By Alfred Majek and Julio C. Adjunta
High-performance personal computer (PC) and intelligent remote terminal unit (IRTU) technology have optimized the remote control of gas lift injection and surveillance of over 1,000 offshore production wells at Lake Maracaibo, Venezuela.
Maraven S.A., a subsidiary of Petroleos de Venezuela, operates this central Lake Maracaibo light oil and gas production.
In its 3-year program, Maraven expects a 27,000 b/d increase in oil production by reducing deferred production and optimizing gaslift injection by as much as 20%. In addition, real time data on well performance will enhance production management as well as allocation of operational and maintenance resources.
In Lake Maracaibo, Maraven S.A. operates 52 flow gathering stations. Each has between 15 to 25 gas lift driven wells.
The gas lift loop is a complex network of suction pipelines from the gathering station to several compressor plants where lift gas enters 42 distribution manifolds at 1,600 psig.
At the gas lift distribution manifold, a 2-in. line feeds gas to a well through a needle-type manual choke valve. A circular-chart, differential-pressure recorder installed in series with the choke valve allows the operator to set the gas injection rate for a particular well. The setting is authorized on a daily basis by production engineers in an onshore office at the Maraven headquarters in the town of Lagunillas.
The Lake Maracaibo production system has been known for its long and ineffective well response time. Offshore, operating crews rely on boats for transportation. This greatly affects their ability to set the manual choke valve and collect gas lift intake circular charts for each well.
To make matters worse, many wells are up to 1 mile from a gas lift distribution manifold and the flow gathering station. In some wells the tubing-head pressure varies by as much as 300 psig or more. While these pressure fluctuations can be caused by a number of different circumstances, most fluctuations are mainly caused by the fixed position of a manual choke that allows the gas to first fill the long gas lift line and then to fill the annular space of the well. Therefore, a time lag occurs between the actual injection demand and the tubing-head pressure.
To automate Maraven's manned production operation at Lake Maracaibo, a PC/IRTU-based system was developed by Maraven S.A. and International Automation Resources, Inc. (IAR). Fig. 1 illustrates a typical gas lift loop that includes both the manual and remote-control gas lift branches to a well platform.
The remote control system consists of a solar-powered wellhead monitor (WHM) installed on each well platform. At each flow gathering station within a two mile range of a family of wells, a host terminal unit polls and stores the well data with low power, 250 mW radios. From a remote location, 60 miles onshore, an operator interface implemented by a PC-based Realflex/QNX master polls the host units for real time data with 5-watt radios operating in the 900 megahertz band.
Key factors confronting the detail design were:
Hardware and software management of the WHM efficiently manage power consumption to a point that two small 10-in. x 7-in. solar panels and two 8 amp-hr batteries allow the monitor 4 days operational autonomy even in the most severe and cloudy weather conditions.
Specifically, a gas-piston driven, high-resolution throttles the gas lift choke valve. This eliminates heavy, power-hungry actuators. For maximum power savings, hardware and software control the sleeping/active and sampling modes of the electronics.
In addition to the wide weather operational autonomy, the monitor is designed to continuously operate even in during a lost telecommunication link between the WHM and the host, or between the host and the onshore master. Loss of production would become critical if a well or family of wells lost control during radio communication failures.
The WHM behaves as a self-sustained flow controller, capable of storing up to four days worth of 2-min interval real time well data.
A branch-tree communication structure effectively organizes an enormous volume of real time data from the wells.
At 2-min intervals, a WHM uploads eight process variables, calculates and controls the gas lift injection setting, and estimates the gas-to-oil ratio (GOR) and water mass flow rate at the well discharge. In addition, the WHM's software handles the master operator downloading of all its operational parameters.
A 1,000 well family can generate more than 10,000 time-tagged data points every 2 min. To deal with this significant data volume, a branch-tree data flow pattern was configured to run the overall system.
While for purposes of accurate flow control the WHM samples and time tags the well variables every 2 min, the host interrogates the WHM every 6 min for the last data sets. Conversely, the master interrogates the host every 6 min and trends the well status using the available Realflex software. This configuration has reduced the data handling points by 66% of the normal volume and yet has maintained adequate well surveillance.
The wellhead platform requires minimum electronics maintenance and servicing. This design was of prime importance so that the abundance of electronic and position servo components did not burden maintenance crews. To ease this concern, the WHM is designed on a modular basis. Each WHM is composed of five main modules:
Module repairs are intended to be done onshore. Consequently, work at the offshore well location is limited to replacing defective modules.
Gas lift flow rate control within 1% of set point value and mass flow rate trending for the well net output determine the stability of the overall field production. The WHM application software resolves the standard non-linear flow equation:
C * Y * B^2 * Fa * Fx Q = N * ------------------------------ * d^2 * F(D) * SQRT(H) SQRT(1 - B^4) Where: Element is an orifice or ISA short nozzle Q = Volume or Mass flow N = Unit of measurement Factor C = Element Discharge Coefficient Y = Gas expansion Factor B = Element Diam-to-pipe Diam. ratio Fa = Element Thermal Expansion Coeff. Fx = Liquid-in-gas block Coeff (Murdock factor) d = Element discharge diameter F(D) = Density Function H = Diff. Pressure across the Element
On the injection side of the well, the gas flow rate is computed by reading the gas lift pressure, temperature, and differential pressure across an orifice plate. A proportional pulse length algorithm determines the flow rate deviation from the downloaded set point and sets the position of the automatic choke within a 1% dead band around the desired value.
The length of the pulse and the dead-band width determine the close-loop gain of the controlling action. To account for choke hysteresis, a correction pulse is factored within this algorithm whenever the actuator changes the throttling direction.
On the well discharge, a short radius ISA nozzle has been provided. This element is primarily intended for the determination of the flow/no-flow condition of the well. Although other sophisticated flow/no-flow detectors exist, a nozzle offers advantages particularly in multiphase flow correlation. In addition, the differential pressure across this element gives substantial resolution in observing actual flow patterns and tendency behavior of the well.
From a mechanical viewpoint, a nozzle has proven to have less erosion impact due to solid suspension in the flow stream than a plane orifice. Multiphase flow correlation is possible if the GOR and percent water are known. Such information is usually obtained from a test separator located at the flow station.
The majority of Lake Maracaibo well flow patterns fall within 60:40 gas/liquid mass flow rate. With the actual GOR and %H2O known, the WHM algorithm estimates a blocking factor Fx (Murdock) that correlates the differential pressure across the nozzle to the multiphase mass flow rate at the discharge side of the well.
Gas lift optimization has long been a major issue in Maraven's operational division at Lake Maracaibo. If a compressor plant loses a compressor train, the remaining gas lift must be distributed.
The well automation system accomplishes this task by downloading specific commands to the WHM from the master computer at Lagunillas, cascading its traffic through the host unit at the flow station.
Fig. 2 depicts typical production characteristics of a given well. This functional well characteristic of the gas lift-vs.-crude, CR = F(GL), is plotted by placing the WHM in a manual remote mode and allowing the well to flow into a test separator at the flow station. An operator can then vary the gas lift flow rate and plot several points to derive the true production curve of the well, consequently detailing the optimum gas lift injection rate (Qopt).
This procedure, which may take a few hours to complete, yields the steady-state response of the well along with actual GOR and %H2O by directly measuring the gas from the separator and a centrifugal separation of a sample of the liquid phase.
Once the steady-state flowing characteristics of the wells are known, a computer program approximates each of the well curves with a third degree polynomial of the type:
CR = C0 + C1(GL) + C2(GL)^2 + C3(GL)^3
To further complete the well definition, the field engineer must determine the degree of importance relative to overall field performance of each well within a family. Subsequently, each well of a flow station family is given a basket classification number (BKT) between 1 and 5. The 5 is assigned to the most important wells.
This value determines the wells that should have gas lift reduced whenever there is insufficient gas for all wells. The selective reduction of gas lift to a given well is referred to as "trimming" for the gas lift shortage.
Wells operating in a trimmed mode reach a state of minimum production during which the casing stays gas packed until the shortage is over.
Parameters that the operators download from the master station to each of the WHMs for proper operation include:
The operator at the master station can then issue operational commands, either directionally by selecting a particular well, or to a whole family of wells. Operational Commands are of the following type:
Optimization is achieved using Qopt during daily operations, and the trim capability when supplies of injection gas are low.
Because of these circumstances, this project serves as an excellent example of the issues that may be manifested during any similar work.
Maraven S.A. recognized the need for local engineering support for obtaining the feedback required for future expansion. In addition, national interest is best served by the development of a local technology base. Speed of response to warranty questions and overall cost are also valid concerns. These issues were resolved by establishing a close relationship between the original manufacturer and the group, Recursos Electronicos de Venezuela (REV).
Because programming a master station requires an interactive approach with the client, REV was assigned this particular task. In addition, REV handled questions regarding the system hardware and installation.
Close attention to all facets of the effort was absolutely essential in attaining the overall goal of a compact, rugged, modular construction, low-power wellhead monitor.
Implementation of an intermediary unit, the HRTU, offered the solution. The HRTU functions as an autonomous data gathering unit constantly acquiring information from the monitors for subsequent transmission to the master station. This approach allows for low-power radios within the wellhead devices while the HRTU is outfitted with high-power equipment.
Ultimately, the final disposition of these design questions could only be achieved through cooperation between a client and vendor. Sound ties are imperative in work of this nature, and must be given the level of consideration worthy of the most critical component of system design.
Maraven's oil and gas production stability depends greatly upon the timely availability and safety of its facilities at Lake Maracaibo. Therefore, Maraven is conscious of the economic advantages of production automation and has established a sound international reputation in this field.
The ability to determine if a well is off production within a short period of time enables field operators and maintenance crews to correct the anomaly at least during the next day, thus increasing revenues by decreasing production down time.
The return in investment and revenue increase for Maraven S.A., therefore, is because of this factor and the following:
Every day, more economic advantages are seen possible with the implementation of this project. Because the daily clerical effort of administering production including collecting data from the well platform, adjusting gas lift flow rates at the distribution manifolds, and integrating circular charts is virtually eliminated, operators can concentrate their efforts on analyzing the well data and optimizing behavior.
Maraven is also conscious that oil spillage may contaminate lake waters if an oil line from the well to the station should break. The WHM could detect such an anomaly and shut the gas lift injection down, thus minimizing the contamination impact.
Similar corrective action needs to be taken in the event of a fire in a flow station. This would require a whole family of wells to be shut down in the shortest time possible.
Because of the overall success of the project to date, the future is very bright for continued expansion of the system on Lake Maracaibo.
Julio C. Adjunta held the position of project technical advisor for Maraven in Lagunillas, Venezuela, during the WHM development. He has 34 years' experience in system integration and project management. Adjunta has an MS in electrical engineering with the University of Michigan (1972).
Alfred Majek is chief executive officer of International Automation Resources, Inc., Houston. His experience includes 5 years with supervisory control and data acquisition (scada) systems as a project engineer with Mobil Corp. and 13 years on various aspects of industrial automation design and manufacturing. Majek received a BS in electrical engineering from Rice University in 1976. He is a registered professional engineer.