Forecaster Gas


Forecaster is the application designed to meet the needs of short-term forecasting for the Gas market. It automatically provides daily forecasts and monthly programs for each network point. It also updates a wide range of indicators for ongoing performance monitoring and operating by exceptions only on points that do not perform in an optimal way.

Business Context

The natural Gas market requires accurate forecasts, because of tight regulations which compel the users of the National Gas Pipeline to provide the network operator with a detailed daily schedule of the withdrawals for each point of delivery. An error in these estimates causes significant imbalance costs. Each delivery point has different profiles and different temperature sensitivity, due to the household/industrial mix of customers below the delivery point. Large industrial customers, directly connected to the transport network, or power plants, are extreme examples of completely industrial profiles. Every profile, any shift in customer portfolio composition, must be taken into account because inaccuracies produce unbalancing costs, market risk exposure, and negatively affect other areas of the gas shipping activity, such as storage or LNG management.

ì4C Application

Forecaster Gas is designed to easily and accurately meet the forecasting needs of the Gas market. It provides daily withdrawal estimates and monthly planning per point of delivery, and offers a full range of functions, graphics, and indicators for a valid and simple management of the forecasting process. Time series can be analysed at any level, starting from the customer, up to significant aggregates  that are completely under user’s control. The system offers a full range of statistical methods, including highly market-specific forecasting methods such as  authority profiles and NDM (non-daily metered) algorithms. The user has automated support throughout the process, from customer data acquisition, through measurement/weather data processing and the assignment of forecasting logics, up to the production and deployment of the forecasting outcomes.

Business Pain – Key Features

  • Produce accurate forecasts to reduce unbalancing
  • State-of-the-art statistical forecasting methods
  • Arima, ARMAX, AREG, Exponential Smoothing
  • Forecasts at any level of detail
  • Flexible forecasting horizon
  • Statistical KPI’s
  • Minimize the statistical know-how required
  • Automated modeling for less confident users
  • Stepwise selection, self-identification
  • modeling logic based on meter clustering
  • KPI driven intervention strategy
  • Not a black box: advanced users have free access to models
  • Reduce management effort
  • All operations can be scheduled
  • Picker for data acquisition
  • KPI for accuracy assessment
  • Alerts
  • Take into account Gas Market peculiarities
  • Uses different version of TSO’s allocation data (from forecasted actuals, to M+3)
  • Automatic building of weather and calendar driver
  • Standard Profiles
  • NDM treatment
  • Modeling for direct TSO customers
  • Manage market under the DP
  • Customer list
  • Customer contracts
  • Contract volumes
  • standard profiles
  • Communicate with system infrastructure
  • Native formats for major TSO networks
  • Available capacitities
  • Automatic pickers from portals