http://sdc.deloittenet.deloitte.com/dcsp93f6x100008apgodjex3l_5s8j/njs.gif?dcsuri=/nojavascript&WT.js=No&WT.tv=8.6.2
43

MarketBuilder 6.1

How the application works

integrated models pgec smIntegrated models for power, world gas,
coal, and emissions

MarketBuilder 6.1 is based on a branch of economic game theory called "agent-based models." MarketBuilder models are built by constructing a network of interrelated "agents" that describe the supply chain for a particular commodity. In most cases, agents will correspond to the major players in the market — for example, drilling companies, refineries, and mid-marketers would each be considered an agent. The "secret sauce" in the model is that each of these agents represents a profit-seeking entity that seeks to maximize its own profit relative to the constraints built into the system (in the same way that the real market behaves). As these agents interact with one another, the resulting equilibrium between supply and demand determines the market price of each commodity.

In the application, economic agents are graphically represented as "nodes." Nodes contain detailed economic fundamentals — logic and data — that can be grouped together to form larger nodes. Links between nodes allow the transmission of commodities and prices throughout the model. Each model is constructed by linking nodes to one another through a graphical drag-and-drop interface. Complex models are built iteratively. For example, a model of a refinery can be built as a source model or an intermediary model in the supply chain process of even larger and complicated models. The simulations are directly based on the model diagram. When the required nodes are connected, a very complex system can be simulated.

Graphical creation process

Users create MarketBuilder models by dragging and dropping components (nodes) onto a map. This creates a visual representation of the system that MarketBuilder 6.1 converts into an economic model. These drag-and-drop components describe the technical considerations of the energy industry. Constraints are what make the job of modeling a physical commodity market extremely complex, but MarketBuilder 6.1 simplifies this modeling dramatically. Fortunately, most components have already been built and can be used off-the-shelf. If necessary, new components can be built up by aggregating simple components or added through the MarketBuilder 6.1 Library using any programming interface. However, in nearly all cases, all of the modeling can be done in the visual interface without any programming.

MarketBuilder Market Components (Nodes)Easy to use — Click-and-drag objects represent key market components

The large number of off-the-shelf components makes constructing economic models much easier. These components have been extensively tested to encapsulate the spatial and technical complexity of physical commodity markets. If a component needs to be added or expanded, this can usually be done directly through the visual interface without any programming whatsoever.

mb network smComplex models — MarketBuilder 6.1 helps
represent processes and geographic dimensions
of complex models

Using the graphical interface, complicated components can be constructed from simpler components. This allows users to reuse preexisting models and build up a component library. For example, a model is shown in the diagram above. This model is composed of smaller components, the click-and-drag objects. The model can, in turn, be used as a node in larger and more complex models.

Once a diagram of a model is created using that graphical interface, the model is completed — the diagram essentially is the model. One advantage of this approach is that every model is fully transparent. It is easy to see what is going on at a high level and then drill down into the details by clicking any element. A second advantage is that, as the model gets changed over time, so does the documentation, so that they are in sync and up-to-date. In many cases where the model corresponds to specific geographic locations, MarketBuilder Desktop enables users to easily construct and visualize the model on a MarketBuilder 6.1 GIS-compatible map.

The picture of the model is the model!

 Southern Europe Natural Gas Supply, Pipe, and DemandAll of Europe supply, pipe, demand in natural gas

MarketBuilder models are all created graphically. This makes complex models easier to understand, visualize, and maintain. A palette of prebuilt components can be easily dragged onto the map to extend the model.

Drill down into a model for details

 Southern Europe Natural Gas Supply, Pipe, and DemandDrill down to Southern Europe supply, pipe, and demand in natural gas

To get details of what is happening within a particular node, it is possible to drill down for a closer look. MarketBuilder 6.1 makes it easy to go from a high-level macroeconomic view down into the technical details of a model. This helps to ensure that the executive summary is consistent with the actual model.

A model hierarchy keeps things organized

Romanian Natural Gas Supply and StorageRomanian Natural Gas Supply and Storage

Entire models — like a refinery — might be represented as a single node on the map. A model hierarchy allows component models to be viewed as nodes of larger models. As shown, the model hierarchy is also represented in a Microsoft Explorer interface on the left, which enables another way to navigate. For example, complex nodes contain details that can be expanded or hidden.

Getting the data out of the model

MarketBuilder 6.1 data can be exported into Excel, Access, Firebird or SQL databases. This makes it easy to maintain a model with MarketBuilder 6.1 and also move data into other applications. The look and feel of MarketBuilder 6.1 is similar to Microsoft Office — you can copy and paste, drill down by using folders, and use any other standard Windows feature. Additionally, many common output graphs and forms are provided to easily report on the models you have customized or built.

Technology behind the application

With the significant computational requirements and data volumes involved with broad-based fundamental market modeling, MarketBuilder 6.1 has been designed to run using standard scalable technologies and easy-to-use, familiar user interfaces.

The application:

  • Runs on Windows platforms
  • Is closely integrated with MS Office
  • Supports 64-bit operating systems
  • Supports multicore, multiprocessor computers

Usability

mb multipleprocessors smWe believe that making a modeling system easy to use is one of the best ways to get the maximum benefit. As a result, MarketBuilder 6.1 is designed to run on desktop computers without programming or specialized hardware. MarketBuilder 6.1 is designed to give users control over the modeling process. Our easy-to-use graphical interface, interoperability with MS Office, and drag-and-drop capabilities make it easy to get results out of MarketBuilder 6.1. It also makes it easy to visualize results, whether in MarketBuilder 6.0's map overlays or MS Office applications, such as Excel and PowerPoint.

Given the close integration between MS Office and MarketBuilder 6.1, users are required to have MS Excel and PowerPoint installed on their systems. MarketBuilder 6.1 is compatible with MS Office versions from Office 2003 onward.

Parallel processing and scalability

MarketBuilder 6.1 has been designed to be usable on computers with modest hardware requirements so that it can run on common desktop and laptop computers. As MarketBuilder models become more detailed, multicommodity, and multiregional or global, the processing time to run the model increases because of the increased number of required computations. Designed for parallel processing and scalability, MarketBuilder 6.1 takes advantage of multiprocessors and 64-bit memory addressing to reduce computing times. When run on Windows multicore and multi-CPU systems, MarketBuilder 6.1 automatically spreads the processing across the multiple processors, reducing computing times with nearly linear scalability, i.e., the processing time is close to single processor computing time divided by the number of processors.