See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/327993603

The future of low carbon industrial process heat: A comparison between solar
thermal and heat pumps

Article  in  Solar Energy · October 2018

DOI: 10.1016/j.solener.2018.08.011

CITATION READS

1 201

3 authors, including:

Steven Meyers Bastian Schmitt

Universität Kassel Universität Kassel
17 PUBLICATIONS   112 CITATIONS    28 PUBLICATIONS   122 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

GREENFOODS View project

SolarAutomotive - Solar Process Heat for the Automotive Industry and its Suppliers View project

All content following this page was uploaded by Bastian Schmitt on 15 October 2018.

The user has requested enhancement of the downloaded file.

 Solar Energy 173 (2018) 893–904

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

The future of low carbon industrial process heat: A comparison between T

solar thermal and heat pumps
⁎
Steven Meyersa, , Bastian Schmitta, Klaus Vajena
a
University of Kassel, Institute of Thermal Engineering, Kurt-Wolter-Str.3, 34109 Kassel, Germany

A R T I C LE I N FO A B S T R A C T

Keywords: Low carbon heat generation is now a major concern for many industries to achieve sustainability targets,
Solar thermal however, it is not always clear which renewable or low carbon process heat technology is the most economical.
Heat pump This article develops a techno-economic assessment methodology to compare the cost effectiveness of solar
Renewable process heat thermal and electricity powered vapor compression heat pumps for process heat generation. Using key invest-
Levelized cost of heat
ment and performance indicators, it clearly elucidates the lower cost renewable or low carbon heat technology
Comparison methodology
under most conditions found in low temperature industries. The analysis also calculates the maximum turn-key
specific investment, inclusive of all material, labor, and financial costs, for solar thermal to remain financially
competitive against heat pumps, serving as a target for the solar thermal industry. The methodology, which is
independent of plant size, process temperature, and technology, reveals key results when applied to three cities
in Europe of varying solar irradiation and current electricity costs. In Seville, the maximum turn-key specific
solar investment is typically greater than 500 €/m2ap, meaning that solar thermal will most likely provide lower
cost heat than heat pumps. The case for Stockholm is the opposite, with the maximum investment being pri-
marily less than 300 €/m2ap, a challenging turn-key solar plant investment target that leads to the superiority of
heat pumps in this region. There is a wide range (230–1000 €/m2ap) of maximum turn-key solar thermal in-
vestment figures for a central German location (Würzburg), indicating that either technology could be selected,
but this is highly dependent on the process and other boundary conditions. Therefore, at any time now or in the
future, the developed methodology can flexibly compare solar thermal and heat pumps so that the lower cost
process heat technology can quickly be selected, while also providing a plant investment target for the solar
thermal industry.

1. Introduction
As industries begin to reduce their carbon footprints in accordance
with corporate, national, and international goals, renewable and low
carbon heat will replace current fossil fuel options. However, it is often
unclear what the most economical technology measures are. To help
industries with their investment decisions, this article develops a robust
methodology to determine which technology can produce low tem-
perature process heat at a relatively lower cost: solar thermal (ST) or a
vapor compression heat pump (HP) powered with grid electricity.
General economic assessments for renewable energy have been
frequently published, with some studies focused specifically on re-
newable heat for industrial processes and their calculated heat costs
and investment payback time (Silva et al., 2014; Lima et al., 2015;
Ommen et al., 2015; Sharma et al., 2018). However, few studies have
comparatively assessed ST and HP. A national study of renewable and
low carbon energy options for Australian industrial gas users shows
multiple levelized cost of heat (LCOH) scenarios of various process
temperatures against natural gas (Lovegrove et al., 2015). These results
indicate that solar process heat at 200 °C is less expensive when natural
gas costs > 36 €/MWh, and at 100 °C, when natural gas costs > 12 €/
MWh, at a site with an annual irradiation of 1850 kWh/m2a. Turn-key
ST plant investments are estimated to be between 350 and 1000 €/kWp
or 500 and 1400 €/m2ap (assuming 0.7 kWp/m2ap, a standard conversion
adopted from the International Energy Agency – Solar Heating and
Cooling research community) for flat plate and vacuum tube collectors.
In the same study (Lovegrove et al., 2015), electrically driven vapor
compression HPs are economically competitive when producing heat at
100 °C against a natural gas cost > 24 €/MWh, though this is highly
dependent upon Coefficient of Performance, electricity cost, and op-
erational hours. A similar report details renewable and low carbon
energy integration in industry, focusing on ST, HP, and biomass (Taibi

⁎
Corresponding author.
E-mail addresses: solar@uni-kassel.de, steven.andrew.meyers@gmail.com (S. Meyers).

https://doi.org/10.1016/j.solener.2018.08.011
Received 15 May 2018; Received in revised form 1 August 2018; Accepted 4 August 2018
0038-092X/ © 2018 Elsevier Ltd. All rights reserved.
S. Meyers et al. Solar Energy 173 (2018) 893–904

Nomenclature LCOH levelized cost of heat, €/kWh

OM operation and maintenance, %
Abbreviations ST
qsol solar thermal plant specific yield, kWh/m2ap
Q thermal energy, kWhth
CO2e carbon dioxide equivalent emissions RMSE root-mean-square error, normalized, %
EU European Union SD degradation rate, %
HP heat pump STelec electrical used by solar thermal, kWhel
IEA International Energy Agency θ0,1,2,3,4,5 model coefficients
MC Monte Carlo simulation Tamb ambient temperature, °C
PV photovoltaic TP,flow process load flow temperature, °C
ST solar thermal Tsink heat sink temperature for the heat pump, °C
Tsource heat sink source temperature for the heat pump, °C
Symbols UAstorage heat losses for the thermal storage, W/K
W work (electric), kWhel
COP coefficient of performance, kWth/kWel
Cel electricity cost, €/MWhel or €/kWhel Scripts
inflation of electricity costs, %
DR discount rate, % ap aperture
PV
esol specific annual PV yield, kWhel/kWp,el el electrical
ηCOP heat pump efficiency gen generalized calculation of term
εgen validation error, €/m2ap HP,ideal ideal operational case of a heat pump
′ , HP
FST correction factor for practical load cases (ST vs HP) local local calculation of term
F1 parameter of FST ′ , HP correction factor market market conditions of term
F2 parameter of FST ′ , HP correction factor sol generated solar energy
FLH full load hours, h/a th thermal
I specific plant investment, €/m2ap, €/kWth, €/Wp,el val validation set of term
ICR investment cost ratio, (€/m2ap)/(€/MWh)
LCOE levelized cost of energy/electricity, €/kWhel

et al., 2012). The calculated “break even” price for an installed ST plant
against other fuel stocks varies between 200 and 300 €/kWth for low
(1200 kWh/m2) and high (2000 kWh/m2) solar irradiation locations
respectively, or 140–210 €/m2ap when assuming a 0.7 kW/m2ap conver-
sion. The authors state that HP will compete with ST as a low carbon
heat technology but they do not directly compare the two technologies
to determine under which conditions one would produce heat at a
lower cost than the other. An assessment of subsidies and costs in the
European Union (EU) energy sector, undertaken by Ecofys (2014),
shows that when ST is coupled with a gas boiler, it is typically less
expensive than an air source HP, though in the domestic, not the in-
dustrial, sector. A wide reaching study by IRENA (2015) compiles the
renewable energy options for industry by 2030, predicting that ST
generated heat (< 150 °C) could cost anywhere between 30 and 90 €/
MWh, and HP generated heat anywhere between 20 and 50 €/MWh.
While this study provides an indirect technology comparison, its results
allow limited flexibility in ST specific yield, technology cost, electricity
cost, and HP operation time, with the latter set to a very optimistic
7000 h. The ongoing ENPRO project, led by AEE INTEC, takes this as-
sessment one step further, providing support for energy audits, re-
newable heat integration, and techno-economic analysis of both ST and
HP (Wilk et al., 2017) in the industrial sector. Pérez-Aparicio et al.
(2016) develop a comparison methodology between ST and a photo-
voltaic (PV) powered HP, similar to the work of Meyers et al; Meyers
et al. (2016; 2018). While the methodology from Pérez-Aparicio et al.
(2016) is applied to a handful of sites around the world, it only assesses
a HP at a fixed performance and process temperature at 60 °C, serving
as a preheater to achieve 200 °C with PV electrical resistance heating. In
addition, the results do not allow for flexibility in future technology
costs and are not supported by energetic simulations. Neyer et al.
(2018) use their developed assessment tool to compare a PV powered
air-water heat pump against a solar thermal system for domestic needs
in Madrid, indicating a slight economic advantage for the former but
did not address potential application for industrial process heat de-
mand.
Previous economic assessments have generally focused on com-
paring current process heat technology, i.e. fossil fuel boilers, with a
potentially lower cost renewable or low carbon heat technology, ST or
HP. In the coming years, the question will shift from “Can renewable or
low carbon process heat be less expensive than fossil fuel heat?” to
“Which renewable or low carbon heat technology is better suited to
achieve the industry’s carbon reduction or sustainability goals?” In this
light, this article’s main objective is to develop a robust methodology
coupled with an exemplary case study in Europe that determines which
renewable or low carbon process heat technology, ST or HP, can pro-
duce heat at a relatively lower cost, taking into account various fi-
nancial and industrial process conditions.
2. Methodology
This section develops a methodology to assess which renewable or
low carbon process heat technology, ST or HP, can produce lower cost
heat under primary boundary conditions. In Section 2.1 the main in-
fluencing parameters of both ST and HP are presented. Section 2.2
describes the techno-economic comparison methodology, using said
parameters. Section 2.3 uses the methodology to create assessment
calculations to compare the two technologies under ideal and non-ideal
conditions, leading to a universally applicable process for technology
comparison, both now and in the future.
2.1. Primary boundary conditions
The energetic performance of ST and HP is subject to very specific
industrial parameters that greatly influence each technology’s ability to
generate heat. Some, but not all, of these parameters include the in-
dustrial heat load profile and schedule, demanded temperature, and
waste heat source. The type of technology selected, ranging between
flat plate and compound parabolic concentrators for ST to various re-
frigerants, pumps, and pressure levels for HP, determines the potential
thermal energy generation per year at a given cost. Due to the near

894
S. Meyers et al. Solar Energy 173 (2018) 893–904

infinite quantity of techno-economic comparisons per technology, in- Table 2

dustrial demand, and location, a flexible general assessment framework A selection of high temperature heat pumps and their operating parameters.
is proposed, based on key inputs that can be easily obtained through Sources: 1 – Viessmann (2016), 2 – Ochsner Energie Technik (2016), 3 – Viking
publicly available tools, information sheets, and a general under- Heat Engines (2017), 4 – Tveit (2017).
standing of the factory in question. These input parameters are listed in Model Refrigerant Tsink (°C) Tsource (°C) ηCOP COP
Table 1, with the HP coefficient of performance (COP) and electricity
cost (Cel) representing average annual values. 350 HT Pro R1234ze 35–90 0–15 & 0.35–0.5 1.9–6.0 1
10–50
IWWHSS R134a/FKW 70–90 8–45 & 0.37–0.5 2.3–6.5 2
2.1.1. Solar thermal R2R3 ÖKO 1 8–55
ST
The specific solar yield (qsol ) is defined as the quantity of annual Heat Booster HFC245fa 90–130 40–50 & 0.34–0.53 2.1–4.9 3
collected solar energy per square meter of collector aperture area de- 70–80
Heat Booster 1336mzz(Z) 110–150 60–70 & 0.31–0.48 2.0–4.7 3
livered to the industrial process, inclusive of thermal losses from pipes
ST 90–100
and storage. Values of qsol range considerably, based on collector type, SPP High R-704 180 100 0.46 2.1 4
temperature range, storage size, process load profile/schedule and lo- Lift
cation. Typical values can vary from less than 300 kWh/(m2ap·a) in low
solar irradiation regions to more than 1200 kWh/(m2ap·a) in high solar
irradiation regions (Kalogirou, 2003; Martínez et al., 2012; Schmitt The COP efficiency (ηCOP) has been calculated for a number of
et al., 2015; Bellos et al., 2017; Furbo et al., 2018) for industries such as commercially available high temperature HPs (Table 2). The average
food and beverage manufacturing and textiles (Farjana et al., 2017, COP efficiency for all listed HPs is approximately 45%, leading to COPs
2018). Estimates of the specific solar yield can be carried out by the between 2 and 6, dependent on Tsink and dT. Similar COPs have been
System Advisor Model (NREL, 2017) software and on the Solar Pro- measured by other researchers (IEA Annex 35, 2014; Wolf et al., 2014;
zesswärme website (University of Kassel, 2017). A quick assessment Ommen et al., 2015; Bamigbetan et al., 2017; Arpagaus et al., 2018;
tool is also discussed in the dissertation of Meyers (2018). Bamigbetan et al., 2018). It should be noted that HPs often do not
The Solar Plant Investment (IST) is the turn-key price for a complete operate at their design capacity for the entire year, adjusting refrigerant
ST process heat plant, inclusive of thermal storage, installation, process flow to match transient waste/source heat energy flows. As such, an
integration, etc. and any locally available subsidies. Typical invest- instantaneous COP can vary due to the changing efficiency of the
ments for larger plants (> 500 m2ap) range between 300 and 400 €/m2ap electrically driven compressor at different flow rates (Li et al., 2015;
for flat plate collectors and 400–700 €/m2ap for vacuum tube collectors Waddicor et al., 2016). Therefore, research into HP technologies
(Ritter et al., 2017), with agreement from Eiholzer et al. (2017). Very (Aynur, 2010; Sarkar and Agrawal, 2010; Del Col et al., 2015) and
large plants (> 10,000 m2ap) built in Denmark for district heating with product offerings by companies now incorporate either variable speed
near optimal conditions can currently be constructed under 250 €/m2ap or multiple parallel compressors, or both, which helps to maintain a
(Hansen and Vad Mathiesen, 2018). Investment targets without sub- more constant COP over a range of flow rates. The referenced COP is an
sidies for solar process heat have been set by the European Solar average value over the operating year, taking into account any variable
Thermal Industry Federation (2014) at 250 €/m2ap for non-con- load performance that can be calculated by the HP supplier or through
centrating plants and 300 €/m2ap for concentrating plants by 2020. their data sheets. In addition, for industrial HP applications, the cold
side heat source is typically from waste heat created during the man-
2.1.2. Heat pumps ufacturing process, as indicated by van de Bor et al. (2015) and Law
The coefficient of performance (COP) is the main parameter used to et al. (2016). It is theoretically possible, but unlikely, to use air or
describe the ability of a HP to lift heat from a lower to a higher tem- ground source heat pumps due to cold side heat source availability and
perature. The parameter is ideally defined (Carnot) by the temperature its relatively low temperature level compared to the high process
of the sink (Tsink, or the process heat demand) in Kelvin divided by the temperatures required for industry. The methodology presented here is
temperature difference between the sink and source (Tsource, or the site- still valid for such HPs as long as one knows the annual average COP
specific waste heat), named the temperature lift (dT). Due to unavoid- and the annual number of full load operating hours to generate useable
able thermodynamic inefficiencies, typical COPs are much lower and heat by the industrial process.
calculated with a COP efficiency modifier (ηCOP), defined by the mea- In terms of generating lower carbon heat, ST is a fitting technology
sured COP of an operating HP, which is the ratio of useable heat pro- by virtue of its ability of convert carbon-free solar irradiation to thermal
duced (Qth) to the electrical work of the compressor (Wel), divided by energy, with a minor amount of electricity required and carbon emitted
the Carnot COP. The equation to calculate a HP COP is based on sink to operate fluid pumps (Wesselak et al., 2017). HPs also produce lower
and source temperatures (in Celsius), shown in Eq. (1), taking into carbon heat in comparison to a natural gas boiler, typical in industry, as
account the COP efficiency modifier. Only electric vapor compression long as a certain criterion is met. A grid electricity driven HP produces
HPs are investigated in this study. lower carbon heat as long as the HP COP is greater than the ratio of grid
electricity to natural gas equivalent carbon emission (CO2e) per unit of
Qth T + 273.15
COP = = ηCOP · sink generated energy. The current electrical equivalent carbon emissions of
Wel Tsink−Tsource (1)

Table 1
The main input parameters for the comparison of renewable and low carbon heat technologies.
Tech. Parameters Units Potential Source

ST Specific Solar Yield ST

(qsol ) kWh/m2ap·a Lauterbach (2014), Schmitt et al. (2015), University of Kassel (2017) and Meyers (2018)
Solar Plant Investment (IST) €/m2ap Local installer/supplier or (University of Kassel, 2017; AEE INTEC, 2017)

HP HP Plant Investment (IHP) €/kWth Local installer, supplier or (Wolf et al., 2014)
Coefficient of Performance (COP) kWth/kWel Manufacturer data sheets
Full Load Operation Hours (FLH) hours/year Basic energy audit
Electricity Cost (Cel) €/MWhel Basic energy audit
€/kWhel

895
S. Meyers et al.
1400
1200
1000
Specific 800
Investment
(€/kW) 600
400
200
0 intermittent, very low (< 500) FLHs are not out of the realm of possi-
10 100 1000
Heating Capacity (kW)
Fig. 1. A summary of specific turn-key heat pump investment (IHP), adapted
from IEA Annex 35 (2014).
most European countries are between 0.1 and 0.6 t CO2e/MWhel and is
approximately 0.225 t CO2e/MWhth for natural gas (Meyers, 2018).
When assuming a HP COP of 2.75 or greater, HPs produce lower carbon
heat as compared to a natural gas boiler for most countries, including
the worst case of grid electricity equivalent carbon emissions (0.6 t
CO2e/MWhel). If lower carbon electricity is available, the HP COP can
follow suit while ensuring a reduction in emitted carbon.
In addition, some HPs, so-called “integrated heat pumps,” can
provide both a cooling and heating source simultaneously. Such HPs are
often found in the food and beverage industry due to product cooling
(refrigeration) and heating (cooking, cleaning) requirements. In this
work, the focus is primarily on process heat applications and the in-
dicated COP is only for heat. If an integrated HP were to be used in the
described analysis and methodology, the COP would have to be in-
creased by approximately 60% to capture the economic benefit of
cooling (Wolf et al., 2014). This would significantly increase the eco-
nomic competitiveness of HPs.
HP plant investment (IHP), in units of €/kWth, is defined as the total
price of a HP, including all components related to its realization, such as
heat exchangers, thermal buffer storage, process integration for both
cold source and heat sink sides, and installation labor, divided by its
nameplate thermal capacity. Similar to ST, the investment costs are
variable, dependent on integration difficulty, size, storage requirement,
etc. A compilation of case studies for vapor compression industrial HPs
in Germany from the IEA Heat Pump Programme (IEA Annex 35, 2014)
is shown in Fig. 1. The specific HP investment as a function of plant size
generally ranges between 300 and 900 €/kWth for HPs larger than 100
kWth, which is the focus of this study. Wolf et al. (2014) and Wolf
(2017) indicate a similar range for turn-key HP investments, between
300 and 1000 €/kWth, with 400 €/kWth a current industry average for
off-the-shelf products to custom built multi-megawatt plants. On the
low end of the investment spectrum are results for China, with values
between 200 and 250 €/kWth (Zhang et al., 2016). Such low turn-key
investments will likely not be repeated in Europe or other countries in
the foreseeable future.
The last two main HP parameters that define its economic perfor-
mance are the grid electricity cost (Cel) to the industrial consumer and
the number of full load hours a HP can operate per year (FLH). While
electricity costs vary widely between companies due to their negotiated
Table 3
The effect of thermal storage losses on heat pump energy production.
Storage Type Heat Pump Tsink Tamb
(kWth) (°C) (°C)
Buffer 500 50–90 15
12 Hour 500 50–90 15
Solar Energy 173 (2018) 893–904
contracts and different cost structures (demand/capacity and con-
sumption charges), Eurostat (2017) provides typical electricity costs
(excluding value-added tax) in the EU. As of the end of 2017, costs
ranged from 0.065 €/kWhel to 0.155 €/kWhel for Sweden and Italy,
respectively, with an EU-28 average of 0.114 €/kWhel. The number of
full load hours (FLH) a HP can operate depends on the heat source and
sink availability, their concurrence, and potential storage size. If a
manufacturing process has one-shift operations, five days a week, a HP
can operate approximately 2000 h, assuming the source and sink are
available during the entire time. More continuous processes with
readily available heat sources can have higher FLHs, up to 6000 or more
in some cases (Brückner et al., 2015). If heat sources are limited or
10000 bility. In some situations, by incorporating a thermal storage, a smaller
HP can be used to achieve larger FLHs, but this must be included in the
turn-key IHP.
The use of a small storage tank is recommended in the design of a
HP plant to provide a small buffer from transient temperature spikes in
the source and sink streams, allowing better and more consistent
compressor and heat exchanger performance. Industry leading HP
vendors recommend a buffer tank to be sized between 10 and 20 L per
kWth of HP nameplate capacity (Fischer et al., 2016). In some in-
frequent cases, a larger tank is required to store hot water when the
source and sink streams are not concurrent. The use of a storage tank
will incur thermal losses. Two typical cases are presented in Table 3 to
estimate the impact of storage thermal losses on the overall energy
production of a HP: one for a small buffer storage tank and the other for
a large and “worst case” twelve-hour storage, representing a complete
non-concurrence between source and sink availability. Assuming a ty-
pical storage tank heat loss U-value of 0.86 W/m2 K (Cruickshank and
Harrison, 2010; Fan and Furbo, 2012b, 2012a), results show that the
thermal energy lost due to the inclusion of storage is quite small, a
quarter percent for the buffer storage and less than two percent for the
worst case twelve-hour storage. Other storage losses will vary between
these two values, depending on the number of storage hours required.
Due to the small impact of thermal storage losses on overall HP energy
production, they are neglected in the subsequent analysis for simplifi-
cation purposes.
2.2. Financial comparison calculations
The methodology described in this section builds on the metho-
dology for financial comparison between ST and PV, as developed in a
prior publication by Meyers et al. (2018). In this previous work, a fi-
nancial comparison term is coined, called the Investment Ratio, which
is a function of meteorological and industrial process inputs that was
derived by thousands of energetic simulations. Using this successful
analytical process, it is applied for a specific set of ST and HP process
heat plant parameters (Table 1 and 4) and the Levelized Cost of Heat
(LCOH) is calculated for both, using their turn-key investments, in-
clusive of subsidies. While the absolute LCOHs are helpful to know, only
the relative values are important when determining which of the two
technologies is relatively less expensive. To do this, it is first determined
at which point the LCOHs are equivalent, by setting LCOHST = LCOHHP,
using Eqs. (2) and (3). With known parameters of a HP (IHP, FLH, COP,
ST
Cel) and ST (qsol ) for a certain case, an iterated value of IST is determined
to make the LCOHs equivalent. Two new parameters are included in the
Storage Volume U-value UA-value HP Thermal Energy Lost
(l/kWth) (W/m2K) (W/K)
15 0.86 18.2 0.27%
257 0.86 121 1.82%
896
S. Meyers et al. Solar Energy 173 (2018) 893–904

LCOH calculations to more accurately account for operation costs: the
quantity of electricity used by the ST plant (STelec) and the inflation rate
of electricity cost (), as compared to Meyers et al. (2018). STelec is es-
ST
timated by dividing the annual solar specific (qsol ) yield by 80, a typical
average value for larger ST plants (Wesselak et al., 2017). LCOH results
are calculated for three different electricity inflation rates (0, 3, and
6%) to allow for flexibility in local conditions.
inf n
N OMST ·IST + STelec·Cel·(1 + Cel )
IST + ∑n = 1 (1 + DR)n
LCOHST =
q ST ·(1 − SD)n
N
∑n = 1 sol(1 + DR)n
N
IHP + ∑n = 1
OMHP·IHP + ( )·C ·(1 + C
FLH
COP
el
inf n
el )
(1 + DR)n
LCOHHP = N FLH
∑n = 1 (1 + DR)n
The key results from the iterations are the ST investment (IST) and
the grid electricity cost (Cel), as they are the two largest influencing
factors in the overall heat cost for the respective technology. When
combined into a ratio, dividing IST by Cel, a new term is coined, called
the Investment Cost Ratio (ICR), Eq. (4), valid only when
LCOHST = LCOHHP.
2 ICRlocal are the OMHP and the degradation rate of the ST plant. Higher
⎛ €, £, $/map ⎞
ICR ⎜ ⎟ = IST / Cel
⎝ €, £, $/MWh ⎠
The ICR is a powerful term that can be quickly applied at an in-
dustrial site to determine which technology provides lower cost heat.
This ratio has four main subscripts defined for certain uses: local,
market, gen, and val. The “local” subscript comes from the results of one
specific simulation run and is specific to that one local case, which is
the true or exact value when the process heat costs are identical. The
“market” subscript is used to denote a case when a turn-key investment
for the specified technology is known, drawing on the market price of
ST and electricity for a location of interest, inclusive of local subsidies
and other financial incentives. Regression analysis is used to generalize
and approximate the iterated ICR from many cases, denoted by “gen” in
the subscript. A validation simulation is conducted to confirm the
methodology’s accuracy and its results have a “val” subscript. The
“local” and “gen” subscript can be used interchangeably, as “gen” is
calculated to be a very accurate approximation of “local.” These sub-
scripts can also be used with IST and Cel to depict certain cases.
When the ICRgen and Cel,market are known for a specific case, they can
be multiplied together in Eq. (4) to obtain IST,gen. If an ST plant can be
built turn-key in a specific market (IST,market) for less than the calculated
IST,gen, ST will provide heat at a lower cost, shown in Eq. (5). This is
because the IST,gen value represents a point when LCOHST = LCOHHP,
indicating the maximum investment of an ST plant to remain compe-
titive against a HP. When the market price is lower (IST,market), so is
LCOHST relative to LCOHHP, ensuring ST is the lower cost low carbon
process heat solution.
⎧ ST chosen, IST , market < IST , gen
⎨ HP chosen, IST , market > IST , gen
⎩ (5)
The described financial analysis and subsequent results are sensitive
(2)
to the parameters in the LCOH equations which are not varied, namely
the discount rate (DR), operation and maintenance rate (OMST,HP), de-
gradation rate (SD), and years of operation (n). A brief sensitivity
analysis is conducted to determine if a fifty percent increase or decrease
(3) in these parameters’ values would greatly influence the ICRlocal, with
results shown in Fig. 2. The base case is taken for the following tech-
ST
nology parameters, typical for both an ST and HP plant; qsol : 600 kWh/
m2ap·a, STelec : 7.5 kWhel/m2ap·a, COP: 3.5, FLH: 2000 h/a, Cel: 0.10 €/
kWhel, : 3%, and IHP: 400 €/kWth.
The two most sensitive parameters are the cost to operate and
maintain the ST plant (OMST) and the number of operational years (N)
of both plants. The parameters that induce the least change in the
discount rates (DR) negatively influence the ICRlocal because the total
(4) present cost of purchased electricity, the main contributor to the pro-
duced heat cost, is reduced. In conclusion, the magnitude of the ICRlocal
change is relatively small, indicating that the results are robust under
typical financial parameters. If large deviations from these parameters
were to occur, a new simulation study with more fitting results would
have to be undertaken.
2.3. ST and HP comparison simulation and model generalization
When applying the ICR methodology for technology comparison, a
two-step process is recommended. First, the ICRlocal is calculated for an
infinite load or ideal case with fixed industry parameters (Section
2.3.1). A correction factor is then derived and calculated to adjust the
ICRlocal to a wider range of parameters, giving it universal applicability
(Section 2.3.2). Graphical tools (Section 3) provide an easy-to-use
method to determine the lower cost renewable or low carbon heat
producer.
2.3.1. ICR calculation, ideal case
The IST,local and ICRlocal are calculated using Eq. (4) when comparing

OM ST 3% 2% 1%

N (years) 15 20 25

DR 8% 6.4 % 4%

OM HP 1.25 % 2.5 % 3.75 %

SD 0.6 % 0.4 % 0.2 %

-10% -8% -6% -4% -2% 0% 2% 4% 6% 8% 10%

Change in ICRlocal
Fig. 2. A tornado chart representing the ICRlocal sensitivity analysis.

897
S. Meyers et al. Solar Energy 173 (2018) 893–904

Table 4
The input parameters for the LCOH calculations.
Parameter Definition ST Value HP Value

IST,HP Plant Investment (/m2ap, /kWth) Variable

OMST,HP Operation and Maintenance (% of IST,HP) 2% (VDI, 2014) 2.5% (VDI, 2012)
Cel Grid Electricity Cost 0.05–0.20 (Eurostat, 2017)
Celinf Inflation of Electricity Cost 0–3–6% (Schlesinger et al., 2014)
FLH Number of annual full load operation hours 500–8760
COP Coefficient of Performance 2–6
ST
qsol Specific Annual Solar Yield 200–1400
kWh/(m2ap·a)
STelec ST electricity Use ST
qsol /80 (Wesselak et al., 2017)
kWhel/m2a
DR Discount Rate Herrando and Markides (2016) and Allouhi et al. (2017)
SD Degradation Rate 0.4%/a (Fraunhofer ISE, 2016)
N Years of Operation 20 (Duffie and Beckman, 2013) 20 (VDI, 2012)

20·Cel·FLH
and iterating the two LCOH equations as set equal to each other for the IHP +
COP
span of expected COPs and qsol ST
(Table 4), while holding fixed the FLH, LCOHHP I Cel ⎞ ⎛ COP ⎞
′ , HP
FST ≈ ≈ 20·FLH
≈ ⎛ HP + · ⎜ ⎟

IHP, and Cel (8760 h/a, 200 €/kWth, 50 €/MWhel), respectively. This is LCOHHP, ideal 200 +
20·Cel·8760
⎝ FLH COP ⎠ ⎝ Cel ⎠
COP
repeated for multiple electricity inflation rates (). These values are 20·8760 (8)
symbolic of the ideal case when HPs will produce the lowest cost heat,
HP , ideal ′ , HP ≈ F1·F2 + 1
FST (9)
meaning that the ICRlocal (renamed ICRlocal , indicating the ideal HP
ST
case) is at its lowest possible value for a given COP and qsol . This implies F1 = IHP / FLH (10)
that the iterated IST,local is also at its lowest possible value (when Cel:
50 €/MWh), a useful cost target for the solar process heat industry to F2 = COP / Cel (11)
ensure it will be competitive with HP process heat. The analytically
HP , ideal F1 is the IHP divided by FLH. These two terms are paired together be-
calculated ICRlocal is generalized through a multiple linear regression
ST cause as IHP increases and FLH decreases (deviations from the ideal HP
analysis using the two inputs of COP and qsol , in Eq. (6). Results are
′ , HP both increase. There is a similar trend for F2: as COP
case), F1 and FST
graphically displayed in Section 3. ′ , HP both increase. By reducing the
increases and Cel decreases, F2 and FST
2 dimensionality to two instead of four, the results can be graphically
HP , ideal ⎛
€, £, $/ map ⎞ ST ST depicted, shown in Section 3.2, though only as a calculation tool and
ICRgen ⎜ €, £, $/MWh ⎟ = θ0 + θ1·COP + θ2·qsol + θ3·qsol / COP
⎝ ⎠ (6) not a sensitivity analysis.
Linear regression is employed to develop a mathematical relation-
ship betweenFST ′ , HP , F1 and F2 in the form of Eq. (12).

2.3.2. Correction factor for universal application ′ , HP ≈ θ4 ·(F1·F2) + θ5

FST (12)
The ideal HP operating parameters (FLH, IHP, and Cel) are not re- It is of final interest to calculate the maximum IST,local for realistic
presentative of typical conditions in industry. Typical HPs operate less cases, as this serves as a target for the ST community to reduce plant
than 8760 h, cost more than 200 €/kWth turn-key, especially when costs and compare against a known IST,market. By multiplying FST ′ , HP
considering installation costs, and use electricity from the grid at a cost (calculated and graphically displayed in Section 3.2) by ICRgen HP , ideal
, an
greater than 50 €/MWhel. To account for these differences, a correction accurate ICRgen is determined under case dependent operating condi-
factor (FST′ , HP ) is calculated dependent upon these three parameters and tions, in Eq. (13). The subscript “gen” indicates that its calculation
the COP, using non-linear regression and a Monte Carlo (MC) simula- comes from a generalized form of ICRlocalHP , ideal
′ , HP in Eq.
in Eq. (6) and FST
tion to obtain the required input data. (12) due to the regression analysis employed. This holds true for IST,gen
First, a 5000 run MC simulation is conducted in which the main in Eq. (14).
parameters for the LCOH equations are randomized per run (FLH:
2
500–8760 h/a, IHP: 200–800 €/kWth, Cel: 50–200 €/MWhel, COP: 2–6, ⎛ €, £, $/ map ⎞ HP , ideal
′ , HP
ST
qsol : 200–1400 kWh/m2ap·a). During each randomized run, the LCOH
ICRgen ⎜ ⎟ = ICRgen ·FST
⎝ €, £, $/ Wp ⎠ (13)
equations (Eqs. (2) and (3)) are set equal to each other and IST,local is
solved (Section 2.2), along with ICRlocal in Eq. (4). For the same MC The maximum investment for ST to remain the lower cost process
HP , ideal heat technology, IST,gen, is subsequently calculated by rearranging Eq.
simulation run, ICRgen is calculated in Eq. (6), using the randomized
ST
COP andqsol . The relative difference between these two values is the (4) combined with Eq. (13), shown in Eq. (14), when a market cost for
correction factor between the ideal and actual HP case, expressed by electricity is known (Cel,market).
the ratio in Eq. (7). 2
IST , gen (€/ map HP , ideal
) = ICRgen ′ , HP ·Cel, market
·FST (14)
HP , ideal
′ , HP =
FST ICRlocal / ICRgen (7)
3. Results
This correction factor is expressed in terms of FLH, IHP, Cel, and COP.
This relationship is derived by recognizing that FST′ , HP is approximately The results of the developed methodology are discussed in this
a ratio of the LCOHHP between the non-ideal and ideal case because the HP , ideal
section. Section 3.1 displays the calculated ICRlocal results for the
LCOHST remains the same, shown in Eq. (8). This equation can be re- ideal HP case. Section 3.2 details the regression results for the correc-
HP , ideal
′ , HP ) to adapt the ICRlocal
duced, ignoring the electricity cost inflation, discount, and degradation tion factor (FST to any possible HP operating
rate, to a simplified equation in Eq. (9), containing two reduced factors, case. Section 3.3 validates the methodology and Section 3.4 introduces
F1 and F2 (Eqs. (10) and (11), respectively). the possibility of a PV powered HP. Nomograms are developed to

898
S. Meyers et al. Solar Energy 173 (2018) 893–904

Table 5 Table 6
HP , ideal ST ′ , HP ).
Coefficients to calculate the ICRgen from qsol and COP using Eq. (6). Non-linear model coefficients to estimate the correction factor (FST

Celinf (%) θ0 θ1 θ2 θ3 RMSE Celinf (%) θ4 θ5 RMSE

LCOH 0 0.00073 −0.00018 0.00035 0.00879 0.02% LCOH 0 0.096 0.837 4.69%
3 0.00086 −0.00022 0.00032 0.01145 0.02% 3 0.077 0.867 3.75%
6 0.00111 −0.00028 0.00027 0.01522 0.02% 6 0.061 0.894 2.88%

IST,gen
[(€/m2ap)/(€/MWh)] (€/m2ap) F2
1400 4 8 400 (COP / Cel)
8

3
7

5
6

7 350 18

2
5 0.05

14
4
1
1200

18
8

12
3

16
10
6
100 16
6 300
4 3
1000 14
5

5 250
80 14
12 12
10
(kWh/m2ap·a)800 3 4 200

8
2

5
10

4
1
4

6
3
COP Cel 60
3 150 (kWth/ (€/kWh)
600 8
2 kWe)
3 2 100 8
6
1 40 5 6
400 4
2 1 1 50 3 4

2
5

1
200 1 20 4 2
2 3 4 5 6 3
2 0.20 2 0
0 0.5 1 1.5 2
HP , ideal
Fig. 3. The calculation of the Investment Cost Ratio (ICRgen ) and IST,local 200 IHP (€/kW) 1000
under ideal HP conditions with Celinf : 3% F1
(IHP/FLH) FLH
8760 500
(hrs/year)
provide a graphic tool for fast and simple technology comparison.
Fig. 4. A graphical representation of the calculated correction factor FST ′ , HP .
This graph is not to be interpreted as a sensitivity analysis; rather it serves as a
3.1. Ideal HP investment cost ratio calculation correction factor calculation tool once the COP is already known for a certain
HP , ideal ′ , HP
application. The same COP must be used for both the ICRgen and FST
HP , ideal calculation.
The Investment Cost Ratio (ICRlocal ) is calculated for a span of the
ST
two main performance parameters of ST and HP, qsol and COP, re-
spectively, while under ideal HP conditions (FLH: 8760 h/a, Cel: 50 €/ Under these conditions, ST process heat plants must generally be below
kWhel, IHP: 200 €/kWth). Three electricity inflation rates () are con- 300 €/m2ap turn-key, often less than 200 €/m2ap, to be the lower cost heat
sidered (0, 3, and 6 percent). supplier, as shown in the yellow and red areas of Fig. 3. While this serves
HP , ideal
The coefficients of the ICRgen regression model are shown in as an aggressive cost target for the ST industry to ensure that ST always
Table 5 for use in Eq. (6). One case (: 3%) is graphically depicted in produces lower cost heat than a HP, it is not a very realistic comparison
Fig. 3. The root-mean-square errors (RMSE) of the least squares re- because HPs nearly always consume electricity above 50 €/MWhel, op-
gression are very small for the LCOH cases, indicating a good regression erate less than 8760 h a year, and cost more than 200 €/kWth turn-key to
model. The p-values of each model coefficient are well below 0.05, purchase and install. To adapt the results from Fig. 3 and Eq. (6) for a
indicating significance. ′ , HP is calculated using the
more realistic comparison, a correction factorFST
It is recognized that the LCOH model fit is near perfect and could be four main HP parameters (FLH, IHP, Cel, and COP) through a MC analysis.
overfit. However, when removing the model parameters in succession,
the RMSE increases to greater than 5% in each case, so all terms are
included for improved model prediction at the detriment of a longer 3.2. Correction factor
equation.
HP , ideal
To use Fig. 3 for the calculation of ICRgen HP , ideal
, first determine the The required correction factor equation to adapt ICRgen to non-
annual average COP of a desired HP plant and locate its value on the X- ideal and more realistic operating conditions is derived from Eqs.
ST
axis. Then estimate a qsol of a potential ST plant installed on-site (8)–(11). By using a MC simulation for all possible values of FLH, IHP,
through a number of publicly available tools (Lauterbach, 2014; Cel, and COP, the correction factor (FST′ , HP ) is calculated in Eq. (7) and a
Schmitt et al., 2015; NREL, 2017; University of Kassel, 2017; Meyers, linear regression is conducted in Eq. (12) to simplify the results into a
HP , ideal
2018). The intersection of these two points is the ICRgen , determined concise and useable form. Regression results of Eq. (12) are shown in
by the contour lines and color bar on the right-hand side. The maximum Table 6 and are graphically displayed in Fig. 4.
IST,gen is also shown, as the Cel,market is already known for the ideal case The X-axis of Fig. 4 is the first dimensionality reduced term of the
(50 €/MWhel). An example is shown in Fig. 3 for a COP: 3.5 and qsol ST
: ′ , HP derivation, named F1, and is simply the IHP/FLH in Eq. (10). The
FST
2 HP , ideal
700 kWh/(map·a). The methodology produces a calculated ICRgen of Y-axis is the second term, named F2, and is the COP/Cel in Eq. (11). Cel
2.51 [(€/m2)/(€/MWhel)] and an IST,gen of 125 €/m2, when Cel,market: in this graph is in units of €/kWhel and not €/MWhel. The same COP is
50 €/MWhel. to be used as in the first calculation of the ICRgen HP , ideal
(Fig. 3), along with
HP , ideal
The calculated ICRgen serves as a starting point for the comparison a “market” known Cel,market, and not 0.05 €/kWhel for the ideal case. If a
between ST and HP for renewable and low carbon industrial process heat different COP were to be used, the ICRgen HP , ideal
must first be recalculated
generation, in particular when HPs are the most competitive against ST. in Eq. (6). In addition, this graph is not to be interpreted as a sensitivity

899
S. Meyers et al. Solar Energy 173 (2018) 893–904

Table 7 Table 8
′ , HP .
An example calculation of FST Validation results of the multiple linear regression analysis from the Monto
Carlo simulation.
F1 ′ , HP
FST
Celinf (%) RMSE (%) RMSE (€/m2)
IHP (€/kWth) 500 0.333 1.47
FLH (h/a) 1500 LCOH 0 4.7% 23.47
F2 3 3.7% 21.93
6 2.9% 20.64
COP (kWth/kWel) 3.5 23.333
Cel,market (€/kWhel) 0.15
maximum ST investment. In addition, the IST,gen is calculated using Eqs.
(6), (12) and (14). The difference between these two is the methodology
validation error (εgen = IST,gen − IST,val in €/m2ap). Its distribution for one
Percent of
case is shown in Fig. 5 with the validation error (εgen) along the X-axis
Validation Simulations
(2 €/m2ap bar width) and the frequency of error on the Y-axis. The error
(εgen) is rather Gaussian (red line) and well centered at nearly zero. The
RMSE of the error for all cases, shown in Table 8, is generally less than
4% or approximately 20 €/m2ap, smaller than the uncertainty within the
solar energetic simulations and thus considered acceptable. However,
some larger errors are observed, greater than 50 €/m2ap, which may
falsely indicate one technology’s preference over the other. If the ab-
solute difference between IST,gen and IST,market is less than 50 €/m2ap, an
individual calculation of LCOH for both technologies can eliminate this
uncertainty.
Through the use of the described methodology, the maximum ST
investment (IST,gen) can be quickly and accurately calculated. To de-
termine if ST can produce process heat at a lower cost than a HP, this
calculated value can be compared to a known market investment
(IST,market) in the location of interest. If IST,market < IST,gen, ST is the
lower cost process heat technology, in Eq. (5).
A nomogram (Fig. 6) is created to summarize the entire metho-
dology to determine the maximum IST,gen under all possible conditions
Validation Error, İgen (€/m2ap) for ST to remain the lower cost process heat technology (when
Fig. 5. The distribution of error from the use of the described methodology IST,market < IST,gen). The use of the nomograms is as follows:
instead of direct calculation, with a Gaussian distribution highlighted in red.
HP , ideal
Each error bin is 2 €/m2ap. (For interpretation of the references to color in this 1. Determine the ICRgen in the upper right box
figure legend, the reader is referred to the web version of this article.) • Draw a line from the top X-axis (e.g. qsol ST
: 700 kWh/m2ap·a) to meet
the known COP (3.5), indicated with the red lines, its intersection
analysis, it only serves to calculate a correction factor (FST′ , HP ), which is labeled “A”
the intersection of F1 and F2 on the graph, dependent on the four HP 2. Apply the Cel,market in the upper left box
inputs, shown on the color bar to the right-hand side. This can be also • Continue the line from the qsol ST
and COP intersection (“A”) and
move left until the intersection with the green line that indicates
calculated with Eq. (12), using model parameters in Table 6. For ex-
the known grid electricity cost, Cel,market: 0.15 €/kWhel, its inter-
ample, if a given industrial site intended to install a HP with a COP of
section labeled “B”
3.5 (same as the example in Fig. 3) at a turn-key investment of 500 €/
3. Determine FST′ , HP in the lower right box
kWth that operates 1500 h per year, using the market electricity at
0.15 €/kWhel, then F1: 0.333 and F2: 23.333, resulting in FST ′ , HP : 1.47, • Calculate F1 and F2 with known values of IHP (500 €/kWth), FLH
(1500 h/a), COP (3.5), and Cel,market (0.15 €/kWhel) from Table 7
summarized in Table 7.
This value corrects the ICRgen HP , ideal
of 2.51 [(€/m2ap)/(€/MWhel)], • Draw a line from the bottom X-axis (F1: 0.333) to meet the known
previously calculated in Section 3.1, to determine an accurate ICRgen F2: 23.333, indicated with the blue lines, its intersection labeled
under any HP conditions in Eq. (13). By applying the market electricity “C”
cost (Cel,market) of 150 €/MWhel in Eq. (14), the maximum ST investment 4. Determine IST,gen in the lower left box
(IST,gen) is determined for that case, shown in Eq. (15). If the market • Draw a line down from intersection point “B” and left from in-
investment (IST,market) for a turn-key ST plant in this region is less than tersection point “C” until they intersect each other
this value (IST,gen), ST is the lower cost process heat technology, in Eq. • This point, highlighted with the red circle, lies on black diagonal
(5). lines which indicate the IST,gen results, in this case IST,gen : 553 €/
m2ap.
2 HP , ideal 2
IST , gen (€/ map ) = ICRgen ′ , HP ·Cel, market = 2.51·1.47·150 = 553 €/ map
·FST
(15) The developed nomogram presents a useful tool to quickly de-
termine the maximum investment for ST to remain economically
competitive against a HP within the parameters used. In the final de-
3.3. Model validation and nomogram development
cision box located in the bottom left, two boxes are displayed in the
corners, indicating which technology is the clear and obvious choice in
A second MC simulation is conducted to determine the validity of
a specific geographical region. ST will always be chosen when the de-
the assessment methodology described in Section 2.3. The simulation
rived IST,gen is greater than 1200 €/m2ap, and HP when the IST,gen is below
consists of 10,000 runs with randomized parameters from Table 1, it-
200 €/m2ap, due to current and probable future market technology costs.
erated to solve for the IST,val (“val” for validation in the subscript to
Between these two values are black lines, indicating the IST,gen used to
indicate its source) when LCOHST = LCOHHP, which is the “true”

900
S. Meyers et al. Solar Energy 173 (2018) 893–904

compare against IST,market to reach a technology decision in Eq. (5).

One can infer from the nomogram how the primary parameters in-
fluence the technology selection. When the electricity cost (Cel,market),
located in the upper left box, increases from 0.05 to 0.20 €/kWhel, it
shifts the intersection point (point B) right, increasing the IST,gen value
and making HPs less competitive. The same holds true for the solar yield
ST
(qsol ) in the upper right box, which, when increased (assuming a constant Cel,market
COP), will also shift its intersection point (point A) right, thus inducing (€/kWhel) IPV (€/Wp,el)
point B to shift as well, finally increasing IST,gen (red dot). The contrary is
true when the COP increases. The correction factor (FST ′ , HP ) is slightly
counterintuitive, seen in the lower right box. For F1, the relationship is
clear. When the HP investment increases (IHP > 200 €/kWth) and the HP
operates fewer hours (FLH < 8760 h/a) compared to the ideal case, F1
increases and the blue lines trend positive, moving point C upwards. This
shifts the IST,gen in a positive direction as the HP generated heat is more
expensive. For F2, the case is the opposite. As COP increases and Cel,market (kWhel/kWp,el )
decreases (larger F2 value), HPs become more competitive, leading to a
decrease in IST,gen, but the opposite is depicted in FST ′ , HP . However, since Fig. 7. A nomogram to calculate the market cost of on-site PV generated
the COP and Cel,market terms were already used earlier in the nomogram electricity, when the PV annual yield and specific investment are known.
and their values are fixed, only one F2 blue isoline can be used and serves
as it was denoted, as a correction factor from the assumptions made LCOE (Y-axis), or Cel,market, is quickly determined by knowing the turn-
when first calculating ICRgen HP , ideal
. This subsequently shifts the IST,gen in- key PV investment (in €/Wp,el) shown in the contour lines and the es-
tersection point (red dot) to minimize error when employing the linear timated annual yield along the X-axis, using Fig. 7. Once calculated, this
regression model (Eq. (12), Table 6, and Fig. 4). While counterintuitive Cel,market can be used in a nomogram similar to Fig. 6 when the inflation
for F2, the overall methodology is successful in predicting IST,local with an rate of electricity () is zero.
RMSE between 2 and 4%.

3.4. Use of photovoltaics as a heat pump electricity source
Instead of using grid electricity, whose future cost is unknown, a HP
can utilize renewable electricity from a local PV power plant, essen-
tially “buying” PV power at its locally generated cost. The levelized cost
of electricity (LCOE) generated by a local PV plant has been calculated
previously (Meyers et al., 2018) as a function of estimated solar PV
PV
yield (esol , gen ) and site latitude. This cost is only for local generation and
self-consumption and does not include taxes or other externalities. The
4. Case study
To indicate the current status of low carbon industrial process heat,
this section applies the assessment methodology developed in Section 2
and displayed in Section 3 to a number of possible cases. It focuses on
three European cities of interest (Stockholm [Sweden], Würzburg [Ger-
many], and Seville [Spain]), which vary in terms of annual solar irra-
diation and average industrial electricity prices. These cities are used to
demonstrate the functionality of the described methodology to determine
the maximum investment for ST (IST,gen). The case studies are currently

Cel,market (€/kWh) (kWh/m2ap·a)

400 600 800 1000 1200 1400

COP

B A

ST
F2:
COP/Cel,market

C
HP
0.25 0.5 0.75 1 1.25 1.5 1.75 2
F1 :IHP / FLH

Fig. 6. A nomogram tool to calculate the IST,gen from plant parameter inputs, assuming a 3% electricity cost inflation.

901
S. Meyers et al. Solar Energy 173 (2018) 893–904

900 Seville
0.11
(kWh/m2a)
500 Würzburg
0.14
Cel
(€/kWh) 400 Stockholm
0.06
COP IHP
(kWth/kWel) (€/kWth)

600
3
IST,gen 200
(€/m2ap)

COP IHP
(kWth/kWel) (€/kWth)

2
400
4

Fig. 8. Summary results of the comparison methodology for three city and typical parameter values for both technologies to generate heat at a lower cost. The HP and
ST investments are turn-key.

valid for the 2018 electricity costs. If the electricity costs are to sig- steep increase, especially for FLHs less than 2000 h. This is expected due
nificantly change in the coming years due to a variety of factors, such as to the LCOH calculation for the HP in Eq. (3), with FLH in the de-
increased national grid integration and renewable electricity installations, nominator. The case study results indicate high sensitivity to IHP when
one must implement the described methodology with the most current FLHs are less than 2000 (top graph, Fig. 8), noticeable by the wide color
values available to guarantee an accurate technology comparison. band. This is because the majority of the overall heat cost is attributed
The solar yield (qsolST
) is set at 900, 500, and 400 kWh/(m2ap·a) for to the initial HP investment and not the operating cost when FLHs are
Seville, Würzburg, and Stockholm respectively, estimated from solar minimal. The opposite is true for the bottom graph of Fig. 8, which
simulations previously conducted by Meyers et al. (2018). The required shows high COP sensitivity at high FLHs. In this case, the heat cost
electricity cost for the HP (Cel,market) is determined from Eurostat (2017) depends primarily on HP operating parameters and not upfront in-
at 0.11, 0.14, and 0.06 €/kWhel for each country respectively, shown in vestment.
Fig. 8. The rate of electricity cost inflation is assumed to be zero. The At 8000 FLH, or high HP utilization, IST,gen is at its lowest, providing
parameters for COP and IHP are varied in two plots. The top plot in the most competitive climate for HPs. In this case, ST only makes fi-
Fig. 8 holds the COP constant at 3 while varying IHP between 200 and nancial sense in Seville or in other high solar resource regions. From
600 €/kWth. The bottom plot holds IHP constant at 400 €/kWth while 4000–8000 h, the IST,gen remains fairly constant. This implies that op-
varying COP between 2 and 4. This allows for a visualization of a band erating a HP for more than 4000 h per year does not make it
of potential IST,gen values, given that these parameters are very industry
specific, and sensitive at different HP FLHs. The upper limit of the co-
Table 9
lored area represents the “ST favored” case and the lower limit re- The calculated IST,gen values for various cases often found in industry for the
presents the “HP favored” case. An “average case” is also calculated, LCOH comparison.
when COP: 3 and IHP: 400 €/kWth, indicated by the thicker line bi-
Stockholm Würzburg Seville
secting the colored bands. The X-axis of Fig. 8 shows the number of
FLHs the HP may operate. The Y-axis shows the determined IST,gen. COP 4 3 2 4 3 2 4 3 2
Seville is shown in light red, Würzburg in blue, and Stockholm in green. Cel,market (€/kWhel) 0.06 0.14 0.11
The other colors located between these cities (purple, dark green) are IHP (€/kWth) 200 400 600 200 400 600 200 400 600
instances when the bands of results overlap. The regression model error IST,gen (€/m2ap)
(Table 8) of less than 5% is not included as it is much smaller than the
range of IST,gen values for the various case studies. FLH (h/a) 500 207 367 531 344 569 822 562 950 1372
1000 129 217 313 247 381 550 387 611 883
The results from Fig. 8 indicate a few key trends, accompanied by 2000 90 141 204 198 287 414 299 442 638
numerical IST,gen outcomes for select cases in Table 9. The number of 4000 71 104 150 174 240 346 255 357 516
FLHs that a HP can operate heavily influences the IST,gen, indicated by its 8000 61 85 123 162 216 312 233 315 454

902
S. Meyers et al.
significantly more competitive against ST. At 2000 FLH, or medium HP
utilization, HP is the preferred technology in Stockholm (IST,gen:
90–200 €/m2ap), while ST is more cost efficient in Seville (IST,gen:
300–630 €/m2ap). In Würzburg, the choice between both technologies is
dependent on the IHP and COP. When FLHs are less than 2000 h, the
IST,gen sharply increases, clearly making ST the lower cost process heat
choice. At low HP utilization (500 FLH), the range of IST,gen is
562–1372 €/m2ap for Seville, 344–822 €/m2ap for Würzburg, and
207–531 €/m2ap for Stockholm, indicating that ST is the lower cost heat
producer in nearly every case. This case study shows that a flexible
methodology is required to quickly estimate IST,gen as the numerous
parameters greatly influence each technology’s feasibility. An extended
quantitative analysis of these and other results are found in the dis-
sertation of the primary author (Meyers, 2018).
In addition, the cost of electricity greatly influences the calculated
IST,gen. While both Stockholm and Würzburg have similar solar yields
(400 and 500 kWh/m2a, respectively), the calculated IST,gen for
Stockholm is between 40% and 60% lower, primarily due its sig-
nificantly lower cost of electricity. In such low-cost electricity regions,
the competitiveness of ST is greatly challenged unless a very high solar
resource is available.
The solar yield exhibits its influence on IST,gen, as evident from the
Seville results. The yield is nearly double that of Würzburg and, even
though electricity is three cents less expensive per kWhel (favoring HP),
the IST,gen remains 45–65% higher. Renewable or low carbon process
heat plants located in high solar resource regions are typically better
suited for ST, unless a HP can operate with a COP > 4 for more than
half of the year (> 4000 FLH) at lower electricity costs (< 0.11 €/
kWhel).
Current specific ST investment (IST,market) varies widely depending
on plant size, difficulty of integration, and installer expertise, to name
just a few influencing factors. For larger (> 500 m2ap) well-built plants
in Germany, an IST,market < 300 €/m2ap is possible for flat plate collec-
tors, and < 500 €/m2ap for evacuated tube collectors without subsidy
(Ritter et al., 2017). At these investment levels, ST should not be in-
stalled in Stockholm, should only be installed in Würzburg when
FLH < 1000, and can be installed in most cases in Seville. If subsidies
are locally available, the range of ST feasibility will increase, depending
on the level of financial support. The European Solar Thermal Industry
Federation (2014) has identified an IST,market goal of 250 €/m2 by 2020.
If this were to be consistently achieved, ST would clearly be the pre-
ferred renewable heat choice in Seville and Würzburg and for some
cases in Stockholm if the turn-key IHP were high and the COP and FLH
low.
The described methodology and case study indicate similar results
to very limited research that directly compares both technologies for
useful process heat generation. Pérez-Aparicio et al. (2016) compares
an ST and a PV coupled HP for process heat generation at 200 °C. In
their case study for Seville, the researchers indicate that ST is the
preferred technology in most cases, in agreement with the results
shown in Table 9, though with different operational boundary condi-
tions. In the Nordic countries and in particular Stockholm for this case
study, HPs are typically the lower cost technology. This finding is
confirmed by Rämä and Wahlroos (2018), who demonstrate that HPs
are the more economical option in Helsinki for heating requirements
between 80 °C and 110 °C. The agreement with prior research lends
credibility to the methodology and case study presented here and helps
justify its applicability to compare lower carbon process heat technol-
ogies around the world.
5. Summary and conclusion
The need for low cost process heat to reduce energy use and carbon
emissions in industry is critical to minimizing the detrimental effects of
climate change. Two such technologies, ST and HP, can already provide
lower carbon heat, as evident by the successful implementation of
903
Solar Energy 173 (2018) 893–904
numerous plants around the world. As the technology matures and
more companies decide to reduce their carbon footprints, a clear and
robust methodology is needed to determine which technology provides
relatively lower cost heat, using key performance parameters.
This paper has presented such a methodology, using plant turn-key
investments and electricity costs, specific ST yield, COP, and annual HP
operating hours as the main inputs. Through subsequent financial
analysis and data modeling, an easy-to-use process and nomograms
have been created to determine the maximum investment for ST to be
competitive against HPs. This value can be used in two ways: to com-
pare it to the current market investment of ST plants or as an ST
technology development goal to reduce costs and preserve future
competitiveness.
A case study has been conducted, using the nomogram and mod-
eling results, to understand the current competitive landscape between
ST and HPs in three European cities with different ambient tempera-
tures, available solar irradiation levels, and average industrial elec-
tricity cost. Results are split into three main groups, low, medium, and
high HP utilization, based on the number of FLHs. In the low HP uti-
lization (500 h/a) case, ST is currently the lower cost heat technology,
except in Stockholm, if a plant’s COP is high (4) and technology cost
(IHP) is low (200 €/kWth). The opposite is true in high utilization cases
(FLH > 4000 h/a), where HPs produce lower cost heat in nearly all
cases, except in Seville, when the COP is low (2) and technology cost
(IHP) is high (600 €/kWth). When operating for 2000 h a year (middle
utilization case), HP produced energy is the lower cost option in
Stockholm and in some cases in Würzburg, while in Seville, ST is the
most cost-effective heat provider. In any future case study assessment,
new values for electricity costs must be applied to the methodology as
their values will change over time.
It is obvious that the low carbon renewable and low carbon process
heat assessment is extremely case and parameter dependent. The de-
veloped methodology serves as a valuable guide to quickly determine a
preferred lower carbon heat solution, while also providing cost targets
for both technologies to remain competitive against the other. In the
near future, the price of emitted carbon from industrial facilities may
play a more critical role in technology selection. As such, the developed
comparison methodology can be reconfigured to compare the levelized
cost of abated carbon and determine which technology reduces carbon
emissions at a lower cost.
Acknowledgements
This work was supported by the People Programme (Marie Curie
Actions) of the Seventh Framework Programme FP7/2007e2013/ under
REA grant agreement no. 317085 [PITN-GA-2012-317085], commonly
known at the SHINE (Solar Heat INtegration Network) Program
(https://www.uni-kassel.de/projekte/solnet-shine/home.html).
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the
online version, at https://doi.org/10.1016/j.solener.2018.08.011.
References
AEE INTEC, 2017. SHIP plants. Database for applications of solar heat integration in
industrial processes. < http://ship-plants.info/ > .
Allouhi, A., Agrouaz, Y., Benzakour Amine, M., Rehman, S., Buker, M.S., Kousksou, T.,
Jamil, A., Benbassou, A., 2017. Design optimization of a multi-temperature solar
thermal heating system for an industrial process. Appl. Energy 206, 382–392.
Arpagaus, C., Bless, F., Uhlmann, M., Schiffmann, J., Bertsch, S.S., 2018. High tempera-
ture heat pumps. Market overview, state of the art, research status, refrigerants, and
application potentials. Energy 152, 985–1010.
Aynur, T.N., 2010. Variable refrigerant flow systems. A review. Energy Build. 42 (7),
1106–1112.
Bamigbetan, O., Eikevik, T.M., Nekså, P., Bantle, M., 2017. Review of vapour compression
heat pumps for high temperature heating using natural working fluids. Int. J. Refrig.
S. Meyers et al.
80, 197–211.
Bamigbetan, O., Eikevik, T.M., Nekså, P., Bantle, M., Schlemminger, C., 2018. Theoretical
analysis of suitable fluids for high temperature heat pumps up to 125 °C heat de-
livery. Int. J. Refrig.
Bellos, E., Tzivanidis, C., Belessiotis, V., 2017. Daily performance of parabolic trough
solar collectors. Sol. Energy 158, 663–678.
Brückner, S., Liu, S., Miró, L., Radspieler, M., Cabeza, L.F., Lävemann, E., 2015. Industrial
waste heat recovery technologies. An economic analysis of heat transformation
technologies. Appl. Energy 151, 157–167.
Cruickshank, C.A., Harrison, S.J., 2010. Heat loss characteristics for a typical solar do-
mestic hot water storage. Energy Build. 42 (10), 1703–1710.
Del Col, D., Azzolin, M., Benassi, G., Mantovan, M., 2015. Energy efficiency in a ground
source heat pump with variable speed drives. Energy Build. 91, 105–114.
Duffie, J.A., Beckman, W.A., 2013. Solar engineering of thermal processes, 4th ed. John
Wiley & Sons.
Ecofys, 2014. Subsidies and costs of EU energy. Final report. < https://ec.europa.eu/
energy/sites/ener/files/documents/ECOFYS%202014%20Subsidies%20and
%20costs%20of%20EU%20energy_11_Nov.pdf > (accessed 29 September 2017).
Eiholzer, T., Olsen, D., Hoffmann, S., Sturm, B., Wellig, B., 2017. Integration of a solar
thermal system in a medium-sized brewery using pinch analysis. Methodology and
case study. Appl. Therm. Eng. 113, 1558–1568.
European Solar Thermal Industry Federation, 2014. Solar Heating and Cooling
Technology Roadmap. European Technology Platform on Renewable Heating and
Cooling. < http://www.rhc-platform.org/fileadmin/user_upload/Structure/Solar_
Thermal/Download/Solar_Thermal_Roadmap.pdf > (accessed 27 September 2017).
Eurostat, 2017. Electricity prices for non-household consumers. < http://appsso.eurostat.
ec.europa.eu/nui/show.do?dataset=nrg_pc_205&lang=en > (accessed 29
September 2017).
Fan, J., Furbo, S., 2012a. Buoyancy driven flow in a hot water tank due to standby heat
loss. Sol. Energy 86 (11), 3438–3449.
Fan, J., Furbo, S., 2012b. Thermal stratification in a hot water tank established by heat
loss from the tank. Sol. Energy 86 (11), 3460–3469.
Farjana, S.H., Huda, N., Mahmud, M.P., Saidur, R., 2017. Solar process heat in industrial
systems – a global review. Renew. Sustain. Energy Rev.
Farjana, S.H., Huda, N., Mahmud, M.P., Saidur, R., 2018. Solar industrial process heating
systems in operation – current SHIP plants and future prospects in Australia. Renew.
Sustain. Energy Rev. 91, 409–419.
Fischer, D., Lindberg, K.B., Madani, H., Wittwer, C., 2016. Impact of PV and variable
prices on optimal system sizing for heat pumps and thermal storage. Energy Build.
128, 723–733.
Fraunhofer ISE, 2016. Entwicklung beschleunigter Alterungstestverfahren für so-
larthermische Kollektoren und deren Komponenten. Fraunhofer-Institut für Solare
Energiesysteme ISE.
Furbo, S., Dragsted, J., Perers, B., Andersen, E., Bava, F., Nielsen, K.P., 2018. Yearly
thermal performances of solar heating plants in Denmark – measured and calculated.
Sol. Energy 159, 186–196.
Hansen, K., Vad Mathiesen, B., 2018. Comprehensive assessment of the role and potential
for solar thermal in future energy systems. Sol. Energy 169, 144–152.
Herrando, M., Markides, C.N., 2016. Hybrid PV and solar-thermal systems for domestic
heat and power provision in the UK Techno-economic considerations. Appl. Energy
161, 512–532.
IEA Annex 35, 2014. Applications of Industrial Heat Pumps. Final Report, Part 1 and 2.
Report No. HPPAN35-1 and 2; ISBN 978-91-88001-92-4 and 978-91-88001-93-1.
IRENA, 2015. A background paper to “Renewable Energy in Manufacturing”. < http://
www.irena.org/remap/IRENA_RE_Potential_for_Industry_BP_2015.pdf > (accessed
29 September 2017).
Kalogirou, S., 2003. The potential of solar industrial process heat applications. Appl.
Energy 76 (4), 337–361.
Lauterbach, C., 2014. Potential, system analysis and preliminary design of low-tem-
perature solar process heat systems. Kassel University Press, Kassel.
Law, R., Harvey, A., Reay, D., 2016. A knowledge-based system for low-grade waste heat
recovery in the process industries. Appl. Therm. Eng. 94, 590–599.
Li, Y., Liu, M., Lau, J., Zhang, B., 2015. A novel method to determine the motor efficiency
under variable speed operations and partial load conditions. Appl. Energy 144,
234–240.
Lima, T.P., Dutra, J.C.C., Primo, A.R.M., Rohatgi, J., Ochoa, A.A.V., 2015. Solar water
heating for a hospital laundry. A case study. Sol. Energy 122, 737–748.
Lovegrove, K., Edwards, S., Jacobson, N., Jordon, J., Peterseim, J., Rutowitz, J., Saddler,
H., Watt, M., Wyder, J., 2015. Renewable energy options for australian industrial gas
users. Australian Renewable Energy Agency. < http://arena.gov.au/assets/2017/
05/ITP-RE-options-for-industrial-gas-users-Summary.pdf > (accessed 29 September
2017).
Martínez, V., Pujol, R., Moià, A., 2012. Assessment of medium temperature collectors for
process heat. Energy Procedia 30, 745–754.
Meyers, S., 2018. Methodology development and assessment of lower carbon industrial
process heat through solar energy and heat pumps. PhD Dissertation. < http://nbn-
resolving.de/urn:nbn:de:hebis:34-2018052455520 > (accessed 13 August 2018).
Meyers, S., Schmitt, B., Vajen, K., 2016. Competitive assessment between solar thermal
904
View publication stats
Solar Energy 173 (2018) 893–904
and photovoltaics for industrial process heat generation. In: Proceedings of
EuroSun2016. International Solar Energy Society, pp. 1–11.
Meyers, S., Schmitt, B., Vajen, K., 2018. Renewable process heat from solar thermal and
photovoltaics: the development and application of a universal methodology to de-
termine the more economical technology. Appl. Energy.
Neyer, D., Ostheimer, M., Dipasquale, C., Köll, R., 2018. Technical and economic as-
sessment of solar heating and cooling – methodology and examples of IEA SHC Task
53. Sol. Energy.
NREL, 2017. System Advisor Model (SAM).
Ochsner Energie Technik, 2016. High temperature heat pump. IWWHSS R2R3. < http://
ochsner-energietechnik.com/wp-content/uploads/2015/08/TI_MD_IWWHSS_R2R3_
Hochtemperatur_W%C3%A4rmepumpen_DE_20160427_V02-Internet.
pdf > (accessed 29 September 2017).
Ommen, T., Jensen, J.K., Markussen, W.B., Reinholdt, L., Elmegaard, B., 2015. Technical
and economic working domains of industrial heat pumps: Part 1 - Single stage vapour
compression heat pumps. Int. J. Refrig. 55, 168–182.
Pérez-Aparicio, E., Lillo-Bravo, I., Moreno-Tejera, S., Silva-Pérez, M., 2016. Economical
and environmental analysis of thermal and photovoltaic solar energy as source of
heat for industrial processes. In: AIP Conference Proceedings, pp. 180005.
Rämä, M., Wahlroos, M., 2018. Introduction of new decentralised renewable heat supply
in an existing district heating system. Energy 154, 68–79.
Ritter, D., Schmitt, B., Vajen, K., 2017. Analyse und Erschließung des Marktes für solare
Prozesswärme in Deutschland (in German Language). Abschlussbericht
Förderkennzeichen 03MAP286.
Sarkar, J., Agrawal, N., 2010. Performance optimization of transcritical CO2 cycle with
parallel compression economization. Int. J. Therm. Sci. 49 (5), 838–843.
Schlesinger, M., Hofer, P., Kemmler, A., Kirchner, A., Koziel, S., Ley, A., Ulrich, P., 2014.
Entwicklung der Energiemärkte–Energiereferenzprognose. Studie im Auftrag des
Bundesministeriums für Wirtschaft und Technologie. < https://www.prognos.com/
uploads/tx_atwpubdb/140716_Langfassung_583_Seiten_Energiereferenzprognose_
2014.pdf > (accessed 3 October 2017).
Schmitt, B., Lauterbach, C., Ritter, D., Meyers, S., Vajen, K., 2015. Fast feasibility as-
sessment for solar thermal systems in industry. In: Proceedings of the EuroSun 2014
Conference. International Solar Energy Society, pp. 1–10.
Sharma, A.K., Sharma, C., Mullick, S.C., Kandpal, T.C., 2018. Financial viability of solar
industrial process heating and cost of carbon mitigation. A case of dairy industry in
India. Sustain. Energy Technol. Assess. 27, 1–8.
Silva, R., Berenguel, M., Pérez, M., Fernández-Garcia, A., 2014. Thermo-economic design
optimization of parabolic trough solar plants for industrial process heat applications
with memetic algorithms. Appl. Energy 113, 603–614.
Taibi, E., Gielen, D., Bazilian, M., 2012. The potential for renewable energy in industrial
applications. Renew. Sustain. Energy Rev. 16 (1), 735–744.
Tveit, T., 2017. Application of an industrial heat pump for steam generation using district
heating as a heat source. 12th IEA Heat Pump Conference 2017. HPC 2017,
Rotterdam.
University of Kassel, 2017. Solar Process Heat Website (in German Language). < http://
www.xn–solare-prozesswrme-ztb.info/vorauslegung/ > .
van de Bor, D.M., Infante Ferreira, C.A., Kiss, A.A., 2015. Low grade waste heat recovery
using heat pumps and power cycles. Energy 89, 864–873.
VDI, 2012. 2067 – Economic Efficiency of Building Installations. Verein Deutscher
Ingenieure.
VDI, 2014. 6002 – Solar Heating for Potable Water. Verein Deutscher Ingenieure (ac-
cessed 13 July 2017).
Viessmann, 2016. Vitocalc 350-HT Pro. < https://www.viessmann.de/de/gewerbe/
waermepumpe/grosswaermepumpen/vitocal-350-ht-pro.html > (accessed 29
September 2017).
Viking Heat Engines, 2017. HeatBooster. High-temperature heat pump that generates
clean heat up to 160 °C. < http://www.vikingheatengines.com/upl/files/
142690 > (accessed 29 September 2017).
Waddicor, D.A., Fuentes, E., Azar, M., Salom, J., 2016. Partial load efficiency degradation
of a water-to-water heat pump under fixed set-point control. Appl. Therm. Eng. 106,
275–285.
Wesselak, V., Schabbach, T., Link, T., Fischer, J., 2017. Handbuch Regenerative
Energietechnik. Springer Berlin Heidelberg, Berlin, Heidelberg.
Wilk, V., Windholz, B., Jentsch, R., Fleckl, T., Fluch, J., Grubbauer, A., Brunner, C., Lange,
D., Dietrich, W., Ponweiser, K., 2017. Valorization of industrial waste heat by heat
pumps based on case studies of the project EnPro. 12th IEA Heat Pump Conference
2017. HPC 2017, Rotterdam.
Wolf, S., 2017. Integration von Wärmepumpen in industrielle Produktionssysteme.
Potenziale und Instrumente zur Potenzialerschließung.
Wolf, S., Fahl, U., Blesl, M., Voß, A., 2014. Analyse des Potenzials von Industrie-
wärmepumpen in Deutschland. Universität Stuttgart, Institut für Energiewirtschaft
und Rationelle Energieanwendung. < http://www.ier.uni-stuttgart.de/
publikationen/veroeffentlichungen/forschungsberichte/downloads/141216_
Abschlussbericht_FKZ_0327514A.pdf > .
Zhang, J., Zhang, H.-H., He, Y.-L., Tao, W.-Q., 2016. A comprehensive review on advances
and applications of industrial heat pumps based on the practices in China. Appl.
Energy 178, 800–825.