Puigjaner L. (ed.) Syngas from Waste

Подождите немного. Документ загружается.

1. Level I: Corresponds to multiple phases closed systems, where pollutants do not

react, i.e., are conserved in their chemical form. Each phase is considered as a

closed vessel that attains thermodynamical equilibrium, see Chap. 2 of Mackay

[25].

2. Level II: Corresponds to steady state multiple phase open systems, where pollu-

tants are subject to advective ﬂows (related to ‘the direct movement of a chemical

by virtue of its presence in a medium that happens to be ﬂowing’), chemical

reactions (biodegradation, hydrolysis, oxidation and photolysis) and attain

physicochemical equilibrium. Each phase is considered as a CSTR where outlet

concentrations equal phase concentrations, see Chap. 2 of Mackay [26].

3. Level III: Corresponds to steady state multiple phase open systems, where

pollutants are subject to advective ﬂows, chemical reactions and diffusive ﬂows

between environmental compartments, so chemical equilibrium is used but not

attained, see Mackay [26].

4. Level IV: Corresponds to level III models where some compartments are taken

into non-steady state conditions.

In all MCMs where equilibrium is hypothesised the partitioning of a chemical

between environmental phases is described using the concept of fugacity for the

description of mass transfer and reaction phenomena.

2.3.3 Environmental Impact Estimation

The concept of environmental impact is closely related to the concept of risk,

which in many cases is embedded in the way fate, dose and impact of a chemical

compound are calculated. In the case of risk there are two analytical tools available

for such analysis: Environmental risk assessment (ERA) and Impact pathway

analysis (IPA). Both tools put emphasis on impacts to humans, in the case of ERA

emphasis is put on ingested dose, while in the case of IPA the focus is on air

concentration, see Sonnemann [27, p. 27].

Risk in the environmental sense is deﬁned by Allen and Shonnard [28]as‘the

probability that a substance or situation will produce harm under speciﬁc con-

ditions’. This risk will be the combination of two factors (Chap. 9 of Cameron and

Raman [29]): (i) the probability that the adverse event will occur and (ii) the

consequences/effects of such event. It is generally accepted that risk is a function

of a given hazard and the exposure to such hazard; considering that hazard is the

potential of a given substance/situation to produce harm or adverse effects in

people or the environment, while exposure is the contact time or exposition to such

hazard. To assess the risk, the following items have to be addressed properly:

• Hazard assessment, which addresses the question of which are the adverse

effects that a given substance or situation produces (mortality, shortened life-

span or impairment).

• Dose response is the mathematical relationship between the dose of a given

substance and the appearance of negative effects.

238 M. Pérez-Fortes and A. D. Bojarski

• Exposure assessment is linked to dose measurement and it studies how much

and which subjects are ‘exposed’ to the substance or situation.

• Risk characterisation addresses how big the adverse impact of the chemical/

situation is.

Most of the environmental metric methodologies consider a given set of emis-

sions into some compartments which are modelled using a given environmental

model. These emissions are assessed in terms of hazard/dose/exposure/risk and a

given characterisation factor (CF) is obtained which relates the emission to its

impact. Several environmental metrics have been developed within the LCA context

for LCIA (life cycle impact assessment), where two important terms are crucial to be

deﬁned appropriately, these are: impact category and environmental mechanism

(EM). An impact category represents environmental issues of concern to which

some LCI (life cycle inventory) results may be assigned. According to de Haes et al.

[30], all physical process and variables starting from extractions, emissions or other

types of interaction between the product/process system and the environment, which

are connected with a given impact category, are called the EM of that impact

category. Within and connected to a given EM it can be distinguished:

• Areas of protection (AoPs), these are variables of direct societal concern, also

known as classes of endpoints which have some well recognisable value for

society. Each impact assessment methodology has a predeﬁned set. Common

AoPs are: human health, natural resources, natural environment and man-

made environment [30].

• Category mid-points: these variables which appear within the EM of an impact

category ﬁt between environmental interventions and the impact category

endpoints. Examples are: concentration of toxic substances, deposition of

acidifying substances, global temperature or sea level.

Regarding environmental impact assessment two schools of methods have

evolved [22, 31]:

• Problem oriented or mid-point methods like CML [32, 33], EDIP [34, 35] and

TRACI [21, 36], which restrict quantitative modelling to relatively early stages

in the EM to limit uncertainties and classify and characterise emission results in

mid-point categories. Themes are common mechanisms (e.g., climate change)

or commonly accepted grouping (e.g., aquatic ecotoxicity).

• Damage oriented or endpoint methods such as Eco-indicator 99 [37] or EPS

[38] try to model the EM up to the damage to a given area of protection,

sometimes with high uncertainties. These methods differ on the way endpoint

impacts are measured and in the way that weights are assessed for each impact.

Moreover not all methods consider the same AoPs, or how each mid-point

indicator affects the endpoint.

The actual environmental impact can then be interpreted in terms of AoPs

(EI

AoP

) or category midpoints (EI

cat

), and its calculation can be performed using

Eqs. 7–9.

Global Clean Gas Process Synthesis and Optimisation 239

cat

all sinks

all species

cat

ð7Þ

AoP

all sinks

all species

AoP

ð8Þ

AoP

all cats

AoP;cat

cat

ð9Þ

where characterisation factors could be available for all different environmental

interventions (m

) in terms of mid-point impact categories (CF

cat

) or in terms of

areas of protection (CF

AoP

). Some impact assessment methods also provide with a

set of weights (W

cat;AoP

) to transform mid-point environmental impacts into

environmental impacts at the areas of protection level as in Eq. 9. In the context of

LCIA, normalisation is conducted to obtain a comprehensive view of impact

category indicator results. Normalisation values in LCIA are calculated on the

basis of chosen reference systems, e.g., all society’s activities in a given area and

over a speciﬁed period of time, or the interventions of the world as a whole in a

certain year [39, 40].

Besides emission factors, the most common way of estimating an emission is to

measure it or to model it using process simulation. The use of both methods has

been shown extensively in the literature.

The former sections discussed the different available metrics for measuring the

key parameters of a given design, the next sections discuss how to deal with such

information in two different ways: for generating different options by using multi-

objective optimisation and how to decide among different optimised and non-

dominated solutions.

3 Generation and Identiﬁcation of Alternatives

Different objective functions could be used according to the decision maker’s cri-

teria. Multiple objective programming methods aim at ﬁnding suitable solutions of

mathematical problems with multiple conﬂicting objective functions, and different

alternative strategies can be applied to solve a multiobjective problem [41, 42].

One typical approach consists of optimisation of alternating objectives, that is,

solving the problem for one objective, and next an additional objective function

subject to constraints for the objectives already optimised. The main drawback of

the former consists of the optimisation process usually leading to different solu-

tions depending on the order in which the objectives are selected. Another

approach consists of aggregating the different objectives in a single objective

function with varying numerical weights. Unfortunately, these coefﬁcients usually

240 M. Pérez-Fortes and A. D. Bojarski

lack of physical meaning, and entail an arbitrary assignment of values, see

Sect. 3.1. Thus, there is no unique optimal solution for multiobjective problems,

but rather a set of feasible solutions which may be suitable. The preferred approach

consists of providing a set of Pareto optimal solutions: a Pareto solution is one for

which any improvement in one objective can only take place if at least another

objective worsens. Pareto optimal solutions are also termed dominating solutions,

while the remaining possible optimisation solutions are dominated. This latter

approach implies that the decision maker is interested in all possible trade-off

solutions resulting from no previous prioritisation of the objective functions.

Particularly in the case of objective functions related to the environment, economic

metrics are always prioritised in companies and constraints on the environmental

interventions (emissions, concentrations and others) are given by stringent envi-

ronmental policies. However, a view of process operation that sees environment as

an objective and not just as a constraint on operations can lead to the discovery of

operating policies or plant designs that improve both environmental and economic

performance [43].

3.1 Techniques Applicable for MOO

The techniques for generating a set of Pareto optimal solutions should have some

desirable properties. Namely, they should be able to ﬁnd all available Pareto

points, generate them evenly along the possible solutions in the feasible region

(understood as the collection of points that satisfy all problem constraints), and

they should not generate and explore dominated solutions [44]. However, all the

available techniques present deﬁciencies in some of the former aspects. For

example, the weighted sum must be carefully applied since it does not generate all

available Pareto points, and the Pareto frontier does not represent an evenly set of

solutions of the feasible region [45]. Finally, normal boundary intersection (NBI)

[46] and normal constraint method (NC) [44] generate points that are not in the

Pareto frontier, but NBI is more prone to generate dominated solutions. In general,

all previous procedures require of a ﬁltering step to distinguish and classify

dominated from non-dominated solutions. In general the preferred approach for

generating Pareto fronts (PF) is the use of a normalised constraint method.

3.2 Problem Analysis from Pareto Efﬁcient Solutions and MCDA

Once a certain number of alternatives are generated the selection of the ‘best com-

promise’ alternative from the set of non-dominated alternatives requires input about

the values and preferences of the people responsible for making the decision. Thus,

design teams working on a problem with multiple objectives are faced with the need

to apply multi-attribute decision analysis (MADA) or MCDA techniques. These are

brieﬂy outlined in this section based on a review given by Seppala et al. [1].

Global Clean Gas Process Synthesis and Optimisation 241

• Elementary methods: these methods do not require any inter criteria weighting,

and some cases nor a relative ranking.

– Maxi-min: it assigns total importance to the attribute with respect to which

alternative performs worst, its equivalent to assess the strength of an alter-

native by its weakest link.

– Maxi-max: it assigns total importance to the attribute with respect to which

alternative performs best; assessing the strength of an alternative by its

strongest link.

– Conjunctive and disjunctive methods are screening methods that select

different alternatives given that attributes are exceeding threshold values for

all alternatives (conjunctive) or for some (disjunctive). In general they allow

for selection of satisfactory alternatives instead of best alternatives.

– Lexicographic, it requires a ranking of attributes, and selects the best alternative

by choosing the one that has the best value for the ﬁrst ranked attribute.

• Multi-attribute utility theory (MAUT) methods, these methods require that the

stakeholders’ articulate preference according to strict preference or in differ-

ence relations, this approach provides a clear axiomatic foundation for rational

decision making.

• Multi-attribute value theory (MAVT) is considered a special case of MAUT,

where there is no uncertainty in the consequences of an alternative; while in

MAUT, it is explicitly incorporated. Both approaches can use a simple

formulation (additive) to asses the value (or utility) of a given alternative

(see Eqs. 10 and 11).

Vða

Þ¼

all criteria

ðx

ða

ÞÞ ð10Þ

Vða

Þ¼

all criteria

ðx

ða

ÞÞ



; ð11Þ

where V(a

) represents the value of alternative a

, w

are attribute/criteria weights

and v

() are single attribute functions. If u

() is used instead of the calculation of

the utility of the option (U(a

)) can be assessed. Typical examples of v

() are

found in Eqs. 12 and 13.

ðx

ða

ÞÞ ¼

ða



ð12Þ

ðx

ða

ÞÞ ¼

ða

Þx



 x

ð13Þ

Equation 12, scales scores according to distances the origin and the best/highest

option (x



), while Eq. 13 scales scores relative to the distances between lowest (x

)

and highest scores (x



242 M. Pérez-Fortes and A. D. Bojarski

• Outranking methods: to use this methodologies the decision maker is forced to

express his preferences when comparing one alternative to other. If such binary

relations hold, then by performing pairwise comparisons between each pair of

alternatives under consideration for each criterion the decision of which

alternative is best can be achieved. Several frameworks are available, such as

ELECTRE, PROMETHEE, MELCHIOR and ORESTE. All these methods

have calculation methods reﬂecting the idea that beyond a certain level, bad

performance on one criterion cannot be compensated for by good performance

on another criterion; a non-compensatory approach to decision-making is

adopted, but they lack strong theoretical foundations [1].

• Other methods

– The analytical hierarchy process (AHP) proposed by Saaty [47] calculates

criteria scores (weights) through pairwise comparison using a pre-speciﬁed

1–9 point scale that quantiﬁes verbal expressions of strength of importance

between attributes or preference between alternatives. Having evaluated all

comparisons, weights are calculated via a so-called principal eigenvalue

method; consistency of preferences can also be assessed via an index also

provided by the method.

– In the technique for order by similarity to ideal solution (TOPSIS), proposed

by Hwang and Yoon [48], the best solution is selected according to the

alternative that has the shortest distance (euclidean) from the ‘utopian’ best

possible alternative, formed by the best possible scores for each attribute.

The same metric but measured ﬁxing the worst (nadir) possible alternative

also provides an alternatives ordering.

The selection of the MCDA technique used depends on each case, but the

central question regarding sustainability is whether a compensatory or non-com-

pensatory approach should be used. As Seppala et al. [1] exempliﬁes can clean air

compensate for dirty water? in this sense the authors explicitly consider conditions

under which non-compensatory methods have to be used.

• Single, all-important indicator: if there is one criterion whose importance is

deemed to be overriding; then by that deﬁnition any compensation is forbidden,

consequently lexicographic methods should be used. This is related to the

question of strong-sustainability (non compensation) and weak-sustainability

(compensation is allowed).

• Criteria of ranked importance combined with performance uncertainty: quan-

tiﬁed uncertainty can aid decision makers in setting threshold values of dif-

ference and conﬁdence that are required to distinguish between alternatives

(eliminating possible ties). After ties are eliminated non-compensatory methods

requiring rank order should be used.

• Performance thresholds, in this case the assessment of such thresholds can help

in identifying situations where compensation does not hold.

In the case of selection of the appropriate multi-criteria decision framework, the

most important question that requires to be answered is: are the criteria able to

Global Clean Gas Process Synthesis and Optimisation 243

compensate each other? This question separates broadly the methods to be used,

in the case that compensation is allowed then several techniques are available

which depend on the amount of effort that the decision maker is willing to spend in

devising his/her preferences.

4 Application of MCDA and MOO to Electricity Production

The case study selected aims at showing the capabilities of superstructure scenario

based optimisation using different criteria considers the use of alternate fuels such as

petroleum coke and residual biomass, as well as the evaluation of the use of natural

gas as fuel for the combined cycle (as in a NGCC power plant). Several different

metrics can be calculated from the results of an LCA as discussed in chapter ‘‘

Modelling Superstructure for Conceptual Design of Syngas Generation and

Treatment’’ and different economic and engineering-based metrics are comple-

mentary to give a global techno-economic and environmental point of view.

The ﬂexibility of IGCC power plants has been already commented in chapter

‘‘ Main Puriﬁcation Operations’’, and evaluated from different points of view

simultaneously with several conﬁgurations of carbon capture and storage (CCS)

installations in the literature. In this sense we can mention its modelling including

removal and considering the improvement of the efﬁciency [49], or reduction

of operating costs [50, 51], and different types of coal as feedstock have also been

evaluated [52]. NGCC power plants have been evaluated as well for a large

amount of scenarios [53, 54]. Life cycle assessment (LCA) has been already used

as an environmental impact quantiﬁcation tool for IGCC power plants [17, 55, 56].

The considered scenarios here to show the superstructure capabilities are the

base case (SC1), the IGCC conﬁguration with feedstock composition based on

45% coal, 45% petcoke and 10% orujillo (SC2), and the NGCC conﬁguration with

two layouts, differing in costs (SC3) (see Table 3). The analysis of different energy

conversion performances is done from the point of view of operating cost, energy

conversion efﬁciency and environmental impact.

• Process efﬁciency is a key parameter to evaluate the ﬁnal product to the market

and the energy integration of the plant, as already discussed. The energy

conversion efﬁciency is calculated in this case study using Eq. 7 in chapter

‘‘ Modelling Superstructure for Conceptual Design of Syngas Generation

and Treatment’’ .

• The economic metric considered is the COE of the plant, as described in Sect. 2.1.

Moreover, as a means of exemplifying the use of the ISO14040 guidelines,

the study is organised following the ISO standard 4-steps as follows, for the

environmental impact point of view:

• Goal deﬁnition and scope. The study is focused on the environmental contribution

change attained by the different raw material composition feeds and associated

244 M. Pérez-Fortes and A. D. Bojarski

topological ﬂowsheet changes. The simulated co-gasiﬁcation plant and natural gas

power production considers extraction and processing of raw material, up to the

power production to the grid, which constitutes a ‘cradle to gate’approach.

Regarding system boundary, it is worth mentioning that emissions from waste

water treatment plants and inert solid disposal (ashes) are considered. Moreover,

sulphur obtained from Claus plant is considered to be an environmental credit, by

reducing its production from virgin materials. All scenarios are compared based on

a functional unit of 1 MJ. Regarding total power plant production a similar time

horizon was set and the same production hypothesis as in the economic aspects

were done.

• The LCI is gathered for the different simulated ﬂowsheets, considering the

different superstructure topologies. Simulation results are used to estimate ﬂue

gas and plant wide emissions. The use of simulation software makes mass

balances and energy balances to be met without requiring further data checks;

making this approach a fairly robust one. This step constitutes a conservative

approach (i.e., it overestimates emissions), given that the industry complies

with all legal emission requirements, which in our analysis are disregarded. For

all other echelons studied, which are related to production/extraction of raw

materials and transport, their corresponding LCIs are retrieved from the Eco-

invent database. In the current case, production of: coal, petcoke, natural gas

and other commodities (sulphuric acid, sodium hydroxide and others) are

required given their consumption for plant operation, as shown in Table 2.

Ecoinvent LCI information was gathered considering that the plant is located in

Spain (ES), consequently geographical data for that region was considered

when available, while the European average (RER) was selected otherwise.

Table 2 results show that lower fuel consumption is achieved for the case of

IGCC co-gasifying biomass. Solid wastes are considered to be inert and sent to

Table 2 Summary of different scenario LCI results, expressed in (kg/MJ)

SC1 SC2 SC3

Inputs

HardCoal\ES 0.0550 0.0499 0

PetCoke\RER 0.0550 0.0499 0

Limestone milled\RER 0.00382 0.00359 0

Natural gas\RER 0.00030 0.00030 0.05129

O decarbonised\RER 0.551 0.556 0.739

Consumables (Inorganics\RER) 0.000601 0.000606 0.000072

Ouputs

0.2233 0.2119 0.1373

3.12E-05 2.55E-05 7.08E-06

9.99E-05 8.68E-05 1.77E-04

Particles 4.20E-07 4.23E-07 6.62E-07

Solid inert waste 0.02960 0.02960 0.00015

Water to waste treatment 0.55978 0.55978 0.66555

Sulphur produced 0.00323 0.00321 0

Global Clean Gas Process Synthesis and Optimisation 245

a landﬁll, while waste water efﬂuent is sent to a waste water treatment plant.

Consumables are considered to be inorganic chemicals and a proxy LCI is used

considering the average production for ﬁrst 20 most important inorganic

compounds. Regarding sulphur by-product, it is common practise to allocate

emission and consequently impacts based on some criteria. However, in this

case it has been decided to use a LCI of sulphur production and consider its

production from virgin materials, consequently lowering the overall impact of

IGCC operation.

• Afterwards, the LCIA, based on the previously gathered LCI (see Table 2), is

performed. Several environmental impact indicators are available, regarding

different environmental issues, for the case of electricity generation relevant

environmental issues involve the estimation of impacts related to climate

change using global warming potentials (GWPs), regional acidiﬁcation using

acidiﬁcation potentials (APs), and virgin resources consumption using abiotic

depletion potentials (ADPs). These metrics can be calculated using different

ready to use LCIA methodologies such as the CML 2 [32] or the Impact 2002+

[31]. Other commonly used metrics are the ecological footprint (EF) [57] and

the cumulative energy or exergy demand (CED or CExD) [58]. These metrics

have an aggregating effect and serve in many cases of proxy of more complex

environmental metrics. In the proposed methodology, the former metrics are

calculated using Simapro [59], which allows gathering of LCI information from

the Ecoinvent database [60] and its subsequent organisation for LCIA calcu-

lation. Table 3 shows the LCIA results for CED, CExD, EF and CO

-eq

emission as well as the Impact 2002+ overall environmental impact values.

• Finally, the LCA results interpretation is done. It is found that operation using

NG instead of solid fuels reduces the CED and CExD. In the ﬁrst case the

operation using IGCC requires approximately 60% more than the total amount

of exergy required in the NGCC. In the case of CExD, nearly three times more

energy is required for IGCC than for NGCC; which clearly shows the low

quality of energy produced from IGCC compared to NGCC. The differences

found in the case of EF are not as great as in the CExD case and the same

happens for the CO

-eq. Analyzing the Impact 2002+ results, it is found that the

lowest impacts refer to the case of operation using NGCC followed by IGCC

Table 3 Key performance

indicator results for the

considered scenarios

Metric SC1 SC2 SC3

CED (MJ

) 5.0 4.5 2.8

CExD (MJ

) 9.2 8.3 3.1

EF (m

a) 0.70 0.69 0.41

IPCC GWP (kgCO

2eq

) 0.27 0.27 0.16

IMPACT 2002+ (lPts) 69.8 65.7 40.3

COE (€/kWh) 0.0619 0.0593 0.0537

TCR (€/kW) 3,119 2,987 1,026

TOC (€/kW

year

) 273 261 108

Eff

global

(%) 42.1 41.6 53.12

246 M. Pérez-Fortes and A. D. Bojarski

operation using olive pomace together with coal and coke as feedstock.

Figure 3 shows the results of Impact 2002+ assessment methodology distrib-

uted along the different mid and endpoint categories of the LCIA methodology

and along the different supply chain echelons. In the case of mid-point cate-

gories the largest impacts are associated to: non-renewable energy and GWP,

which are mimicked by the endpoint categories: resources and climate change.

Small differences are found for the endpoint categories human health and

ecosystem quality, which cannot be traced directly to a single mid-point impact

as in the case of resources and climate change. Regarding endpoint impacts, in

the three electricity production cases, human health impact and ecosystem

quality accounted for less than 13% of the total impact, while the remaining

was almost evenly partitioned between resources and climate change impacts.

In the case of IGCC operation the biggest overall impact is related to the plant

operation itself. In second and third place, raw material production is found.

Despite the fact that both IGCC scenarios use the same percentage of coal and

coke, their associated impact is different. Coal production is found more

environmental friendly than coke production. In all the former environmental

metrics, the life cycle stages associated to most impact are: raw materials

production (coal, coke and NG, respectively for each scenario), for the case for

resources, while climate change is because of the IGCC/NGCC echelon.

Table 3 presents the summary of all considered performance indicators.

Economic values are referred to €

2007

. The environmental metrics are related to

Fig. 3 Results for environmental impact estimation using Impact 2002+ assessment method-

ology. The impact is shown distributed along mid- and end-point indicators as well as along

different SC echelons

Global Clean Gas Process Synthesis and Optimisation 247

Puigjaner L. (ed.) Syngas from Waste - Emerging Technologies

Подождите немного. Документ загружается.