Potential of electrical building heating as thermal energy storage in Sweden

Potential of electrical building heating as thermal energy storage in Sweden

PONTUS DAHLSTR ¨ OM

Master’s Degree Project

Stockholm, Sweden June 2019

Potential of electrical building heating as thermal energy storage in Sweden

Pontus Dahlstr¨ om

Master Thesis

MJ241X Degree Project in Energy Technology Academic Supervisor EGI: Bj¨ orn Palm

Academic Supervisors EPE: Lars Herre and Lennart S¨ oder Industrial Contact SWECO: Johan Bruce

Examiner: Bj¨ orn Palm Date: July 2019

TRITA-ITM-EX 2019:525

Abstract

The purpose of this Master thesis is to investigate the potential of using electricity based

building heating as thermal energy storage in Sweden and its applications. Building data

and statistics along with literature were the basis for data collection and processing. The

work was then carried out by selecting a thermal energy storage model to represent dif-

ferent building types that are equipped with electricity based heating systems. This

aggregate thermal energy storage model was applied to the Swedish building stock, his-

torical weather data and typical thermal comfort zones. The power and energy capacity

of the thermal energy storage were studied and the model was used to evaluate Demand

Response (DR) both as Price Based DR and Emergency DR. This thesis gives an approx-

imation of the potential of both power and energy capacity which has not been clearly

quantified in previous studies for thermal energy storage in buildings of Sweden. The

thesis was carried out for the Department of Energy Technology (EGI) at the Division

of Electric Power and Energy Systems (EPE) in collaboration with SWECO as part of

the North European Energy Perspectives Project (NEPP).

Sammanfattning

Syftet med detta examensarbete p˚ a Masterniv˚ a ¨ ar att utreda potentialen av att anv¨ anda

v¨ armesystem drivna av elektricitet i svenska byggnader som lagring av termisk energi

och m¨ ojliga till¨ ampningar. Datainsamlingen och behandlingen baserades p˚ a byggnads-

data och statistik tillsammans med l¨ amplig litteratur. Arbetet utf¨ ordes sedan genom

att v¨ alja en modell f¨ or lagring av termisk energi f¨ or att representera olika byggnadstyper

som har v¨ armesystem drivna av elektricitet installerade. Den samlade termiska energila-

gringsmodellen till¨ ampades f¨ or det svenska byggnadsbest˚ andet med historiska v¨ aderdata

och termiska komfortzoner. Effekten och energikapaciteten fr˚ an den termiska energila-

gringen studerades och modellen anv¨ andes sedan f¨ or att utv¨ ardera Demand Response

(DR) b˚ ade baserat p˚ a elpris och vid n¨ odfall. Examensarbetet uppskattar potentialen av

b˚ ade effekten och energikapaciteten vilket i tidigare studier av lagring av termisk energi

i svenska byggnader ej kvantifierats tydligt. Detta examensarbete har utf¨ orts f¨ or insti-

tutionen f¨ or energiteknik (EGI) vid institutionen f¨ or elkraftteknik (EPE) i samarbete

med SWECO inom ramen f¨ or North European Energy Perspectives Project (NEPP).

Acknowledgements

The author would like to thank several people who have contributed to the thesis: Bj¨ orn

Palm has provided thoughtful comments regarding the work and been supportive in the

master thesis process. Lars Herre has assisted in constructing and implementing the

model and has throughout been helpful in the work and writing of the thesis. Lennart

S¨ oder has provided insightful ideas for the direction of the project and connected the

work to Emil Nyholm at Chalmers and Johan Bruce at SWECO. Egill T´ omasson and

Alessandro Crosara has provided data from the Flex4RES project. Others at KTH have

also provided their perspectives on the master thesis.

Contents

1 Introduction 1

2 Data 3

2.1 Data Collection . . . . 3

2.1.1 Building Properties of SFDs . . . . 3

2.1.2 Share of SFDs . . . . 4

2.1.3 Heating Type Data . . . . 5

2.2 Data Processing . . . . 5

2.2.1 Assumptions . . . . 5

2.2.2 Processed Data . . . . 8

3 Methodology 10 3.1 Individual building Thermal Model . . . . 10

3.2 Aggregate Thermal Energy Storage Model . . . . 11

3.3 Price Based Demand Response . . . . 12

3.4 Emergency Demand Response . . . . 13

4 Results 16 4.1 Thermal Energy Storage . . . . 16

4.2 Price Based Demand Response . . . . 24

4.3 Emergency Demand Response . . . . 24

5 Discussion 27

6 Conclusion 31

7 Appendix 32

List of Figures

2.1 Nordpool pricing areas in Sweden . . . . 7

3.1 Temperature profile 060219 . . . . 13

3.2 Electricity price 060219 . . . . 14

4.1 Aggregate Baseline Power versus outdoor Temperature . . . . 16

4.2 Aggregate heating power with baseline power as reference plane versus time until empty thermal battery charge . . . . 17

4.3 Aggregate heating power with baseline power as reference plane versus time until empty thermal battery charge . . . . 18

4.4 Aggregate heating power with baseline power as reference plane versus time until empty thermal battery charge . . . . 18

4.5 Aggregate heating power with baseline power as reference plane versus time until empty thermal battery charge . . . . 19

4.6 Maximum power reduction versus outdoor temperature . . . . 19

4.7 Aggregate heating power with baseline power as reference plane versus time until normal thermal battery charge . . . . 20

4.8 Aggregate heating power with baseline power as reference plane versus time until normal thermal battery charge . . . . 21

4.9 Aggregate heating power with baseline power as reference plane versus time until normal thermal battery charge . . . . 21

4.10 Aggregate heating power with baseline power as reference plane versus time until normal thermal battery charge . . . . 22

4.11 Maximum power increase versus outdoor temperature . . . . 22

4.12 Aggregate heating power consumption on February 6th 2019 . . . . 25

4.13 Energy of thermal battery during DR . . . . 25

7.1 Nordpool pricing areas in Sweden . . . . 33

7.2 Type nr 1 . . . . 34

7.3 Type nr 2 . . . . 34

7.4 Type nr 3 . . . . 35

7.5 Type nr 4 . . . . 35

7.6 Type nr 5 . . . . 36

7.7 Type nr 6 . . . . 36

7.8 Type nr 7 . . . . 37

7.9 Type nr 8 . . . . 37

7.10 Type nr 9A . . . . 38

7.11 Type nr 9B . . . . 38

7.12 Type nr 10 . . . . 39

7.13 Type nr 11 . . . . 39

7.14 Type nr 12 . . . . 40

7.15 Type nr 13 . . . . 40

List of Tables

2.1 V¨ armek excel input data . . . . 4

2.2 Percentage of house types per decade . . . . 5

2.3 Percentage of electricity based heating per decade . . . . 6

2.4 Pricing areas for Sweden by L¨ an . . . . 7

2.5 Number of SFDs in each price area . . . . 8

2.6 SFDs with heating type per price area . . . . 8

3.1 Time constants for different SFD types . . . . 11

3.2 Generation Adequacy Data . . . . 14

3.3 Comparison of study input parameters . . . . 15

4.1 Maximum Power decrease and time to empty thermal battery charge from normal cycle operation . . . . 20

4.2 Maximum Power Increase and time until normal thermal battery charge . 23 4.3 Maximum Power Decrease and time from full to empty thermal battery charge . . . . 23

4.4 Maximum Power Increase and time from empty to full thermal battery charge . . . . 23

4.5 Sensitivity analysis results . . . . 24

4.6 Emergency Demand Response . . . . 26

7.1 Temperature profiles 6th February 2019 . . . . 32

Nomenclature

∆P Power difference [W ]

∆T Temperature Difference [K]

∆t Time step [s]

Q ˙ T otal Total energy flow [W ] Q ˙ _vent Ventilation energy loss [W ] η Efficiency [.]

η _elm Efficiency of electric motor [.]

λ Thermal Conductivity [W/(mK)]

ρ Density [kg/m ³ ]

τ Thermal Time Constant [h]

A Area [m ² ]

ACH Air changes per hour [.]

c _p Specific Heating Capacity [J/kg]

C _th Thermal Capacitance [J/K]

COP Coefficient of performance [.]

D Duty cycle [.]

d Thickness [m]

E Operating energy [J ]

E ^t Battery state of charge [J ]

E _max Maximum energy [J ]

E _min Minimum energy [J ]

P _b ^t Baseline power consumption [W ] P _el Electrical Power [W ]

P _el ^t Mean power consumption [W ] P _max Maximum power [W ]

P _min Minimum power [W ]

P _th Thermal heating power [W ] Q ₁ Heating energy output [J ] R _th Thermal Resistance [K/W ]

SCOP Seasonal coefficient of performance [.]

T Temperature [C]

t Time [s]

T _h Temperature upper bound [C]

T _i Indoor temperature [C]

T l Temperature lower bound [C]

T _o Outdoor temperature [C]

t _{OF F} Time heating is off [s]

t _ON Time heating is on [s]

V Volume [m ³ ] x Heating variable [.]

Q _house Energy content of SFD [J/m ² ]

Q independent Independent energy content [J ]

U _total Total U-value [W/(Km ² )]

Chapter 1 Introduction

This master thesis work will investigate the potential for electricity based building heat- ing as thermal energy storage for single family dwellings (SFDs) in Sweden. The thesis is carried out for the Department of Energy Technology (EGI) at the Division of Electric Power and Energy Systems (EPE) in collaboration with SWECO as part of the North European Energy Perspectives Project (NEPP). The thesis work has also collaborated with research projects in Flex4RES.

The electricity based heating system will be represented by electrically driven heat pumps or in cases where this is not available electric radiators. The study is limited to SFDs since multi-family dwellings (MFDs) to a larger degree are connected to district heating and therefore has a low extent of electric heating systems and is not appropriate for a study of this kind [1]. The study is based on the house size definitions of the Swedish statistical agency where the term SFDs is used to represent the sm˚ ahus house definition in the report [2].

The work will be carried out by simulating the thermal behavior of buildings of dif- ferent types and from different decades and estimating their heating energy use. Knowl- edge of this is important for estimating the use of SFDs as demand response (DR) in the electricity system, by altering the times for heating based on the system load profile.

Investigations of this type could in effect contribute to ecological sustainability by mak- ing the electricity system more fit to meet the varied output of renewable energy sources such as wind power. Using weather data and thermal comfort zones the constraints of energy use for the system can be estimated and to ensure social sustainability it is important that the implementation does not compromise factors such as human health and comfort. Knowledge of the buildings thermal behavior will allow flexibility of when to heat the buildings for optimal performance and the cost of electricity delivered for heating can be evaluated in order to meet economical sustainability.

For the literature study other works concerning the heating and cooling of buildings

were considered. A major part focused on the simulation of the 14 Swedish building types

and their thermal response [3] and furthermore understanding thermal models of this

type using heat transfer literature [4]. Studies made of the response of air conditioning

systems in buildings were analyzed since the thermal modeling is similar when comparing

heating and cooling systems in buildings [5]. Energy statistics regarding electrically

heated SFDs were considered for the year 2016 [6].

Previous studies have investigated the demand response potential of SFDs in Sweden for years 2010 and 2012, by allowing the temperature to increase above the normal levels for intervals of time [1]. This work intends to extend this concept by studying the potential of electric heating in SFDs were indoor temperatures are allowed to alter both above and below preferred indoor temperature within the constraints for human well being. Further studies of Swedish SFDs have modeled each building in one or two thermal zones, with the conclusion that further resolution in the model increases the calculation requirements and time [7]. For this work a single thermal zone or node has been deemed sufficient based on the model used for the building stock and scope for the work in a Master thesis.

The thesis could contribute to viability studies for the use of SFD electric heating

for demand response purposes by providing data and figures of the specific power and

energy that could possibly be achieved for the Swedish system which has not clearly been

presented in the literature previously. Studying the possibility of emergency demand

response of the electric system could for example benefit from this kind of data allowing

estimations to be made about what potential the electricity based heating could have as

means for demand response. Also since the stock of electrically heated SFDs is changing

each year a study using recent statistics will estimate the power available in the present.

Chapter 2 Data

2.1 Data Collection

2.1.1 Building Properties of SFDs

The thesis work started from the thermal behavior of the buildings in Sweden. This behavior was previously studied in a report for Svenska V¨ armeverkens Ekonomiska F¨ orening (V¨ armek). The calculation for the SFDs is hence based on a excel file from the V¨ armek work which was constructed to investigate heating of buildings in Sweden [3].

In the file, 14 different house types of different materials from different decades are sim- ulated and the goal is to match the SFD building stock of Sweden to these house types.

The calculation defines different parameters for the materials in the house such as density ρ, specific heating capacity c _p and thermal conductivity λ and then uses this as a basis for calculating the energy content in the house. The energy content in the house is based on a set temperature difference ∆T between the initial indoor and outdoor temperatures.

The analysis is based on calculating the total U-value U total and then the energy flow Q ˙ _{T otal} in the building in equation (2.1) where A is area and d thickness.

Q ˙ _{T otal} = U _total A∆T = 1 P _d

λ

A∆T (2.1)

Energy content of the house Q _house itself is also considered for the walls and roof of the building and is calculated using an equation of the general form as (2.2) based on the starting temperature difference.

Q _house = c _p dρ∆T (2.2)

When summing all the energy content, two cases are made: with or without the

slab foundation energy content. This is due to that it is uncertain to what degree the

foundation influences the total energy content and losses of the SFDs. The energy losses

occur through leakage in the envelope and partly through the ventilation system as ˙ Q _vent

calculated in equation (2.3) with volume V [8].

Q ˙ _vent = V ρc _p ∆T ACH

3600 (2.3)

The calculation includes different values for the air changes per hour ACH of SFDs but due to the health constraints of Folkh¨ alsomyndigheten stating that at least an ACH = 0.5 is recommended only one value was deemed appropriate for the purposes of this study [9]. Certain solar gains are also considered in the excel file in the form of a yearly solar constant value of radiation transmitted through the building windows and also roof windows for some SFDs [10].

The different types of houses are based on the following input parameters in Table 2.1.

Pictures of the house types are found in the appendix: Type nr 1 Fig. 7.2, Type nr 2 Fig. 7.3, Type nr 3 Fig. 7.4, Type nr 4 Fig. 7.5, Type nr 5 Fig. 7.6, Type nr 6 Fig. 7.7, Type nr 7 Fig. 7.8, Type nr 8 Fig. 7.9, Type nr 9A Fig. 7.10, Type nr 9B Fig. 7.11, Type nr 10 Fig. 7.12, Type nr 11 Fig. 7.13, Type nr 12 Fig. 7.14 and Type nr 13 Fig. 7.15.

Table 2.1: V¨ armek excel input data

Nr Building Year Type Foundation Ventilation Envelope

11 1880 Fri 2 Crawl S Lumber

4 1935 Fri 1.5 Basement & Crawl S Wood

2 1950 Fri 1.5 Basement S Wood

5 1951 Fri 2 Basement S Concrete

12 1962 Rad 1 Slab F Wood

3 1972 Fri 1.5 Basement S Wood

13 1970 Rad 2 Slab F Wood

6 1975 Fri 1.5 Slab F Wood

1 1981 Fri 1.5 Slab F Wood

7 1988 Fri 1 Crawl S Wood

9A 2005 Rad 3 Slab F-VP Wood

9B 2005 Rad 3 Slab F-VP Steel-Gypsum

8 2006 Fri 2 Slab FTX Wood

10 2007 Fri 2 Slab FTX Wood

2.1.2 Share of SFDs

The next step is to attempt to match these house types to the SFDs present in Sweden.

Statistiska Centralbyr˚ an (SCB) uses the term sm˚ ahus to describe a number of different buildings all with small size such as separate single family dwellings and townhouses.

Also buildings with two families are included in this data but this was not deemed to be

MFDs for the purposes of this study [2]. Hence all the buildings in Sweden of this size is

represented in the thesis and referred to by the general term SFDs although this included

a number of ”double family dwellings” as well still however of appropriate size for the

analysis. Since statistics were available by building year, the starting point was to assume shares for each house type per decade. The shares were based on the general assumptions that older SFD building types are not continued into the next decade if a new type is present and a constant building rate for each year. Furthermore data regarding the building materials, type of foundation and ventilation used in SFDs for different time periods were available and could be used as basis for further assumptions [11]. For the most modern house types the shares were based on comparing the house types to those present in studies from Chalmers university utilizing the BETSI (Byggnaders energianv¨ andning, tekniska status och inomhusmilj¨ o) database and adjusting to match these ratios [7]. The acronym BETSI is used in the report to refer to this study along with the paper [1].

In Table 2.2 shares of house type per decade are presented.

Table 2.2: Percentage of house types per decade

Type Nr 1 Nr 2 Nr 3 Nr 4 Nr 5 Nr 6 Nr 7 Nr 8 Nr 9A Nr 9B Nr 10 Nr 11 Nr 12 Nr 13

-1930 0 0 0 0 0 0 0 0 0 0 0 100 0 0

31-40 0 0 0 50 0 0 0 0 0 0 0 50 0 0

41-50 0 10 0 90 0 0 0 0 0 0 0 0 0 0

51-60 0 95 0 0 5 0 0 0 0 0 0 0 0 0

61-70 0 80 0 0 12 0 0 0 0 0 0 0 8 0

71-80 0 55.5 0 0 22.25 0 0 0 0 0 0 0 22.25

81-90 72.2 0 0 0 0 0 27.8 0 0 0 0 0 0 0

91-00 14.8 0 0 0 0 85.2 0 0 0 0 0 0 0

01-10 10.2 0 0 0 0 0 14.8 19.5 33.9 2.1 19.5 0 0 0

11- 0 0 0 0 0 0 0 32.5 33 2 32.5 0 0 0

2.1.3 Heating Type Data

Next the electicity based heating systems in the SFDs needed to be investigated. Once more, data was available for different time periods of house construction for systems of direct and water based electrical heating and ground source heat pumps. For other types of heat pumps however only totals were available, hence an assumption of equal distribution of these across the SFDs of different time periods was required to combine the two data sets [6].

2.2 Data Processing

2.2.1 Assumptions

The gathered data required some additional processing before it could be used in the

model. The heating system performance and abundance was presented from the different

sources. When combining two data sets for electricity based heating system amounts the

shares were normalized so that the share of non electric heating in the system would be

kept at the same numbers to not overestimate the presence of electric heating systems

in Sweden. Hence assumptions for the average share of electric heating systems in SFDs of different time periods could be produced and are shown in Table 2.3 where P _el is the electrical power.

Table 2.3: Percentage of electricity based heating per decade

Heating A-AHP A-W HP EX HP GSHP Direct EL Water EL

SCOP 2.0/3.1 2.3 2.4 3.4 η = 97% η = 95%

P _el [kW] 6.5 8.6 5.3 8.2 10 10

-1930 10.1 3.3 2.4 19.6 7.2 6.1

1931-1940 10.1 3.3 2.4 19.6 7.2 6.1

1941-1950 14.8 4.9 3.5 16.6 2.7 7.1

1951-1960 14.8 4.9 3.5 16.6 2.7 7.1

1961-1970 10.3 3.4 2.4 17.4 7.2 13.3

1971-1980 7.7 2.5 1.8 14.7 28.2 6.7

1981-1990 13.0 4.3 3.1 4.1 8.1 20.1

1991-2000 19.1 6.3 4.5 5.7 5.7 14.7

2001-2010 17.9 5.9 4.2 10.7 2.5 15.7

2011- 29.1 9.6 6.8 8.0 0 7.3

The performance of the electric heating systems were assumed based on different sources. The direct electric heating system works directly by converting electricity in coil radiators to heat and is assumed to have a very good efficiency of η = 97%. The water-based electric heating system has an electric boiler pumping water to radiators in the house and the efficiency is assumed to be slightly lower than the direct system at 95% [12]. The electrical power was assumed based on performance values for water based electric heating systems present on the consumer market.

For the heat pumps, performance was measured using the SCOP (seasonal coefficient of performance) which is the yearly heat output divided by the operating energy. This can be compared to the normal COP (coefficient of performance) for heat pumps COP ₁ = Q ₁ /E meaning heating energy output over operating energy [13]. Conservative averages for these values were taken from Swedish testing data for Air-water (A-A HP), Exhaust air (EX HP) and Ground source systems (GSHP) with two performance values for cases of Air-air heat pumps (A-A HP) in northern (SE1 and SE2) and southern (SE3 and SE4) Sweden. The average electrical power for heat pumps was also available but in the case of air-air heat pumps only the compressor power average was presented [14]. Therefore in order to calculate the electrical power, the electric motor efficiency was assumed to be η _elm = 0.85 for the air-air heat pumps [15].

The statistics for amount of SFDs in Sweden were available by building year and

area such as L¨ an or kommun. In order to find the amount of houses in each pricing area

a division of the areas into the pricing areas based on a map was first done [16]. If a

considerable amount of the L¨ an was outside a pricing area it was further divided into

parts using the borders of the kommun. The division is presented in Table 2.4. A map

of the pricing area can be seen in Fig. 2.1 and enlarged in the appedix in Fig. 7.1.

Figure 2.1: Nordpool pricing areas in Sweden

Table 2.4: Pricing areas for Sweden by L¨ an

SE1 SE2 SE3 SE4

Norrbotten J¨ amtland Dalarna Sk˚ ane

V¨ asterbotten (part) V¨ asterbotten (part) V¨ astmanland Blekinge V¨ asternorrland V¨ armland Kronoberg G¨ avleborg Orebro ¨ J¨ onk¨ oping (part)

V¨ astra G¨ otaland Kalmar (part) Gotland Halland (part) Stockholm

Uppsala

S¨ odermanland

Osterg¨ ¨ otland

J¨ onk¨ oping (part)

Kalmar (part)

Halland (part)

The division of L¨ an could then be used to sum the statistical data for SFDs into the pricing areas in Table 2.5 [2].

Table 2.5: Number of SFDs in each price area

Time period SE1 SE2 SE3 SE4

-1930 10779 50271 241063 110795 1931-1940 6328 18924 84077 31434 1941-1950 6760 16440 83893 30376 1951-1960 10833 22876 96258 34097 1961-1970 13146 24968 179936 71370 1971-1980 20905 43322 260457 102960 1981-1990 9830 20746 131648 51436 1991-2000 3745 6451 65473 22854 2001-2010 1942 7198 75250 29138

2011- 1017 3327 30495 10448

2.2.2 Processed Data

Now with knowledge of the share of house types and electricity based heating for each decade, assuming the same distribution of heating systems across the country, the amount of SFDs of a certain house type and with a specific heating system could be calculated per decade. This was then summed into Table 2.6.

Table 2.6: SFDs with heating type per price area

Pricing Area Type Nr 1 Nr 2 Nr 3 Nr 4 Nr 5 Nr 6 Nr 7 Nr 8 Nr 9A Nr 9B Nr 10 Nr 11 Nr 12 Nr 13

SE1 A-A HP 1067 2702 894 1218 242 358 1017 164 216 13 164 1404 108 358

A-W HP 353 894 296 403 80 118 336 54 71 4 54 464 36 118

EX HP 249 632 209 285 57 84 238 38 50 3 38 328 25 84

GSHP 343 3655 1704 1631 365 683 323 55 85 5 55 2735 183 683

Direct El 613 1056 3268 392 129 1310 412 9 16 1 9 1006 76 1310

Water El 1535 2181 781 626 249 313 1062 83 128 8 83 856 140 313

SE2 A-A HP 2267 5508 1852 3141 477 742 1992 566 757 46 566 6016 205 742

A-W HP 750 1822 612 1039 158 246 659 187 250 15 187 1990 68 246

EX HP 530 1288 433 734 111 174 466 132 177 11 132 1407 48 174

GSHP 745 7365 3531 4314 712 1416 661 198 310 19 198 11718 348 1416

Direct El 1289 2074 6772 1080 248 2715 811 35 61 4 35 4308 145 2715

Water El 3258 4322 1619 1631 481 649 2131 299 463 28 299 3668 266 649

SE3 A-A HP 15619 29543 11132 15401 2928 4463 17409 5510 7501 455 5510 28512 1478 4463

A-W HP 5166 9771 3682 5093 969 1476 5758 1822 2481 150 1822 9430 489 1476

EX HP 3652 6908 2603 3601 685 1044 4070 1288 1754 106 1288 6666 346 1044

GSHP 5257 41676 21230 20789 4563 8511 5846 2005 3174 192 2005 55537 2509 8511

Direct El 8466 13104 40714 5063 1692 16322 6454 367 639 39 367 20420 1042 16322

Water El 21682 26281 9734 7937 3221 3903 17279 3023 4739 287 3023 17383 1920 3903

SE4 A-A HP 6021 11102 4401 5626 1131 1764 6352 2005 2774 168 2005 12741 586 1764

A-W HP 1991 3672 1455 1861 374 583 2101 663 917 56 663 4214 194 583

EX HP 1408 2596 1029 1316 265 413 1485 469 649 39 469 2979 137 413

GSHP 2028 15838 8392 7625 1776 3365 2149 757 1209 73 757 24818 995 3365

Direct El 3284 5085 16094 1869 666 6452 2387 142 247 15 142 9125 413 6452

Water El 8408 10128 3848 2904 1263 1543 6404 1139 1803 109 1139 7768 762 1543

Next this can be used to simulate for each individual house type and heating combi-

nation. An important constraint is the indoor temperature which outside of the range

18 − 24 ^◦ C according to Swedish authorities is disturbing to human health [17].

Chapter 3

Methodology

3.1 Individual building Thermal Model

The methodology is founded on the data and calculations from the V¨ armek excel file.

This calculation is based on ∆T = 20K and the excel file summarizes each buildings stored thermal energy in the envelope and contents as well as the thermal energy losses.

The thermal time constant τ for the house is calculated for the different house types from the stored thermal energy over the thermal energy losses.

The time constant is then the basis for solving an equation (3.1) for temperature T using starting indoor and outdoor temperatures T _i , T _o and time t.

T = T _o + (T _i − T _o )e ^−t/τ (3.1)

Using the electrical analogy, we have the time constant τ = R th C th based on the thermal resistance R _th and thermal capacitance C _th [4].

Adding the heating system to the equation we get (3.2) with the variable x set to 1 or 0 depending on if the heating is running or not. The house is heated based on the thermal heating power and thermal resistance. The thermal heating power P _th is related to the electrical power by P _el = P _th /SCOP for the heat pumps and P _el = P _th /η for the electricity based heating systems [18].

T = T _o + (T _i − T _o − xR _th P _th )e ^−t/τ + xR _th P _th (3.2) From the V¨ armek data the time constants are calculated. Then using only the house energy content, which is dependent on ambient temperature, subtracting from the total the independent content Q independent such as constant temperature hot water one can calculate the thermal capacitance in (3.3) and in turn the thermal resistance from the thermal time constant of the house type.

C _th = Q _total − Q independent

∆T (3.3)

In Table 3.1 we have the time constants and R _th , C _th values for the appropriate ven-

tilation flow of ACH = 0.75 without slab foundation for the different building types.

The simulation with the slab foundation included did not produce results for all building types and therefore for consistency across the building types only the constants without slab are presented here and used in the thesis.

Table 3.1: Time constants for different SFD types House Type

τ [hours] C _th [kWh/K] R _th [K/kW]

Nr 11 Lumber 1880 36.72 7.89 4.66

Nr 4 Wood 1935 41.04 5.93 6.92

Nr 2 Wood 1950 45.12 4.45 10.14

Nr 5 Concrete 1951 144.72 30.36 4.77

Nr 12 Wood 1962 38.4 3.89 9.87

Nr 3 Wood 1972 49.68 6.41 7.76

Nr13 Wood 1970 66.48 6.95 9.56

Nr 6 Wood 1975 30.96 4.39 7.05

Nr 1 Wood 1981 29.04 4.66 6.23

Nr 7 Wood 1988 43.68 4.53 9.64

Nr 9A Wood 2005 101.76 13.67 7.44

Nr 9B Steel-Gypsum 2005 69.36 8.69 7.98

Nr 8 Wood 2006 35.52 4.71 7.53

Nr 10 Wood 2007 40.08 6.56 6.11

3.2 Aggregate Thermal Energy Storage Model

Next these time constants from the individual modeling were used in producing the aggregate thermal energy storage model summarizing the SFD electric heating system in Sweden. The model is based on representing the aggregation of heating in each pricing area as having a thermal battery state of charge E ^t shown in (3.4) where over a time step ∆t we have mean power consumption P _el ^t and baseline power consumption P _b ^t

E ^t+1 = E ^t + (P _el ^t − P _b ^t )∆t (3.4) The state of the thermal battery is changed based on the sign of the power difference

∆P = P _el ^t − P _b ^t where a positive value indicates charging of the battery by heating the SFDs and negative value reducing the charge of the aggregate power by allowing the SFDs to cool down.

The mean power consumption P _el ^t can be varied between P _max which is the sum of all

the individual SFDs power capacities and P _min which in theory is zero. The energy in

the thermal battery is also confined between E _max where all SFDs are heated and E _min

when the temperature reaches the lower setpoint. Hence the operation of the aggregate

electric heating systems is constrained by the following conditions in (3.5) and (3.6).

P _min < P _el ^t < P _max (3.5)

E _min < E ^t < E _max (3.6)

For normal running conditions the heating systems are assumed to operate in duty cycles D which can be calculated based on the on t _ON and off t _{OF F} time for heating from the individual thermal model of the house. This gives P _b ^t = P _el ^t D. The energy content in one SFD is then in (3.7) calculated [5].

E ^t = P _el (1 − D)t _ON (3.7)

At normal operation with all SFDs in their duty cycles the thermal battery is assumed to be at E ^t = 0.5(E _max +E _min ). The model can then be set up to calculate the aggregate power available to unload the thermal battery for a set time period or power and time required to increase the charge of the battery to full capacity.

The model equations were written in MATLAB code to be used for simulation with the pricing areas each being separated in different programs. The temperatures in the individual SFDs were allowed to vary between the lower bound of T _l = 20 ^◦ C and higher bound of T _h = 22 ^◦ C. This temperature range is according to ASHRAE (American Society of Heating, Refrigerating, and A-C Engineers) standards generally within the comfort zone for winter which is the primary time of use for the electric heating. This range is also within the previously stated Swedish standard limits for human health.

3.3 Price Based Demand Response

One possible application for this thermal energy storage model is in demand response.

Previous studies on heating in Sweden for demand response have indicated that the winter is where such a solution can have the largest impact [1]. Simulation is therefore planned to be made for the winter months in order to find the maximum potential for this type of solution. One case would simulate the coldest day based on temperature data [19]. These simulations will then based on load profiles and electricity pricing data give the possible time periods for heating and cost of electricity [20]. Since these values are expected to be the maximum for the solution it will give an indication of the range of results available for demand response for heating systems in SFDs.

The temperature in the pricing areas are varied but for purposes of simulation one temperature will be assumed across the whole pricing area. This is because the individual SFDs specific geographical location is not known beyond its placement in one of the pricing areas. One measuring station located at the closest airport were chosen for each pricing area with Lule˚ a for SE1, Sundsvall for SE2, Bor˚ as for SE3 and Malm¨ o for SE4.

These geographical locations correspond well to the testing data for performance of the

heat pumps considered in the study. A cold day was chosen for the heating season

between September 2018 and May 2019 and since the north of Sweden generally has

the coldest weather the SE1 temperatures determined the day. February 6th 2019 was chosen to represent a cold day and a system of demand response was therefore set up for the four pricing areas for 24 hours. The temperature profiles for this day are shown in Fig. 3.1. A table of the temperature profiles is found in the appendix Table 7.1.

Electricity price data from the Swedish Nordpool market is shown in Fig. 3.2.

Figure 3.1: Temperature profile 060219

The goal of the demand response simulation was in this case peak shaving and valley lifting of the historical load profile in order to achieve the lowest possible cost for the day. Cost was calculated for the electricity consumed for the SFDs and the Demand Response was simulated in GAMS (General Algebraic Modeling System) software with baseline and optimum system at normal cycle conditions at the beginning and end point of the 24 hours simulated.

3.4 Emergency Demand Response

Another important application of the thermal energy storage model is demand response for emergency power shortages. The Emergency Demand response was based on a study of Generation Adequacy for Sweden in 2050 where a JULIA Monte Carlo simulation [21]

predicts the energy shortage that may be present in the electricity system. The thermal

battery model can be applied to provide this energy shortage assuming that a similar

SFD system to the one describes in the thesis remains in the year 2050. The energy

shortage in the year is shown for the pricing areas in Table 3.2 as a yearly value for

expected energy not supplied. Data from the generation adequacy study is based on an

Figure 3.2: Electricity price 060219

assumption of outdoor temperature -10 C for SE3 and SE4 to test the limitations of the system [22].

Table 3.2: Generation Adequacy Data Pricing Area Energy shortage [MWh]

SE1 0

SE2 0

SE3 7508

SE4 2148

The results from the modeling can be compared to previous work in the field. A

summary of the input value differences between this thesis and the previous BETSI

study is shown in Table 3.3 and will be important for understanding the similarities and

differences in the results.

Table 3.3: Comparison of study input parameters

Parameter Thesis BETSI

Ti [C] 20-22 21.2-24

Base Year 2016 2012

SFDs in total [million] 1.1 1.3

Building Types 14 574

Chapter 4 Results

4.1 Thermal Energy Storage

The thermal battery model was executed in MATLAB as a separate program for each of the four pricing areas. For the calculation temperature T _o was varied between 10 ^◦ C and −20 ^◦ C and the aggregate baseline power P _b ^t consumed in normal cycle operation changed as shown in Fig. 4.1 below. The baseline electric power required is the largest

Figure 4.1: Aggregate Baseline Power versus outdoor Temperature

when it is cold outside and SE3 having the most SFDs shows the greatest variation in power.

Next it was investigated how the model would respond to aggregate power consump-

tion below the baseline meaning that the SFDs would be cooled towards temperature T _l

from their normal cycle operation. The power consumption was varied up towards the baseline resulting in the time when the energy in the SFD aggregation is consumed. A 3D plot was produced for each of the pricing areas.

In Fig. 4.2 we have the aggregate heating electrical power P _el for SE1 and time from normal cycle operation to empty thermal battery with the baseline power displayed as a plane for different temperatures. In a similar fashion we have 3D plots for SE2 Fig. 4.3,

Figure 4.2: Aggregate heating power with baseline power as reference plane versus time until empty thermal battery charge

SE3 Fig. 4.4 and SE4 Fig. 4.5.

Comparing the 3D plots it is observed that each pricing area shows a similar trend with regard to outdoor temperature. The scale in power however is different as SE3 has the largest and SE1 has the smallest allowed power difference reduction. This power difference ∆P is the difference between the baseline plane and electrical power of the system and the maximum values for each of the pricing areas are displayed in the graph Fig. 4.6. It is observed that cold weather gives rise to the largest potential power reductions across the pricing areas.

The power reduction can only be kept for a limited time before the thermal energy depleted which is shown in Table 4.1.

When the thermal battery is at zero charge the power consumption must be larger than the baseline power in order to recharge the battery by heating the SFDs. To illustrate this 3D plots were produced to display heating electrical powers above the baseline and time from empty battery to normal charge for each of the pricing areas.

In Fig. 4.7 we have the aggregate heating electrical power P _el for SE1 and heating

time from empty thermal battery to normal charge with the baseline power displayed

Figure 4.3: Aggregate heating power with baseline power as reference plane versus time until empty thermal battery charge

Figure 4.4: Aggregate heating power with baseline power as reference plane versus time

until empty thermal battery charge

Figure 4.5: Aggregate heating power with baseline power as reference plane versus time until empty thermal battery charge

Figure 4.6: Maximum power reduction versus outdoor temperature

Table 4.1: Maximum Power decrease and time to empty thermal battery charge from normal cycle operation

E

SE1 0% SE2 0% SE3 0% SE4 0%

T

[C] ∆P [MW] Time [min] ∆P [MW] Time [min] ∆P [MW] Time [min] ∆P [MW] Time [min]

-20 -208 67 -523 65 -2853 66 -1138 66

-15 -183 76 -459 74 -2506 76 -1000 75

-10 -157 89 -395 86 -2158 88 -860 87

-5 -132 106 -331 102 -1809 105 -721 104

0 -106 131 -267 127 -1461 130 -582 129

5 -81 172 -204 167 -1112 171 -443 170

10 -55 251 -140 243 -764 249 -304 248

as a plane for different temperatures.

Figure 4.7: Aggregate heating power with baseline power as reference plane versus time until normal thermal battery charge

The same type of 3D plots can be seen for SE2 Fig. 4.8, SE3 Fig. 4.9 and SE4 Fig. 4.10.

The 3D plots show that the thermal battery can be recharged using a large aggregate

power, however this is at the maximum limit of the capacity of the heating systems and

may not be realistically achievable. The power difference should be taken as an upper

bound in this case. The power difference ∆P maximum values from the modeling for

each of the pricing areas are displayed in the graph Fig. 4.11 below. Considering the

safety of the system it would be preferable to heat at lower levels than the maximum

values in the graph. Still the higher the heating power, the quicker the thermal battery

Figure 4.8: Aggregate heating power with baseline power as reference plane versus time until normal thermal battery charge

Figure 4.9: Aggregate heating power with baseline power as reference plane versus time

until normal thermal battery charge

Figure 4.10: Aggregate heating power with baseline power as reference plane versus time until normal thermal battery charge

Figure 4.11: Maximum power increase versus outdoor temperature

will be recharged. The times for charging the battery at the maximum power difference is shown in Table 4.2.

Table 4.2: Maximum Power Increase and time until normal thermal battery charge

E

SE1 50% SE2 50% SE3 50% SE4 50%

T

∆P [MW] Time [min] ∆P [MW] Time [min] ∆P [MW] Time [min] ∆P [MW] Time [min]

-20 178 78 430 79 2773 68 1089 69

-15 203 69 494 69 3120 61 1227 61

-10 229 61 558 61 3468 54 1367 55

-5 254 55 622 54 3817 49 1506 50

0 280 50 686 49 4165 45 1645 45

5 305 46 749 45 4514 42 1784 42

10 331 42 813 42 4862 39 1923 39

Special cases can also be constructed such as preheating the thermal battery from normal cycle operation to full charge which numerically requires the same power and time as presented in Fig. 4.11 and Table 4.3. When the aggregation of SFDs is preheated to full charge it is possible to reduce the power potential for a longer time compared to if the system starts at normal cycle operation as shown in Table 4.3.

Table 4.3: Maximum Power Decrease and time from full to empty thermal battery charge

E

SE1 0% SE2 0% SE3 0% SE4 0%

T

∆P [MW] Time [min] ∆P [MW] Time [min] ∆P [MW] Time [min] ∆P [MW] Time [min]

-20 -208 134 -523 130 -2853 133 -1138 132

-15 -183 153 -459 148 -2506 152 -1000 151

-10 -157 178 -395 172 -2158 176 -860 175

-5 -132 212 -331 205 -1809 210 -721 209

0 -106 262 -267 255 -1461 260 -582 259

5 -81 345 -204 334 -1112 342 -443 340

10 -55 503 -140 487 -764 499 -304 496

Finally if the thermal battery is at zero charge heating up to full charge will require the following parameters Table 4.4.

Table 4.4: Maximum Power Increase and time from empty to full thermal battery charge

E

SE1 100% SE2 100% SE3 100% SE4 100%

T

[C] ∆P [MW] Time [min] ∆P [MW] Time [min] ∆P [MW] Time [min] ∆P [MW] Time [min]

-20 178 157 430 158 2773 137 1089 138

-15 203 138 494 138 3120 122 1227 123

-10 229 122 558 122 3468 109 1367 110

-5 254 110 622 109 3817 99 1506 100

0 280 100 686 99 4165 91 1645 91

5 305 92 749 91 4514 84 1784 84

10 331 84 813 84 4862 78 1923 78

In Table 4.5 we have sensitivity analysis performed for thermal model on MATLAB

file for SE1, outdoor temperature −20 ^◦ C and initial system at normal duty cycle con-

sumption. Analysis is done by increasing each parameter by 1 % and studying the response in output power difference from the initial absolute value of 208.8 MW.

Table 4.5: Sensitivity analysis results

Parameter increased by 1% Percentage change in ∆P Absolute value of ∆P [MW]

τ = RC -0.9578 206.7765

T _l 0.227455 209.2511

T _h 0.275928 209.3523

T _l T _h 0.502063 209.8244

From the sensitivity analysis it is observed that increasing the time constant τ and in turn thermal resistance decreases the output power of the model to an almost equal degree. Increasing the lower temperature bound has a small but significant impact on increasing the power potential and increasing the higher temperature bound has an even larger effect. Increasing both temperature bounds results in an on average higher indoor temperature and produces the largest temperature change effect for the output power potential.

4.2 Price Based Demand Response

The price based demand response was simulated for 24 hours on a cold day (6 February 2019) and results are shown in Fig. 4.12 for normal baseline operation power consump- tion and the DR operation for the combined pricing areas of Sweden. Here the power differences can be observed as the DR power is altered based on the price of the hour.

The maximum power increase is 6.5 GW and the maximum power decrease is 2.9 GW in the hours when the DR shuts off all the heating in the thermal battery. Furthermore the variation of the thermal battery charge during DR operation is shown in Fig. 4.13 where the changes from initial charge of 50 % to full and empty battery can be observed. In the final hour the system is returned to normal operation at baseline power consumption.

The economic savings achieved for 24 hours are about 1.45 million SEK comparing the Nordpool electricity cost for the baseline and DR case for the consumption of the 1.097 335 million SFDs. This results in about 1.32 SEK per SFD and day in savings. The savings are achieved even though the baseline simulation consumes slightly less energy in total with 65503.9 MWh/day compared to 65505.3 MWh/day for the DR system.

Hence the solution increases total energy use slightly.

4.3 Emergency Demand Response

Comparing the energy shortage calculated in a previous study for 2050 gave the following

outcome. The results of Emergency demand response can be seen in Table 4.6 for SE3

and SE4 where the energy shortage occurred. The energy for the normal case of baseline

cycle operation and a special case of preheated SFDs are shown. A part of the shortage

Figure 4.12: Aggregate heating power consumption on February 6th 2019

Figure 4.13: Energy of thermal battery during DR

can be met in the normal case and if the emergency can be forecasted it is possible to preheat the SFDs to 100% in about 1 hour assuming normal cycle operation before the charging. This would then mean that the thermal battery could reduce a larger percentage of the shortage.

Table 4.6: Emergency Demand Response

Area/Charge Energy for DR [MWh] Energy shortage [MWh] Reduction in [%]

SE3/ 50% 3165 7508 42.1

SE4/ 50% 1247 2148 58.1

SE3/ 100% 6330 7508 84.3

SE4/ 100% 2508 2148 100

Chapter 5 Discussion

In general the aggregate system has a higher power consumption as the outdoor tem- perature decreases. This in turn means that the possible potential power decrease value will be higher in cold days however only for a short amount of time since the battery is depleted more rapidly at low temperatures. The variations between pricing areas are mostly in terms of scale of the system as the aggregations behave in a similar fashion.

The case study for price based demand response can be compared to previous work about SFD heating systems in Sweden. This thesis results included a maximum power increase of 6.5 GW and maximum power decrease of 2.9 GW saving in total about 1.23 SEK per SFD and day. The model simulated in [1] when used for price based demand response can result in 4.4 GW of increased load and 5.5 GW of decreased load compared to this thesis. The SFDs are controlled depending on the price signal to reduce the energy costs but only minor savings are achieved of up to about 3800 SEK per year which corresponds to about 10 SEK per day on average for each SFD. The performance values which are not included in BETSI were taken from a reference from the Swedish National Board of Housing Building and Planning. An important constraint is the indoor temperature which in this case is allowed to vary from the Swedish measured average SFD temperature of 21.2 to 24 ^◦ C. Hence a higher maximum and minimum temperature bound was used compared to the the T _l = 20 ^◦ C and T _h = 22 ^◦ C of this study. In total the BETSI model summarized the energy from 1.3 million Swedish SFDs corresponding to the 1.29 million weighted buildings in the BETSI database. More buildings were considered for the year 2012 which the study is based on compared to the 1.1 million SFDs in this thesis from 2016 statistics.

Comparing data used in the studies, [1] studied 571 out of the in total 826 BETSI

buildings. The extracted SFDs all share the characteristic of having some kind of elec-

tricity based space heating system, however the performance data for these systems is

not included in the database. The SFDs are placed in the different pricing areas and

quantified with a weighting value indicating the amount of SFDs similarly to the 14

types used in this thesis. The work in [1] is based on modelling each SFD as a one-zone

energy balance using parameters from BETSI. The indoor temperature is dependent on

BETSI factors of heating equipment power, ventilation system heat transfer, transmis-

sion losses depending on the SFDs’ U-value, natural cooling, internal energy gains and

solar irradiation. The rate of temperature increase is governed by the total thermal mass of the BETSI building in question. The internal gains are based on the power profiles of the lighting, appliances, ventilation fans and occupants of the building. The model can be updated for each time step to account for changes in the climate and use of the build- ing. The energy use for each of the 571 SFDs is calculated and then using the weighting, scaled up for the total space heating electric demand. This could be considered more detailed than the analysis in this thesis as more heating factors are considered.

Further differences are that this thesis building simulation only considers solar gains in the yearly energy calculations as a single solar irradiation value per square meter while BETSI has detail regarding window shading and framing. Furthermore this thesis considers internal gains from radiators and furniture while BETSI has the power profiles of a number of internal gains mentioned above. Both calculations however consider the ventilation system losses and the influence of the building thermal mass. Comparing the insulation for similar buildings the BETSI building and building type of this study have U-values of the same size. The two buildings also have comparable size in terms of floor, surface and window area.

Comparing the results of this thesis to the BETSI model one can observe that the maximum potential load increase and decrease are in the same order of magnitude which indicates that the studies have comparable size. However since the studies are based on different years and since the BETSI model considers 200 000 more SFDs there is a difference in the results. The BETSI model can conclude a larger potential for load decrease while this thesis has the larger potential for load increase. The differences are surely in part due to the different indoor temperatures allowed in the study as BETSI heats up to 2 ^◦ C and never allows the temperature to drift below the average value, while still allowing a larger temperature bound range compared to this thesis. Also assumptions of system equipment heating power and performance very much influence the results but were not available for comparison. Both studies result in small savings per SFD for price based demand response in the order of magnitude of 1 to 10 SEK per SFD and day. The economical sustainability of the solutions is therefore questionable since these magnitude of savings are not expected to be enough incentive for families to support DR on their own. Some other economic solution is therefore required to finance the equipment required to control the heating systems as possible compensation to connected SFDs. This has not been investigated in the study but it is probable that this investment would need to be taken by either the electricity providing companies or responsible government authority to ensure implementation of this type of solution.

Futhermore effects of price feedback on the Nordpool market as the demand is shifted has not been investigated as the simulation was based on historical data.

The thermal battery model is limited in that it cannot account for the variations in

heating systems and their individual constraints that may be present. Some heat pump

systems for example may not be able to switch between power modes within the minute

resolution of the model due to technical limitations. Also the maximum increased power

assumes that heating systems are running at full power which in the long run may be

damaging to the systems. The value should therefore be seen as an upper threshold value

and should probably not be replicated in a real application. Furthermore considering the empty and full thermal battery in a real application the system would be at the boundary of its capabilities and some kind of safety margin to these states is suggested to avoid risk of SFDs temperatures drifting beyond the upper and lower bound values. Also in a real application there would be considerable variation in the response of individual SFDs which further suggests having safety margins in the system. A complete match with the thermal behavior in the 14 house types is not expected for real houses and some outliers may be present which will not store the expected amount of thermal energy. Also the actual technical control may be limited for some electric heating systems, especially in SFDs with manually controlled heating temperatures.

Considering the social sustainability of the thesis, if the SFDs respond as the ther- mal battery model predicts, thermal comfort should be achieved for the winter indoor temperatures as previously demonstrated which is an important aspect for families to accept this type of solution in the long run. Considering the ethics of the solution we have the question of how much control a family is willing to accept since the indoor temperature in the SFDs will be controlled between a lower and higher bound removing the freedom for an individual specific indoor temperature for the SFD. However health should not be compromised since the constraints of human health have been met in the modeling.

From the results of the emergency DR the potential for the solution was good. There are however limitations in the assumptions that the model for SFDs will be similar in Sweden 2050 where the solution is needed. Some variation is to be expected in the building stock and technology but it is surely possible that SFDs with electric heat pumps are in abundance in 2050 and can contribute to emergency DR. The full technical implications of the solution has also not been investigated, only the overall viability.

Effect of the load pick up after the emergency DR and the question of whether it is possible to forecast emergencies and preheat SFDs for greater effect remain.

The sensitivity analysis has shown the significant effect of temperature variations on the model which indicates that formulations of boundary values for the allowed indoor temperatures and their variations in time are key to the results that are to be expected from this solution. Application of this type of technology will need to consider this in combination with constraints of thermal comfort and health when deciding the temperature allowed limits.

Looking at the ecological sustainability of this thesis, such solutions may be impor-

tant in the future energy system especially with the increased introduction of renewable

energy generation. Applications such as emergency DR can reduce problems of energy

shortage which are likely when there is more electricity generation of varying output

such as wind power. Although the solutions of demand response may result in slightly

higher energy consumption this could be considered negligible compared to the system

size and should not compromise the sustainability. Also failure in developing solutions

of this kind could result in continued power generation from fossil sources to meet the

shortages which would not be preferable. Ideally this type of energy storage solution

could replace the need for quick fossil fueled emergency power plants such as gas turbines

when there is a mismatch between the demand and generation in the electricity system.

As the thesis title suggests the work has focused on finding the possible potential of

thermal energy storage of SFDs and therefore has limited detail beyond the maximum

values that are likely to be achieved. This is an effect of the limited time and resources

available for a Master Thesis but as has been discussed above the general results found

are comparable to studies of larger size such as the BETSI study.

Chapter 6 Conclusion

The purpose of this Master thesis was to apply a model to find the potential of using

electricity based building heating as thermal energy storage in Sweden and applications

for Demand Response (DR) purposes. The theoretical results in terms of power and

energy potential were compared to a similar model [1]. The prospect of electricity

based building heating in Single Family Dwellings (SFDs) as thermal energy storage can

constitute a considerable amount of power and energy capacity and a power decrease

of up to 2.9 GW compared to the baseline consumption was found for cold days. The

flexibility potential of electric heating was studied in two different Demand Response

settings: Price Based DR and Emergency DR in case of capacity deficit. For Price Based

DR, the economical savings were not large and therefore the financial incentive for an

individual SFD is small. Emergency DR may have a high benefit in the future 2050

energy system, where the risk of capacity deficit can be partially addressed by using

stored thermal energy in the Swedish SFDs. Outside of the economic considerations no

other factors were found that impair the sustainability and ethics of the thesis. In order

to obtain more accurate results, the detailed building property data from the previous

study [1] could be used with recent statistical data and indoor temperature bounds

that do not compromise thermal comfort. Furthermore, the technical infrastructure and

business model solution required for actual implementation of thermal energy storage

in SFDs could be studied. This would include constraints on the heating systems such

as cycle limitations of heat pumps. These questions need to be answered before real a

world implementation of the solution is possible.

Chapter 7 Appendix

Table 7.1: Temperature profiles 6th February 2019 Time [hours] SE1 T _o [C] SE2 T _o [C] SE3 T _o [C] SE4 T _o [C]

00:00:00 -22.3 -22.8 -1.4 2.9

01:00:00 -23.1 -22.2 -1.6 2.8

02:00:00 -24.5 -21.9 -1.8 2.6

03:00:00 -26.0 -21.3 -1.9 2.2

04:00:00 -25.2 -21.9 -1.9 2.1

05:00:00 -26.7 -20.9 -2.1 2.2

06:00:00 -27.8 -21.4 -2.2 2.3

07:00:00 -28.0 -21.8 -2.1 1.9

08:00:00 -28.4 -20.9 -1.7 2.2

09:00:00 -28.0 -19.1 -1.5 2.5

10:00:00 -25.8 -16.6 -1.1 2.6

11:00:00 -24.5 -14.3 -0.5 2.0

12:00:00 -25.2 -12.6 -0.4 0.9

13:00:00 -24.7 -12.5 -0.3 1.6

14:00:00 -25.3 -14.1 -0.2 1.8

15:00:00 -24.2 -14.2 -0.1 2.1

16:00:00 -25.7 -14.7 -0.1 2.5

17:00:00 -25.8 -14.1 -0.1 2.7

18:00:00 -27.2 -14.0 0.3 2.9

19:00:00 -26.9 -12.5 0.6 3.0

20:00:00 -27.6 -11.5 1.1 3.1

21:00:00 -27.4 -10.6 1.4 3.1

22:00:00 -27.3 -10.0 1.6 3.4

23:00:00 -26.3 -8.8 2.2 3.6

Figure 7.1: Nordpool pricing areas in Sweden

Figure 7.2: Type nr 1

Figure 7.3: Type nr 2

Figure 7.4: Type nr 3

Figure 7.5: Type nr 4

Figure 7.6: Type nr 5

Figure 7.7: Type nr 6

Figure 7.8: Type nr 7

Figure 7.9: Type nr 8

Figure 7.10: Type nr 9A

Figure 7.11: Type nr 9B

Figure 7.12: Type nr 10

Figure 7.13: Type nr 11

Figure 7.14: Type nr 12

Figure 7.15: Type nr 13

Bibliography

[1] Emil Nyholm, Sanket Puranik, ´ Erika Mata, Mikael Odenberger, and Filp Johns- son. Demand response potential of electrical space heating in swedish single-family dwellings. Building And Environment, 96(1):270–282, 2016.

[2] Statistikdatabasen. Antal l¨ agenheter efter region, hustyp och byggnadsperiod. Tech- nical report, SCB, Stockholm, 2016.

[3] Bj¨ orn Dahlroth. S¨ akrare v¨ armef¨ ors¨ orjning! Technical report, Svenska V¨ armeverkens Ekonomiska F¨ orening, Stockholm, 2007.

[4] J P Holman. Heat Transfer. McGraw-Hill, New York, 2010.

[5] Johanna L Mathieu, Marina Gonz´ alez Vay´ a, and G¨ oran Andersson. Uncertainty in the flexibility of aggregations of demand response sources. In 39th Annual Confer- ence of the IEEE Industrial Electronics Society, Vienna, 2013.

[6] Energimyndigheten. Energistatistik f¨ or sm˚ ahus. Technical report, Statisticon AB, Eskilstuna, 2016.

[7] Emil Nyholm. The role of Swedish single-family dwellings in the electricity system.

PhD thesis, Chalmers University of Technology, 2016.

[8] Hans Havtun and Paulina Bohdanowics. Sustainable Energy Utlilisation. KTH Energy Technology, Stockholm, 2014.

[9] Folkh¨ alsomyndigheten. Folkh¨ alsomyndighetens allm¨ anna r˚ ad om ventilation. Tech- nical report, Folkh¨ alsomyndighetens f¨ orfattningssamling, Stockholm, 2014.

[10] Joachim Claesson. Thermal Comfort and Indoor Climate. KTH Energy Technology, Stockholm, 2017.

[11] Boverket. Teknisk status i den svenska bebyggelsen. Technical report, Boverket, Karlskrona, 2010.

[12] U.S. Department of Energy. Furnaces and boilers, 2019.

https://www.energy.gov/energysaver/home-heating-systems/furnaces-and-boilers.

[13] Eric Granryd. Refrigerating Engineering Part II. Department of Energy Division of Applied Thermodynamics and Refrigeration Technology Royal Institute of Tech- nology, KTH, Stockholm, 2005.

[14] Energimyndigheten Testlab. Tester a-¨ o, 2018.

http://www.energimyndigheten.se/tester/tester-a-o.

[15] Eric Granryd, Ingvar Ekroth, Per Lundqvist, ˚ Ake Melinder, Bj¨ orn Palm, and Peter Rohlin. Refrigerating Engineering. Department of Energy Division of Applied Ther- modynamics and Refrigeration Technology Royal Institute of Technology, KTH, Stockholm, 2009.

[16] Lantm¨ ateriet. Natomraden.se, 2010. Svenska kraftn¨ at.

[17] Folkh¨ alsomyndigheten. Folkh¨ alsomyndighetens allm¨ anna r˚ ad om temperatur in- omhus. Technical report, Folkh¨ alsomyndighetens f¨ orfattningssamling, Stockholm, 2014.

[18] Johanna L Mathieu, G¨ oran Andersson, Maryam Kamgarpour, John Lygeros, and Duncan S. Callaway. Arbitraging intraday wholesale energy market prices with ag- gregations of thermostatic loads. IEEE Transactions on Power Systems, 30(2):736–

772, 2015.

[19] SMHI. Ladda ner meteorologiska observationer, 2019.

https://www.smhi.se/klimatdata/meteorologi/ladda-ner-meteorologiska- observationer/param=airtemperatureInstant,stations=all.

[20] Nord Pool. Historical market data, 2019.

https://www.nordpoolgroup.com/historical-market-data/.

[21] Alessandro Crosara. Calculating the risk of power shortage in the nordic power system, 2018.

[22] Nordic Energy Research. Flex4res, 2019. https://www.nordicenergy.org/flagship/flex4res/.