OTC-18678-PP

A Field-Wide Integrated Production Model and Asset Management System for the
Mumbai High Field
S.K. Moitra and Subhash Chand, Oil & Natural Gas Corporation Limited, India,
Santanu Barua, Deji Adenusi and Vikas Agrawal, Schlumberger Data & Consulting Services, India

Copyright 2007, Offshore Technology Conference

This paper was prepared for presentation at the 2007 Offshore Technology Conference held in
Houston, Texas, U.S.A., 30 April3 May 2007.
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, as
presented, have not been reviewed by the Offshore Technology Conference and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Offshore Technology Conference, its officers, or members. Papers presented at
OTC are subject to publication review by Sponsor Society Committees of the Offshore
Technology Conference. Electronic reproduction, distribution, or storage of any part of this
paper for commercial purposes 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 where and by whom the paper was presented. Write Librarian, OTC, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
To increase the profitability and field production an Integrated
Asset Modeling project was initiated. A high fidelity, fieldwide, Integrated Production Model (IPM), was developed for
Oil and Natural Gas Corporations (ONGC) giant offshore
field of Mumbai High. The network based IPM enables
superior engineering analysis and more efficient decisionmaking compared to classical single branch and nodal analysis
approaches. This is done by capturing, in the model, the effect
of all the network interactions that occur in the production
network.
Such a model helps to quickly identify and accurately quantify
bottlenecks and other opportunities to improve production.
Production improvements can be made by making daily or
weekly operational changes as in gas-lift-allocation, surface
backpressures, pipeline lineup etc as flow conditions change in
the field. The IPM model would also assist in reliably
evaluating field operational and re-development plans, and
thus expedite the engineering decision processes.
The project-plan also called for an Integrated Asset Model
(IAM) by integrating the IPM to reservoir simulation models
and to a suite of fit-for-purpose optimization tools. The IAM
model will then enable model based field-wide future
development projection studies and planning scenario studies
over time. Network based optimization such as network-gaslift optimization can be easily done to find the optimum
operation set points that maximize production without
violating user defined operational and capacity constraints.
The model will eventually evolve into an online IAM. Here

the On-Line utilities will effortlessly update data from the

application historian to the IAM with the latest or requested
data set. The historian stores recurrent data from field-wide
SCADA-DCS systems and other sources.
This paper describes the development of the IPM for ONGCs
giant offshore Mumbai High field and the technical and
project management challenges faced. The IPM includes the
production, water-injection, gas-lift supply and transportation
networks. The integrated model also includes simplified
models of processing facilities at each process complex.
Work to extend the IPM to an IAM is ongoing. Factors that
were found to be critical to the success of the project are
identified and discussed in this paper. This will provide a
better understanding and appreciation of the complex project
implementation issues and challenges that need to be resolved.
Introduction
ONGCs Mumbai High field is located 160 Km W-NW
offshore from Mumbai in the Western Coastal shelf of India,
with an aerial extent of 1200 sq km (Figure 1). The field was
discovered in 1974 and put on production in 1976; it is
currently on a pressure maintenance scheme using water
injection (since 1984). The field is currently in a decline phase
with problems like high water cut and high GOR (gas oil
ratio).
Mumbai High Field is divided into two major blocks, North
and South. The field is a multi-layered reservoir with several
exploitable zones of oil and some gas (Figure 2). At present oil
and gas is being produced through 5 processing complexes, 98
unmanned wellhead platforms and 2900kms of pipeline
network. There are about 650 oil and 40 gas producing strings,
and 200 water injection wells in Mumbai High Field. The field
is divided into two parts; Mumbai High North and Mumbai
High South (fig 3a and 3b) More than 80% of the oil
producers are on gas lift. Presently, around 250,000 bpd of
crude oil and 15 MMSCMD of gas are produced from the
Mumbai High Field (Figure 4).
The bulk production from unmanned wellhead platforms is
transported to one of the 5 process complexes (BHS, SHP,

ICP, NQO and SCA) through collector lines for separation,

conditioning and subsequent transport to shore. The bulk
fluids received in the process complexes are separated in two
stages, high pressure and low pressure, to separate crude oil,
natural gas and water. The separated and conditioned crude oil
is then metered and exported to Uran onshore terminal through
two major trunk lines. The gas separated at each process
platform is dehydrated, compressed and sent to onshore
terminal at Uran. A part of the gas is used for internal
consumption as fuel gas and as recycle gas lift operations. A
small quantity of low-pressure gas is flared for safe operation
of the field. Produced water is treated in the produced water
conditioner and then it is disposed off into the sea. For water
injection, sea water is treated prior to injection.
Project Execution
The IAM Project implementations challenges are
multidimensional
(a) Data required to build the field-wide IPM resides in
different departments and further distributed in the
administration units formed by each processing platform.
The data are not centralized and therefore not easily and
efficiently accessible. The large size and associated
complexity that comes with a brown field development
and surface facilities only exacerbates the difficulties.
(b) Integration of surface and sub-surface models implies that
data need to be collected from both these groups.
(c) Most importantly, the introduction of the On-Line IAM
into the organization implies alignment of new
technologies, work processes, workflows with people
within the organization to form a more efficient
engineering organization. Managing this change over the
transition period is a daunting but resolvable challenge.
For success of the project, it is essential to have a proper
project plan, provide sufficient strategic attention and planning
time, emphasis on coordination with relevant groups - office
and field engineers, operators and managers. Even though the
modeling environment will reside in the office, the field
support for such a model is critical to successfully realize the
true potential of the Integrated Asset Model (IAM). The field
is where the data collection begins and their sustained due
diligence and trust in office based engineering decisions is
critical to the success of the project over time. It is easier to
gain sustained conscientious field support for data collection
and reporting if these groups also derive appreciable tangible
benefits from the project. Potential derivable benefits that can
be motivating are
(a) Improved field work processes and workflows to make
their routine data collection and reporting work easier and less
tedious and risky.
(b) Provide engineering operational directives/guidance that
are accurate, timely and reliable based on modeling results
from the model.
(c) In case there is a need for modeling in the field-sites, make
the model easily available to the appropriately trained field
personnel for them to use.
It is critical that the model data collection efforts do not

OTC-18678-PP

become an additional burden to already busy and hectic

schedule in the field.
The project to build this modeling system was divided into
three distinct phases.
Phase 1: Offline Integrated Production Model for the entire
MH field (MH-IPM). With the IPM, it would be possible to
run modeling scenarios which can vary in scope -- field-wide
or processing-platform-wide or production-platform-wide or a
single well.
Phase 2: The IPM will be enhanced with optimization
functionality and made integration ready for coupling to
reservoir model(s) (ECLIPSE and/or tank models). With this
offline Integrated Asset Model (IAM), it would be possible to
optimize production, improve production operations decision
making and predict future scenarios to help in field
redevelopment plans.
Phase 3: The Offline IAM model will be positioned to run in
an On-line environment. Here, the latest recurrent data for the
model will be accessible via the computer network and the
model will be automatically updated with the required data.
An ever-green model of the field would then be available ondemand, boosting modeling efficiency, engineering study turn
around times and planning decision cycles several fold.
Extensive training and work association periods and field
visits were included in the project execution plan. The purpose
was to start training the ONGC engineers across various
groups so that they familiarize and begin taking ownership of
the models, as the models become available. This enables the
engineers to become involved and begin using the models
early and benefit from early model results as the project
progresses. Field visits helped in getting a clear understanding
of the current field work processes and drivers and to better
understand the work realities in the field. This helped in
identifying opportunities for stream-lining and standardizing
the data collection processes better. The IAM projects footprint on the current field work processes should be minimized
and if possible improved and standardized field-wide. Direct
tangible benefits need to be realized early during the IAM
implementation to effectively build the necessary support
groups of well co-coordinated, knowledgeable and motivated
IAM stakeholders throughout the organization before project
implementation phase comes to a close.
The Integrated Production Model
The Integrated Production Model (IPM) of Mumbai High field
is a comprehensive, high fidelity, rigorous multi-phase flow
blackoil model. The model includes wells, production network
and facilities, gas-lift delivery network, water injection system
and oil & gas transportation network which connects to the onshore terminal at Uran. The entire IPM was developed using
well and network modeling software PIPESIM.
The Mumbai-High field contains 5 processing complexes.
Each complex receives the production from wells connected to
it and is under its administration and management. The
separated gas is then compressed. Part of the compressed gas

OTC OTC-18678-PP

is used for gas-lift, as fuel in the offshore process-complex

utilities and the remainder is transported to shore for
consumption.
Regular engineering efforts have been ongoing to maximize
production by optimizing gas lift on a well by well basis. Most
of the time, results of the recommendations made on singlewell analysis basis did not deliver the gains as expected, at the
process complex level. This shortfall, which is due to the
interaction effects of adjacent wells and pipelines, could not be
analyzed in the absence of an integrated production network
model. This is a key issue in gas lifted fields, where the gas lift
gas is returned along with the produced fluids to the central
production facilities, adding to the back pressure on the
wells. This aspect of the behavior is naturally captured in a
network model
The Integrated Production Model for Mumbai High was
developed in an organized and phased manner so that the huge
amount of data and modeling work involved could be
managed efficiently and smoothly. Major phases were:
1.
2.
3.
4.
5.
6.

Data Collection and Validation

PVT Data Calibration
Well Model Construction and Calibration
Network Model construction (Production, Gas lift
and Water Injection)
Network and Well Model Integration.
History matching of the Model at different levels.

Data Collection and Validation

Since the field is structurally and otherwise a dynamic
environment it was essential that the model building and
validation should be done by matching model results to a
certain cut-off date instead of trying to match a moving target.
Additionally model building exercise for such a big brown
field is time consuming and complex. Therefore, it was
decided that the model would first be made as per the
conditions existing on target date to verify the model accuracy
and validate the model. Subsequently the model will be
updated to current conditions during the follow-up work
association period after Phase 1 is completed.
Data gathering is the first and foremost requirement of a
model building effort. For the physical skeletons of the well
models a massive data collection and data storage effort was
required for the 800+ wells. In order to ensure speed and
efficiency on the data gathering process, a detailed list of data
requirements was prepared upfront. A networking workshop
among various disciplines and groups was organized to ensure
clear understanding of data requirements, links and no
duplication. The data was manually collected from various
user groups within ONGC. Considerable project time and
effort was spent in interfacing with the respective groups,
persistently searching for required model building data. The
major types of data collected included
o

Production and Injection Plan as of cut-off target

date,

o
o
o
o
o
o
o

Well Location Maps,

Pressure Study,
PVT data and reports,
Flow Gradient Surveys (Pressure/ Temperature
surveys),
Latest Completions Diagrams, downhole equipment
details, deviation Survey of each well,
Production test history for each well,
Field Layout Diagrams and Pipeline Details, Process
Flow Schemes etc.

Before using the gathered data for modeling purposes, various

consistency checks were applied on each data set to confirm
the quality and accuracy of the data gathered. For missing/
inconsistent data engineering judgments and indirect methods
were used to generate the data.
PVT Data Calibration
Before well modeling work started, calibration of the fluid
properties was done by the software using laboratory PVT
data available for three major reservoirs viz. L-III South, L-III
North and L-II North. A multi-point calibration for all the
fluid properties was carried out for the three PVT Datasets and
following correlations were found to be the best match.
Lasater, for Solution gas
Standing, for Oil FVF
Beggs & Robinson, for Oil viscosity
Lee et Al, for Gas viscosity
Since, multipoint calibration data was available for each of the
three reservoirs, above mentioned correlations were calibrated
to match the available PVT data using calibration coefficients.
(Fig 5)
This step is a critical step because errors in PVT modeling
affect the multi-phase flow modeling results significantly and
can ultimately affect the fidelity of the model itself.
Well Model Construction and Calibration
To develop an accurate high fidelity model of the production
network, it is essential to model the wells correctly because it
is the relative well performance that largely determines the
network performance. In order to achieve higher degree of
accuracy, it was necessary to properly model, analyze and
calibrate the well performance at each wellhead before the
well is integrated to the network model. This was done by
constructing a valid stand-alone well model for each well.
Dual string wells were considered as two separate wells for the
purpose of modeling.
Well model building task was carried out in two phases
implemented in a relay workflow: (a) Well model skeleton
construction and (b) calibration of the model using available
pressure/temperature surveys and well tests and production
history.
Well Model Skeleton Construction: During the first phase,

basic skeleton of each well was developed. It was necessary to

carry out this separately because this involves careful
handling, scrutiny and validation of voluminous static data
that goes into making the well skeleton. Major data required
for this phase were reservoir description, PVT correlations,
well completion details, well deviation survey, downhole
equipment details, Choke etc.
At the completion, the reservoir inflow performance
relationship (IPR) was based on PI equations, using Vogel
corrections for flow below the bubble-point pressure.
Deviation survey data was used to define well trajectory in the
model. Flow in the well was modeled starting with the
reference point of flow entry in the well bore. The reference
point was taken as mid-perforation depth in well bore with
single/multiple perforation sets. In case of multilaterals, the
reference point was taken at the average of true vertical depths
of the different laterals. The horizontal well completions
consisted of completed cased sections followed by
uncompleted, unsupported open-hole perforated liner sections.
The reference point in such wells were at the end of the
completed section since the uncompleted sections were
inaccessible and therefore their conditions over time was
unknowable.
Downhole equipment modeling data include SSSV details
(Bean ID), ESP details (pump type, frequency of operation
etc), Gas Lift Valves details (Type, Model, Manufacturer, Port
size, PTRO values), and Gas Injection Point details (Valve
port diameter). The gas injection in the wells is modeled with
parameters like surface injection rate, operating surface
injection pressure, specific gravity and surface injection
temperature of the injected gas. Tubular were configured using
the latest available completion diagram. Internal diameter and
wall thickness data are defined as tubular specifications. Flow
type in the different sections is selected as
tubing/annular/combination flow, as applicable. Once the
skeleton of the model was completed, it was passed on to be
calibrated during second phase as described below.
Well Model Calibration: In the second phase, calibration of
the well modeled was done using the Well-test and production
history and pressure / temperature gradient surveys.
To preserve the inherent fidelity of the model, the proper
choice of calibration parameters is important. If an acceptable
match can be obtained by tuning the most uncertain input data
to the model, and these parameters are tuned within reasonable
physical ranges, the inherent fidelity of the model can be
preserved. For example, tuning pipe roughness or overall heat
transfer coefficient to get a better pressure drop and heat
transfer match is a fidelity preserving calibration option.
Tuning pipe diameter for situations where there is direct
evidence of diameter reduction due to scaling and/or wax
deposition will also preserve the model fidelity.
If a sufficiently close match cannot be obtained after this step,
black box calibration parameters such as efficiency and other
blanket correction factors may have to be used. Care should be
taken in the use of such parameters as it can lead to significant

OTC-18678-PP

loss in model fidelity if used unscrupulously. This can

seriously impact accuracy of future predictions as model
conditions change.
In this case, no black-box type tuning parameters were used as
an adequate match was obtained, which commensurate with
the data quality and quantity available. During the
identification of the most appropriate multi-phase flow
correlation model, special attention was paid to account for the
effect of flow angle in a pipe on the pressure drop predictions.
Since most of the wells were highly deviated, standard and
well respected vertical flow correlations failed to model the
wells adequately. From past modeling experience, it has been
experimentally observed and field verified that flow structure
and associated flow physics changes significantly as the flow
angle changes from vertical to deviated in the vicinity of 70
degrees flow angle from horizontal, other factors remaining
the same. When this was accounted for when choosing a
combination of correlations for a well, based on deviation
angle, a significant improvement was observed in the models
predictions capability for the deviated wells in the field when
matched with measured Flow-Gradient-Survey (FGS) data.
More specifically the Modified Hagedorn-Brown (Dun & Ros
Map) model or Ansari vertical upward flow correlation in
combination with the Beggs & Brill model or the semimechanistic Mukherjee-Brill model for inclined upward flow
was found to match the FGS data very well.
During this phase, if adequate FGS data was not available, a
detailed gas lift diagnostics was carried out by the software to
identify the point of gas injection in the string. This also
helped in identifying a number of wells producing suboptimally due to sub-optimal performance of gas lift system in
the well. Where adequate FGS data was available point(s) of
injection could easily be identified by inspecting plotted FGS
data. For each gas lifted well, two different well models were
developed. The first well model was configured for doing
detailed gas-lift analysis, design, diagnostics and troubleshooting. The second model was derived from the first model
and was configured to inject gas through a specified point
only. This well model was matched to well test production
data for verification and then integrated to the IPM network
model. From modeling point of view, the flow mechanics and
pressure profile inside the tubing remains the same as the first
well model as long as point of injection and gas injection rate
remains unchanged.
Finally, well models (oil and gas producers; water injectors)
were calibrated to the latest available measured production test
data, including lift gas injection rate and flowing wellhead
pressure, available as of cut-off date. Well performance was
evaluated using the system analysis approach, on the
production tests available over the last one year. For example:
for each production well test data acquired over last one year,
a combination of PI and reservoir pressure was generated for
respective liquid production rate, water cut, formation GOR,
gas lift injection rate and pressure (Figure 6). Assuming,
reservoir pressure relatively constant over last one year,
variation of PI of the well over time was analyzed to get better
understanding of well performance. A confidence level on the

OTC OTC-18678-PP

fidelity and accuracy of each well model was assigned by

taking into account the following factors:
Quality of match of reservoir pressure vs. PI of the
well using the production test data acquired over the
last one year.
Quality / consistency of the production data available.
Gas-lift performance evaluation.
A flowchart illustrating the well model building and
calibration process is presented in figure 7

Boundary condition considerations: Care must be taken to

ensure that each of these sub-network problems is well posed
in order to have a solution. The boundary node sinks (well
reservoir sinks) in the water injection network model were
specified as fixed pressure sinks since the reservoir pressure
and Injectivity (IPR) is a model given. For the producing well
source boundary nodes, the local reservoir pressure was
specified. The back pressure to the production networks were
the processing complex first stage separator pressures.
The gas flow networks consisted of two main parts.

When reservoir pressure and IPR was unavailable past

production test data sets were used to generate flowing
bottom-hole pressure versus flow rate trends. A non-linear
Excel Optimizer was then used to fit a Vogel corrected PI
curve through the selected points essentially a constrained
auto-regression curve fit. The curve fit yielded a reservoir
pressure and a PI consistent with measured data. (See figure 8)
Network Model
The Mumbai-High field contains 5 processing complexes.
Each complex receives the production from wells connected to
it and is under its administration and management. Gas-lift gas
from a given processing complex may be compressed and
delivered into wells belonging to another process complex.
Similarly, the water injection may cross Process complex
boundaries to wells under the purview of another Processing
complex. To preserve the flow and administrative structure in
the model and to simplify model navigation to various parts of
the model, the network layout was logically organized using
PIPESIMs folder option. The IPM model was built using the
following six layers architecture (from the top to the
bottom).This is illustrated in figure-9 and 10.
Layer 1: Interchangers Layer containing the MumbaiHigh production system and the water injection system
with platform-crossing fluid exchanges.
Layer 2: Trunk lines Layer with inter platform
connections and the transportation network.
Layer 3: Process Complex Layer containing the
production-separation processing -gas-lift network
interfaces.
Layer 4: Flow lines Network Layer containing flow lines
connecting to various wellhead platforms within a process
complex.
Layer 5: Wellhead Platform Layer containing various
wells in a wellhead platform.
Layer 6: Well Layer containing the detailed well model.
The separators at the process complex were modeled as a node
separator in IPM. A node separator enables the model to
separate the gas or water stream and re-inject it at a
downstream network location and continue to model the flow
of the separated streams. This technique is used to
conveniently integrate the multi-phase production network, the
gas distribution and transportation networks and the oil
transportation networks. (See figure 10c)

(a) The gas-lift distribution network where the boundary sink

nodes represent the well-head-platform gas-injection manifold
where a specified gas flow rate was delivered. Here the model
will calculate the manifold pressure setting required for the
specified flow rate.
(b)In the gas sales network portion, where the sink is at the
onshore terminal at Uran, the delivery pressure is specified in
the model.
The Integrated Network model that was constructed, included
all the wells, well head platforms, connecting flow lines
(active/ inactive), risers, down comers, and major process
equipments such as separators, compressors, heaters, pumps
etc. To predict the pressure and temperature losses in all the
multiphase flow lines, Beggs and Brill correlation was used. It
was found to be sufficiently good match when compared to the
limited observed pressure drop data available. Since snap-shot
surface flow line measured flow rate and pressure data was
unavailable at this time, further calibration efforts were not
expended at this stage. Similarly moody correlation was used
for single phase flow lines. Hagedorn-brown revised
correlation was used for vertical riser flow. Further tuning of
the models will be justified in the future when the necessary
measured data is expected to become available more readily.
Network and Well Model Integration
Once all the well models were calibrated, they were imported
into the network model within each well head platform,
providing a seamless integration between well and network
models. This makes the sixth layer of the layered architecture
of this model. Both the well models and surface network
models are solved simultaneously as an integrated network
model when solved at the IPM level. Since, each well is
allowed to have its own set of PVT model, multiphase
correlation and detailed completion details, results obtained
are more accurate when the model is solved at the network
level.
History Matching of the Model at different levels
History Matching is the process of getting a model to produce
Pressure, Temperature and rate values in the model that match
measured values at corresponding specific points in the field at
a given time. The idea is to make the model behave in the
same way that the field normally behaves, so that the field and
model should respond similarly to specific changes. This is
done by running the network model subject to specific

OTC-18678-PP

constraints. As PIPESIM models are steady state, therefore the

input data should ideally be from a system in steady-state
production. It is virtually impossible to get the snap-shot
steady state data manually for such a large and ever-changing
system, so trended data, individual branch data and
engineering judgment was used to match the IPM surface
components of the model.
History matching is very critical for this type of study and
requires careful analysis of the different aspects of behavior of
each object in the model, viz.: well performance, pressure drop
across pipes and flow lines, heat loss along main pipelines,
compressors performance, etc. For matching/calibration
purposes, the entire model was looked at, at the process
complex level and wellhead platform level. All the five
process complexes, gas lift network and water injection
network were matched to the production / injection data
available on the cut-off date. The predicted production
network flow rates with backpressure set at separators in the
process complexes matched well with the measured data (see
Table 1). Additional pressure and flow rate snapshot data at
different points in the network was unavailable for further
tuning of the model at this time.
Due to inadequate data and the inherent dynamic behavior of
the Gas lift distribution system as indicated by the available
data no further calibration effort was expended at this time. No
further calibration effort was expended at this time for the
water injection system due to unavailability of sufficient
adequate data. This was expected since available manifold
pressure data available was not snap shot data but distributed
over a long time period making the data inconsistent and
unsuitable for history matching purposes. Also, due to the
dynamic flow conditions in the gas-lift delivery network
indicated by the data and the production models severe
slugging analysis, it was futile to try and match the steadystate flow based gas-injection network model results to what
appears to be dynamic field data.
Note that in general, the network models for single phase fluid
are field proven and sufficiently accurate and robust with little
calibration if the data input into the model is correct. So the
model can be used to predict scenarios with good accuracy and
reliability. It was decided to further refine and calibrate the gas
and water injection network model at a later stage when
snapshot data at steady-state operations become available in
the future.
Major Findings from IPM
During the course of modeling building and validation, a
number of production issues were identified. They may be
broadly categorized into (See Table 2):

Production optimization opportunities for gas-lift

wells.
High back-pressure (bottlenecks)
Severe Slugging locations at Riser bases.

Wells with unusually low Productivity (Injectivity)

Index, which requires further investigation
Gas-lifted wells experiencing multiple point injection
Erosion risks (tubular integrity issues)

The IPM is being used by ONGC to study and analyze

alternate pipe lineups and flow connectivitys, resizing
pipelines and redesigning dysfunctional gas-lift well
installations.
Training and Work Association:
On completion of the model building and calibration phase the
model was delivered by Schlumberger, the technology
provider, and made accessible to ONGC engineers. A
comprehensive training on multi-phase flow and network
modeling using the modeling software was conducted soon
after final model and modeling report delivery.
Subsequent Phases and Challenges
Consistent with the production architecture of the field, the
field management and administration of the asset is organized
around the process complexes. Field engineers and managers
are assigned tasks and responsibilities based on the processing
complex they belong to. Field engineers within a group may
belong to surface or subsurface. Inter-complex issues are
handled through group meetings. Office based field-wide
engineers and managers are responsible for addressing asset
wide issues. ONGC is in the process of forming a dedicated
IAM modeling group who will be responsible for the models
upkeep and use. They will be responsible for maintaining an
ever-green model of the field production system which will be
made available on demand.
In the next phase (Phase 2), optimization functionality will be
added to the IPM network model. The primary optimization
target is gas-lift optimization where optimum and feasible gas
allocation is achieved based on the compressor capacities,
production network and gas-delivery network behavior.
Operational constraints will be specifiable. Optimization
studies can be done at various levels individual well level,
wellhead platform level, process complex level and field wide;
based on the need, using fit-for-purpose optimization tools
using linear-programming, non-linear programming and
Neural-network methods. Additionally, the model will be
made reservoir integration ready in an Integrated Asset
Modeling environment (AVOCET) so that ONGC will be able
to model and generate results for future planning and
development planning studies and analysis.
The sheer scale of the model poses a model maintenance
(model updating) challenge when done manually. For example
approx 10 wells are single-point production tested daily in
Mumbai High. So the well test data, on an average, is repeated
every three months. In addition, well work over and re-drilling
continue as per ONGCs drilling plans.
To address this high volume updating data issue the Integrated
Asset Model for Mumbai High field will evolve into a On-

OTC OTC-18678-PP

Line model during Phase 3 of the project where data update

and model maintenance are largely automated. This is done by
storing measured and entered data in Historians and
automatically feeding the model with the requested data on
demand. The field is currently in the middle of implementing a
field-wide SCADA system for data acquisition and monitoring
and storing. Data collected by the enterprise-wide SCADA
system and individual DCS systems and by other means made
accessible in the wide-area-computer network. The culled data
will be accessed by the model and this data will be used to
seamlessly update the IAM model. The model will be current
automatically, and ready for simulation and analysis.
Challenges: Phase 2 and phase 3 project activities have
begun. Several issues relating to modeling specifics for these
phases require further discussion, analysis and final
clarification between ONGC and the technology provider.
Discussions and clarification activities have started for the
integration of Reservoir simulation models with the IPM
model.
Design of the data and data base model and the on-line
modeling architecture is currently being designed and
evaluated. Well established standard commercial software is
available to build a historian and an automated updating
system.
Current workflows and processes vary, between processing
platforms. There is error prone duplication of effort for
recording and storing data. Greater interaction with production
engineers, field engineers and reservoir engineers have been
initiated and is on-going at this time to design better and
standardized workflows and processes prior to the Phase 3
implementation of the On-line IAM model. It is imperative
that the new work flows are properly aligned with the task
requirements and does not pose an additional burden on the
field engineers and operators.
As typical for any Brownfield, the greatest challenge is to get
ONGCs engineering and management organization to
efficiently move to a more modern and efficient production
modeling and analysis workflow environment and adapt to an
environment where all the relevant groups will have access to
all the models and data. When completed the IAM modeling
environment will enable all the related and interested groups
to work well together and be on the same page. A review of
such real-time optimization projects worldwide1 indicates that
current work processes for engineering analysis takes 80% of
the time to find and collect data and build models (nonproductive time) and only 20% is used for actual engineering
simulation and analysis (productive time). With a successful
online system the ratios are expected to reverse with 20% time
spent on model building and data collection efforts and 80%
will be available for simulation and engineering analysis. In
effect it is expected to boost engineering productivity.
To effect and generate support for such a wide ranging
change, efforts are underway today, to familiarize both
management and engineering with where the project is today
and where it is headed tomorrow and in the future and how it

fits in with the organizations overall engineering and

management vision. The transition cannot be imposed
suddenly but has to be done gradually over a transition period.
The expectation is, an early start would make the ONGC
Mumbai organization ready to welcome these changes by the
time the Phase 3 of the project delivery comes to a close.
Conclusions:
The following conclusions can be drawn from our experience
to date in implementing a large scale field wide Integrated
Asset Modeling project
1. Collecting data for a giant offshore brown-field and
building a 800+ well high fidelity network model can
be successfully done with current technology.
2. Calibration of the well models where relatively more
data (test data) was available (compared to surface
lines) is critical to building a accurate high fidelity
model.
3. Such field-wide Real-time modeling projects require
changes in an organizations workflows and
processes. Ensure these changes are not burdensome.
If possible the changes should be designed to make
things more rational and easier to build sustainable
field support.
4. Extra care and planning need to be expended to
address people, process and workflow issues to build
the company-wide support in order to keep the
project viable, even during the implementation
phases.
5. To insure the long term viability of the project and
use of the model, the model is dependent on the
sustained support and due diligence from several
allied groups in the organization. The model must
visibly show tangible benefits to all these groups
however small to have a naturally sustainable
commitment to support, maintain and manage the
model into the future.
Acknowledgments
The Authors wish to thank the management of Mumbai High
Asset, Oil and Natural Gas Corporation Ltd (ONGCL) and
Schlumberger for their encouragement and support during
project execution. We would also like to thank ONGC and
Schlumberger for permission to publish this paper.
A special thanks to the IAM project engineers in
Schlumberger and the ONGCL staff, from Mumbai High.
They provided extensive help in data collection, model
building and simulation.
References
1. Mochizuki, S., Saputelli, L.A., Kabir, C.S., Cramer,
R., Lochmann, M.J., Reese, R.D., Harms, L.K., Sisk,
C.D.,Hite,
J.R.,
Escorcia,
A.,
Real-Time
Optimization: Classification and Assessment,
November 2006, SPE Production & Operations,
Volume 21, Number 4

OTC-18678-PP

Figure-1 Location of Mumbai High

Existing Platform

Existing Producer

Old Location of ST Well

Existing water injector

Existing Duel Water Injector

Existing Duel Producer

New Platform

Figure-2: Mumbai High Stratigraphy

MUMBAI HIGH NORTH

WELL LOCATION MAP

z 7H

z 6H

New Producer

New Water Injector

New Duel Water Injector

New Duel Producer

z2

b2

2Z

z2

z3
5
z P1

z 6H

z P2

z6

z4

WI3

z1
TW1
WIN

z4

z3

z4

N2

b3

z2

z P1

z2

z1

z3

z3

z2

z1

N1

z1

 P5

z2

z3

 P2

 P2
z3

NU

z3

 P1

z8

5
 P2

NI

z9

8

9
z2

z1
z6

 P7
z2

z3

z4

2
z1

z5

 P3
3

z9

z4

b4
z4

NS

z5
b5

NX

5
z9

z7

 P5
z1
 P2

 P1

N11

TW2

7
z3

 P3 TW1
 P4

zB

z6

6
z5

z7

z8

z6

5 km
1 km

z8
z5
P1
z2

z3
zH
P2

TW1

1
z6

LD

z6

4

z P4
z3

NJ

z z1
z H1

N7

z1
5
z H2

zE

z2

z 2 z H3
4

 N12

N6

 P1
b7
z P2 z 7

11

TW1

P1
3
z 3 LB
z zH z 1 z 1 7
z9
z2
z
z P2
z2
z 5H
z4 2

z8

 P3
z7
z6
TW1
z P2
z1

z6
z8

z7
z6

3
5

z 6 NR
4
5

7 8H

 P1

z4

N5

 10
z4

z4

z6

 P4

z P5

z P8

z8

z 2Z
1

NV

5
z5

z7

z N10

 P6

 P4

NQ

9
z P1

z P6

z P7

z2
z1

z P3
NQO

z5

N9 z

NA
BHN

z4

z4

 P1

z4
z 6H

z7
6
9

 P3

4
z5 z4
z P2

NW

z3 z8

z4
4

5
3

NB

WI4

z P1
 BSS
z 1 z H-1
z P1
z9
zA
TW1
z P2
3

z P1
z2
z2
z P3
z 3 LC
LA
z4
7
1
7 z1
P2
z6
TW1
z8
z5
z 6H
z4
3
8
5

z6

TW2

TW1

NP

z5

8 z 1

9
z P2

z8 5

2X

z6

z5

z2

N4 z

z3

z1

NO

8

z3

z7

 P1

z 12

z 11
z1

z9

NC

z5

9

z3

b4

z8

z6
2

 P1

WI5

z7

z8
z6

z2

ND

z2

z2

N8

z7
z4

z1

N3

 z9

z 10

z5

8
z4

z9

TW1
b4
z3

z4

4
2

z6

WI2

z3

NE

NH

5
z 5Z

z1

z 5H

1 1Z

z2

z3

z4

LE

z2

z4
 8H
z1
 9H

2
z 8H

NT

z1

N o d al A nal ysi s G r o up

NK

Figure-3a: Mumbai High North Well Location Map

N o v- 0 1

OTC OTC-18678-PP

WELL LOCATION MAP

MUMBAI HIGH SOUTH

RS5
WA

RS4

WI6

RS1
ZB

RS2

BH-13

NL

SK

RS3

BH-45

BH-37

SI
SJ

IP

IU
NM

BH-10

IA

IS

IF

WI7

SC

IQ

EA
SN

RS11

RS7

BH-23

SB
SM

IG
BH-64

BH-16

RS8

IB
SQ

EB

SA
BH-24

SP

ZA
WI8

SD

S1-5

BH-36

BH-61

IC

SF
EC

IH

BH-6

SS
SE
WB

SR

WI9

S1-4
ID
SU

II

BH-35

SG

ED

ST
WI10
IJ
BH-26

SH
EE
SW
Exploratory Well
Platform

WI11

BH-5

S1-6

SV
IK

BH-11

IN
BH-52

IE
SY

Producer

IM

WI12
IW

Injector

IL
Released Well
BH-12

IT

Proposed Producer
BH-17

Proposed Injector
BH-21

Gas Well

BH-7

Figure-3b: Mumbai High South Well Location Map

Figure-4: Mumbai High Field layout

10

OTC-18678-PP

Figure-5: Multi-point PVT Calibration

Figure-6: Reservoir pressure vs PI chart

Reservoir Pressure Estimation (Well : XX)

Test Data
Calculated data

Flowing B ottomhole Pressure

2000.00

1500.00

1000.00

500.00

0.00
0.

500.

1000.

1500.

2000.

2500.

3000.

Liquid rate

Figure-7: Well Model building and calibration Workflow

Figure-9: IPM Model layers Structure

Figure-8: Reservoir pressure estimation using

Non-linear excel optimizer

Fig 10a: Layer 1: Exchangers Layer (IPM- MH)

3500.

400

OTC OTC-18678-PP

11

Fig 10b: Layer 2: Trunk lines Layer (IPM- MH)

Fig 10c: Layer 3: Process Complex Layer (IPM- MH)

Fig 10d: Layer 4: Flow line Network Layer (IPM- MH)

Fig 10e: Layer 5: Wellhead Platform Layer (IPM- MH)

Fig 10e: Layer 6: Well Layer (IPM- MH)

12

OTC-18678-PP

MPIP
Well Test
Process Complex Liquid Oil Water Liquid Oil Water Liquid
ICP
138,037 55,988 82,049 144,605 61,128 83,477 150,889
SCA
40,465 12,730 27,735 39,863 12,476 27,387 36,762
BHS
108,963 34,203 74,760 110,015 34,418 75,597 90,441
SHP
162,748 46,540 116,208 162,618 49,135 113,483 155,866
NQO
165,519 64,366 101,153 157,271 63,850 93,421 155,534
Total Field-BHN 656,197 226,557 429,640 654,235 233,483 420,752 626,254

DPR
Oil
58,501
9,898
33,600
49,687
72,511
234,095

Model
Water Liquid Oil Water
92,388 139,689 54,479 85,210
26,864 38,093 11,568 26,525
56,841 111,062 36,059 75,003
106,179 161,820 46,928 114,892
83,023 161,746 65,174 96,572
392,159 652,565 227,056 425,509

Table: 1a: Flow rate History Match results w.r.t Daily production report, MPIP and well tests data

Process Complex
ICP
SCA
BHS
SHP
NQO
Total Field-BHN

MPIP
Well Test
DPR
Liquid
Oil
Water Liquid
Oil
Water Liquid
Oil
1.2% -2.7% 3.9% -3.4% -10.9% 2.1% -7.4% -6.9%
-5.9% -9.1% -4.4% -4.4% -7.3% -3.1% 3.6% 16.9%
1.9% 5.4% 0.3% 1.0% 4.8% -0.8% 22.8% 7.3%
-0.6% 0.8% -1.1% -0.5% -4.5% 1.2% 3.8% -5.6%
-2.3% 1.3% -4.5% 2.8% 2.1% 3.4% 4.0% -10.1%
-0.6% 0.2% -1.0% -0.3% -2.8% 1.1% 4.2% -3.0%

Water
-7.8%
-1.3%
32.0%
8.2%
16.3%
8.5%

Table: 1b: Flow rate History Match Error (%) w.r.t Daily production report, MPIP and Well tests data
Wellhead
Platform*

Production
Optimization

Severe
Slugging

NQO
SHP
BHS
SCA
ICP
Total

9
2
1
4
25
41

10
9
5
3
27

Erosion
High
Risk
Backpressure
Number of Findings
7
1
24
1
9
1
1
1
22
3
63
7

Low
PI / II

Multi-Point
Injection

43
24
10
12
52
141

14
16
12
4
30
76

Table: 2: Summary of Production issues identified while modeling during phase1