Integrated planning of electricity, gas and heat supply to municipality

8/13/2019 Integrated planning of electricity, gas and heat supply to municipality

1/18


2/18

110 B. H. BAKKEN et al., Integrated planning of electricity, gas and Strojarstvo55 (2) 109-126 (2013)

Nomenclature/ Nomenklatura

Parameters/ Parametri

fkt A - Constraint coefficients for component k in timestep t/ Koeficijent ogranienja za

komponentu k i vremenskom trenutku t

ft b - Restrictions on resources/capacities in timestep t/ Ogranienja na resurse/ kapaciteteu vremenskom trenutku tinvd c - Investment cost for investment d [Euro]/ Investicijski trokovi za investiciju d [Euro]

COP - Coefficient of performance/Faktor hlaenja - Annual discount factor/ Godinja diskontna stopa; ( )r += 1/1

C - Carnot power factor/ Faktor snage Carnotovog procesa [-]

Cls - Power factor for cooling supply s/ Faktor snage izvorahlaenja s [-]

d - Lifetime of investment alternative d/ ivotni vijek investicijske alternative d [years]Cl

lt L - Cooling load at load node l in timestep t/ Rashladnooptereenje za optereenje l u

vremenskom trenutku t [MWh/h] c

- Carnot-efficiency-factor/ Stupanj efikasnosti Karnotovog procesa

Cll - Connection loss factor for cooling load l/ Faktor gubitka prikljuka za rashladno

optereenje lCh p

- Connection loss factor for Chiller p/ Faktorgubitka prikljuka za hladnjak p

start - The first year in the first timestep in the planning horizon/ Prva godina prvog

vremenskog koraka za period planiranja end - The first year in the final timestep in the planning horizon/ Prva godina zadnjeg

vremenskog koraka za period planiranja step - The number of years in each timestep in the planning horizon/ Broj godina u svakom

vremenskom periodu za period planiranjaPenCl - Cooling deficit penalty/Penali za manjak hlaenja [Euro/MWh]r - Interest rate/ Kamatna stopa [pu]

CSupsS - Source Type (ambient air or water) in cooling supply s/ Vrsta izvora (zrak ili voda)

za hlaenje s T0 -

Evaporating temperature in K/ Temperatura isparavanja u KTK - Condensation temperature in K/ Temperatura kondenzacije u K

CSupsTair - Air Temperature in cooling supply s/ Temperatura zrakaizvora hlaenja s [

oC]CSup

sTwater - Water Temperature in cooling supply s/ Temperatura vodeizvora hlaenja s [oC]

CSupsTreq -

Required Temperature in cooling supply s/ Potrebna temperatureizvora hlaenja [oC]

CSupsT max -

Max Air Temperature Difference/ Maksimalna razlika temperature zraka [K]w

- Weight factor for length of segments/ Teinski faktor duine segmenata [days]

Variables/ Varijable

kt c - Operating cost of component k in timestep t/ Trokovi voenja komponente k uvremenskom trenutku t

pC - Operating cost for different technologies/ Trokovi voenja razliitih tehnologija [Euro]; ologiesTechn p

opesC

- Operating cost in a given state s, period and time segment / Trokovi voenja uodreenom stanju s, periodu i vremenskom segmentu [Euro]

opesc

- Annual operating costs for state s in period/ Godinji trokovi voenja za stanje s u periodu [Euro]

* C

- Minimum net present value for period through ( end + step)/ Minimalna neto

sadanja vrijednost za period kroz ( end + step) [Euro]


3/18

Strojarstvo55 (2) 109-126 (2013) B. H. BAKKEN et al., Integrated planning of electricity, gas and 111

Cllt

D - Cooling deficit in cooling demand in timestep t/ Nedostatak hlaenja u vremenskom

trenutku t [MWh/h]

- Rest value of investments/ Vrijednost ostatka investicije [Euro]

d I - Binary variable that identifies investments. d I = 1 if the investment Dd has been

carried out in period, and d I = 0 otherwise./ Binarna vrijednost koja identificira

investicije. d I = 1 ako je investicija d D provedena u periodu a inae d I = 0.scrapd I - Binary variable that identifies scrapping of equipment.

scrapd I = 1 if the equipment

from project Dd has been scrapped in period, and 0 otherwise./ Binarnavarijabla koja identificira otpisivanje opreme,scrapd I = 1 ako je oprema projekta

Dd bila otpisana u period, 0 inae

- Identifier for investment periods given as first year in each period/ Identifikator

investicijskog perioda danog kao prva godina u svakom periodu Elst P

- Electric energy input to cooling supply s or Chiller p/ Elektrina energija za radhladnjaka p [MWh/h]

Ld ijt P

- Load_flowijt: Energy flow from network node i to load node j in timestep t/Energetski tok iz mrenog vora i u vor optereenja j u vremenskom trenutku t[MWh/h]

Locijt P

- Local_flowijt: Energy flow from supply node i to load node j in timestep t/ energetskitok izopskrbnog vora i u vor optere enja j u vremenskom trenutku t [MWh/h]

N N ijt P

2

- Net2net_flowijt :Energy flow between network nodes i to j in timestep t/Energetski

tok od mrenog vora i do mrenog vora j u vremenskom trenutku t [MWh/h]Sup

sit P - Supply_flowijt: Energy flow from supply node i to network node j in timestep t/

Energetski tok od opskrbnog vora i do mrenog vora j u vremenskom trenutku t [MWh/h]

Q0 - Evaporation Heat (cooling power)/ Toplina isparavanja (rashladna snaga) [MWh/h]

QK - Condenser Heat/ Toplina kondenzacije [MWh/h]

S -

State identifier/ Identifikator stanja; StatesS t - Index for timesteps (e.g. hours) within operational model/ Indeks vremenskog koraka

(npr. sati) u operacijskom modelu, stepsTimet _

- Index for years within an investment period/ Indeks godina u investicijskom periodu,step

,,1 Clst U - Use of cooling energy at supply point s in timestep t/ Koritenje energije za hlaenje

u toki opskrbe s u vremenskom trenutku t [MWh/h]Ch pt U

- Supply of cooling energy from Chiller p in timestep t/ Opskrba sa energijom zahlaenje iz hladnjaka p u vremenskom trenutku t [MWh/h]

Heat pt U

- Supply of excess heat from Chiller p in timestep t/ Opskrba vikom topline izhladnjaka p u vremenskom trenutku t [MWh/h]

W - Input work (Carnot cycle)/ Ulazni rad (Carnotov ciklus) [MWh/h]W el - Compressor work/ Rad kompresora [MWh/h] S - Entropy difference/ Razlika entropija [J/K]

kt x - Decision variable for component k in timestep t/ Varijabla odluke za komponentu k uvremenskom trenutku t

- Index for load segments within a year/ Indeks za segmentoptereenja u godini

Sets/SetoviCold_loads - Set of cooling loads/ Set rashladnihoptereenja D - Set of investment alternatives/ Set investicijskih alternativaEmissions - Set of (predefined) emission types; Emissions = [CO2, CO, NOx, SOx]/ Set

(unaprijed odreenih) vrsta emisija; Emisije = [CO 2, CO, NOx, SOx]


4/18


Load_points - Set of load and market nodes/Set vorova optereenja i trinihoptereenja Net2load - Set to define connections between network nodes and load nodes/ Set koji definira

spojeve meu mrenim vorovima i vorovima optereenja Net2net - Set to define connections between two different networks/ Set koji definira spojeve

izmeu dvije razliite mree

Network_nodes -

Set of network nodes/ Set mrenih vorova Periods - Set of investment periods/ Set investicijskih periodaStates - Set of system states (alternative system designs)/ Set stanja sustava (alternativnih

oblika sustava)Segments - Set of load levels within a year/ Set razinaoptereenja u godiniSupply2load - Set to define direct connections between supply nodes and load nodes/ Set koji

definira izravne veze izmeu vorova opskrbe i vorova optereenja Supply2net - Set to define connections between supply nodes and network nodes/ Set koji definira

veze izmeu vorova opskrbe i mrenih vorova Cold_supplies - Set of supply points for cooling/ Set toaka opskrbe za hlaenje Technologies - Set of technology modules contributing to the object function; Technologies =

[Cold_supplies, Cold_loads, Chiller, ...]/ Set tehnolokih modula koji doprinosefunkciji objekta; Tehnologije = [Cold_supplies, Cold_loads, Chiller, ...]

Time_steps - Set of hours in the operating model, typically [1, ..., 24]/ Set sati u operativnommodelu, uobiajeno [1,,24]

Optimal expansion planning in general energy systemsis traditionally performed with large scale optimisationtools for regional or global system studies likeMARKAL/TIMES, EFOM, MESSAGE and similarmodels ([13]-[16]). In such large scale tools the energysystem is typically represented with an aggregated typeof modelling with energy balances per energy carrier,

and with resources deployed on one side and end useextracted on the other side. Various technologies aremodelled with emissions and energy losses.A much higher level of detail is required for expansion planning of local energy supply systems with multipleenergy supply options. Various conversion technologiesand infrastructure alternatives within the geographicalarea of concern have to be optimized together toquantify the complementarities and differences between possible combinations of solutions. Geography,topology and timing are all key elements in local energy planning. It is not only a question of which energyresources and which amounts to use, but alsowhere in

the geographical system the investments take place.2. The eTransport ModelThe optimization model "eTransport" is developed forexpansion planning in local energy systems whereseveral alternative energy carriers and technologies areconsidered simultaneously ([17]-[19]). The model usesa detailed network representation of technologies andinfrastructure to enable identification of singlecomponents, cables and pipelines while calculating theoptimal hourly operation of a given energy system.Then an optimal expansion plan for the energyinfrastructure is calculated over a planning horizon ofseveral decades. The approach is not limited tocontinuous energy transport like lines, cables and

pipelines, but can also include discrete transport by sea,road or rail.

2.1 Model overvieweTransport is separated into anoperational model (energy system model) and aninvestment model , asillustrated in Figure 1 [17]. In the operational modelthere are sub-models for each energy carrier and forconversion components. The planning horizon isrelatively short (1-3 days) with a typical time-step ofone hour. The operational model finds the cost-minimising hourly operation for a given infrastructureand for given energy loads. Annual operating costs fordifferent energy system designs are calculated bysolving the operational model for different seasons (e.g. peak load, low load, intermediate etc.), differentinvestment periods (e.g. 2-5 year intervals) and relevantsystem designs. Annual operating and environmentalcosts for all different periods and energy system designsare then used by the investment model to find theinvestment plan that minimises the present value of allcosts over the planning horizon.


5/18


Figure 1. Combination of operation and investmentoptimization in eTransport

Slika 1. Kombinacija optimizacije vo enja i investicije ueTransportu

The sub-models for different components are connected by general energy flow variables that identify the flow between energy sources, network components fortransport, conversion and storage, and energy sinks likeloads and markets. The connections between these arecase-specific as defined by the user. The modules areadded together to form a single linear optimisation problem where the objective function is the sum of thecontributions from the different modules, and therestrictions of the problem include all the restrictionsdefined in each of the modules. This is formulated as ageneral minimization problem:

=t k

kt kt ope xcC min (1)

subject to

ft kt fkt k

b x A , [ ]m f ,,1 , t

ft kt fkt k

b x A = , [ ]nm f ,,1+ , t

The minimum operational cost of Eq. (1) is aggregatedover different seasons (Segments) to give a minimumannual cost for each alternative system design (state),see also Section 2.3. Emissions are calculated from asubset of components that are emitting CO2, NOx, CO

and SOx (typically power plants/CHP, boilers, road/shiptransport etc.). Further environmental consequences can be implemented. Emissions are accounted for asseparate results, and when emission penalties areintroduced by the user (e.g. a CO2 tax) the resultingcosts are included in the object function and thus addedto other operating costs.The task for theinvestment model is then to find theoptimal set of investments during the period of analysis, based on the investment costs for different projects andthe annual operating costs for different periods andstates. The user defines a set of investment alternativeswhere each alternative consists of one or more physicalcomponents with predefined connections to the rest ofthe energy system. The same component can beincluded in several competing investment alternatives,thus making the alternatives mutually exclusive. Suchalternatives are identified by the model in the search forthe best expansion plan, but the user can further restrictthe solution both with respect to timing and combinationof alternatives. The optimal investment plan is found asthe plan that minimises the discounted present value ofall costs in the planning period, i.e. operating costs plusinvestment costs minus the rest value of investments.The investment algorithm is briefly outlined in Section2.3, while a more detailed documentation can be foundin [18].The combined operational and investment analysisenables a very flexible time resolution as illustrated inFigure 2. The user specifies the hourly profiles of pricesand loads for one or more Days , which are aggregated

into one or more seasonalSegments (e.g. winter,summer, spring and autumn). The sum of the Segmentsequals oneYear , which is the basis for the results fromthe operational analysis. Yearly values of costs andemissions are input into the investment analysis, whereone or more years define an Investment period (wherethe model is allowed to make investments). The sum ofall investment periods, which do not have to be of equallength, is thePlanning horizon of the case. In principleall energy sources can be implemented in the model:Primary fuels (coal, gas, oil, biomass etc.), renewableenergy sources (wind, hydro, solar, wave etc.) andsecondary energy supply (electricity, heat etc.).


6/18


Figure 2 . Time resolution in eTransportSlika 2. Vremenska rezolucija u eTransportu

Sources are implemented either as an hourly energy profile (e.g. in the case of hydro or solar energy), a price profile (e.g. electricity market prices) or a combinationof energy and price profiles. However, since the lowesttime resolution of the model is one hour, intra-hourfluctuations of e.g. wind and wave energy cannot berepresented.The operational model uses a combination of linear programming (LP) and mixed integer programming(MIP) to optimize the hourly operation of each possibleenergy system design. The investment model then usesdynamic programming (DP) to find the optimalinvestment plan over the planning horizon. A prototypeversion with stochastic DP to handle uncertainty infuture parameters like fuel prices and demand levels isalso available. The operational model is implemented inthe AMPL programming language with CPLEX as thesolver, while the investment model is implemented in

C++. A modular design ensures that new technologymodules developed for the operational model areautomatically embedded in the investment model. Afull-graphical Windows interface is developed for themodel in MS Visio. All data for a given case are storedin a database.

2.2 Recent model improvements

Due to the modular design of eTransport, a number ofimprovements have been implemented over the lastyears for very different technologies. In particular, themodel has been used for optimization of theinfrastructure for CO2 capture and storage where massflow was included in the network structure [20]; amixed integer module for biomass supply chains is

developed to handle time dependent drying processes[21]; and a new module for cooling networks isdeveloped [22]. In the context of this paper, someexamples of the latter have been included to illustratethe modelling structure.The cooling network module developed in a Masterthesis in cooperation with the Clausthal University ofTechnology in Germany currently has six newtechnology models [22]:

Cooling Demand: Industrial cooling or airconditioning

Cooling Supply: Source for natural cooling(cold water or ambient air)for a cooling network orsingle loads

Compression Chiller: Simplified model of a one-

level refrigeration machine Heat Sink: Condenser with heat

recovery option Cooling Network: District or local cooling

network

Cool Heat Supply: Link between coolingnetwork and heat pump

Figure 3 shows the symbolic cooling technology modelsimplemented in eTransport. The cooling networkmodels can be combined in different ways and allow formodeling for both district cooling purposes and localcooling applications. Both industrial cooling down tolow temperatures and air conditioning can be modeled.


7/18


8/18


+==net Net snn

N N nst

net Supplysii

Supist

Clst

Cls

Elst PPU P

2),(:

2

2),(:

(6)

stepsTimet supplyCold s _ , _

There are no operating costs associated with theCoolingSupply model itself since the cost for the requiredelectricity Elst P is accounted for in the electricity supplymodel.

2.2.3 Compression ChillerThe Compression Chiller is a model of a one-levelrefrigeration machine producing cooling energy in acooling agent circuit by using electricity. It is the mostdetailed model in the cooling network module. To allowfor proper work of the chiller model, the heat output point (condenser heat) always has to be connected to the Heat sink model. Both the input and the output nodes ofthe chiller belong to the set Network_nodes .

The theoretical cycle used to describe the refrigeration process is the Carnot-cycle. In the ideal Carnot-cycle,the input heat plus the used work equals the output heatsince no losses appear. Thus, the Carnot-power-factorcan be determined by the temperature difference between evaporator and condenser:

( ) 00

0

00

T T

T

S T T

S T

W

Q

K K C

=

== (7)

Detailed calculation of a vapor compression cycle withall changeable variables as described in [23] is too

detailed for a linear eTransport technology module.

Therefore, the compression cycle is not applied in themodel and the modeling of the compression chiller is based on energy flow and the temperature level inevaporator and condenser.One known input variable is the cooling energyQ0 which has to be delivered to cover the demand of theconnected cooling load. The condenser heat is equal tothe cooling energy output (evaporation heat) pluscompressor work:

elK W QQ += 0 (8)

The COP (Coefficient of Performance) determines the performance of a refrigeration machine. It is the ratio ofcooling power (evaporation heat) and compressor work.Compared to the ideal Carnot-cycle several losselements appear in a real refrigeration machine whichare summed up by the Carnot-efficiency-factor c. Bymultiplying the Carnot-power-factorc with the Carnot-efficiency-factor c the COP for a real refrigeration process can be determined by the following equation:

elC C W

QCOP 0== (9)

The flow equations for delivered cooling energyCh pt U ,

electricity consumption El pt P and condenser heat Heat pt U

for the Chiller are given in Equations (10a)-(10c),

respectively:

+=load Net l pl

Ld plt

net Net n pn

N N pnt

Ch pt PPU

2),(:2),(:

2 , (10a)

+==

net Supply pss

Supspt

net Net pnn

N N npt

Ch p

Ch pt

p

El pt PPU COP

P2),(:2),(:

21 , (10b)

+=+=load Net l pl

Ld plt

net Net n pn

N N pnt

El pt

Ch pt

Heat pt PPPU U

2),(:2),(:

2 . (10c)

stepsTimet Chillers p _ ,

With the cooling energy output required and theevaporating and condensation temperatures known, theCOP can be calculated and thus the electric powerneeded to run the compressor. Default values for theCarnot efficiency factor c dependent on system size aregiven in the model based on [24].The temperature levels in the evaporator and thecondenser are defined by the user. The input parametersare the evaporator outlet temperature (supplytemperature to load) and the condenser inlettemperature. For both temperatures four differentdefault alternatives considering different load andcondenser types are given in the model:

Evaporator outlet temperature;

Industry Supply (-18 C) Building Supply (Air System) (15 C) Building Supply (Water System) (10 C) Cooling Network Supply (low temperature

network) (7 C).

Condenser inlet temperature; Ambient Air (= Outdoor Temperature) Indoor Air (20 C) Water; once through (river or lake) (10 C) Water; recirculating (water circuit e.g. in

combination with a cooling tower) (20 C).

Both in the evaporator and the condenser a temperaturedifference is required to enable heat transfer. The

.


9/18


default value is 5 K but can be modified by the user.With the temperature differences, the evaporator outletand the condenser inlet temperature defined by the userthe evaporating and condensation temperatures arecalculated and the operating conditions for thecompression chiller are found. There are some moreaspects implemented in the model restricting theevaporating and condensation temperatures (e.g. due tochiller construction, data sheet etc.). Further details arefound in [22].The compression chiller model is the only model in thecooling module which causes emissions due to thecooling agent in the refrigeration system. Emissionscaused both by leakage and incomplete recycling of thecooling agent are considered. The two characteristicnumbers for the calculation of emissions caused by thecompression chiller applied in the model are the GWP(Global Warming Potential) and the TEWI (TotalEquivalent Warming Impact). Default values for theGWP of different cooling agents and the mass of thecooling agent dependent on the rated capacity of therefrigerator (data sheet) are given in the model. If anemission penalty is defined in the chiller model forthese emissions, the cost is added to the total operationcost.

2.3 Investment model

For each State s (layout of the system), investmentPeriod and seasonalSegment , all technologies that

contribute to operational costs (e.g. Eq. (3)) are addedtogether, Eq. (11):

=esTechnologi p

popes C C (11)

SegmentsPeriodsStatess ,, This equation corresponds to the general formulation ofEq. (1) for each Segment. If more than one Segment isdefined, the model will go to the next Segment andupdate load profiles etc. Then the operational model issolved again. When the operational model has beensolved for all segments within the year, the annualoperating costs are calculated as the sum of the dailyoperating costs multiplied by the weighing factorw forthe number of days represented by the respectivesegments, Eq. (12):

= Segmentsopes

opes wC c

(12)

PeriodsStatess , A matrix of annual operational costs for all States andPeriods is sent to the investment model. The investmentmodel uses dynamic programming to find theinvestment plan that minimizes the present value ofoperation and investment costs minus the scrappingvalue of new investments over the planning horizon.The dynamic programming formulation of theinvestment model is given in equations (13) (17) [18].

( ){ }

+=

step

start

Dd d

invd

opes I ccS C

,,1,

1* min

( )+ +++ 1* 1 S C start stepend (13)

Periods

where)

+ +=

Dd

scrapd d

d ord I I S S 1)(

1 2 (14)

[ ]

+=

d

stepend

Dd d

invd I c

end start

1;0max,

(15)

0* 1 =+ end C (16)

0= start S (17)

3. Municipal Case StudyThe municipality of Flora is located on the westerncoast of Norway. About 8000 people live in and aroundthe main town in the area defined by the system bordersof this analysis. In addition to residential consumers, thearea includes commercial and public buildings, industry, public schools, a sports arena, a hospital and an airport.It is expected that there will be a significant increase inthe number of inhabitants and activities in the coming

years, and the municipality is currently updating their plans for future energy, water and road infrastructures.

The Planning horizon of the analysis is 25 years(2010-2034), split into 5 investment periods of 5 yearseach. Investments are allowed at the start of eachinvestment period. The Norwegian Ministry of Financecurrently recommends a discount rate between 4% and6% for public investments depending on risk. Thediscount rate for the Flora case is set to 5%.To reflect the main seasonal variations in the coastal

region of Flora, three hourly profiles for electricity andheat demand for different customer groups are


10/18


established. These profiles are aggregated into threeyearly segments (see also Figure 2):Peak load of 29days, Intermediate load of 243 days and Low load of 92

days. The profiles are calculated from measured annualenergy demand in the area, disaggregated with nationalload profiles for relevant customer groups [25].

Figure 4. Example of residential heat demand profiles (150 detached houses)1 Slika 4. Primjeri profila potranje za toplinskom energijom u kuanstvima (150 odvojenih kua) 1

1 The new buildings are assumed to have an average size of 90 m2 and to be built according to the national building standard of 2008 with a totalannual energy demand of 125 kWh/m2 (Energy class C).

As an example, Figure 4 shows the residential heatdemand profiles for 150 new buildings planned as thefirst stage of a new suburb. The total area load isassumed to increase 6% per 5 years during the period ofanalysis.Initially, also existing cooling demand in commercialand public buildings was supposed to be included in theanalysis. However, this had to be omitted due to a lackof sufficient data for the demand.

3.2 Alternatives for energy supplyIn this case study, six different supply alternatives areconsidered for the municipality of Flora:

0. No alternative heating systems beyond currentelectricity and oil boilers

1. District heating (DH) grid with gas firedcogeneration2. District heating (DH) grid with gas fired boiler

3. District heating (DH) grid with biomass/pelletsfired boiler

4. Distribution of natural gas5. Distribution of low temperature heat from sea

water

In addition, two specific cases are analysed where onlyone suburb is considered for heating based on either acentralised heat pump (HP) or a gas fired boiler. Thedesign of the five main supply options is shown in more

detail in Figure 5 while an overview picture of thecomplete case model is included in Appendix 1.The heat centrals are designed with a separate 60% baseload capacity and a 40% peak load with back upcapacity for the base load only. Investment costs are based on national data and previous analyses in the area[26, 27]. Figure 6 shows how the investment costs forheat generation, distribution system and customerinstallations are split for the different alternatives. Notethat in the cases of district heating, investments costs forthe network are identical while the costs for the heatcentrals differ. The low temperature distribution grid based on sea water heat is a special concept that isalready in operation in a municipality not far fromFlora. Sea water pumped from 50 m depth with anannual temperature range of 7-11oC is exchanged withfresh water for distribution. Compared to the hightemperature district heating systems, the technicalsolution of a low temperature network operating around10 oC is very cheap, with only plastic pipes instead ofinsulated steel pipes. A disadvantage of this system isthe need for heat pumps at each customer location toraise the temperature to supply the buildings heatsystem. However, cooling can be obtained directly fromthe low temperature water with only a simple heatexchanger.


11/18


Figure 5. The five main energy supply options for FloraSlika 5.

Pet glavnih opcija za opskrbu Flore energijom

Figure 6. Investment costs for the different investment alternatives (Euro)Slika 6. Investicijski trokovi razliitih investicijskih alternativa (Euro)


12/18


The low temperature distribution grid based on seawater heat is a special concept that is already inoperation in a municipality not far from Flora. Sea water pumped from 50 m depth with an annual temperaturerange of 7-11oC is exchanged with fresh water fordistribution. Compared to the high temperature districtheating systems, the technical solution of a lowtemperature network operating around 10oC is verycheap, with only plastic pipes instead of insulated steel pipes. A disadvantage of this system is the need for heat pumps at each customer location to raise thetemperature to supply the buildings heat system.

However, cooling can be obtained directly from the lowtemperature water with only a simple heat exchanger.

3.3 Energy prices

The average energy prices as presented in Figure 7 are based on projections from existing national statistics[26], adjusted to local conditions in dialog with theregional energy company. The electricity price to endusers consists of electricity spot price plus grid tariff. No taxes are included in the prices.

Figure 7. Average end user energy prices (Euro/MWh)Slika 7. Prosjene cijene energije za korisnike (Euro/MWh)

3.4 Main results

When eTransport is run with initial assumptions asdescribed above, the resulting ranking of alternatives isgiven in Figure 8. The numbers in parenthesis in thelegend indicate the year of investment. At firstinspection, the difference between the alternatives

seems very small but the resulting annuities are not theonly factors that need to be considered before makingan investment decision. The resulting annuities can besplit into investment cost and operational cost perinvestment period.

Figure 8. Ranking of investment alternatives (shown as annuities for the whole period)Slika 8. Rangiranje investicijskih alternativa (za cijeli period)


13/18


Figure 9 shows the optimal investment timing of thealternatives. Only the gas distribution (#4) and the lowtemperature distribution (#5) alternatives are realised inthe first investment period (2010). District heating withgas boiler (#2) is realised in the second period (2015)and district heating with cogeneration (#1) in the third period (2020), while the last three investments arerealised in the final period.When investments are delayed beyond the first period,this implies that the municipality will have to use thecurrent energy supply of electricity and oil boilers for

one or more periods before it is economically feasible torealise a new design. This is however not consistentwith the municipalitys intent to combine a new energysupply system with major updates of roads andwater/sewage systems in the coming years. If allinvestments are forced in the first investment period, the price differences between them increase but the rankingorder does not change. Thus, in principle, only gasdistribution and low temperature distribution arefeasible solutions for the area.

Figure 9. Distribution of investments (shown as annuities for each investment period)Slika 9. Raspodjela investicija (prikazano za cijeli period)

Emissions of CO2, NOx, SO2 and CO for eachalternative and period are shown in Figure 10. Allalternatives involving combustion of natural gas havehigher emissions than the alternatives without gas. Sinceemission taxes are not included in this analysis, thedifferent emissions do not influence the annuities ofFigure 8 directly but it further emphasizes the advantageof the low temperature distribution solution. Note thatdirect emissions from existing oil boilers in the area are

not included due to lack of data, so the emissions fromthe existing solution (#0) do not include these.

3.5 Operational resultsDetailed operational profiles for all components in allseasons and for all system solutions can be inspected tohelp the decision maker to better evaluate the utilizationof components and search for critical bottlenecks orundesirable operational situations that are notobservable from the investment conclusions.


14/18


Figure 10. Annual emissions for each alternative and investment periodSlika 10. Godinje emisije za svaku alternativu i investicijski period

Figures 11a and 11b show one example of how the heatload is covered for one selected customer, in this casethe local hotel, during peak load and low load for thetwo most relevant alternatives. In the Low temperaturealternative (Fig. 11a) the heat pump is not dimensionedto cover the maximum load so an oil boiler is neededduring peak load hours. During low load, however, the

heat pump is sufficient to cover all heat demand. In theGas distribution alternative the gas boiler isdimensioned to additionally cover maximum load sothat no supplementary energy source is needed duringthe peak load (Fig. 11b).

Figure 11a. Heat supply to hotel in peak and low load periods for Low temp. alternative.Slika 11a. Opskrba hotela toplinskom energijom u vrnim i niskim periodima za alternativu s niskom temperaturom


15/18


Figure 11b. Heat supply to hotel in peak and low load periods for Gas distribution alternative.Slika 11b. Opskrba hotela toplinskom energijom u vrnim i niskim periodima za alternativu sa distribucijom plina

3.6 Sensitivity analysisAs part of the case study, four different sensitivityanalyses were performed: i) Gas price increased to equalfuel oil price, ii) 25% reduced investment cost fordistrict heating with gas boilers (alternative #2), iii) CO2 emission tax added and iv) changes in electricity price.The main conclusions are summarized in the followingsections.

3.5.1 Increased gas priceWhen the gas price is increased to equal the fuel oil price of Figure 6, the gas fired alternatives lose their

competitive advantage, and the Low temperaturealternative (#5) emerges as a clear winner. Referring tothe initial ranking in Figures 8 and 9, the Lowtemperature alternative is not affected by the increase ingas price and remains in the first investment period,while all other alternatives are postponed to the finalinvestment period.

3.5.2 Reduced investment cost for district heatingwith gas boilersThe district heating alternative with gas boilers (#2) hasthe lowest heat central investment of the three DHalternatives (network investments are assumed to beequal, see Fig. 6), and ranks as third in the initialinvestment analysis of Figure 8. A 25% reduction of(total) investment cost for this alternative is necessary tomake it competitive with respect to gas distribution andit is moved up from the second to the first investment period. However, 25% is a larger reduction than can beexpected by combining construction work with roadsand water.

3.5.3 Addition of CO 2 taxIf emission taxes are added in the model, the cost ofemissions will be included in the operational costs. Thismight influence the investment ranking. A CO2 tax of25.88 /tonne CO2 for fuel oil and 25.63 /tonne CO2 for natural gas (currently proposed by the Norwegianauthorities) will reduce the profitability of all

alternatives that involve use of fossil fuels. Thealternative with gas distribution (#4) remains in the firstinvestment period, but the Low temperature alternativenow emerges as a clear winner.

3.5.4 Changes in electricity priceElectricity is a dominant energy source in the area, alsowhen alternative heating systems are introduced. Thus,the total operational costs of all alternatives are quitesensitive to electricity price.In the initial ranking, the average electricity price is56.25 /MWh over the period of analysis. If the averageelectricity price is reduced to 50 /MWh and then to 47

/MWh (while price profiles are kept unchanged), theinvestments are delayed, first to 2015 then to 2020. Lowtemperature distribution and gas distribution remain thetwo most profitable ones, though.If average electricity price is increased to 65.60 /MWh,district heating with cogeneration (#1) moves up as the best alternative, both due to low electricity demand andthe ability to sell surplus electricity back to the grid.

4. ConclusionThis paper has presented the results of a case study toinvestigate different alternatives for the energy supplyto a municipality on the western coast of Norway. The

case study is performed with the optimization modeleTransport which is developed for the planning of localenergy systems with several alternative energy carriers.The optimal investment ranking of Figure 8 shows smallnumerical differences between the alternatives, butfurther considerations give important insights to enablea clearer conclusion from the study. Only distribution ofnatural gas with local boilers and low temperature waterdistribution with local heat pumps are implemented inthe first investment period. Since the municipality wantsto combine the development of the new energy supplysystem with construction work on roads andwater/sewage systems, these two alternatives are theonly realistic ones. Sensitivity analyses influence thecosts of the alternatives, but the ranking of them more


16/18


or less remains the same. Furthermore, since it is verylikely that a CO2 tax will be imposed on all use of fossilfuels in the near future, the low temperature distributionalternative emerges as the preferred solution. Addingthe possibility of the direct cooling of office buildings(which is not included in this analysis) gives furtheradvantages for this alternative.The regional energy company SFE has continued withmore detailed technical studies of the Flora area, andfinally applied to the Norwegian Water and EnergyAdministration for concession to establish a combineddistrict heating and cooling system based on sea waterheat, heat pumps and electricity boilers for peak load[28]. They received concession for such a system on the4th of February, 2011 [29].

AcknowledgementsThe authors gratefully acknowledge the support fromthe Research Council of Norway, the industrial sponsorsof the 'eTransport' model, SFE Nett and theMunicipality of Flora.

REFERENCES[1] CONNOLLY, D.; LUND, H.; MATHIESEN, B.V.;

LEAHY, M.: A review of computer tools for analysing theintegration of renewable energy into various energysystems , J Applied Energy 87 (2010), 1059-82.

[2] HENNING, D.: MODEST An Energy-SystemOptimisation Model Applicable to Local Utilities andCountries , J Energy 22 (1997) 12, 1135-50.

[3] SANDOU, G.; FONT, S.; TEBBANI, S.; HIRET, A.;MONDON, C.:Short term optimization of cogenerationsystems considering heat and electricity demands , Proc15th Power Systems Computation Conference, Liege,Belgium, 2005.

[4] GEIDL, M.; ANDERSSON, G.:Optimal power dispatchand conversion in systems with multiple energy carriers ,Proc 15th Power Systems Computation Conference, Liege,Belgium, 2005.

[5] GEIDL, M.; ANDERSSON, G.: A modeling andoptimization approach for multiple energy carrier power

flow, IEEE PES 2005 PowerTech Conference, St.Petersburg, Russian Federation, 2005.

[6] AKI, H.; MURATA, A.; YAMAMOTO, S.; KONDOH,J.; MAEDA, T. et al.:Penetration of residential fuel cellsand CO2 mitigation Case studies in Japan by multi-objective models , J Hydrogen Energy 30 (2005), 943-52.

[7] AKI, H.; YAMAMOTO, S.; KONDOH, J.; MAEDA, T.;YAMAGUCHI, H. et al:Fuel cells and energy networksof electricity, heat and hydrogen in residential areas , JHydrogen Energy 31 (2006), 967-80.

[8] BRUCKNER, T.; GROSCURTH, H.M.; KUEMMEL, R.:Competition and synergy between energy technologies inmunicipal energy systems , J Energy 22 (1997) 10,1005-14.

[9] LINDENBERGER, D.; BRUCKNER, T.; GROSCURTH,H.M.; KUEMMEL, R.:Optimization of solar districtheating systems: Seasonal storage, heat pumps andcogeneration , J Energy 25 (2000), 591-608.

[10] LINDENBERGER, D.; BRUCKNER, T.; MORRISON,R.; GROSCURTH, H.M.; KUEMMEL, R.:

Modernization of local energy systems , J Energy 29(2004), 245-56.

[11] OESTERGAARD, P.A.; LUND, H.: A renewable energysystem in Frederikshavn using low-temperature

geothermal energy for district heating , J Applied Energy88 (2011), 479-87.[12] OESTERGAARD, P.A.; MATHIESEN, B.V.;

MOELLER, B.; LUND, H.: A renewable energy scenario for Aalborg Municipality based on low-temperaturegeothermal heat, wind power and biomass , J Energy 35(2010), 4892-901.

[13] BELLER, M.: The Applications of Energy Systems Analysis to Policy and Technology Studies , Proc Int.Conf. of Energy Systems Analysis, Dublin, Ireland, 1979.

[14] GOLDSTEIN, G.A.; KANUDI, A.; LOULOU, R.: MARKAL: An Energy-Environment-Economic Model forSustainable Development , International Resources Group,

NESCAUM, Boston, USA, 2003.[15] MESSNER, S.; STRUBEGGER, M.: Model MESSAGE:

Command Line User Manual , Version 0.18, 15. Feb.2001, International Institute for Applied SystemsAnalysis, Laxenburg, Austria, 2001. Online:http://www.iiasa.ac.at/Research/ECS/docs/MESSAGE_man018.pdf, Last accessed 2011-12-05.

[16] SEEBREGTS, J.; GOLDSTEIN, G.A.; SMEKENS, K.: Energy/Environmental Modeling with the MARKALFamily of Models , In: 2001 Int. Conf. on OperationsResearch (OR 2001), editors:Chamoni, P.; Leisten, R.;

Martin, A.; Minnemann, J.; Stadtler, H. , Duisburg,Germany, 2001.

[17] BAKKEN, B.H.; HOLEN, A.T.: Energy service systems: Integrated planning case studies , Proc IEEE PES 2004General Meeting, Denver, USA, 2004.

[18] BAKKEN, B.H.; SKJELBRED, H.I.; WOLFGANG, O.:eTransport: Investment Planning in Energy SupplySystems with Multiple Energy Carriers , J Energy 32(2007), 1676-89.

[19] BAKKEN, B.H.:Planning of Distributed Energy Systemswith Parallel Infrastructures: A Case study , In:Hammonds, T.J., editor, Renewable Energy , 2009, ISBN978-953-7619-52-7.

[20] BAKKEN, B.H.; VELKEN, I.v.S.: Linear Models forOptimization of Infrastructure for CO2 Capture andStorage , IEEE Trans. on Energy Conversion, Vol. 23;3,September 2008.

[21] VAN DYKEN, S.; BAKKEN, B.H.; SKJELBRED, H.I.: Linear mixed-integer models for biomass supply chainswith transport, storage and processing , J Energy 35(2010), 1338-50.

[22] VAN DYKEN, S.: Entwicklung eines Kltemoduls frdas Programmwerkzeug eTransport zur Planung lokaler

Energiesysteme (Development of a cooling module for thecomputer tool eTransport for planning of local energysystems) , Master Thesis, Clausthal University ofTechnology, Germany, 2007 (in German).

[23] WANG, S.K.: Handbook of air conditioning andrefrigeration , 2. edition, New York: McGraw-HillProfessional, 2000, ISBN: 978-007-0681-67-5.
http://www.iiasa.ac.at/Research/ECS/docs/MESSAGE_man018.pdfhttp://www.iiasa.ac.at/Research/ECS/docs/MESSAGE_man018.pdfhttp://www.iiasa.ac.at/Research/ECS/docs/MESSAGE_man018.pdfhttp://www.iiasa.ac.at/Research/ECS/docs/MESSAGE_man018.pdf


17/18


[24] ULSETH, R.: IEA Project District Heating and Coolingin Future Buildings, Part II. Reduction of Electric EnergyConsumption and Peek Power by District Coolingcompared to Conventional Local Cooling Plants , 1999.http://www.iea-dhc.org/download/dhcv_4/02FinRepPart2.pdf Last accessed 2011-12-05 .

[25] ...: USELOAD - a Windows program for calculation ofelectrical load divided into end-use , Online:www.sintef.no/useload Last accessed 2011-12-05 .

[26] HOFSTAD, K.; RYEN, K.; STENSBY, K.E.:Kostnader for produksjon av kraft og varme. Hndbok 01/2007 NVE(Costs for production of power and heat, NVE Handbook,in Norwegian) , ISBN 978-82-410-0646-3. Online:http://www.nve.no/Global/Konsesjoner/Fjernvarme/hand bok1-07.pdfLast accessed 2011-12-05.

[27] NESSLER, M.: Analyse av alternative lsninger forutnyttelse av naturgass i en kystkommune (Analysis ofalternative solutions for utilization of natural gas in acoastal municipality) , Master thesis, NorwegianUniversity of Science and Technology, Norway, 2006 (in

Norwegian).[28] ...: Konsesjonssknad for fjernvarme i Flor (Application for concession for district heating in Flor) , SFEProduksjon AS, 16.07.2010 (in Norwegian).

[29] ...: Fjernvarmekonsesjon i medhold av energiloven - lovav 29. juni 1990 nr. 50. Meddelt: SFE Produksjon AS(District heating concession according to the Energy Actof 29 June 1990 to SFE Produksjon AS) , NorwegianWater and Energy Administration, 04.02.2011 (in Norwegian).

Appendix: The eTransport Model of FloraThe following figure shows the complete case model ofFlora in eTransport. Squares are sources and black/dark

grey circles are loads. Each physical load is split intothree segments; Peak, Intermediate and Low load periods. Existing technologies at the start of the analysisare drawn with solid lines, while alternatives (to beoptimized by the model) are dashed. The alternativeenergy sources are highlighted by grey shadows.
http://www.ieadhc.org/download/dhcv_4/02FinRepPart2.pdfhttp://www.ieadhc.org/download/dhcv_4/02FinRepPart2.pdfhttp://www.sintef.no/useloadhttp://www.nve.no/Global/Konsesjoner/Fjernvarme/handbok1-07.pdfhttp://www.nve.no/Global/Konsesjoner/Fjernvarme/handbok1-07.pdfhttp://www.nve.no/Global/Konsesjoner/Fjernvarme/handbok1-07.pdfhttp://www.nve.no/Global/Konsesjoner/Fjernvarme/handbok1-07.pdfhttp://www.sintef.no/useloadhttp://www.ieadhc.org/download/dhcv_4/02FinRepPart2.pdfhttp://www.ieadhc.org/download/dhcv_4/02FinRepPart2.pdf


18/18

Documents

Integrated planning of electricity, gas and heat supply to municipality