

ne group and analysed together. The value of the error included

he accuracy of Student’s t-test, Fisher’s exact test, and the per-

entage of days in the year when the building is not in operation

n typical regimes such as holidays.

.2. Determining the time zones with peak, minimum, and average

eat load for daily profiles of DHW heat use

To implement energy management in buildings, it is essential

o identify the typical duration and boundaries of time zones with

eak load, minimum, and average heat load during the day. To

olve this issue, we proposed to perform statistical grouping of the

ourly heat use of the DHW system based on the method pre-

ented by Nakhodov in [40] . Initially, this method is used for iden-

ification of the tariff zones of electricity energy use in the power

ystem. In this article, we adapted the method for analysis of DHW

eat use in buildings. The method allowed us to divide the hours

f DHW heat use into several groups with statistically different

ean values within each group. It is based on an iteration pro-

edure and analysis of the mean values of DHW heat use by ap-

lying Student’s t-test. In this case, DHW heat use profile was con-

idered as a statistical sample e . The sample contained N = 24 ele- ents (hours) with DHW heat use in these hours equal e j (where

j was DHW heat use in the j-th hour. j was the number of the

lement in the sample). The flowchart for the algorithm for deter-

ining the time zones with peak, minimum, and average heat load

or daily profiles of DHW heat use is shown in Fig. 2 .

The detailed algorithm of the method for determining the time

ones was as the following:

Step 1. Sorting the elements of the sample in the order of their

increase

The elements e j in the sample e were sorted in the order of

heir increase. Such an arrangement of elements from smaller val-

es of hourly DHW heat use to bigger values allowed us to obtain

D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070 5

Fig. 2. Flowchart for the algorithm for determining the time zones with peak, min-

imum, and average heat load for daily profiles of DHW heat use.

t

n

t

m

M

s

E

s

t

T

i

T

c

T

t

e

p

o

c

s

c

b

o

l

s

i

g

h

i

t

s

t

c

p

t

s

i

t

t

z

b

E

E

w

u

M

a

d

T

a

u

m

E

w

u

u

2

i

D

f

p

t

h

b

i

3

c

T

t

s

he sorted sample E with N elements E i (where E i +1 > E i , i is the umber of element in sample E ).

Step 2. Identifying the initial groups for the elements that could be

considered statistically similar

Based on the sample E , an iterative procedure of generating of

wo statistical subsamples R 1 and R 2 with variable number of ele-

ents was applied. For each step of iteration, sample R 1 contained

elements, while R 2 should have M + 1 elements. The elements in amples R 1 and R 2 were taken consistently from the initial sample

. With each iteration, the number of elements M in these sub-

amples increased by one. The value of M varied from 1 to 23.

For each step of these iterations the value of Student’s t-test for

wo subsamples R 1 and R 2 were calculated by using Equation (1) .

For instance: iteration 1) R 1 = [ E 1 ] , R 2 = [ E 1 , E 2 ] , M = 1, and cal 1 ; iteration 2) R 1 = [ E 1 , E 2 ] , R 1 = [ E 1 , E 2 , E 3 ] , M = 2, and T cal 2 ;… teration 23) R 1 = [ E 1 , E 2 . . . E 23 ] , R 1 = [ E 1 , E 2 . . . E 24 ] , M = 23, and cal 23 ;

Step 3. Checking the possibility of merging the closest groups ac-

cording to Student’s t-test

Based on the iteration procedure of Step 2, the series of t-

riteria for all the combinations of the subsamples R 1 and R 2 ,

cal = [ T cal 1 , T cal2 . . . T M ] were found. If an ordered sample of hourly DHW heat use was monotonous,

hen the numerical values of elements in this sample increase

venly. In that case, the series of t-criteria obtained by iteration

rocedure would also be monotonous. This means that the values

f t-criteria obtained by Equation (1) would decrease monotoni-

ally with each next iteration ( T cal1 > T cal2 . . . > T M ). If the ordered

ample of hourly DHW heat use was uneven, then a monotonic de-

rease of the calculated values of the t-criteria would be violated

y periodic abrupt growth ( T cali < T cali +1 ). Thus, the identification f points of growth of the calculated values of the t-criteria al-

owed us to determine between which hours there is a noticeable

tatistical difference of DHW heat use. This assumption allowed us

nitially to divide hours in the profile of DHW heat use into several

roups. Each of these groups was the sample of data, where DHW

eat use data varied monotonously. Created in this way, neighbour-

ng groups of hourly DHW heat use could be checked in terms of

he possibility for their further merge. For this purpose, the data

amples of two neighbouring groups were assessed by Student’s

-test (see Equation (1) ). As a result, the calculated value of the t-

riteria, T cal , could be compared with critical value, T cr . This com-

arison could lead to the three possible situations:

- If T cal ≤ T cr ( n grou p1 + n grou p2 − 2 , k = 0 . 05 ) , then the mean values of the two groups were similar and should be merged;

- If T cal ≥ T cr ( n grou p1 + n grou p2 − 2 , k = 0 . 01 ) , then the mean values of the two groups were different and they should be considered

separately;

- If T cal ≤ T cr ( n grou p1 + n grou p2 − 2 , k = 0 . 01 ) and T cal ≥ T cr ( n grou p1 + n grou p2 − 2 , k = 0 . 05 ) , then the mean values of the two groups could be considered as similar. However, the

final decision should be done based on the knowledge of

researcher.

After we merged the groups based on explained above condi-

ions, the new set of groups was created. The calculations of Step 3

hould be repeated from the beginning with the new set of groups

n the sample. Iterative calculations of Step 3 was continued until

he t-test showed that no groups can be merged together and that

he total number of groups could not be reduced.

Step 1. Based on the groups with the elements, identifying the crit-

ical borders that separated the DHW heat use profile into zones

with peak, average, and minimum heat use

Critical borders that separated the DHW heat use profile into

ones with peak, average, and minimum heat use can be identified

y the following:

min = Ē group . 1 + T cr . 1 ( M group . 1 + 1 − 2 , k = 0 . 01

)√ S 2 group . 1

M group . 1 (5)

max = Ē group . K −1

+ T cr . K −1 ( M group . K −1 + 1 − 2 , k = 0 . 01

)√ S 2 group . K −1

M group . K −1 (6)

here Ē grou p. 1 , Ē grou p.K−1 were the mean values of the DHW heat se in the first group and the next to the last group. M group .1 ,

grou p.K−1 were the numbers of the elements in the first group nd the next to the last group. S 2

grou p. 1 , S 2

grou p.K−1 were the standard eviations in the first group and the next to the last group. T cr .1 ,

cr.K−1 were the critical values of the t-criteria for the first group nd the next to the last group. The hours in which the DHW heat

se was below E min should be considered as zone with the mini-

um DHW heat use. If the DHW heat use was between E min and

max , it could be assumed that in these hours the DHW heat use

as in a zone of average heat use. The hours with the DHW heat

se higher than E max lied within the zone of the maximum heat

se.

.3. Determining the seasons of DHW heat use

The method described in Section 2.2 can be applied in order to

dentify the groups of months with similar characteristics of the

HW heat use. In this case, in contrast to the sample of 24 hours

or each daily profile as considered in Section 2.2 , the initial sam-

le contains 12 elements for the monthly DHW heat use during

he year. The basic principles and procedure of calculations in both

ourly and monthly analysis was the same. As a result, the num-

er of seasons of the DHW heat use in the year and the months

ncluded in each season could be identified.

. Description of buildings

One year of hourly measured data for the DHW heat use were

ollected from three nursing homes located in the Eastern Norway.

he characteristics and work regimes of the nursing homes were

ypical for Norwegian conditions and was expected to be repre-

entative for DHW heat use in the similar types of buildings.

6 D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070

T a

b le 2

M a

in p

ro p

e rt

ie s

o f

m e

a su

re d n

u rs

in g h

o m

e s.

ID

A re

a [m

2

] N

u m

b e

r o

f ro

o m

s H

e a

t so

u rc

e

D is

tr ib

u ti

o n h

e a

ti n

g

S to

ra g

e

M e

a su

re m

e n

t

N H

1

3 3

2 7

5 2

E le

c tr

ic w

a te

r h

e a

te rs 3 ×

2 5 k

W

E le

c tr

ic h

e a

t tr

a c in

g

3 ×

6 0

0 li

te r

E le

c tr

ic it

y e

n e

rg y to w

a te

r h

e a

te rs

N H

2

2 6

0 9

2 5

E le

c tr

ic w

a te

r h

e a

te rs 2 ×

2 5 k

W

E le

c tr

ic h

e a

t tr

a c in

g

2 ×

9 0

0 li

te r

E le

c tr

ic e

n e

rg y to w

a te

r h

e a

te rs

N H

3

6 7

7 4

5 0

L o

c a

l a

re a h

e a

ti n

g +

e le

c tr

ic w

a te

r h

e a

te rs 3 ×

1 5 k

W

C ir

c u

la ti

o n (s

h o

rt -c

ir c u

it e

d )

3 ×

4 0

0 li

te r

E le

c tr

ic e

n e

rg y to w

a te

r h

e a

te r

a n

d th

e rm

a l

e n

e rg

y fr

o m d

is tr

ic t

h e

a ti

n g

a

t

i

d

b

t

w

d

h

t

c

m

d

d

m

i

o

f

t

u

4

c

v

i

S

a

p

i

o

w

4

u

s

o

e

m

i

m

a

F

e

r

p

t

D

b

p

t

p

h

a

t

[

h

t

Table 2 shows the main properties of the observed buildings,

nd Fig. 3 shows the principle layout of the DHW plants, including

he measurement points. The energy meters are marked with EM

n Fig. 3 .

For all the buildings, the measured heat use was the total heat

elivered into the system, i.e. including the heat losses. The two

uildings, NH1 and NH2 did not have hot water circulation sys-

ems, but electric heat traces. The power use of the heat tracers

ere not included in the measurements, which means that the

istribution losses were not accounted. The third building, NH3,

ad a circulation system, but the system was short-circuited close

o the heating plant, which means that the thermal losses in the

irculation were minimal. Based on this, it was assumed that the

easured heat use for the DHW in all the buildings were without

istribution losses, and thereby compared on equal ground.

The main differences between the nursing homes was the room

ensity (the total area per room), with a range from 64 to 136

2 /room. All nursing homes have private rooms only, all with the

ndividual bathrooms, and the nursing homes are normally fully

ccupied. Therefore, the number of rooms was also representative

or the amount of people living in the buildings. For investigation,

he weather data obtained from the closest weather station were

sed.

. Results and Analysis

The section is divided in several subsections that consider spe-

ific steps of the method explained in Section 2 . The analysis of the

ariation of DHW heat use in the nursing homes, as well as the

ndicators that explains its variability, was shown in Section 4.1 .

ection 4.2 investigates the nursing homes DHW heat use profiles

ggregated by similar days of the weeks and seasons. The hours of

eak, average and minimum heat use for these profiles were stud-

ed. In Section 4.3 , the standards were compared with the profiles

btained from the measurements. The drawbacks of the standards

ere highlighted.

.1. Initial analysis of DHW heat use in the nursing homes

Even within the same building type, the characteristics of heat

se may vary. To compare buildings with different characteristics,

pecific heat use may be used. Specific heat use is actual heat use

f the building divided by certain physical indicator. This indicator

xplains variability of the DHW heat use in different buildings, and

akes them comparable with each other. For this purpose in build-

ngs, the specific heat use per number of rooms or area is com-

only used. To choose which of these indicators to use in further

nalysis, the box plots of daily heat use were analysed as shown in

ig. 4 .

The results in Fig. 4 show that the relative difference in the av-

rage daily use is 67 % per area and 41% per room. Since the main

eason for DHW use at nursing homes are related to hygienic pur-

oses and nourishment of the residents, it is reasonable to think

hat the number of rooms is better parameter for describing the

HW heat use. Accordingly, in the further analysis, attention will

e paid mainly to the specific energy use per room. Only in the

arts of the article dedicated to the standards, where it is relevant,

he heat use per m 2 also will be considered. The DHW heat use

er room is quite high (see Fig. 4 b)) since rooms in the nursing

omes have large area from 64 to 136 m 2 /room.

The nursing homes considered in the article had similar trends

nd regimes of the DHW heat use. The difference in variance in

heir DHW heat use was within 30%. The energy distance test

41] showed that distributions of the DHW heat use in nursing

omes were identical and it provided a foundation for further sta-

istical analysis. Therefore, in order to simplify analysis and make

D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070 7

Fig. 3. Principle layout of the three DHW plants.

Fig. 4. Box plot of daily DHW heat use in the nursing homes, where: a) DHW heat use per m 2 , b) DHW heat use per room.

Fig. 5. Average hourly DHW heat use in three nursing homes.

t

t

a

i

t

S

S

f

s

w

p

e

s

o

t

a

2

t

s

i

i

f

f

d

u

T

4

a

m

h

h

he results more representative, the average DHW heat use of the

hree nursing homes was investigated. One-year data of the aver-

ge specific DHW heat use for the three nursing homes are shown

n Fig. 5 .

From Fig. 5 it can be noted that the DHW heat use during

he year were varying, and seasonal influence was clearly present.

easonality of the DHW heat use will be explained in detail in

ection 4.2 . In addition, some spikes may be noted in the data,

or example on September 9 th , 2018 at 24:00 o’clock. These spikes

howed untypical behaviour of the DHW heat use. Untypical spikes

here taken in account in the analysis of DHW heat use. Another

oint that was taken into account in the analysis was the differ-

nce in behaviour on holidays compared to ordinary days. Fig. 6

hows the DHW heat use in the nursing homes in the week with-

ut holidays (from January 1 st to January 13 th ), the week that con-

ained Christmas holidays (from December 24 th to December 30 th ),

nd days which are official public holidays (from December 25 th to

6 th December, and January 1 st ).

As we can see from Fig. 6 , the shapes of the DHW heat use pat-

erns during the public holidays on December 25 th and 26 th were

imilar to the patterns in the weekends. The DHW heat use dur-

ng the week that contained Christmas holidays was lower than

n a regular week. This can be explained by the fact that some

amilies took their elder relatives home from the nursing homes

or Christmas celebrations. Finally, on the last day of holidays el-

er people were arriving back to the nursing home. Therefore, Jan-

ary 1 st , the DHW heat use was becoming similar to a regular day.

hereby, during the holidays, water use was usually reduced.

.2. DHW heat use profiles aggregated by similar days of the weeks

nd seasons

Fig. 7 shows average daily DHW heat use per room for each

onth and corresponding outdoor temperature.

From Fig. 7 , strong negative correlation between monthly DHW

eat use and outdoor temperature may be noted. In nursing

omes, it is expected that the routines for DHW use are simi-

8 D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070

Fig. 6. DHW heat use within the Christmas holidays.

Fig. 7. Daily DHW heat use and outdoor temperature for different months over the year.

Fig. 8. Cold and warm seasons of DHW heat use in nursing homes.

y

a

p

T

i

s

c

p

n

lar around the year, and the variation on monthly heat use for

DHW can be described by the variation in cold fresh water inlet

temperature [42] . Through our investigation of the correlation be-

tween the monthly heat use and the lagged monthly average out-

door temperature, the highest coefficient of determination, 0.96,

was found between the monthly heat use and the average out-

door temperature of the previous month. This fits well with the

fact that the cold inlet water temperature has a slow response to

the outdoor temperature. Further, this effect leads to seasonal vari-

ation of the DHW heat use in the nursing homes. Therefore, to take

into account variation of the DHW heat use in the nursing homes

over a year, the seasonality was investigated. The number of sea-

sons during the year and the months associated with each season

were identified based on the average daily DHW heat use for nurs-

ing homes in different months, applying the method described in

Section 2.3 . Using Student’s t-test, the months of the year were di-

vided into two groups with substantially different mean values of

the heat use within each group. The results of the seasonality iden-

tification are shown in Fig. 8 . The groups represent the cold and

warm seasons. The cold season included the following months:

January, February, March, April, May, November, and December.

Meanwhile, June, July, August, September, and October were as-

signed to the warm season. Finally, for these seasons were devel-

oped separate profiles of DHW use.

As explained in the method, Section 2.1 , at the next step of

the investigation, the days of the week were assessed for simi-

larity. The DHW heat use data from nursing homes were divided

into separate weeks. In total, there were 52 full weeks within the

ear. According to the method in Section 2.1 , within each week,

ll combinations of daily DHW profiles were systematically com-

ared among themselves by Student’s t-test and Fisher exact test.

he matrix of matching of daily profiles is shown in Table 3 .

In order to find the critical value that shows when the profiles

n different days of the week could be considered as statistically

imilar, the three following factors were taken in account: the ac-

uracy of Student’s t-test, the accuracy of Fisher’s exact test, and

ercentage of days in the year when the buildings operation was

ot typical, including holidays. The accuracy of Student’s t-test and

D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070 9

Fig. 9. Profiles of DHW heat use in the nursing homes divided by day of week and seasons.

Table 3

Matrix of matching daily DHW heat use profiles in nursing homes.

Mo. Tu. We. Thu. Fr. Sa. Su.

Mo. 100 − − − − − − Tu. 93 100 − − − − − We. 97 97 100 − − − − Th. 87 97 93 100 − − − Fr. 95 97 97 97 100 − − Sa. 32 59 32 55 51 100 − Sun. 30 71 48 71 61 97 100

F

i

u

t

t

s

s

t

t

u

t

E

a

o

t

e

t

fi

S

z

isher’s exact test were accepted equal to 5%. In addition, taking

nto account the number of the days with untypical DHW heat

se, the values of the acceptable error (see Section 2.1 ) was es-

imated as 14%. Therefore, the days of the week in nursing homes

hat have statistically similar profiles in more than 86% of the con-

idered weeks were identified, see Table 3 . Based on this conclu-

ion, the following groups of the days were identified:

- The first group: 1) Monday, Tuesday, Wednesday, Thursday and

Friday,

- The second group: 2) Saturday and Sunday.

Detailed DHW heat use profiles organized by similar days of

he weeks and seasons are shown in Fig. 9 . For these profiles,

he time zones were identified based on average daily DHW heat

se by the method explained in Section 2.2 . Fig. 9 demonstrated

he time zones with a peak heat load (heat use above Emax, see

quation 6 ), minimum (heat use below Emin, see Equation 5 ) and

verage (heat use in the range between Emin and Emax) heat load

f DHW. The borders between time zones in Fig. 9 are shown in

he form of the horizontal lines.

The identification of the time intervals when minimum, av-

rage, and peak heat use occurred during the day was one of

he key information from the analysis of the DHW heat use pro-

les. Thereby, the application of the method presented in the

ection 2.2 allowed us to determine the following borders of time

ones:

1) The peak heat use of the DHW heat use occurred when the

heat use was higher than: 0.19 kWh/room for Monday-Friday

in the cold season, 0.168 kWh/room for Saturday-Sunday in the

cold season, 0.147 kWh/room for Monday-Friday in the hot sea-

son, and 0.137 kWh/room for Saturday-Sunday hot season;

2) The minimum heat use of the DHW heat use occurred when

the heat use was less than: 0.0 6 6 kWh/room for Monday-Friday

10 D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070

h

s

h

h

fi

i

a

b

F

w

s

I

n

p

u

T

b

s

2

2

u

h

t

m

u

D

D

i

s

t

i

e

3

t

h

c

p

u

c

t

fi

d

h

a

r

t

h

d

m

d

o

b

p

t

i

b

v

h

o

D

i

N

in the cold season, 0.065 kWh/room for Saturday-Sunday in the

cold season, 0.053 kWh/room for Monday-Friday in the hot sea-

son, and 0.052 kWh/room for Saturday-Sunday in the hot sea-

son;

3) The average heat use of the DHW heat use occurred when

it was between: 0.0 6 6 kWh/room and 0.19 kWh/room for

Monday-Friday in the cold season, between 0.065 kWh/room

and 0.168 kWh/room for Saturday-Sunday in the cold season,

between 0.053 kWh/room and 0.147 kWh/room for Monday-

Friday in the hot season, and between 0.052 kWh/room and

0.137 kWh/room for Saturday-Sunday in the hot season.

From Fig. 9 it can be observed that the hourly values of the

DHW heat use, as well as its peak, were much higher from Mon-

day to Friday compering to Saturday and Sunday. In general, DHW

heat use during the cold season was higher than in the warm sea-

son. Moreover, in the different seasons, there are some shifts in in-

tensity of the DHW heat use between the hours. From Monday to

Friday in the cold season, the peak of the DHW heat use occurred

from 9:00 to 15:00 o’clock, with the maximum heat use from 9:00

to 11:00 o’clock. Opposite, the evening peak in the cold season was

not clear and cannot be observed easily. Sunday and Saturday in

the cold season, the maximum of the DHW heat use was much

lower and may be noticed at 11:00 o’clock. Furthermore, the low

peak heat use appeared at 20:00 o’clock. Meanwhile, in the work-

ing days in the warm season, the peak of DHW heat use occurred

from 9:00 to 14:00 o’clock, with the maximum heat use at 10:00

o’clock and the values that are close to the maximum at 9:00 and

11:00 o’clock. In addition, two small peaks could be observed at

17:0 0 and 20:0 0 o’clock in the warm season. In the weekends in

the warm season, the peak was from 9:00 to 12:00 o’clock, and

at 14:00 and 20:00 o’clock. The minimum of the DHW in all the

profiles was at night, usually from 2:00 until 5:00 o’clock.

Changes of the DHW heat use intensity and the occurrence of

the peak values of the heat use in different profiles in Fig. 9 could

be explained by different work regimes in the nursing homes at

the weekends and the working days, as well as at different seasons.

In general, our study showed that dividing the DHW heat use pro-

files by seasons and days of the week was reasonable. The profiles

obtained in this way were more informative and allow us retrieve

additional information about DHW heat use in buildings.

4.3. Comparison of the standard profiles for DHW heat use with the

profiles obtained based on measurement data

In this section, two standards were compared with the profiles

obtained from the measurements and analysis in Section 4.2 in the

nursing homes. The Norwegian standard, “SN/TS 3031:2016: En-

ergy performance of buildings. Calculation of energy needs and en-

ergy supply” [43] is a national standard for calculations of build-

ings energy need and heat losses. Among different information,

this standard gives recommendation on DHW heat use profiles per

m 2 in nursing homes that should be used as an input for energy

demand calculation [43] . The standard “NS-EN 12831-3:2017: En-

ergy performance of buildings” [44] is European standard, which

is recommended for application in Norway. NS-EN 12831-3 pro-

vides reference profiles of DHW heat use per person in nursing

home. As mentioned earlier, in Norway, each room in the nursing

homes is occupied by only one person. Thus, heat use per room

is approximately equal to heat use per person. The profiles in the

both standards show DHW tap heat use without losses in the stor-

age tank and the system. Meanwhile, typically the measurements

in the nursing homes include losses in the storage tanks. For this

reason, to remove the losses from the profiles obtained by mea-

surements, the method proposed in [45] was used in this study.

This method is based on the assumption that the hourly DHW

eat use with minimum values represents system losses [43] . Con-

equently, extracting the minimum DHW heat use during these

ours from measured data gives us approximate value of the DHW

eat use without system and tank losses. Accordingly, using pro-

les in Fig. 9 , the hour with the minimum DHW heat use was

dentified. After that, the DHW heat use profiles were recalculated

ccording to the method in [45] . The DHW system losses obtained

y this method were approximately 20% of the total DHW heat use.

or the comparison, both profiles obtained by the measurements

ith adjustments according to the losses, and the profiles from the

tandards SN/TS 3031 and NS-EN 12831-3 are presented in Fig. 10 .

n addition, for a better understanding of the DHW heat use in the

ursing homes, the box plots of hourly DHW heat use per m 2 and

er room are presented in the Fig. 11 .

Fig. 10 indicates on the big difference between the DHW heat

se profiles obtained from the measurements and both standards.

he comparison with actual profiles showed the following draw-

acks of the standards: 1) standards are not taking into account

easonality and influence of the day of the week on DHW heat use,

) standards significantly overestimate average daily DHW heat use

) for certain hours the profiles in the standards overestimate or

nderestimate DHW heat use, 3) standards can not properly reflect

ours with peak and minimum DHW heat use.

The profile in the standard SN/TS 3031, see Fig. 10 . a), overes-

imated the daily DHW heat use in the nursing homes approxi-

ately 3.5 times. Even if we compare it with the maximum heat

se in the nursing homes, shown in the box plot, see Fig. 11 . a), the

HW heat use in the standard SN/TS 3031 was still much higher.

espite this fact, the standard making the assumption that there

s no DHW heat use from 1:00 to 5:00 o’clock. The actual profiles,

ee Fig. 10 , showed a small amount of DHW heat use even at night

ime.

Information about magnitude and timing of the peak heat use

n the buildings is crucial for solving a number of issues in en-

rgy planning. However, from Fig. 10 . a) we can see that SN/TS

031 is not representing this information in a proper way. From

he standard profile, we could assume that the morning peak of

eat use occurred from 7:00 to 8:00 o’clock, and the similar peak

ould be observed from 18:00 to 19:00 o’clock. Meanwhile, in the

rofile based on actual measurements, the maximum DHW heat

se was from 9:00 to 11:00 o’clock, and the evening peak was not

learly visible. The peak value in the standard is 3.7 times higher

han in the measured profile. These differences between the pro-

les were significant and they show the drawbacks of the stan-

ard SN/TS 3031. It should be noted that a sample of three nursing

omes is probably not enough to be sure that the measurements

re representative for the national average. However, this sample

epresented well the DHW heat use in nursing homes in the cen-

ral part of Eastern Norway.

The standard, NS-EN 12831-3, overestimated the daily DHW

eat use by 1.65 times, see Fig. 10 . b). Unlike SN/TS 3031, the stan-

ard NS-EN 12831-3 shows DHW heat use at night time, which

akes it more realistic. The values in the NS-EN 12831-3 stan-

ard are closer to the maximum than the average hourly values

f the DHW heat use in the nursing homes presented in Fig. 11 .

). From Fig. 10 . b) it may be noted that the timing of the actual

eaks of the DHW heat use did not match perfectly the informa-

ion in the standard NS-EN 12831-3. The morning peak of heat use

n the standard is shown from 7:00 to 8:00 o’clock. It is shifted

y two hours compared with the actual one, see Fig. 10 . b). The

alue of the maximum DHW heat use in the standard is 2.7 times

igher than in the profile based on measurement. The behaviour

f DHW in the evening time was similar to the measured profile.

espite the fact that NS-EN 12831-3 is the international standard,

t explains the DHW nursing home heat use much better than the

orwegian national standard SN/TS 3031.

D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070 11

Fig. 10. Hourly profiles of DHW heat use according to the standards and measurements in the nursing homes, where a) standard SN/TS 3031 b) NS-EN 12831-3.

Fig. 11. Hourly profile of DHW heat use obtained by measurements, where: a) DHW heat use per m 2 , b) DHW heat use per room.

fi

o

t

b

s

c

i

l

t

a

w

s

b

t

a

c

t

i

a

D

e

h

m

l

u

t

p

fi

f

i

b

m

w

m

h

5

W

t

t

A

There could be several reasons for the inaccuracy of the pro-

les in the standards. First, the majority of the standards are based

n information and data obtained decades ago [17] . The introduc-

ion of new types of DHW appliances, changes in routines and

ehaviours in the nursing homes are also likely changing the as-

umed values from the standards. Consequently, standards cannot

orrectly display the current state of the DHW heat use in build-

ngs, because the standards are developed to give limits and guide-

ines and cannot determine the real use. The other reason is that

he profiles given in the standards are usually too simplified to en-

ble their easier implementation by practitioners. These profiles

ere created for certain categories of buildings: nursing homes,

chool, hotel, offices, etc. However, even within one category of the

uildings, the DHW heat use can behave differently. The location of

he building in different parts of the country with specific temper-

ture conditions is also factor that could lead to uncertainty.

The above mentioned standard profiles are commonly used for

alculation of the building performance against national regula-

ions. If the standard profiles deviates significantly from the reality,

t may lead to unwanted effects. For example in Norway, there is

demand that above 60% of the energy demand for heating and

HW should be covered by a centralized system without fossil fu-

ls. In cases with highly insulated buildings, the standard DHW

eat demand may represent above 60% of the total heating de-

m

and. If the real DHW use is much lower than the standard calcu-

ation, the standard requirements on the system design will have

nwanted effects on choosing energy supply systems and sizing

he energy infrastructure.

Therefore, this study showed that dividing the DHW heat use

rofiles by season and days of the week is reasonable. These pro-

les should be based on accurate and up-to-date statistical data

rom real buildings and reliable methods of processing available

nformation. The potential for energy saving, can be achieved by

etter DHW system sizing, introducing of demand-side manage-

ent, and other energy saving measures. Representative profiles

ill form a basis for the proper implementation of energy saving

easures and increasing the efficiency of DHW heat use in nursing

omes.

. Conclusions

DHW system is a significant consumer of energy in buildings.

ith the introduction of highly insulated building structures and

echnologies of passive houses, the share of the DHW heat use in

he total energy balance of the buildings is continuously increasing.

ccordingly, reducing the DHW heat use in buildings becoming a

ore important target.

12 D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070

W

F

t

W

A

e

c

R

m

D

G

s

R

The review of the literature showed that there is a gap in

knowledge about actual DHW heat use in buildings. Specific heat

use in the nursing homes is one of the highest comparing to other

types of buildings. Therefore, analysis of the DHW heat use in

nursing homes is particularly relevant for Norway. To increase en-

ergy efficiency in the DHW systems in Norway, an extensive anal-

ysis should be carried in various types of buildings. One of the

most critical problems of such analysis is the development of up-

to-day profiles of the DHW heat use. These profiles should accu-

rately reflect DHW heaty use in the buildings and fill gaps in exist-

ing standards. In this article, the relevant problem was investigated

for nursing homes located in the Eastern Norway.

Analysis of the measurements in three nursing homes showed

a strong negative correlation between the monthly DHW heat use

and the outdoor temperature. Consequently, seasonality is an es-

sential factor that should be taken into account for DHW heat use

profiles for nursing homes. The other significant factor identified

in the article was the day of the week. For the DHW heat use

analysis, the statistical approach that allowed us to develop uni-

fied profiles divided by months and days of the week with simi-

lar behaviour of DHW heat use was suggested. Based on this ap-

proach the months of the year for the nursing homes were divided

into two groups: the cold season (January, February, March, April,

November, December) and the warm seasons (June, July, August,

September, October). Comparison of the profiles in different days

of the week showed that weekends and working days should be

considered separately. Furthermore, a method for determining the

time zones with the peak, the minimum, and the average heat use

in the daily profile of the DHW heat use was applied.

For the nursing homes, the profiles obtained by seasons showed

that the DHW heat use in the cold season was higher than in the

warm season. Besides, nursing homes used less heat for DHW in

the weekends than in the working days. The maximum DHW heat

use in nursing homes usually occurred from 9:00 o’clock to 11:00

o’clock, and minimum from 2:00 to 5:00 o’clock.

Finally, the DHW heat use profiles obtained from the measure-

ments in the nursing homes were compared with profiles from

national standard SN/TS 3031:2016 and international standard NS-

EN 12831-3:2017. The comparison showed that the European stan-

dard, NS-EN 12831-3, overestimated the daily DHW heat use by

1.65 times, and the Norwegian standard, SN/TS 3031, overestimated

it by 3.5 times. The magnitude and timing of the peak heat use

in the buildings was also different from the standards. The Euro-

pean standard explains much better the actual DHW heat use in

the nursing homes than the Norwegian standard. For practical ap-

plication and relevant decisions related to building energy supply

systems, preference should be given to profiles obtained on the ba-

sis of statistical data collected in real buildings.

The study in this work was limited to only three nursing homes.

For this reason, in the future work, the analysis in larger amount

of nursing homes and other types of buildings will be performed.

For a larger amount of buildings, the application of different clus-

tering methods for the analysis of DHW heat use will be tested. In

addition, the question of predicting the DHW heat usage profiles

will be considered in further studies.

Declaration of Competing Interest

The authors declare that they have no known competing finan-

cial interests or personal relationships that could have appeared to

influence the work reported in this paper.

CRediT authorship contribution statement

Dmytro Ivanko: Conceptualization, Methodology, Formal anal-

ysis, Software, Investigation, Writing - original draft, Visualization,

riting - review & editing. Harald Taxt Walnum: Data curation,

ormal analysis, Writing - review & editing. Natasa Nord: Concep-

ualization, Formal analysis, Writing - original draft, Supervision,

riting - review & editing.

cknowledgement

This article has been written within the research project "En-

rgy for domestic hot water in the Norwegian low emission so-

iety". The authors gratefully acknowledge the support from the

esearch Council of Norway (ENERGIX-programme), SINTEF Com-

unity, Department of Energy and Process Engineering at NTNU,

rammen Eiendom, Omsorgsbygg, Boligbygg, OBOS, Olav Thon

ruppen, Armaturjonsson, Høiax, Geberit, Uponor and FM Matts-

on.

eferences

[1] European Commission, Energy Performance of Buildings Directive,

Available from: https://ec.europa.eu/energy/en/topics/energy- efficiency/ energy- performance- of- buildings (Accessed October 16, 2019).

[2] M. Jamieson, O. Brajterman, Y. Verstraeten, J. Arbon, EPBD Compliance Study - Final Report, Publications Office of the European Union, (2015) pp. 138.

[3] European Commission, Energy Roadmap 2050, Available from: https://eur- lex.

europa.eu/legal- content/EN/TXT/PDF/?uri=CELEX:52011DC0885&from=EN (Ac- cessed October 16, 2019).

[4] E. Abel , P.-E. Nilsson , L. Ekberg , P. Fahlen , L. Jagemar , R. Clark , O. Fanger , K. Fitzner , L. Gunnarsen , P.V. Nielsen , Achieving the desired indoor climate-en-

ergy efficiency aspects of system design, Lund: Studentlitteratur: Commtech Group (2003) .

[5] E. Fuentes , L. Arce , J. Salom , A review of domestic hot water consumption pro- files for application in systems and buildings energy performance analysis, Re-

newable and Sustainable Energy Reviews 81 (2018) 1530–1547 .

[6] V. Foldvary , L. Borisova , D. Petras , Energy use and thermal comfort of two apartment buildings before and after refurbishment in Slovakia, REHVA Eu-

ropean HVAC Journal 50 (5) (2013) 52–56 . [7] T. Peters, J. Weyer, Architectural design for low energy housing: Experiences

from two recent affordable housing projects in Denmark, in: Proceedings of Sustainable cities and buildings: 7th Passive House Norden Conference, DTU

Byg and KADK Copenhagen, Denmark, 2015, pp. 1-10.

[8] J. Hirvonen , J. Jokisalo , J. Heljo , R. Kosonen , Towards the EU emissions targets of 2050: optimal energy renovation measures of Finnish apartment buildings,

International Journal of Sustainable Energy 38 (7) (2019) 649–672 . [9] S. Medved , S. Domjan , C. Arkar , Sustainable Technologies for Nearly Zero En-

ergy Buildings, Springer International Publishing (2019) . [10] S. Attia, P. Eleftheriou, F. Xeni, R. Morlot, C. Ménézo, V. Kostopoulos, M. Betsi,

I. Kalaitzoglou, L. Pagliano, M. Almeida, Overview of challenges of residential

nearly Zero Energy Buildings (nZEB) in Southern Europe, in, Sustainable Build- ings Design Lab, Technical Report, Liege, Belgium, 2016, pp. 87.

[11] B. Bøhm , Production and distribution of domestic hot water in selected Dan- ish apartment buildings and institutions. Analysis of consumption, energy ef-

ficiency and the significance for energy design requirements of buildings, En- ergy conversion and management 67 (2013) 152–159 .

[12] G. Cossali, M. Ratcliffe, E. Routledge, T. Karayiannis, J. Fielder, The cost of

Legionellosis and technical ways forward, in, Proceedings of CIBSE Technical Symposium. Liverpool John Moores University, Liverpool, UK, 2013, pp. 1-9.

[13] I. Knight , N. Kreutzer , M. Manning , M. Swinton , H. Ribberink , European and Canadian non-HVAC electric and DHW load profiles for use in simulating the

performance of residential cogeneration systems, International Energy Agency, Annex 42 (2007) .

[14] S. Tyler , L. Schipper , The dynamics of electricity use in Scandinavian house-

holds, Energy 15 (10) (1990) 841–863 . [15] K. Ahmed , P. Pylsy , J. Kurnitski , Monthly domestic hot water profiles for energy

calculation in Finnish apartment buildings, Energy and Buildings 97 (2015) 77–85 .

[16] G. Levermore , W. Chong , Performance lines and energy signatures: review and analysis, Building Services Engineering Research and Technology 10 (3) (1989)

105–114 .

[17] T.-A . Koiv , A . Mikola , A . Toode , DHW design flow rates and consumption pro- files in educational, office buildings and shopping centres, Smart Grid and Re-

newable Energy 4 (03) (2013) 287–296 . [18] G. Curculacos, G. Turchetto, U. Mazzali, F. Peron, P. Romagnoni, F. Cappelletti,

F. Bauman, M. Scarpa, On the Energy Performance Design of a Skilled Nursing Facility Building, in: Proceedings of 3rd International High Performance Build-

ings Conference at Purdue, 2014, pp. 1-10. [19] J. Tindall, J. Pendle, Are we significantly oversizing domestic water systems?,

in: Proceedings of CIBSE Technical Symposium, London, UK, 2015, pp. 1-23.

[20] M. Kouveletsou , N. Sakkas , S. Garvin , M. Batic , D. Reccardo , R. Sterling , Simulat- ing energy use and energy pricing in buildings: The case of electricity, Energy

and Buildings 54 (2012) 96–104 . [21] A. Reynaud , M. Pons , C. Pesado , Household Water Demand in Andorra: Impact

of Individual Metering and Seasonality, Water 10 (3) (2018) 321 .

D. Ivanko, H.T. Walnum and N. Nord / Energy & Buildings 222 (2020) 110070 13

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

22] X. Yang, S. Svendsen, H. Li, Supply of domestic hot Water at comfortable tem- peratures by low-temperature district heating without risk of Legionella, in,

Technical University of Denmark, Department of Civil Engineering. DTU Civil Engineering Report., 2016, pp. 140.

23] H. Ahmad, Energy Efficiency Solutions for Domestic Hot Water, Online pub- lisher: bigee.net, (2017) p. 1-5.

[24] L. Tronchin , K. Fabbri , Energy performance building evaluation in Mediter- ranean countries: Comparison between software simulations and operating

rating simulation, Energy and buildings 40 (7) (2008) 1176–1187 .

25] N. Djuric , V. Novakovic , Identifying important variables of energy use in low energy office building by using multivariate analysis, Energy and Buildings 45

(2012) 91–98 . 26] K. Ahmed , P. Pylsy , J. Kurnitski , Hourly consumption profiles of domestic hot

water for different occupant groups in dwellings, Solar Energy 137 (2016) 516–530 .

[27] I. Verhaert, B. Bleys, S. Binnemans, E. Janssen, A methodology to design do-

mestic hot water production systems based on tap patterns, in: Proceedings of CLIMA 2016: 12th REHVA World Congress, Aalborg, Denmark, 2016, pp. 1-10.

28] E. Wilson, C. Engebrecht-Metzger, S. Horowitz, R. Hendron, 2014 Building America house simulation protocols, in, National Renewable Energy Laboratory,

Technical Report, Golden, United States, 2014, pp. 91. 29] F. Lombardi , G. Silvagni , P. Sirini , R. Spagnuolo , F. Volpi , Daily water demand,

An Interdisciplinary Journal of Applied Science 13 (5) (2018) 1–11 .

30] N. Good , L. Zhang , A. Navarro-Espinosa , P. Mancarella , High resolution mod- elling of multi-energy domestic demand profiles, Applied Energy 137 (2015)

193–210 . [31] D. Fischer , T. Wolf , J. Scherer , B. Wille-Haussmann , A stochastic bottom-up

model for space heating and domestic hot water load profiles for German households, Energy and Buildings 124 (2016) 120–128 .

32] J. Widen , M. Lundh , I. Vassileva , E. Dahlquist , K. Ellegard , E. Wackelgard , Con-

structing load profiles for household electricity and hot water from time-use data - Modelling approach and validation, Energy and Buildings 41 (7) (2009)

753–768 . [33] L.G. Swan , V.I. Ugursal , I. Beausoleil-Morrison , Occupant related household en-

ergy consumption in Canada: Estimation using a bottom-up neural-network technique, Energy and Buildings 43 (2011) 326–337 .

34] E. Pieterse-Quirijns , A. Van Loon , H. Beverloo , E. Blokker , E. Van der Blom , J. Vreeburg , Validation of non-residential cold and hot water demand model

assumptions, Procedia Engineering 70 (2014) 1334–1343 . [35] I. Pieterse-Quirijns, Manual SIMDEUM Pattern Generator. Tool for water de-

mand and discharge patterns for residential and non-residential buildings, in: KWR Water Research Institute, The Netherlands, 2014, pp. 99.

36] C. Agudelo-Vera , E. Pieterse-Quirijns , W. Scheffer , E. Blokker , New method to design domestic water systems, REHVA European HVAC Journal 6 (2013)

12–16 .

[37] J. Wiehagen, J. Sikora, Domestic hot water system modeling for the design of energy efficient systems, NAHB Research Center, Upper Marlboro, Technical re-

port, (2002) 67. 38] E. Vine , R. Diamond , R. Szydlowski , Domestic hot water consumption in four

low-income apartment buildings, Energy 12 (6) (1987) 459–467 . 39] H. Kauko , O. Stavset , M. Bantle , N. Nord , Energy use in Norwegian non-residen-

tial buildings: building regulations, calculations and measurements, in: Pro-

ceedings of 8th International Conference on Heating, Ventilation and Air-Con- ditioning: Cold Climate HVAC (2015) 1–8 .

40] V.F. Nakhodov , A.I. Zamulko , M. Al Sharari , D.O. Medintseva , Analysis of dura- tion and border of existing tariff zones, Power engineering: Economics, Tech-

nology, Ecology 2 (2016) 88–97 . [41] M.L. Rizzo , G.J. Szekely , Energy distance, wiley interdisciplinary reviews, Com-

putational statistics 8 (1) (2016) 27–38 .

42] O. Gerin , B. Bleys , K. De Cuyper , Seasonal variation of hot and cold water con- sumption in apartment buildings, in: Proceedings of International Symposium

of CIB W062 Water Supply and Drainage for Buildings, Brazil (2014) 1–9 . 43] SN/TS 3031:2016, Energy performance of buildings. Calculation of energy

needs and energy supply, Standard Norge (2016) 168 . 44] NS-EN 12831-3:2017 Energy performance of buildings - Method for calculation

of the design heat load - Part 3: Domestic hot water systems heat load and

characterisation of needs, Module M8-2, M8-3, in, 2017, pp. 64. 45] A.J. Marszal , C. Zhang , M.Z. Pomianowski , P.K. Heiselberg , K. Gram-Hanssen ,

A.R. Hansen , Simple methodology to estimate the mean hourly and the daily profiles of domestic hot water demand from hourly total heating readings, En-

ergy and Buildings 184 (2019) 53–64 .