SPE logo

Production Forecasting

6 – 7 March 2013

Rio de Janeiro, Brazil | Hotel Pestana

Technical Agenda

Wednesday, 6 March, 0900-1000

Introduction and Objectives of the Workshop

Program Chairperson: Daniel Miranda, Petrobras

  • Current status
  • Review of Berlin & Cadiz
  • Review of goals
  • Brief discussion of definitions


Session I: Beginning with the End in Mind

Session Chairpersons:  Marcia Rosa, OGX; Claudio de Lima, Statoil

Description: Generating production and injection forecasts for hydrocarbons fields is a work process where three elements are present:

  • Estimated production and injection potentials
  • Forecast of production and injection efficiency
  • Production and injection forecasts for both short and long term

Production and injection potential are mainly related to reservoir and dynamic understanding while production and injection efficiency are related to historical plant performance. Both elements, potential and efficiency, should be considered in the process with an uncertainty range. Consistency with previous processes must be checked.

The production forecasts should have expected, P10 and P90 values based on the uncertainty ranges for potential and efficiency. For it, a multidisciplinary uncertainty analysis should be performed according to the complexity of the case and the available data and information, regardless the methodology used for generating the profiles. The purpose of uncertainty analysis is to allow better decisions when using forecasts to avoid downsides and grasp the upsides.

Since input from different disciplines is required for generating the forecasts, it is recommended that all assumptions used in the forecasting process are clearly communicated (provided/informed), supported and approved by the leaders as early as possible. This will help avoid last minute changes and guarantee sufficient time for the final forecast and its approval.

Confirmed speakers:

  • Matei Negrescu, Statoil
  • Daniel Dias, Schlumberger


Session II: Obtaining the Right Data

Session Chairpersons:  Sezai Ucan, YPF;  Alberto Sampaio de Almeida, HRT

Problem oriented data gathering and interpretation will lead to problem-oriented models construction, data integration and validation, and forecasting.

  • Data sources
  • Data quality, adequacy & consistency
  • Audit and storage

Confirmed speakers:

  • Cesar Ortiz, YPF
  • Carlos H.L. Bruhn, Petrobras
  • Yngve Belsvik, Statoil


Session III: Which Forecast Method Should I Use?

Session Chairpersons: Toni Uwaga, Centrica; Kamel Bennaceur, Schlumberger

With several forecasting methods available to the Industry the challenge is usually deciding which method to use. It is important to look at the challenge from a global perspective but with a local and field specific mind set.

  • Decline curve/analytical
  • Simulation
  • Analogies/other

Confirmed speakers:

  • Ivan Leitao, Statoil
  • Mickaele Le Ravalec, IFP Energies Nouvelles
  • Claudia Amorim, Shell

Thursday, 7 March, 0830-1030

Session IV: How Should I Combine Multiple Forecasts?

Session Chairpersons: Stephan Weber, Shell; Djuro Novakovic, Chevron

In order to combine and aggregate forecasts at the corporate/country/asset level, they must be based on common definitions. In previous workshops it has come to light that different companies have different definitions for probabilistic forecasts and this has slowed progress in defining an aggregation process. The following will be addressed:

  • How to define base vs. investment forecasts
  • How to define low/mid/high (resp. P90/P50/P10) forecasts
  • How to define long term/short/medium term forecasts
  • How to define train-wrecks or black Swan events and include them in the forecasts at the appropriate level

Participants are requested to come with their current internal definition for the above, view on how individual components are derived, and what kind of problems they may have encountered in aggregating them into forecasts at different level of aggregation (ideally, with examples as available). Presentation of alternative definitions will be followed by joint effort in workgroups to assess pros and cons and reach a group consensus, followed by presentation of findings and recommendations.

The goal is to agree on a set of definitions that will allow for an aggregation process for probabilistic forecasts that is both pragmatic and has technical rigor as well as being compatible with SPE-PRMS (Petroleum Resource Management System).

Confirmed speakers:

  • Matt Kedian, Shell
  • Delton Sy, Shell


Session V: Does My Forecast Meet the Needs of the Business?

Session Chairpersons: Omar Ceroi, Petrobras; Dave Reid, Shell

Description:  Given the business needs in terms of production delivery and expected risk, the most important aspects to be solved can be summarized by answering the following questions:

  • How do the next processes use the production forecast?
  • Are the techniques used to model the reservoir/uncertainties compatible with the expected output production forecast range?
  • At portfolio level, how robust are your the P10/P90 representations? What are the processes used to assure the robustness/comparability of the forecasts?

Confirmed speakers:

  • Omar Ceroi, Petrobras
  • Diego Diaz Peralta, YPF Argentina
  • Mauro da Silva Sant'Anna, Petrobras
  • Alexandre Castellini, Chevron


Session VI: How Can We Learn and Improve Forecasts?

Session Chairpersons: Denis Schiozer, UNICAMP; Regis Romeu, Petrobras

Description: This session is dedicated to exploring how we can improve forecasts. We should be attentive to benefits of retrospective analysis (from comparing forecasts with actual results) and of special training programs. Discussions during previous sessions can be revisited and the contributions will be oriented to develop SPE forecast guidelines.



Wrap-Up and Recommendations