You are on page 1of 9

OTC-25457-MS

Model Based Multiphase Metering and Production Allocation


Piyush Patel, Kongsberg Oil & Gas Technologies
Havard Odden, Kongsberg Oil & Gas Technologies
Biljana Djoric, TOTAL E&P NORGE AS
Robert Dennis Garner, TOTAL E&P NORGE AS
Hans Kristian Vea, TOTAL E&P NORGE AS

Copyright 2014, Offshore Technology Conference

This paper was prepared for presentation at the Offshore Technology Conference Asia held in Kuala Lumpur, Malaysia, 2528 March 2014.

This paper was selected for presentation by an OTC program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect any position of the Offshore Technology Conference, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Offshore Technology Conference is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of OTC copyright.

Abstract
Subsea field developments are the fastest growing area in the Oil & Gas Industry. Fields with different owners are often part of
the same infrastructure and the demand for production monitoring and production allocation to the respective owners raises the
necessity for accurately metering the production. Multiphase Flow Meters (MPFM) are often used for this application.

MPFMs are a high CAPEX item as they often require a dedicated over-trawlable structure, and also involve relatively high
OPEX as the intervention of an ROV during the lifetime of the field may be necessary to give reliable readings as production
declines and new flow regimes prevail. Most MPFMs also require subsea sampling to keep the thermodynamic tables up to
date and to achieve the MPFM performance quoted by the vendor.

Model Based Multiphase Metering (MBMM) is a new technology for production allocation which might eliminate the demand
for conventional MPFMs, especially for smaller and marginal fields. MBMM uses low cost reliable measurements and takes
advantage of an accurate process model and sophisticated reconciliation and allocation techniques. Low installation and
operation cost compared to conventional MPFMs, results in substantial (USD 5-10 mill in) savings per measuring point.

This paper will focus on the principals behind the MBMM technology and present a case study of a successfully installed
project for Totals Atla field in Norway.

Introduction
Model based allocation systems have been around for two decades. Most of them are based on a steady state approach [5], but
some are also dynamic, meaning that the system will have a solution also in transient conditions. These systems have mostly
been pipeline models reaching from well to inlet facilities on a production unit. This technology goes under lots of different
names, but Virtual Metering seems to be the most common term. The basic principles of such systems is that the pipeline
model is fed with real time measurements from the DCS or SCADA into model boundaries, enabling it to simulate process
states in the pipeline in real time.

The problem with the conventional model based allocation technology is that it requires accurate measurements to be read into
the model boundaries. If a transmitter is starting to drift or fail, the allocation system is also generating wrong measures.
Additionally the software itself hasnt been robust enough to work as a real time system. These limitations and constraints
have been known since its evolution, but there hadnt been any solution to overcome it. Enhanced compilation capabilities of
computers and increased quality and robustness of dynamic simulation software over the last couple of years, have made it
possible to develop algorithms to overcome the problems with model based allocation systems.

Virtual Metering the traditional way


A virtual plant or copy of the production process is the engine of a traditional dynamic model based metering system. The
model boundaries are usually the well upstream and the reception facilities at the plant. As it simulates alongside the plant, it
provides a wide range of information that cannot be measured, such as pressure, flow and hold-up profiles inside the wells and
2  OTC-25457-MS

flow lines. The real-time model receives data from the process control system through an on-line connection. Measurements,
signals and set points from the control system are used to synchronize, or align the model with actual process conditions. This
is done by continuously updating the real-time models boundary conditions (e.g. inlet and outlet pressures) and states (such as
valve positions and/or controller set-points).

Other measurements are used, over a longer time-scale, to tune the models of the pipelines and other equipment. This tuning is
essential to keep the model tracking the real plant and the same is achieved by adjusting parameters (such as the ambient
temperatures, the CV of a valve or the roughness of a pipe) to match measured pressure drops, fluid temperatures and flow
rates.

The data read into the model from the control system is usually validated using the information attached to the actual
measurement, like the healthy status bit. A typical traditional Virtual Metering system is shown in Figure 1.

Control System

P,F,T History data


valve positions

Transmitter
Validation

Figure 1 Typical Dynamic Model for Virtual Metering

Model Based Multiphase Flow Metering The Innovation


As mentioned, the model based metering system are governed and guided by actual measurements from real field, any
deviation will result in disaster outcome for decision making. The problem with the conventional model based allocation
technology is that it requires accurate measurements to be read into the model boundaries but the quality of measurements are
not monitored with precision. If a transmitter is starting to drift or fail, the allocation system is also affected.

The key differences to the traditionally implementation of such systems [5], which has been developed and successfully
installed on the Total Atla project, is:

1. The dynamic model is much bigger, reaching from wells and subsea infrastructure through to inlet facilities and main
process systems as well as utility systems.
2. A transmitter validation layer has been introduced.

These two items solves the problem with unreliable measurements as drivers to the dynamic model. Measurements in the plant
are validated using other measurements AND the dynamic model. With this over- specified mathematical system, it is possible
to calculate the deviations for each transmitter which is covered by the model. Then the system can neglect measurements
which are believed to be out of range and use other, more reliable, measurements instead. Known optimization algorithms and
statistics are used for these calculations.

Comparison to Conventional Multiphase Flow Meters


A Model Based Multiphase Flow Meter (MBMM) and a conventional Multiphase Flow Meter (MPFM) can be compared one
to one in the following four groups:

Architecture
Technology
Installation & Maintenance
Robustness & Performance
Value
OTC-25457-MS 3

System Architecture
A typical architecture for MBMM and MPFM are shown in Figure 2 and Figure 3.

Figure 3 Architecture of Conventional MPFM Figure 2 Architecture of MBMM

A conventional MPFM, as shown in figure 3, has a complex unit which basically consists of various measurements as
pressure, differential pressure, temperature and gamma rays or echo sounder to determine the phases and velocity. All these
measurements go into a flow computer where the mass rates for each phase (3 phases in this case) are calculated using
thermodynamic packages.

An MBMM uses existing transmitters on the wells and topside from inlet facilities to export. A high fidelity dynamic model in
combination with allocation techniques are used to determine the phase splits and the velocities for each pahse.

Technology
In case of MPFM, the volume flow rate of each phase is represented by the area fraction multiplied by the velocity of each
phase. This means that a minimum of six parameters has to be measured or estimated. MPFMs commonly employ a
combination of two or more of the following measurement technologies and techniques:

Electromagnetic measurement principles: Microwave technology, Capacitance, Conductance


Gamma ray densitometry or spectroscopy
Neutron interrogation
Differential pressure using Venturi, V-cone or other restriction
Positive displacement
Ultrasonic
Cross-correlation of electromagnetic, radioactive, ultrasound signals (to calculate flow velocities)

Some MPFMs assume that either two or all three phases travel at the same velocity, thus reducing the required number of
measurements. Phase density of each phase is calculated based on PVTdata.

MBMM is based on driving the high fidelity dynamic model of production system with real time measurements from the plant
and a set of data analysis methods. The metering and back allocation is not only based on the local measurements and metering
at custody meter but it is reconciliated based on overall mass balance across the system.

An MBMM has the following software platforms:

LedaFlow a multiphase flow tool developed by Total, ConocoPhillips and Sintef.


K-Spice A dynamic process simulation tool developed by KONGSBERG.

The MBMM utilizes an integrated solution of LedaFlow and K-Spice to provide a high fidelity dynamic simulator. Both
models are then integrated seamlessly in K-Spice.

A high fidelity model is built to accurately reflect the field development. The subsea wells, flowlines and risers are modelled
in LedaFlow and the process units, such as, subsea process units, slug catchers, pumps, compressors, oil and gas trains,
metering units, etc. are modelled in K-Spice. The dynamic simulation model is then synchronized and run in parallel with
4 OTC-25457-MS

the real plant and is calibrated using automatic and routine manual methods so that it tracks the process. An Analytic layer is
embedded in the system and it is used to monitor and display results calculated from the simulator and to help optimize the
system.

DYNAMIC PROCESS MODEL


Well P, T,
Chokes, roughness ..

Production KPIs

On/off signals.,
valve positions,.. ..
Loop status

System Diagnosis Calibration


& Loops
Historian
Reporting
Estimated

Pressures,
flows,
temperatures..

Reconciliation Target calculations


Measured. Transmitters (in clusters) Deviation from
target

Every model
time step

Target.

Figure 4 MBMM working philosophy

Figure 5 gives an overview of the method by which MBMM works. As mentioned, a high-fidelity dynamic process model is
the basis of the system. The heat and material balance is maintained in the system. The model is synchronized with real plant
status from the historian (Examples of synchronization signals are valve positions and on/off status flags). The measured
values for pressures, flows and temperatures are read from the Historian and are subjected to a priory validation (example:
validation quality code received from the historian and out-of-range check).

To achieve high redundancy and to validate the measurements against each other, it is highly recommended to include all
available measurements (Temperature, Pressure, Flow and Density) for data reconciliation. Every single measurement is
configured with a Reconciliation transmitter, which keeps all relevant information, knows the corresponding reconciled model
value at the measurement location and can calculate residuals and statistics on deviations. Each reconciliation transmitter
evaluates an output status signal (good, suspect, bad DCS signal, suppressed etc.).

Apart from validation of individual measurement, physical relevant measurements that can be interpreted to have little or no
transport or hold-up delay between them are configured in clusters. A least-squares data reconciliation problem is solved for
the measurements that belong to the cluster. The output of this calculation is a target value. This target value is calculated
dynamically and will depend on the current dynamic state of the plant. Each reconciliation transmitter belonging to a cluster
will be presented with its target value. The target values can be interpreted to be the condition the model should converge to.
Once the target values have been defined, functionality to achieve the target values must be in operation. This function is
achieved by calibration loops that will adapt the battery limits or model parameters. Examples of calibrated properties are:
Well temperature and pressure, ambient temperatures, pipe roughness, choke CV.

The calibration loops are basically set up to be in automatic mode to ensure that target is achieved in real-time mode.
However, some properties are configured to be manually calibrated. The calibration philosophy will be system dependent. The
status of the calibration loops must be evaluated.

A system diagnosis & reporting tool generates automatic periodic reports. The content of reports will be KPIs:
OTC-25457-MS 5

Calculated on data extracted from the dynamic model. Examples are rates and accumulated production: per well,
chemicals, water/gas injection rates, export
based on results from configured smart agents (steady state indicators, hydrate monitors, slugging indicators)
based on sensors and cluster performance of the system
based on an overall test of the residuals in the system, i.e. .the system performance tracking test, that will indicate
how good the model is matching the received measurements

Back allocation of flow is also handled by the reconciliation solver which supports two types of flow allocation methods, the
KONGSBERG Method and the API 85 method. The KONGSBERG method is a modified version of API 85 in which it
allocates more to the wells with high production than the flow uncertainty requires. API 85 allocates flow based on uncertainty
calculations only, which can give wells with low production unreasonable high reallocation compared to its production.

Installation & Maintenance


Compared to MPFM, MBMM installation is less demanding as no physical devices must be installed. K-Spice Meter has
two components to complete the system installation. High fidelity system model is installed in a server and the same is
connected to OPC server (client side) to facilitate communication with data historian. K-Spice Meter can be installed in CCR
or onshore networks parallel to operator consoles to facilitate the alternative production monitoring. It is highly recommended
to install parallel system to the metering or hydrocarbon accounting station. As this system is connected to data historian, not
directly with plant DCS, the DCS network security is not compromised. As the data acquisition is achieved based on OPC
protocol, redundancy on data acquisition is ensured on an MBMM. Along with plant start- up, the model in an MBMM must
be calibrated to match the actual production data.

To achieve consistence accuracy despite the changes in flow regimes, increased salinity and change in fluid composition,
MPFM requires periodic calibration and maintenance. The physical maintenance on such systems is of course the most costly
and complex as it may need ROV intervention. An MBMM doesnt require physical maintenance in the same way. To a
certain extent, MBMMs are calibrated automatically. The non- physical calibration of an MPFM and an MBMM is
comparable. Both systems need to be validated and adjusted at regular intervals.
___________________________________________________________________________________________

The non- physical calibration of an MBMM and an MPFM is comparable. Since an MBMM is installed in CCR or
onshore environment and requires no service in the hazardious area, it can be maintained at relatively low cost.

Robustness & Performance


K-Spice Meter is a real time system that calculates the most probable flows including water and gas content for oil and/or
gas production facilities. The calculations in K-Spice Meter are dynamical, and the calculations will work both in steady
state and during transients. All metered flow from a fiscal metering will be back allocated to the producing wells. K-Spice
Meter will use all available measurements. It will calculate the redundancy for the different parts of the system. The calculated
accuracy of the reconciled flows can be presented.
If some of the measurements are failing, the system will automatically leave out the measurements with their accuracies. But
the remaining valid measurements will continue with the reconciliation and back allocation as before. It will be some increase
in uncertainty. The system has the functionality to perform reruns of the calculations based on the historical values. This is
important in cases where well compositions have changed without these being detected automatically.

In the table below, the uncertainty limits specified in NORSOK STANDARD I-104 Fiscal measurement systems for hydrocarbon
gas [2] is listed:
6 OTC-25457-MS

K-Spice Meter will normally achieve accuracy in gas metering between 3 - 5% of std. volume and between 5 - 10% of mass
for liquid. A drawback of K-Spice Meter is that its actually hard to measure the accuracy on a normal test bench for MPFM.
This is due to the fact that the technology requires measurements from a production facility to function. However, industry
experience from similar model based systems indicates that the above numbers are in line with what can be expected. The
accuracy will also be a product of the degradation of the measurements being used as input to the system.

It is very hard to state something accurately about the robustness and accuracy of MPFMs. The number of different
experiences among Operators seems to equal the number of installations. There seems to be a trend, though, that the accuracy
of an MPFM is significantly reduced after 12 months of operation. It is also a fact that an MPFMs accuracy reduces when the
fluid composition changes. MPFMs seem to be especially sensitive to fluid composition changes related to higher watercuts.

Value
As a consequence of which an MBMM doesnt introduce additional hardware (transmitters) to an existing facility, has no
equipment in the hazardous area and can be maintained from a PC from anywhere, an MBMM is an attractive solution for
production allocation. It saves CAPEX as the solution doesnt involve expensive hardware and it saves OPEX in which
complex and comprehensive maintenance is avoided offshore. An operator will typically save in the range of USD 5-10 mill in
life time cost for each conventional MPFM replaced by an MBMM.
___________________________________________________________________________________________

K-Spice Meter saves CAPEX and OPEX in the range of USD 5-10 mill per allocation point

Case Study Atla Field in Norway


Atla is one of the smallest fields ever developed on the Norwegian continental shelf. This small gas-condensate field was not
large enough to justify a dedicated pipeline and separator on a host platform, and the project had to make use of existing
subsea infrastructure (fig. 6).

Cost-benefit calculations showed that the reduced uncertainty gained from using multiphase meter technology, instead of a
virtual flow meter based on computer algorithms, to determine the flow could not be commercially justified.

Partners and authorities were involved at an early stage and kept informed throughout the process as different options for
model based multiphase flow metering were evaluated. Finally, with their consent, the opportunity to utilize non-conventional
technology and save a considerable amount of CAPEX and OPEX was seized, and the project became one of the first to
develop a field with model based multiphase flow metering as the only tool for production allocation metering.
OTC-25457-MS 7

Figure 5 Atla tieback

Model Based Multiphase Flow Metering


The Skirne/Atla Allocation Module (SAM) is a dynamic model of the subsea system, consisting of three wells and a single
pipeline to the inlet separator at the host facilities.

Several simulation tools could have been used, but the decision was made to build the system, delivered by Kongsberg Oil and
Gas Technologies (KOGT), on an existing model of the Skirne and Byggve field. The process model is built on the K-Spice
process modeling simulator combined with LedaFlow for simulating the wells and pipelines (fig. 7). Combined, these two
applications make a powerful online adaptive dynamic simulator which follows the actual measurements from the plant.

Figure 7 LedaFlow

Modeling includes the wells, Xmas trees, flowlines, inlet facilities and first-stage separator, together with all the valves in the
system. By including the common pipeline, inlet facilities and first-stage separator in the model, more accurate separator
metering measurements can be back-allocated to the wells and the overall accuracy of the model is improved. A total of 180
signals (plant measurements and valve position signals) are used as input to the model. The measurements are weighted in
accordance with the performance of the individual instrument to optimize the performance of the simulator.

More than ten years of R&D work between Leda partners resulted in the first in-house test pilot on the Skirne and Byggve
fields. This pilot project was launched and funded by R&D in Total E&P Norge in 2010. In the next phase, the Skirne/ Atla
Allocation Module was based on the existing Skirne-Byggve test pilot and further developed in the Atla project to add the
features needed for it to become a fully functional allocation tool.
8 OTC OTC-25457-MS

In 2011, the Norwegian Research Council recognized the innovation and the challenges associated with this new technology,
and awarded the project funding through the Demo 2000 program. The project aims to demonstrate that LedaFlow
integrated with K-Spice can be successfully used as a dynamic reconciliation system for product allocation purposes.

Outlooks
Due to the criticality and commercial importance of the SAM application, a set of Key Performance Indicators has been
developed to give full insight into the performance of the system. This includes the validity of thermodynamic tables
(composition), the validity of gas/oil ratios, the models ability to track plant-measurement data and the performance of
adaptive tuning functions. In addition, a calibrator package has been incorporated to evaluate and find the optimum setting for
fixed parameters. The new features introduced by the project and developed in collaboration with KOGT will be of benefit to
the whole industry.

The proven results obtained from the model based multiphase flow metering system resulted in CAPEX savings related to the
acquisition and installation of the alternative MPFM hardware and reduced the OPEX for maintenance and intervention
activities. In addition, the solution contributed to meeting the challenging deadlines of a fast-track project by minimizing the
scope for fabrication of long-lead items, installation, scheduling and commissioning.

The results regarding performance of the actual virtual metering systems are yet to be proven when the system is in operation
in commingled mode. It is currently being tuned and the initial results from the testing under normal production conditions are
promising. Results from the first month of production can be seen in figure 3. Performance levels for an MPFM are included
for comparison. The results are well within the acceptable performance requirements with regards to the assumptions used in
cost benefit calculations for normal production.

SAM - 1 month production


5 2500

4 2000
Gas [MSm3/d]

3 1500

m3/d
2 1000

1 500

0 0
21.10.12 26.10.12 31.10.12 5.11.12 10.11.12 15.11.12
Gas - SAM Gas - Plant Condensate - SAM Condensate - Plant MPFM performance

Figure 6 SAM performance over 1 month of operation

Conclusion
Driving the dynamic model with real time measurements from the plant opens a new world of information for the operator in
terms of model estimated data and enables him to take better decisions.

Without the enthusiasm behind this innovative approach, the project would most probably have opted for a more traditional
route, using multiphase meter technology to determine the flow split between the two commingled fields.

Virtual Metering Systems and Reconciliation Systems will undoubtedly continue to play a role in the future for Production
Management Systems, and provide decision-making support for production optimization. In addition, the system could be used
as a complementary technology for validating readings from installed multiphase meters. The Atla project aims to prove that
less conservative performance figures can be used for virtual metering systems when evaluating different metering solutions. If
it is successful, this type of system will find itself competing with alternative technologies like MPFMs and is likely to often
appear as the sensible and cost-friendly solution for marginal fields. The potential cost savings for upcoming projects appear to
be significant.

References
1. API RECOMMENDED PRACTICE 85. Use of sub-sea Wet-gas Flow meters in Allocation Measurement Systems, FIRST
EDITION, AUGUST 2OO3
OTC-25457-MS 9

2. NORSOK STANDARD I-104 Fiscal measurement systems for hydrocarbon gas http://www.standard.no/PageFiles/1232/I-
104r3.pdf
3. NORSOK STANDARD I-105 Fiscal measurement systems for hydrocarbon liquid http://www.standard.no/PageFiles/1233/I-
105e3.pdf
4. TechnoHub Totals Exploration & Production Techniques Magazine
5. SPE paper from TOTAL

You might also like