The key is simple - come up with an intelligent daily operations system and then work it every day.

Production optimization ensures that wells and facilities are operating at their peak performance at all times to maximize production or to maximize revenues. The current manual production optimization approaches are both time consuming and error prone due to the complexity and large volumes of data that have to be considered. Frequent changes in well and surface equipment downtime, maintenance work, evolving reservoir conditions, etc., usually make it impossible for engineers to keep the asset tuned for optimal operating conditions. And production enhancement studies that are not linked to an automated optimization system have limited value because their recommendations quickly become out of date.

Recognizing these deficiencies, a new well and surface network optimization system has been developed to provide sustainable production optimization. Called i-DO for intelligent daily operations, it links real-time well, surface and corporate data sources to ensure that reservoir, well and facility models are constantly monitored and updated to reflect actual operating conditions. A sequential linear programming calculation engine then simultaneously optimizes hundreds of critical parameters. The system, originally developed for gas lift optimization, is being used in three large assets to handle very complex producing networks having hundreds of wells. The ability to gather and interpret data online as well as implement many of the changes electronically means production optimization can be sustained effectively with minimum manpower and expense. The key advantage of the system is keeping the asset operating at peak performance.

With the three online systems running today, engineers are streamlining and automating a number of production engineering tasks associated with asset management. Since no two fields or reservoirs behave identically, the optimizer is not tailored to solve specific problems. The system can be applied without modification to a wide range of production optimization problems, including downhole controls in smart wells and intelligent completions.

The technology leverages the investment of SCADA and other communications and control systems, which are being installed in an increasing number of fields worldwide. Used on a daily basis to manage mature assets and make operational decisions, the system integrates production data management and reservoir modeling with transient pressure analysis, well modeling and surface network modeling and optimization (Figure 1). The flexible open software architecture allows third-party and legacy applications to be incorporated with relative ease.

Rather than having engineers perform individual tasks, a number of processes are automated in pre-defined schedules, triggered by specific events or initiated manually:

• Well models are automatically updated to match production tests;
• Network facility models are updated for current conditions;
• Daily production data are allocated to wells;
• Root cause analysis of unscheduled downtime determines the focus of maintenance efforts for maximum impact;
• New reservoir models can be created, short-term production forecasts generated, transient analysis of downhole pressure data determines current reservoir pressure and estimates permeability and skin; and
• Data are stored and data trending, production and reservoir surveillance reports are generated.
The system can be successfully deployed on several types of assets:
• Mature, artificially lifted oil fields producing at high rates with continuous gas lift;
• Fields with intermittent gas lift;
• Gas condensate and light/volatile oil fields; and
• Dry gas fields.

Real-time in the digital oilfield

There are many views of the meaning of real-time optimization. These often stem from a specific view of the process given the technology being considered at the time. A resent SPE Technical Interest Group defined the term in SPE83978 as follows:

"Real-time optimization is achieved when the measure-calculate-control cycle is implemented at a frequency which maintains the system at optimal operating conditions at all times."

This allows for considerable variation in the meaning of real-time optimization from a cycle time of a few minutes for process optimization, to a day for production operations, to months or years when considering reservoir optimization. Of course, in order to optimize the system it is crucial that the value of variables be known over the required time-scale, e.g. oil price, if we want to maximize financial drivers. Predicting the oil price over the scale of one day is certainly easier than predicting it over a 20-year life of a field development plan. This, again, points us to making a serious effort at production optimization.

Data rich - Action poor

Moore's Law for the semiconductor industry, which states that computer processing capability of silicon chips doubles every 18-24 months, has held true form around 30 years, and looks set to do so for some time. In the same period, advances in sensors, telemetry systems and data storage and retrieval ensure that we are swamped with data. Data is the source for information, knowledge, decision-making and actions to improve the outcome of any system, but data in itself provides little or no value without appropriate processes to convert to a useful result.

The data acquisition-to-action cycle must be within the time scale of the system we are trying to optimize. Therefore, if daily production optimization is our objective, then the data acquisition and subsequent processes must produce actionable decisions in hours and minutes. Many companies have cycle times of weeks while attempting to optimize daily. It doesn't add up.

Moore's Law means that increasingly the bottlenecks to improvement are no longer "not enough data," but rather too much data awaiting conversion to information, decisions and action. The focus should be on reducing the cycle time, and therefore using a system that provides appropriate and reliable output within the required cycle time.

The benefits of sustainable production optimization are significant and have been proven with the three implementations to date (Figure 2). Gains achieved are a 2% to 5% improvement in uptime along with a 3% to 7% improvement in produced volumes and overall reduction of lifting costs by 5% to10%. In the three implementations to date, benefits like these can substantially change the overall economics of the asset.