A Multi-Paradigm Energy Model for Liquid Natural Gas Analysis

Bri-Mathias S. Hodgea; Joseph F. Peknya,b; Gintaras V. Reklaitisa    aSchool of Chemical Engineering, Purdue University, 480 Stadium Mall Drive, West Lafayette, IN 47907, USA
be-Enterprise Center, Discovery Park, Purdue University, 203 S. Martin Jischke Drive, West Lafayette, IN 47907, USA

1 Introduction

The current global energy system is both highly complex and increasingly coupled. Changes in the energy supply or demand in one region of the world, or one sector of the economy, can have far reaching and often unforeseen consequences. Energy policy decisions at both the government and corporate level play a critical role in the determination of energy usage and the availability of suitable forms of energy where necessary. These policy decisions may only benefit from a deeper understanding of the interactions of entities within the system as well as the system capabilities and limitations. As the size and importance of the system prevents direct experimentation, insights must be gleaned through the modelling and simulation of energy systems.

The most prominent large scale energy system models already in use come from international and national organizations such as the International Energy Agency (IEA, 2007) and the United States Energy Information Administration (EIA, 2007). Both of these models are highly detailed mathematical programming models which produce single solutions to the given problem instead of the spectrum of possible future scenarios which the large uncertainty involved should dictate. System Dynamics, which uses highly aggregated data with defined feedback loops to forecast system behaviour, has also been widely used in energy modelling. Typical system dynamics models for energy systems analysis are the Energy 2020 (Backus et al., 1995) and Fossil2 (Naill, 1992) models. Agent-based models work from a bottom-up approach, as opposed to the top-down approach of system dynamics, and specify the behaviour and the means of interaction between individuals within the system. With the individual’s behaviour specified it is then the combination of the interactions between system components that is studied in order to better understand the system. A framework for an agent-based overall energy system model has been developed and applied at the sub-national level (Hodge et al., 2008), while the agent-based system concept has also been applied to electricity systems (Koritarov, 2004). A more comprehensive review of past work in energy systems modelling can be found in (Wei et al., 2006) which contrasts the approaches which have been examined and gives examples of applications at varying geographical scale.

All of the modelling paradigms mentioned above have some inherent limitation. Agent-based methods with their disaggregated approach can suffer from scaling problems when there are a very large number of individual entities which must be modelled separately. Mathematical programming produces only individual solutions instead of looking at multiple possible future scenarios and is often limited to linear model components. In many cases the formal mathematical relationship necessary for a system dynamics model may not be readily apparent. The aggregation that is common in system dynamics models may miss key differences between system components which help to explain the system behaviour. It is thought that by combining these, and other, techniques in a multi-paradigm modelling system each modelling standard may be allowed to work on only those segments of the problem for which it is most suited. Instead of allowing the choice of modelling technique to dictate the structure of the problem we can allow the structure of the problem to drive the decision of which modelling paradigms are used.

The multi-paradigm approach allows the development of models with multiple objectives, multiple levels of aggregation and multiple perspectives (Zeigler & Oren, 1986). Hybrid systems, which mix continuous and discrete time, may be considered an important subset of multi-paradigm models that has received much attention. A good review of hybrid systems theory may be found in (Barton & Lee, 2002). This approach has been applied to physical systems (Mosterman & Biswas, 2002) such as supply chains (Pathak et al., 2003), as well as software systems (de Lara & Vangheluwe, 2002). The combination of multiple modelling paradigms allows the level of abstraction and the formalism used in sub-models to differ from that of the meta-model with the intended goal of a more realistic representation of the system under study. An overview of the concepts behind multi-paradigm modelling, abstractions and transformations between formalisms can be found in (Vangheluwe et al., 2002).

2 A Multi-Paradigm Energy Model Framework

A decomposition of the energy system must be undertaken in order to allow the use of multiple modeling paradigms within the same model. Subsystems must be characterized so that the modeling style which best fits each component of the system may be determined. The system has been broken down into three critical components: markets, supply and demand. An agent based modeling structure is used in order to represent each of these modules of the energy system at the highest level of abstraction. This paradigm was chosen for the ease with which communication between different modules may be facilitated. Messages may be sent between subsystems through standardized ports which can recognize relevant information and ignore non-relevant noise. The agent based methods of communication are the most natural fit for integrating multiple modeling styles of the methods considered. At the meta-model level agents act as communication wrappers which encompass modules of the system, allowing sections of the model generated through varied modeling paradigms to effectively share the information which is needed by diverse subsystems.

2.1 Market

The market modules are at the center of the energy system framework. Markets serve as a meeting place for supply and demand and are thus the main means by which information is exchanged between the two. The market facilitates communication between diverse supply and demand modules through the use of a common language: the bid. Bids to buy or sell an energy product may be submitted and contain all the information relevant to the prospective trading partner. Bids consist of six important pieces of information: the bidder’s name, the market to which the bid is submitted, the product type for which a bid is placed, whether the bidder is buying or selling as well as the offered or requested amount and price. The market module operates as a double blind auction in which each participant receives the equilibrium price and bids may be partially fulfilled. The fulfillment of a bid can be seen as a contract to supply or receive an amount of the product during the next time frame. The auction mechanism is not dynamic but operates once per each discrete time step, or for each tick in the meta-model agent-based framework. Once supply has been matched with demand and a market price has been established the market must then communicate the results of the auction to the participants. This is again easily accomplished by sending bid messages to those participants whose bid has been accepted notifying them directly of the amount of their bid which has been fulfilled and the market price that they will pay or receive per unit of commodity.

2.2 Supply

The producers who supply the product within an energy system have been modeled using an agent-based approach. The agent-based approach represents suppliers as autonomous entities which make their production decisions based upon rules established to govern their behavior and interactions with the other players within the system. Since the supply side consists of a small number of large producers as opposed to the large number of small users on the consumption side, individual behavior is more easily generalized into rules such as profit maximization. In addition the utility of the product to the suppliers, the money received from the sale, is much simpler than the utility of the myriad possible uses for the product on the consumption side. This reduces the complexity of strategic decisions and helps make the case for using agent based modeling to represent the supply side.

2.3 Demand

The system dynamics modeling approach has been chosen as the most applicable paradigm for the demand side of the energy system under study. System dynamics works with aggregated groups of entities...