Journal of Cleaner Production 142 (2017) 3117e3130

Contents lists available at ScienceDirect

Journal of Cleaner Production

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

Carbon risk and optimal retrofitting in cement plants: An application

of stochastic modelling, MonteCarlo simulation and Real Options
Analysis
Luis M. Abadie a, Nestor Goicoechea b, Ibon Galarraga a, *
a
Basque Centre for Climate Change (BC3), Spain
b
Escuela de Ingeniería de Bilbao, University of the Basque Country UPV/EHU, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The cement sector is highly intensive in CO2 emissions and is the second biggest industrial sector in
Received 31 May 2016 terms of emissions after the electricity generation sector. It emits CO2 from the combustion of fossil fuels,
Received in revised form the calcination process and, indirectly, from electricity consumption. The ambitious climate change
24 October 2016
policy in the EU means that carbon prices and fuel prices are two very important sources of uncertainty
Accepted 26 October 2016
Available online 28 October 2016
that may affect the competitiveness of the sector. This paper focuses on understanding the risk associ-
ated with the future price of European Union Emission Trading System allowances. This is done by
modelling a stochastic process with parameters calculated using market prices. Risks are valued with the
Keywords:
Cement plants
Expected Shortfall and Value at Risk measures over the lifetime of a plant using MonteCarlo simulation:
Uncertainty two well-known risk measures in financial economics. Risks are greater and returns lower in the wet
MonteCarlo simulation process than in the dry one. The paper includes a sensitivity analysis of the effects arising from changes
Carbon emissions in the prices of allowances as a consequence of a hypothetical drastic change in climate policy, including
Real options jumps, withdrawing of free emission allowances and changes in future carbon prices. In this case impacts
Expected shortfall will also be higher in the wet process. Finally, the paper illustrates the optimal conditions for retrofitting
a wet cement plant to convert it to a dry cement plant under uncertainty of the price of carbon al-
lowances. This is done using Real Options Analysis. The trigger price is V114 million for a plant with
remaining lifetime of 25 years and a production of one million tonnes.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction On the other hand, in regions where the industry has being
going through a recession phase, such as Europe, production is
The cement sector is highly emission-intensive. Apart from the much lower than capacity. This is the case of the cement industry in
use of fossil fuels, this is due to the calcination process itself, pro- the EU 28, where a maximum of 190.8 Mtonnes of grey clinker was
ducing clinker that releases a considerable amount of CO2. In fact, produced in 2007 (World Business Council for Sustainable
this sector is considered to be the world's second biggest source of Development, 2015) (See Table 1).
anthropogenic emissions (Cembureau Activity Report, 2015). The economic crisis, climate policies and technological im-
Emerging countries produce a considerable amount of cement provements in cement production have recently led to a significant
(Fig. 1). China in particular has emerged as an industrial super- drop in emissions in Europe. This trend is illustrated for the EU(28)
power and has become the world's second largest economy ac- in Fig. 2. Note that for the same production of 225 million tonnes of
cording to the economy ranking for 2010 (World Development cement 1995 and 2009 the CO2 emissions are nearly 15% lower in
Indicator Database, 2016). The total capacity of its 798 plants was the latter year. The drop experienced in cement production has also
1409.8 Mtonnes of cement per year in 2013 (Global Cement pushed down CO2 emissions significantly.
Directory, 2015). In order to better understand the source of CO2 emissions in the
cement industry one must be aware that the clinkering or calci-
nation process is the decomposition of CaCO3 to CaO and CO2. The
* Corresponding author. high quantity of emissions in this sector stems from this chemical
E-mail addresses: [email protected] (L.M. Abadie), nestor.goikoetxea@
ehu.eus (N. Goicoechea), [email protected] (I. Galarraga).
reaction, where the source of CO2 is related to the carbonates and

http://dx.doi.org/10.1016/j.jclepro.2016.10.155
0959-6526/© 2016 Elsevier Ltd. All rights reserved.
3118 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130

Fig. 1. World cement production 2014 by region and main countries (%) (Cembureau Activity Report, 2015).

Table 1
Trends in clinker production and emissions in the EU 28 (WBCSD, 2015).

Year 1990 2000 2005 2006 2007 2008 2009 2010 2011 2012 2013

Clinker (Mtonnes) 177 173 177 183 191 178 143 142 140 125 120
Emissions (Mtonnes) 164 157 158 163 171 158 126 125 122 109 103

Fig. 2. EU-28 (96% Coverage 2013) Grey and white cement production and emissions (excluding CO2 from on-site power generation) (drawn up by the authors with data from CSI,
2015).

the fuel used in a stoichiometric relationship in line with the law of material being mixed with water in the former.
conservation of mass (Masterton, 1989). At the burning stage the slurry is introduced into the kiln. The
The raw materials used in the pre-burning stage, during slurry fuels most commonly used in kilns for pyroprocessing are coal,
preparation, are calcareous materials including limestone, chalk, natural gas and petroleum coke. In a wet process more heat is
magnesium carbonate, silica and others. The two main groups of required than in a dry process to obtain the same amount of clinker
technology used are known as wet processes and dry ones, and the (Gartner, 2004).
main difference between them is found at this stage, with raw In the last ten years the European cement industry has striven to
 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130 3119

mitigate CO2 emissions by adopting different strategies: carbon (2015) evaluate overall CO2 emissions from the cement industry
capture and storage (CCS) demonstration, employing more energy based on detailed information from all China's 1574 cement en-
efficient processes, shifting to more efficient plant processes, using terprises in 2013. They conclude that ownership of cement com-
alternative products as raw materials and/or using lower-emitting panies should be carefully considered in policy preparation as
fuels. In fact there has been a major effort to reduce the use of publicly-owned plants usually tend to emit less.
fossil fuels and some plants are now using up to 80% alternative The type of fuel used to obtain energy is a very important factor
fuels. in determining CO2 emissions and both fuel and carbon prices can
Table 2 shows the long-term trend for the EU(28) in the significantly affect emissions. Accordingly, Pardo et al. (2011)
composition of fuels used in cement kilns. Fossil waste fuel and analyse three scenarios: a baseline scenario representing the cur-
biomass have gained in share over time, reflecting an effort towards rent trend in the cement sector and two alternative scenarios
greater sustainability. representing the sensitivity of fuel prices and CO2 emission prices,
It has been estimated that by 2050 only 40% of kiln energy may respectively. They conclude that increases in fuel prices or CO2
potentially come from traditional sources, i.e. coal (30%) and pet- emission prices could lead to significant decreases in the compet-
coke (10%): the remaining 60% would be provided by alternative itiveness of the European cement industry. García-Gusano et al.
fuels, 40% of which could be biomass (Cement Technology (2015) discuss trends in the cement industry in Spain using
Roadmap, 2009). A number of policy modifications will be cement demand projections and an energy optimisation model,
required to achieve these reductions (Cembureau, 2015). Some and find that the deployment of the EU ETS Directive will decrease
examples are the use of Best Available Technologies (BAT) such as CO2 emissions by 8Mt per year.
multistage preheaters, precalcination technology, retrofit wet The paper presented here looks at the potential for reducing
production processes and long dry kilns or the use of alternative emissions by switching fuels (traditional, alternative and biomass)
fuels, with special attention to waste heat recovery to avoid land- and analyses the risk that carbon policies may pose for the cement
filling. Understanding the drivers of these changes is very industry. The methodology enables us to deal with both un-
important. certainties derived from fuel prices and prices of EU ETS Allow-
Several authors have also argued in favour of carbon capture and ances. The EU ETS puts a price on CO2 emissions to enhance carbon
storage (CCS), especially post combustion CO2 capture (IPPC 2016). emission reductions and acts as a technology-neutral, cost-effective
Interesting papers on this topic include Barker et al. (2008), driver for low-carbon investments (Directive 2009/29/EC). The
Gutierrez et al. (2014) and Benhelal et al. (2013). The latter ana- European Commission regularly reviews the system and a number
lyses energy and emission savings from more efficient processes of free allowances are allocated to prevent carbon leakage. Thus,
and considers CCS as an effective means of reducing CO2 emissions. together with fuel costs the cement industry faces uncertainty
Romeo et al. (2011) propose the integration of a power plant, a about the price of emission allowances and even about the amount
cement plant and a CO2 capture system. The integration of a coal- of rights that may be freely allocated. This paper applies a stochastic
fired power plant with an Enhanced Oil Recovery (EOR) facility model consistent with market prices and MonteCarlo simulation
and CCS is studied in Abadie et al. (2014) from the financial point of for investment risk appraisal. MonteCarlo simulation is used in
view. However, despite the progress made toward their commer- combination with other methodologies of risk analysis. This
cialisation none of these carbon capture technologies is yet avail- method enables us to obtain a good approximation of the theo-
able (Hills et al., 2015). retical distribution in all cases, even when there is no analytical
Ali et al. (2011) review emission sources in the cement industry solution or the solution is very complex such as in the case of jump
and different techniques for reducing emissions and recognise that, processes (Abadie and Chamorro, 2013). In fact, with Montecarlo
apart from CCS, using alternative fuels to replace fossil fuels may be simulation it becomes relatively easy to calculate Expected Shortfall
a sensible way to reduce carbon emissions. Regarding improve- (ES) with a high degree of accuracy and for a great number of tra-
ments in cement plants, Hasanbeigi et al. (2012) study the infor- jectories. The method is also suitable for the analysis of American
mation available for eighteen emerging technologies in the cement options (Longstaff and Schwartz, 2001). In addition, a Real Options
industry, focussing on describing the processes, the energy savings Analysis (ROA) exercise is also presented. This work comprises both
and the environmental benefits. Economic instruments such as an optimisation and a cement valuation model under uncertainty.
carbon trading mechanisms are recognised as very interesting tools The methodology presented here offers a solid grounding for
for enhancing emerging low-carbon technologies. rigorous, sophisticated financial analysis of investment decisions
Energy consumption is also determinant for CO2 emissions. Xu and represents a significant contribution to the literature. Similar
et al. (2012) highlight the importance of using BAT that allow for applications have been drawn up for energy investment analysis in
a 26% potential energy saving and 33% CO2 mitigation. Cai et al. Abadie and Chamorro (2008, 2013), for CCS in Abadie et al. (2014)

Table 2
EU 28 Cement plant energy consumption in gigajoule and weighted average fuels (CSI, 2015).

Energy Consumption (grey and white cement) in EU 28 GJ Weighted average % total energy

Year Fossil waste Biomass Fossil fuel Total Fossil waste Biomass Fossil fuel

1990 15,950,790 1,527,993 719,717,929 737,196,712 2.2% 0.2% 97.6%

2000 49,926,614 9,028,415 611,655,833 670,610,862 7.4% 1.3% 91.2%
2005 73,437,064 25,526,798 580,536,115 679,499,977 10.8% 3.8% 85.4%
2006 94,801,150 31,854,404 574,722,020 701,377,574 13.5% 4.5% 81.9%
2007 109,318,171 31,738,749 597,665,771 738,722,691 14.8% 4.3% 80.9%
2008 112,928,557 36,396,292 537,084,427 686,409,276 16.5% 5.3% 78.2%
2009 119,321,039 35,032,885 398,368,097 552,722,021 21.6% 6.3% 72.1%
2010 131,757,284 33,612,772 380,140,775 545,510,831 24.2% 6.2% 69.7%
2011 136,618,459 47,484,574 359,515,557 543,618,590 25.1% 8.7% 66.1%
2012 124,063,320 52,971,428 312,598,114 489,632,862 25.3% 10.8% 63.8%
2013 115,224,304 60,859,047 285,614,348 461,697,699 25.0% 13.2% 61.9%
3120 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130

and for commodity prices in Kolos and Ronn (2008). Stochastic calculated as the standard deviation of the daily returns Rt: (Abadie
processes and ROA are being used increasingly for resource man- and Chamorro, 2013).
agement investment decisions and represent a growing field of

research (Trigeorgis, 1996; Longstaff and Schwartz, 2001). St
Rt ¼ ln (1)
The rest of the paper is structured as follows: Section 2 focuses St1
on carbon prices, their stochastic model, the calculation of the
parameters, the risk model, fuel prices and emissions. Section 3 where St and St-1 are the prices of two consecutive trading days.
addresses the stochastic model of carbon prices and the calcula- This gives the value of sd. To obtain the annualised volatility s
tion of the parameters, together with the calculation of the carbon the following standard finance equation (2) is used:
price risk for two types of technology (wet and dry). This risk pffiffiffiffiffiffiffiffiffi
assessment is conducted for the two technologies using the ES s ¼ sd 252 (2)
measure. Values for Value at Risk (VaR) are also given for com-
parison. The impacts of a jump in the price of emission allowances where 252 is the number of trading days.
and a potential reduction in the free distribution of allowances are Volatility prices using a 50-day window are represented in
also studied as an illustration of a drastic change in climate policy. Fig. 4. The 50-day window is calculated for each day with the
Additional sensitivity analysis is performed for unexpected changes volatility of returns of that trading day and the 49 days preceding,
in spot and futures prices. Finally the real option of retrofitting a as is commonly done in financial economics (Abadie and Chamorro,
wet plant to a dry plant is also analysed. Section 4 offers some 2013). Using the prices for 2015 included in the sample it is possible
discussion on the results and the potential use of the methodology to calculate s ¼ 0.3111.
presented here. With this information we now develop a stochastic model to
estimate the present CO2 cost in the following section.

2. Material and methods

2.2. The risk model
2.1. Carbon prices
Risk measurements are useful analytical tools in situations of
For a better understanding of carbon markets the trend in uncertainty and are often used in economics to account for price
emission spot prices from 04/01/2010 to 24/07/2015 is represented uncertainties in investment projects (Abadie and Chamorro, 2013).
in Fig. 3. This is long enough for the type of analysis performed in Due to the related volatility in emission prices this tool suits the
this paper. Periods of high volatility and a general upward price cement model well.
trend in the last few months of the series used can be observed, Risk-neutral measurements are usually used in derivative pric-
with a spot price of S0 ¼ V7.99/tonne on the last day of the series. ing models and cash flows can be discounted at the risk-neutral
Spot prices are very low but seem to be very volatile. Low carbon interest rate by altering the drift. On the other hand risk mea-
prices and their high volatility do not greatly stimulate investments surement methods rely on stochastic models such as the one
in mitigation such as improvements in energy efficiency (Abadie developed in the previous section. The two main risk metrics are
and Chamorro, 2008). Value at Risk (VaR) and Expected Shortfall (ES). Expected Shortfall
Volatility is a manifestation of uncertainty (Abadie and is also called Conditional Value at Risk (CVaR) and Expected Tail
Chamorro, 2008). It plays a determining role in valuing real op- Loss (ETL). Both measures enable us to deal with low probability
tions and risk. When trading day quotas are involved it is usually but high impact situations, as explained later.

Fig. 3. Carbon dioxide emission spot prices (from 01/04/2010 to 07/24/2015).

 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130 3121

Fig. 4. Carbon spot price volatility (from 01/04/2010 to 07/24/2015).

2.2.1. Value at risk (VaR) reactions can take place at 2000  C. The main thermal energy
This is a measurement that is widely used for calculating and expenditure takes place during the burning process (Marciano,
managing the risk of a portfolio V. It requires two elements: a time- 2003), which can be obtained from coal, fuel oil, petroleum coke,
frame Tt and a confidence level x%. It is based on calculating a natural gas or diesel (Madlool et al., 2011). The fact that the clinker
value a that has a probability of only 100%-x% of being exceeded. In pyro process offers the possibility of burning different types of non-
our case we have V ¼ V(t1,t2). The values of x% most frequently used conventional fuels is a twofold advantage: the fuel selection choice
are 95%, 97.5% and 99%. That is, VaR(95%) gives the value above is wider and the process itself can become more environmentally
which the 5% of worst cases (i.e.low probability events) arise. friendly. So the use of alternative fuels has been increasing quickly.
An accurate result can be obtained if a GBM distribution process
is used to generate log normal distributions for each time frame. 2.3.1. Petroleum coke
Petcoke is a by-product of the refinery industry. Its main char-
2.2.2. Expected shortfall (ES) acteristics are its high gross calorific value and its low price. In-
This is the expected value in those cases in which the value dustry is the main user of petcoke, primarily in electric power
obtained is greater than VaR(x%) and it is a much more suitable risk generators (Wang et al., 2004). In terms of sulphur content, petcoke
measure than VaR. In fact, despite its widespread use, VaR has poor depends on the nature of the crude residue and on the coking
mathematical properties for optimisation applications as it is non- feedstock used in the oil refinery. Two types of petcoke are avail-
convex. (Artzner et al., 1999). able: high- and mid-sulphur. During the calcination process clinker
A coherent risk measurement R(D) in which D represents losses takes up all the sulphur and negative effects on the air such as SO2
or damage must meet the conditions of monotonicity, sub- emissions are minimised.
additivity, positive homogeneity and translation invariance, The price of high-sulphur petcoke is close to US$40-45/tonne
where the risk is represented by a negative value. Those conditions FOB US Gulf (Global Cement Directory, 2015) and that of mid-
are that it must be: sulphur petcoke is around US$55/tonne. Its being more expensive
means that the latter is not often used.
a) monotonous: R(D)0. Petcoke production is independent of market demand because
b) Sub-additive: R(D1þD2)R(D1)þR(D2). petcoke is a byproduct of oil refineries. If heavy fractions are turned
c) Positive homogeneous: R(kD) ¼ kR(D). into light fractions such as aviation fuels, less petcoke is produced
d) Translation invariant: R(D þ l) ¼ R(D)-l. (Santos and Silva, 2008). Curiously, fluctuations in petcoke prices
depend largely on changes in the aluminium sector. The electrodes
VaR as a measurement of risk lacks several desirable properties, used in aluminium manufacturing are made of petcoke and this
for instance it is not sub-additive, so it is not coherent in the sense affects its price. Venezuela is a major producer of petcoke and the
of Artzner et al. (1999). Other authors who argue in favour of using range of its prices is considerable and can fluctuate greatly. It has
ES are Szego € (2002) and (2004). Moreover, Yamai and Yoshiba even dropped to US$6/tonne at times (Personal Interview).
(2005) illustrate how the tail risk of VaR can cause serious prob- In this paper we use a price of US$42.5/tonne, with an average
lems in certain cases, in which ES can serve more aptly in its place. exchange rate for 2014 of 1.3286 $/V (European Central Bank),
ES is however a coherent risk measurement, as it meets all the giving us a price of V31.99/tonne.
above conditions. ES is the expected value when the damage is
greater than VaR (x%) for the case of the damage considered here. 2.3.2. Alternative fuels and biomass
The use of biomass and alternative fuels benefits resource sav-
2.3. Fuel prices ings, waste management alternatives, improvements in the local
economy and mitigation of climate change.
Cement kilns require a large amount of energy so exothermic A wide range of different raw materials are used to produce
3122 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130

alternative fuels. Cementers or outsourced stakeholders therefore references the amount of clinker produced and uses default values
need to collect, sort and size-adjust depending on the stage at instead of Total Organic Carbon (TOC) for calculating emissions.
which the kiln is to be fed. All these tasks are carried out in a pre- A cement plant directly produces emissions from:
treatment plant which is usually located close to the cement plant.
The nature of alternative fuels varies according to what burnable  The calcination process.
products are produced and available locally, so supply channels  The combustion of fuels
differ widely from country to country. The regulations on RDF  Indirect emissions.
(Refuse Derived Fuels) also vary from country to country.
The fuel prepared must be homogeneous and needs to be To calculate total CO2 emissions the amount for each source is
controlled in order to avoid blockages in cyclone towers (pre- calculated independently as shown below.
calciner) due to chlorine or sulphated agents (Tokheim et al., 2001).
Note that temperatures and retention times must be high enough 2.4.1. Emissions from raw material calcination
to eliminate problems, such as those related to dioxins and furans Emissions are produced by the calcination process itself and the
(Lockwood and Ou, 1993). combustion of organic carbon in the raw material, with CO2 being
In 2004 6.1 million tonnes of different types of waste were used released into the atmosphere according to Equations (3) and (4):
in Europe. Availability, the level of gate fees and the reasons
explained above are the main drivers of the economics of RDF us- CaCO3 (þ950  C) / CaO þ CO2 (3)
age. In this paper information provided by a cement company
(Personal Interview) leads us to consider a possible price for MgCO3 (þ950  C) / MgO þ CO2 (4)
alternative fossil fuels of V22/tonne and V20/tonne for biomass
fuels. where CaCO3 is calcium carbonate, MgCO3 is magnesium carbonate,
CaO is calcium oxide and MgO is magnesium oxide.
2.4. Emissions from cement plants In the absence of better data, a default of 525 kg CO2/t clinker is
recommended by the Cement Sustainability Initiative (CSI) in
There are four processes for manufacturing cement: dry, semi- simple output method B1 (World Business Council for Sustainable
dry, semi-wet and wet. The main differences depend largely on Development, 2015). This value is comparable to the IPCC default
whether the raw material to be placed in the kiln is dry or wet. The (IPCC, 2006) (510 kg CO2/t) corrected for the typical MgO content in
higher the moisture content, the more energy is consumed. clinker of 2%.
A large part of world clinker production depends on wet pro- The IPCC method correlates with a default CaO contained in
cesses. However, 75% of the processes for grey cement in the Eu- clinker of 65% including a 2% correction for discarded dust.
ropean Union are dry (Personal Interview).
As can be seen in Table 3, the dry process increased its share in 0.785  Fraction CaO ¼ 0.785  0.65 ¼ 0.510 t CO2 / t clinker (5)
Europe from 72% in 1990 to 84% in 2013, with the share accounted
for by the dry process with preheater and precalciner rising where 0.785 is (44.01 g/mole CO2)/(56.08 g/mole CaO).
particularly. The IPCC recommends 2% for discarded dust, which means
The design of cement plants varies widely around the world, 0.525  1.02 ¼ 0.5355 tonnes CO2/tonne clinker. In addition, the
even for the same process. Even in Europe a large variety of facilities CO2 from organic carbon in raw material per clinker unit is
can be found. In order to ensure reasonable calculations, a proto- included:
type kiln factory that produces 3000 tonnes per day or close to
1,000,000 tonnes per year is analysed in this paper. We use data 1.55  Organic Fraction x  44/12 (6)
from a stylised cement plant that can perfectly well represent an
average plant in Europe. Note however that the model can be easily where 1.55 is the raw material/clinker ratio and “Organic Fraction”
adapted to different characteristics to calculate emissions from any refers to the fraction of total organic carbon in the raw material,
specific plant. with a default value of 0.2%. 44/12 is the conversion between CO2
The cement plant emission model proposed is a bottom-up and C. In short, the calcination process is as follows:
model. The methodology used to calculate emissions is consistent
with The Cement CO2 and Energy Protocol proposed by the World EFc ¼ 0.547 t CO2 / t clinker (7)
Business Council for Sustainable Development (WBCSD); more
precisely the method is based on clinker output method B1. This Where EFc is the emission factor per tonne of clinker corresponding
method does not take account of the raw meal consumed: it to the calcination process.

Table 3
EU(28) total production of grey cement per kiln type (World Business Council for Sustainable Development, 2015).

Year Total dry Dry with preheater & with precalciner Dry with preheater & without precalciner Other dry Total wet

1990 72% 26% 40% 6% 28%

2000 75% 36% 36% 3% 25%
2005 77% 42% 32% 3% 23%
2006 77% 43% 31% 3% 23%
2007 78% 43% 32% 3% 22%
2008 79% 43% 33% 3% 21%
2009 81% 46% 32% 3% 19%
2010 82% 44% 33% 5% 18%
2011 82% 44% 33% 5% 18%
2012 82% 46% 29% 7% 18%
2013 84% 48% 29% 7% 16%
 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130 3123

This value is a realistic approximation, although it can vary from according to the BAT, along with the fuel share or mix to be used,
place to place as the CaO content can vary from region to region. the number of kg of fuel to be fed in the kiln is calculated according
Cement kiln dust, a grained solid material that flows out in cement to its GCV. The specific energy consumption per tonne of clinker
exhaust gases, can also lead this value to vary due to the ratio used already takes into account that the kiln is fed using alternative and
in the recommended adjustment, which is around 2e6%. biomass fuels. Thus, applying the default values determined in
Table 6 and the fuel prices mentioned in subsection 2.4, the values
2.4.2. Emission from fuel combustion shown in Table 5 are obtained. This table summarises the emissions
The combustion emission factor of a given kiln depends on the and costs from fuel consumption per tonne of clinker using the
cement production processes, energy efficiency and fuels used 2013 mix as per Table 2 for wet and dry processes.
(Shen et al., 2014). Canales et al. (2003) consider that the best The costs in the dry process are therefore lower and a quarter of
available process is a rotary kiln fitted with multistage cyclone a tonne of CO2 is saved per tonne of clinker produced. If a typical
preheaters and precalcinating. These authors acknowledge that the plant produces 3000 tonnes/day of clinker, roughly equivalent to
thermal energy required in this process is around 2900e3200MJ/ one million tonnes/year, the savings in emissions in fossil fuel
tonne clinker. According to the Best Available Technologies (BAT, combustion obtainable by changing from a wet to dry process
2013), the energy demand for clinkerisation is 3500 MJ/tonne amount to 250,830 CO2 tonnes/year. If the remaining lifetime is 25
clinker in the dry process and 6400 MJ/tonne clinker in the wet years, the savings add up to nearly 6.27 million tonnes of CO2.
process.
The fossil fuels used can be coal, petcoke, fuel oil or natural gas. 2.4.3. Indirect emissions from electricity consumption
Regions such as the United States and the former Soviet Union use a The cement industry consumes a large amount of electricity.
higher percentage of gas in their fuel than Europe, where the share Electricity generators based on fossil fuel directly emit considerable
of gas is insignificant due to its high price. In that sense, as stated quantities of CO2. Thus, consumption of electricity can be consid-
above, the use of non-conventional fuels offers new possibilities in ered in some cases as indirectly emitting CO2. Many cement plants
terms not only of availability costs but also of emission costs. have implemented a self generating electricity system involving
The emission factors of the fuels (Ff) and the Gross Calorific waste heat recovery (WHR). In these cases the electricity consumed
Value (GCV) of fossil, alternative and biomass fuels are shown in from WHR should be taken into account for electricity-related
Table 4. The default values are those proposed by IPCC Guidelines emissions (Shen et al., 2014).
for National Greenhouse Gas Inventories (IPCC, 2006) and when no As equipment is not standardised across all plants it is not rare
such values are defined the values from the 2011-2014 WBCSD for electricity consumption to differ from one plant to another.
Cement Sustainability Initiative (CSI)/ European Cement Research Generally speaking the consumption of a modern dry cement plant
Academy (ECRA) Cement CO2 and Energy Protocol are used (*). It is about 110e120 KWh per tonne of cement (Alsop, 2014).
must also be noted that for calculating the gross calorific value the Electricity is consumed mainly in the operations to grind raw
simple average is used when the value is contained within a range materials prior to their burning and clinker additions to obtain
(**). cement. Between them these two operations account for around
The CO2 emissions from biomass fuel are considered as climate- 75% of the electricity consumed in the plant, with the other 25%
neutral and therefore not taken into account. On the other hand, being due to transportation of materials, gas impulsion and electro-
CO2 emissions from alternative fossil fuels are not a priori climate filters.
neutral. The electricity consumption analysis is divided according to the
In order to calculate emissions and fuel cost a simple analysis is cement processes: pre-burning, burning and grinding stages.
required. Once the heat required in the process is determined According to Avami and Sattari (2007) electricity consumption

Table 4
Different types of fuel: Emission default value and gross calorific value. Source: Own work using data from (IPCC, 2006) and (CSI, 2015).

Category IPCC default value KgCO2/GJ Gross calorific Value KJ/Kg

Fossil fuels

Coal - anthracite 96.1e98.3 24,500 **

Petroleum coke 97.5 34,100
(Ultra) heavy fuel 77.4 * 43,000
Diesel oil (Bunker oil) 74.1 44,800
Natural gas (dry) 56.1 43,000
Oil shale 107 e
Lignite 101 18,400 **
Motor gasoline 69.3 47,300
Alternative fossil fuels (simple average) 94.76 23,217
Waste oil (lubricants) 73.3 34,100 **
Tyres 143 29,900 **
Plastics 75* 31,000 **
Solvents 74* 21,000
Impregnated sawdust 75* 22,600 **
Mixed industrial waste 143 28,603
Other fossil based wastes 80* e
Biomass fuels(simple average) 97.71 14,825
Dried sewage sludge 54.6 10,000
Wood, non impregnated sawdust 112 15,900
Paper, carton 110* 15,000
Animal & bone meal 100 18,400
Agricultural, organic, diaper waste, charcoal 100e112 e
Other biomass 95.3e112 e
3124 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130

Table 5
Fuel cost and emissions per tonne of clinker: Wet and dry processes.

Product Mix PCS J/Kg Emissions KgCO2/GJ Wet process Dry process

Fuel cost V Emissions Kg Fuel cost V Emissions Kg

Pet coke 61.8% 34,100 97.5 3.77 391.66 2.03 210.89

Alternative fuel 25.0% 23,217 94,76 1.54 153.98 0.83 82.92
Biomass fuel 13.2% 14,825 0,00 1.16 0,00 0.62 0.00
Total 100% 28,835 6.47 544.64 3.48 293.81

in dry and wet processes is as shown in Table 6. Table 8

Note that the unit used is kWh/cement, and to turn it into kWh/ Total cost with remaining useful lifetime.

clinker a conversion factor is needed. That factor depends heavily Remaining useful lifetime Cost Wet process Dry process
on the characteristics of the plant. In this case we use a clinker 25 years Total (million V) 248.51 111.33
conversion factor in the cement of 0.756 (Cement Technology Fuels 161.64 87.04
Roadmap, 2009), so consumption is 154.76 kWh/tonne clinker in Emissions 86.86 24.29
the dry process and 197.10 kWh/tonne clinker in the wet process. 30 years Total (million V) 302.41 134.77
Fuels 193.97 104.45
Thus, the indirect emissions from electricity consumption are 0.033
Emissions 108.44 30.32
tonnes of CO2/tonne clinker in the wet process and 0.026 tonnes of 35 years Total (million V) 357.57 158.68
CO2/tonne clinker in the dry process. Fuels 226.30 121.85
The means of transport are not taken into account as a second Emissions 131.67 36.82
40 years Total (million V) 415.31 183.08
indirect CO2 emission factor in this study, as we are considering a
Fuels 258.63 139.26
plant located close to a quarry and no clinker is outsourced for Emissions 156.68 43.82
cement production.

obtain the total cost. Thus, if the output of the plant is one million
2.4.4. Total emission and cost
tonnes/year the current value of the savings over 30 years is around
Table 7 shows the total emissions taking into account the three
V167.64 million. The savings for a remaining useful lifetime of 25
sources for the two types of process:
years amount to V137.18/tonne clinker. We take 25 years as the
If emission allowances of 0.766 tonnes CO2/tonne of grey clinker
standard case from here onwards, even though the model proposed
are considered, the costs would be 0.100 tonnes CO2/tonne clinker
can be adapted to different figures.
in the dry process for grey clinker and 0.359 tonnes CO2/tonne
clinker in the wet process. The emission allowance for electricity
consumption is considered to be paid within its cost and the price 3. Theory and calculations
of those allowances is considered as indirectly passed on to clinker
manufacturing. 3.1. Stochastic emission price model
If the fuel price is considered to grow at the risk-free interest
rate r and the expected value of the price of emissions is that of the In the real world the model is as follows:
stochastic process, then initially the real value is obtained for the
costs per tonne of clinker for a remaining useful lifetime as set out dSt ¼ aSt dt þ sSt dWt (8)
in Table 8:
where St is the carbon price at time t,a is the drift in the real world,
Changing from a wet to a dry process means a saving of V167.64
per tonne of clinker produced per year (V89.52 fuel cost and 78.12
s is the volatility and dWt stands for the increment to a standard
Wiener process.
emission allowances) over 30 years. It must be taken into account
The risk-neutral version of the model is:
that the cost increases each year due to the emission allowances,
whose value is expected to increase over the risk-free rate. The dSt ¼ ða  lÞSt dt þ sSdWt (9)
values shown in Table 10 must be multiplied by annual output to
where lSt is the market price of risk. Kolos and Ronn (2008) esti-
Table 6
mate the market price of risk for energy markets.
Kilowatt hour per tonne of cement. If X ¼ lnS, and Ito's lemma is applied, then the following is
obtained:
Dry process Wet process
 s
1.-Pre-burning 48 63
dXt ¼ a  l  dt þ sdWt (10)
2.-burning 29 33 2
3.-Grinding 53 53
Total 130 149 In this case, F(T,t) the value of a future with maturity T at time tis
obtained from equation (11):

Table 7 Fðt; TÞ ¼ St eðalÞðTtÞ (11)

Total emissions per tonne of clinker.
In equation (11) the price of the future depends only on the
Wet process Dry process initial value St, the trend a-l and the time to maturity T-t.
Calcination 0.547 0.547 Equation (11) shows that a commodity with a Geometric
Fuel consumption 0.546 0.294 Brownian Motion (GBM)-type behaviour in its spot price should
Electricity 0.033 0.026 exhibit prices on the futures market that increase by larger
Total 1.125 0.866
amounts in absolute value as the maturity period increases. This
 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130 3125

behaviour can be used to identify whether a commodity such as Table 9

carbon is a good candidate for GBM modelling. Parameters of the gbm model.

Fig. 5 and Table 9 show the estimated values of al for each Parameter S0 a-l s r t1 t2
trading day. The trend can be estimated using the prices of the 2015 Value 7.99 0.0219 0.3294 0.0071 0 25
futures market included in the sample (see Fig. 6), where the
following is obtained: a-l ¼ 0.0219,with a t-ratio of 80.88. This
shows a substantial model adjustment with real futures quotes, and
The final values obtained must be close to the theoretical
this a-l value shows low growth in futures prices as maturity
average and variance:
increases.
The trend in futures prices for different maturities is repre-
sented in Fig. 6. E0 ðS25 Þ ¼ S0 e25ðflÞ ¼ 13:81 (14)
Another parameter needed is the risk-free rate r. We use the July
2015 interest rate for 10-year German government bonds: And the standard deviation
r ¼ 0.0071.
The parameter values of the stochastic emission price model are pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
S0 e25ðalÞ es t  1 ¼ 44:21
2
summarised as shown below in Table 9: (15)
The present value of a tonne of CO2 emitted every year between
The values obtained by simulating a million trajectories are
t1 and t2 is calculated as per Abadie and Chamorro (2013): V13.80 and V43.96. The current value of a yearly emission allow-
  ance over 25 years is V241.68, a figure that is very close to the
eðalrÞt2  eðalrÞt1 theoretical value of V241.72. Values for the VaR (95%) ¼ V708.34
Vðt1 ; t2 Þ ¼ S0 (12)
alr and ES(95%) ¼ V1256.19 are obtained. In other words, when the
annual average per tonne over 25 years is V241.68, it will exceed
The results obtained for the case of a cement plant according to
V708.34 in 5% of the cases. When this value is exceeded the average
its remaining useful lifetime are summarised in Table 10.
value will be V1256.19, which is the result of the high volatility of
This information is crucial for guiding decision-making under
carbon prices, which involves a considerable risk. As already ana-
uncertainty. To complement this view, two risk measures are esti-
lysed, 0.100 tonnes CO2/tonne clinker are needed for a dry process,
mated in the following section.
while the figure is 0.359 tonnes CO2/tonne clinker for a wet process,
which gives the values in Table 11:
3.2. Carbon price risk in a cement plant The risk for each specific plant would have to be multiplied by
its annual clinker production, for example, by a million if that is the
The different equiprobable trajectories of the performance of annual output. Changing from a wet to dry process reduces the
emission allowance prices can be simulated by using a standard risks notably. Fig. 7 shows the distribution of the 5% of worst cases.
Montecarlo simulation. Now we introduce a more complex risk model:


pffiffiffiffi
al12s2 Dtþs Dtεt a) Assume that there is a 50% probability in 2020 (within five
StþDt ¼ St e ; (13) years) of withdrawing 0.100 emission allowance per tonne of
clinker, which would then be auctioned.
Where Dt is the step size measured in years and εt is the standard b) In 2020 there could be a jump in the price of emission allow-
normal variate. ances with an expected value of V5/tonne COS and 0.20

Fig. 5. Daily estimate of drift.

3126 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130

Fig. 6. Carbon futures prices.

Table 10 happens 0.200 would be necessary for the dry process and 0.459 for
Carbon dioxide emission cost and remaining useful lifetime per tonne of carbon the wet process.
per year.
As can be seen in Table 12, the risks have increased considerably,
Remaining useful lifetime Present value CO2 cost which would make the dry process more favourable than the wet
25 years 241.72 V process from the carbon risk price perspective, as the risks have
30 years 301.75 V increased more in the wet process than in the dry process. Greater
35 years 366.39 V uncertainty increases the risks of less efficient technologies.
40 years 435.99 V
Although the actual prices of carbon futures are coherent with

Table 11
Carbon risk at a cement plant per tonne of clinker per year.

Carbon price risk Wet process risk Dry process risk

Mean 241.68 86.76 24.17

Value at risk VaR (95%) 708.34 254.29 70.83
Expected Shortfall (95%) 1256.19 450.97 125.62

volatility in the jump. In other words, the jump has a normal demand and supply as well as with expected future behaviour, one
distribution. could argue that unexpected changes may occur as a consequence,
c) This is all superimposed on the stochastic model used. for instance, of changes in climate policy, technology or other socio-
economic factors. If such a change affects prices over a long period
For t ¼ 5 the following is obtained before the possible jump: of time the value of a e l may change. If the unexpected change is

 sudden and takes immediate effect the carbon spot price S0 may
pffiffiffiffi change. Table 13 shows a sensitivity analysis for an unexpected
al12s2 Dtþs Dtεt
S ¼ S5Dt e ; (16) change in both parameter a e l and S0. If these values change, the
5
parameters of the model have to be recalculated but the sensitivity
analysis shows that the model is stable.
And after the jump; Sþ 
5 ¼ S5 þ J (17)
Where Jt ¼ J(mJ,sJ)dq, mJ is the expected value of the jump, sJ is 3.3. The option of retrofitting a wet plant to a dry plant
the volatility of the jump and dq has a value of 1 when t ¼ 5 and
0 otherwise. The reduction of CO2 is an important aspect to consider with a
On the other hand the withdrawal has a 50% probability of view to reducing emissions of greenhouse gases. The use of waste
affecting the allowances needed per tonne of clinker. If this heat as an alternative source of energy has already been discussed.
 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130 3127

Fig. 7. Queue log-normal distribution 95th percentile.

Table 12 1970 with an estimated capacity of three million tonnes per year
Carbon risk cement plant per tonne of clinker per year (with jumps). were retrofitted to dry kilns with a clinker production capacity of
Wet process risk Dry process risk 2.3 million tonnes (Joint Implementation Project Design, 2007).
Mean 142.48 50.95
In this section we analyse the option of retrofitting a wet plant to
Value at risk VaR (95%) 398.14 149.00 a dry one using Real Option Analysis (ROA). This methodology en-
Expected Shortfall (95%) 683.51 261.74 ables us to calculate the value of the option to retrofit as well as the
maximum cost that makes the retrofit a good alternative. We use
the price of carbon markets for this analysis. The numerical calcu-
Table 13 lation instruments used here are very similar to the ones used for
Carbon risk sensitivity at a cement plant per tonne of clinker per year. financial options. All details on the construction of binomial tress
Wet process risk and backward optimisation are described in Abadie and Chamorro
a-l 0.010 0.015 0.020 0.025 0.030 (2013) and Trigeorgis (1996).
Mean 74.36 79.27 84.61 90.43 96.78 With a remaining useful lifetime of 25 years, retrofitting a wet
Value at risk VaR (95%) 213.28 229.41 247.12 266.47 287.77 plant to a dry plant means obtaining savings (emission allowance
Expected Shortfall (95%) 371.92 402.99 437.13 474.65 515.92
Dry process risk
and fuel) with a net present value of V137.18 million. Therefore if
a-l 0.010 0.015 0.020 0.025 0.030 the current cost of retrofitting a cement plant is lower than that the
Mean 20.71 22.08 23.57 25.19 26.96 Net Present Value (NPV) is positive and the decision would be to
Value at risk VaR (95%) 59.41 63.90 68.84 74.23 80.16 invest in retrofitting immediately. However, this may not be the
Expected Shortfall (95%) 103.60 112.25 121.76 132.21 143.71
optimum time for investment under uncertainty. The optimum
Wet process risk time under Real Option Analysis is usually when the cost is lower
S0 7.00 8.00 9.00 than a frontier value or trigger price (I*).
Mean 76.01 86.87 97.73
Value at risk VaR (95%) 222.78 254.61 286.44
To calculate that value we construct a binomial tree that rep-
Expected Shortfall (95%) 395.09 451.54 507.98 resents the performance under uncertainty of the price of emission
Dry process risk allowances. In our case we have a remaining lifetime of T and we
S0 7.00 8.00 9.00 use a step size of Dt ¼ 60
1 ; which is equivalent to 60 steps per year or
Mean 21.17 24.20 27.22
5 steps per month. If the savings generated during the remaining
Value at risk VaR (95%) 62.06 70.92 79.79
Expected Shortfall (95%) 110.05 125.78 141.50 lifetime are invested in the plant the savings A(t,S) for a plant that
produces (P) one million tonnes of clinker a year will initially be:
 
Another potential way of mitigating CO2 is to use energy saving eðalrÞT  1
technologies. According to Caruso (2007) and Ogbeide (2010) AðT; S0 Þ ¼ S0 Pð1:125  0:866Þ þP
alr
bigger emission savings can be achieved by switching from a wet
 Tð6:47  3:48Þ (18)
process to a dry one. This assertion is shown in Table 10. A wet
process is not really an option for newer plants, but retrofitting Where the first summand is the current value of the emission
existing plants from wet to dry is an interesting option, and one allowance price saved over T years, with 1.125e0.866 being the
which has in fact been adopted at several plants, such as in the one number of allowances saved per tonne of clinker.
in Podilsky (Ukraine). The six kilns built there for the wet process in The second summand is the fuel saving.
3128 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130

When time passes within the binomial tree and n steps have Table 14
elapsed the saving is lower due to a shorter remaining time T-t ¼ T- Trigger price for retrofitting a wet plant.

nDt, but the initial price of the emission allowance may have taken Remaining lifetime I* NPV ¼ 0
another value St, which may be lower or higher. In that case the 20 92.84 107.97
saving at an intermediary node of the binomial tree is: 21 97.10 113.76
  22 101.33 119.59
eðalrÞðTtÞ  1 23 105.53 125.47
24 109.71 131.39
AðT  t; St Þ ¼ St Pð1:125  0:866Þ þP
alr 25 113.86 137.35
26 117.99 143.36
 ðT  tÞð5  2:69Þ (19)
27 122.10 149.42
28 126.17 155.51
After one period the emission allowance price can go either up
29 130.22 161.66
or down: 30 134.26 167.85
pffiffiffiffi pffiffiffiffi 31 138.27 174.09
Sþ ≡uS ¼ S0 es Dt ; S ≡dS ¼ S0 es Dt ; (20) 32 142.26 180.38
33 146.23 186.72
34 150.17 193.11
with probabilities:
35 154.10 199.54
pffiffiffiffi 36 158.01 206.03
eðalÞDt  es Dt 37 161.91 212.58
pu ¼ pffiffiffiffi pffiffiffiffi ; pd ¼ 1  pu ; (21) 38 165.78 219.17
es Dt  es Dt 39 169.64 225.82
First we assess the decision to invest at the end. Thus, if we have 40 173.48 232.52
the option to invest at the expiry time T and the cost of the in-
vestment is I we have NPV ¼ A(0,ST)-I and our decision is means that no allowances are required for such fuel).
maxðAð0; ST Þ  I; 0Þ, the result is logically going to be zero at the The cement industry faces both risk and profitability challenges.
terminal nodes. Improved profitability comes from the mix of fuels and from the
At the intermediary nodes the decision will be to invest at that technology used. The dry process described in this paper is much
time t or to postpone the possibility of investing until later, more efficient than the wet process, given that it uses less fuel and
whichever has the higher value. also emits less.
  Two of the main risks facing cement plants result from uncer-
maxðAðT  t; St Þ  I; erDt pu Wþ þ pd W (22) tainty as to the prices of fuel and emission allowances. In particular,
they are subject to uncertainty in future climate policy, which could
Where Wþ and W are the values at the following nodes: affect the amount of emission allowances available on the market
At the initial time the value is: and consequently their price. But they are also subject to future
uncertainly as to the amount of allowances that they can receive for
 
maxðAðT; St Þ  I; erDt pu Wþ þ pd W (23) free.
This paper presents a cost model for clinker production based on
The question that now arises is this: given S0¼7.99, what is the emission allowances and fuel prices, as well as the fuel mix used for
threshold investment cost I below which it is optimal to invest production. It is a risk assessment optimisation and control model
immediately? This will indeed be the best course of action when under uncertainty.
the NPV exceeds the continuation value. Note, though, that both The paper focuses on the effects of uncertainty on CO2 emission
parts of the comparison depend on I. The value of I depends on the allowance prices. The model is applied to both wet and dry process
remaining lifetime of the plant T. for a given fuel mix.
The results of the trigger price I* and the NPV according to the We show how to value risks through the ES by means of Mon-
remaining lifetime are set out in Table 14 and Fig. 8. tecarlo simulation, with parameters obtained by calibrating the
As can be observed, when the remaining lifetime is shorter a model using market prices. The results show a greater risk and a
lower cost I* is required to make the investment and when the lower return in the wet process than in the dry one. The risks are
remaining lifetime is longer there are greater differences between also shown to grow more sharply in the wet process when there is a
the cost which leads to NPV ¼ 0 and the trigger price I*. possibility of climate policy becoming stricter, which could lead to
jumps in allowance prices, and/or when there is a likelihood of
changes in the allowances allocated for free to cement plants.
4. Discussion and results Similarly climate policy can cause changes in spot and futures
prices, which would have most impact in wet plants. A sensitivity
The cement industry has the second greatest impact of all in- analysis is included to account for this.
dustries on CO2 emissions into the atmosphere and the calcination In particular, the risk measured for ES(95%) in our base case is
process significantly contributes to it. In this regard, technological V451/tonne of clinker in the remaining useful lifetime of a standard
improvements help to lower those emissions and also significantly wet cement plant. This figure decreases significantly to V126/tonne
impact on the profitability of cement plants due to lower fuel costs in a standard dry plant. The ES(95%) ranges from V395-508/tonne
and the corresponding reduction in emission allowances needed. of clinker when the spot carbon price is around V7.00e9.00/tonne
The fuel mix used for plants of this type in clinker production for a wet plant, while the ES(95%) interval for the same price in-
has recently shifted towards greater use of alternative and biomass terval is V110-141/tonne for a dry one.
fuels, which has impacted on the price of the mix given that It is tempting to compare these results with traditional scenario-
alternative and biomass fuels are both usually lower in price than based analysis but unfortunately this is not possible, mainly
traditional fossil fuels such as petcoke. Another impact of this mix is because they have no probability distribution. The method used
the need for fewer emission allowances as a result of the use of here is equivalent to having an infinite (or very large) number of
biomass fuel which are considered to be climate-neutral (which scenarios weighted with a probability that is consistent with
 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130 3129

Fig. 8. Graph showing the trigger price I* and the net present value criteria for retrofitting a wet process cement plant.

market data. In our case calculations are consistent with market his guidance in understanding the cement industry.
quotes. Those quotes are based on actual and expected future pa-
rameters. Of course, unexpected changes may occur that would References
require the risk analysis to be redone, as always occurs in any other
financial model. Abadie, L.M., Chamorro, J.M., 2008. European CO2 prices and carbon capture in-
Finally, a Real Options exercise is performed by calculating the vestments. Energy Econ. 30, 2992e3015. http://dx.doi.org/10.1016/
j.eneco.2008.03.008.
trigger price of the investment cost I* under which a wet plant Abadie, L.M., Chamorro, J.M., 2013. Investment in Energy Assets under Uncertainty.
would be retrofitted to a dry plant. That value is notably lower than Publisher: Springer, London.
the value for an NPV of zero, given that a lower cost is needed to Abadie, L.M., Galarraga, I., Rübbelke, D., 2014. Evaluation of two alternative carbon
capture and storage technologies: a stochastic model. Environ. Modell. Softw.
outstrip the value of the wait option. This explains why investments 54, 182e195. http://dx.doi.org/10.1016/j.envsoft.2014.01.002.
are sometimes not made under uncertainty even though they have Ali, M.B., Saidur, R., Hossain, M.S., 2011. A review on emission analysis in cement
a positive NPV. In particular the trigger price I* for optimal retro- industries. Renew. Sust. Energy Rev. 15, 2252e2261. http://dx.doi.org/10.1016/
j.rser.2011.02.014.
fitting a wet cement plant to a dry one is V113.9 million for a
Alsop, P., 2014. Cement Plant Operations Handbook for Dry Process Plants, sixth ed.
cement plant with a remaining lifetime of 25 years and a produc- Tradeship Publications TTd, Portsmouth United Kindom.
tion of one million tonnes of clinker per year. In such a case, the Artzner, P., Delbaln, F., Eber, J.M., Heath, D., 1999. Coherent measures of risk. Math.
Financ. 9, 203e228. http://dx.doi.org/10.1111/1467-9965.00068.
retrofit will be undertaken if the cost is lower than V113.9 mill.
Avami, A., Sattari, S., 2007. Energy conservation opportunities: cement industry in
Note that the Real Options criterion is more stringent than the NPV. Iran. Int. J. Energy 1, 65e71.
For the latter, the investment will only take place if the cost is lower Barker, D.J., Turner, S.A., Napier-Moore, P.A., Clark, M., Davison, J.E., 2008. CO2
than V137 mill. This difference is due to uncertainty being incor- capture in cement industry. In: Proceedings of the 9th International Conference
on Greenhouse Gas Control Technologies (GHGT-9). Greenhouse Gas Control
porated into the analysis with ROA but not in NPV. The results that Technologies 9, Washington DC, USA, 16-20 November 2008.
investments with positive NPV may not be worth undertaking Benhelal, E., Zahedi, G., Shamsaei, E., Bahadori, A., 2013. Global strategies and po-
when uncertainty is considered is a very interesting one which il- tential to curb CO2 emissions in cement industry. J. Clean. Prod. 51, 142e161.
http://dx.doi.org/10.1016/j.jclepro.2012.10.049.
lustrates the usefulness of the methodology applied in this paper. Best Available Techniques (BAT), 2013. Reference Document for the Production of
This model can be applied for the calculation of the carbon risk of Cement, Lime and Magnesium Oxide. Joint Research centre, Institute for Pro-
cement plants with different characteristics such as consumption of spective Technological Studies Sustainable Production and Consumption Unit
European IPPC Bureau, 2013 ISBN: 978-92-79-32944-9. (Accessed 2 August
different fuels (fossil fuels, alternative fuel, biomass), annual 2016).
output, remaining useful lifetime and/or technology. Cai, B., Wang, J., He, J., Geng, Y., 2015. Evaluating CO2 emission performance in
China Cement industry: an enterprise perspective. Appl. Energy 166, 191e200.
http://dx.doi.org/10.1016/j.apenergy.2015.11.006.
Acknowledgments Canales, C., Gimenez, O., Avellaneda, A., Monfa , E., Silva, S., Mora, P., Romay, M.,
Serra, M., 2003. Guía de Mejores Te cnicas Disponibles en Espan ~ a en la fab-
ricacio n de cemento. Centro de Publicaciones Secretaría General Te cnica,
Luis M. Abadie and Nestor Goicoechea are grateful for financial Ministerio de Medio Ambiente, Spain, 2004 ISBN: 84-8320-264-6. http://www.
support from The Spanish Ministry of Science and Innovation prtr-es.es/data/images/Gu%C3%ADa%20MTD%20en%20Espa%C3%B1a%20Sector%
(ECO2015-68023). Luis M. Abadie and Ibon Galarraga are grateful 20Cemento-BA18C5917BE0DC9D.pdf (Accessed 2 August 2016).
Caruso, H., 2007. Reduction of Carbon Dioxide Emissions from Cement Plants. Level
for financial support received from the Basque Government via of Thesis. University of Waterloo, Canada.
project GIC12/177-IT-399-13. Special thanks to Nicolas Gaminde for Cembureau Activity Report. Available online: http://www.cembureau.be/activity-
3130 L.M. Abadie et al. / Journal of Cleaner Production 142 (2017) 3117e3130

reports (Accessed 17 December 2015). 2042e2060. http://dx.doi.org/10.1016/j.rser.2011.01.005.

Cembureau. The Role of Cement in the 2050 Low Carbon Economy. Available online: Marciano, E., 2003. Sustainable Development and the Cement and Concrete In-
http://www.cembureau.be/role-cement-2050-low-carbon-economy (Accessed dustries. Level of thesis. University of Sherbrooke, Ottawa. http://savoirs.
17 December 2015). usherbrooke.ca/handle/11143/1754 (Accessed 1 January 2012).
Cement Technology Roadmap, 2009. Carbon Emissions Reductions up to 2050. Masterton, I., 1989. Chemical Principle with Qualitative Analysis. Publisher: Hart-
World business council for sustainable development, International energy court School. ISBN 0-03-006138-5.
agency. Available online: http://www.iea.org/publications. Accessed 28 August Ogbeide, S., 2010. Developing an optimization model for CO2 reduction in cement
2015. production process. J. Eng. Sci. Technol. Rev. 3, 85e88.
CSI (Cement Sustainability Initiative) GNR Projet. Available online: http://www. Pardo, N., Moya, J.A., Mercier, A., 2011. Prospective on the energy efficiency and CO2
wbcsdcement.org. (Accessed 25 August 2015). emissions in the EU cement industry. Energy 36, 3244e3254. http://dx.doi.org/
Directive 2009/29/EC of The European Parliament and of the Council of April 2009 10.1016/j.energy.2011.03.016.
amending Directive 2003/87/EC so as to improve and extend the greenhouse Romeo, L.M., Catalina, D., Lisbona, P., Lara, Y., Martínez, A., 2011. Reduction of
gas emission allowance trading scheme of the community. greenhouse gas emissions by integration of cement plants power plants, and
García-Gusano, D., Cabal, E., Lecho n, 2015. Y. Long-term behavior of CO2 emissions CO2 capture systems. Greenh. Gases Sci. Technol. 1, 72e82. http://dx.doi.org/
from cement production in Spain: scenario analysis using an energy optimi- 10.1002/ghg3.5.
sation model. J. Clean. Prod. 99, 101e111. http://dx.doi.org/10.1016/ Santos, A.R., Silva, R.J., 2008. An
alisis del consumo de coque de petro  leo en algunos
j.clepro.2015.03.027. sectores industriales. Inf. Tecnol. 19, 93e101. http://dx.doi.org/10.4067/S0718-
Gartner, E., 2004. Industrially interesting approaches to “low-co2” cements. Cem. 0718-07642008000200011.
Concr. Res. 34, 1489e1498. http://dx.doi.org/10.1016/j.cemconres.2004.01.021. Szego€, G., 2002. Measures of risk. J. Bank. Financ. 26, 1253e1272.
Global Cement Directory. Available online: http://www.globalcement.com/ Szego€, G., 2004. Risk Measures for Risk Measures for the 21st Century. Publisher:
directory (Accessed 28 December 2015). Wiley.
Gutierrez, P., Vigil-Montan ~ o, M., Burgos-Rodriguez, S., 2014. CO2 capture technol- Shen, L., Gao, T., Zhao, J., Wang, L., Wang, L., Liu, L., Chen, F., Xue, J., 2014. Factory-
ogies. Dyna 89 (4), 360e365. http://dx.doi.org/10.6036/7003. level measurements on CO2 emission factors of cement production in China.
Hasanbeigi, A., Price, L., Lin, E., 2012. Emerging energy-efficiency and CO2 emission- Renew. Sust. Energy Rev. 34, 337e349. http://dx.doi.org/10.1016/
reduction technologies for cement and concrete production: a technical review. jrser.2014.03.025.
Renew. Sust. Energy Rev. 16, 6220e6238. http://dx.doi.org/10.1016/ Tokheim, L., Gautestad, T., Axelsen, E.P., 2001. Energy recovery from wastes: expe-
j.rser.2012.07.019. rience with solid alternative fuels combustion in a precalciner kiln. In: 3rd
Hills, T., Leeson, D., Florin, N., Fennel, P., 2015. Carbon capture in the cement in- International Symposium on Incineration and Fuel Gas Treatment Technologies.
dustry: technologies, progress, and retrofitting. Environ. Sci. Technol. 2016 (50), Brussels, Belgium, July 2-4 2001.
368e377. http://dx.doi.org/10.1021/acs.est.5b03508. Trigeorgis, L., 1996. Real Options: Managerial Flexibility and Strategy in Resource
IPCC Special Report on Carbon Dioxide Capture and Storage. Available online: Allocation. The MIT Press, Cambridge.
http://www.ipcc.ch/pdf/special-reports (Accessed 2 January 2016). Wang, J., Anthony, E.J., Abanades, J.C., 2004. Clean and efficient use of petroleum
IPCC, 2006. Guidelines for national greenhouse gas Inventories. In: Eggleston, H.S., coke for combustion and power generation. Fuel 83, 1341e1348. http://
Buendia, L., Miwa, K., Ngara, T., Tanabe, K. (Eds.), Prepared by the National dx.doi.org/10.1016/j.fuel.2004.01.002.
Greenhouse Gas Inventories Programme. Published: IGES, Japan. http://www. World Business Council for Sustainable Development, the GNR Project of Cement
ipcc-nggip.iges.or.jp (Accessed 2. August 2016). Sustainable Development.Available on line: http://www.wbcsdcement.org/
Joint Implementation Project Design “Switching from Wet to Dry Process at index.php/en/key-issues/climate-protection/gnr-database (Accessed 17
Podilsky Cement, Ukraine”, 2007. PDD version 2.1, dated 2 February 2007. December 2015).
Kolos, S.P., Ronn, E.I., 2008. Estimating the commodity price of risk for energy World Development Indicators Database, World bank, Gross domestic product
prices. Energy Econ. 30, 621e641. http://dx.doi.org/10.1016/jeneco.2007.09.005. 2016, Available online: http://www.worldbank.org/reference (Accessed 15
Lockwood, F.C., Ou, J.J., 1993. Review: burning refuse-derived fuel in a rotary cement February 2016).
kiln. J. Power Energy 207, 65e67. http://dx.doi.org/10.1243/pime_proc_1993_ Xu, J.H., Fleiter, T., Eichhammer, W., Fan, Y., 2012. Energy consumption and CO2
207_008_02. emissions in China's cement industry A perspective from LMDI decomposition
Longstaff, F.A., Schwartz, E.S., 2001. Valuing american options by simulation: a analysis. Energ Policy 60, 821e832. http://dx.doi.org/10.1016/
simple least-square approach. Rev. Financ. Stud. 14 (1), 113e147. j.enpol.2012.08.038.
Madlool, N.A., Saidur, R., Hossain, M.S., Rahim, N.A.A., 2011. A critical review on Yamai, Y., Yoshiba, T., 2005. Value-at-risk versus expected shortfall: a practical
energy use and savings in the cement industries. Renew. Sust. Energy Rev. 16, perspective. J. Bank. Financ. 29, 997e1015.