| dc.description.abstract | Background and Motivation
Unlike what is common in Europe and the rest of the world, Norway has traditionally met most of its stationary energy demand (including heating) with electricity, because of abundant access to hydropower. However, after the deregulation of the Norwegian electricity market in the 1990s, the increase in the electricity generation capacity has been less than the load demand increase. This is due to the relatively low electricity prices during the period, together with the fact that Norway’s energy companies no longer have any obligations to meet the load growth. The country’s generation capacity is currently not sufficient to meet demand, and accordingly, Norway is now a net importer of electricity, even in normal hydrological years. The situation has led to an increased focus on alternative energy solutions.
It has been common that different energy infrastructures – such as electricity, district heating and natural gas networks – have been planned and commissioned by independent companies. However, such an organization of the planning means that synergistic effects of a combined energy system to a large extent are neglected. During the last decades, several traditional electricity companies have started to offer alternative energy carriers to their customers. This has led to a need for a more comprehensive and sophisticated energy-planning process, where the various energy infrastructures are planned in a coordinated way. The use of multi-criteria decision analysis (MCDA) appears to be suited for coordinated planning of energy systems with multiple energy carriers. MCDA is a generic term for different methods that help people make decisions according to their preferences in situations characterized by multiple conflicting criteria.
The thesis focuses on two important stages of a multi-criteria planning task:
- The initial structuring and modelling phase
- The decision-making phase
The Initial Structuring and Modelling Phase
It is important to spend sufficient time and resources on the problem definition and structuring, so that all disagreements among the decision-maker(s) (DM(s)) and the analyst regarding the nature of the problem and the desired goals are eliminated. After the problem has been properly identified, the next step of a multi-criteria energy-planning process is the building of an energy system model (impact model). The model is used to calculate the operational attributes necessary for the multi-criteria analysis; in other words, to determine the various alternatives’ performance values for some or all of the criteria being considered. It is important that the model accounts for both the physical characteristics of the energy system components and the complex relationships between the system parameters. However, it is not propitious to choose/build an energy system model with a greater level of detail than needed to achieve the aims of the planning project.
In my PhD research, I have chosen to use the eTransport model as the energy system model. This model is especially designed for planning of local and regional energy systems, where different energy carriers and technologies are considered simultaneously. However, eTransport can currently provide information only about costs and emissions directly connected to the energy system’s operation. Details about the investment plans’ performance on the remaining criteria must be found from other information sources. Guidelines should be identified regarding the extent to which different aspects should be accounted for, and on the ways these impacts can be assessed for each investment plan under consideration. However, it is important to realize that there is not one solution for how to do this that is valid for all kind of local energy-planning problems. It is therefore necessary for the DM(s) and the analyst to discuss these issues before entering the decision-making phase.
The Decision-Making Phase
Two case studies have been undertaken to examine to what extent the use of MCDA is suitable for local energy-planning purposes. In the two case studies, two of the most well-known MCDA methods, the Multi-Attribute Utility Theory (MAUT) and the Analytical Hierarchy Process (AHP), have been tested. Other MCDA methods, such as GP or the outranking methods, could also have been applied. However, I chose to focus on value measurement methods as AHP and MAUT, and have not tested other methods. Accordingly, my research cannot determine if value measurement methods are better suited for energy-planning purposes than GP or outranking methods are.
Although all MCDA methods are constructed to help DMs explore their ‘true values’ – which theoretically should be the same regardless of the method used to elicit them – our experiments showed that different MCDA methods do not necessarily provide the same results. Some of the differences are caused by the two methods’ different ways of asking questions, as well as the DMs’ inability to express clearly their value judgements by using one or both the methods. In particular, the MAUT preference-elicitation procedure was difficult to understand and accept for DMs without previous experience with the utility concept. An additional explanation of the differences is that the external uncertainties included in the problem formulation are better accounted for in MAUT than in AHP. There are also a number of essential weaknesses in the theoretical foundation of the AHP method that may have influenced the results using that method. However, the AHP method seems to be preferred by DMs, because the method is straightforward and easier to use and understand than the relatively complex MAUT method.
It was found that the post-interview process is essential for a good decision outcome. For example, the results from the preference aggregation may indicate that according to the DM’s preferences, a modification of one of the alternatives might be propitious. In such cases, it is important to realize that MCDA is an iterative process. The post-interview process also includes presentation and discussion of results with the DMs. Our experiments showed that the DMs might discover inconsistencies in the results; that the results do not reflect the DM’s actual preferences for some reason; or that the results simply do not feel right. In these cases, it is again essential to return to an earlier phase of the MCDA process and conduct a new analysis where these problems or discrepancies are taken into account.
The results from an MAUT analysis are usually presented to the DMs in the form of expected total utilities given on a scale from zero to one. Expected utilities are convenient for ranking and evaluation of alternatives. However, they do not have any direct physical meaning, which quite obviously is a disadvantage from an application point of view. In order to improve the understanding of the differences between the alternatives, the Equivalent Attribute Technique (EAT) can be applied. EAT was tested in the first of the two case studies. In this case study, the cost criterion was considered important by the DMs, and the utility differences were therefore converted to equivalent cost differences. In the second case study, the preference elicitation interviews showed, quite surprisingly, that cost was not considered among the most important criteria by the DMs, and none of the other attributes were suitable to be used as the equivalent attribute. Therefore, in this case study, the use of EAT could not help the DMs interpreting the differences between the alternatives.
Summarizing
For MCDA to be really useful for actual local energy planning, it is necessary to find/design an MCDA method which: (1) is easy to use and has a transparent logic; (2) presents results in a way easily understandable for the DM; (3) is able to elicit and aggregate the DMs' real preferences; and (4) can handle external uncertainties in a consistent way. | nb_NO |
