Value of Information

Arguably the primary function of a Practising Reservoir Engineer is to provide information to Asset Management in order for decision making to optimised. The value of this  information will be associated with different management option scenarios and influenced by production performance uncertainty.

In his books Dake pointed out that Asset related data acquisition is required to give the engineer a fair chance of providing meaningful information. And, that as data acquisition is expensive it was the engineer’s responsibility to convince managers of the necessity of that data.  This was in the context of application of quantitative engineering practice.

In the 21st Century, if the growth in related technical publications is anything to go by, the practice of probabilistic engineering is increasing . The influence of uncertainty, if known and which could lead to meaningless quantitative reservoir engineering evaluations, can now be evaluated with mathematical rigour. However, scrutinising the assumption that uncertainty and its assigned probabilities can itself be known with any confidence might be prudent (see Subjective Probabilities)

In recent years the reservoir engineer has been encouraged, in the SPE literature, to seek data acquisition not as a means to reduce uncertainty but as a means to enable optimisation of decision making. There is no point in reducing uncertainty if it has no influence on Asset related decision making. The literature promotes ‘Value of Information (VOI)’ analysis to justify the expense of data gathering on which the information is contingent.

Bratvold et al’s SPE paper No 110378 ‘Value of Information in the Oil and Gas Industry; Past, Present and Future’ rigorously defines VOI analysis in mathematical terms. Beware, apparently the SPE literature contains different and sometimes erroneous definitions.

In the paper’s abstract the authors reflect on the efforts and expense of data gathering and analysis undertaken to generate information pertinent to decision making and  comment;

“Yet, considerably less time and resources are expended on assessing the profitability of this information”

The question ‘Why is that?’ is posed. It is a question that may not be too difficult to answer in the decision making environment of the UKCS. One possibility is offered  here.

First published in 2007 Bratvold et al’s paper cites 30 SPE publications and points out that only 7 are associated with real decision making applications. Most papers are associated with acquisition of seismic data. Further details of SPE publications pre-2007 can be found here.

The paper by Kumar et al SPE No 96171 “Value of Information for Appraisal of Multiple Prospects” presents an application relevant to Reservoir Engineers. It addresses the dilemma that;

“..the time and effort needed to update the analysis for new information is nearly as great as for the first complete iteration”

Bratvold et al’s paper usefully provides details of of VOI misconceptions such as; the determined value of information is not an expected value but the difference of two expected values. An appropriate value of information estimate is dependent on the attitude of the decision maker(s).

Since 2007 two further papers illustrating VOI analysis for data gathering associated with production performance have been published.

A Procedure for Assessing the Value of Oilfield Sensors

SPE 109628, 2007                                              R.B Gilbert et al

This paper provides examples of VOI analysis of deployment of downhole sensors under a number of recovery technologies; Mature Field, CO2 injection, SAG application, MHF in Tight gas and waterflooding a Deepwater reservoir.

The selection of 'posterior' probabilities associated with uncertainties in reservoir performance is well discussed. However, comparison of the assigned values to the 'prior' probabilities is not discussed. An article,under the Chinese Whispers theme discusses the mature waterflood scenario. It  makes the case that there is some knowledge ‘fuzziness’ because of an unrealistic change from assigned 'prior' to 'posterior' probabilities.

The Value of Well Testing – Optimisation of the West Brae Field

SPE 124271, 2009                                             M Tharagonnet

This paper presents a more simplistic approach in assessing the VOI associated with repairing a multi-phase flow meter on a North Sea Satellite development. The value of information contingent on a meter is associated with the ability to optimise offtake strategy. However, the reader may be left wondering why a subsea manifold meter was necessary given that “The West Brae field is metered independently on arrival at the Brae Alpha Platform”.

The paper provides a valuable example of a simplistic VOI application. And, the reader is also rewarded with an insight into the uncertainty that can be associated with well production data. To date this aspect of probabilistic engineering may not have received the attention, in published literature, it warrants.

Theuveny et al’s SPE paper 102351 “Real-Time Production – A Virtual Dream or Reality? The Case of Remote Surveillance of ESP and Multiphase Flowmeters” will be of interest to those who recognise that data interpretation is dependent not only on acquiring the data but also having contextual information (well name, depth, historic interpretations etc) available.