Analysis of a hybrid heating system with alternative control strategies

(1)

Analysis of a hybrid heating system with alternative control strategies

Glenn Stahre

Master of Science Thesis

KTH School of Industrial Engineering and Management Energy Technology EGI-2016-087

Division of Applied Thermodynamics and Refrigeration SE-100 44 STOCKHOLM

(2)

Master of Science Thesis EGI 2016:087

Analysis of a hybrid heating system with alternative control strategies

Glenn Stahre

Approved Examiner

Hatef Madani

Supervisor

Nelson Sommerfeldt

Commissioner

Bengt Dahlgren AB

Contact person José Acuña

Abstract

The potential of a hybrid heating system with the possibility to utilize both district heating and a ground source heat pump is of great interest due to the high prices for district heating in dense populated areas.

By controlling a hybrid heating system in a more cost effective manner there might be a possibility to trim the lifetime cost of the entire system. By using the different heating sources to their full strength, the ground source heat pump for space heating needs and the district heating for peak coverage and domestic hot water a more efficient heating system could be achieved.

The overall objective for this thesis is to construct and test alternative control strategies for a hybrid heating system in place at the multi-family housing cooperative BRF Artilleriberget 8 in Stockholm, Sweden. These control strategies were to be investigated from a lifetime cost and COP perspective over a time period of 18 years. Two alternative control strategies was constructed and tested by comparing them to the control strategy already in place. The simulation software TRNSYS was used to model and simulate the hybrid heating system. A model of the hybrid heating system with the same control which is used in the actual system was created and validated with measured data from the site.

By applying a control strategy where the ground source heat pump is used to cover the full space heating demand instead of only being operational during a period between 1^st September to 30^th April it was shown in this study that the total cost of the 18 simulated years could be decreased by 4%. It was also shown that the the higher load on the boreholes did not affect the temperature inside the boreholes to any high extent, instead the temperature in the borehole decreased throught the simulated time with 5 ˚C.

-2-

(3)

Sammanfattning

Potentialen för ett hybridvärmesystem med möjligheten att använda sig av både fjärrvärme och bergvärmepumpar är av stort intresse speciellt då priserna för fjärrvärme är höga i tätbebodda områden.

Genom att styra ett hybridvärmesystem på ett mer kostnadseffektivt sätt finns det möjligheter att trimma kostnader för systemet. Genom att använda de olika värmekällornas specifika styrkor, bergvärmepumparna för radiatorerna och fjärrvärmen för tappvarmvattnet, är det möjligt att systemet uppnår en högre effektivitet.

Målet med denna uppsats är att skapa och testa alternativa kontrollstrategier för ett existerande hybridvärmesystem installerat hos bostadsrättsförmedlingen Artilleriberget 8 i Stockholm, Sverige.

Kontrollstrategierna ska utvärderas ur ett perspektiv innefattande livstidskostnad och COP över en tidsperiod på 18 år. Två alternativa kontrollstrategier skapades och testades genom att jämföra dem med kontrollstrategin som installerades då värmesystemet sattes i bruk. Simuleringsprogrammet TRNSYS användes för att modellera och simulera hybridvärmesystemet. En modell av hybridvärmesystemet med samma kontroll som är installerad i det verkliga systemet skapades och validerades med uppmätt data från anläggningen.

Genom att använda en kontrollstrategi där bergvärmepumpen är aktiv hela året istället för att enbart vara aktiv under perioden mellan 1:a september till den 30:e april visade det sig att den totala kostnaden över den simulerade 18 års perioden kunde minskas med 4%. Det visade sig även att den mer omfattande användningen av borrhålen inte påverkade temperaturen i borrhålen i någon större utsträckning. Istället visade det sig att temperaturen i borrhålen minskade utöver den simulerade perioden med 5 ˚C.

-3-

(4)

Acknowledgment

I would like to thank José Acuña at Bengt Dahlgren AB for all the help and guidance. I would also like to thank Willem Mazzotti for all the help given in times of need. I would like to thank my supervisor Nelson Sommerfeldt for all the help he has given and also provide my sincerest thanks to Gunilla Dahlqvist, head of the housing cooperative Artillieriberget 8 and Kari Fougman at Gerox AB for providing measured data.

Abbreviation

ASHRAE American Society of Heating, Refrigeration and Air-Conditioning Engineers BHE Borehole heat exchanger

BRF Multi-family housing cooperative

CP Circulation pump

COP Coefficient of performance DH District heating

DHW Domestic hot water

DTRT Distributed thermal response test GHP Geothermal heat pump

GSHP Ground source heat pump

HP Heat pump

NPV Net present value

NRMSD Normalized root mean square deviation PDA Price dependent auxiliary

RMSD Root mean square deviation TRT Thermal response test

Nomenclature

cp Thermal capacity [J/(kg K)]

𝐸𝐸̇ Used energy [kWh]

𝑚𝑚̇ mass flow [kg/s]

T Temperature [°C]

Rb Thermal borehole resistance [K m/W]

Q1 Heating capacity [kW]

Q2 Cooling capacity [kW]

Qrad Radiator effect [kW]

𝑄𝑄̇1 Heating energy [kWh]

𝑄𝑄̇2 Cooling energy [kWh]

-4-

(5)

1 Introduction

In dense populated areas district heating (DH) has often proven itself to be a good solution for a tight and well connected heating network. By combining many residences, connecting them all to the same grid, and controlling the heating on a larger scale, it is possible to move the production of heat to locations where the residents will not be disturbed. It is then possible to distribute from outside city boarders instead of an individual heating solution for every house. District heating has proven to be effective but has left many residences completely dependent on its distribution. With only one heating alternative during cold months, the price takers are in an unconvinient position when the heating demand is increasing.

By also using a secondary heat source to DH in a hybrid heating system it is possible to decrease the total heating cost. District heating’s advantages, such as reliability, the possibility to handle peak heating demands, and high temperatures, should not be underestimated. However, when it comes to base loads demanding lower supplied temperatures it is not necessarily the most price efficient heating solution compared to alternative heating sources.

Ground source heat pumps (GSHP) are an interesting alternative to other heating methods or heat pumps for a hybrid heating system. When extracting heat from the ground the temperature in the ground will decrease and vice versa when heat is injected. By balancing the extraction and injection of heat the ground can work as a thermal store and could be used throughout the lifetime of the buildings' heating/cooling demand. In borehole configurations where recharging not have been applied the performance after the lifetime of the borehole occasionally have turned out to be lower than predicted (Björk, et al., 2013). With a hybrid heating system there will be an opportunity to determining when the GSHP should be operational and thus the impact on the temperature of the borehole would become more of a direct cause of a decision rather than the only option at that given moment.

With a high dependency of district heating for residences in Stockholm, Sweden, a hybrid heating solution with both a geothermal reservoir and a connection to the already vastly branched DH network might be an option to reduce the high heating costs during winter. By using a control strategy which utilizes the GSHP during the peak heating demands in the winter, a reduction in lifetime costs for the entire heating system may be achieved.

1.1 Background for GSHP

With a GSHP the heat is extracted via a borehole in the ground where the source is kept at a more stable temperature. A fluid is pumped through a pipe in the borehole and heat is transferred from the ground through the pipe to the fluid or vice versa for heating and cooling purpose respectively. Sweden in particular is a suitable place for ground source heat pump applications due to the grounds favorable heat transfer characteristics and the cold climate (Björk, et al., 2013).

The borehole and heat pump should be dimensioned specially for each site depending on the ground conditions and the heating/cooling demand. Physical constrains inside a city for boreholes might be a problem if the demand is too high or the ground conditions are not ideal for boreholes. The depth of the borehole, positioning of boreholes and other parameters are either determined numerically or analytically.

The system characteristics of the total heating or cooling system with boreholes, heat pumps, storage tanks and all other components are more time consuming and costly to evaluate. By neglecting how the total system interacts as a unit is a loss of a more comprehensive analysis, also an opportunity for an improvement of the system is missed.

ASHRAE (American Society of Heating, Refrigeration and Air-Conditioning Engineers) provides guidelines for how to dimension a borehole or a borehole field but these guidelines are only valid for a lifetime of 10 years (Philippe, et al., 2010). These guidelines are sometimes used when boreholes with a desired lifetime exceeding 10 years are dimensioned and provides in these cases incorrect dimensions.

These dimensions could provide a lower estimated first cost for the installation and drilling of the boreholes and are therefore competitive on the market. By the time the temperature in the borehole has

-6-

(7)

become too low to be utilized by the heat pump, the extra heating costs might increase and become more expensive than the savings during installation.

TRNSYS, a transient system simulation tool, makes it possible to simulate complete GSHP heating systems. TRNSYS solves transient systems iteratively during quasi-steady state conditions, meaning that the models in TRNSYS are solved during steady state conditions between each time step. In TRNSYS Simulation Studio components can be connected to each other by creating a link from the outlet of a component to an inlet of another. When modelling in such way the impact of each component can be seen more easily and alternative control system can be tested and evaluated.

1.2 Previous Work in TRNSYS for Total Systems

TRNSYS was originally developed as a simulation tool for solar energy systems and traces of this can be seen today by the number of papers where TRNSYS is used to evaluate and improve a variety of heating/cooling systems with solar. An interest for the software to be used in other application such as energy systems with a GSHP either in combination with a type of solar collector or as a standalone system has been evaluated in earlier papers.

Liu et al. (2015) stated that simulations over a time period as short as one year was not enough to determine the feasibility for a GSHP system but an operation time of ten years is enough to evaluate the system. According to Capozzo et al. (2015) the performance of a GSHP would increase if the ratio of cooling and heating for an unbalanced load would be rebalanced by implementation of auxiliary equipment such as auxiliary heater or chillers, described in a research of long term GSHP for office buildings in Italy. In a paper by Zarella et al. (2014) a long-term analysis for two ground source heat pumps over a period of 10 years concluded that the field arrangements for the borehole will have higher impact on the fluid temperature when the load profile included both heating and cooling.

The use for a GSHP to cover the total heating demand both for space heating and domestic hot water (DHW) investigated by Sebarchievici et al. (2014) showed that the coefficient of performance (COP) for the system was decreased from a value which was higher than four to between three and four when utilizing the GSHP for both domestic hot water and space heating. Kim et al. (2012) introduced a systematic method for verification of the actual performance for a water to water GSHP by comparing it to ISO standards. By modelling a total system and the use of indicators it was possible to see problems affecting the GSHP and the study showed that the performance for the actual system was lower compared to the specified ISO standard.

Combined solar collectors and GSHP has shown to be a good solution for the coverage of the entire heating demand. Using solar collectors for the domestic hot water and to recharge the borehole during the winter was the optimal design in a paper by Kjellsson et al. (2009). In this paper the performance of the model was compared to experimental measurements for one cooling day and the highest deviation for the TRNSYS simulation was 2%. A model by Xi et al. (2011) simulated a similar system under the weather conditions in Beijing for a time period of 20 years and concluded that recharging the borehole is a necessity to maintain the borehole temperature but will affect the parasitic losses of the system due to increased pumping power. In a master thesis by Ericsson (2015) the influence of control strategies and storage tanks on the system performance for a ground solar heat pump installation was investigated. It was shown in this paper that the most efficient control of such a system was to use the solar heat directly for DHW production.

-7-

(8)

2 Objectives

This thesis will investigate if it is economically motivated to utilize a hybrid heating system to cover the space heating demand for a multi-family housing cooperative BRF Artilleriberget 8 in inner city Stockholm. With hourly fluctuating electricity prices and monthly dependent prices for district heating following questions will be answered:

• How is the lifetime cost of a DH house affected by implementing a GSHP solution in collaboration with the already installed district heating connection?

• How would the lifetime cost and COP of the GSHP be affected if the GSHP is used more frequently than it is at the moment? How will the boreholes be affected by a higher heating load?

• Is there a possibility to achieve a lower lifetime cost by using intelligent control strategies or adding a third heating source such as electrical heaters?

3 Methodology

The housing cooperative BRF Artilleriberget 8 in Stockholm, Sweden, is in possession of a hybrid heating system with several thermocouples and flow meters at critical points which log values for the temperature and flow with a couple of minutes’ interval. This valuable data resource is used to achieve the objectives outlined above in the following workflow;

• A TRNSYS model is constructed and validated with measured values.

• A simulation of the hybrid system over an 18-year time period with the same control which is used in the actual system is created as a baseline scenario.

• Three alternative control strategies are tested over the same time period and compared with the baseline:

o The full space heating demand is met by district heating, the DH scenario.

o A scenario where the GSHP is used to its fullest capacity for space heating throughout the simulation time named Full HP.

o A scenario where the auxiliary space heating when the GSHP is active and the space heating when the GSHP is not active will be covered either with DH or by an electrical heater dependent on the price for respective heating source at an hourly basis named Price Dependent Auxiliary (PDA).

4 Limitations

Due to the vastness which is a full simulation of a heating system, limitations are needed to maintain a reasonable scope of the thesis. The limitations for this paper are described below.

• DHW demand has not been considered due to that it is covered by district heating and therefore is not affected by the changes in the control strategies for the GSHP.

• Hourly electricity prices for the period 1996-2013 are used, the same period is also the base for the results. By simulating for a period of 18 years the market volatility will be considered.

• Monthly DH prices and yearly subsciption fees are estimated by using the 2016 price structure by Fortum and average DH prices for the Stockholm region between the period 1996-2013.

• Hourly outdoor temperatures for the Stockholm region is regarded for the same period 1996- 2013.

• The results from this thesis are only valid for the GSHP system in place at BRF Artilleriberget 8 in Stockholm, Sweden.

-8-

(9)

5 BRF Artilleriberget 8

BRF Artilleriberget 8 is a multi-family housing cooperative in Stockholm, Sweden. On the 1^st of September 2015, they replaced their heating system from exclusively district heating to a hybrid heating system using a GSHP for space heating and DH for both domestic hot water and peak heating demands.

The system began operating the 15^th of September 2015, was turned off the 2^nd of May 2016 where only DH is used to cover heating demands, and will be restarted in the autumn of 2016 when a demand for space heating returns. Building loads and specifics about the GSHP for a calculated normal year can be seen in Table 1.

Table 1. Heating demand and new heat ratio (Fortum, 2015).

Heating demand

Total heating need 375 [MWh/year]

Domestic hot water 75 [MWh/year]

Peak space heating effect 120 [kW]

New heat ratio

DH Domestic hot water 75 [MWh/year]

DH space heating 20 [MWh/year]

GSHP space heating 280 [MWh/year]

A simplified illustration of the layout for the space heating system can be seen in Figure 1 below where the brine circuit is highlighted in blue and the water circuit in purple. A more detailed explanation of the system is described in the different parts of this chapter.

-9-

(10)

Figure 1. Layout of the heating system without DHW.

Not illustrated in Figure 1 is the tracking system which is implemented in the system. Values for critical temperatures and flows are used for the model of the heating system. These values are the foundation of the whole survey and used to explain the heating system.

5.1 Heat Pumps

To be able to deliver the required heating demand for the space heating and to cope with varying loads two heat pumps (HP) were installed. The first one called EB100 (HP 1) has a heating capacity of 60 kW and the second one called EB101 (HP 2) of 40 kW. With a total heating capacity of 100 kW, around 83%

of the peak space heating demand can be delivered by the heat pumps. Both heat pumps utilize a two- stage capacity by using two equally sized scroll compressors with on-off controls, meaning they can run on full or half capacity (NIBE, 2016). An illustration of a two cycled heat pump with two independent circuits is shown in Figure 2, in this case HP 2 40 kW.

-10-

(11)

Figure 2. HP 2 40 kW with two independent cycles and compressors.

The sizes and capacity levels for each heat pump yields four possible nominal heating outputs from the condenser side of each heat pump; 30, 30, 20 and 20 kW. The combination results in eight possible power outputs, each of which can be achieved with a single or two different compressor operations. Depending on the demand for space heating, a compressor will kick in to cover the demand with the condition to keep the running hours for the compressors at the same level to reduce the wear for one specific compressor according to a Master-Slave relationship (NIBE, 2016). The eight power levels and compressor combinations can be seen in Table 2.

Table 2. Running conditions for the heating system.

Partial power [kW] Total power [kW]

20 20

30 30

20 & 20 40

20 & 30 50

30 & 30 60

20, 20 & 30 70 20, 30, & 30 80 20, 20, 30 & 30 100

The heat pumps are connected to the boreholes on the evaporator side and with the secondary heating water loop on the condenser side. Since the heating system consist of two heat pumps of different capacity, automatic calculations of the required flows is not possible. Instead the flows are manually configured and the resulting values of these configurations will be described in chapter 5.2 and 5.3 (NIBE, 2016).

-11-

(12)

5.2 Brine Circuit

Eight 300 m deep boreholes have been installed under the front pavement outside of the building of the housing cooperative. The boreholes have varying inclination in relation to each other to maximize the distance between the boreholes (Geoborr Geoenergi AB, 2015). The boreholes are coupled in parallel and connected with the heat pumps by insulated pipes under the building. The incoming brine flow KB1 is a mix of water and 28% ethanol and is lead to the evaporator side of the heat pumps. Figure 3 below is a zoomed in picture of the brine circuit with the boreholes and heat pumps visible along with circulation pumps (CP), temperature sensors (abbreviated BT in the layout) and valves.

Figure 3. Layout of the brine circuit.

In Figure 3 three circulation pumps are visible, the circulation pump CP10 is connected to the total brine circuit and two more, CP11 and CP12 one for each heat pump, HP 1 respectively HP 2. It is worth noting that there is only one circulation pump per heat pump. The CPs are operating if any of the two compressors inside the heat pump are running thus with only one compressor operating the flow will still be divided to the two cycles of the heat pump but only half the flow would transfer heat to one of the evaporators. With both heat pumps operating the flow will be divided between them with a factor 60/40 in favor for the larger heat pump HP 1 due to power difference between CP11 and CP12. From measured data, three running conditions have been obtained for the main circulation pump CP10. These operating conditions with the power consumption for the three circulation pumps are presented in Table 3 below.

-12-

(13)

Table 3. Running conditions for the brine circuit circulation pumps (NIBE, 2016).

Operating

heat pumps Total brine flow through CP10

[m³/h]

Power consumption

CP10 [W]

CP11 [W]

CP12 [W]

HP 2 12.2 190 0 700

HP 1 12.7 200 1200 0

HP 1 & HP 2 16.8 300 700 1000

From measured data it is shown that CP11 and CP12 runs at 100% capacity and the power consumption seen in Table 3 is the value from the charts in the installation guide of the heat pumps (NIBE, 2016). The low value of CP10 indicates that the pump is controlled to just maintain the flow in the brine circuit. A larger part of the work is provided by the pumps CP11 and CP12.

5.3 Space Heating

A secondary system called VS1 (secondary heating) is utilized for space heating and consists of a water circuit between the radiators in the house, the heat pumps and a district heating connection. During periods when the heat pumps are not operational, the water in the radiator system will be circulated through a by-pass which does not include the heat pump circuit. During these periods the demand for space heating is covered by district heating where the water is pumped through a water-to-water heat exchanger connected to the DH network. The same heat exchanger can boost the temperature of the distributed water to the radiators during periods when there is a peak heating demand which cannot be met by the heat pumps. An illustration of the space heating with the by-pass highlighted can be seen in Figure 4 below.

-13-

(14)

Figure 4. Space heating distribution layout.

The by-pass seen in Figure 4 also works in such a way that when the heat pumps are operational the circulation pumps for the heat pumps will generate a pressure drop for the water in the t-pipe to the right of the by-pass (corner from the radiators). This design leads part of the flow through the active heat pumps rather than just circulate it in a single radiator circuit through the by-pass, which would be the case if no heat pump is operational. As the heating demand increases so will the pressure drop and flow in the heat pump circuit to such an extent that the flow in the heat pump circuit exceeds the flow in the circulation loop. At this point the direction of the flow in the by-pass would reverse to keep the circulation flow in the radiator circuit at a constant value. A fraction of the heated water from the heat pumps will then be lead back in the by-pass and mixed with the return water from the radiator.

The recirculation loop is supposed to be kept at a constant flow according to the limitations of the radiator system. A circulation pump VS1-CP13 is installed with the intention to maintain the flow at an even level. However, throughout the first operation year the flow has varied due to numerous reasons. For the first month the system was not calibrated correctly and as a result the flow in the circulation loop was

-14-

(15)

below the average value for the first operational period which is flow of 6.9 m³/h. In addition, an estimation error of 5% during normal operation and 10% in worst case might occur due to faulty calculated flow values by the pump (Grundfors, 2016).

Each heat pump has two circulation pumps on the space heating circuit, one for each heating cycle. These operate when the compressor for that cycle is active. Measured values of the flow for all possbile HP outputs have been obtained, and in Table 4 flows for all the possible HP operations are shown.

Table 4. Running conditions for the HP circuit.

Capacity heat pump [kW] Flow heat pump circuit [m³/h]

20 3.3

30 4

40 5.2

50 6.4

60 7.3

70 8.3

80 8.8

100 9.6

The measured value of the flow for each capacity level seen in Table 4 is the flow during operation where fluctuation in flows due to times when the compressors are starting and stopping is neglected. Power consumption for every circulation pump including CP13 for the different operation modes are shown in Table 5 below.

-15-

(16)

Table 5. Power consumption circulation pumps HP circuit.

Heat pump capacity [kW]

Power consumption CP13 [W]

Power consumption CP1HP 1 [W]

Total power consumption [W]

20 120 - - 90 - 210

30 120 90 - - - 210

40 120 - - 90 90 300

50 120 90 - 90 - 300

60 120 90 90 - - 300

70 120 90 - 90 90 390

80 120 90 90 90 - 390

100 120 90 90 90 90 480

Seen in Table 5 is that the power consumption is the same for all the conditions regardless the size of the heat pump which explains the steady increase of the flows in Table 4. The flows are set to maintain a temperature difference of 5-10 °C for the water flow in the heat pumps. (NIBE, 2016)

The set point for the provided hot water temperature to the heat pumps are given from a heating curve which is decided by the contractor. The set point for the temperature is directly linked to the outdoor temperature and can be seen in Figure 5 below.

Figure 5. Set point temperature as a function of the outdoor temperature.

Degree-minutes are calculated and are the difference of the set point temperature and the actual temperature of the water provided from the heat pumps over time. When the degree-minutes value exceeds 60 it is the signal for the heat pumps to start operate. The formula for the degree-minutes is seen in Equation (1) below.

𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑚𝑚𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = �𝑇𝑇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠− 𝑇𝑇𝑑𝑑𝑠𝑠𝑠𝑠𝑠𝑠𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑠𝑠𝑠𝑠𝑑𝑑� ∙ 𝐷𝐷𝐷𝐷𝑚𝑚𝐷𝐷_{𝑚𝑚𝑠𝑠𝑠𝑠} (1)

10 20 30 40 50 60 70

-30 -20 -10 0 10 20 30

Set point temperatrue °C

Outdoor temperature °C

-16-

(17)

The heat pump which has been inactive the longest will kick in to raise the distributed temperature, and if the degree-minutes still increase due to inadequate heating another heat pump will begin operating at 60, 120, 180, and 240 degree-minutes. When enough heat pumps are in operation the temperature of the supplied water will be higher than the set point temperature, the degree-minutes will decrease, and the heat pump that has been running the longest will shut down. Activation of DH for peak heating occurs when the value for degree-minutes exceeds 640.

In the case for BRF Artilleriberget 8 the heating curve seen at the webpage of the contractor is not the actual curve but instead the curve has been lifted due to complains from the residents. The contractor only has the legal responsibility to supply a temperature matching the heating curve directly from the heat pumps and has no responsibility for the actual temperature supplied to the building. In the TRNSYS model described in Chapter 7 the heating curve available on the webpage of the contractor will work as a base for the temperature supplied.

6 District Heating and Electricity Prices

The prices for district heating and electricity are time dependent. The price for DH is determined by factors such as legislation, taxes, the current prices for alternative heating sources and the demand from the customers (Energimyndigheten, 2011). BRF Arilleriberget 8 is connected to the district heating network provided by Fortum AB. The subscription used by Brf Artilleriberget 8 is called “Fjärrvärme Trygg” and the price is determined yearly. The customer also has the possibility to determine their own subscribed maximal effect (in kW) which is the energy used daily (in kWh) divided by 24 hours. If the maximal effect is overpassed the customer is penalized with an additional fee per kilowatt (Fortum, 2016).

Details for the 2016 DH prices are shown in Table 6.

Table 6. District heating prices for 2016.

Type Price

Maximal effect 501 [SEK/kW, yearly]

Delivered variable energy:

Jan-Mar, Dec 708 [SEK/MWh]

Apr, Oct-Nov 465 [SEK/MWh]

May-Sep 282 [SEK/MWh]

Fixed charge

0-250 MWh 0 [SEK/year]

251-1250 MWh 2044 [SEK/year]

With an increasing demand for electricity more expensive power plants needs to operate and will therefore raise the price. This is also a reason why the electricity price fluctuates at an hourly basis, due to a demand which overreaches the normal level of demand. (Energimarknadsinspektionen, 2016)

Due to the many assumptions regarding the price for DH and electricity, an estimation for a price development for the Swedish market is too complex and outside the scope of this thesis. A price development for the Swedish electricity price on 167.4% during a ten-year period (between 2001-2011) and a substantial drop the year after indicates how difficult it could be to estimate the future prices (Nordpool, 2016). With a planned shutdown for the Swedish nuclear program and no new hydro-plants, Sweden might be in a scenario where they need to import electricity in the future. The dependency on the hydro-plants and their seasonal changes might result in a shortage for the electricity production. With a

-17-

(18)

low variable price at 219 SEK/MWh at June 2016 (Bixia, 2016) an export of electricity might even be feasible. (Energimarknadsinspektionen, 2016)

A study done at Linköping University by Djuric et al. (2009) proclaims that the DH companies anticipate a lower demand per household due to an increased deployment of energy efficiency buildings but also that climate change would have an impact. In the Stockholm area, the demand is assumed to be at the same level today as in the future due to that an increased number of people and houses will compensate for the more energy efficient households. With the partly shared, partly monopolized situation for DH in Stockholm it is problematic to estimate future costs.

In this thesis the hourly electricity price throughout the period 1996–2013 with corresponding temperature for the same period will be used for an 18-year simulation. Estimated prices for DH for the period is used for the same time period.

7 TRNSYS Model

There is wide library of analytically developed components in TRNSYS. A standard library of components is included with the purchase of the software and there is the possibility to buy components from the TESS library or the entire library which consists of components not available in the standard library. The user could also construct their own components and distribute them freely. All components used in the model created are seen in Table 7.

Table 7. Description of the different types used in the model.

Description Type Borehole heat exchanger 246

Controller 2b

Forcing function 14h

Pump 3d

Pipe 31

Flow diverter 11f

T-piece 11h

Data reader 91

1-D Interpolation 81

In this chapter the inputs (values of temperatures, length, dimensions for the boreholes) for the models and how the system is modelled are described. The system is then validated against measured values from the site. It is followed by results which will provide other solutions and an evaluation of the system. The layout of the heating system without the control part is shown in Figure 6.

-18-

(19)

Figure 6. Heating system TRNSYS.

7.1 Control System

The control system used in the model uses the same method as described in chapter 5.3 with degree- minutes. Degree-minutes are calculated at every time step with equation (2):

𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑚𝑚𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 = �𝑇𝑇𝑤𝑤𝑤𝑤𝑠𝑠𝑠𝑠− 𝑇𝑇𝑠𝑠𝑑𝑑𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠� ∙ 𝐷𝐷𝐷𝐷𝑚𝑚𝐷𝐷 𝐷𝐷𝐷𝐷𝐷𝐷𝑠𝑠[ℎ𝐷𝐷] ∙ 60[𝑚𝑚𝐷𝐷𝐷𝐷/ℎ𝐷𝐷] (2) Twant is the value from the heating curve and Tsupply is the actual supplied temperature to the radiator. The data reader Type 91 provides the outdoor temperature with an hourly time step. Type 93 input value recall is used to sum up previous degree minutes and add or subtract the calculated degree-minute at every time step. Five controllers of Type 2b are used to separately decide the control signal which is dependent on the different degree-minute levels, Type 2b generates a Boolean data type either true or false. Equation (3) is the equation used for the first controller, the control signal will either be true (1) or false (0) (TRNSYS, 2014).

𝑐𝑐𝑐𝑐𝐷𝐷𝐷𝐷𝐷𝐷𝑐𝑐𝑐𝑐 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑠𝑠𝑐𝑐 = 30[𝑑𝑑𝐷𝐷𝐷𝐷𝑚𝑚𝐷𝐷𝐷𝐷] ± 30[𝑑𝑑𝐷𝐷𝐷𝐷𝑚𝑚𝐷𝐷𝐷𝐷] (3) A hysteris of 30 degree-minutes will keep the output signal (1) over the span of 0-120 degree-minutes and the signal will become true (1) first after that the value of the degree-minutes has exceeded 60 degree- minutes. This control is applied in the same manner for the other degree-minutes levels as well. For degree-minute level 120 it will provide a true value first after exceeding 120 and stay true if the degree- minutes remains in the span of 60-180 degree-minutes.

In reality the heat pump that should be turned off is determined by its running time, mentioned in Chapter 5.1. In the model certain heat pumps are more used than others and which heat pump that should run at a certain time is determined with an implemented variation signal. With eight possible heating

-19-

(20)

capacitates and four available output signals from the degree-minutes calculations a variation variable is introduced in a pre-controller. This variation is in the form of a time dependent forcing function Type 14h which connects the output signal for each controller to a specific heat pump capacity. The signal from respective controller is sent to the pre-controller and the capacity of the heat pump array is described in Table 8.

Table 8. Degree-minutes levels.

Degree-minutes Capacity heat pump [kW]

60 20 or 30

120 40, 50 or 60

180 70 or 80

240 100

640 District heating

For the degree-minutes 60 and 180 the time dependency is one hour, thus during 60 simulated minutes one of the possible heating capacities will be prioritized over the other, these conditions change for the next one-hour period. When the degree-minutes are in the span between 120 and 180 there are three possible heating capacities, 40, 50 and 60 kW. Where 50 kW is the most common one due to that it consists of two half capacities of the two heat pumps. For one hour 50 kW is prioritized and the next hour 40 respectively 60 kW is alternating with a 30 minutes’ interval. This is a generalization of the reality but simplifies the model by prioritizing certain heat pumps in the model and therefore simplifies the controllers.

The output signal from the pre-controller representing the heating capacity demanded is used by the HP circulation pump, brine pump and the diverters which control the flow. With a specific flow for both the brine and water circuit attached to a certain heat capacity according to the measured data the model becomes constrained and will be worse at adapting to fluctuations that would occur in reality.

7.2 Heat Pumps

The heat pumps, the heart of the hybrid heating system is modelled by using performance maps from the supplier Nibe to model the heating and cooling capacity for the heat pumps (NIBE, 2016). The heating and cooling capacity is derived from the incoming brine temperature and supplied water temperature. A polynomial correlation derived from the values from the performance maps was used to create an equation which represents the heat pumps in the model.

The polynomial correlations were curve fitted quadratic to the performance in similar fashion as Corberan et al. (2010) with the assumptions that with two similar cycles in each heat pump the total heating and cooling capacity could be divided by a factor of two to represent each separate compressor. A general polynomial equation is used for the performance maps for both the 60 and 40 kW heat pumps. The general equation can be seen in Equation (4):

𝑄𝑄₁, 𝑄𝑄₂= 𝐴𝐴₀+ (𝐴𝐴₁∙ 𝑇𝑇_{𝑑𝑑𝑑𝑑𝑠𝑠𝑠𝑠𝑠𝑠}) + (𝐴𝐴₂∙ 𝑇𝑇_{𝐵𝐵𝑑𝑑𝑠𝑠𝑠𝑠𝑠𝑠}²) + (𝐴𝐴₃∙ 𝑇𝑇_{𝑊𝑊𝑤𝑤𝑠𝑠𝑠𝑠𝑑𝑑}) + (𝐴𝐴₄∙ 𝑇𝑇_{𝐵𝐵𝑑𝑑𝑠𝑠𝑠𝑠𝑠𝑠}∙ 𝑇𝑇_{𝑊𝑊𝑤𝑤𝑠𝑠𝑠𝑠𝑑𝑑}) (4) In Equation (4) the constants A0 to A4 are determined by the regression analysis tool in Excel to minimize the quadratic error between the heating and cooling capacity from the performance maps values and the ones calculated. With TBrine constrained to (-5, -2.5, 0, 2.5, 5, 7.5, 10, 12.5, 15 °C) and TWater (35, 50, 65 °C) the root mean square deviation of the error (RMSD) for the capacities from the performance maps and calculated capacities is illustrated in Table 9.

-20-

(21)

Table 9. RMSD of the heat pump capacities compared to the performance maps.

Heating form

Heating capacity [kW] QHeat RMSD QCool ratio RMSD

40 0.6 0.52

60 0.6 0.72

The small values of RMSD for each heat pump and each heating form shows that the polynomial

equation provides an acceptable estimation of the performance maps. In the model all the heat pumps are bundled together into one single heat pump. The equation which represents all the heat pumps is coupled to the control determining which heat pump that should be active. Figure 7 to Figure 10 below illustrates the errors of the calculated capacities compared to the capacities from the performance map.

In the model Twater is an output from the heat pump and to use it as an input for the same TRNSYS component is not possible. Therefore, a mean value between Twant and the output temperature Twater from the heat pump at an earlier time step is used in order to prevent a runaway solution.

Figure 7. Q1 40 kW with a ± 5% deviation. Figure 8. Q2 40 kW with a ± 5% deviation.

Figure 9. Q1 60 kW with a ± 5% deviation. Figure 10. Q2 60 kW with a ± 5% deviation.

30 35 40 45 50 55 60

30.00 40.00 50.00 60.00

Qperfromance map [kW]

Qcalculated [kW]

Q1 40kW

15 20 25 30 35 40 45 50

15.00 25.00 35.00 45.00

Qperformance map [kW]

Qcalculated [kW]

Q2 40kW

45 50 55 60 65 70 75 80 85

45.00 55.00 65.00 75.00 85.00 95.00

Qcalculated [kW]

Q1 60kW

20 30 40 50 60 70

20.00 40.00 60.00

Qcalculated [kW]

Q2 60kW

-21-

(22)

7.3 Heating Distribution System

The water circuit or the heating distribution system consists of a single speed pump (Type 3d), two pipes (Type 31), two flow diverters (Type 11f), two T-pieces (Type 11h) and two inserted equations representing the heat pumps and the radiators. The hourly outdoor temperature needed in the simulation is provided in a data reader (Type 91) and interpolated between values given by the heating curve with a 1-D Interpolation (Type 81).

In the model, the entire flow is controlled by one pump instead of five as in the actual system. This pump (HPpump) has a maximum flow of 9,600 kg/h as presented in Table 4, the same flow as the maximum flow when the heat pumps are operating at full capacity. Due to misreading from the pump of the flow in the radiator circuit and uncertainties of how the flow is controlled, the flow is curve fitted as a linear equation from measured data to fit the values. As the entire system is dependent on the outdoor temperature it naturally follows that the volumetric flow rate in the radiator circuit, Radflow is also dependent on the outdoor temperature. The equation used in this estimation is illustrated in Equation (5):

𝑅𝑅𝑠𝑠𝑑𝑑𝑅𝑅𝑐𝑐𝑐𝑐𝑅𝑅 �𝑚𝑚³

ℎ � = −0.0522 ∙ 𝑇𝑇^{𝑂𝑂𝑑𝑑𝑠𝑠}+ 6.8687 (5) This estimation yielded a slightly better estimation of the flow with a RMSD at 0.72 for the equation when compared to the measured flow and 0.86 if the flow had been assumed to be constant at an average value of 6.9 m³/h throughout the simulation.

Instead of the flow being divided in the pipes by pressure differences, a flow diverter defined as a fraction between the flow from the pump and the radiator circuit flow is implanted to withhold a proper flow. The flow diverter is controlled with a value between 0-1 which determines how the fraction of the incoming flow will be divided to the two outputs (TRNSYS, 2014). The diverter, which leads the water back through the by-pass, is only operational when the flow in the HP-circuit exceeds the calculated value of the radiator flow, Radflow. In this case the diverter will lead the flow fraction above the wanted value back to T-piece 1. The heat pumps and how they affect the flow can be seen in Table 10 along with the fraction for the pump where the active heat pump is highlighted. The index for each HP represents which heat pump it is and which cycle in that heat pump.

Table 10. Fraction of the flow in the HP circuit, the highlighted cells indicates which heat pump that is operational.

HP capacity

[kW] Total flow

[kg/h] HP1,1

[kg/h] HP2,1

[kg/h] HP1,2

[kg/h] HP2,2

[kg/h] Fraction HPpump

20 3,300 0 3,300 0 0 0.34

30 4,000 4,000 0 0 0 0.42

40 5,200 0 2,600 0 2,600 0.54

50 6,400 3,200 3,200 0 0 0.67

60 7,300 3,650 0 3,650 0 0.76

70 8,300 2,767 2,767 0 2,767 0.86

80 8,800 2,933 2,933 2,933 0 0.92

100 9,600 2,400 2,400 2,400 2,400 1.00

Half of HP 1 and HP 2 is active for most of the operational time. It is because the running conditions which were described in Chapter 7.1, where the choice of which heat pump that should go operational is determined by a time dependent control. While operating, the pump will have a flow which is a fraction of the maximum flow according to Table 10, diverting with an even flow for respective heat pump.

The difference of the real energy demand of the house and the calculated demand used in the model will generate an error which is hard to optimize. The radiator in the model is described as an equation which

-22-

(23)

utilizes the energy equation to derivate the outlet temperature of the flow from the inlet temperature, heating demand and the incoming flow which can be seen in Equation (6).

𝑇𝑇𝑠𝑠𝑑𝑑𝑠𝑠,𝑑𝑑𝑤𝑤𝑑𝑑= 𝑇𝑇𝑠𝑠𝑠𝑠,𝑑𝑑𝑤𝑤𝑑𝑑− 𝑄𝑄_{𝑑𝑑𝑤𝑤𝑑𝑑}

𝑚𝑚̇𝑑𝑑𝑤𝑤𝑑𝑑∙ 𝑐𝑐𝑠𝑠𝑤𝑤𝑤𝑤𝑠𝑠𝑠𝑠𝑑𝑑 (6)

Where the desired heating demand Qrad is estimated by an energy signature from earlier heating bills and is dependent on the outdoor temperature, the function for Qrad is seen in Equation (7).

𝑄𝑄_{𝑑𝑑𝑤𝑤𝑑𝑑}= −5.5 ∙ 𝑇𝑇_{𝑠𝑠𝑑𝑑𝑠𝑠}+ 58.5 (7)

No simulation of the entire housing cooperative is performed. This will generate a modulation error due to the complexity that the house and its thermal capacity provides to the heating system.

7.4 Pipes

Four pipes (Type 31) are used in the model to provide thermal inertia to the system, by using pipes with enough volume a simulation closer to the reality can be achieved. Due to insufficient info regarding the systems pipe lengths and thermal conductivity the sizing of the pipes is estimated for the heating distribution side. The volume of the heating system is estimated using a method for dimensioning expansion tanks (Soma Therm, 2016). By extracting the heating demand (115 kW) at dimensioning outdoor temperature for Stockholm of -18°C a volume of 1.5 m³ is estimated for the piping.

For the borehole loop the borehole protocol provided by the drilling company report detailed values about the volume of fluid in the system. 5,240 liters in total, 4,800 in the boreholes and 440 liters fill the pipes in the brine circuit (Geoborr Geoenergi AB, 2015).

The pipes in the borehole circuit are well isolated, and the heat loss in pipes are already considered in the total heating demand provided by the energy signature. The total loss for the pipes in the model is thus set to zero. The length of the pipes is half the length needed to sustain the total volume. Length and number of the pipes are shown in Table 11.

Table 11. Pipe sizing in TRNSYS.

Description Water pipe 1 Water pipe 2 Pipe B1 Pipe B2

Diameter [mm] 65 65 80 80

Length [m] 226 226 44 44

7.5 TRNSYS component borehole heat exchanger Type 246

The BHE types available in TRNSYS are limited. In the standard library there are no BHE type. Instead it is possible to purchase an add-on called Geothermal Heat Pump (GHP) which consists of numerous types of BHE from the TESS library. For more information about the use and performance of the BHE types in TRNSYS see (Thorén, 2016). Thorén (2016) evaluated the available BHE in the Geothermal Heat Pump add-on. For multiple boreholes there were three models available at the time, Type 557a, Type 557b and Type 246. In this thesis the BHE Type 246 is used.

Type 246 is a borehole type developed at École Polytechnique de Montreal. The model is not yet published and therefore not available on the market. Type 246 has the possibility to simulate for multiple boreholes in varying geometries and also the possibility to have one or two U-tubes inside the borehole.

The g-functions describing the geometry of the borehole field can be attached as an external file providing the opportunity to model non-uniform fields, which is a limitation in Type 557.

The load scheme consists of three blocks, one large, one medium and one small which contains the load and time period presented by (Liu, 2005). The idea is that the fluctuations of the current load have a

-23-

(24)

higher impact on the model than the past load therefore the past load is packed into larger and medium sized blocks to reduce computing time. The large block stretches over the longest time period similar to the ASHRAE method which uses the same load scheme to reduce the simulation time (Philippe, et al., 2010). The heat extraction/injection at specific moments are added to a small block, small blocks are then added in medium blocks and medium blocks in large blocks. This load scheme is set in the parameter properties of the components (Godefroy, 2014). The g-function used for these thesis was generated in the software GSD provided by José Acuna and Willem Mazzotti at Bengt Dalhgren AB, Stockholm.

Type 246 has some aspects that do not fully match with Swedish standards. Adjustments were made on the grouting side of the boreholes. Grouting boreholes with materials such as cement and bentonite-based materials is currently not common in Sweden. Instead they use the natural ground water as a filling in the borehole and will therefore not necessarily represent constant values for the equivalent thermal conductivity at every depth, due to possible groundwater convection caused by temperature dependent water density (Acuña, 2013).

For boreholes in the Swedish climate, the thermal resistance Rb is more preferably to be used as an input than the thermal conductivity of the grout. Thermal response tests (TRT) and distributed thermal response tests (DTRT) are ways to obtain Rb for a specific borehole and commonly used during borehole installations. Values for the borehole resistance are most common in the span 0.06-0.12 K m/W (Acuña, et al., 2013). In type 246, Rb for the borehole is calculated and therefore an output from type 246 but can be tuned by adjusting the thermal conductivity of the grout to maintain the borehole resistance at a proper value.

By keeping the grout thermal conductivity at a constant value of 1 W/mK (a typical value for still water is 0.58 W/mK (Toolbox, 2016)) to compensate for the natural convection of the groundwater a thermal resistance of the borehole of 0.085 K m/W is fixed. This value was obtained iteratively and the temperature from the borehole model was validated and used in the simulations further ahead by being compared to measured values. Input values used in the model for Type 246 is shown in Table 12.

Table 12. Input values for Type 246.

Value Unit Comment

Borehole characteristics

Nr. Of borehole 8 (Geoborr Geoenergi AB, 2015)

Borehole active length 293 m -

Borehole buried depth / Distance

to groundwater 7 m -

Borehole Radius 57.5 mm -

Outer radius u-pipe 20 mm -

Inner radius u-pipe 17.6 mm -

Shank spacing 36 mm Assumed

Initial ground temperature 11 °C Observed temperature out from borehole during first operation day

Ground properties

Ground thermal conductivity 3 W/mK (Björk, et al., 2013) Ground thermal capacitance 2,000 kJ/(m³/K) Assumed

Grout thermal conductivity 1 W/mK Interpolated value

-24-

(25)

Grout thermal capacitance 4,200 kJ/(m³/K) Table value for water Pipe thermal conductivity 0.42 W/mK Table value for PE

Pipe roughness 0.01 mm Assumed

Brine

Ethanol 28%

Water 72%

7.6 Borehole Circuit

The borehole circuit consists of the borehole model Type 246, a single speed pump (Type 3d), a flow diverter (Type 11f), two pipes (Type 31), a t-pieces (Type 11h) and a connection to the heat pump equation. The fluid which is a mix of 28% ethanol and 72% water has different characteristics than water.

This has not been considered in the implementation of the physical pump which would not have any effect on the flow in m³/h but with a density of 945 kg/m³ instead of water (1000 kg/m³) it would affect the mass flow in kg/h. How the heat pumps affect the flow and the fraction for the pump as well as for the diverter is illustrated in Table 13.

Table 13. Fraction of the flow in the borehole circuit, the highlighted cells represent active heat pumps.

HP capacity

[kW] Total flow

[kg/h] HP1,1

[kg/h] HP2,1

[kg/h] HP1,1

[kg/h] HP2,2

[kg/h] Fraction Brine

pump Fraction

diverter

20 11,529 0 5,765 0 5,765 0.73 0.5

30 12,002 6,001 0 6,001 0 0.76 0.5

40 11,529 0 5,765 0 5,765 0.73 1

50 15,876 4,763 3,175 4,763 3,175 1.00 0.5

60 12,002 6,001 0 6,001 0 0.76 1

70 15,876 4,763 3,175 4,763 3,175 1.00 0.7

80 15,876 4,763 3,175 4,763 3,175 1.00 0.8

100 15,876 4,763 3,175 4,763 3,175 1.00 1

7.7 Heating with District Heating

At BRF Artilleriberget 8, when the heating demand is met by district heating, it is shunt controlled with a signal change for every 10^th second. Due to a simulation time for the model of two minutes, it was not possible to implement in the system. Therefore the demand for district heating is modelled in such a way that the DH maintains the temperature of the distributed water at the desired level Twant and the amount of energy provided from district heating is calculated according to Equation (8):

𝑄𝑄_{𝐷𝐷𝐷𝐷}= 𝑚𝑚̇_{𝑓𝑓𝑠𝑠𝑠𝑠𝑤𝑤}∙ 𝑐𝑐𝑠𝑠_{𝑤𝑤𝑤𝑤𝑠𝑠𝑠𝑠𝑑𝑑}∙ (𝑇𝑇_{𝑤𝑤𝑤𝑤𝑠𝑠𝑠𝑠}− 𝑇𝑇𝑑𝑑𝑠𝑠𝑓𝑓𝑠𝑠𝑑𝑑𝑠𝑠 𝐷𝐷𝐷𝐷) (8) This equation is used for both the auxiliary heating with DH as well as heating with DH when the heat pumps are not operational. These two conditions are followed by their own mass flow rates according to Equation (5) which will yield different heating demands.

7.8 Validation of TRNSYS Model

The model created was compared to measured data from the site for key measurements in order to test the viability of the simulated values over a total time of 4,390 hours from the 10^th of October 2015 to the 31^th of Mars 2016. This time-span leaves out the first period of operation from 15^th of September to 30^th of September, which were the first 16 days due to error-prone or non-existent measured data. The time

-25-

(26)

step used in the simulation is two minutes. A lower simulation time would be preferred to ensure convergence but a smaller simulation time interfered with the convergence of the borehole model Type 246. Due to some measurements (such as the heating capacity) are logged at an interval of 10 minutes the measured values gathered for that interval are compared with the simulated values gathered with a 10 minutes interval as well. The key evaluated measurements are described in Table 14 below.

Table 14. Key measurements monitored during this study.

Key measurements Si-unit Heating capacity Q1 kW Cooling capacity Q2 kW Temperature into borehole °C Temperature out of borehole °C Temperature to radiator °C Temperature from radiator °C Total produced heating kWh Total produced cooling kWh

Normalized root mean square deviation (NRSMD) is used to evaluate the model. With the model being a simplification of the true site and that the actual heating system is heavily dependent on time due to the degree-minute control, the simulated values are often out of sync with the measured values. For example, the temperature to the radiator floats over and under the dimensioned temperature causing the root mean square deviations to increase. An average value of six measured and modelled values (a time period of one hour) was used for the NRMSD which are calculated as Equation (9).

𝑁𝑁𝑅𝑅𝑁𝑁𝑁𝑁𝐷𝐷 = �∑ (𝑦𝑦�^𝑠𝑠− 𝑦𝑦_𝑠𝑠)²

𝑠𝑠 𝐷𝐷

𝑠𝑠=1

𝑦𝑦𝑚𝑚𝑤𝑤𝑚𝑚,𝑚𝑚𝑠𝑠𝑤𝑤𝑠𝑠𝑑𝑑𝑑𝑑𝑠𝑠𝑑𝑑− 𝑦𝑦𝑚𝑚𝑠𝑠𝑠𝑠,𝑚𝑚𝑠𝑠𝑤𝑤𝑠𝑠𝑑𝑑𝑑𝑑𝑠𝑠𝑑𝑑

(9)

The NRMSD is expressed as a percentage over the total time period and will provide a smaller value due to that the range of the values have been taken out of the comparison which results in a smaller goodness of fit for the values. The NRSMD for the key measurements is seen in Table 15.

Table 15. NRMSD for key measurements.

Key measurements NRMSD [%]

Heating capacity Q1 12.7 Cooling capacity Q2 12.8 Temperature into borehole 6.5 Temperature out of borehole 4

Temperature to radiator 5 Temperature from radiator 7.7

In Table 15 it is visible that the temperatures to and from the radiator is differing. This is a combined result of the uncertainties of the real flow to the radiators and due to that the volume of water transported in the radiator circuit does not match the real volume. As the whole system is dependent on the supplied

-26-

(27)

temperature to the radiators it results that the heat pumps will run at different times to supply the temperature needed, which will not match the times the heat pumps are running from the measured values.

Also the minimum runtime of the heat pumps is assumed to be automatically controlled with the degree- minutes control. The supply temperature is increasing during times when the degree-minutes are rising due to a constant calculation in TRNSYS, which yields a capacity change in times when there is no need for it. And with the iterative nature in TRNSYS all capacity changes are occurring during a time step, (two minutes). This might not be an optimal or even possible time difference for a heat pump to be operational.

In conclusion, the facts above result in an elevated error for the modelled heating and cooling capacity which in turn affects the temperatures in and out of the borehole. This is investigated more thoroughly over a larger time span in order to estimate the error from the model. In Figure 11 the average brine temperature for one day is illustrated.

Figure 11. Simulated and measured average brine temperatures for one day.

The brine temperatures, which are the averages of the temperature into and from the borehole, are correlated with a value of 0.99. Seen in Figure 11 is that the error is elevated during times when the load on the borehole is at its highest. This does not seem to affect the overall average temperature of the borehole due to the short duration. In Figure 12 and Figure 13 below, the average value of the heating and cooling capacity over one day throughout the inspected time is investigated.

0 5 10 15 20 25 30 35 40

0 1 2 3 4 5 6 7 8 9

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162 169 176

Error [%]

Brine temperature [°C]

Days

Tbrine Measured Tbrine TRNSYS Error

-27-

(28)

Figure 12. Simulated and measured heating capacities and error for Q1 with an average value for one day.

Figure 13. Simulated and measured cooling capacities and error for Q2 with an average value for one day.

Seen in Figure 12 and Figure 13 is that the simulated average heating and cooling affect is in sync with the measured average capacities with correlated values of 0.96 for Q1 and 0.93 for Q2. It can also be seen that the high error for the brine temperature seen in Figure 11 is due to the higher value of Q2 seen in Figure

0 10 20 30 40 50 60 70 80 90 100

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162 169 176

Error [%]

Average heating capacity (Q1) [kW]

Days

Q1 Measured Q1 TRNSYS Error

0 10 20 30 40 50 60 70 80 90 100

0 10 20 30 40 50 60 70

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162 169 176

Error [%]

Average cooling capacity (Q2) [kW]

Days

Q2 Measured Q2 TRNSYS Error

-28-

(29)

13. So the high error for the brine temperatures occurs during time periods when the modelled Q2 is higher than the measured due to the higher load on the borehole.

In conclusion the model generates a lesser total produced energy with a total produced energy which is 5.5% less than the actual heat produced by the heat pumps on the site and a total Q1/Q2 ratio of 95%

when compared to measured values. There is still an error which indicates that the model behaves incorrectly. This error might seem high, at some few points around 30% for an average value of one day which is a result of the obstacles mentioned earlier in this chapter. However, the error is mostly around or under 15% for the heating and cooling capacity and with the high correlation and low difference in produced energy the model is accepted.

8 Results

The results consist of several charts where key values such as the COP, borehole temperature and cost of the system are illustrated for the different scenarios. The different scenarios are firstly the Baseline scenario which is when the hybrid heating system is controlled as the contractor decided. With the GSHP operational between the period from the 1^st September to the 30^th of April, DH is active during the remainder of the period and for peak space heating. The Full HP scenario is when the GSHP is controlled to be used as much as possible for the entire simulation with DH as peak coverage.

The Price Dependent Auxiliary scenario is when the auxiliary heating and the space heating are covered by either an electrical heater or DH during the period when the GSHP is not active. Which heating source that should be used is determined by which heating source is cheapest at the start of every hour. During times when the GSHP is active the COP of the heat pump from an earlier time step in the simulation is used to estimate the electricity needed thus determining which heating source should be utilized. The District Heating scenario is when the full space heating demand is covered by DH, the heating method before the hybrid heating system. The COP for the GSHP over the inspected time period is calculated according to Equation (10).

𝐶𝐶𝐶𝐶𝐶𝐶 = 𝑄𝑄̇1

��𝑄𝑄̇1 − 𝑄𝑄̇2� + 𝐸𝐸̇𝑠𝑠𝑠𝑠,𝐶𝐶𝐶𝐶� (10) The cost in SEK for the system is calculated according to Equation (11).

𝐶𝐶𝑐𝑐𝐷𝐷𝐷𝐷 = 𝐷𝐷𝐷𝐷𝑓𝑓𝑠𝑠𝑚𝑚𝑠𝑠𝑑𝑑 𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠+ 𝐷𝐷𝐷𝐷𝑚𝑚𝑠𝑠𝑠𝑠𝑠𝑠ℎ𝑠𝑠𝑠𝑠,�𝑁𝑁𝐸𝐸𝑆𝑆

𝑘𝑘𝑘𝑘ℎ� ∙ 𝑄𝑄̇ 1𝐷𝐷𝐷𝐷+ 𝐸𝐸𝑐𝑐𝐷𝐷𝑐𝑐𝐷𝐷𝐷𝐷𝐷𝐷𝑐𝑐𝐷𝐷𝐷𝐷𝑦𝑦 𝑠𝑠𝐷𝐷𝐷𝐷𝑐𝑐𝐷𝐷ℎ𝑠𝑠𝑑𝑑𝑑𝑑𝑠𝑠𝑠𝑠�𝑁𝑁𝐸𝐸𝑆𝑆 𝑘𝑘𝑘𝑘ℎ�

∙ (𝐸𝐸̇𝑠𝑠𝑠𝑠,𝑐𝑐𝑠𝑠𝑚𝑚𝑠𝑠𝑑𝑑𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑑𝑑+ 𝐸𝐸̇_{𝑠𝑠𝑠𝑠,𝐶𝐶𝐶𝐶}+ 𝐸𝐸̇𝑠𝑠𝑠𝑠,𝐴𝐴𝑑𝑑𝑚𝑚𝑠𝑠𝑠𝑠𝑠𝑠𝑤𝑤𝑑𝑑𝑠𝑠 ) (11) Where the the Electricity pricehourly is illustrated in Figure 14 and DHmontly is according to Figure 15. The DH subsciptional prices are estimated by using the prices from 2016 seen in Table 6 and then extrapolated with prices during the time period between 1996-2013 (Holgersson, 2016). The DHfixed,cost differs between the DH scenario and the rest with a maximal effect of 133 kW for the DH scenario and 33 kW for the other scenarios due to a higher usage of DH water for the DH scenario.

-29-

Analysis of a hybrid heating system with alternative control strategies