Energy Reports 11 (2024) 624–636

Contents lists available at ScienceDirect

Energy Reports
journal homepage: www.elsevier.com/locate/egyr

Research paper

Energy efficiency investment in Swiss industry: Analysis of

target agreements
Navdeep Bhadbhade *, Martin K. Patel
Chair for Energy Efficiency, Department F.-A. Forel for Environment and Aquatic Sciences, Institute for Environmental Sciences (ISE), University of Geneva, Switzerland

A R T I C L E I N F O A B S T R A C T

Keywords: Slowing down the energy efficiency (EE) improvement rate over the past half a decade and relatively low EE-
Profitability related investments highlight a need for a more targeted approach to encourage EE improvement investment
Cost-effectiveness in the Swiss manufacturing sector. The current work is based on an analysis of already implemented energy
K-means clustering
efficiency measures (EEMs) via target agreements. Four segments of the Swiss manufacturing sector are iden­
Switzerland
tified using the K-means clustering based on the energy consumption profiles of the establishments. The EEMs are
Energy-efficiency investment decision-making
categorized through text analysis with the help of a machine learning algorithm. The EEMs are ranked using two
indicators that can inform different stakeholders: i) EEMs with Internal rates of return well above the reported
risk-adjusted hurdle rates are identified for each segment to help industrial decision-makers take ad-hoc, profit-
based investment decisions on EE improvement. ii) Levelized costs of saved energy are presented as value ranges
for the EEMs that have the largest impact on annual total final energy (TFE) reduction for each segment to help
policymakers design a more targeted approach for the promotion of high-impact EEMs. EE system improvements
such as insulation of steam systems and process optimization, as well as EE equipment improvements like process
equipment insulation, installation of variable frequency drives (VFDs), smart controls, and IE3 motors, were
found to be the most profitable options across all clusters. For the electricity cluster, CHP was found to be the
most impactful way of EE improvement (high energy savings) but not cost-effective. For fuel-consuming clusters,
waste heat recovery measures were found to be the most cost-effective and among the most impactful options for
EE improvement, along with fuel substitution measures with varying degrees of cost-effectiveness.

the EU industry, the Swiss industry has experienced a slowdown in the

1. Introduction rate of EE improvement from 2015 to 2020, which may be further
faltered due to the pandemic-induced economic downturn since 2020
1.1. Background (ODYSSEE-MURE, 2021). Based on the sectoral studies conducted for
energy-intensive subsectors (Cement (Zuberi and Patel, 2017), Chem­
Swiss energy and climate policy has successfully managed to icals (Zuberi and Patel, 2018), Metals (Bhadbhade et al., 2019b), and
decouple economic and population growth from energy consumption, Food (Bhadbhade and Patel, 2020)) and cross-cutting technologies
thus, setting an example for other countries with similar economic (EMDS (Zuberi et al., 2017), Waste heat recovery (Zuberi et al., 2018)
structures (IEA, 2018; Bhadbhade et al., 2019a). As part of a set of within the Swiss industry (see Table 1), there is still a substantial amount
measures along with renewable energy integration to achieve the of untapped potential for the EE improvement that can be achieved
climate targets, Swiss energy and climate policies promote Energy effi­ through the implementation of commercially available Energy Effi­
ciency (EE) improvement, which is often dubbed as the first fuel for a ciency Measures (EEMs). Based on the sensitivity analyses presented in
sustainable global energy system (Energy, 2023). Since the Swiss in­ the sectoral studies, the cost-effectiveness of an individual EEM is
dustry is responsible for 19 % (SFOE, 2020a) of total final energy (TFE) influenced by the energy prices and CO2 levy. The establishment char­
consumed in Switzerland and 11% (EEA, 2020) of total CO2 emissions, acteristics, such as size, location, and the fuel mix, in turn, influence the
energy efficiency improvement of Swiss industrial processes is likely to energy prices and CO2 levy paid by the establishment and therefore,
play an important role in achieving the targets set by climate policies. strongly influence the EE improvement-related investment
However, much in line with the trends of EE improvement observed for decision-making process (e.g., the choice between investing in measures

* Correspondence to: Lucerne University of Applied Sciences and Arts (HSLU), Switzerland.
E-mail addresses: Navdeep.bhadbhade@unige.ch, Navdeep.bhadbhade@hslu.ch (N. Bhadbhade).

https://doi.org/10.1016/j.egyr.2023.12.021
Received 22 September 2023; Received in revised form 22 November 2023; Accepted 6 December 2023
Available online 19 December 2023
2352-4847/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

Nomenclature I Initial investment (CHF)

Ia Annualized initial investment cost (CHF)
A Age of equipment at the time of replacement (years) I,er Energy-relevant investment cost investment (CHF)
ANF Annuity factor IRR: Internal rate of return
B Annual benefits of the measure (CHF) L Lifetime of measure y (years)
Clvl, Levelized costs of saved energy by technology or measure y LCSE Levelized cost of saved energy
(CHF cents/kWh) NPV Net present value of measure y (CHF)
CA Annual potential CO2 savings by technology or measure y O&M Annual operations and maintenance costs of measure y
(tCO2 /year) (CHF)
CF Annual cash flow for the year t (CHF) Pe Price of electricity (CHF/GJ)
CHF Swiss Franc Pfi Price of fuel i (CHF/GJ)
ECx Total final energy consumption of establishment x (GJ/yr) PCO2 CO2 price in year t (CHF/t CO2)
EFi Emission factor of fuel i (tCO2/GJ) r Real discount rate
ES Annual energy savings achieved by measure (GJ/yr) RES Relative energy savings for the measure
ES or FS Electricity or Fuel savings reported for the measure t year

related to process heat versus building envelope) (Abadie et al., 2012). Schönenberger, 2019), the most prominent barriers specific to EE in­
Moreover, based on the analysis of investments triggered by voluntary vestment identified for the Swiss industry were the prioritization of
target agreements, EE-relevant investment represented 33% of total other investments over EE-related investments, internal financial con­
investment, whereas the dominating share of 67% of investments were straints such as limited access to capital, low financial attractiveness,
made for the measures where either energy efficiency was not a primary and low energy costs possibly leading to higher hurdle rates for
purpose (e.g., capacity expansion) or where it was necessary to replace EE-related investments. In order to design effective policies to overcome
process equipment because it had reached the end of its lifetime. Thus, the barriers, a detailed analysis of the energy efficiency measure’s
there is a need to facilitate investment in industrial EE improvement. specificity and evaluation from both the company’s and policymaker’s
An extensive body of literature exists discussing the insufficient in­ perspective is required (Cagno and Trianni, 2014). Table 2 summarizes
vestment in industrial energy efficiency (Cagno and Trianni, 2014; the approach, perspective, methods and indicators presented in existing
Backlund et al., 2012; Solnørdal and Foss, 2018; Trianni et al., 2020), literature as well as for the current analysis.
possible barriers to the closure of the resulting energy efficiency gap The present body of literature can be broadly classified into two
from a company and policy perspective (Trianni et al., 2020; Fleiter categories namely, analysis of EE improvement potentials and analysis
et al., 2011; Steve Sorrell and Nye, 2011; Safarzadeh et al., 2020), EE of EE investment decision-making. The studies related to the EE
investment-related decision-making in industry and the influence of improvement potentials rely on quantitative analysis of target agree­
firm characteristics on the EE-related investment decisions (Sandberg ments using indicators such as Levelized costs and payback times.
and Söderström, 2003; Henri and Peter Nijkamp, 2001; Lawrence et al., However, the entire industry or industry sub-sectors are typically
2019, 2018; Bruce Tonn, 2000). The most prominent barriers to the analyzed as monoliths in these studies. In other words, the entire in­
investment in industrial EE improvement identified from the literature dustrial sub-sector is assumed to have similar technical and economic
are imperfect information (such as cost and saving potentials for EEMs), characteristics during the techno-economic analysis of the EEMs. While
limited availability of capital, risk and uncertainty (leading to high the studies related to the EE-related investment decision-making do take
hurdle rates), split incentives, and bounded rationality. According to the establishment characteristics into account, these studies are qualitative
survey conducted in the context of the M-Key project (Cooremans and and rely on the literature survey and interviews. As a result of the

Table 1
Potential for EE improvement in the Swiss industry.
Sector/Technology group Technical potential Technical potential represented by Cost-effective potential Sensitivity analysis results
(based on BAT)a commercial EEMsb
Low energy High energy
prices prices

EE potentials for most energy-consuming subsectors of the Swiss industry

Chemical manufacturing (Zuberi and n.a. 15% 15% 11% 15%
Patel, 2018)
Food and beverage (Bhadbhade and Patel, 25% 18% 14% 7% 16%
2020)
Cement manufacturing (Zuberi and Patel, 20% 18% 14% 7% 15%
2017)
Metals manufacturing (Bhadbhade et al., 19% 16% 11% 9% 12%
2019b)
EE potentials for cross-cutting technologies in the Swiss industry sectorc
Electric motor-driven systems (EMDS) ( n.a. n.a. 17% n.a. n.a.
Zuberi et al., 2017)
Excess heat recovery (Zuberi et al., 2018) n.a. 12% 12% (5% for stricter n.a. n.a.
economic criteria)
a
Overall total technical potential is estimated by comparing current (status quo) specific energy consumption with that of best available techniques (BATs). Per­
centages are presented relative to total the final energy demand of the sector in the year of the study.
b
Technical potential represented by commercial technologies is a cumulative potential of all the commercially available technologies irrespective of cost-
effectiveness.
c
The sector-specific EE potential and potentials for cross-cutting technologies are not mutually exclusive. The percentages are presented relative to the total
electricity demand for EMDS and the total heat demand of the entire industry sector for excess heat recovery.

625
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

Table 2
Summary of literature survey and current research.
The existing Topic Approach Perspective Method Indicators
body of
literature

EE investment decision- Qualitative, empirical analysis Industrial establishment, Literature survey,

making Policy Interviews
EE improvement potentials Quantitative, target agreements Industrial establishment, Discounted cash flow Payback time, Levelized
analysis Industry sub-sector, Policy analysis (Techno-economic cost
analysis)
Current research EE investment decision- Statistical analysis of target Industrial establishment, Discounted cash flow Internal rate of return,
making using techno- agreements specific to industry Policy, Cluster analysis, Machine Levelized cost, Relative
economic indicators clusters learning energy savings

influence of financial, technological, and organizational characteristics results about the entire industry sector and individual clusters. Sub-
of the industrial establishment (i.e., industrial sites), different sets of Section 3.1 presents the results at the level of the entire industry sector.
EEMs are economically and technically viable for different types of in­ Sub-Sections 3.2 to 3.5 present the results for individual clusters of in­
dustrial establishments (DeCanio, 1998). Despite several publications dustrial establishments (identified based on energy mix and size). Sec­
related to the two aforementioned categories, there hasn’t been a study tion 4 concludes the paper.
so far analyzing the techno-economic effectiveness of EEMs based on the
establishment characteristics within the Swiss industry. To facilitate the 2. Methodology
investment in EE improvement of industrial processes, there is a need to
understand the economic and technical effectiveness of EEMs for various 2.1. Data
types of industrial establishments (referred to as clusters henceforth).
Hence, the current study presents a novel approach for estimating the 2.1.1. Techno-economic data from EnAW database
economic and technical effectiveness of the EEMs for various clusters of The EnAW database provides techno-economic data for approx.
industrial establishments that are identified based on the establishment 14,000 EEMs implemented across 12 manufacturing subsectors by
characteristics that are relevant to EE related investment approximately 1100 establishments from 1999 to 2016. The database is
decision-making. These clusters provide an additional dimension of split into two phases, i.e., phase 1 from 1999 to 2012 and phase 2 from
categorization of industrial establishments compared to the traditionally 2013 to 2016. Approximately 72% of establishments can be categorized
used classification based on economic activities. The approach is as large consumers based on their annual electricity consumption (>500
demonstrated using the data from the target agreements collected for MWh per year; above the C4 category of Elcom2). EEMs implemented to
the Swiss industry. Moreover, the EEMs are categorized using a machine save fuel represent approximately 53 % of total cumulative energy
learning algorithm, thus providing a novel approach of automized text savings, while all other EEMs target electricity savings (see Fig. 1). The
mining which is applicable for the classification of EE improvement breakdown of the establishments’ TFE consumption by industry sub-
actions reported in target agreement /energy audits. sector is similar to the national statistics (SFOE, 2020a). The EnAW
database can therefore be reasonably assumed to represent the TFE
consumption profile of the entire Swiss industry sector (see Appendix A
1.2. Aim and objectives for comparison).

Given this background, the main objectives of the present study are:

• From a company perspective, identification of the most profitable

options (without assuming a discount rate) for EE improvement
within each cluster.
• From a policy perspective, identification of the most impactful (see
Section 2.5 for definition) options for EE improvement within each
cluster.

The present study is based on the analysis of a database of energy

efficiency measures implemented via voluntary target agreements un­
dertaken by EnAW1 in the Swiss industry (see Section 2.1.1). The
analysis is conducted to inform decision-makers in the industry about
the profitable EE improvement options by type (cluster) of industrial
establishments as well as to provide policymakers with relevant infor­
mation about the most impactful technology options for the identified Fig. 1. Composition of EnAW database a) Distribution of consumer size based
clusters within the Swiss industry, e.g., as the basis for designing more on total final energy consumption b) Share of EEMs implemented for electricity
targeted subsidy policies (such as competitive tenders). and fuels.
The paper is structured in the following way: Section 2 explains the
methodology and data sources used for the analysis. Section 3 presents
results and discussion, which is further divided into subsection for the

2
Elcom publishes electricity prices for 7 different consumer categories (C1 to
1
EnAW stands for “Energie-Agentur der Wirtschaft” in German or “Energy C7; commercial and industry) for each canton. The categories are based on the
Agency of the Swiss Private Sector” in English. It was set up by the private annual electricity consumption and max required power (28. ElCom, F.E.C.,
sector in order to help companies avoid paying the CO2 levy by entering into Electricity prices in Switzerland. 2020: https://www.strompreis.elcom.admin.ch
the target agreement with the Swiss administration for CO2 emission reduction. /.).

626
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

The database contains techno-economic data such as total initial 2.2. Measure categorization
investment, and annual absolute energy savings along with auxiliary
information such as the year of commissioning, a brief description of the Within the EnAW database, the EEMs are classified into 14 broad
measure, broad classification of EE improvement categories (see Section categories of EE improvement. EEMs are further classified at the level of
2.2.), and the share of EE-relevant investment.3 The investment costs technology (e.g., VFDs, IE3 motors, smart controls etc.) and type of in­
obtained from the EnAW database are converted to CHF 2020 values vestment (e.g., optimization, new equipment or maintenance and
using producer price indices (FSO, 2021) in order to ensure consistency repair) using machine learning algorithms (text mining). In addition, the
with the 2020 energy prices used in the analysis. The EnAW database EEMs are also classified according to the purpose of investment (e.g., EE-
also provides data on the energy consumption by energy carriers for all relevant investments and other investments) using the auxiliary data
participating establishments. The energy consumption data for the first provided in the EnAW database.
year of participation is used for the clustering analysis.
While the average Internal rates of return (IRR) for the EEMs are 2.2.1. Technology classification – machine learning approach
estimated by the combined analysis of both phases, the clustering was A machine learning algorithm (Support Vector Machines) is
performed separately for two phases due to the different formats of employed for further detailed classification of EEMs using the natural
unique identifiers used for establishments in the two phases (see Section language processing library (nltk)4 in python. Support Vector Machines
2.3). was chosen for the classification algorithm as it provides an accuracy of
82%. Other classification algorithms tested were Random Forest Clas­
2.1.2. Data cleaning criteria sifier (68% accuracy), Logistic regression (73% accuracy), Stochastic
The simple payback time was used as a criterion for primary data Gradient Descent classifier (74% accuracy), and Multinomial Naïve
cleaning: EEMs with unrealistic payback time values (>50 years) were Bayes classifier (49% accuracy). The phase 2 database provides the
excluded from the analysis. The energy efficiency measures for which classification at a higher level of detail than phase 1 (although, still at
the investment costs and energy savings are clubbed together for mul­ the level of broad EE improvement categories). This database was used
tiple EE improvement actions are also excluded from the analysis since it as a training dataset and the EEMs from Phase 1 were then classified into
is not possible to attribute energy savings and investment costs to in­ more detailed categories. A training dataset in Machine learning algo­
dividual action. Furthermore, energy efficiency measures with ambig­ rithm is pre-tagged dataset of expected categories. For example, in the
uous measure descriptions were excluded to improve the results of the context of this research, a measure with a description related to a
classification algorithm (see Section 2.2). As a result of data cleaning, compressed air network is tagged as “compressed air”. This dataset is
the dataset includes approximately 8000 EEMs. used to train the algorithm to learn the correlation between keywords in
the measure description and the EE improvement category (or tag). The
2.1.3. Energy prices, CO2 emission factors, CO2 levy and technology keywords are identified during the vectorization step. These identified
lifetime keywords are then used to extract the tags for further detailed equip­
The annual EE improvement-related benefits generated due to the ment and system-related EEM classification. Table 3 provides the list of
implementation of an EEM are estimated as the annual energy costs equipment and system-related EEMs classified within each broad EE
saved through energy savings achieved, thereby using the energy prices improvement category identified using the machine-learning algorithm.
for the year 2020. The average prices paid for natural gas within Broad EE improvement categories represent the typical end uses of in­
Switzerland are monitored for nine standardized consumer categories dustrial energy consumption. An EEM is classified in a certain broad EE
(six household and commercial and three industrial) and are obtained improvement category if its implementation results in the reduction of
from (Federal Department of Economic Affairs, 2020). By analogy, the energy consumption from the corresponding end-use. For example,
prices paid for electricity by the Swiss industry are obtained from Elcom compressors with variable frequency drives (VFD) are categorized under
(ElCom, 2020); refer to Appendix C for details). Prices for other fuels are compressed air systems or process cooling depending on the concrete
obtained from IEA and extrapolated to 2020 based on the methodology end-use.
followed for earlier subsectoral studies (Zuberi and Patel, 2017). EE improvement action is categorized as either EE equipment
The annual benefits from fuel-saving measures also depend on the improvement or EE system improvement. EE equipment improvement is
cost savings generated by CO2 abatement. Annual CO2 abatement- defined as the modification or replacement of a specific equipment (e.g.,
related cost savings are estimated using the CO2 abatement achieved dryer, compressor, lights, etc.) to improve the energy efficiency at the
and the CO2 levy. The CO2 emission factors by fuel according to the process level. EE system improvement is defined as an action taken to
Swiss national inventory report (FOEN, 2021) and emission factor improve the energy efficiency at the system level (e.g., process optimi­
database (IPCC, 2020) are used in combination with fuel savings re­ zation, process heat integration, leakage repairs, insulation etc.). The
ported in the EnAW database to estimate the CO2 abatement achieved by specific classifications as well as the broad classification of the equip­
the implementation of an EEM. The CO2 levy is assumed to be 120 ment and system improvements are identified using the machine
CHF/tonne CO2 (SFOE). The technology lifetimes used in the economic learning algorithm.
analysis are obtained from EnAW for broad categories whereas various
sources are used for more detailed technologies (SFOE, 2020b; ASHARE, 2.2.2. Type of investment
2017). Along with the technical classification, categorization based on the
specific implications of the EE improvement action improves the
decision-making process (Trianni et al., 2020). Based on the descriptions
provided in the database an investment in EE improvement can be made
either to replace or retrofit the existing equipment with new equipment,
to optimize the process by changing process parameters or during the
3 maintenance and the repair of the equipment (also identified by ma­
As defined by EnAW, this represents either i) the share of EE-related cost
used for the replacement of pre-existing equipment or ii) in case of replacement
chine learning).
of larger subsystem of the process (e.g., packaging facility), an expert judge­
ment of what can be considered as additional EE improvement. For i), the EE-
related investment Ier cost is calculated from the total investment cost I by
accounting for the replaced equipment’s age A and its technical lifetime L: Ier = 4
Natural language toolkit is a natural language processing library in python
I × (1 − AL) used for text processing and classification.

627
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

Table 3 investment. These shares along with the measure descriptions are used
Technology-specific categories for EE improvement. to categorize the type of investment. The EEMs with an EE-relevant
EE improvement EE equipment EE system improvements share of investment above 70% and EEMs with EOL replacements
category improvements (where the measure descriptions explicitly mention that more energy-
Lighting • LED lights • Repair and maintenance efficient equipment was purchased) are categorized as EE-
• Motion sensors improvement-related investment, whereas all other investments are
Building envelope • Heat pump for space heat • Insulation of building categorized as EOL and other investments.
envelope
• Window replacement
Compressed air • VFD retrofit for • Pressure optimization 2.3. Clustering analysis (K-means algorithm)
system compressors • Leakage repair and
• Screw compressors maintenance
• VFD on compressor • Waste heat recovery (WHR) Annual shares of energy carriers are estimated for each establish­
• Smart control system for compressor ment in the EnAW database. Overall energy costs paid by the estab­
compressor load lishments depend on the type of energy carrier, which in turn affects the
management cost-effectiveness evaluation of EEMs. In order to facilitate the techno-
Process heat • Condensate return – Heat • Steam network pressure
economic analysis specific to establishment characteristics, they are
-generation exchanger optimization
• Feed water preheat – • Insulation –steam segmented according to the energy mix and their sizes. The establish­
Economizer distribution network ment sizes are categorized according to annual electricity and TFE
• Burner replacement in consumption. Based on the shares of energy carriers in their respective
boiler
TFE consumptions, each establishment is assigned to one of four clusters
• Insulation - boilers
Process heat - • Furnace replacement • Optimization – process using an unsupervised clustering K-means algorithm. The resulting
consumption • Burner replacement in parameters, furnace clusters formed thus represent a group of establishments with one
furnace parameters (temperature, dominant energy carrier within their energy mix (e.g., electricity, nat­
• Process equipment pressure) ural gas, fuel oil etc.; Section 3.2). For example, establishments assigned
upgrade (more efficient • Process heat integration
to electricity cluster consume an average of 75% electricity. The opti­
equipment)
• Process equipment mum number of clusters is determined using the silhouette analysis (see
insulation Appendix D for details). The clustering is primarily carrier out to present
Other core • Process controls • Process scheduling cluster-specific techno-economic analysis (see Section 3.4) of the EEMs
process-related • IE3 motors for process optimization
included in the EnAW database.
mills
Process cooling/ • New refrigeration system • Leakage repair
Refrigeration – Switching to ammonia • Pressure optimization
2.4. Economic effectiveness criteria
or CO2 cooling
• Free cooling
• Compressor with VFD An economic evaluation of an energy efficiency investment can be
Ventilation and • VFD fans • Insulation performed using payback time, Internal rate of return (IRR) or a Lev­
air conditioning • Controllers for fans • Maintenance and repair elized cost of saved energy (LCSE). The simple payback time is the most
• Optimization of air flow
Motors • IE3 motors • Maintenance and repair of
commonly used economic criterion (Qiu et al., 2015; Rolf et al., 2017;
• VFDs motor systems Dobbs, 2009) but it does not take the time value of money into the ac­
• Smart controls for load count. LCSE on the other hand, uses discounted cash flow (DCF) anal­
management ysis, i.e., a discount rate is used to annualize the initial capital cost.
Substitution • Steam production – boiler
However, sector-wide DCF analyses assume a uniform discount rate for
replacement
• Burner replacement all establishments of a given sector. This is often not representing reality
• Process electrification – as each establishment assigns a different risk factor to EE-related in­
Heat pumps vestments, e.g. reflected by implied discount rates that can be as high as
20% to 40% (Qiu et al., 2015; Dobbs, 2009) and to a large extent the
choice of risk factor is driven by the establishment characteristics such
2.2.3. Purpose of investment
as energy mix and sizes.
Energy efficiency gains can accrue from an investment made spe­
Thus, in order to present economic evaluation from the company’s
cifically for energy efficiency improvement i.e., EE-improvement related
perspective, the IRR is used as the criterion to evaluate the profitability
investments or from an investment made for other purposes such as
of the EE-related investment. The IRR is defined as the discount rate at
capacity expansion or productivity improvement (Other investments).
which the net present value (NPV) of the investment becomes zero.
Three types of investments are categorized as EE-improvement-related
Hurdle rates (or minimum acceptable rates of return) are set by the in­
investments: 1) investments made to replace equipment during its
vestment decision-makers for each industrial establishment. In­
technical lifetime, 2) investments made specifically for EE improvement
vestments that exhibit higher IRR values than the set hurdle rates are
(i.e., WHR, process optimization, insulation, etc.), and 3) investments at
accepted. On the other hand, acceptance of only the most profitable
the end of the lifetime (EOL)5 of the equipment to replace it with more
investments implies the existence of high hurdle rates which in turn
energy efficient equipment than the standard option (e.g., compressor
indicate a high risk perceived by the company for the EE improvement-
with VFD vs standard compressor). The EnAW database provides for
related investment. The IRR for an EEM is estimated by finding the
each EEM the share of EE-relevant investment out of the total capital
discount rate at which the NPV becomes zero using the trial-and-error
method and Eq. 3.
In addition, LCSE is used in the current work to present the cost-
5 effectiveness of energy saving measures from the policymaker’s
An equipment can be replaced at the end of its technical life or during its
lifetime. While an investment made to replace the equipment during the life­ perspective based on a uniform assumed discount rate of 15%. LCSE is
time can be considered as EE improvement related investment, an investment estimated using Eq. 1 or Eq. 8, with negative values implying that the
made to replace the equipment at the end of the lifetime is considered as EE investment is cost-effective. A narrow range of LCSE implies more pre­
improvement related only if the existing equipment is replaced by more energy dictable economic performance for the EEM in the given market, thus
efficient option. making the investment safer.

628
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

ANF × NPV By combining Eqs. 1 to 5, the equation for LCSE can be represented
Clvl = (1)
ES as:

Where, Clvl =
I × ANF + O&M − B
(8)
Clvl, = Levelized costs of saved energy by an EEM (CHF cents/kWh). ES
ANF = Annuity factor calculated by Eq. (2). ES = Energy savings reported for the implantation of an EEM.
NPV = Net present value of an EEM calculated by Eq. (3) (CHF).
ES = Final energy saving achieved by implementing an EEM 6 (GJ)
2.5. Technical effectiveness criteria
L
(1 + r) × r
ANF = (2) The technical effectiveness of the EEM is presented as relative energy
(1 + r)L − 1
savings (RES) i.e., the percentage of the total final energy saved relative
And to the total final energy consumption of the establishment for the year of
∑
L implementation of that EEM.
NPV = CF tr × (1 + r)− tr+t
(3)
ES
t=tr RES = (9)
TFEe
Where,
CFtr = Annual cash flow for year t calculated by Eq. (4) (CHF). Where,
r = Real discount rate (%).7 RES = Relative energy savings of an EEM.
L = Technological lifetime of an EEM (in years). TFEe = Total final energy of the establishment e.
tr = running year (e.g., 2020). Depending on the percentage of TFE savings achieved, the technol­
t = year of implementation ogy or the measure are categorized as high-impact (>1% of annual TFE
reduction), medium-impact (0.5% <annual TFE reduction<1%) and
CF tr = I + O&M − B (4) low-impact (<0.5% annual TFE reduction). Fig. 2 presents a simple
workflow of the overall analysis.
Where,
I = Initial investment for implementation of an EEM.
O&M = Annual operation and maintenance cost of an EEM. 2.6. Score normalization
B = Annual net benefits earned by implementing an EEM calculated by
Eq. (5). The average profitability (or IRR) and technical effectiveness (or
relative energy savings; RES) are estimated for each broad EE
∑ improvement category based on the values characterizing each EEM
n
B = (ELS × Pe ) + ( (FSi × Pi )) + (CA × PCO2 ) (5)
i implementation included in that category. The average profitability and
technical effectiveness values for each EE improvement category are
Where, quantified using normalized scores between 1 and 0. The overall score
ELS = Electricity savings achieved by an EEM. for each category is estimated as a weighted score with equal weights
FSi = Saving achieved for fuel i by an EEM. assigned to both profitability and technical effectiveness (50:50).
Pi = Price of fuel i.
Pe = Price of electricity e. 3. Results
PCO2 = CO2 levy.
CA = CO2 abatement potential of an EEM calculated by Eqs. 6 or 7. The following subsections present the results of the analysis of EE
The CO2 emission reduction potential of measures whose main improvement investments organized in the order from the results at the
purpose it is to abate CO2 (i.e., fuel substitution, integration of renew­ level of the entire industry (Section 3.1) to the level of individual
ables) is estimated using Eq. (6).
CA = FS × EFi (6)

Where,
CA = carbon dioxide abatement achieved by an EEM (t-CO2/year).
FS = Fuel savings for the EEM reported in EnAW database (GJ/year).
EFi = CO2 emission factor for fuel i (t-CO2/GJ).
Or
CA = (FCx × EFx − FCy × EFy ) (7)

Where,
EFx = CO2 emission factor for fuel x (x = fuel oil, natural gas, coal) (t-
CO2/GJ).
EFy = CO2 emission factor for fuel y (y = natural gas, industrial waste,
district heating) (t-CO2/GJ).
FCx = Consumption of fuel x (GJ/year).
FCy = Consumption of fuel y (GJ/ year).

6
Energy savings for substitution measures are estimated as net energy sav­
ings and used for estimation of technical effectiveness.
7
A discount rate of 15% is chosen for estimating LCSE from the private
perspective. Fig. 2. Simplified workflow of overall analysis.

629
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

Table 4
Overall rankings for broad EE improvement categories.8
Overall rank EE improvement category Profitability Technical Effectiveness Economic rank Technical rank

1 Process heat 58% 1.5% 1 4

2 Substitution 37% 2.6% 8 2
3 Other process related 51% 1.6% 5 3
4 HVAC 57% 1.1% 2 5
5 Space heating 52% 1.0% 4 6
6 Process cooling 48% 1.0% 6 7
7 Power generation 16% 2.7% 11 1
8 Compressed air 53% 0.5% 3 9
9 IT and appliances 39% 0.4% 7 10
10 Other motors 36% 0.3% 9 12
11 Lighting 25% 0.3% 10 11
12 Building envelop 16% 0.7% 12 8

clusters (sub-Sections 3.2–3.5). identified with the aid of the k-means algorithm based on the dominant
energy carrier in the energy mix of the establishment. Fig. 3 presents the
3.1. Ranking of energy efficiency investment categories based on average composition of the energy mix for a typical establishment
normalized scores within the resulting clusters. The clusters are named after the dominant
energy carrier used by the cluster. On average, 75% of the TFE demand
Table 4 presents the overall rankings for broad EE improvement of the establishments within the electricity cluster is met with elec­
categories at the level of the entire industry based on the methodology tricity. By analogy, natural gas and fuel oil represent, on average, 67%
outlined in Section 2.6. The EE improvement categories are arranged in and 69% of the TFE demand of the respective clusters. The “other energy
ascending order of the overall rank which results in the EE improvement carriers cluster” comprises establishments that rely on energy carriers
category with the highest weighted score at the top of the table. In such as district heating, propane, butane, diesel and renewables (approx.
addition to the overall ranks, the table also presents economic ranks 70%). Process heat is the most dominant energy-consuming end-use in
(based on average IRR for particular category) and technical ranks all the clusters, representing the largest share within the “other energy
(based on average RES for particular category) for each category. Green carriers cluster” (69% of TFE) and the smallest share (57%) in the
cells represent EE improvement categories with the highest average electricity cluster (Fig. 4). EMDS are the second most important type of
profitability (>40%) and high-impact measures (see Section 2.5 for end use, with final energy use between 19% and 24%.
definition). In contrast, red cells represent EE improvement categories
with the lowest average profitability (<20%) and low-impact measures, 3.3. Investment patterns by cluster
and yellow and orange cells represent the EE improvement categories
with medium average profitability (between 40% and 20%) and Fig. 4 displays the results of the analysis of past investment prefer­
medium-impact measures. According to the combined score for tech­ ences in various EE improvement categories for the establishments
nical effectiveness and economic attractiveness, EE improvements in the belonging to four identified clusters in the Swiss industry. These results
category of process heat rank first (first column). Improvement in pro­ highlight the investment trends identified so far for each establishment
cess heat demand is most commonly realized by process heat integra­ cluster. The establishments belonging to electricity cluster preferred to
tion, heat pump integration, replacement of process equipment or improve energy efficiency by investing most prominently in compressed
through measures like insulation of process equipment, process air systems (on average 12% of total investment), building envelope (on
parameter optimization and process scheduling optimization (see average 12% of total investment) and process heat-related EE
Table 3). While being technically most effective (i.e., capable of improvement measures (on average 12% of total investment). In the
achieving largest annual TFE reduction), the investment in power gen­ natural gas cluster, establishments preferred to invest most prominently
eration measures and fuel substitution measures are amongst the least in process heat related EEMs (on average 19% of total investment),
profitable ways of improving the energy efficiency of the establishment while the companies falling into the fuel oil cluster had a preference for
due to very high capital costs (refer to appendix E for capital costs). On process heat and other process-related measures as well as the invest­
the other hand, fuel substitution ranks very well (2nd) according to the ment in fuel substitution measures (amongst all clusters highest in the
combined score, followed by Other process-related measures as well as fuel oil cluster). Despite being technically and economically attractive,
Ventilation and air conditioning. The lighting and building envelope- EEMs associated with power generation and process cooling (see
related EE improvement measures rank lowest based on the combined Table 4) were the least preferred EE improvement-related investments
scores. The EEMs involved in these categories are some of the least by the establishment that belong to electricity cluster. EEMs associated
profitable investments as well as have a relatively lower impact in terms with the building envelope and with lighting were among the most
of overall energy savings achieved. However, lighting and building preferred investments despite their low score for both economic and
envelop measures experienced rather large investment shares within the technical criteria (see Table 4).
Swiss industry compared to some of the other more profitable and Optimization measures are, on average, more profitable than
technically effective ways of EE improvement (see below, Fig. 5), indi­ investing new equipment for all clusters (see Fig. 6). Nevertheless, es­
cating lower hurdle rates for EE improvement related to these tablishments belonging to all clusters except for the other energy car­
categories. riers cluster preferred to invest in new equipment (see Fig. 7). Analysis
of the purpose of investment (based on EE-relevant share), however,
3.2. Clustering indicates that only one-third of all cases contributed clearly to EE
improvement, and two-thirds (“EOL or other investments”) may be
The establishments in the EnAW database are segmented in 4 clusters considered as standard equipment (see Fig. 8). As a caveat, it should be
noted that EEMs for which the descriptions provided in the database are
vague were not included in the percentage of EE-relevant investments
8
Profitability and technical effectiveness columns represent average IRR and and that the economic energy savings potential estimated in this analysis
RES values for each category. (Fig. 8) may therefore be underestimated. A very small percentage of

630
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

Fig. 5. Shares of investment within each cluster for EE improvement categories. (Abbreviations: BE - Building envelop, CA - Compressed air, VAC – Ventilation and air
conditioning, IT & EA – IT and electrical appliances, OM – Other motors, OPM - Other process measures, PG - Power generation, PH - Process heat, PC - Process cooling, SH -
Space heating, Subst. – Fuel substitution).

Fig. 3. Average energy mix compositions of various clusters.

investments is related to repair and maintenance measures (Fig. 7). economic behavior.

3.4.1. Electricity cluster

3.4. Most profitable and impactful technologies and measures for EE The EE-related investments in this cluster are typically characterized
improvement for clusters9 by the low or medium-impact on annual TFE reduction (e.g., lighting,
building envelope and ventilation) with few exceptions of investment in
Sections 3.1–3.4 presented the results based on the past trends of EE- high-impact measures like power generation, fuel substitution and heat
related investments in the Swiss industry. In contrast, the present section pumps (Table 4). The most profitable EE investment options for the
identifies technically and economically the most impactful EEMs based electricity cluster are presented in Table 5. Installation of ventilation
on aforementioned economic and technical for all identified clusters. controls was found to be the most profitable investment for EE
The profitable options to achieve EE improvement (presented in tables) improvement in the electricity cluster based on its IRR. Pressure opti­
are ranked based on the average IRR, and they offer guidance to mization and leakage repairs in compressed air networks closely follow
decision-makers in the industry about the most profitable ways of ventilation controls and can generate returns that are significantly above
improving EE for a company that belongs to a particular cluster. The plausible risk-adjusted hurdle rates (typically reported to be in the range
distribution of LCSE for the most impactful EE improvement options is of 20% to 40% (Qiu et al., 2015); Dobbs, 2009). However, these in­
presented as box plots for all identified clusters to help policymakers vestments are medium or low-impact investments i.e., together, they
design a more targeted approach to reduce the uncertainty in their potentially save only up to 1% of annual TFE consumption. Replacing
the chillers with a new, more efficient chiller in the process cooling
system is one of the profitable as well as a high-impact investment (1%
9
Abbreviations used in this section: EE- Energy efficiency, EEM- Energy ef­ TFE savings by this measure only). Installing VFDs for the exhaust fans
ficiency measures, IRR- Internal rate of return, RES – Relative energy savings, in the ventilation systems (along with control systems for ventilation)
TFE – Total final energy, VFD – Variable frequency drive, CHP – Combined heat results in optimum speed of the fan and approximate TFE savings of
and power.

631
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

100% 3%
5% 4% 5%

90%
24% 24% 19%
24%
80%
3%
70% 5% 6%
7%
60%
Other processes
50% EMDS
Ligh�ng, HVAC and appliances
40%
61% 61% 69% Process heat
57%
30% Space heat and warm water

20%

10%
6% 6% 7% 4%
0%
Electricity cluster Natural gas Fuel oil cluster Other carriers
(n=360) cluster (n=377) (n=197) cluster (n=99)

Fig. 4. Shares of end use consumption of final energy within each cluster (estimated using Prognos,2021).

Other energy carriers cluster

Fuel oil cluster

Natural gas cluster

Electricity cluster

0% 20% 40% 60% 80% 100%

Average profitability

Repair and maintenance Optimization New equipment

Fig. 8. Purpose of investment by cluster.
Fig. 6. Average profitability of investment types for various clusters.

p.a. of the TFE consumption of the establishment (presented below the

about 0.5% per year at 55% profitability. bars in Fig. 9), power generation via CHP was found to be the most
Fig. 9 presents the range of LCSE for the most impactful EEMs impactful way of reducing TFE demand for the electricity cluster.
identified in the electricity cluster. With a potential to reduce up to 9.4% However, this technology is capital-intensive and its cost-effectiveness
depends on the quality of the available excess heat as well as the
installed capacity of the CHP unit (Zuberi et al., 2018), leading to a wide
distribution of LCSE values implying less predictable economic perfor­
mance. Fuel substitution for steam production, typically replacing
oil-fired boilers with natural gas boilers also represents a substantial
potential (2.2% TFE savings p.a.) to reduce the TFE consumption of the
establishments in the electricity cluster. Despite being relatively
capital-intensive and often unprofitable (depending on the replacement
carrier), a considerable number of instances of implementation can be
observed for this measure within the EnAW database. Investing in heat

Table 5
Most profitable EEMs in electricity cluster (n = 304).
Energy efficiency measures Profitability Technical Effectiveness

Ventilation controls 73% 0.4%

Compressed air network optimization 70% 0.6%
Compressed air network leakage repair 70% 0.4%
New chillers 58% 1.0%
VFD for fans 55% 0.5%
Fig. 7. Type of investment by cluster.

632
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

Fig. 10. Distribution of levelized cost of saved energy for most impactful EEMs
Fig. 9. Distribution of Levelized cost of saved energy for most impactful EEMs for Natural gas cluster (n = 232; Horizontal lines=Medians, dots= mean values;
for electricity cluster (n = 141; Horizontal lines=Medians, dots= mean values; the percentages represent average annual TFE savings relative to the TFE de­
the percentages represent average annual TFE savings relative to the TFE de­ mand for each EEM category).
mand for each EEM category).

pump integration was found to be one of the most impactful (3.3% final Table 6
energy savings p.a.) as well as cost-effective ways to reduce the TFE Most profitable EEMs in natural gas cluster (n = 285).
consumption for the establishments falling into the electricity cluster. Energy efficiency measures Profitability Technical Effectiveness
The majority of the establishments included in the electricity cluster are Dryer modifications 72% 1.7%
categorized as large consumers (tariff C4 and higher categories), which Process optimization 65% 0.9%
pay lower electricity prices, making it more likely that the heat pump Process equipment insulation 64% 0.5%
Steam network leakage repair 62% 0.2%
will be cost-effective. Investments in compressor improvements like VFD
Steam pipes insulation 59% 0.5%
retrofit and smart controllers were found to be medium-impact (0.6%
and 0.5% final energy savings p.a. respectively) and cost-effective ways
of reducing TFE consumption within the electricity cluster. The average the natural gas cluster. However, due to its limited application (only
impact of IE3 motors10 was low (0.3% final energy savings). However, a within the subsectors that employ evaporation processes e.g., food and
large spread for the savings achieved by IE3 motor implementation was beverage) and very high initial investment costs, there were very few
observed, ranging from 1.5% to 0.1%, depending on both the end-use instances of implementation of this technology in the EnAW database.
where the technology was implemented and the size of the motor. The Substitution of fuel for steam generation (typically from fuel oil to
highest savings were achieved for upgrading the rolling mill with an IE3 natural gas or to district heating) can on average, save 2.2% of annual
motor. TFE demand. However, these measures exhibit a wide range of LCSE
which is largely influenced by the fuel which is being substituted. Based
3.4.2. Natural gas cluster on the subsectoral composition of the natural gas cluster, drying oper­
The typical EE-related investments in natural gas cluster are char­ ations are highly energy intensive end-uses. Modifications in the drying
acterized by high and medium-impact with relatively low investment in operation such as replacing part of heat demand for drying by a me­
low-impact EEMs (see Fig. 10). Table 6 presents the most profitable EE chanical press or by the introduction of silica gel can result in approxi­
investment options for the natural gas cluster. Modification of drying mately 1.7% p.a. reduction in TFE consumption for the natural gas
operation in various subsectors was found to be the most profitable as cluster establishments in a cost-effective manner. Process heat integra­
well as a high-impact options for EE improvement. While resulting in a tion was found to have a significant impact on the annual TFE savings
medium-impact for TFE savings, Process optimization measures were (2.0% p.a.) and utilization of waste heat for preheating the boiler feed
amongst the most profitable investments due to the rather low initial water and recovering waste heat from compressors for the production of
investment costs. Replacing or improving insulation for process equip­ hot water are also cost-effective ways of reducing TFE for the natural gas
ment and steam pipes were found to have a medium-impact in terms of cluster, albeit with slightly lower impacts on annual TFE consumption.
energy savings but they exhibit IRRs well above the typical risk-adjusted
hurdle rates. Due to very low investment costs and low complexity, 3.4.3. Fuel oil cluster
leakage repair within the steam network was found to be one of the most The fuel oil cluster is mainly characterized by investments in high-
profitable investments but it has a low-impact on annual TFE savings for impact and capital-intensive measures (see appendix B for capital
the natural gas cluster. costs) like substitution as well as cost-effective measures like process
Fig. 10 presents the range of LCSE for most impactful EEMs falling
into the natural gas cluster. Replacing part of heat demand for evapo­
ration by reverse osmosis was found to be the most impactful way of Table 7
reducing TFE energy consumption (3.9% p.a.) for the establishments in Most profitable EEMs in fuel oil cluster (n = 199).
Energy efficiency measures Profitability Technical Effectiveness

Steam system insulation 87% 1.7%

10
IE represents an international classification of the energy efficiency of IE3 motors for boiler 78% 0.3%
motors. IE1 represents standard efficiency (90%). IE2 represents high efficiency Dryer modification 63% 3.3%
Process optimization 58% 4.7%
(92%). IE3 presents premium efficiency (95%) and IE4 represents super pre­
Process equipment insulation 54% 1.7%
mium efficiency (97%).

633
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

heat integration. Table 7 presents the most profitable EE investment Table 8

options for the fuel oil cluster. Insulation of steam pipes was found to be Most profitable EEMs in the cluster of Other energy carriers (n = 106).
the most profitable as well as a high-impact investment (1.7% p.a. final Energy efficiency measures Profitability Technical Effectiveness
energy savings) for the fuel oil cluster. Installing IE3 motors in the boiler
Steam system insulation 78% 0.7%
house contributes low final energy savings (0.3% p.a.) but was found to Ventilation WHR 68% 0.7%
be one of the most profitable investments. Modifications of dryer along Process control 62% 0.8%
with the process optimization were found to be highly profitable as well Heat pump 51% 0.6%
as very impactful measures (3.3% p.a. and 4.7% p.a. TFE savings Process optimization 50% 0.6%

respectively) for the fuel oil cluster. Similar to other clusters, process
equipment insulation as a result of relatively low initial investments was 3.4.4. Other energy carriers cluster
found to be a very profitable investments for the fuel oil cluster. Similar to the fuel oil cluster, typical EE-related investments associ­
Fig. 11 presents the range of LCSE for the most impactful EEMs ated with this cluster include high-impact and expensive options like
identified in the fuel oil cluster. The impact of the options displayed in substitution and cost-effective options like process heat integration.
Fig. 11 was found to be much higher compared to other fuel consuming- Table 8 presents the most profitable EEMs for the cluster of Other energy
clusters, possibly due to relatively inefficient process heat use. Amongst carriers. Insulating the steam pipes was found to be the most profitable
the high-impact EEMs, process heat integration was found to be the most investment with a medium-impact on the annual TFE reduction. In terms
impactful (5.4% p.a.) as well as one of the most cost-effective ways of of profitability, this is followed by medium to low-impact investments
reducing TFE consumption for the establishments in the fuel oil cluster. like installing process control, waste heat recovery from ventilation
Boiler modifications such as installation of economizers and condensate systems, heat pumps for space heating, and process optimization. All
recovery were found to also have a remarkable technical effectiveness, these options are characterized by profitability levels that are consid­
namely of 5.1% p.a. However, these technologies were observed to be erably above the typical risk-adjusted hurdle rates reported for EE-
unprofitable in few cases, especially for smaller boiler sizes (boiler related investment decisions (Qiu et al., 2015; Dobbs, 2009).
output below 6000 kW). Fuel substitution technologies (typically from Fig. 12 presents the range of LCSE for the most impactful EEMs
fuel oil to natural gas) offer an annual TFE saving potential of approx­ falling into the Other energy carriers cluster. Waste heat recovery
imately 3.8% but they can imply very high capital costs (see appendix B measures like process heat integration and WHR from compressors were
for capital costs), therefore representing a relatively less cost-effective found to be amongst the most impactful measures in terms of annual
way of reducing the TFE consumption amongst the other high-impact energy savings achieved (5.3% and 0.8% respectively), with all cases
options for the fuel oil cluster. However, as a result of the high CO2 being economically attractive (exclusively negative cost of at least − 2.7
intensity of fuel oil, the fuel substitution measures were identified as CHF cents/kWh). This was generally also the case for the optimization of
more cost-effective compared to the electricity or the natural gas cluster. furnace parameters (temperature, pressure) and process equipment
It should also be noted that a rather large distribution of LCSE values was insulation with relatively few instances of these measures showing lower
observed for the substitution in the fuel oil cluster implying less pre­ cost-effectiveness, however mostly still offering profitable operation; the
dictable economic performance. The fuel substitution is carried out by associated energy savings were found to be still significant (1.5% p.a.
either replacing the burners or replacing the entire boiler. While having and 1% p.a. energy savings respectively). Finally, while being less cost-
a similar effect on annual TFE savings, the boiler replacement is less effective on average (more cases of net positive costs) compared to other
cost-effective than replacing the burner. Similar to other clusters, high-impact options of this cluster, fuel substitution (typically from
however, investment in fuel substitution technologies was quite popular propane, butane, or diesel to natural gas) was identified as the second
within the fuel oil cluster due to their high potential for CO2 abatement. most impactful option (2.8% p.a. final energy savings).
Replacement of furnace burners was also one of the least cost-effective
but high-impact options for reducing TFE consumption. Process opti­
mization and dryer modifications were found to be relatively cost- 3.5. Levelized cost of saved energy for select electricity-saving
effective amongst the high-impact EE-saving investments for the fuel technologies for small consumers
oil cluster.
The vast majority of the establishments included in the EnAW
database are characterized as large consumers based on their annual
electricity demand. However, according to the electricity tariff data
provided by Elcom, large consumers pay relatively low tariffs as
compared to small consumers. Evidence from the EnAW database
(average IRRs for large consumers 15% and for small consumers 25%)
and previously published studies (Cagno and Trianni, 2014; Qiu et al.,
2015; Dobbs, 2009) show that small consumers tend to be more
risk-averse and that they (implicitly) apply higher discount rates when
assessing the cost-effectiveness of investments. As a result, the
cost-effectiveness of electricity-saving technologies is influenced by the
size of the establishment. Table 9 provides the average LCSE for selected
electricity-saving technologies (across all sectors) for both large and
small consumers. Since the EnAW database does not contain a suffi­
ciently large sample size of EE improvements implemented by small
consumers, the LCSEs for the small consumers are estimated using the
typical electricity prices paid by small consumers and initial capital
cost11 and energy savings data for the measures reported by the large

Fig. 11. Distribution of levelized cost of saved energy for most impactful EEMs
11
for fuel oil cluster (n = 210; Horizontal line – Medians, Black dots: Mean values; Initial capital cost for EEMs like heat pump, compressors and IE3 motors,
the percentages represent average annual TFE savings relative to the TFE de­ capital cost for smaller capacities were assumed to represent the capital costs
mand for each EEM category). for small consumers.

634
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

each cluster (however, such analyses are available from the previous
work of the authors for selected subsectors of Swiss manufacturing;
(Zuberi and Patel, 2017, 2018; Bhadbhade et al., 2019b; Bhadbhade and
Patel, 2020). While it was possible to validate results for the Levelized
costs of saved energy for various EEMs by comparing the results of the
current study with other sub-sectoral studies from the literature (Zuberi
et al., 2017, 2018), the profitability of the various EEMs by energy de­
mand clusters could not be validated due to the lack of comparable
studies in the present body of literature. In addition to economic benefits
generated through the reduction of TFE demand and related CO2
emissions, EE improvement can also potentially result in other social (e.
g., alleviating energy poverty, improving quality of life), environmental
(e.g., reduction in local air pollution) and economic benefits (e.g.,
impact on innovation of new technologies, employment generation,
impact on GDP) (Reuter et al., 2020). Monetary quantification and in­
clusion of these benefits in the analysis of the Levelized cost of saved
energy will make some of the capital-intensive EEMs economically more
Fig. 12. Distribution of levelized cost of saved energy for most impactful EEMs attractive. Although the inclusion of some of the multiple benefits (e.g.,
for other energy carriers cluster (n = 92; Medians, Black dots: Mean values; the local air pollution reduction and employment) is relatively straightfor­
percentages represent average annual TFE savings relative to the TFE demand ward using simple multipliers, monetary quantification of other multi­
for each EEM category). ple benefits such as health impact or innovation impact warrants
separate analyses. As a result, due to the lack of data, multiple benefits
could not be included in the present analysis.
Table 9 Despite its limitations, the current study presents a framework for
Levelized costs of saved energy for electricity saving technologies for large and providing guidance to industrial decision-makers and policymakers for
small consumers (mean values).
short-term industrial energy efficiency improvement investment de­
Levelized cost of saved energy (CHF cents/kWh) cisions through a targeted approach based on the establishment char­
Energy Efficiency Measures Large consumers Small consumers acteristics such as energy carrier consumption mix and size of the
establishment. Given the availability of data similar to the EnAW data­
Heat pump -11.35 3.71
VFD for compressors -6.01 -1.9 base, such analysis can be performed for other countries or at the
VFD for fans -5.16 -2.37 regional levels. This may lead to a more solid understanding and more
Controls for compressor -3.85 1.92 robust conclusions for decision-makers in the industry sector and the
Chillers -3.66 2.68
energy, climate, and innovation policy domain.
IE motors -3.23 9.27
Free cooling -1.59 3.66
LED lights 8.64 14.46 5. Conclusions

This study employs K-means clustering to categorize Swiss

consumers. In order to replicate the decision-making at the establish­ manufacturing establishments into four energy consumption clusters
ment level, the average profitability of the investments made by small and identifies the most profitable and the most technically effective
consumers within the EnAW database was chosen as the discount rate energy efficiency measures (EEMs) for each. The clustering approach
(25%) for cost-effectiveness analysis. Except for VFD installations for provides a new dimension of categorization of the industrial establish­
fans and compressors, none of the technologies displayed in Table 7 ments compared to traditionally used classifications based on economic
appear to be cost-effective for small consumers. This is a consequence of categories. Based on the analysis of energy consumption by end uses, the
the higher discount rate assumed for small consumers, which over­ process heat demand emerges as a key energy consumer across all
compensates for the effect of the higher fuel prices that they pay clusters. The majority of establishments opt for new equipment in­
compared to large consumers. While heat pump integration was found to vestments, however only at the end of their lifetime or for a purpose
be the most cost-effective technology for large consumers, it results in other than EE improvement. While fossil fuel cluster establishments
net costs from the perspective of small consumers. The installation of have to rely more on capital-intensive options for EE improvement, the
LED lights was neither found to be cost-effective for large nor for small electricity cluster benefits from smaller, less capital-intensive measures.
consumers despite being one of the most widely implemented The most profitable and effective EE improvement options vary
technologies. significantly across the clusters. Profitable EEMs, such as insulation,
process optimization, smart controls, VFDs, and IE3 motors exhibit high
4. Discussion internal rates of return, making them economically viable even for the
most risk-averse establishments in Switzerland. The most technically
By offering a ranking of the most attractive EE improvement options, effective EEMs have a wider range of the Levelized costs of saved energy
the study offers first guidance in pinpointing EE options for any given for the electricity cluster compared to the fossil fuels clusters. This im­
establishment. The first step is to assign the establishment to one of the plies, the EEMs for fossil fuels cluster exhibit relatively more stable
four clusters based on its energy mix. Once this is known, the most economic performance and are typically safer investment options. EE
promising EE options in terms of energy savings and/or cost- improvements related to process heating, especially waste heat utiliza­
effectiveness can be identified based on the tables and graphs pre­ tion, prove both cost-effective and technically impactful across all
sented in this paper. Both decision-makers in companies and policy- clusters. Due to the high CO2 intensity and the resulting high CO2 levy
makers should ensure that especially cost-effective measures with high payments associated with fuel oil and other energy carriers such as
to medium-impact in terms of energy savings are implemented. propane, butane, and diesel, the process heat and steam-production
Although the present study identifies the most profitable and most related EEMs (such as WHR and substitution) appear more cost-
impactful options for EE improvement by cluster, data limitations made effective, more impactful, and more widely adopted for fuel oil and
it impossible to quantify the current levels of the uptake of the EEMs for the Other energy carriers cluster as compared to the natural gas cluster

635
N. Bhadbhade and M.K. Patel Energy Reports 11 (2024) 624–636

despite their high capital costs. Despite higher electricity prices for Bhadbhade, N., Zuberi, M.J.S., Patel, M.K., 2019b. A bottom-up analysis of energy
efficiency improvement and CO2 emission reduction potentials for the swiss metals
smaller consumers, electricity-saving technologies are less cost-effective
sector. Energy 181, 173–186.
due to inherently high discount rates implemented. Bruce Tonn, M.M., 2000. Industrial energy e$ciency decision making. Energy Policy 28.
This study provides a framework for guiding industrial decision- Cagno, E., Trianni, A., 2014. Evaluating the barriers to specific industrial energy
makers and policymakers in short-term energy efficiency investments efficiency measures: an exploratory study in small and medium-sized enterprises.
J. Clean. Prod. 82, 70–83.
based on establishment characteristics. Despite limitations, the Cooremans, C., Schönenberger, A., 2019. Energy management: A key driver of energy-
approach offers valuable insights and could be replicated using similar efficiency investment? J. Clean. Prod. 230, 264–275.
data in other countries or regions, helping to achieve more informed DeCanio, S.J., 1998. The efficiency paradox: bureaucratic and organizational barriers to
profitable energy-saving investments. Energy Policy 26 (5).
decision-making. Dobbs, I.M., 2009. How bad can short termism be?—A study of the consequences of high
hurdle discount rates and low payback thresholds. Manag. Account. Res. 20 (2),
CRediT authorship contribution statement 117–128.
EEA, E.E.A., Greenhouse gas emissions database. 2020.
ElCom, F.E.C., Electricity prices in Switzerland. 2020: https://www.strompreis.elcom.
Navdeep Bhadbhade: Conceptualization, Data analysis, original admin.ch/.
draft preparation Martin K. Patel: Supervision and reviewing the Energy, S.F.O.o. Initial package of measures.
Federal Department of Economic Affairs, E.a.R.E., Gas price monitoring. 2020: http://gasp
article. reise.preisueberwacher.ch/web/index.asp.
Fleiter, T., Worrell, E., Eichhammer, W., 2011. Barriers to energy efficiency in industrial
Declaration of generative AI and AI-assisted technologies in the bottom-up energy demand models—A review. Renew. Sustain. Energy Rev. 15 (6),
3099–3111.
writing process
FOEN, F.O.f.t.E., National Inventory Report: Switzerland’s Greenhouse Gas Inventory
1990–2019. 2021.
During the preparation of this work the author(s) used a Grammarly FSO, F.S.O., Producer and Import Price Index, Price Index of Total Supply, Total. 2021: htt
in order to make sure grammatical accuracy throughout the manuscript. ps://www.bfs.admin.ch/bfs/en/home/statistics/prices/producer-prices-import-pric
es.assetdetail.21424933.html.
After using this tool/service, the authors reviewed and edited the con­ Henri, L.F. de Groot, Peter Nijkamp, E.T.V., 2001. Energy saving by firms: decision-
tent as needed and take full responsibility for the content of the making, barriers and policies. Energy Econ. 23.
publication. IEA, I.E.A., Energy policy of IEA countries - Switzerland 2018. 2018, IEA.
IPCC, I.P.o.C.C., EFDM Emission factor database. 2020.
Lawrence, A., et al., 2019. Effects of monetary investment, payback time and firm
Declaration of Competing Interest characteristics on electricity saving in energy-intensive industry. Appl. Energy 240,
499–512.
Lawrence, A., Karlsson, M., Thollander, P., 2018. Effects of firm characteristics and
None of the authors of this paper has a financial or personal rela­ energy management for improving energy efficiency in the pulp and paper industry.
tionship with other people or organizations that could inappropriately Energy 153, 825–835.
influence or bias the content of the paper. ODYSSEE-MURE, ODYSSEE database (subscription based access). 2021.
Qiu, Y., Wang, Y.D., Wang, J., 2015. Implied discount rate and payback threshold of
energy efficiency investment in the industrial sector. Appl. Econ. 47 (21),
Data Availability 2218–2233.
Reuter, M., et al., 2020. A comprehensive indicator set for measuring multiple benefits of
energy efficiency. Energy Policy 139.
The data that has been used is confidential.
Rolf, Iten, et al., 2017. Management as a Key Driver of Energy Performance. Swiss
National Science Foundation (SNSF), National Research Programme “Managing
Acknowledgments Energy Consumption” (NRP 71). Infras, Université de Neuchaˆtel, Impact Energy,.
Safarzadeh, S., Rasti-Barzoki, M., Hejazi, S.R., 2020. A review of optimal energy policy
instruments on industrial energy efficiency programs, rebound effects, and
The authors would like to thank the Energy Agency of the Swiss government policies. Energy Policy 139.
Private Sector (Energieagentur der Wirtschaft, EnAW) for providing Sandberg, P., Söderström, M., 2003. Industrial energy efficiency: The need for
valuable information and access to the database. This analysis was investment decision support from a manager perspective. Energy Policy 31 (15),
1623–1634.
funded by the Chair for Energy Efficiency at University of Geneva, and it SFOE, S.F.O.o.E., Energieverbrauch in der Industrie und im Dienstleistungssektor. 2020a.
was also supported by the programme SWEET-DeCarbCH (Decarbon­ SFOE, S.F.O.o.E. Exemption from the CO2 levy subject to a reduction commitment: step by
isation of Cooling and Heating in Switzerland). step.
SFOE, S.F.O.o.E., CO₂-Abgabebefreiung ohne Emissionshandel. 2020b.
Solnørdal, M., Foss, L., 2018. Closing the energy efficiency gap—A systematic review of
Appendix A. Supporting information empirical articles on drivers to energy efficiency in manufacturing firms. Energies 11
(3).
Steve Sorrell, A.M., Nye, Sheridan, 2011. Barriers to Industrial Energy Efficiency: a
Supplementary data associated with this article can be found in the Literature Review. United Nations Industrial Development Organization (UNIDO).
online version at doi:10.1016/j.egyr.2023.12.021. Trianni, A., Accordini, D., Cagno, E., 2020. Identification and categorization of factors
affecting the adoption of energy efficiency measures within compressed air systems.
Energies 13 (19).
References
Zuberi, M.J.S., et al., 2018. Excess heat recovery: an invisible energy resource for the
Swiss industry sector. Appl. Energy 228, 390–408.
Abadie, L.M., Ortiz, R.A., Galarraga, I., 2012. Determinants of energy efficiency Zuberi, M.J.S., Patel, M.K., 2017. Bottom-up analysis of energy efficiency improvement
investments in the US. Energy Policy 45, 551–566. and CO 2 emission reduction potentials in the Swiss cement industry. J. Clean. Prod.
ASHARE, Ashare Life Expenctancy chart. 2017. p. www.ashrae.org. 142, 4294–4309.
Backlund, S., et al., 2012. Extending the energy efficiency gap. Energy Policy 51, Zuberi, M.J.S., Patel, M.K., 2018. Cost-effectiveness analysis of energy efficiency
392–396. measures in the Swiss chemical and pharmaceutical industry. Int. J. Energy Res. 43
Bhadbhade, N., et al., 2019a. The evolution of energy efficiency in Switzerland in the (1), 313–336.
period 2000–2016. Energy. Zuberi, M.J.S., Tijdink, A., Patel, M.K., 2017. Techno-economic analysis of energy
Bhadbhade, N., Patel, M.K., 2020. Analysis of energy efficiency improvement and carbon efficiency improvement in electric motor driven systems in Swiss industry. Appl.
dioxide abatement potentials for Swiss Food and Beverage sector. Resour., Conserv. Energy 205, 85–104.
Recycl. 161.

636