at SciVerse ScienceDirect

Energy 40 (2012) 84e99

Contents lists available

Energy

journal homepage: www.elsevier .com/locate/energy

Energy efficiency in the German pulp and paper industry e A model-basedassessment of saving potentials

Tobias Fleiter a,*, Daniel Fehrenbach b, Ernst Worrell c, Wolfgang Eichhammer a

a Fraunhofer Institute for Systems and Innovation Research, Breslauer Strasse 48, 76139 Karlsruhe, Germanyb European Institute for Energy Research, Emmy-Noether-Str. 11, 76131 Karlsruhe, GermanycCopernicus Institute of Sustainable Development, Utrecht University, Heidelberglaan 2, 3584 CS Utrecht, The Netherlands

a r t i c l e i n f o

Article history:Received 29 August 2011Received in revised form20 January 2012Accepted 14 February 2012Available online 17 March 2012

Keywords:Energy efficiencyPulp and paperBottom-upEnergy saving potentialsEnergy-efficient technologiesConservation supply curves

* Corresponding author. Tel.: þ49 721 6809 208; faE-mail addresses: Tobias.Fleiter@isi.fhg.de (T. F

(D. Fehrenbach), E.Worrell@uu.nl (E. Worrell), Wol(W. Eichhammer).

1 We define energy efficiency as improvemenconsumption of particular energy services (e.g. produ

0360-5442/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.energy.2012.02.025

a b s t r a c t

Paper production is an energy-intensive process and accounted for about 9% of industrial energy demandin Germany in 2008. There have only been slow improvements in energy efficiency in the paper industryover the past twenty years. Policies can accelerate the progress made, but knowledge about theremaining efficiency potentials and their costs is a prerequisite for their success.

We assess 17 process technologies to improve energy efficiency in the German pulp and paper industryup to 2035 using a techno-economic approach. These result in a saving potential of 34 TJ/a for fuels and12 TJ/a for electricity, which equal 21% and 16% of fuel and electricity demand, respectively. The energysavings can be translated into mitigated CO2 emissions of 3 Mt. The larger part of this potential is foundto be cost-effective from a firm’s perspective. The most influential technologies are heat recovery inpaper mills and the use of innovative paper drying technologies. In conclusion, significant savingpotentials are still available, but are limited if we assume that current paper production processes willnot change radically. Further savings would be available if the system boundaries of this study wereextended to e.g. include cross-cutting technologies.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Energy efficiency1 is considered a key element in sustainabledevelopment. It particularly contributes to reducing energyresource depletion rates and mitigating greenhouse gas emissions.It is a central greenhouse gas abatement option at low specific costs[1], while it also improves firms’ competitiveness.

Paper production accounts for about 9% of industrial energydemand and around 2.5% of all energy-related greenhouse gasemissions in Germany. It recorded high growth rates in the past 20years resulting in an increase of paper production by 75% from 1991to 2008. The paper industry is considered an energy-intensiveindustry with energy costs at around 13% of total productioncosts [2]. It experienced rapid e cost-driven e improvements inenergy efficiency (EEI) in the second half of the 20th century.However, these came to a halt in recent years. Between 1991 and

x: þ49 721 6809 272.leiter), fehrenbach@eifer.orgfgang.Eichhammer@isi.fhg.de

ts in the specific energyction of paper).

All rights reserved.

2008, the specific energy consumption per ton of paper decreasedby only 5.7% [3]. Implementing suitable policies might accelerateenergy efficiency progress in the future. However, knowledge aboutavailable EEI potentials and their costs is a prerequisite to designingeffective and efficient policies.

This paper analyzes available energy saving potentials in theGerman paper industry. We conduct a scenario analysis usinga technology-specific, bottom-up modeling approach and combineit with a thorough review of the literature on energy efficiencymeasures (EEM).

While several studies on thepaper industry havebeen conductedin recent years, they differ in focus, scope and appliedmethodology.Davidsdottir and Ruth [4] analyze the impact of capital turnover andthe vintage structure on energy demand in an econometric modelfor theUS pulp andpaper industry. They focus onpolicy impacts andconsider technologies only in a stylized way. A group of studies[5e7] established an end-use energy demand model-based onenergy flows for the US paper industry. Although the model istechnology-specific, they do not use the model to calculate savingpotentials through technology improvement, instead they focus onallocating energy consumption to the distinct end-uses. Szabó et al.[8] studied the impact of carbonprices on greenhouse gas emissionsof the global paper industry. Although their model also contains

T. Fleiter et al. / Energy 40 (2012) 84e99 85

technical information, like specific energy consumption (SEC), theyfocus on the market dynamics and paper demand.

Studies of the German paper industry focused on a review oftechnology improvement options. IUTA et al. [9] review a largenumber of technologies and provide guidelines for energymanagers, but they do not calculate aggregated saving potentials.Similar studies are available for Austria [10], the US [11] or theEuropean Union [12]. Other engineering studies concentrate onsingle aspects of the paper production chain e.g. [9,13,14,15].

A broad engineering review of the US paper industry [16] ismore comparable to our approach. It considers technology costsand calculates saving potentials for the entire paper industry. Asimilar study was conducted by Möllersten et al. [17] for theSwedish paper industry. They assessed CO2 mitigation potentials ofa set of new technologies. However, they do not explicitly considerthe development over time and consequently draw no conclusionson the timeframe the potentials would need to unfold. Also Far-ahani et al. [18] analyze the impact of new technologies on CO2emissions in the paper industry by comparing Sweden to the US,focusing on the more efficient use of black liquor.

In this paper, we assess if there are technologies available tofurther improve energy efficiency in the German paper industry inthe long term (i.e. 2035), and estimate the economic and technicalpotentials for EEI.

We first review the literature on available energy-efficienttechnologies. For triangulation of technology characteristics wealso conducted interviews among paper mills and technologysuppliers. Next, the technology-specific information is integrated ina bottom-up model, allowing aggregated EEI potentials and theircost-effectiveness to be calculated.

We explicitly consider the diffusion of technologies over time,which allows transparency about the degree of maturity of thetechnologies and yields more detailed policy recommendations.This is particularly important, given the long capital lifetime in thepaper industry and has direct implications on the economics.

The analysis is limited to so-called process-specific technologies.While their counterparts, cross-cutting technologies (e.g. motors,pumps, lighting, boilers, ventilation), are applied across industrialsectors, process-specific technologies are particular to a chosenindustrial sector or process. They are typically deeply rooted in theproduction process and an in-depth analysis of the process isrequired in order to assess EEI potentials.

2. Modeling approach

2.1. Introduction

The approach is based on technology-specific modeling ofenergy demand. Suchmodels are typically referred to as bottom-upmodels and have been applied in energy system analysis since theearly 1980s [19e21]. Bottom-up models derive final energydemand from changes in the technological structure over time.Exogenous activity parameters like industrial production aretranslated via technical parameters into energy consumption. Thelevel of detail of technology representation varies among models.While some models only consider aggregated energy intensity,others consider the useful energy demand (e.g. mechanical energy)and estimate the final energy demand as a function of the efficiencyof the technical system.

Most recent bottom-up models explicitly consider technologiesand their diffusion, and some also stock turnover [22e24]. Newtechnologies change the technical system over time, resulting inchanging energy demand. The advantage of bottom-up models isthe transparency of the underlying technology development, whichensures a “realistic” development path. In most models, diffusion

mainly depends on the cost-effectiveness of the technologies. Ifother determinants like barriers to energy efficiency are consid-ered, they are most often integrated using an ad-hoc approach likean increased discount rate [23].

On the other side, technology diffusion over time is only rarelyconsidered in technology-specific analyses of saving potentials forparticular industries (e.g. see [17,25]). Most studies estimate thepotential energy savings of new technologies, but do not considerthe time required for diffusion. However, particularly in theindustrial sector, with often long technology lifetimes, diffusion isan important aspect of saving potentials [26] and including it in themodel may result in better policy recommendations.

In this paper, we explicitly consider technology diffusion for themodel-based assessment of EEI potentials to compare the impact ofdifferent diffusion scenarios on energy demand. The diffusion pathsare considered as fixed exogenous variables based on expertinterviews and the literature. Endogenous modeling of technologydiffusion would include a huge degree of uncertainty as manyfactors beyond the financial cost-effectiveness influence the speedof diffusion (see below) and take the focus of our study away fromthe analysis of EEI potentials.

2.2. Model description

The model applied in this study, IS Industry, follows thephilosophy of technology-specific bottom-upmodeling. It explicitlydefines technologies and considers investment costs. The model isused as an accounting model and technology diffusion is an exog-enous model parameter. Accounting models do not allow fore-casting, but are used for scenario analysis to compare alternativefutures and draw conclusions on the drivers of energy demand.

The scenarios in our study differ in the assumed diffusion rate ofenergy efficiency measures (EEM). The resulting differences inenergy demand between the scenarios are the saving potentialsachievable by accelerated technology diffusion.

With regard to technology structure, IS Industry distinguishesprocesses and EEM. The former are characterized by their specificenergy consumption (SEC) and a production output. EEM aredefined as technologies or behavioral changes that reduce the SECof a particular process. Thus, each EEM addresses a specific process.

The annual energy savings (ES) of an EEM in year t for onescenario (Sc) are calculated based on the specific saving potential(sp), the diffusion (Diff) of the saving option in year t and theindustrial production (IP) of the related process (p).

ESt;p;EEM;Sc ¼ spEEM*�Diff t;EEM;Sc � Diff t¼2007;EEM

�*IPt;p (1)

Diffusion is an exogenous assumption and derived from pastdevelopment and expectations, including assumptions abouttechnology turnover and lifetime, as well as barriers and costs.

The resulting energy demand (ED) of a scenario is then calcu-lated as the reference SEC corrected for the sum of energy savingsover the different processes for this scenario for year t and the levelof production.

EDt;Sc ¼Xnp¼1

IPt;p*SECt¼2007;p �

XnEEM¼1

ESt;p;EEM;Sc

!(2)

Further, cost-effectiveness of the EEM is calculated and used toconstruct a scenario. Cost-effectiveness for a given year is deter-mined on the basis of all cash flows in that year. The criterion forcost-effectiveness is the specific cost of EEI (c) in a given year. It iscalculated as the total (or net) costs (C) in year t divided by theinduced annual energy savings in year t (ES). The specific costs for

Table 1Assumed specific fuels and electricity consumption (SEC) and resulting total energy consumption and CO2 emissions in 2007.c

Energy consumption CO2 emissionsb

SEC electricity[kWh/t]

SEC fuels[GJ/t]

Total electricity[PJ]

Total fuels[PJ]

Total[TJ]

Share[%]

Indirect[Mt CO2]

Direct[Mt CO2]

Total[Mt CO2]

Chem. pulp 639 12.6 3.5 19.5 23.1 10% 0.6 1.0 1.6Mech. pulp 2200 �2.0a 11.5 �2.9 8.6 4% 2.1 �0.2 1.9RCF pulp 260 0.5 14.7 8.5 23.2 10% 2.6 0.4 3.1Paper 530 5.5 44.5 128.2 172.7 76% 8.0 6.6 14.6Sum 74.3 153.4 227.7 100% 13.3 7.9 21.2

Source: own calculations based on [12,37e39].a The negative fuel consumption represents heat recovery that is used in integrated paper mills for drying paper.b CO2 emissions are calculated using the emission factors given in Table 13. Indirect CO2 emissions comprise those resulting from electricity generation (on-site and offsite)

and direct CO2 emissions comprise all fuel combustion processes in the paper mills (excluding electricity generation).c The German energy balances do not provide energy demand by process, but only for the pulp and paper industry as a whole, which is used to cross-check if the

assumptions on the process-level are realistic. These assumptions are further broken down in the following tables.

Table 2Assumed specific electricity and fuel consumption (SEC) per process step formechanical pulp (GWP).

Process step Electricity Fuels/heat

[kWh/t] [GJ/t] [kWh/t] [GJ/t]

Wood handling 50 0.18 42 0.15Grinding 1800 6.48 e e

Washing 50 0.18 e e

Bleaching 100 0.36 e e

Heat recovery e e �375 �1.35Sum 2000 7.20 �333 �1.2

Source: own assumptions based on [12].

Table 3Assumed specific electricity and fuel consumption (SEC) per process step for theproduction of RCF pulp.

Process step Electricity Fuels

[kWh/t] [GJ/t] [kWh/t] [GJ/t]

Pulping 40 0.14 e e

Screening 50 0.18 e e

De-inking (Flotation) 80 0.29 e e

Concentration and dispersion 40 0.14 150 0.54Bleaching 30 0.11 e e

Others 20 0.07 e e

Sum 260 0.94 150 0.54

Source: own assumptions based on [12].

Table 4Assumed specific electricity and fuel consumption (SEC) per process step for paperproduction.

Process step Sub-step Electricity Fuels

[kWh/t] [GJ/t] [kWh/t] [GJ/t]

Stock preparation Pulper 10 0.04 e e

Refiner 130 0.47 e e

Screening 30 0.11 e e

Wet end Head box 40 0.14 153 0.55Forming section 30 0.11 e e

Dry end Press section 100 0.36 e e

T. Fleiter et al. / Energy 40 (2012) 84e9986

EEI (c) are similar to the cost of conserved energy usually calculatedin such an analysis of EEI potentials. These are typically presentedas conservation supply curves (see Section 6.3).2 The only differ-ence is that the energy cost savings are already considered in thespecific costs as we calculate them. This is necessary, because weconsider a number of different energy carriers with different pricesand are not able to disaggregate them at a later stage.

ct;EEM ¼ Ct;EEMESt;EEM

(3)

The total annual costs for EEI (C) in year t comprise the invest-ment costs (CI), the running costs (CR), the saved energy costs (CE),as well as the saved costs for emission certificates (CC). Theinvestment costs are annualized. The interest rate r is often used inenergy demand models to consider barriers to the implementationof cost-effective EEM [23,28]. In this case, a discount rate higherthan the firms’ profit expectations is assumed.

Ct;EEM ¼ CIt;EEM*

ð1þ rÞLtEEM*rð1þ rÞLtEEM�1

þ CRt;EEM � CE

t;EEM � CCt;EEM (4)

The investment costs (CI) consider both new plants as a result ofcapacity expansion as well as replacement of retired plants. Theassumed specific investment costs (cI) decline over time accordingto an exogenous annual cost reduction coefficient, representingtechnical learning (see Table 5). The calculation of investment costsbased on the specific costs of year t (cI) and the total savingpotential in year t (ES) as in Formula (5) is clearly a simplification, asthe savings in year t also result from investments in earlier periods,

2 The calculation of the specific cost of conserved energy dates back to Meier[27]. A recent application similar to our approach has been conducted for the Thaicement industry [25].

when the specific investment costs might have been higher. Thissimplificationwas necessary, as we are not using a stock model andthus do not know the characteristics of single vintages. However, inorder to derive robust interpretations of the cost-effectiveness, weare calculating a sensitivity analysis of the specific costs, varying thetime horizon as well as the discount rate (see Chapter 6.3). Thesimplification still allows for a transparent and realistic depiction ofthe cost-effectiveness.

CIt;EEM ¼ cIt;EEM*ESt;EEM (5)

The modeling approach also poses restrictions on the systemboundaries of the analysis. Fig. 1 presents different variables thataffect energy demand. Of these, we exclusively analyze the specificenergy consumption (SEC). Changes in the other variables, likechanging paper production, changing product mix toward recycled

Dryer section 90 0.32 1069 3.85Coating and finishing 40 0.14 153 0.55

Others Other processes (effluents,compressed air)

60 0.22 153 0.55

Sum 530 1.91 1528 5.5

Source: own assumptions based on [9,10,12,37].

Productionlevel Product mix

Specificenergy

consumption

Specificmaterial

consumption

Fig. 1. Factors determining energy demand in an industry (adapted from Möllersten[17]).

Fig. 2. Material flow and process chain for the paper production as modeled.

T. Fleiter et al. / Energy 40 (2012) 84e99 87

fibers, or reduced fiber material consumption are unchangedamong all scenarios.

We further narrow the system boundaries to process-specifictechnologies, excluding cross-cutting technologies (e.g. motorsystems, lighting or space heating). These are not particular to thepaper industry and have been studied widely. Thus, in reality theEEI potential will be higher than included in the model in thispaper.3

3. The German pulp and paper industry

3.1. Overview

With a production of 21 million tons in 2009, Germany is by farthe single largest producer of paper in the European Union [2].Worldwide, only the United States, China and Japan produce morepaper. The major share of paper production was in the form ofgraphical paper (9.2 million tons) and paper and paperboard forpackaging (9.1 million tons). The share of both technical paper andtissue paper is relatively low, at 1.5 and 1.4 million tons,respectively.

The paper industry is heterogeneous in company size, rangingfrom small and medium-sized enterprises (SME) to large compa-nies. Out of the 104 firms, 57% have an annual production output ofless than 50,000 tons, while 12 firms produced more than500,000 tons of paper in 2009. Production quantity is concentratedin these 12 firms which are responsible for 65% of the total outputin 2009.

The German paper industry is highly integrated in the Europeanmarket. Although domestic paper production exceeded consump-tion by 13%, significant trade flows exist. The total export accountedfor 12.4 million tons of which 79% were exported to other EUmember states. Similarly, 83% of imports originated from other EUcountries amounting to a total of 10 million tons in 2009 [2].

The German paper industry experienced a substantial produc-tion growth of 79% from 1991 to 2008. Industry representatives donot expect these growth rates to continue. Driven by the risingoutput, energy demand increased simultaneously over the sameperiod: by 75% from 1991 to 2008. Consequently, the SEC remainedmore or less constant (�5.7% between 1991 and 2008). However,CO2 emissions decreased significantly, mainly due to fuel switch toless carbon-intensive energy carriers (renewable energy andnatural gas replaced hard coal and crude oil).

3.2. Technology adoption in the paper industry

Empirical evidence exists that firms often do not adopt energy-efficient technologies despite their cost-effectiveness [29]. This gapbetween the available potential and the real implementation infirms is also referred to as the energy efficiency gap or the no-regretpotential [30]. From a policy point of view, the no-regret potential isvery attractive, due to the net benefit it implies for the technology

3 The processes in the paper industry particularly use mechanical energy forrolling the paper web and pumping pulp and water. Thus, options like replacinginefficient electric motors or pumps have a high EEI potential in the paper industry.

adopter [31]. Several studies have analyzed the structure of thefactors or barriers hampering technology adoption. Schleich [32]analyzed barriers in the German service sector and found lack ofstaff time, investment priority-setting, information deficits andsplit incentives to be major barriers. The latter two were also foundby de Almeida [33] for the electric motor market. Further importantbarriers are competition with alternative investment opportunitiesand uncertainty with regard to future technology and pricedevelopment [34].

Although the pulp and paper industry is grouped among theenergy-intensive industries with energy costs accounting for morethan 10% of the production cost, here too non-economic barriershamper the adoption of cost-effective, energy-efficient technolo-gies [35].

The structure of barriers varies greatly, depending on tech-nology and firm characteristics. If technologies are integrated intocomplex production processes, the intensity of barriers is differentcompared to technologies that are applied somewhat detachedfrom the production process, like space heating or lighting. Thol-lander and Ottosson [35] confirm this view and find technical risk(of production disruptions) as the main barrier in the Swedish pulpand paper industry. This is followed by hidden costs throughproduction losses and other inconveniences. Further importantbarriers are lack of time or other priorities and lack of access tocapital.

Del Río González [36] analyzed the adoption of clean technol-ogies4 using a survey among 46 paper producers in Spain. The threemajor barriers for technology adoptionwere all related to high costs(long payback time, high initial investment, not cost-effective). Thisis further supported by the interviews with German paper industryrepresentatives, indicating that two years is the maximum paybacktime acceptable for energy efficiency investments.

4. Paper production

4.1. Overview

Paper is produced based onwood and recovered paper. Themostenergy-intensive process steps are the production of pulp and thefurther processing of this semi-finished product to the paper web.Chemical or mechanical pulp is produced from wood, while RCF(recovered fibers) pulp is produced from recovered paper. The pulpis processed in the paper mill to produce the paper web. Additionalnon-fiber resources like fillers or additives are used in lower quan-tities and their production is not included in this study (see Fig. 2).

4 Although energy-efficient technologies are a sub-group of clean technologies,they are somehow idiosyncratic, due to the different motivation for adoption. Whileclean technologies are mostly adopted to comply with environmental regulation,energy-efficient technologies always affect energy and cost savings, and are lessdriven by regulation.

T. Fleiter et al. / Energy 40 (2012) 84e9988

The three pulp production lines distinguished differ in terms ofenergy intensity and product characteristics. Also, the core paperproduction process differs depending on the paper grade andbetween integratedandnon-integratedmills, but thedifferences areminor when compared to the differences in pulp production lines.

The assumed reference energy consumption per process is givenin Table 1. While the production of mechanical as well as chemicalpulp shows the highest SEC, the paper production process has thehighest total consumption, accounting for 76% of final energydemand of the paper industry in Germany in 2007. Virgin pulp ismainly imported from Scandinavian countries, while the domesticproduction focuses on RCF pulp.

The definition of energy consumption used in this study isbased on German energy balances (AGEB), and assumes finalenergy demand. Thus, primary energy used for on-site electricitygeneration is not considered, instead the total net consumedelectricity is included (including self-generated and purchasedelectricity).

4.2. Pulp production

Chemical pulp. The wood chips are cooked together withchemicals and water in a digester at around 130e150 �C. Lignin isseparated from the fibers (defibration), while the structure of thefibers remains intact. Chemical pulp production is further distin-guished into sulphite and sulfate (kraft) pulp. In Germany, fourintegrated paper mills use the sulphite pulp process, while twokraft pulp plants were built in 1999 and 2004. The latter havea combined capacity of 975 kt of pulp per year. The sulphiteproduction capacity is around 600 kt.

Mechanical pulp. The mechanical breakdown of wood into fibersby grinding or refining yields pulp with different characteristicscompared to chemical pulping. As the lignin remains in the pulp,yield is higher and paper strength is lower. Mechanical pulpproduction is typically integrated into the paper mill.

In Germany, mainly two alternative processes are used formechanical pulp: ground wood pulp (GWP) and thermo-mechanical pulp (TMP). The ground wood pulp process (GWP)consumes 4e7.9 GJ of electricity per ton of pulp [12]. Thermo-mechanical pulping relies on refining at elevated temperatures,resulting in higher pulp quality, at an SEC of 6.5e13 GJ electricity/tof pulp [12]. In 2008, 1.45 million tons of mechanical pulp wereproduced in Germany, of which 30% TMP and 70% GWP [38].

The typical process steps for GWP and their assumed SEC aregiven in Table 2. The most energy-intensive step is grinding (forTMP it is refining). Both GWP and TMP processes mostly consumeelectricity. During mechanical processing, large amounts of wasteheat are released, which are typically used in the drying section ofthe paper mill.

Recycled fibers (RCF) pulp. In Germany, the largest share of pulpproduction is based on recovered paper. In 2008 a total of 16milliontons of recycled fiber pulp (RCF) was produced [38]. The processsteps and their SEC depend on the pulp quality. The values assumedfor our analysis are shown in Table 3.

4.3. Papermaking

In papermaking, pulp and other raw materials are processed toa paper web. Two main production steps can be distinguished: thestock preparation and the paper machine (see Table 4). In non-integrated paper mills, stock preparation begins with the pulpingof the delivered (dry) fibers to produce a fibrous slurry. Dependingon the required paper grade, the fibers are again refined to e.g.increase paper strength. Before screening, the slurry enters the firstpart of the paper machine, thewet end, where the fibers are filtered

and the paper web is formed. It has a solid content of 16e25% at thisstage. In the dry end, the paper web enters the pressing sectionwhere the solids content is further increased to about 50e55%. Theremaining water cannot be removed by mechanical means, but isevaporated in the dryer section, through conventional dryingcylinders. The thermal drying is by far the most energy-intensiveprocess step. The finished paper web has a solids content of morethan 90%.

Note that the average values used for the modeling maydiffer significantly from the situation in individual paper mills,because:

� The SEC may differ significantly, depending on the paper gradeproduced [see for example 9].

� Certain process steps such as pulp drying or pulping are notrequired in integrated paper mills. Also waste heat fromrefining and grinding may be used in the papermaking processin integrated mills.

� The number of finishing steps differs (e.g. calendaring, coating).� The structure and quality of the fibers.� The past implementation of EEM.

However, the publicly available data does not allow us to furtherdifferentiate paper grades or integrated from non-integrated mills.Indeed, the assumed average values reflect a realistic situation in an“average” German paper mill taking the shares of the differentpaper grades in Germany into consideration.

5. Review of energy-efficient technologies

5.1. Chemical pulp

Black liquor gasification (1). Chemical pulp plants are favorablefor the production of a variety of “green” chemicals as well as bio-fuels. This extension of a pulp plant into a bio-refinery would allowimproved use of wood residues. Bio-refining is intensively dis-cussed in the literature, especially in connection with black liquorgasification [40].

In a conventional chemical pulp plant, the black liquor,a mixture of lignin and chemicals, is concentrated and then burnedin a recovery boiler. These boilers have strict thermodynamiclimitations resulting in a low electrical efficiency of about 10e15%[12]. Alternative processes like a gasification of black liquorallows a combined cycle to be used, resulting in higher efficienciesfor electricity generation [41]. It is assumed that this could evendouble electricity generation, while heat production remainsconstant [1]. Although demonstration plants are in operation, it isstill a challenge to integrate the technology into the productionprocess of a chemical pulp plant.

5.2. Mechanical pulp

Energy demand for the production of mechanical pulp ischaracterized by a high consumption of electricity for the woodgrinding or refining, and releasing considerable amounts ofwaste heat, as 95% of the mechanical energy used is transformedinto heat. Consequently, energy efficiency improvementsconcentrate on efficient grinding and refining as well as wasteheat use.

Heat recovery (TMP, GW) (2). In integrated plants, the wasteheat can be used in the paper machine. Depending on the processdesign and grinding intensity, about 20e40% of the electricityconsumed can be recovered in the form of steam, and a further20e30% in the form of hot water [12]. This concept, however, is

T. Fleiter et al. / Energy 40 (2012) 84e99 89

standard in new plants and thus the remaining diffusion potentialis limited.

Effluent water from the bleaching process is a potential heatsource at a lower temperature level which was not as extensivelyused in the past. However, for integrated paper mills with a highdemand for process heat, it might be cost-effective. Also, certainheat sinks are not fully exploited, like the preheating of bark orsludge that is burned in the bark boiler [42].

High-efficiency grinding (GW) (3). Different concepts aim at EEI ofwood grinding design. One of these is to use fully metallic grinderswith optimized grinding surface patterns instead of stone orceramics. First results from pilot plants at the Finnish researchinstitute KLC attained electricity savings of around 50%, while pulpcharacteristics were unchanged and production capacitydoubled [43].

Enzymatic pre-treatment (TMP) (4). Pre-treating wood chipsusing enzymes reduces the mechanical energy needed for woodprocessing. A variety of processes and enzymes have been dis-cussed since the 1980s, but no single dominant process design hasevolved so far [44]. New approaches combine the use of enzymeswith low-intensity refining to improve the penetration of theenzymes into the wood. Electricity savings are expected of10e40%, depending on the type of enzymes and the processdesign [44].

High-efficiency refiner, pre-compression and use of wood shav-ings (TMP) (5). Many technologies aim to improve refining effi-ciency. In the past 20 years, research activity was aimed atrefining and several innovations entered the market. Oneexample is the high-efficiency refiner RTS from Andritz, which isclaimed to save up to 10e15% of energy compared to conven-tional disc refiners [45]. Gorski et al. [46] found that compressionprior to refining could reduce specific electricity demand by up to20%, and the first applications are commercially available.Another option is the use of wood shavings instead of woodchips, which could reduce electricity consumption by around25% [47].

5.3. Pulp from recovered fibers (RCF pulp)

High consistency pulping (6). In pulping, most of the energy isused to circulate and move the slurry. Consequently, by increasingthe consistency of the slurry, the electricity demand of the pulpercould be decreased, due to reduced mass flow. Electricity savings of2e10 kWh per ton of deinked pulp are expected if the solidscontent is increased from a typical 5e7%5 to 20% [37]. Highconsistency pulping is already used in a number of plants inGermany. Further savings in pulping are possible if the spiral coil ishydro-dynamically optimized and driven at lower speed. Depend-ing on the explicit process design, these optimizations could resultin an EEI of up to 20% [48].

Efficient screening (7). Improvements were made in the field ofscreening and filtering. Increasing the slurry consistency from 1.5to 2.5% results in considerable EEI [37]. Further optimization ofthe screening process showed energy savings of 5e30%,depending on the plant characteristics [48].

Waste heat recovery from bleaching (8). Bleaching waste waterhas an increased temperature. This heat can be recovered topreheat fresh water. Steam savings of around 30 MJ/t of pulp arereported [37]. According to Blum et al. [37], several plants inGermany already apply this technology.

5 The different levels of solid content (see number 7 and 9) refer to differentprocess steps.

De-inking flotation optimization (9). The most important processstep in the de-inking process is flotation.While the solids content isvery low, at around 1%, chemicals are added and air separates theink particles from the fibers. Energy demand during flotation ismainly a result of pumping slurry. Better demand-related control ofpumps as well as a reduced flow speed enable significant energysavings.

Efficient dispersers (10). Dispersion is a post-treatment of fibersuspension to improve the strength and to separate remainingparticles from the fibers. Energy-efficient dispersers are beingincreasingly used and show savings of around 20% when comparedto conventional dispersers [48].

5.4. Paper

Efficient refiners and optimization of refining to reduce idle time(11þ12). According to Blum et al. [37], the reduction of idle-running comprises a large saving potential. They expect thatnew refining concepts will allow reduction of the idle-runninglosses by up to 40%. An example of such a refining concept isthe Papillon refiner [49]. It uses a cylindrical form that separatesthe refining process from the fiber transportation process, result-ing in a better control of the refining process. Compressionrefining that changes the fiber structure by compression forcescould improve refining efficiency even further. Dekker [50]expects electricity savings of up to 30% compared to conven-tional single disk refiners - based on results from a pilot plant.Currently, a demonstration plant has been set up and marketintroduction is planned [50].

Chemical modification of fibers (13). Chemical modification offibers is based on a new understanding of the adhesive forcesbetween fibers. Following conventional theory, the adhesiveforces are based on hydrogen bonds, which depend on the size ofthe fiber surface. This in turn is increased by energy-intensivefiber refining. The alternative idea of modifying the fiberschemically instead aims to influence other binding effectsbesides hydrogen bonds to improve paper strength. Currentresearch is experimenting with alternative fiber-bindingprocesses [51].

The (partial) substitution of refining may result in electricitysavings of about 100 kWh/t of pulp, depending on the papergrade. Further, given a similar paper strength, the water retentionvalue is lower than for refining, which results in lower energydemand for dewatering and drying [51,52]. Furthermore, thedensity of the paper web could be reduced, which resulted infiber savings of 5e15% [51]. These considerations do not includethe energy needed to produce the chemicals. The technologycould have considerable co-benefits, like reduced pulp costs andincreased productivity. The technical feasibility in a large-scaleplant and the market entry is expected for the coming years. Itis assumed that it will be possible to upgrade a paper machinealready in use.

Steam box (14). The steam box preheats water to reduceits viscosity, improving dewatering efficiency and allowinghigher dry contents to be attained in the press section. Asa result, less water needs to be evaporated in the dryer section[37,53]. It is assumed that a temperature increase of 10K results inincreased of dry contents of about 1% [9]. Voith mentions steamsavings of up to 4%.6 Steam boxes are common in modern papermills.

6 http://www.voithpaper.com/applications/documents/document_files/ 464_d_modulesteam_d_ 06_07_150dpi.pdf (access July 8th 2010).

Base year

Historicdevelopment

Diff

usio

n of

ene

rgy

effic

ient

tech

nolo

gies

BAUDiffusion

TechnicalDiffusion

time

FrozenEfficiency

Cost-effectiveDiffusion

No-regretpotential

Technicalpotential

Cost-effectivepotential

BAUpotential

Fig. 3. Definition of diffusion paths and saving potentials.

T. Fleiter et al. / Energy 40 (2012) 84e9990

Shoe press (15). The shoe press is integrated into the papermachine’s press section, improving dewatering of the paper web byan increased pressing surface between the two rollers. This reducesthe demand for thermal drying, while the electricity demandincreases slightly. As rule of thumb, it is assumed that a 1% increasein the paper web’s dry content results in 5% steam savings in thedrying section [53]. Apart from energy savings, shoe presses haveother benefits, such as increased production capacity, improvedproduct quality and space savings due to shorter thermal dryingsections. These are the main drivers for technology adoption [54].Although, shoe presses are widely diffused, further potentials forapplication remain.

New drying techniques (16). As drying the paper web is the majorenergy-consuming process in a paper mill, the R&D efforts for EEIare concentrated on the drying section. The literature discussesvarious new drying concepts that might result in EEI. However,contradictory opinions about the possible energy savings areobserved, as well as uncertainty about market entry. Examples aresteam/air impingement drying, condensing belt drying andimpulse drying.

Impulse drying combines the effects of pressing and contactdrying by running thewet paper web through a heated pressing nip(150e500 �C, 100 ms, 0.3e7 MPa), resulting in a steam explosion atambient conditions behind the pressing nip. Although very effec-tive, this drying method is offset by low achievable paper quality.Expected EEI range from 20% [15] to 0% [12]. Again, productivityincreases are the primary motivation for research. Despite 25 yearsof research activities, including several pilot plants, the technicalbarriers7 have not yet been overcome [55].

Another technology is steam and air impingement drying,where superheated steam or hot air (w300 �C) is blown at highspeed against the paper web. Laurijssen et al. [14] do not expectsignificant energy savings compared to conventional drying cylin-ders. The evaporated water would be available as steam that couldbe used for heat recovery. Despite long research activity in thisfield, market entry is still uncertain [14,56].

7 Blistering on the paper surface as well as delamination.

For condensing belt drying, steam savings of around 10e20% areexpected and two commercial plants have been running in Finland(since 1996) and South Korea (since 1999) [16]. However, in thepast years no further plants were equipped with condensing beltdrying [14].

Even if one of the discussed drying technologies would becommercially viable, the diffusion through the papermachine stockwould take a long time, as the dryer section of a paper machinetypically has a lifetime of 20e40 years [14].

Heat recovery and integration (17). Heat recovery and the use ofwaste heat are widespread in the paper industry. Large potentialsare found in the use of waste heat from refiners and grinders, butalso from the dryer section in the paper machine and the effluentwater. In particular, the use of low temperature heat still showsfurther potential, but also the steam system is often not adequatelyoptimized.

The LfU and PTS [57] conducted an energy audit in an integratedGermanpapermill focusingontheuseof lowtemperaturewasteheat.Using pinch-analysis, they found several opportunities forwaste heatuse thatwould amount to steam savings of up to 25%. These includedthe external use of heat for district heating. All measures hada payback time of less than four years, whilemanywere even shorterthan one year. Further studies confirmed these saving potentials.Another study found a cost-effective saving potential of 7e13% byoptimizing and replacing heat exchangers in three paper machines[58,59]. A recent thermodynamic optimization of Dutch paper millsreported potentials to reduce the steam demand for paper drying by32% [14]. Anoptimizationof heatflows infive papermills inGermanyresulted in average steam savings of 9.3% [9]. Bujak [60] empiricallyfinds fuel savings of 8% due to modernizing the steam system ina corrugated boardmill in Poland. The investment of around 100,000euros showed a payback time of one year and had as a co-benefita lower consumption of water and chemical agents.

A survey among 46 paper mills in Germany [9] found thataround 70% of firms use waste heat from the paper machine toheat the supply air. The use of waste heat to preheat clear waterand white water is implemented by 30 and 40% of the firms,respectively. Waste heat from the coater is less used. 20% of thefirms use it to preheat the supply air and only 5% to preheat thehot water.

8 Ideally this should not be a simple factor but rather follow the “experiencecurve” methodology as outlined for example by Weiss et al. [62]. However, asvirtually no such empirical analyses exist for the considered technologies [63],learning rates would have to be assumed and the more sophisticated approachwould probably not result in more robust results.

T. Fleiter et al. / Energy 40 (2012) 84e99 91

6. Scenario analysis and results

6.1. Scenario definition

For the scenario analysis, alternative “futures” are constructed,based on differing assumptions of scenario parameters. Comparingthe scenarios enables to learn about the potential impact of variousassumptions and developments.

The study assesses EEI potentials through the diffusion ofenergy-efficient technologies. We are defining scenarios bychanging the speed of technology diffusion, which is used as anexogenous parameter. The differences between the resultingenergy demand in the scenarios equal the EEI potentials. An over-view of the scenarios and the related saving potentials is given inFig. 3. Four scenarios are considered and described in the following.

Frozen-efficiency scenario. For the frozen-efficiency scenario,energy intensity remains at the 2007 level, see Table 6. It is thebaseline to estimate the EEI potentials. Energy demand in thisscenario is only impacted by changes in the production output.

Business-as-usual (BAU) diffusion scenario. The BAU scenarioassumes that barriers to technology diffusion remain high in thefuture and represents an extrapolation of past trends. Theexogenous technology diffusion rates are based on the pastdevelopment as well as on discussions with paper industryrepresentatives. These diffusion rates are typically lower thanthey would be in case firms decided purely on the basis of cost-effectiveness (see chapter 3.2).

Cost-effective diffusion scenario. The cost-effective diffusionscenario assumes homo economicus behavior and the imple-mentation of all cost-effective technologies. Cost-effectiveness isassessed on the basis of the investments annuity, using a discountrate of 15%. It implies the removal of all non-financial barriers. If thetechnology is calculated to be cost-effective, the exogenous diffu-sion path from the technical scenario is considered. In case it is notcost-effective, the diffusion path from the BAU scenario is assumed.This approach ensures that, even if the technology is not cost-effective for all mills, it is applied in a certain number of firms,mainly because it might be cost-effective in a niche of technologyadopters, as a result of heterogeneity.

The intensity of barriers in this scenario depends on the level ofthe discount rate. For example, a discount rate of around 15%implies the above mentioned homo economicus investmentbehavior. However, a higher discount rate of 30 or even 50% wouldrepresent higher barriers. This approach of adjusting the discountrate is indeed widely used in bottom-up models to account forbarriers [28]. In order to show the impact of the discount rate, wecalculate conservation supply curves (CSC) for varying discountrates.

The definition of costs has important implications on the cost-effectiveness. Instead of using the full investment costs of newtechnologies, we used differential costs. These are calculated as thedifference between the costs of conventional technology andenergy-efficient technology delivering the same energy service.Consequently, the investment in new technologies is only allowedin case the existing technology has to be replaced. In other words,there is no premature technology replacement and the stockturnover rate is not adjusted. Furthermore, we exclude co-benefits,such as quality or capacity improvements. Considering co-benefitscould further improve the cost-effectiveness of many technologies(for a discussion of co-benefits, see [61]).

Technical diffusion scenario. The fourth scenario is namedtechnical diffusion. It does not include cost considerations for thediffusion of technologies. No premature stock replacement isallowed and thus the diffusion can still be considered “realistic”,although ambitious. Given the long lifetime of certain processes, it

can take a long time for the full saving potential to be realized,even in the technical scenario. The scenario may therefore betermed a “realistic” technical diffusion scenario as it does notinclude completely unrealistic technology options and diffusionpaths.

6.2. Scenario input parameters

Tables 5 and 6 give an overview of the resulting assumptions onthe modeling input data per EEM. The EEM are sorted according toprocess and process step they are allocated to.

The relative saving potential is given as the share of energydemand of the corresponding process step. The absolute specificsaving potentials are defined per production output of the corre-sponding process. All costs are defined as differential costs, i.e. thedifference between the costs of conventional technology and theenergy-efficient option. As cost data is rarely found in the literature,we mainly relied on estimates by representatives from paper millsand technology providers. The specific costs are input to the model,while the payback times help to more easily judge the cost-effectiveness of the EEM. As specific investment costs typicallyfall when technologies diffuse and become more mature, we haveincluded a cost reduction factor, which varies according to thematurity of the EEM.8 The economic lifetime is used for theinvestment calculations and is not to be confused with the realtechnology lifetime, as given in Table 6.

The exogenous assumptions on technology diffusion are givenin Table 6. The future diffusion is derived from literature sources (asdescribed in chapter 5) and discussions with experts from paperproducing firms, technical research institutes and technologysuppliers. Issues like the technology lifetime and the replacementcycle are taken into account. These assumptions may includeuncertainties, but a scenario is not a forecast, but rather a methodto assess the impacts of different assumptions on future develop-ments. For some technologies, even in the technical scenario, thediffusion remains low. This is mainly due to technical processrestrictions and heterogeneity in the process. Enzymatic pre-treatment, for example, is applied in the TMP process, which onlyaccounts for 30% of all mechanical pulp production plants.

Further model input is the SEC per process in the base year 2007as discussed and presented in Chapter 4.

In order to calculate the total EEI potentials for the entire paperindustry, we take interactions among technologies into account.These might occur in two ways. First, technologies are alternatives(which mutually exclude each other) and second, two technologiesaddress the same energy flow. Alternative technologies areaccounted for by restricting the maximum diffusion of the tech-nologies. In the second case, the technology implemented firstreduces the energy flow and, simultaneously, the remaining savingpotentials of the technologies implemented afterward. To accountfor this effect, we calculated the technology saving potentials step-by-step, accounting for changes in the addressed energy flow.Table 7 shows the saving options concerned by interactions and theimplementation order assumed. While this approach is clearlya simplificationof the real circumstances inpaperproducingfirms, itdoes capture the main effects of technology interaction, which issufficient for our interpretation of the scenario results at nationallevel.

Table 5Summary of technology assumptions: specific saving potentials and costs per energy efficiency measure.

Process Process step Energy efficiency measure (EEM) Specific saving potentialb Differential costsc

Electricity Fuels Initial costs(2007) [V/t]

Costreduction[%/a]

Runningcosts[V/ta]

Paybackperiodd

(2007) [a]

Economiclifetime [a]

[%]e [MJ/t] [%]e [MJ/t]

Chemicalpulp

Chemical pulp 1 Black liquor gasification 16 2000 e e 440.0 1.6 4.7 15.0 10

Mechanicalpulp

TMP refiner andGW

2 Heat recovery (TMP, GW) e e 38 3475 35.9 0.0 e 1.2 10

Grinding (GW) 3 High-efficiency grinding (GW) 40 2592 e e 352.5 1.6 e 8.0 5TMP refiner 4 Enzymatic pre-treatment 30 1862 e e 433.7 1.6 2.8 15.0 10TMP refiner 5 Efficient refiner and pre-

treatment20 1552 e e 105.5 1.0 e 4.0 10

Recoveredfibers

Pulping 6 High consistency pulping 14 20 e e 2.4 0.4 e 7.0 10Screening 7 Efficient screening 36 65 e 5.5 0.4 e 5.0 10Bleaching 8 Heat recovery from bleaching e e 6 30 1.3 0.4 e 5.0 10De-inking 9 De-inking flotation optimization 17 50 e e 0.9 1.0 e 1.0 10Concentration 10 Efficient dispersers 15 22 e e 1.1 0.4 e 3.0 10

Paper Refiner 11 Efficient refiners 30 118 e e 15.2 1.4 e 7.6 10Refiner 12 Optimization of refining 16 75 e 0.6 0.0 e 0.5 10Refiner, press,drying

13 Chemical modification offibers

15/40a 164 5 185 4.1 1.0 3.0 3.0 10

Drying section 14 Steam box e e 5 180 4.0 0.4 e 2.6 10Drying section 15 Shoe press e e 12 480 28.9 1.0 e 7.0 10Drying section 16 New drying techniques e e 20 6.67 86.0 1.6 e 15.0 1Paper 17 Heat recovery and integration e e 20 1071 13.8 0.0 e 1.5 0

a 15% is for the press section and 40% for the refiner.b Main sources for specific saving potential: 1: double electricity generation [1]; 2: 20e40% of electricity consumed [12]; 3: around 50% electricity savings [43]; 4: 10e40%

electricity savings expected [44]; 5: 10e15% [45], up to 20% [46], around 25% [47]; 6: 2e10 kWh/t deinked pulp [37], up to 20% [48]; 7: 5e30% [48]; 8: Steam savings of around30MJ/t pulp [37]; 9: assumption based on product description; 10: 20% [48]; 11: reduction of idle-running losses by up to 40%. [37]; 12: 30% expected savings [50]; 13: savingsestimated from expert interview; 14: steam savings of up to 4% according to Voith product description; 15: 14% steam savings found in case study [9]; 16: Wide range oftechnologies, e.g. for impulse drying the expected savings range from 20% [15] to 0% [12]; 17: 7e13% [58,59], steam savings of up to 25% [57], 32% [14], 9.3% [9] and 8% [60]. Fora more detailed description of sources please see Section 5.These sources provided the basis for the assumed values and were verified in expert interviews resulting indeviating assumptions for some EEM in the table.

c Only very limited empirical data is available for costs. The costs weremainly estimated based on typical payback periodsmentioned in interviewswith plantmanagers andtechnology experts. Assuming typical energy prices, the specific costs can be calculated from the payback period. The annual cost reduction was assumed to be higher for lessmature technologies.

d Main sources: 10: about 1 year from case study [9]; 15: 9 years from refurbishment case study [9]; 17: many <1 year and max 4 years [57] 1 year [60].e The relative saving potential is related to the fuel/electricity demand of the corresponding process step.

Table 6Summary of technology assumptions: technology diffusion scenarios per energy efficiency measure.

Process Process step Energy efficiency measure (EEM) Diffusion parametersb Diffusion:BAUscenariob

Diffusion:technicalscenariob

Technology life cycle Technicallifetime [a]

Diffusion2007a [%]

2020 2035 2020 2035

Chemical pulp Chemical pulp 1 Black liquor gasification R&D 20 0 0 16 5 45

Mechanicalpulp TMP refiner and GW 2 Heat recovery (TMP, GW) Standard 20 92 95 98 100 100Grinding (GW) 3 High-efficiency GW Demonstration 5 0 4 10 25 67TMP refiner 4 Enzymatic pre-treatment R&D 15 0 0 5 10 17TMP refiner 5 Efficient refiner and pre-

treatmentCommercial 20 7 12 18 22 33

Recoveredfibers Pulping 6 High consistency pulping Commercial 20 30 43 58 70 100Screening 7 Efficient screening Commercial 20 20 30 42 55 100Bleaching 8 Heat recovery from bleaching Commercial 20 20 30 42 55 100De-inking 9 De-inking flotation optimization Demonstration 15 0 13 25 40 100Concentration 10 Efficient dispersers Commercial 15 30 43 60 83 100

Paper Refiner 11 Efficient refiners Commercial 20 5 13 22 30 80Refiner 12 Optimization of refining Commercial 20 30 45 60 85 100Refiner, press, drying 13 Chemical modification of fibers R&D 20 0 3 13 10 70Drying section 14 Steam box Standard 15 64 67 70 80 80Drying section 15 Shoe press Standard 20 57 63 69 70 81Drying section 16 New drying techniques R&D 20 0 5 16 15 75Paper 17 Heat recovery Standard 20 50 65 80 100 100

a Diffusion is defined as the share of technology stock that is equipped with the related energy efficiency measure.b There are only very few empirical sources for the technical lifetime and diffusion (2007 as well as scenarios). They result from discussions with experts from paper

producing firms, technology providers and research institutes and clearly involve a substantial degree of uncertainty. Often only qualitative information was available. In thiscase we discussed it in Section 5. The scenarios business-as-usual (BAU) and technical diffusion are presented in Section 6.1.

T. Fleiter et al. / Energy 40 (2012) 84e9992

Table 7Assumed implementation order of energy efficiency measures (EEM) addressing thesame energy flow.

Process Process step Order of EEM

Paper Refiner 12, 11, 13Paper Drying section 15, 14, 13, 16, 17Mechanical pulp Refining and grinding 5, 4, 3, 2

Table 8Production output as assumed for all scenarios by process [kt].

Process 2007a 2010 2015 2020 2025 2030 2035

Chemical pulp 1545 1520 1520 1490 1346 1217 1100Mechanical pulp 1456 1383 1383 1355 1225 1107 1001RCF pulp 15,737 15,378 16,476 17,242 17,785 17,929 17,799Paper 23,319 22,509 24,008 25,040 25,536 25,567 25,280

a Source for 2007 values: [2].

T. Fleiter et al. / Energy 40 (2012) 84e99 93

Other input parameters like the production output, the energyand emission certificate prices and the CO2-intensity of energycarriers are shown in the Appendix and are derived from a recentenergy scenario study prepared for the German government [64].We use the average CO2-intensity of German electricity generationto calculate CO2 emissions based on energy demand. It changesover time, due to fuel switch and efficiency improvement in theelectricity generation sector.

The total level of energy demand is influenced by productionoutput per process. The future paper production was estimatedtogether with experts from the paper industry at the level ofpaper grades.9 The assumed development of the different pulpproducts was derived from the paper production taking paperrecycling and the use of fillers into account.10 Import and exportshares are assumed to remain constant. The resulting productionper process is given in Table 8. The shift to recovered fibers andthe increased use of fillers reduces the demand for virgin pulpover time and, according to our assumptions, also the productionin Germany.

6.3. Scenario results

The resulting energy savings across all four processes are givenin Fig. 4. Excluding economic considerations, the EEI potential isestimated at 21% for fuels and 16% for electricity when comparingthe technical diffusion scenario to the frozen-efficiency scenario until2035. The cost-effective EEI potential amounts to 15% for fuels and13% for electricity, when compared to the frozen-efficiency scenario.

Fig. 5 depicts the EEI potentials in relation to the total electricityand fuel demand. While the frozen-efficiency energy demandincreases slightly until it peaks around 2025, the total energydemand falls continuously when the cost-effective or technicalpotentials are exploited. It can be observed that, for fuel savings, thecost-effective scenario is closer to the technical scenario in mostyears than is the case for electricity savings. There are two mainexplanations. First, the higher increase in fuel prices compared toelectricity prices (see Table 11) and, second, fuel savings implya reduction of direct CO2 emissions for which the producer isrequired to deliver emission certificates.11

The contribution of single technologies to the aggregated EEIpotentials is given in Table 9. High impact options are heat recoveryand integration, new drying techniques and the chemical modifi-cation of fibers.

The resulting specific fuel and electricity demand in the fourscenarios is shown in Table 10. The SEC for RCF pulp changes slowly,which is the result of the fact that EEI potentials are distributedamong a large number of smaller options. These are difficult toaccount for in our analysis so that besides the five EEM considered

9 The assumed average annual growth rates are: graphical paper 0.3%, packagingpaper 0.9%, tissue paper: 0.9%, technical paper: 1.2%.10 The share of “fillers” increases from 17 to 19% and the share of recycled paperfrom 69 to 71% between 2009 and 2035.11 Note that the CO2 emission certificate price, which is already integrated in theassumed electricity prices, will also affect electricity prices.

further potentials are probably available. Furthermore, RCF processsteps also depend on the paper grade. For example, bleaching andde-inking are not applied in the production of packaging paper-board. Thus, the presented average savings will be different fordifferent paper grades.

For chemical pulp, the SEC for fuels remains constant, becausewe only considered black liquor gasification as EEM, which resultsin a more efficient on-site electricity production (and does notaffect final energy demand of the process). Chemical pulp is mainlyproduced in two relatively new plants in Germany, which alreadyapply recent technology. Although this does not imply that no EEIpotentials are available, they are relatively low compared to theother processes.

6.3.1. Resulting costs for CO2 emissions abatementAs energy efficiency is regarded as a major greenhouse gas

(GHG) abatement option, often available at low cost, we alsodiscuss the resulting abatement costs (see Fig. 6). The abatementcost curve12 shows the abatement potential available by 2035 andits related costs for a discount rate of 15%. The discount rate is usedto calculate the annuity of the distinct investments and thusinfluences the specific costs of CO2 abatement (see Equation (4)). Itis not used to discount all financial flows to the base year 2008. Thespecific costs comprise both avoided energy costs as well as CO2emission certificate costs.

Fig. 6 shows that most of the EEM are cost-effective for thegiven assumptions. However, the total cost-effective potentialdepends largely on key abatement options like “new dryingtechniques”.

The distinction into three classes of technologies, namelyprocess optimization, best available technology (BAT) and processinnovations not only allows conclusions to be drawn on policyrecommendations, but also on the reliability of technology data. Forprocess innovations, data mainly originates from experience gainedin pilot plants and these were extrapolated to industrial applica-tions, taking the expectations of technology experts into consid-eration. Process optimization and BAT assumptions areconsiderably more reliable as they are often based on results fromcase studies in paper mills.

6.3.2. Sensitivity analysis of cost-effectivenessFor a better understanding of the costs-effectiveness, we

calculate conservation supply curves (CSC)13 for varying discountrates and years. Changes in other factors influencing the specificcosts, like energy prices, potentials for EEI, or investment costsresult in proportional changes in the specific conservation costs.The curves are similar to Fig. 6, except that they are not related toCO2 emissions but to energy demand.14

12 For a critical review of the use of abatement cost curves for policy-making see[65].13 For a broader discussion of CSC, see [66].14 Using the underlying CO2-intensity of energy carriers as shown in Source: owncalculations based on reference scenario from [64] Table 13 allows converting onecurve into the other.

2%4% 4%

8%

4%

11%13%

15%

6%

12%

16%

21%

0%

5%

10%

15%

20%

25%

Electricity Fuels Electricity Fuels

53020202

Shar

e of

ele

ctric

ity/fu

els

dem

and BAU potential

Cost-effective potentialTechnical potential

Fig. 4. Resulting saving potential across all processes by scenario as share of the electricity/fuel demand in the frozen-efficiency scenario.

0

10

20

30

40

50

60

70

80

90

2007

2015

2020

2025

2030

2035

Electricity[PJ]

0

20

40

60

80

100

120

140

160

180

2007

2015

2020

2025

2030

2035

Fuels[PJ]

BAUpotential

Additionalcost effectivepotential

Additionaltechnicalpotential

Remainingconsumption

Fig. 5. Energy demand of all processes in the frozen-efficiency scenario and resulting saving potentials by scenario.

Table 9Technical EEI and CO2 mitigation potentials by process and EEM in relation to the frozen-efficiency electricity/fuel/heat demand.

Energy savingsa CO2 mitigationa

Electricity [PJ] Fuels [PJ] Electricity [%] Fuels [%] [kt CO] [%]

2020 2035 2020 2035 2035 2035 2020 2035 2035

Pulp and paper industry 4.85 12.25 19.25 33.84 16% 21% 1803 2985 19%Chemical pulp 0.15 0.99 e e 39% 0% 25 108 11%Black liquor gasification 0.15 0.99 e e 25 108

Mechanical pulp 1.45 2.47 0.38 0.28 31% 14%b 264 282 37%Heat recovery (TMP, GW) e e 0.38 0.28 19 14High efficiency grinding (GW) 0.88 1.74 e e 148 189Enzymatic pre-treatment 0.25 0.32 e e 42 35Efficient refiner and pre-treatment 0.32 0.41 e e 54 45

RCF pulp 1.07 2.33 0.18 0.43 14% 4% 189 275 12%High consistency pulping 0.14 0.25 e e 23 27Efficient screening 0.39 0.92 e e 66 101Heat recovery e e 0.18 0.43 9 21De-inking flotation optimization 0.34 0.89 e e 58 97Efficient disperser 0.20 0.27 e e 33 29

Paper 2.18 6.46 18.70 33.13 13% 24% 1325 2320 19%Efficient refiners 0.74 2.24 e e 124 244Optimization of refining 1.03 1.33 e e 173 144Chemical modification of fibers 0.41 2.90 0.46 3.28 93 476Steam box e e 0.72 0.73 37 35Shoe press e e 1.0 2.92 82 144New drying techniques e e 2.50 12.64 128 616Heat recovery and integration e e 13.41 688 660

a The relative potentials are related to the electricity/fuel demand or CO2 emissions of the process in the frozen-efficiency scenario. The CO2 emissions comprise directemissions from fuels combustion and indirect emissions from electricity generation.

b The 14% fuel savings for mechanical pulp represent the increase in the excess-heat already recovered in this process.

T. Fleiter et al. / Energy 40 (2012) 84e9994

Table 10Resulting specific energy and fuel consumption (SEC) by scenario and process.

Process Base year BAUdiffusion

Cost-effectivediffusion

Technicaldiffusion

2007 2020 2035 2020 2035 2020 2035

Specific electricity consumption (SEC) [GJ/t]Chemical pulp 2.3 2.3 2.0 2.3 2.0 2.2 1.4Mechanical pulp 7.9 7.7 7.4 7.6 7.2 6.8 5.5RCF pulp 0.9 0.9 0.9 0.9 0.8 0.9 0.8Paper 1.9 1.9 1.8 1.8 1.7 1.8 1.7

Specific fuel consumption (SEC) [GJ/t]Chemical pulp 12.6 12.6 12.6 12.6 12.6 12.6 12.6Mechanical pulp �2.0 �2.1 �2.2 �2.3 �2.3 �2.3 �2.3RCF pulp 0.5 0.5 0.5 0.5 0.5 0.5 0.5Paper 5.5 5.3 5.0 4.8 4.6 4.8 4.2

T. Fleiter et al. / Energy 40 (2012) 84e99 95

The discount rate is often used in energy demand models asan ad-hoc way to consider barriers to cost-effective EEM. Ahigher discount rate represents barriers like information deficits,capital constraints, capacity and knowledge constraints, or moregenerally bounded rationality. Fig. 7 shows how an increasingdiscount rate increases the slope of the CSC and consequentlyreduces the number of cost-effective saving options. However,the resulting potential is not fundamentally different whencomparing a 15% with a 50% discount rate. The main reason isthat the EEM with very short payback times remain cost-effectiveeven for high discount rates. Discussions with paper mill repre-sentatives showed that a payback time of 2 years is often used asa threshold for investment decisions, while at the same time,most of the equipment has a technical lifetime of more than 10years. However, longer payback times are accepted for strategicinvestments that also improve the production process and gobeyond pure energy efficiency motivated retrofits (see alsoRef. [67]).

Fig. 7 also shows the change of the CSC depending on the year.Generally, technologies become more cost-effective as a result offalling investment costs and increasing energy prices. The size ofthe potentials also increases, due to continuing technologydiffusion.

Fig. 6. CO2 abatement cost curve

7. Discussion and conclusions

7.1. Energy-saving potentials

The analysis demonstrates that further EEI potentials exist in thepaper industry. Although the SEC of paper production has hardlyimproved in the past 15 years, new technologies are being devel-oped that could result in considerable EEI. The analyzed 17 tech-nologies would result in an EEI of 21% (34 PJ/a) for fuels and 16%(12 PJ/a) for electricity until 2035 e as compared to the frozen-efficiency development. These EEI would result in CO2 abatement ofabout 3 Mt/a, or 19% of the frozen-efficiency development. Inaddition, cross-cutting technologies (not considered in this study)may result in further savings.

Technologies with the greatest EEI potentials mainly address thecore paper production process. Heat integration promises a fuelsaving potential of 13.5 PJ/a, of which most is realized before 2020.Innovative drying techniques could result in substantial savings ofaround 13 PJ/a. However, this is a long term option, which developsits effects only after 2020. Two options, efficient refiners and thechemical modification of fibers, contribute most to the electricitysavings, with more than 2 PJ/a each until 2035.

Technology diffusion rates are critical assumptions for thecalculated potentials. New technologies only enter the technologystock when old technologies are retired. Consequently, we consid-ered the differential costs between a conventional technology andthe energy-efficient technology. This assumption generally resultsin comparably slow technology diffusion, but low costs.

Most of the assessed technologies were found to be cost-effective, given the assumed development of energy and CO2 pri-ces. Cost-effectiveness was defined from a firm or decision-makerperspective, assuming a discount rate of 15%. Given the ambitiousexpectations of firms on payback time, as well as the non-monetarybarriers discussed, it is likely that large parts of these potentials willnot be implemented in a BAU scenario.

Rising energy and CO2 certificate prices would further improvethe cost-effectiveness of the technologies considered. However,within the European emissions trading scheme (EU ETS), the paperindustry is not obliged to purchase all CO2 emission certificates, atleast up to 2020, in order to maintain a level playing field withinternational competitors. Instead, paper mills receive a free

for 2035 (discount rate 15%).

Fig. 7. Sensitivity of Conservation supply curves to different discount rates for the year 2035 (left) and to different years with a 15% discount rate (right).

T. Fleiter et al. / Energy 40 (2012) 84e9996

allocation up to a certain CO2 benchmark according to their CO2

efficiency. From an economic theory point of view, the financialincentive to mitigate emissions as induced by the EU ETS is notaffected. Firms with emissions above the benchmark are obliged topurchase certificates on the market; firms below the benchmarkhave the opportunity to sell the remaining certificates. In bothcases, an emission certificate has the same value and thus inducesthe same incentive to mitigate emissions. In practice it might bedifferent. As discussed in Section 3.2, the adoption of energy-efficient technologies not only depends on financial profitability,but also on a number of other factors and barriers. In a similar wayas for energy prices, the incentives from emission certificate pricesare also not likely to take full effect [68]. However, so far, only littleempirical evidence is available for the impact of the emissionstrading scheme on the paper industry. A first analysis of the impactof the EU ETS on the paper industry found that although paper millsclearly take the EU ETS into account in their technology adoptiondecision, other market factors are more influential [69].

As many of the observed barriers to the adoption of energy-efficient technologies are non-monetary, additional policy instru-ments could further contribute to exploiting the cost-effectivesaving potential (no-regret potential). These range from energymanagement to R&D support. The close collaboration between thepaper mill and the technology supplier is essential, particularly forcomplex process technologies. A significant share of the calculatedsaving potentials was due to process innovations, which stilldepend on successful technology development. Thus, technologysuppliers represent an important target group for energy efficiencypolicy in the pulp and paper industry. Furthermore, risk ofproduction interruption is a main barrier in the paper industry [35].Demonstration plants could contribute by demonstrating the reli-ability of energy efficiency innovations and support their marketentry. Significant potentials are available also in the short termthrough optimization of production plants in place. The mainoption is a better integration of heat sources and sinks. Energymanagement and intensive energy audits as well as improvedprocess control systems help exploiting these short term potentials.

Technology-specific assessments of EEI potentials are typicallyunable to cover all potential possibilities for EEI. All options beyondthe chosen system boundaries are excluded from the analysis (seeFig. 1). Among them are fuel switch, increased use of recycled fibers(see Ref. [70]), radical process innovations like dry-sheet forming,reduction of paper consumption,15 EEI in cross-cutting technologies

15 For example through e-book readers, printing on demand or even in-officepaper recycling [71e73].

and end-of-pipe solutions, like carbon capture and storage. Further,it is very likely that new innovations that were unknown at thetime of this study will emerge in the future.

7.2. Modeling methodology

The methodology used allows a transparent comparison of theimpact of different energy-efficient technologies and aggregation ofEEI potentials for the entire paper industry. Explicitly consideringSEC of processes and process steps improved the accuracy of theassumptions. The focus on technology diffusion over time added animportant dimension to the analysis of EEI potentials, which isfrequently ignored. It allowed us to distinguish between the degreeof maturity of the technologies in the assumptions as well as in theresults.

However, the paper industry shows a great degree of hetero-geneity at the plant level with regard to specific energy demand,but also firm size and technology structure. Consequently, certaintechnologies are cost-effective in single paper mills, while they arestill too expensive in others. This effect is not considered with thechosen approach based on average values. This becomes moreimportant, if technology costs were considered as an endogenousvariable, which they are not in our current model. In this case,further technology diffusion would result in falling costs, whichagain accelerates diffusion.

A further critical input is the breakdown of the average energydemand to the level of processes and process steps in Section 4,because the savingpotential is givenas a shareof theenergydemandin a process step. Again here, the lack of data and the heterogeneityof paper processes (e.g. among paper grades) increases the uncer-tainty of the results. Distinguishing paper grades in such an analysiswould help to reduce the level of uncertainty, but wouldmost likelynot affect our general conclusions.

The approach of using an adjusted discount rate to simulatebarriers to energy efficiency showed certain disadvantages. Amongthem is the fact that a higher discount rate only increases the slopeof the CSC, and the low cost options remain cost-effective even forvery high discount rates. In reality, however, even these options arenot implemented in all cases, due to the existence of a variety ofbarriers. We assumed an exogenous diffusion rate for the BAUscenario. While this approach allowed us to extrapolate pastdiffusion rates on a technology level, it certainly has its inconve-niences as well, among which is the non-elasticity of energydemand and technology diffusion for changes in energy prices.

Furthermore, the technology stock is not explicitly modeled byrepresenting different vintages. Technology stock turnover is onlyimplicitly assumed in the exogenous assumptions on technology

T. Fleiter et al. / Energy 40 (2012) 84e99 97

diffusion. Explicitly considering the technology stock would improveboth the transparency and the dynamics of themodeling approach. Itwould further allow differentiating more explicitly between tech-nologies that enter the market via modernization and those tech-nologies that are bound to the plant turnover. The latter wouldexperience a slower diffusion.

8. Conclusions

In this paper we assessed options to improve energy efficiencyin the German pulp and paper industry up to the year 2035. Thispaper combines engineering studies of energy-efficient technolo-gies with bottom-up modeling of energy demand and savingpotentials. 17 process technologies are assessed, resulting ina technical saving potential of 34 TJ/a for fuels and 12 TJ/a forelectricity by 2035. These represent 21% of the fuel demand and 16%of the electricity demand of the pulp and paper industry. The largerpart of this potential was found to be cost-effective from a firmperspective. In terms of CO2 mitigation, these energy savingstranslate to 3 Mt CO2 in 2035. The most influential technologieswere heat recovery in the paper mill and the use of innovative,highly efficient paper drying technologies. In conclusion, significantsaving potentials are still available in the German pulp and paperindustry. However, the potentials are limited if we assume thatcurrent paper production processes would not undergo radicalchanges. Further savings would be possible if the system bound-aries of this study were extended to include cross-cutting tech-nologies, paper recycling or the increased replacement of fibers byless energy-intensive additives.

Table 11Energy carrier prices as assumed for all scenarios [V/GJ].

2007 2015

Electricity 19.72 15.71Light fuel oil 16.33 17.75Hard coal 4.03 4.71Lignite 4.03 4.71Natural gas 7.69 9.47Heavy fuel oil 7.16 11.87Liquefied petroleum gas, refinery gas 7.69 9.47Waste 2.01 2.35Biomass 4.83 5.65District heat 13.15 15.28

Source: Own calculations based on reference scenario from [64].

Table 12Assumed development of CO2 emission certificates for all scenarios.

2007 2015

Price EUAs [V/tCO2] 1a 15

a Certificate prices fell dramatically at the end of the first trading period in 2007.Source: Own calculations based on reference scenario from [64].

Table 13CO2-intensity per energy carrier [t CO2/GJ].

Energy carrier 2007 2015

Electricitya 0.179 0.172Light fuel oil 0.074 0.074Hard coal 0.094 0.094Lignite 0.112 0.112Natural gas 0.056 0.056Heavy fuel oil 0.078 0.078Liquefied petroleum gas, refinery gas 0.060 0.060Waste 0.046 0.046Biomass 0 0District heat 0.082 0.08

a Source: own calculations based on reference scenario from [64].

The modeling methodology proves useful for the analysis;however, certain potentials for improvement remain. Furtherinsights could be gained, by explicitly considering the technologystock and the age distribution of technologies for the modeling ofthe technology diffusion path. Furthermore, relying on averagevalues does not adequately represent the huge heterogeneity in thepaper industry. If energy consumption data were available, differ-entiating among paper grades would enable more explicit consid-eration of the structure and niches in the paper industry. We haveshown that the assessment of cost-effectiveness greatly dependson the discount rate assumed, as well as the shape of the cost curve.As long as the cost curve consists of average values of single EEM,adapting the discount rate to simulate barriers to energy efficiencyis only a very rough estimation that does not adequately reflectreality. This is particularly the case when large EEM instantlybecome cost-effectivewith only little price or discount rate changes(penny-switching effect).

Acknowledgments

This research article is partly based on work conducted ina research project funded by the German Environmental Agency.

We thank the anonymous reviewers of this paper for theirconstructive and helpful comments and suggestions.

Appendix

2020 2025 2030 2035

13.89 15.56 17.22 19.1718.39 20.16 21.94 23.225.02 5.6 6.18 6.85.02 5.6 6.18 6.8

10.28 10.97 11.67 12.3614.01 15.9 17.8 18.8410.28 10.97 11.67 12.362.51 2.8 3.09 3.46.02 6.72 7.41 8.16

16.29 18.17 20.06 22.08

2020 2025 2030 2035

20 25 30 35

2020 2025 2030 2035

0.168 0.151 0.133 0.1090.074 0.074 0.074 0.0740.094 0.094 0.094 0.0940.112 0.112 0.112 0.1120.056 0.056 0.056 0.0560.078 0.078 0.078 0.0780.060 0.060 0.060 0.0600.046 0.046 0.046 0.0460 0 0 00.078 0.07 0.062 0.051

T. Fleiter et al. / Energy 40 (2012) 84e9998

References

[1] IEA. Energy technology perspectives: scenarios and strategies to 2050. Paris:OECD/IEA; 2008.

[2] VDP. Papier 2010-Ein Leistungsbericht. Bonn: Verband Deutscher Papierfab-riken e.V; 2010.

[3] ODYSSEE Database. Energy efficiency indicators in Europe; 15-7-2011.[4] Davidsdottir B. Capital vintage and climate change policies: the case of US

pulp and paper. Environmental Science and Policy 2004;7(3):221e33.[5] Giraldo L, Hyman B. An energy process-step model for manufacturing paper

and paperboard. Energy 1996;21(7e8):667e81.[6] Giraldo L, Hyman B. Energy end-use models for pulp, paper, and paperboard

mills. Energy 1995;20(10):1005e19.[7] Ozalp N, Hyman B. Energy end-use model of paper manufacturing in the US.

Applied Thermal Engineering 2006;26(5e6):540e8.[8] Szabó L, Soria A, Forsström J, Keränen JT, Hytönen E. A world model of the

pulp and paper industry: demand, energy consumption and emissionscenarios to 2030. Environmental Science and Policy 2009;12(3):257e69.

[9] IUTA, PTS, LTT, and EUTech. Branchenleitfaden für die Papierindustrie. Duis-burg, Institut für Energie- und Umwelttechnik e.V.; PTS PapiertechnischeStiftung; RWTH Aachen Lehrstuhl für Technische Thermodynamik. EUtechEnergie & Management GmbH; 2008.

[10] Brand J, Pinter H, Sattler M, Schweighofer M, Starzer O, Tretter H. Chancen derPapierindustrie im Rahmen der Klimastrategie: Energieeffizienz und CO2Emissionen der Österreichischen Papierindustrie - Entwicklungen undPotenziale. Wien: Austropapier; 2005.

[11] Kramer K, Masanet E, Worrell E. Energy efficiency opportunities in the U.S.Pulp and paper industry. Energy Engineering: Journal of the Association ofEnergy Engineering 2010;107(1):21e50.

[12] European IPPC Bureau. Integrated pollution prevention and control (IPPC) -draft reference document on best available techniques in the pulp and paperindustry. European Commission; 2010.

[13] Bakhtiari B, Fradette L, Legros R, Paris J. Opportunities for the integration ofabsorption heat pumps in the pulp and paper process. Energy 2010;35(12):4600e6.

[14] Laurijssen J, De Gram FJ, Worrell E, Faaij A. Optimizing the energy efficiency ofconventional multi-cylinder dryers in the paper industry. Energy 2010;35(9):3738e50.

[15] Martin A. Energy analysis of impulse technology: research-scale experimentalpapermaking trials and simulations of industrial applications. AppliedThermal Engineering 2004;24(16):2411e25.

[16] Martin N, Anglani N, Einstein D, Khrushch M, Worrell E, Price LK. Opportu-nities to improve energy efficiency and reduce greenhouse gas emissions inthe U.S. pulp and paper industry. Berkeley: Ernest Orlando Lawrence BerkeleyNational Laboratory (LBNL); 2000.

[17] Möllersten K, Yan J, Westermark M. Potential and cost-effectiveness of CO2reductions through energy measures in Swedish pulp and paper mills. Energy2003;28(7):691e710.

[18] Farahani S, Worrell E, Bryntse G. CO2-free paper? Resources, Conservation andRecycling 2004;42(4):317e36.

[19] Chateau B, Lapillonne B. Long-term energy demand forecasting a newapproach. Energy Policy 1978;6(2):140e57.

[20] Worrell E, Price L. Policy scenarios for energy efficiency improvement in firms.Energy Policy 2001;29:1223e41.

[21] Lapillonne B, Chateau B. The MEDEE models for long term energy demandforecasting. Socio-Economic Planning Sciences 1981;15(2):53e8.

[22] Daniels BW, Van Dril AWN. Save production: a bottom-up energy model forDutch industry and agriculture. Energy Economics 2007;29(4):847e67.

[23] Fleiter T, Worrell E, Eichhammer W. Barriers to energy efficiency in industrialbottom-up energy demand modelsea review. Renewable and SustainableEnergy Reviews 2011;15(6):3099e111.

[24] Murphy R, Rivers N, Jaccard M. Hybrid modeling of industrial energyconsumption and greenhouse gas emissions with an application to Canada.Energy Economics 2007;29(4):826e46.

[25] Hasanbeigi A, Menke C, Therdyothin A. The use of conservation supply curvesin energy policy and economic analysis: the case study of Thai cementindustry. Energy Policy 2010;38(1):392e405.

[26] Worrell E, Biermans G. Move over! Stock turnover, retrofit and industrialenergy efficiency. Energy Policy 2005;33(7):949e62.

[27] Meier KA. Supply curves of conserved energy. Berkeley: Lawrence BerkeleyLaboratory, University of California; 1982.

[28] Worrell E, Ramesohl S, Boyd G. Advances in energy forecasting models basedon engineering economics. Annual Review of Environment and Technologies2004;29:345e81.

[29] DeCanio SJ. The efficiency paradox: bureaucratic and organizational barriersto profitable energy-saving investments. Energy Policy 1998;26(5):441e54.

[30] Jaffe AB, Stavins RN. The energy-efficiency gap - what does it mean? EnergyPolicy 1994;22(10):804e10.

[31] Ostertag K. No-regret potentials in energy conservation - an analysis of theirrelevance, size and determinants. Heidelberg: Physica-Verlag; 2003.

[32] Schleich J. Barriers to energy efficiency: a comparison across the Germancommercial and services sector. Ecological Economics 2009;68(7):2150e9.

[33] de Almeida E. Energy efficiency and the limits of market forces: the exampleof the electric motor market in France. Energy Policy 1998;26(8):643e53.

[34] de Groot HLF, Verhoef ET, Nijkamp P. Energy saving by firms: decisionmaking, barriers and policies. Energy Economics 2001;23(6):717e40.

[35] Thollander P, Ottosson M. An energy efficient Swedish pulp and paperindustry: exploring barriers to and driving forces for cost-effective energyefficiency investments. Energy Efficiency 2008;1(1):21e34.

[36] del Río González P. Analysing the factors influencing clean technologyadoption: a study of the Spanish pulp and paper industry. Business Strategyand Environment 2005;14(1):20e37.

[37] BlumO,MaurB,ÖllerH-J.Revisionof best available technique referencedocumentfor the pulp & paper industry: use of energy saving techniques; 2007. München.

[38] VDP. Papier 2009-Ein Leistungsbericht. Bonn: Verband Deutscher Papierfab-riken e.V; 2009.

[39] Worrell E, Neelis M, Price L, Galitsky C, Nan Z. World best practice energyintensity values for selected industrial sectors. Berkeley: Ernesto OrlandoLawrence Berkeley National Laboratory; 2007.

[40] Joelsson JM, Gustavsson L. CO2 emission and oil use reduction through blackliquor gasification and energy efficiency in pulp and paper industry.Resources, Conservation and Recycling 2008;52(5):747e63.

[41] Naqvi M, Yan J, Dahlquist E. Black liquor gasification integrated in pulp andpaper mills: a critical review. Bioresource Technology 2010;101(21):8001e15.

[42] Ruohonen P, Hippinen I, Tuomaala M, Ahtila P. Analysis of alternative secondaryheat uses to improve energy efficiencyecase: a Finnish mechanical pulp andpaper mill. Resources, Conservation and Recycling 2010;54(5):326e35.

[43] Leinonen A. A long search for energy efficient wood grinding - MikaelLucander was awarded for his innovative research. Link - KCL’s customermagazine for re-inventing paper 2006; 1/2006: 10e12.

[44] Viforr S. D1.1.11 Enzymatic pre-treatment of wood chip for energy reductionsat mechanical pulp production - A review. 2008. Ecotarget: New and inno-vative processes for radical changes in the European pulp & paper industry,project within the EU Sixth Framework Programme. 25-3-2010.

[45] Sabourin M. Energy savings in TMP using high efficiency refining. Appleton;2006.

[46] Gorski D, Hill J, Engstrand P, Johansson L. Review: reduction of energyconsumption in TMP refining through mechanical pre-treatment of woodchips. Nordic Pulp and Paper Research Journal 2010;25(2):156e61.

[47] Viforr S, Salmén L. From wood shavings to mechanical pulp - a new rawmaterial? Nordic Pulp and Paper Research Journal 2005;20(4):418e22.

[48] Brettschneider W. Energiekostenreduzierung - auch in der Stoffaufbereitung,eine Herausforderung. Twogether - Magazin Für Papiertechnik 2007;24:11e5.

[49] Gabl H. Papillon - a new refining concept. IPW - The Magazine for theInternational Pulp & Paper Industry 2004;2:43e50.

[50] Dekker JC. Kompressionsmahlung: Die Entwicklung einer neuen Mahltech-nologie zur Energieeinsparung bei der Papierherstellung, PTS Paper Sympo-sium 2008, 2008.

[51] Erhard K, Arndt T, Miletzky F. Einsparung von Prozessenergie und SteuerungvonPapiereigenschaften durch gezielte chemische Fasermodifizierung. Euro-pean Journal of Wood and Wood Products 2010;68(3):271e80.

[52] Stumm D. Untersuchungen zum chemischen Wasserrückhaltevermögen undzur Trocknungsfähigkeit von Papierstoffen unter besonderer Ber-ücksichtigung der Rolle von chemischen Additiven. Fachbereich Chemie,Technische Universität Darmstadt; 2007.

[53] Bos JH, Staberock M. Papierbuch: Handbuch der Papierherstellung. ECA Pulpand Paper; 2006.

[54] Luiten E, Blok K. The success of a simple network in developing an innovativeenergy-efficient technology. Energy 2003;28(4):361e91.

[55] Luiten E, Blok K. Stimulating R&D of industrial energy-efficient technology.Policy lessonseimpulse technology. Energy Policy 2004;32(9):1087e108.

[56] De Beer J, Worrell E, Blok K. Long-term energy-efficiency improvements in thepaper and board industry. Energy 1998;23(1):21e42.

[57] LfU and PTS. Klimaschutz durch effiziente Energieverwendung in derPapierindustrie - Nutzung von Niedertemperaturabwärme. München: Bayr-isches Landesamt für Umweltschutz (LfU); 2002.

[58] Sivill L, Ahtila P, Taimisto M. Thermodynamic simulation of dryer section heatrecovery in paper machines. Applied Thermal Engineering 2005;25(8e9):1273e92.

[59] Sivill L, Ahtila P. Energy efficiency improvement of dryer section heat recoverysystems in paper machines - a case study. Applied Thermal Engineering 2009;29(17e18):3663e8.

[60] Bujak J. Energy savings and heat efficiency in the paper industry: a case studyof a corrugated board machine. Energy 2008;33(11):1597e608.

[61] Worrell E, Laitner JA, Ruth M, Finman H. Productivity benefits of industrialenergy efficiency measures. Energy 2003;28:1081e98.

[62] Weiss M, Junginger M, Patel MK, Blok K. A review of experience curve anal-yses for energy demand technologies. Technological Forecasting and SocialChange 2010;77(3):411e28.

[63] Jardot D, Eichhammer W, Fleiter T. Effects of economies of scale and experi-ence on the costs of energy-efficient technologies: a case study of electricmotors in Germany. Energy Efficiency 2010;3(4):331e46.

[64] Prognos, EWI, and GWS. Energieszenarien für ein Energiekonzept der Bun-desregierung. Basel/Köln/Osnabrück: Prognos AG; EWI;GWS; 2010. ProjektNr. 12/10 des Bundesministeriums für Wirtschaft und Technologie.

[65] Kesicki F, Strachan N. Marginal abatement cost (MAC) curves: confrontingtheory and practice. Environmental Science and Policy 2011;14(8):1195e204.

T. Fleiter et al. / Energy 40 (2012) 84e99 99

[66] Stoft SE. The economics of conserved-energy supply curves. The EnergyJournal 1995;16(4):109e37.

[67] Cooremans C. Make it strategic! Financial investment logic is not enough.Energy Efficiency; 2011:1e20.

[68] Schleich J, Rogge K, Betz R. Incentives for energy efficiency in the EU emissionstrading scheme. Energy Efficiency 2009;2(1):37e67.

[69] Rogge KS, Schleich J, Haussmann P, Roser A, Reitze F. The role of the regulatoryframework for innovation activities: the EU ETS and the German paperindustry. International Journal of Technology, Policy and Management 2011;11(3):250e73.

[70] Nathani C. Modellierung des Strukturwandels beim Übergang zu einermaterialeffizienten Kreislaufwirtschaft - Kopplung eines Input-Output-

Modells mit einem Stoffstrommodell am Beispiel der Wertschöpfungskette“Papier”. Heidelberg: Physica; 2002.

[71] Counsell T, Allwood J. Desktop paper recycling: a survey of novel technologiesthat might recycle office paper within the office. Journal of Materials Pro-cessing Technology 2006;173(1):111e23.

[72] Hekkert M, Vandenreek J, Worrell E, Turkenburg W. The impact of materialefficient end-use technologies on paper use and carbon emissions. Resources,Conservation and Recycling 2002;36(3):241e66.

[73] Moberg Å, Johansson M, Finnveden G, Jonsson A. Printed and tablet e-papernewspaper from an environmental perspective e A screening lifecycle assessment. Environmental Impact Assessment Review 2010;30(3):177e91.