Multicriteria Decision Analysis for Energy infrastructure planning: the study-case of Natural Gas in Italy

13 
Introduction: Natural Gas in Italy 
In this chapter a brief overview of the Italian natural gas system is given. In 2011 in Italy 184,2 
Mtep of primary energy were consumed, of which the 35% came from natural gas.  
 
Figure 1: Primary energy matrix. (Ministero dello Sviluppo Economico, 2012) 
Even though in the last years natural gas consumption in Italy decreased, due to the economic 
crisis, it still remains the primary source for electricity generation, accounting for more than a half 
of national electricity production, and its consumption has consistently increased over the last two 
decades.  
0.1. Internal production 
The indigenous production of natural gas in Italy is led by Eni S.p.A, which is the principal owner of 
the deposit (more than 90% of the total fields). Due to the scarcity of resources the internal 
production represents actually just a 12% (2012) of the total supply. In the decade 2000-2009 a 
decrease of 53% in production was observed, however this tendency is currently changing and the 
last two years showed an increase in extraction activities. 
solid 
9% 
natural gas 
35% 
oil 
37% 
renewables 
13% 
primary 
electricity 
6% 
Primary energy matrix, 2011
14 
 
Figure 2: Natural gas production 1991-2011. (Ministero per lo Sviluppo Economico, 2012) 
The wells  are small, very fragmented and often located at great depths or offshore, and this makes 
it more difficult both their search and their exploitation. The most important fields are located in 
Sicily and offshore of the island, others are sited in Basilicata and Emilia Romagna (the Orsini field, 
near Ravenna). Proven reserves account for 60.000 Mm3 and the estimated recoverable ones 
slightly more than 150.000 Mm3. 
0.2. Natural gas market 
Italy imports almost 90% of the natural gas it consumes, and most imports come from non-EU 
countries. 
 
Figure 3: Natural gas per origin. (Autorità per l'Energia Elettrica e il Gas, 2011) 
29% 
12% 
29% 
2% 
6% 
9% 
2% 
0% 
0% 
11% 
Natural Gas per origin, 2009 
Algeria
Libya
Russia
Quatar
Norway
Netherlands
Others
Egypth
Nigeria
Internal prod
15 
From 2003, following the liberalization in the natural gas market, the final user can choose its 
supplier. The freedom to choose the supplier (obviously linked to the fact that there are more 
subjects to the sale operating in the area of reference) is guaranteed by the formal separation 
between different actors in the gas market: 
 Carrier: owns and manages the infrastructure for the transportation of high pressure gas 
from the extraction place to the entry point into the low pressure network. 
 Wholesaler: is the owner of the gas inside the pipelines operated by the carrier. 
 Distributor: is the owner or more often the handler of low and medium pressure 
distribution networks (urban networks). 
 Sales companies: own the gas networks in low-pressure and sell this gas to end customers. 
In summary, therefore, there is a clear separation between the transport operators of the 
infrastructure and those responsible for the gas sale. 
The gas contracts are designed to safeguard the parties through a system of guarantees and 
constraints: 
 clause Take or Pay (ToP) that engages the importer to the supplier to ensure a revenue 
stream, regardless of the volumes taken, for an amount generally ranges between 70% and 
95% of the total value of the contract; 
 duration of between 20 and 25 years; 
 base price of bargaining, depending on the characteristics of the importer; 
 indexation mechanisms and renegotiation linked to the price of a basket of oil products. 
0.3. Infrastructures 
The main import infrastructures will be discussed in Chapter 2.1 – Problem structuring, as they 
constitute the framework from which the decision problem is constructed. Internal gas network is 
shown in the figure below.
16 
 
Figure 4: Italian natural gas network. (Snam) 
Snam Rete Gas is the main operator for transport and dispatching of natural gas. It owns great part 
of the national gas network. The principal direction of the gas flow is from South to North, because 
more than 30% of the supply comes from the South, while consumption is basically more intensive 
in the Center and North. There is just a 20% of the transport gas flow coming from North to South. 
0.4. Natural gas companies 
Companies can have production concessions, can own and manage the national or regional 
network, distribute the gas or have concessions for storage. 
 Production companies: the main ones are Edison and Eni. 
 Transmission companies: Snam Rete Gas (owning the majority of both national and 
regional network), Società Gasdotti Italia, Edison. 
 Distribution companies: the ownership of natural gas distribution facilities remains 
fragmented, with about 250 companies operating in the sector (this compares, however,
17 
with over 430 in 2005). The principal operator is still ENI, which controls 22.9% of the 
market, in terms of distributed volumes. 
0.5. Future projects 
The gas market in Europe has changed considerably over the past few years, due to the demand 
decrease following the economic crisis, the emergence of shale gas and rise of new energy-
consuming countries, particularly China. Even more, there is uncertainty about Russian supplies, 
which now have the EU countries as the only buyers, but could move to Asia, thus decreasing the 
quantities exported to Europe. In this context, Italy could be a crossroads for natural gas that 
passes through the Mediterranean. 
In spite of the uncertainty currently in Europe there are several pipelines projects under study, 
some of them involving Italy as well. 
 
Figure 5: New gas line projects. (Rotondi, 2012) 
The destiny of most of them is still undecided: the Nabucco project (South Stream) will probably be 
abandoned, while it is not clear if ITGI and TAP will be realized or not. Also not all the regasification 
plants present in the picture will likely be constructed. 
The National Energy Strategy (Strategia Energetica Nazionale), proposed in 2012, establishes seven 
objectives, to be reached in 2020: two of them regard the natural gas system directly. In particular, 
one of the priorities is to promote a more competitive natural gas market, becoming the main 
South-European hub for gas. This means that imports should be enhanced, to allow natural gas 
export toward Northern Europe through a flux inversion in the Transitgas pipeline: at the moment 
the pipeline is not used at its full capacity. The objective is to increase its utilization factor by
18 
exporting gas to the Central and Northern Europe markets. This would also require to increase 
imports and to empower the national network in the South-North direction. 
A second objective of the National Energy Strategy is to promote a sustainable development of 
national hydrocarbons production, in order to reduce the dependency from foreign countries. The 
production objectives at 2020 are shown in the figure below.  
 
Figure 6: enhancing national hydrocarbons production. (Ministero dello Sviluppo Economico, 2012)
19 
1. Multicriteria methods 
Management science’s mandate is to improve and support decisions of all kinds by using rational, 
systematic, and science-based techniques. Several methodologies, which differ for number of 
decision-makers (from one to N), number of objectives (from one to N) or criteria, and 
completeness of information (a perfect information leads to deterministic approach, while an 
imperfect information implies stochastic approach) have been developed and employed to handle 
complex decision making problems. According to this, one possible classification of these 
techniques can be presented as follows: 
 Linear programming – one decision-maker, one objective, perfect information; this is the 
case of the cost optimization performed by Pérez and from which this work starts; 
 Stochastic mathematical programming – one decision-maker, one objective, imperfect 
information; 
 Game theory – several decision-makers, one to N objectives, perfect information; the aim is 
to analyze conflicting situations where each of the stakeholders is carrier of a point of view, 
potentially contrasting with the others, and to find solutions to conflicts through 
mathematical models; 
 Stochastic game theory - several decision-makers, one to N objectives, imperfect 
information; 
 Multiple criteria decision analysis (or MCDA) – one to N decision-makers, N objectives, 
perfect information. 
We focus on this last category. MCDA methodology can be again classified in: 
1. Multiobjective – is a continuous analysis, where the solution is not given but has to be 
found among infinite possibilities. It generally includes several decision variables 
(objectives), to which we give weights, and there might be constraints to respect. The aim 
is to find the optimum point amongst infinite possible solutions. 
 Multiattribute – is a discrete analysis, where solution is chosen among a set of possible 
alternatives. The attributes (called criteria) and alternatives (called actions), are linked 
together to find the best outcome, which will be a compromise between different 
preferences. We use this second approach. 
It is important to notice that all complex problems are multicriteria in nature, since every decision 
we take requires a balancing of multiple factors, and a single-criteria problem is just a multicriteria 
one where one objective (for example cost) has been identified as prevailing and 100% weight is 
attributed to it. 
1.1. How does the method work? 
MCDA process starts with the definition of the problem: in fact in reality it is really unlikely that the 
analyst finds itself already provided with a well-defined set of alternatives and criteria. More likely, 
the MCDA process is embedded in a larger process of problem structuring and resolution. The 
messy figure below shows the basic stages of multicriteria analysis, underlying the fact that the 
practice is to move back and forth frequently to modify or revise hypothesis and choices made 
earlier.
20 
 
Figure 7: the MCDA process. (Belton & Stewart, 2002) 
Once the problem has been defined, in short, the model construction consists in selecting a range 
of alternatives and a set of criteria, and evaluating the performances of the alternatives over the 
whole set of criteria. These data is collected into the so called Evaluation Matrix. Then to each 
criteria is attributed a weight, proportional to its relative importance. Weights have to be 
normalized to 100%. Finally, to operate on the evaluations matrix, which contains non-
homogeneous data, some of which are qualitative and other quantitative, expressed with different 
units of measure we need some mathematical functions which make the comparison possible. This 
is done by the Preference Functions. There are several techniques to assess Preference Functions 
and calculate the final ranking of alternatives, the most common of which are: 
 MAVT (Multiattribute Value Theory) – the intention is to construct a way to associate a real 
number to each alternative, in order to produce a preference order of alternatives. In other 
words, we want to associate a value V(a) to each alternative a, in such a way that a is 
judged to be preferred to b taking all criteria into account. This is true if and only if 
 ( )  ( ) 
 MAUT (Multiattribute Utility Theory) – can be seen as an extension of value measurement, 
including also the use of probabilities and expectations to deal with uncertainties.  
 Outranking – once partial preference functions z
i
(a) have been defined for all criteria, the 
method, consisting in a pairwise comparison of alternatives, is applied. The method is 
based on the dominance concept: we say that alternative a outranks b if there is sufficient 
evidence to justify the conclusion that a is at least as good as b, taking all criteria into
21 
account. With formulas, a dominates b if z
i
(a) ≥  z
i
(b) for all criteria, and there is at least on 
criterion for which z
i
(a) >  z
i
(b) (strict inequality). We can immediately conclude that a 
should be preferred to b. This last methodology is the one applied in this study.  
1.2. Selection of the tool 
To support the developing of the study, proper tools to perform the multicriteria analysis are 
required. Several software were considered and compared: the features of the analyzed tools are 
reported in the following table: 
Software 
Suppor
ted 
MCDA 
Metho
d 
Pairwi
se 
Compa
rison 
Tim
e 
Anal
ysis 
Sensit
ivity 
Analy
sis 
Grou
p 
Evalu
ation 
Risk 
Manag
ement 
Web 
Base
d 
Qualita
tive 
And 
Quanti
tative 
User
-
Frien
dly 
Inter
face 
Licence 
1000Minds 
PAPRIK
A 
Yes No Yes Yes No 
Yes 
(Only 
Web 
Base
d) 
No Yes 
Trial  
Academic 
Business 
Analytica   No Yes Yes No Yes Yes ? Yes 
Trial  
Professional 
Education 
Criterium 
Decisionplus 
AHP, 
SMART 
Yes No Yes No Yes No ? ? 
Trial/Student 
Version 
Business 
Decideit MAUT Yes No Yes Yes Yes Yes Yes Yes 
Trial 
Academic 
Business 
Decision 
Lens 
AHP, A
NP 
Yes ? Yes Yes Yes Yes ? Yes 
Demo 
Business 
Visual 
Promethee 
PROME
THEE, 
GAIA 
Yes No Yes Yes Yes 
Not 
Avail
able 
Yet 
Yes Yes 
Launch Edition 
Business 
Academic 
Demo 
Online 
D-Sight 
PROME
THEE, 
UTILITY 
Yes No Yes Yes Yes Yes Yes Yes 
Trial 
Academic 
Online 
Desktop 
Expert 
Choice 
AHP Yes No Yes Yes Yes Yes ? Yes 
Trial 
Online 
Limited 
Academic 
Version 
Desktop 
Hiview3   No No Yes Yes No No ? Yes 
Trial 
Student 
Standard 
Logical 
Decisions 
AHP, 
MAUT 
Yes No Yes Yes Yes No ? ? Trial 
Table 1: Software comparision. (Belton & Stewart, 2002) 
For several of the listed tools free trials were available, so we had the opportunity to test the 
software before making the choice. The most important features considered are the possibility to