SPE 156798

Integrated Production Optimization and Surface Facilities Management

through Advanced Optimization Techniques
Viadana Giorgio, eni e&p
Albani Danilo, eni e&p
Distaso Marco, eni e&p
Sharon Almatasem, eni e&p

Copyright 2012, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE International Production and Operations Conference and Exhibition held in Doha Qatar, 14–16 May 2012.

This paper was selected for presentation by an SPE 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 Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers 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 SPE copyright.

Abstract
The Production Optimization is one of the most complex and multi-disciplinary task in the oil&gas industry from an
operational point of view: it involves the surface asset during all its production life and requires a continuous improvement
process. In the gathering facilities and process plant engineering design, the optimization is driven by inputs that are subjected,
during the asset life, to changes; this fact, coupled to improvements and modifications in surface facilities, creates the
necessity to manage and optimize production scenarios with a more frequent time-frame. Technology improvements have
enabled a widespread use of integrated simulation models for a better asset management to be fully combined with measured
field data. In this paper we present a dedicated workflow for surface facilities – gathering system and process plant –
production enhancement and management coupled with an advanced optimization technique based on a powerful algorithm.
The main feature of this algorithm – and consequently of the proposed workflow – is the ability to control many variables
simultaneously according to the system constraints even with complex, conflicting and non-direct interconnections between
them and the objective to be reached. It has been proven that this algorithm has a robust practice that detects the global
optimum of the feasible area avoiding a premature stop in a local optimum region, a situation quite common in highly-
constrained and non-linear optimization problems in oil&gas industry. The technical contributions of the work are the abilities
to support operations in: definition of field potential, production optimization and de-bottlenecking activities. In the paper a
case study, based on a highly-constrained field with gas treating limitations, is reported and the proposed workflow’s results
are compared with another integrated optimization available from other software, showing an improved ability to detect global
optimum combining well management to the plant capabilities.

Introduction
In this paper a model-based oil field management workflow for complex surface system is presented. Oil & gas field
management and operating optimization is a complex task because of many factors: a multi-component fluid production,
hazard and sourness of released gases, complexity in determine some interconnection between operating variables and
operating limitations, several and sudden temporary modifications. The main feature of the workflow proposed is the ability to
control, by models, many variables simultaneously together with operational constraints and to detect optimized scenarios
thanks to a proven robust optimum searching mechanism.
Two macro areas characterize the production chain system: 1) the gathering system and 2) the process plant. The
optimization and operational management of this oil field is a multidisciplinary task because of the bidirectional influence
between control degrees of freedom and constraints that affect the entire surface system composed by environments ruled by
different laws. In particular, for the gathering system the main principles are related to the fluid mechanic: the reservoir
pressure drives the fluids to the surface and then by pipelines to a delivery or process area; artificial lift methods could be
applied between the pressure source and the delivery point in order to add pressure energy to the system. For the process plant,
necessary to meet the sale specification for the fluids, the main principles are physical and chemical, often non-linear. Figure 1
represents schematically the distinction between these two environments.
This macro categorization, at first sight, seems minimize the required know-how for the surface-production operations and
management, but the number of constraints, interactions and limitations make complex even the comprehension of the entire
2 SPE 156798

system. Several tools exist to support the operator’s decisions, and their use is spreading thanks to the even more difficult
situations encountered in this industry; nevertheless, the tools often remain segregated or managed by different discipline
experts, loosing in overall view-point and considerations. An integrated approach is proposed by the workflow developed in
eni e&p production department, called rabbit™ - Risked Algorithms for Biogenetical Balance Integration Tool*. This
workflow allows at the same time the integration between common industry software and optimization of the global
environment composed by different models.

Figure 1

Case Study
The case study on which has been applied the workflow proposed by this paper is a production train of a complex on-shore oil
field, that sales stabilized oil and associated gas treated in order to match the sale conditions.

Gathering system
The multiphase fluid production, composed by oil, associated gas and water, is gathered by a pipeline network to a process
train. This gathering system collects fluids from seven wells to the process plant inlet manifold. The production fluids have
different characteristics in terms of water cut (WC), gas oil ration (GOR) and sourness of the gases. Table 1 reports these
values. The pipelines routes are affected by several uphill and downhill due to the orographical area characterization. Artificial
lift methods are present in the system: a twin-screw multiphase pump that collects the production of four wells and a well in
gas lift. Two main pipelines carry fluid to the same production manifold. Wells productions are controlled by chokes,
multiphase pump rotating speed and arrival pressure; gas lift system has small control and degree of freedom. No significant
constraints affect directly the gathering system.

Figure 2

* Note: the usage of the rabbit™ trade mark name in this document is not for commercial purposes, only for informative opportunity
SPE 156798 3

Table 1 Main properties of produced fluids

GOR WC %CO2 %H2S
Well 3 3
[Sm /Sm ] [%] In gas In gas
W1 673 1.3 22.19 2.12
W2 800 0.2 24.83 2.1
W3 198 0 4.67 0.84
W4 190 0.3 4.6 0.81
W5 201 1.4 4.62 0.8
W6 299 0 10.85 0.25
W7 310 0 16.34 0.27

Process Plant
The process plant has several production trains that treat oil and gas of the entire field production. The process train
dedicated for the fluids gathered by the pipelines has been projected to separate oil, water and gas, to stabilize oil and treat the
gas in order to match dew point conditions (water and hydrocarbon), to remove sour gases (CO2 and H2S) and compress it to
export pipeline pressure conditions. The separation train is composed by a slug-catcher and two three-phase separators
operated at subsequent lower pressure conditions. A heat exchanger exists between the two separators in case the arrival oil is
too cold. After the separation the oil is sent to the stabilization tower, equipped with a lateral separator and lateral heat
exchanger. Reboiler is fed by process steam. Gases released from low pressure separation and stabilization are compressed and
collected with high pressure gases from other separation stages and then sent to the gas treating section. Gas, characterized by
a high sour content, is sent to an amine contact column, a glycol contact column, a HC-dew point control composed by heat
exchangers and finally compressed to export system. The process system is complex and mainly characterized by bottle-necks
in gas treating section. In fact, the oil separation and stabilization system have spare capacity while the gas section is close to
the overdesign limit. Figure 3 represents schematically the process.

Figure 3

Integrated system management

Because of the highly-constrained conditions found in process plant and a partial flexibility encountered in the gathering
system, the better way to manage, operate and optimize this surface asset is to maintain a global view-point of the situation,
integrating gathering system actions with plant considerations and verifications.
It is easy to understand the difficulty in finding improvements in a system with a bottlenecked process plant in gas section.
Several actions that could be taken in well management to increase oil production have significant impact and limitations in
gas treating section: a choke-up action, or an increase in multiphase pump boosted flow rate, allows increasing in fluid
production, both liquid and gas, making sometime complex the production management. This management is focused in
maximizing oil production and at the same time reducing the GOR of the blend incoming the plant. This task results difficult
because it depends on wells’ behaviour (GOR, productivity, well-head pressure), gathering system pressure drops, interaction
between wells and multiphase pump; furthermore, some flow assurance issues on the pipelines must be taken into account (e.g.
the magnitude of arrival slugs).
Other degrees of freedom are present in the process plant; the most significant are: separators pressures, column reboiler
temperature and column middle temperature. Those variables are strictly involved in oil-gas flash separation (the higher the
separation pressure, the lower the gas released), backpressure in gathering system (the higher the manifold pressure, the higher
the backpressure in flowline) and oil specification (the higher the reboiler temperature, the lower the True Vapour Pressure).
Conflicts and interconnection between variables, constraints and operational limitation must be balanced and solved by an
optimization frame-work with a global sight.

* Note: the usage of the rabbit™ trade mark name in this document is not for commercial purposes, only for informative opportunity
4 SPE 156798

Model-based field management

Simulation software and models help in managing, optimize and control system in several industries. Oil & gas industry uses
these tools to overcome the high complexity of the situations involved in project, optimize and operate such type of systems.
Gathering system and process plant behaviours have been modelled by simulation software and tuned with real field data.

Modelling of the system

The gathering system is modelled by commercial software for network modelling, commonly used in the eni e&p
divisions. In this program, each well is described through specifications for the reservoir fluid extracted, the performance,
layout and length of the perforated well and the length of the tubing: the combination of these characteristics defines the well
production performance to the surface, controlled by the choke valve regulation. In order to solve the network, an end pressure
point, the separator pressure, must be set. Running the simulation, the program returns oil, gas and water produced by each
well, the pressures in each node and the pressure losses in the pipelines. The fluid characterization used for this software is
black oil, considered sufficiently satisfactory for pressure drop calculations.
The process plant environment is modelled by another commercial software equipped with several thermodynamic
packages and several models of the operation units like separators, column, absorber, reboilers and heat-exchangers. The
accuracy of this program is high, but the program management is not straightforward. This tool is one of the most spread
simulation programs used in the oil&gas and chemical industries for process control. It has also logical controls and adjusting
operators that give some automation to the simulation. A compositional fluid characterization is required for process plant
simulation due to chemical-physical principles involvement in this modelling.
Each simulation program used has an internal optimization tool, but separated from the constraints of the other
environment; in particular, the optimization of production from a network point of view cannot account for the problems of the
treating section of the process plant but can only include general considerations (e.g., maximum gas rate, maximum H 2S). On
the other hand, the successful constraints’ satisfaction in the process plant is highly dependent on the input hydrocarbon mass
flow rate, provided by the network environment. The tight interaction between the two systems requires a tool which optimally
integrates the two environments with a controlled view of the entire asset. In this work, an integrated frame-work by rabbit™
is developed for the optimization of the global system; the internal algorithm searches for the optimal variables setting through
a robust evolutionary strategy, guaranteeing a global optimization and a strict constraints satisfaction, avoiding local optimum.
The control performed by rabbit™ on the other software is possible thanks to a pre-imposed collection of variables, constraints
and results in dedicated spreadsheets.
The eni e&p integrated optimization is compared with an optimization made by a third commercial software, released by
the same software-house of the gathering system simulation, called for simplicity from here on Integration Software IS. This
third program allows as rabbit™ the integration between different software both from reservoir, wells, surface facilities and
process plant. Then it could be used to integrate gathering system model with process plant model and optimize the entire
surface asset with its own internal optimization technique, based on a non-linear programming algorithm.
The simulation and results found for both the system are in steady state conditions and with fixed parameters such as
compositions, GOR, WC, wells’ productivities, reservoir pressures and characteristics. The forecast optimization is not
contemplated for operational needs and demanded to other software. This choice is based on high uncertainties on main
features of the proposed work-flow: the tight interaction between the two environments is based on well known fluid
properties that allow a quite precise transition between black oil characterization for gathering system and compositional
composition for process plant.

Operating controls, constraints and objective of the system

The optimization is set up in order to optimize the stabilized oil production satisfying all the existing constraints.
Degrees of freedom are twelve:
Six inlet choke imposed pressure losses (equivalent at choke opening)
o Multiphase pump suction pressure
o Manifold pressure at the process plant
o Slug catcher pressure
o Second three-phase separator pressure
o Stabilizer reboiler temperature
o Temperature at the middle of the stabilizer
The arrival pressure at the manifold is a variable shared between gathering system and process plant software, because it
constitutes their conjunction. In the integrated optimization, the separator pressures become variables, whereas in the separated
one these values are a user-defined necessary condition. Ranges for the variables are reported in Table 3.

* Note: the usage of the rabbit™ trade mark name in this document is not for commercial purposes, only for informative opportunity
SPE 156798 5

Table 2 Variables’ boundaries

Lower Upper Unit
Choke ∆P ±30%
MPP suction 40/35 50 [bar]
Manifold pressure 37 38 [bar]
SC pressure 33 34 [bar]
LP pressure 9 15 [bar]
Stab reboiler T 160 180 [°C]
Stab middle T 80 90 [°C]

Two case studies have been investigated with two different lower boundary for multiphase pump suction pressure. Case 1
has minimum suction pressure at 40 bar while Case 2 allows a major boosting down to 35 bar. These values have been chosen
investigating the pump datasheets.
Choke pressure losses variables are ranged around the actual values in order to reduce the boundary conditions around the
physical capability of the chokes and the wells’ known behaviour. This type of variable could not be considered absolute:
being a pressure loss, it depends directly to downstream pressure, that in turn depends on flow rate (well productivity) and
final network pressure (manifold pressure). The well sustained by gas lift is not considered in the optimization frame-work
because the gas lift system has no rooms of improvements.
The constraints of the system, as previously anticipated, are mainly referred to the process plant. By design specification,
the plant has maximum oil, gas and water capabilities defined as inlet values. Besides that, a significant constraint is the gas
flow sent to the treating section, set up to current maximum gas treated, already in overdesign; other constraints that affect the
acid removal section are the ratio and the sum between the CO2 and H2S fraction of these gases: beyond the specified values,
the unit does not work as required, since the high amount of CO2 degrades the solvent action in the H2S removal. Another
constraint is the TVP specification: it depends on the composition of the inlet fluids and the stabilized reboiler temperature.
Table 5 resumes the plant constraints. Same controls and constraints are used for IS approach.

Table 3 Constraints for the plant

Value Unit
Inlet oil [max] 350 [m3/h]
Inlet gas [max] 85400 [Sm3/h]
Inlet water [max] 92 [m3/h]
Gas flow rate to the
68000 [Sm3/h]
treating section [max]
CO2/H2S to the
21 [-]
treating section [max]
CO2+H2S to the
20 [%]
Treating section [max]
TVP @ 38°C [max] 0.86 [bar]

* Note: the usage of the rabbit™ trade mark name in this document is not for commercial purposes, only for informative opportunity
6 SPE 156798

Results
eni e&p integrated optimization has been used to determine global maximum for Case 1 and Case 2 described in section 3.
Table 4 reports reference case and optimized scenarios.

Table 4 Reference case and results of rabbit™ optimization

Reference Case 1 Case 2
FWHP FWHP FWHP
Well Name [bara] [bara] [bara]
W1 109 122.4 128.7
W2 82.3 75.2 71.2
W3 65 61.6 57.3
W4 lift lift lift
W5 66.6 62.0 59.5
W6 53 49.0 43.6
W7 74.6 67.8 66.3
MPP [bara] 41.0 40.0 35.0
Manifold [bara] 37.0 37.0 37.0
P SC [bara] 34.0 34.0 34.0
P 2° sep [bara] 11.5 9.0 10.8
T reboiler [°C] 161.5 160 160
T h-ex [°C] 84 82 86
Olio Stock-tank [m3/h] 216 220 225.5
Delta prod [bopd] - 600 1400

As could be seen from Table 4, the optimized scenarios require, as expected, a decreasing in multiphase pump suction
pressure. So doing, some well reduces the well-head pressures encountering a lower backpressure. In turn, in order to satisfy
the constraints imposed to the system, which the main one is the gas released sent to the gas treating section, W1 requires in
both cases a consistent choke down. For Case 1 and Case 2, significant increasing are expected, respectively 600 and 1400
bopd. Obviously the multiphase pump plays a major role in the gathering system management and optimization; in fact, the
lower the suction pressure, the higher the production gain expected. In both results for Case 1 and 2, the gas constraint is
almost saturated with a value close to the maximum allowed reported in Table 3.
In order to compare the two integrated optimization techniques presented, several runs have been performed for Case 1
both with rabbit™ and IS with the same gathering system and process plant models.

Figure 4

As could be seen from Figure 4, rabbit™ results remain higher respect several IS runs. IS reaches similar results but with
many runs driven by the user managing the variable ranges. In that case an increased level of knowledge and critical
interpretation of the system is required by the user. In fact, the non-linearity of the system and the conflict between objective
function (oil production) and constraint (gas released) makes complex the optimum search, causing premature stagnation of
the IS optimization algorithm. The high variability of the IS results, respect to the rabbit™ ones (only six runs have been
performed to verify the goodness of them), is directly related to the multiphase pump control: the first software algorithm

* Note: the usage of the rabbit™ trade mark name in this document is not for commercial purposes, only for informative opportunity
SPE 156798 7

reflects difficulty in finding the optimized value keeping higher the pump suction pressure because of the high relationship
between this variable, the objective and the gas constraint.
The only drawback of rabbit™ optimization is the required time for a run (approximately four hours), higher respect a IS
run (30÷40 minutes), but considered acceptable looking at the reliability of the results.

Conclusions
A model-based tool has been developed for field management and operation considering surface facilities such as gathering
system, artificial lift methods and process plant. Integration between these systems has been allowed by eni e&p software
based on an operational philosophy.
A case study characterized by a complex optimization scenario has been selected for the proposed work-flow. Variables,
constraints and items are taken and selected considering real behaviour of the field. Models have been tuned and used to
manage and optimize field production. The optimization evaluates better production scenarios controlling wells’ choke
opening, multiphase pump boost and plant main pressures and temperatures. The expected production gains for the variables
considered are estimated between 1.8 and 4.4% respect to the current production. The integrated optimization proposed by eni
e&p production department has been compared with results obtained with another software available on the market. Results
from this second approach return similar values and scenarios only after a complex and user-driven optimization step-search.
The optimization technique adopted for rabbit™ is then considered satisfactory, robust and reliable, affected by low variability
and with better performances respect to the other one tested.

Acknowledgments
All the authors want to acknowledge the PROD department in eni e&p and field operators for the support received for this
work.

References

[1] Petroleum Production Engineering, Boyun Guo, William C. Lyons, Ali Ghalambor, Elsevier 2008
[2] Sistemi di pompaggio multifase: applicazione al campo di Wafa, Fabio Turconi, master thesis, Politecnico di Milano 2011
[3] Surface Operations in Petroleum Production I, George V. Chilingarian, John O. Robertson, Sanjay Kumar, Elsevier 1987
[4] Surface Operations in Petroleum Production II, George V. Chilingarian, John O. Robertson, Sanjay Kumar, Elsevier 1989

* Note: the usage of the rabbit™ trade mark name in this document is not for commercial purposes, only for informative opportunity