• Nie Znaleziono Wyników

Agent-Based Simulations of Energy Transitions

N/A
N/A
Protected

Academic year: 2021

Share "Agent-Based Simulations of Energy Transitions"

Copied!
10
0
0

Pełen tekst

(1)

Third International Engineering Systems Symposium

CESUN 2012, Delft University of Technology, 18-20 June 2012

Agent-Based Simulations of Energy Transitions

Emile J. L. Chappin1

1Delft University of Technology, Jaffalaan 5, 2628 BX Delft e.j.l.chappin@tudelft.nl

Abstract. The notion of transition receives noticeable attention in political as well

as scientific arenas. In the policy arena in the Netherlands, significant results have not been achieved yet despite all the efforts on ‘managing’ the transition to a sustainable energy sector. Although the scientific literature on transitions contains publications from 25 countries, the US was most important before 2000 and authors from the Netherlands have been dominant since.

Simulations of energy transitions are in its early stages, compared to their potential. Our bibliographical analysis of the transition literature shows that the number of papers mentioning simulations is low, only 19 out of 142, and their young. Most of those papers describe case studies that focus on autonomous, unmanaged transitions; only a few aim their simulations at transition management.

Complex systems theory tells us that energy infrastructures – true socio-technical systems – cannot be designed. Therefore, transition management is a paradox: when transitions are expected to take decades, how could we know what actions to take now in order to shape energy infrastructures in such a way that the preferred transition will occur over decades? And at the end of the day, how could we attribute the result to transition management activities, whether the transition was successful or not?

This paradox is no argument to wait: policy issues regarding energy infrastructures have to be made today. Therefore, we have set out for simulations of energy transition, using agent-based models, as to support energy transition management.

Results from three cases – regarding CO2 reduction from power generation, the electricity-intensity of consumer lighting, and a spot market for LNG trade – have proved that it is possible to gain useful insights in how the myriad of decisions made in energy infrastructures can be influenced in a way that a transition is likely.

Keywords. Energy Infrastructures, Societal transitions, Agent-Based Models,

Socio-technical systems

1 Introduction

1.1. Challenges for energy infrastructures

Energy infrastructures have become the backbone of society. It is widely acknowledged that we have to change our energy infrastructure systems during the 21st century. We have to address issues such as scarcity and the depletion of

(2)

resources, accessibility, affordability, reliability and quality of energy services, and security of energy supply.

The need for change has been addressed, for instance by setting EU and national targets for renewable energy. Despite considerable efforts and budgets of the Dutch government (Ministry of, Nikolic et al), some say that more is required to actually achieve a transition (RMNO, ): How the needed changes can/should be governed is not clear-cut. Safeguarding our infrastructures is not only about technical aspects. Also governance aspects are relevant in order to prevent the improper functioning of markets and ineffective/inefficient realization of long term public values (WRR, ). When decisions are made regarding all the issues concerning our energy infrastructures, how can we be assured now that we do the right thing, in the right way, at the right time?

Fig. 1. Overview of the transition literature. The literature is rapidly growing, but the number of papers on simulation is small.

1.2. Simulation

Change in large systems, such as our infrastructure systems, is the central topic of the scientific literature on transitions (Geels, ) and transition management (Rotmans et al, Loorbach, ). Applying transition management in the real world – if possible at all – is not trivial. Our perspective is that one of the promising approaches is to use simulation models. Only a relatively small effort has been done (see Fig. 1). We hypothesize that – if we are able to capture transitions in simulation models – we may contribute new insights to the body-of-knowledge on transitions and transition management (Chappin, ). At the same time, we may learn about how our energy infrastructures may develop in this century. At the end of the day, we may find opportunities to direct that course of development.

1.3. Structure

The remainder of this paper is structured as follows. In the next section , the bridge from transition literature on the one hand and simulation on the other hand is

(3)

discussed. Afterwards, a discussion on various simulations is presented. The paper ends with conclusions.

2 From transition management to simulation

2.1 Concepts in the transition literature

Energy infrastructures are true socio-technical systems. From a socio-technical system’s perspective, transitions emerge out of the myriad of decisions of actors, their interactions and their behaviour regarding their physical assets. Based on this perspective and an analysis of the many definitions of transitions, we have derived the following definition: “A system transition is substantial change in the state of a

socio-technical system” (Chappin, ). This definition is not related to

Literature regarding unplanned transitions are mainly structured as qualitative transition case-studies. These have led to the recognition of phases in transitions (similar to innovation-diffusion patterns) (Rotmans et al, Wiek et al). Furthermore, three system levels are identified – niche, regime and landscape. A transition is depicted as a regime-shift (Geels, , Holtz, ). We have analysed literature on transitions and transition management, which is growing rapidly.

Fig. 2. Overview of the transition literature. The literature is largely Dutch.

The literature on transitions contains publications from 25 countries, the combined publications of the Netherlands, the UK, and the US count for a share over 70% (see Fig. 2). The US was most important before 2000, but authors from the Netherlands have been dominant since. The publications in the last decade broadened the scope of the transition literature dramatically: the number of authors enormously increased and an international, multidisciplinary field with many perspectives and conceptual models of transitions developed, exploring those concepts by applying them to cases.

Theory regarding unplanned transitions can be distinguished from theory regarding transition management. In our view, public policy regarding energy infrastructures

(4)

relates to transition management: transition management promises, and should allow us to substantially improve our energy infrastructures. However, in the literature on transition management we found that ‘success’ and ‘performance’ of transition management is ill-defined (Chappin &). There are a myriad of prescriptive, and partially conflicting transition management elements. There is a strong focus on sustainability and the role of government is highly debated. Combining these findings with a system’s perspective, transition management is considered to be “the art of

shaping the evolution of socio-technical systems” (Chappin, ).

2.2. Policy interventions as an ingredient of transition management

A strategy to focus the discourse in transition management is to start with policy

interventions. That is a valid starting point from the argument that in-depth

understanding of transitions – and in the long-term effects of specific interventions – is a necessary ingredient of transition management. This means that exploring the merits of transition management translates to tracing the long-term effects policy interventions (Chappin &). If interventions are successful in its intentions to shape the evolution of the socio-technical system, it can be considered an underpinned ingredient for transition management.

2.2. Challenges for tracing policy interventions

The challenge is to measure and assess whether an intervention is likely to meet the requirement of being successful in shaping the system in transition. This challenge has at least the following various components:

Measurement: Appropriate indicators need to be defined

Assessment: A proper assessment needs to be possible

Deep uncertainty: The system is surrounded by deep uncertainty regarding

future developments. This hinders a clear-cut assessment.

System structure: The assessment needs to deal with change in the system’s

structure.

Socio-technical aspects: The response to policy intervention essentially means

changing behaviour of companies and people (i.e. social aspects are relevant). Most behaviour of companies and people is about technology, physical apparatus and the development, selection and use of technology determines the performance of the system (i.e. technical aspects are relevant). Therefore, both technical and social aspects are relevant and should be considered in the assessment.

2.4. Agent-based modelling as paradigm for exploring structural change

Given these challenges, an approach for assessment is to use quantitative simulation. An earlier analysis of various options showed that agent-based modelling (ABM) is the only modelling paradigm that allows for an emergent and changing system structure (Chappin, ). In an agent-based model (ABM), actors are represented as computer-coded agents (Epstein &, Jennings, ), having properties constituting an individual identity (Weiss, , Bussmann et al) or management style.

(5)

There is a large range of types of agent-based models. In order to be able to develop assessments of policy interventions, we have made choices for representation of actors, technology, both in terms of ontological aspects as well as software and hardware.

In our view, agents are equipped with coded decision rules, some that determine their strategic decisions and some that determine their operational decisions. The term

agent is, therefore, reserved for pro-active and autonomous components in the system.

Markets can also be represented as agents if they are institutionalized with their own rules according to which, for instance, prices are determined. An example is a power exchange, where the rules for market clearing are embedded in the juridical and institutional context. Physical components are considered objects. They are represented as computer-coded physical nodes/elements with properties regarding technical capabilities and flexibilities. Both social and physical components interact. Any intervention may affect the agents in the decisions they make. The structure and dynamics of the system emerge from the physical causalities governing the system and from the decision making rules of agents, which make them respond to policy interventions.

Table 1. Three agent-based models, the interventions modeled and the main results obtained.

Case Intervention Type of result

Power Generation CO2 emissions- trading, carbon taxation, secondary policies

Investment risk, income transfer LNG market None, through the market Market structure, contract formation, Consumer lighting Ban on bulbs, incandescent

bulbs taxation, LED subsidy Social network, changing preferences, incomplete and incorrect information

3 Agent-based models of energy infrastructures

In the previous section we showed that 1) we need proper assessments of transition management activities in order to explore the merits of transition management 2) policy intervention is a good starting point for studying transition management, 3) agent-based models are a promising method to develop such assessments.

In this section, three examples will be used to show what agent-based models can be developed to start bridging the worlds of transition and simulation. The three cases are very different in nature, system and focus. An overview is given in Table 1.

3.1 Power Generation

Power generation embodies a large part of global CO2 emissions. A significant part of recent discussions regarding sustainability of energy infrastructures in developed countries is decarbonisation of power generation (European Climate, Stern, ). Significant reduction of CO2 emissions requires investment in clean or green power generation technologies, something that has not happened on large scale yet (Hoffmann, ). Will the transition to a CO2-extensive power generation portfolio be successful under the current European regulation scheme?

(6)

Fig. 3. Snapshot of the Agent-Based Power Generation Model.

In the model (see Fig 3 for an indicative snapshot of the running model), the agents represent power generation companies, operating their existing power generators and investing in new power plants, over the course of decades. The agents are subject to either 1) a CO2 emissions-trading scheme as implemented in the EU, 2) a carbon taxation scheme as proposed as an alternative or 3) no policy intervention, a hypothetical business as usual case. Furthermore, the agents are subject to uncertainty with respect to fuel prices and electricity demand.

It was found that the emissions trading scheme implemented in the EU introduces an investment risk related to the price volatility on the CO2 market (Chappin et al). Under a taxation scheme with an average tax level equal to the CO2-market price, emission reductions are accomplished faster and further, with less income transfer from consumers to producers. This conclusion is robust against many of specific the assumptions behind the study. Such an analysis would have been extremely difficult – if not impossible – with the more traditional modelling approaches or applying the theory qualitatively. The ABM showed its value in surprising the modeller, exploring the effects of a number of policy interventions on energy transition.

(7)

3.2 LNG Market

The liquefied natural gas (LNG) market is traditionally governed by long-term high-volume bilateral contracts. Recently, though, some spot trade has been noted (Morikawa, , Aissaoui, ). Can the transition of the LNG market to spot trade emerge? Four drivers for transition in the LNG market were identified that may initiate such a transition: 1) growth of the market, 2) uncommitted capacity, 3) technological innovations, and 4) the LNG spot market reinforcing itself.

These four drivers were tested in an agent-based model. Agents represent companies active in the LNG market. They can have a portfolio of capacity to ship, to liquefy and to regasify natural gas/LNG. In order to build-up their business, the agents engage into contracts on LNG trade. They do so by optimizing their expectations regarding future options based on their experience and current market conditions.

We have let the market for LNG grow, by starting with companies without any installations, driven by a demand for LNG. By letting the market evolve, the potential of transition towards a market with portfolios of contracts. The agents had the opportunity to choose for either long-term contracts or spot contracts and to build installations before contracting or vice-versa. We observed that a spot market for LNG is likely to evolve and found that the identified drivers growth, uncommitted capacity, and innovation are important for the development of significant levels of spot trade. Contrary to expectations, we have not observed that spot-trade reinforces itself. The ABM showed its value in simulating an evolving market with an emergent structure, driven by demand and by expectations of the participants. This allows for defining experiments quite different from other modelling paradigms. At the end of the day, this may lead to increased understanding of the mechanisms that play a role in the real-world markets we have today.

3.3 Consumer lighting

Energy consumption by lighting is significant. Consumption levels of consumer lighting have not declined despite the availability of alternative technologies. Recently, the EU issued a phase-out of traditional incandescent bulbs. That makes sense, as their efficiency did not improve for a century: it was around 12 lumens of light per Watt electricity input in 1912 and it still is in 2012 (Gendre, , Afman et al). What can we expect from government policies such as this phase-out on a transition to low-electricity consumer lighting?

A social network of heterogeneous consumer agents replace failing lamps based on their individual preferences (see Fig 4 for an indicative snapshot of the running model). Various types of lamps with different properties are for sale and these properties change over time. Agents have a memory of their sales and develop perceptions regarding lamps, technologies, and brands. They exchange these perceptions within their social network.

The simulation results confirm that the implemented phase-out of incandescent bulbs in the EU is the most effective way of achieving a lower electricity usage for lighting. However, the long-run effect of a taxation scheme is equal, but it relieves the purchase peak imposed on consumers. This model allows for showing the effects of perceptions, possibly outdated information, various ways of sharing information between consumers on what effect a certain policy may bring. In product markets

(8)

where such social mechanisms play a role, it can be detrimental for the validity of the outcomes from a quantitative approach.

Fig. 4. Snapshot of the Agent-Based Consumer Lighting Model.

The simulation results confirm that the implemented phase-out of incandescent bulbs in the EU is the most effective way of achieving a lower electricity usage for lighting. However, the long-run effect of a taxation scheme is equal, but it relieves the purchase peak imposed on consumers. This model allows for showing the effects of perceptions, possibly outdated information, various ways of sharing information between consumers on what effect a certain policy may bring. In product markets where such social mechanisms play a role, it can be detrimental for the validity of the outcomes from a quantitative approach.

4 Conclusions

A necessary ingredient for transition management is to choose and implement the ‘best’ policy interventions in the system that needs to transition. We argued that we should be able to assess the long term consequences of such policy interventions in evolving energy infrastructure systems in order to evaluate whether this ingredient for transition management is viable. Such an assessment is difficult because transition

(9)

needs to be measured, there is deep uncertainty, the system structure changes and there are relevant social and technical aspects.

As a promising paradigm, we chose to develop and analyse the output of agent-based models (ABMs). They are suitable to simulate energy transitions, because they can capture change in the system structure and dynamics. Insights gained from ABM simulations show advantages and disadvantages of specific policy interventions in energy infrastructures, by showing the variability in the long-term effects on the affected energy systems. ABMs can determine likely effects of interventions without claiming to perfectly predict future states of socio-technical systems. A wide range of simulations and a strong effort on interpretation and reflection allows for a discussion on the merits of such simulations for policy and business.

The results from three models show that it is possible to successfully design and implement that process, where the modelling effort results in useful insights in how the myriad of decisions made in energy infrastructures may be influenced. Specific interventions affect the many distributed decisions taken by relevant actors in a way that is likely to alter the dynamics and the structure of the socio-technical system along a desired trajectory. Proper experiments with such models may support policy makers, and, at the end of the day, hopefully improve the efficiency and effectiveness of the policies in place.

Acknowledgments. This work was supported by the Next Generation Infrastructures

Foundation (http://www.nextgenerationinfrastructures.eu).

References

Afman, M. R., Chappin, E. J. L., Jager, W. & Dijkema, G. P. J. (2010). Agent-based model of transitions in consumer lighting, 3rd World Congress on Social Simulation: Scientific Advances in Understanding Societal Processes and Dynamics, University of Kassel and Center for Environmental Systems Research, Kassel, Germany.

Aissaoui, A. (2006). Market risks in a changing LNG world: Exploring alternative mitigation strategies for MENA projects, Middle East Economic Survey XLIX(44). http://www.mees.com/

Bussmann, S., Jennings, N. R. & Wooldridge, M. (1998). Multiagent Systems for Manufacturing Control, Springer, Berlin.

Chappin, E. J. L. (2011). Simulating Energy Transitions, PhD thesis, Delft University of Technology, Delft, the Netherlands. ISBN: 978-90-79787-30-2. http://chappin.com/ChappinEJL-PhDthesis.pdf

Chappin, E. J. L. & Dijkema, G. P. J. (2010). Transition management in energy: Design and evaluate transitions with a suitable simulation framework, in M. van Geenhuizen, W. J. Nuttall, D. Gibson & E. Oftedal (eds), Energy and Innovation: Structural Change and Policy Implications, International Series on Technology Policy and Innovation, Purdue University Press. ISBN 978-1-55753-578-8.

Chappin, E. J. L., Dijkema, G. P. J. & Vries, L. J. d. (2010). Carbon policies: Do they deliver in the long run?, in P. Sioshansi (ed.), Carbon Constrained: Future of Electricity, Global Energy Policy and Economic Series, Elsevier, pp. 31–56. ISBN: 978-1-85617-655-2.

(10)

Epstein, J. M. & Axtell, R. (1996). Growing artificial societies: social science from the bottom up, Complex adaptive systems, Brookings Institution Press; MIT Press, Washington, D.C.

European Climate Foundation (2010). Roadmap 2050 – A practical guide to a prosperous, low carbon Europe, European Climate Foundation/Roadmap 2050. Accessed 3 November 2010. http://www.roadmap2050.eu

Geels, F. W. (2007). Transformations of large technical systems – a multilevel analysis of the Dutch highway system (1950–2000), Science, Technology, & Human Values 32(2): 123– 149.

Gendre, M. F. (2003). Two centuries of electric light source innovations. http://www.einlightred.tue.nl/lightsources/history/light_history.pdf

Hoffmann, V. H. (2007). EU ETS and investment decisions : The case of the German electricity industry, European Management Journal 25(6): 464–474. Business, Climate Change and Emissions Trading.

Holtz, G. (2010). Modelling system innovations in coupled human-technology environment systems, PhD thesis, University of Osnabrück, Osnabrück, Germany.

Jennings, N. R. (2000). On agent-based software engineering, Artificial Intelligence 117(2): 277–296.

Loorbach, D. (2007). Transition management – new mode of governance, International Books, Utrecht, the Netherlands.

Ministry of VROM (2001). Een wereld en een wil – werken aan duurzaamheid – Nationaal Milieubeleidsplan 4, Ministerie van Volkshuisvesting, Ruimtelijke Ordening en Milieubeheer, Den Haag.

Morikawa, T. (2008). Natural Gas and LNG Supply/Demand Trends in Asia Pacific and Atlantic Markets, The Institute of Electrical Engineers of Japan.

Nikolic, I., Dijkema, G. P. J. & van Dam, K. H. (2009). Understanding and shaping the evolution of sustainable large-scale socio-technical systems – towards a framework for action oriented industrial ecology, in M. Ruth & B. Davidsdottir (eds), The Dynamics of Regions and Networks in Industrial Ecosystems, Edward Elgar. ISBN 978-1-84720-742-5. RMNO (2010). De volle zaaiershanden – Energietransitie, op naar een volgende fase. Een handreiking voor een effectief energiebeleid. Eindrapportage RMNO-Project Energie & Duurzaamheid, RMNO.

Rotmans, J., Kemp, R. & Van Asselt, M. (2001). More evolution than revolution: Transition management in public policy, Foresight 3(1): 15–31.

Stern, N. (2007). The Economics of Climate Change: The Stern Review, Cambridge University Press. ISBN 05-21-70080-9.

Weiss, G. (2000). Multiagent systems, a modern approach to distributed artificial intelligence, MIT Press, Cambridge, Mass. ISBN 02-62-73131-2.

Wiek, A., Binder, C. & Scholz, R. W. (2006). Functions of scenarios in transition processes, Futures 38: 740–766.

WRR (2008). Infrastructures – Time to invest, number 81 in WRR Rapporten aan de regering, Amsterdam University Press, Amsterdam. ISBN: 978-90-5356-605-3.

Cytaty

Powiązane dokumenty

[r]

Lacan pieni się, s ły ­ sząc nazw isko Junga czy Jaspersa, ale sam przem aw ia ję zy k ie m Mal­ larmego i rysuje obrazeczki, gdzie topografia jaźni przypom ina

[r]

amendments to the Code of Administrative Procedure which allow the initiation and conduct of administrative proceedings electronically, more often the judgements of the

Na otwar- tej we wrzesniu minionego roku wystawie „Skar- by monet w Muzeum Lubelskim” umieszczono 22 najbardziej interesujqce skarby, poczqwszy od

Jeszcze podczas wydawania Piesni ludu polskiego w 1844 roku Kolberg zaprzestal dawania prywatnych lekcji gry na fortepianie, w lipcu tego roku skonczyl takze pracf w

market and institution, the research Use the tools of industrial organization to explain the emergence of electricity power exchanges in Europe, their functioning, and their role

A disadvantage of smaller wings that use cross wind power is that it will be more difficult to restart a laddermill after downtime, because ground winds need to be high enough