Rock cavern as thermal energy storage

Rock cavern as thermal energy storage

Simon Berglund

Sustainable Energy Engineering, master's level

2020

Luleå University of Technology

Preface

This thesis project has been the final part of my master program in sustainable energy engineering with spe-cialization in energy efficiency at Lule˚a University of Technology. The project have been carried out at the department of Engineering Sciences and Mathematics and in cooperation of Skellefte˚a Kraft AB.

I would like to thank Daniel Bystr¨om, my supervisor at Skellefte˚a Kraft for entrusting me with this project and for great advice and guidance. I would also like to thank my examiner Andrea Toffolo for the valuable engagement and his thorough support during the whole project.

I would also like to thank Andreas Johansson at Boliden R¨onnsk¨ar for all the information, help and rewarding discussions.

I would also like to thank Erik B¨acklund at Midroc Milj¨oteknik for providing me with valuable information of the caverns.

I would like to thank Emma, Gerson and Maria for their previous work in their idea study, which makes the foundation for this thesis.

Lastly, I would like to thank my mother, Eva-Lena, for supporting me throughout this project.

Simon Berglund Lule˚a, June 2020

Abstract

In the fall of 2019, a comprehensive idea study was conducted on heat storage in two rock caverns located at N¨asudden in Skelleftehamn and was part of the project course ”Energiteknik, huvudkurs” at Lule˚a University of Technology. This idea study investigated the conditions of using waste heat from Boliden AB:s copper smeltery (R¨onnsk¨ar) and storing this waste heat in two rock caverns and use them as seasonal thermal storage tanks, with the purpose of using the heat in the nearby district heating network, thus replacing some of the oil burned at R¨onnsk¨ar. To investigate this, the authors of the idea study looked at two different storage cycles of seasonal storage and modeled this in ANSYS Fluent to simulate the heat storage and the heat losses. The results from this idea study showed promising results for using these caverns as heat storage and this work is therefore a continuation of the idea study. Since the study provided a good understanding of the conditions for seasonal storage, some questions arose about how the rock caverns will behave during an intermittent operation, which is the planned mode of operating the caverns in case of deployment. In this thesis, intermittent operation of these caverns are explored and how this effects the temperature in the caverns and its surrondings, the charge/discharge speed, how insulated walls affect the operation and how much oil is replaced.

At the beginning of this project a review of the idea study and similar projects was done to gain deeper knowledge about the subject, but also to get a wider grasp on the different problems that could arise during the thesis. Relevant data for the caverns was collected and acquired to get a deeper understanding of its geometry, layout and what kind of modifications are really possible. Further data from the district heating networks of Boliden AB and Skellefte˚a Kraft was acquired. The available waste heat from R¨onnsk¨ar was examined and used to calculate the chargeable energy by hour for the caverns, with the limits of Skelleftehamn district heating network in mind. By examining the different steam boiler patterns, the discharge pattern could be calculated. Using CFD, the unknown global heat transfer coefficient between the cavern water and the cavern wall can be determined. This data was then used with a set of differential equations to model the behavior of the caverns in Simulink. This allowed to determine the behavior for the caverns during normal operation, such as how the heat losses evolve, how the temperatures fluctuate, how much heat the caverns can be charged with and how much they can discharge.

The results from the simulations showed that the caverns discharge a higher amount of energy when operating intermittently than when operating seasonally. Depending on how the caverns are utilized, different amounts of discharged energy are obtained. This range from 2224,7MWh to 7846,1MWh for the different discharging patterns. The usage also affects the efficiency of the cavern giving the efficiency a range between 19% to 53,9%. The heat losses range from around 20kW to 1000kW, depending on operation. Insulating the cavern walls reduces on average the heat losses by a factor of 5. Operating the caverns intermittently would on average remove a total of 29 ktonne CO2 and 88,74 tonne NOxfor its expected lifespan of 30 years. Economically, the rock caverns have good economic potential as they would save about 80 million SEK during their lifetime just from buying less oil.

Sammanfattning

H¨osten 2019 genomf¨ordes en omfattande id´estudie om v¨armelagring i tv˚a bergrum vid N¨asudden i Skelleftehamn och var en del av projektkursen ”Energiteknik, huvudkurs” vid Lule˚a tekniska universitet. Denna id´estudie unders¨okte villkoren f¨or att anv¨anda spillv¨arme fr˚an Boliden AB:s kopparsm¨altverk (R¨onnsk¨ar) och lagra denna v¨arme i bergrummen och anv¨anda dem som s¨asongslagrade ackumulatortankar. Syftet med detta var att anv¨anda v¨armen i det n¨arliggande fj¨arrv¨armen¨atverket och d¨armed ers¨atta en del av den f¨orbr¨anda oljan hos R¨onnsk¨ar. F¨orfattarna utforskade detta genom att unders¨oka tv˚a olika lagringscykler f¨or s¨asongslagring och modellerade detta i ANSYS Fluent f¨or att simulera v¨armelagring och v¨armef¨orluster. Resultaten fr˚an id´estudien visade lovande resultat f¨or s¨asongsbaserad v¨armelagring i dessa bergrum och detta arbete ¨ar d¨arf¨or en forts¨attning av id´estudien. Eftersom studien gav en god f¨orst˚aelse f¨or f¨orh˚allandena f¨or s¨asongslagring, uppstod n˚agra fr˚agor om hur bergrummen kommer att bete sig under en intermittent drift, vilket ¨ar den planerade driften av bergrummen vid en framtida anv¨andning. I detta projekt unders¨oks intermittent drift av dessa bergrum och hur detta p˚averkar temperaturen i bergrummen och dess omgivning, laddnings- / urladdningshastigheten, hur isolerade v¨aggar p˚averkar driften och hur oljef¨orbrukningen reduceras.

I b¨orjan av detta projekt gjordes en genomg˚ang av id´estudien och liknande projekt f¨or att f˚a djupare kunskap om ¨amnet, men ocks˚a f¨or att f˚a ett bredare grepp om de olika problem som kan uppst˚a under arbetets g˚ang. Relevant data f¨or bergrummen samlades in och anskaffades f¨or att f˚a en djupare f¨orst˚aelse f¨or dess geometri, layout och vilken typ av ¨andringar som verkligen ¨ar m¨ojliga. Ytterligare data fr˚an fj¨arrv¨armen¨atverket f¨or Boli-den AB och Skellefte˚a Kraft f¨orv¨arvades. Den tillg¨angliga spillv¨arme fr˚an R¨onnsk¨ar unders¨oktes och anv¨andes f¨or att ber¨akna den urladdningsbara energin per timme f¨or bergrummen, med begr¨ansningarna i Skellefte-hamns fj¨arrv¨armen¨at i ˚atanke. Genom att unders¨oka de olika ˚angpannm¨onstren kan urladdningsm¨onstret ber¨aknas. Med hj¨alp av CFD kan den ok¨anda globala v¨arme¨overf¨oringskoefficienten mellan bergrumsvattnet och bergv¨aggen best¨ammas. Denna data anv¨andes sedan med en upps¨attning differentialekvationer f¨or att modellera driften av bergrummen i Simulink. Detta gjorde det m¨ojligt att best¨amma beteendet f¨or bergrum-men under normal drift, till exempel hur v¨armef¨orlusterna utvecklas, hur temperaturen fluktuerar, hur mycket v¨arme bergrummen kan laddas med och hur mycket de kan ladda ur.

Resultaten fr˚an simuleringarna visade att bergrummen kan ladda ur en st¨orre m¨angd energi ¨an vid en s¨ asongs-betonad drift. Beroende p˚a hur grottorna utnyttjas erh˚alls olika m¨angder urladdad energi. Detta str¨acker sig fr˚an 2224,7MWh till 7846,1MWh f¨or de olika urladdningsm¨onstren. Anv¨andningen p˚averkar ocks˚a grottans effektivitet vilket ger en effektivitet mellan 19% och 53,9%. V¨armef¨orlusterna str¨acker sig fr˚an cirka 1000 kW till 20kw, beroende p˚a drift. Isolering av bergv¨aggarna minskar i genomsnitt v¨armef¨orlusten med en faktor 5. Att anv¨anda grottorna intermittent skulle i genomsnitt ers¨atta totalt 29 kton CO2 och 88,74 ton NOx f¨or den f¨orv¨antade livsl¨angden p˚a 30 ˚ar. Bergrummen har ¨aven god ekonomisk potential eftersom de skulle spara cirka 80 miljoner SEK under sin livstid bara fr˚an minskade oljekostnader.

Contents

1 Introduction 1

1.1 Skellefte˚a Kraft and Boliden AB . . . 1

1.2 Idea study . . . 1

1.3 Purpose and aims . . . 1

1.4 Limitations . . . 2

1.5 Literature study . . . 2

1.5.1 Idea study: Heat storage in rock cavern at N¨asudden in Skellefte˚a . . . 2

1.5.2 Haraholmen . . . 2

1.5.3 Lyckebo . . . 2

2 Theory 3 2.1 Storage tank . . . 3

2.2 Storage capacity . . . 3

2.3 Bedrocks undisturbed temperature . . . 3

2.4 Charging and discharging the cavern . . . 3

2.4.1 Secondary circuit . . . 5

2.5 The effectiveness-NTU method . . . 6

2.6 Cavern efficiency . . . 6

2.7 Insulation heat transfer coefficient . . . 7

2.8 Computational Fluid Dynamics . . . 7

3 Methodology 8 3.1 Cavern information . . . 8

3.2 Energy network Boliden R¨onnsk¨ar . . . 8

3.3 Skelleftehamn district heating . . . 9

3.4 System solution caverns . . . 11

3.5 Charging of caverns . . . 12

3.6 Discharging of caverns . . . 13

3.7 Determination of heat transfer coefficient . . . 16

3.8 Stratification layer calculation . . . 17

3.9 Simulink model . . . 18

3.9.1 HEX and Control system design . . . 19

3.9.2 Cavern design . . . 19

3.10 Oil and CO2 calculation . . . 21

3.11 Sensitivity analysis . . . 21

4 Results 22 4.1 One year simulation . . . 22

4.1.1 Year 2017 . . . 22

4.1.2 Year 2018 . . . 24

4.2 Single cavern simulation . . . 26

4.3 Insulated rock wall . . . 27

4.4 Time to reach maximum charge . . . 29

4.5 Sensitivity Analysis . . . 30

4.5.1 Cavern efficiency . . . 30

4.5.2 Lifetime analysis . . . 30

4.6 Lowered supply temperature . . . 31

5 Discussion 32

6 Conclusions 34

7 Future Work 35

8 References 36

A Cavern geometry 37

Nomenclature

Variable Unit Description

ρ kg/m3 _Density ˙ m kg/s Mass flow V m3 _Volume ˙ V m3_{/s Volume flow} S m2 _{Surface area} h m Height Q kW h Energy ˙ Q kW Power

cp kJ/kgK Specific heat capacity

U W/m2_{K Heat transfer coefficient}

λ W/mK Thermal conductivity

1

Introduction

In January 2018, a new climate law was enforced in Sweden. This law binds Sweden to a series of climate steps, gradually reducing the country’s carbon dioxide emissions by 2045. By then, emissions will be 85% less than 1990 [1].

For district heating production, a form of recovered energy called waste heat has been increasingly utilized in district heating systems. Simultaneously, carbon dioxide emissions for district heating production have declined steadily since 1990 [2]. If the waste heat is produced intermittently, it can be stored in a thermal energy storage and used at a later occasion, increasing the effectiveness of the waste heat.

1.1

Skellefte˚

a Kraft and Boliden AB

Skellefte˚a Kraft AB is a municipally owned company with business areas centering around electricity produc-tion, power distribuproduc-tion, electricity trading, broadband and heating and cooling. The main share of electricity is produced from nature in the form of wind- and hydropower. In Skellefte˚a the company Kraft owns a com-bined heat and power plant that produces electricity and district heating for large parts of the inner city of Skellefte˚a. However, in the outskirts of Skellefte˚a there exists a standalone district heating network supplying the communities of Skelleftehamn and Ursviken. This district heating network is only fed by waste heat from the Boliden R¨onnsk¨ar copper smeltery.

In R¨onnsk¨ar the waste heat originates internally mainly from metallurgical processes and is redistributed in the internal energy network consisting of a district heating network and a steam network. Of these two, the district heating network later supplies some of the heat to the nearby community of Skelleftehamn. However, in some cases the waste heat is not enough to meet the demand. This may be due to several reasons, such as the outside temperature being too low or an internal process requiring large amounts of waste heat. To solve this R¨onnsk¨ar fires an oil boiler to produce steam, which is used to raise the supply temperature for the community district heating network and meeting demand from the community.

1.2

Idea study

In the fall of 2019, a comprehensive idea study was conducted on heat storage in two rock caverns located at N¨asudden in Skelleftehamn and was part of the project course ”Energiteknik, huvudkurs” at Lule˚a University of Technology. This idea study investigated the conditions of using waste heat from Boliden copper smeltery and storing this waste heat in two rock caverns. The purpose was to use them as seasonal thermal storage tanks, using the heat in the nearby district heating network and thus replacing some of the oil burned at R¨onnsk¨ar. To investigate this, the authors of the idea study looked at two different seasonal storage cycles and developed a model in ANSYS Fluent to simulate the heat storage and the heat losses. The results from this idea study showed promising results for using these caverns as heat storage and this work is therefore a continuation of the idea study. Since the study provided a good understanding of the conditions for seasonal storage, some questions arose about how the rock caverns will behave during an intermittent operation, which is the planned mode of operating the caverns in case of deployment.

One of these questions, without an answer so far, is about how the temperatures in and surrounding the caverns behave during an intermittent operation. This is important to know because the heat losses from the stored hot water depend on the temperature of the cavern and the temperature of the surrounding bedrock.

Another question is how the charge/discharge speed of the cavern will change during operation, so in a later stage this can be used to dimension the pumping system for the cavern.

1.3

Purpose and aims

The purpose of this thesis is to increase the efficency of the district heating in Skelleftehamn and make it more sustainable. This is to be achieved by:

• Investigate the caverns operation on a hourly basis. • Determine how the temperatures evolve.

• Determine the resulting charge/discharge speed of the caverns. • Investigate the effect of insulated walls on the rock caverns. • Reduce the amount of fossil CO2emissions.

1.4

Limitations

The following items have not been considered for this thesis. • Heat loss from piping.

• The reinforcements needed for Skelleftehamn district heating network. • The different components effect on the amount charged/discharged. • Volume change in cavern due to temperature.

• Stratification layer growth. • Turbulence’s effect on the cavern.

1.5

Literature study

Throughout the project realisation, literature studies have been conducted to broaden the knowledge of the relevant areas studied. In this section a brief summary of selected literature is presented.

1.5.1 Idea study: Heat storage in rock cavern at N¨asudden in Skellefte˚a

Id´estudie: V¨armelagring i bergrum p˚a N¨asudden i Skellefte˚a, utf¨ord ˚at Boliden R¨onnsk¨ar by Maria Josefsson, Emma Marklund and Gerson Silva. As mentioned, this idea study was conducted as part of the course ”En-ergiteknik, huvudkurs” at Lule˚a University of Technology, investigating the conditions for heat storage in the same caverns of this thesis. In their work they assumed a uniform temperature of the water in the caverns, which means the heat transfer between the hot and cold water has not been accounted for. Their work also shows that a ”heat pillow” will appear after 5 years of operation, mitigating the heat loss to the bedrock. With their seasonal storage approach, they calculated a CO2 reduction of 250 tonne CO2per year by assuming that the energy discharged corresponds to an equal amount of oil saved.

1.5.2 Haraholmen

V¨armelagring i bergrum p˚a Haraholmen i Pite˚a by Sofia Viksten, thesis done at Lule˚a University of Technology 2018 [3]. The thesis investigates the technical and economical conditions for seasonal thermal storage in a rock cavern at Haraholmen (outside of Pite˚a). An important lesson from the work is that the elevated temperature of the water can cause the precipitation of minerals to occur in the water, thus increasing the hardness in the water and increasing the damage on heat exchangers.

1.5.3 Lyckebo

Solv¨arme med s¨asongslager i Lyckebo by Cay ˚Asberg, thesis done at Uppsala University 2011 [4]. This thesis investigates how a seasonal thermal storage in a rock cavern at Lyckebo has operated the past 30 years and how different types of heat losses have affected the cavern.

2

Theory

2.1

Storage tank

Storage tanks are large tanks commonly made of steel with the purpose of storing hot water which are divided into two different types, pressurized storage tanks and depressurized storage tanks. In pressurized storage tanks, as the name suggests, hot water is stored under pressure which enables the tanks to store water above 100C◦. Depressurized storage tanks on the other hand store hot water under atmospheric pressure which means it can only store hot water below 100C◦. An advantage with depressurized storage tanks are that they can be used as expansion tanks for district heating networks and shave of the peak load in the network.

To get the maximal usage of a storage tank the stratification principle is used. This means that the water is layered from top to bottom were the hot water resides at the top and the cold water at the bottom, all this due to the difference in density. Even with still water in a storage tank the stratification layer will grow due to waters condutivity. The stratification layer is found were 80% of the temperature difference is.

There are different methods to reduce the stratification layer were one of them are to get a good ratio of the height and diameter, H/D, of a circular tank. The other method is to use a nozzle to spread out the volumetric flow on a greater area, thus reducing the circulation from discharging and charging the tank. Because the caverns already exist, it would be deemed unnecessary to alter the geometric shape to get a good ratio for the height and diameter [5]. In figure 1 a cavern abiding to these principles can be seen.

Figure 1: A partially filled rock cavern. The hot water is represented by the upper red part, the cold water is represented by the lower blue part and the stratification layer is represented by the color gradient between these two.

2.2

Storage capacity

The theoretical maximum storage capacity at constant volume is calculated according to equation 1 [6]. ˙

Q = V ρcp∆T (1)

2.3

Bedrocks undisturbed temperature

Going downwards in the bedrock the temperature of the bedrock is increasing by 15 to 30 degrees for every kilometer in relation to the mean temperature of the location. This can be seen in equation 2.

Tud = Tm+

Tgradient

1000 hdepth (2)

Were Tud is the undisturbed temperature of the bedrock, Tm is the yearly mean temperature of the location, Tgradientis the temperature increase per kilometer and hdepthis the distance to the ground surface [7].

2.4

Charging and discharging the cavern

To find a suitable way to model the charging and discharging of the cavern we need to simplify the cavern. We assume that the cavern consists of two changeable volumes, one hot (Vh) and one cold (Vc), and assume that the stratification layer (Vlayer) between them remains constant. Because the caverns are closed systems we get

Vcavern= Vh+ Vc+ Vlayer (3) Depending on the charging or discharging operation the caverns will have different mass and energy balances, which will change the volumes and the temperatures of the water in the volumes. This can be seen in Equations 4-6.

When the caverns are charging, mass is added to the hot volume at a certain mass flow, ˙mh,in, and an equal amount of mass is subtracted from the cold volume at the same mass flow, ˙mc,out. The opposite happens when the caverns are discharging. This can be seen in Equation 4.

ρ∂Vh

dt = ˙mh,in = ˙mc,out = −ρ ∂Vc

dt (4)

Mass is added or subtracted from the hot volume depending on the operation of the caverns, so the energy balance changes depends on whether the cavern is charging or discharging, as seen in Equations 5 and 6.

˙

mh,in> 0 (

ρVhcp∂Tdth = ˙mh,incp(Th,in− Th) − ˙Qloss,h,rock− ˙Qlayer ρVccp∂T_dtc = − ˙Qloss,c,rock+ ˙Qlayer (5) ˙ mh,in< 0 ( ρVhcp∂T_dth = ˙Qloss,h,rock− ˙Qlayer

ρVccp∂T_dtc = − ˙mc,incp(Tc,in− Tc) − ˙Qloss,c,rock+ ˙Qlayer

(6)

Qloss,h,rock is the heat loss between the hot volume and the surrounding rocks, Qloss,c,rock is the heat loss between the cold volume and the surrounding rocks and Qlayer is the heat loss between the stratification layer and the hot and cold volume. These are calculated from Equations 7-9. Sh and Scare the surface areas for the hot water and cold water as a function of their respective volumes. Slayer is the surface area for the layer. λ is the thermal conductivity for water and Uwater,rock is the heat transfer coefficient between the water in the cavern and the cavern wall made of rocks. Q˙water,rock is the heat transfer from the water to the cavern wall through its surface area Scavern. Tcavern is the average temperature for the cavern.

˙ Qloss,h,rock= Uwater,rockSh(Vh)(Th− Tir) (7) ˙ Qloss,c,rock= Uwater,rockSc(Vc)(Tc− Tir) (8) ˙ Qlayer= λSlayer(Th− Tc) h (9) Uwater,rock= ˙ Qwater,rock Scavern(Tcavern− Trock)

(10)

Equation 11 is used to calculate Tir. Viris the volume of rocks surrounding the cavern, modelled as the internal rocks. This is included in the model to get something that is somewhat similar to a ”heat pillow” surrounding the cavern.

ρVircp ∂Tir

dt = ˙Qloss,h,rock+ ˙Qloss,c,rock− ˙Qloss,rock (11) Equation 12 is used to calculate ˙Qloss,rock. Sir,bedrock is the interface area between the internal rocks and the surrounding infinite bedrock. Trock is equal to the bedrock undisturbed temperature (Tud) and is a fixed value. This means that the rest of the infinite bedrock surrounding the internal rocks is simply modelled with a constant temperature.

˙

Qloss,rock= Sir,bedrockλ(Tir− Trock) (12)

Figure 2 shows how all these equations interact. The area surrounding the red and blue parts is the volume of the internal rocks and the external area represents the infinite bedrock. The double-headed arrows show how heat transfers that can occure in both ways between the volumes and the single-headed arrows show one-way heat transfers between the volumes. This means that Vh warms Vc through the stratification layer, Vh and Vc warm or are cooled by the internal rocks depending on their respective temperatures.

Figure 2: A cross section of a cavern showing the hot volume, cold volume, stratification layer, internal rocks and the bedrock.

2.4.1 Secondary circuit

During operation there may be periods in which Vh occupies the whole caverns and it needs to be kept warm to remain fully charged, i.e. to be optimally used in case of a discharge operation in the future. A secondary circuit is therefore used to heat Vh, adding a term to the right hand side in Equation 5. The modified expression can be seen in Equation 13 .

ρVhcp ∂Th

dt = mh,incp(Th,in− Th) − ˙Qloss,h,rock− ˙Qlayer+ ˙Qsec (13) where Qsecis calculated according to Equation 14.

˙

Qsec= ˙mseccp(Tsupply− Th) (14)

msec is the mass flow in the secondary circuit and  is the effectiveness of the heat exchanger in the secondary circuit according to the effectiveness-NTU method. In figure 3 the setup for charging and discharging the cavern can be seen.

Figure 3: The solution for charging and discharging the cavern.

2.5

The effectiveness-NTU method

It is easy to model a heat exchanger when the fluid inlet and outlet temperatures are known by using the log mean temperature difference (LMTD) method. Yet, when only the inlet temperatures are known, the LMTD method depends upon a heavy iterative process. Therefore, an alternative approach is preferred, called the effectiveness-NTU method [8]. The heat capacity rate C is defined as the mass flow rate multiplied with the specific heat capacity cp. By calculating Ch and Cc for the hot and cold fluids these can be used in Equation 15 to determine the maximum possible heat transfer rate.

˙

Qmax= min(Ch, Cc)(Th,in− Tc,in) (15)

The number of transfer units (NTU) is a dimensionless parameter is widely used for heat exchanger studies and is defined as in Equation 16.

N T U = U A Cmin

(16) The effectiveness is defined for a heat exchanger as the ratio of the actual heat transfer rate and the maximum possible heat transfer rate (Equation 17).

 = q

qmax

= Ch(Th,i− Th,o) Cmin(Th,i− Th,i)

(17)

In Equation 17 Th,iis the hot fluid inlet temperature and Th,o is the hot fluid outlet temperature. By assuming that Cmin= Chwe obtain:

 = Th,i− Th,o Th,i− Th,i

(18)

If the ratio between Ch and Cc is assumed to be zero the effectiveness can be expressed as:

 = 1 − e−N T U (19)

2.6

Cavern efficiency

Since the caverns will not operate seasonally but intermittently, a metric for cavern efficiency needs to be introduced. This can be seen in Equation 20. By dividing the discharged amount for one year (QDischarged) by the charged amount for one year (QCharged), the losses from the heat exchanger, the heat losses and the previous operation conditions are accounted for to some degree.

η = QDischarged QCharged

2.7

Insulation heat transfer coefficient

If insulation is applied to the inner surface of the cavern, a new overall heat transfer coefficient needs to be calculated. This is done by using Equation 21, where Uwater,rock is the heat transfer coefficient between the water in the cavern and the cavern wall made of rocks, Linsulationthe thickness of the insulation outwards from the wall and λinsulationthe thermal conductivity of the chosen insulation material.

Uinsulation= 1 1 Uwater,rock + Linsulation λinsulation (21)

2.8

Computational Fluid Dynamics

Computational fluid dynamics (CFD) is a numerical method used to solve fluid mechanics problems and allows a detailed analysis of a process flow combined with mass and heat transfer [9]. By assuming some boundary conditions and a geometry for a given system, the geometry for this system can be divided into a set of cells which have a simpler geometry than the system itself. This set of cells forms what is known as a ”mesh”. The mass, heat transfer and the flow is represented by a set of differential equations, which are converted to linear or non-linear equations that describes the mass, heat transfer and the flow depending on neighboring cells in the mesh.

To create a CFD simulation, a certain amounts of steps have to be taken in a well defined procedure, all of which can be seen in Figure 4.

3

Methodology

At the beginning of the thesis a review of similar projects was done to gain deeper knowledge about the subject, but also to get a wider grasp on the different problems that could arise during the thesis. Relevant data for the caverns was collected and acquired to get a deeper understanding of its geometry, layout and what kind of modifications are really possible. Further data from the district heating networks of Boliden AB and Skellefte˚a Kraft was acquired. This data was then used with a set of differential equations to model the behavior of the caverns in Simulink. This allowed to determine the behavior for the cavern during normal operation, such as how the heat losses evolve, how the temperatures fluctuate, how much heat the caverns can be charged with and how much they can discharge.

3.1

Cavern information

The information regarding the cavern was given by Skellefte˚a municipality and Midroc Milj¨oteknik [10, 11]. The bedrock type surronding the cavern is granodiorite.

Volume per cavern [m3] 19000

Height [m] 33

Width [m] 15

Arc height [m] 3,5

Distance between caverns [m] 21,5

Distance to surface [m] 10,5

Water leakage [m3_{/year] 3 100}

Density granodiorite [kg/m3_{] 2650} Specific heat capacity granodiorite [J/kg,K] 790 Thermal conductivity granodiorite [W/m,K] 3,24 Table 1: Given information about the cavern from various sources.

3.2

Energy network Boliden R¨

onnsk¨

ar

The internal energy network at Boilden R¨onnsk¨ar works in such a way that heating oil 5 is pumped from a oil storage tank and then burnt in two boilers to produce steam, which is added to the internal steam network. Some waste heat in the form of steam is also added to this network (Figure 5).

Figure 5: Steam production.

This steam network is partly used in different internal metallurgical processes, but is also used in the district heating network inside Boliden R¨onnsk¨ar. This district heating network is shown in Figure 6.

Figure 6: Adapted P&ID diagram of the energy system for Boliden R¨onnsk¨ar copper smeltery.

Going from left to right, in this district heating network the steam network exchanges heat with condensers 851 and 852 heating the incoming return water, represented by the dotted line. Depending on the demand from the internal processes, some of the heated water from condensers 851 and 852 is cooled off by opening a valve letting it into trim coolers 857 and 858. The rest of the heated water is then transported along the main pipe, where some additional hot water from the acid and fuming process is added. Thereafter, part of the hot water goes to R¨onnsk¨ar east and R¨onnsk¨ar west, two sub-district heating networks at the site. Then, during some parts of the year, the condenser 816 warms up the outgoing water. After all of these steps are made, the remaining hot water passes on to Skelleftehamn district heating network.

3.3

Skelleftehamn district heating

To get a better understanding of the operation of the Skelleftehamn district heating network an overview of the current flowrate, supply temperature and district heating power is offered here. A duration curve was created to see how the flowrate changes during the year (Figure 7). From this figure a reasonable maximum flow can be determined for the Skelleftehamn district heating network, which is used later to determine how much extra water can be added to the Skelleftehamn node.

0 10 20 30 40 50 60 70 80 90 100

Percentage of the year [%]

0 50 100 150 200 250 Flow rate [m 3 /s]

Flow rate duration curve

Figure 7: Duration curve for Skelleftehamn’s flowrate for 2018.

To see how the supply and return temperature are affected by the outdoor temperature, the temperatures were linearized as a function of the outdoor temperature. By setting a constant flow rate in equation 1 a lowered supply temp can be calculated. This is presented in the results section.

-30 -20 -10 0 10 20 30 Outdoor temperature [C°] 40 50 60 70 80 90 100 110 Temperature [C°]

Linearized supply and return temperature Skelleftehamn

Supply temperature Return temperature

Figure 8: The current supply temperature as a linear function of the outdoor temperature.

Since the energy that is supposed to be used to charge the cavern is going to be added on top the current energy going to Skelleftehamn district heating network, the theoretical chargeable energy can be visualized in a load duration curve (Figure 9). Of course this is not optimal because it will increase the current wear on all parts of the network, but it shows a great potential for how much energy could be used to charge the caverns in case

the current waste heat is not enough.

Figure 9: The load duration curve for Skelleftehamn district heating network 2018 and the maximum chargeable energy on top.

3.4

System solution caverns

The system solution for connecting the caverns with the waste heat produced by Boliden R¨onnsk¨ar and the Skellelftehamn district heating network can be seen in Figure 10. The caverns and the Skelleftehamn district heating network are modeled as two separate closed systems, with heat transfer occurring through a series of heat exchangers to ensure constant efficiency. During a charging operation the supply temperature from Boliden R¨onnsk¨ar is used to heat the cold water taken from the bottom of the cavern and then the resulting heated water is added to the top. When the cavern is discharged, the return temperature from the district heating network is heated by the hot water taken at the top of the cavern and then the resulting cooled water is added to the bottom of the cavern.

3.5

Charging of caverns

The charging of the caverns occurs when the heat demand from Skelleftehamn district heating network is lower than the current available waste heat. This means the chargeable amount depends on the load of Skelleftehamn network and R¨onnsk¨ar internal waste heat production. Because the chargeable energy is going to be included in the same pipe going to the community district heating network we need to first look at how much energy is going to Skelleftehamn.

Then, depending on whether the current flow to Skelleftehamn is below a historically reasonable maximum value, we add water from the emergency cooler. Thereafter, the water is heated with the energy cooled in the trim coolers, as they are indirectly cooling excess heat from the steam network. If the amount of heat is enough to raise the temperature of the water to above 120◦C (the maximum temperature for the district heating network) it will only raise it to 120◦C. Then, by subtracting the current district heating energy, the chargeable energy for the caverns is obtained.

Figure 11: The scheme describing how the chargeable energy is calculated.

After this was calculated, a three point centered moving average of 3 hours was applied to the chargeable energy to even out the dramatic changes and make the chargeable energy more akin to how it would behave in reality. All this procedure results in Figure 12.

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time [h] -2 0 2 4 6 8 10 12 Power [MW]

Chargeable energy 2018

3.6

Discharging of caverns

The caverns are discharged with the purpose of replacing the oil burnt in the steam boiler at R¨onnsk¨ar Boliden. To find the discharge pattern we assume that the amount of energy that is discharged from the cavern replaces an equal amount of oil. This means that the periods when oil is burnt correspond to the periods when the caverns should discharge. By using the data for the oil tank level seen in the upper plot of Figure 13 we can calculate the oil level change with the difference between the next timestep and the current, Oil level change = F ill levelt+1− F ill levelt. If the difference is below zero it means the tank is emptied. The absolute value of these are then plotted in Figure 13.

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time [h] 40 60 80 100 Fill level [%]

Oil tank fill level 2018

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time [h] 0 0.02 0.04 0.06

Oil level change

Calculated oil outflow from tank 2018

Figure 13: The level of the oil tank for the year 2018 and the calculated oil level change.

This could be used to determine when to discharge the caverns as the difference in tank level shows the periods in which some oil is burnt. Figure 13 shows that the instants in which the cavern should be discharged are few and instantaneous, making it problematic to implement in a Simulink model, even by flattening the curve with a moving average. Another problem with using this as a basis for the discharge pattern is that the level of the tank is determined by measuring the height of the oil in the tank relative to the height of the tank. With a tank volume of 50m3_{and with an accuracy of one decimal figure, the calculated difference does not give a good} estimate about when oil is burnt. To get a better estimate we can instead look at how much steam is hourly produced by the steam boilers (Figure 14). In this way two measurement points for the amount of oil burnt are obtained, giving a more accurate estimate about when oil is burnt. To smooth out the data, a three point centered moving average of 3 hours is applied to the measured steam flow.

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time [h] 0 20 40 60

Steam flow [tonne/h]

2017

Steam flow [tonne/h]

2018

Produced steam from steam boiler 1 & 3

Figure 14: The measured steam flow from oil boiler 1 and 3 for the year 2018.

These two diagrams are then used as the basis to calculate the discharge pattern, which is made of the instants in which the steam flow is above 0 tonne/h. Since charging and discharging at the same time is impossible, the charge pattern is therefore assumed to be the opposite of the discharge pattern.

With the current available data it is impossible to know how much of the produced steam from the steam boilers is used in the metallurgical processes, but we know that the trim coolers continuously cool off some energy. This is therefore used as the maximal dischargeable energy (Figure 15).

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time [h] 0 5 10 15 20 25 Power [MW]

Maximum dischargeable energy 2018

Figure 15: The maximum chargeable energy for 2018.

burnt. An example of this is seen in Figure 16. 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time [h] 0 5 10 15 20 25 Power [MW]

Dischargeable energy 2018

Figure 16: The chargeable energy for 2018.

By inverting Figure 16 and filling the gaps with the calculated chargeable energy seen in Figure 12, we get a vector representing how much the cavern should charge or discharge at any given moment for one year, which is called Qsupply (Figure 17).

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time [h] -25 -20 -15 -10 -5 0 5 10 15 Power [MW]

Q

2018

3.7

Determination of heat transfer coefficient

ANSYS Fluent 2019 R3 was used to determine the heat transfer coefficient for the water-bedrock interaction. The first step was to create the geometry for the calculations in 2D. This was done to decrease the computational time, but also because the determination of the heat transfer coefficient is not affected by the third dimension. The geometry was meshed with one meter between the nodes in the bedrock and a finer inflation layer mesh between the cavern and bedrock to get a better calculation.

Figure 18: The geometry of the model. The red area represents the rock cavern and the blue area represents the surrounding bedrock. The temperature for the rock cavern is set to 93◦_{C and the bedrock’s temperature is} set to 3◦C.

A transient simulation was performed in which the initial condition for the whole surrounding bedrock was set to the bedrocks undisturbed temperature, for the whole body of water to its maximum temperature and the boundary condition for the sides of the bedrock was set to adiabatic. The bedrock material was set to granodiorite, as it is the most common rock type in the bedrock surrounding the cavern [12]. The temperature in the cavern was measured as a area-weighted average and the heat flux from the cavern water to the rock wall was measured as a area-weighted average. The simulation did not account for the heat loss due to the in-leakage from the groundwater. After the transient simulation was finished the measured heat flux and cavern temperature was imported into MATLAB. By using equation 10 we can calculate the heat transfer coefficient. To calculate the heat transfer coefficient in the case of insulating the rock walls, Equation 21 was used. Since cavern walls have a uneven surface, it was decided to use polyurethane foam because it sticks well to a uneven surface. Its thermal conductivity is 0,026W/m2_{K. The thickness of the insulation was set to 10 cm as this would} lower the heat losses drastically [13]. The resulting heat transfer coefficient for the insulation can be seen in the following equation: Uinsulation= 1 1 Uwater,rock + Linsulation kinsulation = ₁ 1 4,59+ 0,1 0,026 ≈ 0, 25

0 50 100 150 200 250 300 350 Time [h] 4.3 4.35 4.4 4.45 4.5 4.55 4.6 4.65 4.7

Heat transfer coefficient [W/(m

2 K)]

Global heat transfer coefficient between water and granodiorite

Calculated heat transfer coefficient Mean of heat transfer coefficient

Figure 19: The calculated global heat transfer coefficient between the water in the cavern and the surrounding rock walls.

3.8

Stratification layer calculation

Figure 20 was used to calculate the size of the stratification layer. As mentioned in the theory section, the stratification layer is found where around 80% of the temperature difference takes place, which in this figure corresponds to 3 meters[5].

By assuming that the temperature distribution and the relationship between the hot and cold temperature will look similar, even though the bottom temperature of the cavern is lower, we can calculate the height of the stratification layer as the ratio between cavern height and the height of the reference storage tank. This can be seen in the following equation:

hlayer,cavern= hlayer,tank

Hcavern Htank

= 3 ·33 14 ≈ 7

3.9

Simulink model

The model of cavern operation is developed in Simulink, which is a block diagram environment for multidomain simulation, Model-Based Design and also well suited for simulating dynamic systems [14]. Figure 21 shows the top layer of the Simulink model used for testing different inputs for the model. In the top left corner of the figure contains the calculated variables for the model: Qsupplyis a vector containing the discharge/charge pattern and its corresponding discharge/charge amplitude, Tsupply is the calculated supply temperature in Kelvin for the caverns and Skelleftehamn district heating network, Treturn is the calculated return temperature in Kelvin for the caverns and Skelleftehamn district heating network. The bottom of the figure contains three display blocks showing how much energy the cavern receives from the secondary circuit, from the primary circuit and how much is discharged from the cavern in that order. The block named ”All-in-one Scope” shows the plots for the different input and output signals of the model. These are in order: the temperatures Th, Tc, Tir, Tsupply and Treturn, the volumes Vh and Vc, the different heat losses Qloss,h,rock, Qloss,c,rock, Qloss,layer, Qloss,rock and the different powers Qsupply, Qsecondary, Qprimary (Qprimary is the power used to charge the cavern through the primary circuit and Qsecondary the power used to charge the cavern through the secondary circuit).

The model is divided into two subsystems, the two grey blocks named ”cavern” and ”HEX and control”. The equations related to the caverns are implemented in the ”cavern” block. The equations for the primary circuit, the control system and the secondary circuit are implemented in the ”HEX and control” block.

3.9.1 HEX and Control system design

A simple control system was developed to make charging and discharging operations occur as intended. Also, one task of this control system is to ensure that the values in the model stay within feasibility boundaries from the thermodynamic point of view. This control system is shown in Figure 22 in the upper blue area. Depending on what type of operation the cavern is performing, it sends a binary signal to the heat exchanger area that toggles which of the two cavern volumes should be ”filled” or ”emptied”. It is also continuously checking if the hot (cold) volume is completely filled to disable the charging (discharging) operation. The cavern is assumed to be completely filled when it has reached the cavern volume minus the stratification layer volume.

The equation related to the secondary circuit are shown in the red area. The secondary circuit is activated only when Vh is filled, the temperature difference between Th and Tsupply is 10◦C or above, the hot temperature is below 95◦C and when the cavern is in charging mode. Using the NTU-method seen in Section 2.5, the mass flow rate in the secondary circuit is calculated with a lookup table as a function of the heat capacity rate. This mass flow rate is then used in a function block to calculate the effectiveness  with Equation 19.

The heat exchanger model is seen in the yellow area. By using the NTU-method and assuming that we have several heat exchanger we can keep the effectiveness for the heat exchanger model constant (it is assumed to be 90%).

Figure 22: The control system for the caverns.

3.9.2 Cavern design

By using the equations found in Section 2.4 the model related to the cavern temperatures, volumes and heat losses can be implemented. An overview of this can be seen in Appendix B. Assuming that the thickness of the layer of internal rocks is 5m out of the cavern surface, the surface of the interface between the internal rocks

equal to 31487m3. The heat transfer coefficient between the internal rocks and the infinite bedrock is calculated by dividing the thermal conductivity for granodiorite with half of the thickness of the internal rocks. This gives a heat transfer coefficient of 1,34W/m2_K.

Going into the subsystem named ”Qloss,h,rock calculation” seen in Appendix B we can see how Qloss,h,rock is modeled for a single cavern. By using once again the equations in Appendix A we can calculate a lookup table to get the surface area inside the cavern as a function of Vh. When the height of Vh passes the arc height of the cavern it starts adding the corresponding surface area of the box part of the shape. Doing it this way Vh becomes filled from the top to the bottom just as regular storage tanks. Then the calculated area is multiplied by the heat transfer coefficient and the temperature difference. The heat loss from the groundwater leakage is also added, assuming it will mostly leak into the hot volume.

Figure 23: The subsystem for the Qloss,h,rockcalculation.

Going into the subsystem named ”Qloss,c,rock calculation” seen in Appendix B we can see how Qloss,c,rock is modeled for a single cavern. Since the stratification layer is considered as a constant volume, Vc does not have to account for the arc surface of the cavern when its surface area is calculated. To model the surface area as a function of Vc and making the subsystem simpler, a lookup table between Vc and its surface area was made.

Figure 24: The subsystem for the Qloss,c,rock calculation.

The heat loss from the stratification layer, Qlayer, is calculated in its namesake subsystem named according to equation 9.

Since the subsystems for the heat losses are constructed for a single cavern, they are assumed to scale linearly, thus only requiring to be multiplied by two in order to get the heat losses for the two caverns. This is also the case for Qloss,rock. Therefore, Vir and Sir,bedrockare doubled.

3.10

Oil and CO

calculation

The oil spared at Boliden R¨onnsk¨ar by discharging of the caverns is heating oil 5. The amount of heat released by this oil during complete combustion (the heating value) is 10734kWh/m3. The CO2 emission factor for this oil 0,272kg/kWh [15]. The NOx emission factor for this oil is 0,17kg/GJ[16]. These values are multiplied by the amount of heat discharged from the caverns to calculate the m3 of oil replaced and the avoided emissions of CO2 and NOx.

3.11

Sensitivity analysis

In order to gain a better understanding of the different operating conditions in the two caverns, a sensitivity analysis was performed. This is done by calculating a worst, medium and best case scenario based on the different temperatures, flows and powers for the district heating and waste heat. By doing this, the influence from the operating conditions of Boliden R¨onnsk¨ar and the outdoor temperature gets lowered. Also, the discharge pattern was calculated for the two years, 2017 and 2018, to see extensively how the steam boiler operating conditions affect the caverns. Accordingly, a total of 6 different cases are tested. Years 2017 and 2018 are used because they both have enough data to calculate the needed variables.

Then, in order to get a firmer grasp on how the cavern will behave during its expected lifetime of 30 years, a Monte-Carlo simulation was performed. By randomizing the 6 different scenarios and their corresponding amount of energy discharged for a 30 year lifetime we get one unique history of operating conditions for the entire lifetime. This means that we calculate a lifetime for the caverns that statistically will include the worst, average and best cases, thus giving some variety in the expected lifetime.

4

Results

4.1

One year simulation

4.1.1 Year 2017

Figure 25 shows the results of one year of operation with the calculated charging and discharging pattern of 2017. The initial conditions for the simulation are Th = 366K, Vh = 19000m3, Tc = 319K, Vc = 19000m3, Tir = 341K. The simulation is run from the 1st of January to the 31st of December with a time step of one hour.

Charged and discharged energy 2017

Figure 26: The amount charged/discharged after one year of operation with the calculated discharge pattern of 2017.

Figure 26 shows the amount of discharged energy (2492MWh) and the amount of charged energy (12260MWh, of which 7875MWh from the secondary circuit). If the discharged amount is assumed to be equal to the amount of oil saved, this results in 232m3 _{heating oil 5 not burned or, in terms of CO}

4.1.2 Year 2018

Figure 27 shows the results of one year of operation with the calculated charging and discharging pattern of 2018. The initial conditions for the simulation are Th = 366K, Vh = 19000m3, Tc = 319K, Vc = 19000m3, Tir = 341K. The simulation is run from the 1st of January to the 31st of December with a time step of one hour.

Charged and discharged energy 2018

Figure 28: The amount charged/discharged after one year of operation with the calculated discharge pattern of 2018.

Figure 28 shows the amount of discharged energy (7208MWh) and the amount of charged energy (13770MWh, of which 2785MWh from the secondary circuit). If the discharged amount is assumed to be equal to the amount of oil saved, this results in 672m3 _{heating oil 5 not burned or, in terms of CO}

4.2

Single cavern simulation

In case one would like to fill in one cavern to use the area above ground for something else, a simulation for a single cavern was performed. The initial conditions for the simulation were Th = 366K, Vh = 9500m3, Tc= 319K, Vc= 9500m3, Tir= 341K. The simulation is run from the 1st of January to the 31st of December with a time step of one hour.

Charged and discharged energy 2018 - single cavern

Figure 30: The amount charged/discharged after one year of operation with the calculated discharge pattern of 2018.

Figure 30 shows the amount of discharged energy (5583MWh) and the amount of charged energy (9155MWh, of which 1711MWh from the secondary circuit). If the discharged amount is assumed equal to the amount of oil saved, this results in 520m3 _{heating oil 5 not burned or, in terms of CO}

2emissions, 1519 tonne CO2.

4.3

Insulated rock wall

To see the effect of insulating the cavern wall the heat transfer coefficient (Uwater,rock) for Qloss,h,rock and Qloss,c,rockwas changed to 0,2997 W/m2K, as calculated with Equation 21 assuming a thickness of 10cm for the insulation. The initial conditions for the simulation are Th= 366K, Vh= 19000m3, Tc= 319K, Vc= 19000m3, Tir = 341K. The simulation is run from the 1st of January to the 31st of December with a time step of one hour.

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Time [h] 0 200 400 600 800 1000 1200 1400 1600 1800 Power [kW]

Heatlosses hot part mean 2018

Q

loss,h,rock - no insulation Q

loss,h,rock - 10cm insulation

Figure 31: The heat losses for the hot volume of the cavern for a insulated and non-insulated case for the year 2018.

Charged and discharged energy 2018 - insulated caverns

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Energy [MWh] Charged Discharged Primary circuit Secondary circuit

Figure 32: The amount charged/discharged with insulation after one year of operation with the calculated discharge pattern of 2018.

Figure 32 shows the amount of discharged energy (8131 M W h) and the amount of charged energy (9103 M W h, of which 100MWh from the secondary circuit). If the discharged amount is assumed to be equal to the amount of oil saved, this results in 758m3 _{heating oil 5 not burned or, in terms of CO}

4.4

Time to reach maximum charge

In order to see how many days it takes to have the cavern fully charged, two different simulations were conducted for the 2018 operating year. These simulations varied seasonally to see if the rock caverns are seasonally dependent. The summer simulation started the first of June 2018 and the winter simulation the first of December 2018. The initial conditions for the simulations were Th = 283K, Vh = 8100m3, Tc = 277K, Vc = 29900m3, Tir= 280K. To calculate how charged the rock cavern is, Equation 1 was used, calculating ∆T as the difference between Th and the mean of the return temperature (316K).

0 20 40 60 80 100 120 140 Time [days] 0 0.5 1 1.5 2 Energy [GWh]

Time to fully charge the cavern for 2018

Volume hot winter Volume hot summer

4.5

Sensitivity Analysis

By calculating a worst, average and best case scenario between the operating years 2017 and 2018, a range of the operating conditions is calculated. By combining these cases with the discharge patterns for 2017 and 2018 six different scenarios can be simulated. Table 2 shows how much energy is discharged from the caverns for these six scenarios.

Qdischarged [M W h]

2017 2018

worst 2224,7 6567,9 average 2467,3 7221,1

best 2713,0 7846,1

Table 2: The amount discharged from the caverns for the different scenarios.

4.5.1 Cavern efficiency

By using Equation 20, the efficiency for the scenarios in the section above can be calculated (Table 3). η [%]

2017 2018

worst 19,0 50,9

average 20,2 52,5

best 21,4 53,9

Table 3: The calculated efficiency for the discharging operation.

4.5.2 Lifetime analysis

In order to see how much the caverns discharge for the proposed lifespan of 30 years, a Monte-Carlo simulation was performed (Figure 34).

0 10 20 30 40 50 60 70 80 90 100 Cumulative probability [%] 80 100 120 140 160 180 200 220 Energy [GWh]

Energy discharged 30 year lifespan

4.6

Lowered supply temperature

An alternative way to reduce the oil consumption is lowering the supply temperature in Skelleftehamn district heating network and not using the caverns at all. By using some of the water from the emergency cooler, a lower amount of energy would be needed to heat up the return pipe from Skelleftehamn. By keeping the flow rate constant at 144m3_{/s the new calculated supply temperature can be seen in Figure 35 (the supply temperature} in the district heating network should not be below 65◦C).

-30 -20 -10 0 10 20 30 Outdoor temperature [C°] 40 50 60 70 80 90 100 110 Temperature [C°]

Linearized supply and return temperature Skelleftehamn

Supply temperature Return temperature New supply temperature

5

Discussion

As seen in Figures 25 and 27 the behaviour of the caverns depends deeply on the kind of charge and discharge pattern applied to them. For 2017 the cavern hot volume is filled most of the time and, therefore, one might wonder if it is worth operating the rock caverns with two separate volumes (hot and cold) or if it is enough just to operate with one volume at uniform temperature. Since the caverns do not discharge much during this year of operation, the heat loss to the internal rocks does not fluctuate much and could almost be seen as the expected heat loss during steady state. By looking at the lowest plot in Figure 25 one can see that the secondary circuit is in a steady operation mode from 6500 hours and onwards, keeping the caverns completely charged. For 2018 it can be seen that Th drops very fast when Vh is empty, especially when the cavern at that moment is asked to discharge.

In the one cavern case, one can see that the charging and discharging happens faster compared to the corre-sponding two cavern case. In this case it would probably not make much of a difference to operate the cavern as one volume. It is also interesting that the discharged amount from the single cavern is only approximately 80% of the discharged amount for two cavern the same year. This really shows how much waste heat is actually available.

In the simulation it is assumed that the stratification layer is a constant volume and does not increase depending on the charging and discharging of the caverns. This assumption is made because it is difficult to know how the turbulence from charging and discharging the cavern will affect the layer. A compromise could be done by implementing a new subsystem in Simulink, in which the layer volume is non-constant. In this subsystem, the height of the layer would only increase when the secondary circuit is activated and reset only when a certain discharge threshold have been passed. This way, the layer growth will not be affected by the turbulence from pumping water between the cold and hot volumes.

As can be seen in Section 4.3, the heat losses decrease drastically by just insulating the wall. As a result of this, it is possible to discharge an additional 923MWh from the caverns and also to reduce the amount of energy needed to charge the cavern, both from the primary and the secondary circuits. With insulated caverns the secondary circuit would not even be needed as the presumed cost would be far higher than the value it provides. These 923MWh are equivalent to 86m3of oil. With a inner surface area of 9009m2for both caverns and a price of 340SEK/m2 _{for 10cm of insulation, it would cost 3,06MSEK to insulate the cavern wall [13]. With an oil} price of 8000SEK/m3_{[16] the payback time of the insulation is approximately four and a half years. However,} this is not adjusted for emptying both caverns of water, filling them with water and then charging them partly. The time to reach maximum charge for the caverns is not as seasonally dependent as expected, but one can see that for the winter case it takes a dip in February, the coldest month in this area. It could also be more useful to just use the mean for the summer and winter simulation to get a more linear charging. The cavern starts at a negative amount of energy due to the initial conditions of the simulation.

The lifetime analysis shows that the mean energy discharged after a 30 year life span is ≈ 107GWh. This is of course not calculated with the different breakdown probabilities for the different components relating to the cavern. By doing a deeper analysis of these, a better estimate for the lifetime could be made. By also using a statistical model by correlating the discharge pattern to the outdoor temperature a wider range of discharge patterns could be made and provide a better estimation for the lifetime analysis.

As mentioned, the heat exchanger for the primary circuit is modeled with a variable area to get a constant effectiveness. In reality, this means there are several heat exchangers in parallel to get this variable area and ensure a constant effectiveness. But as expected, the investment cost will be far higher for this multiple heat exchanger solution compared to a single heat exchanger. Therefore a cost benefit analysis of the heat exchanger/exchangers choice is recommended to get a better understanding of the costs and how it would impact the operation of the cavern.

Because the implementation of the system only has one input for the charging and discharging pattern it could be expanded to include two input variables for the charging and discharging. In this way when the cavern is completely empty but operating on discharge mode it could instead switch to charging mode. This could be solved by using an optimization algorithm that finds the charging and discharging pattern with the highest reward, or by refining the control system and using e.g auto regressive models to estimate the oil consumption in the future and how the available energy evolves to predict whether to start charging or discharging.

The caverns show great promise in reducing the CO2emissions. In 2017 and especially in 2018 oil consumption would be reduced by a lot. During their expected lifetime operating, the caverns would remove on average a total of 29 ktonne CO2 by not burning 9968m3 of oil. It would also remove on average a total of 88,74 tonne NOx.

In Sweden there exist a grant program for reducing CO2emissions that the Swedish Environmental Protection Agency supervises. In this program housing cooperatives, municipal companies and individuals can apply for a grant. According to their website the mean reduction of CO2per invested SEK is 2,18. Multiplying this with the removed CO2we get a grant worth ≈ 13,4MSEK [17].

According to Andreas Johansson at Boliden R¨onnsk¨ar, around 2.5MW additional waste heat was made available in the fall of 2019 [16]. This has not been investigated in this report as this new solution has only been available for a few months. Currently the district heating network in Skelleftehamn is not connected to the district heating network in Skellefte˚a city. If instead the caverns would discharge depending on the load in the Skellefte˚a district heating network, these networks could maybe be connected. This would increase the available waste heat for the Skellefte˚a city network and thus decreasing the operation of the CHP plant and reducing operation costs. Previous studies have shown that it is important to charge and discharge the cavern from different places in the cavern as this will affect how much energy is stored in the cavern [4]. Implementing this in the rock caverns at N¨asudden is relatively easy due to the layout of the caverns, as there already exists a channel at the bottom of the caverns and a hole at top, both being at the opposite end of each other. This can be seen in Appendix C.

6

Conclusions

The model performs as expected, when the temperature raises in the hot part the heat losses increase. The temperature of the internal rocks stabilizes around the cold volumes temperature, which in turn stabilizes just above the return temperature. The heat losses Qloss,h,rockand Qloss,ir,rock stabilize around 1000kW, while Qloss,c,rockis around 20kW. The charge and discharge speed of the caverns is dependent on the current condition of the caverns, the available waste heat from Boliden R¨onnsk¨ar and the temperatures from Skelleftehamn. By operating the caverns intermittently, a higher amount of energy is discharged than for the seasonal operation seen in the idea study.

Depending on how the caverns are utilized, different amounts of discharged energy are obtained. This ranges from 2224,7MWh to 7846,1MWh for the different discharging patterns. The usage also affects the efficiency of the cavern, giving the efficiency a range between 19% and 53,9%.

The time it takes to fully charge the caverns is different whether the charging starts during the summer or the winter. If it starts at the beginning of the summer it takes around 100 days, if it starts at the beginning of winter it takes around 120 days.

Insulating the cavern walls results in good savings for charging the caverns, but it also increases the dischargeable energy from the caverns, due to lower heat losses. The heat losses are reduced by a factor of 5. As noted in the discussion, this ”extra” energy would have a payback time of 4.5 years calculated with the saved oil cost. Insulating the cavern walls also means that the secondary circuit would have little usage during the year, making the assumed investment cost lower.

The caverns show promising results for the reduction of CO2 emissions as noted in section 4.1, especially compared to the seasonal storage seen in the idea study. Operating the caverns intermittently would remove a total of 29 ktonne CO2 and 88,74 tonne NOxon average for its expected lifespan of 30 years. Economically, the rock caverns have good economic potential as they would save about 80 million SEK during their lifetime just by buying less oil.

7

Future Work

The internal rock volume is calculated by assuming that its thickness is 5m out of the cavern surface. This assumption is arbitrary and it would be interesting to see how different thicknesses would affect the heat losses, as the thermal inertia would be different. Currently the internal rocks are modeled as a shell around each cavern, another approach would be to model the volume between the caverns as one entity and the volumes on the opposite sides as another entity. This approach could give the model a behavior closer to the one in the idea study. Another thing that could be investigated is if the model improves with several layers for the internal rocks, making the model more similar to a CFD simulation.

Currently it is unknown if or how much the Simulink model differs from the one in the idea study, to get a better estimation of the heat losses and if the assumption for the thickness of the internal rocks is reliable. This could be done by using the same conditions found in the idea study and implement them in the Simulink model. Unfortunately, the idea study does not provide enough information for this to be done.

If it is decided to use the caverns as storage tanks, reinforcements should be investigated in Skelleftehamn district heating network, as it is currently not rated for the caverns.

8

References

[1] Joel Bengtsson. Sveriges klimatm˚al och klimatpolitiska ramverk. https://www.naturvardsverket.se/ Miljoarbete- i- samhallet/Miljoarbete- i- Sverige/Uppdelat- efter- omrade/Klimat/Sveriges-klimatlag-och-klimatpolitiska-ramverk/ Retrived 2020-06-06.

[2] Eva Rydegran. Fj¨arrv¨armens minskade koldioxidutsl¨app. https://www.energiforetagen.se/statistik/ fjarrvarmestatistik/fjarrvarmens-koldioxidutslapp/ Retrived 2020-06-06.

[3] Sofia Viksten. V¨armelagring i bergrum p˚a Haraholmen i Pite˚a. Pite˚a, 2018.

[4] Cay ˚Asberg. Solv¨arme med s¨asongslager i Lyckebo, Utredning av v¨armef¨orluster och dimensionering av solf¨alt. Uppsala, 2011.

[5] Per Johan Svenningsson Tove Ekeborg. Ackumulatorsystem vid kraftv¨armeanl¨aggningar. 1991-07-05. [6] Robert Eklund Mohsen Soleimani-Mohseni Lars B¨ackstr¨om. EnBe – energiber¨akningar: formler,

ekva-tioner, data och diagram. 2014.

[7] SGU. Geotermi. https : / / www . sgu . se / samhallsplanering / energi / fornybar geoenergi och -geotermi/geotermi/. Retrived 2020-02-22.

[8] Theodore L. Bergman Adrienne S. Lavine Frank P. Incropera David P. Dewitt. Principles of Heat and Mass Transfer. 8 ed. 2017.

[9] Bengt Andersson et al. Computational Fluid Dynamics for Engineers. Cambridge University Press, 2011. doi: 10.1017/CBO9781139093590.

[10] Emma Marklund Maria Josefsson Gerson Silva. Id´estudie: V¨armelagring i bergrum p˚a N¨asudden i Skellefte˚a utf¨ord ˚at Boliden R¨onnsk¨ar. Lule˚a, 2019.

[11] Erik B¨acklund Midroc Milj¨oteknik. personal communication. Feb. 24, 2020.

[12] Carolina Sahl´en Katja Uusitalo Marie Eriksson Daniel Persson. MKB f¨or avslutande av MKB f¨or Skelleft-ehamn. ”2005-08-10”.

[13] Stefan Strandberg Thermologic. personal communication. May 15, 2020.

[14] Mathworks. Simulink. https://se.mathworks.com/help/simulink/ Retrived 2020-05-24.

[15] Jernkontoret. Jernkontorets Energihandbok - Eldningsolja. https://www.naturvardsverket.se/Stod-i-miljoarbetet/Bidrag/Klimatklivet/Resultat-for-Klimatklivet/ Retrived 2020-06-02.

[16] Andreas Johansson Boliden AB. personal communication. May 6, 2020.

[17] Nanna Wikstr¨om. Resultat f¨or Klimatklivet. https://www.naturvardsverket.se/Stod-i-miljoarbetet/ Bidrag/Klimatklivet/Resultat-for-Klimatklivet/ Retrived 2020-06-02.

A

Cavern geometry

The geometry of the cavern was calculated with the following equations. These are used to calculate the length of the cavern and calculate the surface area of the caverns arc. The basis for this calculation can be seen in figure A.1, were c is the width of the cavern and h is the height of the cavern arc.

Figure A.1: The circular segment of a circle confined between a secant/chord and the arc s [18].

To calculate the area for the circular segment equation 1 is used to calculate the circles radius and then equation 2 is used to get the arcs surface area. This is then used with basic geometry to get the length of cavern.

R = c 2 8h+ h 2 (1) A = R2 arcsin c 2R− c 2R r 1 − c 2 4R2 ! (2)

To calculate the surface area of the arc, the length of s is calculated with equation 3 which is then multiplied with the length of the cavern.

s = 2R arcsin c