ANGOLA ALGERIA CAMEROON CHAD. CONGO EGYPT.. EQUATORIAL GUINEA GABON LIBYA. NIGERIA SOUTH AFRICA SUDAN TUNISIA OTHERS
Data Validation and Reconciliation for Upstream Applications
When the author visited part of
the subsea oil exploration and production (E&P) system of Mexican Petroleum (PEMEX)
in the Gulf of Mexico he did not imagine that computer aided process engineering
(CAPE) technologies could benefit from such upstream process systems.
Indeed, it is questionable as to whether the personnel operating in heavy
conditions should need to consider mathematical models and optimisation methods.
If the crude oil flow rate of a well is sufficiently depleted, the personnel
will bore another well. Several questions arose regarding why well production is
weakening, whether it could be sustained to a higher quality level and where to
drill another well, although this was more than 15 years ago from the time of
press. Computer-assisted oil and gas production is not a futuristic vision
anymore; however, it can benefit greatly from a mature technology that is
successfully applied in the downstream business – the data validation and
reconciliation (DVR) technology.
The objective of this article is to draw upstream experts and decision makers
into that technology, describe it briefly, analyse obstacles and incentives to
its implementation in oil and gas production and report about the benefits
obtained from a first demonstration implementation in a subsea production field
in West Africa.
The Technology
The data validation and reconciliation technology consists of a method and its
corresponding software tool. Its purpose is to:
detect and correct deviations and errors of measurement data so that these satisfy all balance constraints;
exploit the structure and the knowledge of the process system together with the measurement data to compute unmeasured data wherever it is possible, in particular the key performance indicators (KPI); and
determine the post-processing accuracy of measured and unmeasured data including KPIs.
The method
enhances the quality of the information data, which in turn enables sufficient
monitoring decisions and knowledge-based actions.
Figure 1* displays recorded raw data and the
corresponding validated data. It illustrates one of the assets of the DVR
technology, allowing valuable information to be unearthed that would otherwise
be hidden in the ‘noise’ of raw measurement data. In this situation for
instance, it is only after DVR that a functional relationship between the two
displayed variables can be obtained.
Figure 1: Data Reconciliation Unearths Hidden
Information
From such simple facts a control
expert will understand the complementarities of DVR systems and advanced process
control (APC) systems.
Mathematically speaking, the method states and solves a non-linear programming
problem. In upto- date DVR software tools the constraints equations (mass and
energy balances and thermodynamic equilibrium constraints) are automatically
generated when modelling the process at hand using a graphical user interface.
The method implements statistical and thermodynamic principles so that it
differs from simulation methods. DVR determines the performance parameters of
the process system based on measurement redundancy. It therefore does not
compete with simulation methods that can benefit from the said consistency data
processing.
In a recent preliminary study the potential synergy between DVR and data mining
(DM) was analysed.
Figure 2: Soft Sensor
Reproducing Validated KPI
Figure 2
is based on approximately 4,000 DVR results on an oil refinery plant. The purple
curve displays a validated KPI as a function of time and the green curve
presents the corresponding KPI values as predicted by a DM soft sensor during
and after the learning period. Such an illustration demonstrates that DVR brings
new assets for DM systems.
The DVR tool can be used singly but can also be used in synergy with other
tools, as shown in this example. Another important asset in upstream
applications is when the measuring equipment is very expensive – it has been
experienced in downstream applications that the tool can determine a number of
unmeasured data two to three times bigger than the number of measured data. It
can therefore be regarded as producing a valuable virtual measurement set.
The Production
System
An oilfield is an industrial process system with its inputs and outputs. It is
composed of subsystems – the field itself or reservoir that is likely to remain
a fuzzy subset, the incoming streams network with its water, gas and (in
deep-sea) methanol injection systems, the outgoing streams network including
production wells, pipes, manifolds, riser and topside installations for energy
delivery and crude oil separation and pre-treatment. The oilfield itself may be
subdivided into several production areas. The incoming and outgoing stream
networks can cover several field areas (possibly exploited by different oil
companies or owners).
Appropriate DVR software tools can be potentially applied to deal with such
process systems due to their modular structure. At present it is only partly the
case as reported hereafter. To prepare such integrated applications, the
challenges will be analysed and the expected obstacles and incentives
identified.
The Challenge
What matters is to expand the ‘smart wells’ concept to the ‘smart field’
concept. Each subsystem is to be deeply analysed and modelled for itself – all
the subsystems then have to be integrated together to obtain a process
representation synthesis.
The challenge is to manage the inflow system, setting the individual water and
gas (and methanol) injection flow rates and to manage the outflow system,
opening or closing wells or zones to increase integrated oil production and oil
recovery during the life-cycle of the exploited field.
The core subsystem is the reservoir itself – it can be treated as a ‘black box’,
with arrays of input data and arrays of output data. It has been previously
shown that all those data can be validated providing quality information data.
Any mathematical model (simple or complex; static or dynamic) represents a
functional relationship between individual outputs in response to individual
inputs.
Appropriate models result from a compromise – the more complex the model the
more numerous its parameters and to determine more numerous parameters there is
a need for more high quality information data. In an environment of costly
measuring devices, limitations on model complexity can arise.
Smart operations along the field subsystems are the key to steady increment
production performance; however, there is a question over whether it should be
recognised that smart operations rely on the concept of quality and whether
proper actions can be decided on a system where there is a lack of knowledge, or
a partially fuzzy understanding of the system due to a lack of consistency of
the raw information data. The technology exists to unearth reliable and more
precise data interpretation. Such technology can help production managers to use
their own skills more proficiently.
Many questions can potentially be answered applying the DVR technology to
exploit available measurement data. These questions concern:
the flow rates of water and gas in the inflow system possibly in each injection well;
methanol quantity to avoid hydrates drawback;
the sensitivity of the outlet stream of a production well to water inflow in that area;
water leakage; and
indications of water breakthrough.
Implementation
in Upstream Operations
DVR is a proven technology in downstream applications. Payback times have been
reported as ‘always less than one year’; however, the question of whether it
will be readily implemented in the upstream area now arises.
Several obstacles other than suspicion or reasons of unawareness have been
identified. Obstacles encountered and in the downstream include:
there is already a range of software technology deployed;
how DVR complements existing simulation tools; and DVR tool comparison with competitors.
A second series of obstacles originates from general trends in research and development (R&D) financing:
the development effort size needed to adapt or to more competantly dedicate software tools to field systems;
whether this kind of innovative technology is part of the objectives that need to be focused on and whether this answers the specified needs;
whether the
technology can be integrated in current IT systems and whether turnkey
implementations, including modelling, can be offered; and it is important to
know whether the technology has already been applied as companies often do not
want to be the first to try the technology. There is also a question over the
costs incurred if it fails to meet expectations.
DVR implementation incentives:
Much is written about R&D financing decreases; however, innovation is still recognised as a must.
The technology has been proven to be highly beneficial in the chemical industry and in the downstream.
The technology has successfully been applied in offshore facilities in West Africa – it is at field trial stage.
There is a tendency in several upstream majors to address selected R&D topics to services companies.
There is a tendency to encourage collaborations between production companies – this may be indicated when the exploitation of big oilfields is shared.
There is a tendency to outsource noncore functions.
If an application partly fails, there is no risk of heavy consequences because the technology is an add-in; it operates as a watchdog.
Lastly, the technology can be used for the design or for retrofit of measuring instrumentation.
Benefits
Generally speaking, up-to-date DVR technology:
provides the most accurate and reliable values for all sensors;
enables validation of the data by using subsea redundancy and topside measurements when available;
provides calibration of flow meters (choke, pressure and temperature (P&T) models) using other flow meters;
provides an alarm when the difference between measured and validated values of any measurement exceeds a given threshold;
provides a back-up for multiphase flow meters and other sensors, in case of sensor failure;
determines the flow through the entire network (wells, flow lines and risers, etc.);
provides estimated values for any unmeasured value. For instance, it provides validated values for unmeasured flow rates, acting as a virtual flow meter. In addition, the accuracy of the virtual sensor is determined; and
helps the diagnosis by detecting and quantifying water or gas breakthrough.
A Practical
Demonstration Application
DVR technology was successfully implemented in an upstream offshore production
facility on the West African coast. It validates the performance of the water
injection system online every day. An application provider has developed the
model and commissioned the system on site. The validation system has been
integrated with the production historian database (Oracle) to extract the
measured data and is essentially used through an Excel interface. Production
engineers use the system daily to:
produce accurate and coherent production balances;
detect and quantify bypasses;
detect and quantify water leaks;
detect drifting and erroneous sensors and correct them when necessary; and
compare actual water injection rates to set points and take corrective actions.
The system has also been used to locate where additional counters should be placed to further improve the balance quality.
The overall project leads to numerous identified benefits:
increased accuracy of data;
detection of leaks of approximately 2,000 tons/day of injection water. Correcting the problem enabled the operator to bring production back to its nominal level;
improved follow-up of each well (without validation the cause of a decrease of 1,000 barrels per day (bbl/day) compared with a total production of more than 100,000 bbl/day would have been hard to identify);
optimisation of water injection for an improved distribution among the wells;
diagnosis on equipment performance;
setting priorities for instrumentation maintenance;
enhancement of the measurement system by focused maintenance of the instrumentation;
assistance on the locations for new sensors; and
simplified daily work of the reporting crew.
Conclusion
Data validation and reconciliation technology is a mature technology. It can
create value by providing enhanced data and upset detection, resulting in
increased production and better management of the production systems.
Its implementation in upstream business is currently being examined by several
major players in the area.