The Optimizing Gas Management System
A development of MARACO, Inc.
GMAN is a unique set of computer programs
that analyze performance, and optimize development and operation, of gas
production facilities. The GMAN software provides rapid and realistic
projections of physical performance of wells, reservoirs and integrated surface
networks of gas production facilities. In addition GMAN combines such
projections with cash flow analysis using an efficient optimizing logic. Thus,
GMAN is the most fully capable tool available for planning, budgeting and
managing gas supply activities. Its convenient facility for inputting data and
the broad range of graphical and tabular output make GMAN an ideal tool for
managing all sizes of reservoirs, both old and new.
the small scale GMAN analyses individual wells. These analyses
2. effects of liquid production on vertical flow,
of size of production tubing and surface flowlines on deliverability,
and perforation strategies and
5. generation of IPR and tubing performance curves.
On the intermediate scale
GMAN evaluates drilling, production and compression strategies for individual or
groups of reservoirs in a physical, economic or optimal mode. A specially
tailored gas nodal analysis simulator considers flow between and from layered
reservoir blocks up the production tubing and through the surface flowline
network to a central delivery point. Production response to any size central
and/or booster compressor is determined. Evaluations may include impact of :
2. possible distribution of required spare capacity.
On the grand scale GMAN determines the optimal way for a group of reservoirs to supply a time-varying, seasonal demand vector against a possibly time-varying outlet backpressure. The simulator measures interactions between reservoirs in the flow network allowing GMAN to strike the optimal balance between wells and compressors (including timing and location of both). Here, the full scope of GMAN’s capabilities may be brought to bear:
flow segments with specified flow limits and/or chokes at
intermediate nodes (platforms or manifolds)
This realistic physical model is integrated
with GMAN's economic evaluation program, which translates price, cost and flow
rate projections into net revenues, taxes, royalties and economic yardsticks for
each possible investment in a well or a compressor. The physical and economic
calculations are performed in tandem under control of GMAN's optimizing routine
to determine optimal development and production strategies to meet offtake
requirements over a specified planning period. GMAN's scope and versatility is
exemplified by the range of questions that can be studied:
Benefits to engineers and managers are many, and most importantly include understanding of how critical economic and physical outcomes and tradeoffs may affect future business activities. Economic numbers compared to measured desirability and risk, are truly indicative, each being the optimum for the scenario it represents. Sound long-range plans, that guide each year’s budget and operation, can with GMAN be updated at any time to determine optimal response to new information or unexpected business developments. Optimal feedback management becomes an achievable goal. Thus, managers are able to steer the organization along the path of maximum profit and investment return.
GMAN’s TWO OPERATIONAL
1. As a gas
reservoir/trunkline simulator—Given the configuration of
producing formations in the reservoirs, the details of the production tubing,
surface flowlines, compression and separation facilities GMAN calculates the
physical performance of the entire system with an accuracy sufficient for
engineering design and performance assessment. The gas offtake specification for
simulations may be to deliver a targeted rate or to produce at capacity. GMANSIM
performs this function and, in addition, contains a program feature, GMANHST
that facilitates history matching of recorded reservoir performance. The
simulator mode contains an additional, multi-faceted reservoir analysis module,
GMANVFR. This module incorporates the one-, two- and three phase, vertical and
horizontal, flow correlation options available in GMANSIM in subprograms that
compute flow and pressure performance and graph results for:
2. As an optimizer—Given a desired offtake rate plus exogenous limits on
development and/or production, plus the simulation and economic input files,
GMANOPT systematically sorts through the many possible decisions to find the set
that satisfies offtake requirements and maximizes discounted cash flow. In this
mode the simulator and economic evaluator are used as subprograms to determine
the economic desirability of each possible investment decision.
GMANGRID, GMANDAT, GMANSIM, GMANHST, GMANVFR & GMANOPT
GMANGRID — is a program
module that generates a computing grid for a set of ‘well’ points (computing
cells) whose locations are specified by clicking a spot on the screen.
The well point may be dragged to another location if desired. A well point may
be specified as containing zero wells in GMANDAT. In this case the point is a
drainage cell whose hydrocarbons are produced from adjacent, interconnected
cells containing producing wells. In the general case the reservoir’s outer
boundary (green) is a polygon. A cell (red) is a polygon surrounding a well
point (circle). Using a tailored
algorithm GMANGRID draws the red
boundaries of all cells onscreen before the user’s eyes. The grid applies to
all formation layers.
on various objects onscreen brings out an input panel to specify physical data
– boundary segment ®
aquifer presence, size and permeability; well circle ®
vertical/horizontal well, permeability, thickness, porosity. Fault lines and
fracture designators may be inserted manually.
With the cells fitted into an x-y grid and the required input parameters
input, a click by the user causes GMANGRID to calculate interregional flow
coefficients, cell volumes and parameters and aquifer parameters needed for
simulation calculations. These are stored in a text file that is imported into
GMANDAT. Thus, GMANGRID relieves the user of the tedious manual task of
generating a computing grid that limits efficient use of reservoir simulation.
primary objective of GMANGRID, which is currently in the early beta-testing
stage, is to cause GMAN to conform to a reservoir engineer’s normal mode of
analysis, which starts with a study of individual wells. To this end a goal with
GMANGRID is to allow #grid cells ³ #wells.
support GMANGRID we plan to add two submodules, RESERVOIR WELL DATABASE (RWD)
and UNIVERSAL WELL MODULE (UWM). The goal is to put into RWD all well
data needed to support reservoir simulations and supporting well analyses.
Flexibility is a keynote characteristic of RWD. Slots to store additional data
items and input panels to accommodate them can be readily be created.
Likewise for links to other data sources. RWD’s prospective inventory
of data for a well is as follows:
spud date, surface/bottomhole coordinates
Ø list of well logs with link to source
lithology/formation characteristics, e.g., fractures,
pressures/rates from well
production rates or links
to data sources
completion details &
parameters of well’s flow model.
Planned output of UWM is
a flow model for each well that is ported into the simulator (via GMANDAT). The
core of UWM is a collection of computational objects that cover the full scope
of effects needed to simulate the behaviour of any well. Included are routines
vertical, horizontal and lateral wells
completion types including impact of formation
lithology and fracturing
effect of fluid content and composition including
impact of horizontal/vertical permeability ratio,
aggregating production string details to obtain a
compact model for tubing
chokes and surface completion effects
linking UMW to GMAN.VFR to calibrate flow models with
measured data from
is an easy-to-use, Windows-based facility for building a BDS
containing a database for making one or many runs with GMANSIM and the other
GMAN programs. GMANDAT’s Main Windows screen contains a toolbar and a menu bar
for calling out input data templates, tables and boxes for entering all required
data. But GMANDAT’s centerpiece is a graphics window, the Project Network
containing the network analysis diagram of the project in the BDS
file. All data entries are conveniently made by
invoking the data option lists and entry panels, directly from the PNS.
The diagram below shows the PNS when
it is first opened. A graphical layout of the entire model – reservoirs,
regions, flowlines, trunklines and connecting nodes some with compressors – is
easily generated in the PNS via
a series of mouse clicks and typing entries. Major models can be laid out in
minutes! As the diagram is constructed GMANDAT sets up the logic to control all
data input so that all that remains for the user to do is to click icons that
bring up data input panels and to type in the required data. GMANDAT contains
specially designed features that reduce typing to a minimum.
simulates performance of a gas production system. A basic tenet of GMANSIM is
that a full nodal analysis from the aquifer through the formation, production
tubing, surface flowlines and compressors must be performed to obtain an
accurate estimate of instantaneous gas deliverability. Reservoir pressure drives
deliverability, which feeds production, which causes reservoir pressure to fall.
It follows that to predict the declining profile of production over time, a
sequence of nodal analyses must be performed. The basic reservoir and pipe flow
equations are solved in each nodal analysis to determine deliverability. The
corresponding flows enter into material balance equations for the producing
layers to define the traces of reservoir pressures through time.
“GMANSIM uses a novel method of discretizing the flow equations that yield’s solutions rapidly. This speed makes it practical to imbed the gas simulator into the optimizer thereby supporting realistic evaluations of all investment alternatives. The addition of GMANGRID with the resultant smaller cell volumes necessitated a major reworking of GMANSIM’s cell routines to achieve smooth convergence to the correct solution.”
example opposite illustrates a simple two-reservoir, three-region GMANSIM
network simulation model.
Previously up to
(computing cells) could be used in each reservoir, but with the addition of
GMANGRID a larger number of computing cells is required. Although no absolute
upper limit for number of cells that can be handled with GMANSIM has been
established (The largest such grid tested to date contained approximately 50
cells.) an expected upper limit for efficient operation is 100.
Using fewer cells greatly reduces computing time in optimization runs. A few
years ago computing time was a serious consideration, but with modern day PC’s
that limitation has become relatively unimportant. Experience shows that for
most gas reservoirs the loss in accuracy resulting from using only a few grid
cells, rather than the large numbers often used in simulations reported in the
literature, is small. For example, for a moderately large, dry gas reservoir in
Australia, an unpublished comparison of decline in deliverability with
cumulative production predicted by a three-cell predecessor of GMANSIM and a
multi-cell high-end simulator model showed differences to be insignificant.
[The differences pale in comparison to errors resulting from improper
aggregation of layers with different permeabilities.]
This insensitivity to grid size arises because with the relatively small value
of gas viscosity; pressure gradients in a producing gas layer with permeability
greater than 0.5 md. are insignificant. [With
very tight formations the essential problem is to establish an economic well
spacing that yields timely and effective depletion of each drainage volume.]
Pressures differ greatly between layers,
but not within an individual layer. With nearly
uniform pressure a large, multi-cell model
is not required to determine that performance varies in direct correlation with
the average permeability used. The
experience cited in the previous paragraph notwithstanding, the issue here is
not to debate the relative technical merits of a sparsely gridded simulation
model. Instead, the situation is that reservoir models with few cells are used
in GMAN in order to reduce computing time to desired levels. These models
incorporate most of the fundamental equations that describe gas reservoir
behavior and have been demonstrated to give satisfactory results in many
instances. However, if results are available from more extensive simulation
uniform pressure a large, multi-cell model is not required to determine that performance varies in direct correlation with the average permeability used. The experience cited in the previous paragraph notwithstanding, the issue here is not to debate the relative technical merits of a sparsely gridded simulation model. Instead, the situation is that reservoir models with few cells are used in GMAN in order to reduce computing time to desired levels. These models incorporate most of the fundamental equations that describe gas reservoir behavior and have been demonstrated to give satisfactory results in many instances. However, if results are available from more extensive simulation models[which may account for complex effects such as spatial distribution of water influx, drop out of condensates in the formation, or compartmentalization of the producing formation by fractures], the required action is to calibrate GMANSIM’s models to accurately match the more detailed models. The input database for reservoir studies [which for upwards of 50 reservoirs can become very large] is conveniently managed by GMANDAT in a binary data system [BDS] file. Input data files for GMANSIM runs with all or only some of the reservoirs in the BDS file are obtained with only a few clicks of the mouse.
BACKOUT - New Feature
A new BACKOUT feature added to GMANSIM
provides an option to determine which reservoirs are shut in when flow is
throttled at the station to obtain a less-than-capacity production rate. (The
alternate option is to throttle each reservoir individually at the wellhead.)
With the station option low pressure reservoirs may be ‘backed out’ by flow
from a higher pressure reservoir.
GMANDAT & GMANSIM
work together synergistically to quickly provide solutions to a considerable
variety of gas reservoir/trunkline problems. GMANDAT creates and maintains the BDS
file containing the database and generates input SIM
files that are read by GMANSIM, which computes and
displays desired results. The most convenient way to operate is with each
module’s Main Window in view onscreen: Clicking the SIM button in GMANDAT
transfers control for a run with GMANSIM; clicking the DAT button in GMSNSIM
returns control to GMANDAT for changes in data or run type.
— the history matching module that performs three
1. Generates detailed graphical
presentations of results from a GMANSIM run.
GMANHST’s input for this output interpretation is a text [XLS]
file from GMANSIM
is formatted for convenient input into any spreadsheet program]. By
selecting any time step with a click of the cursor and then placing the cursor
over any facility [formation, wellbore, wellhead, flowline,
trunkline segment, compressor, etc.], pressure drop and flow rates in
that facility can be viewed. A right click of the cursor brings up
a pair of options that when clicked yield an
onscreen Windows graph of flow rate and
pressure vs time. Another right click in any graph’s window and hardcopy is
obtained. In this example starting from the top the four
pressures plotted for Region 1 in Reservoir 1 are pres,
The latter is pressure downstream of a wellhead choke.
2. Facilitates preparation of observed data files for history matching [HM] runs. Historical production rates from each producing region are required for a HM run. A separate HST file created by GMANHST contains these rates for each reservoir in the model. A Windows-based input interface is included in GMANHST. The data are entered into specially formatted Windows dialog boxes allowing visual inspection before writing to the file on disk. Onscreen plots of these data vs time [such as the example graphs below but with only the red data points] allow rapid detection of gross typing errors and the contents of any HST file can be called back onscreen for checking and editing. HST files for all reservoirs are input into GMANSIM in a HM run.
3. Generates displays with historical data points overlaying graphs of results computed by SIM in a HM run. GMANHST reads a binary [BIN] file output by GMANSIM plus all of the HST file s. The example shows pwf, pwh and qg vs t and p/Z vs cumulative production for Region 1. Plots such as these are keys for HM. Reservoir properties are adjusted so that differences between calculated and observed become smaller until a satisfactory model is obtained.
GMANIPR — generates an IPR curve [bottomhole flowing pressure vs instantaneous formation deliverability determined by GMANSIM for a specified formation pressure]. GMANIPR also calculates and plots tubing intake (TI) curve(s) [bottomhole flowing pressure vs tubing flow rate] for up to 5 different wellhead pressure(s). For the latter calculation tubing and fluid parameters are obtained from GMANSIM. If included in the GMANSIM input file, observed data points (pwh, pwf, qt) are also plotted on the output graph. The TI curves demonstrate the behavior of the tubing string and the combined plots show the operating point for the different well head pressures. Both Windows-based and Postscript graphs are available and hardcopy of each are readily obtained from the computer’s printer. This example is a Windows graph.
For an input wellhead pressure, pwh,
GMANVFO calculates equilibrium flow rates in the production tubing using a
selected vertical flow option (VFO) [GMANSIM includes 4]
for tubing intake pressures decreasing from just below pwh
to the flow point. Provision is made to input observed data points (pwh,
obtained from well tests. [If pwh
from pwh, pwf
be adjusted for the difference.] In the output graphs the input
points are superimposed on the calculated TI curve allowing “history
matching” to select the VFO and tubing parameters that best fit the
Windows-based and PostScript graphs are available and hardcopy of either is
GMANVFO has now been replaced with GMANVFR,
which is an extended version of the module that also allows the user to model
the subsurface to surface risers, etc. VFR
has a powerful interactive graphics interface that makes constructing a model a
relatively simple task.
— is a program that determines the optimal
development and production
schedule for one or more gas reservoirs
producing through single or multiple compressor stations(s). The optimal
solution specifies the schedule of drilling wells and adding HP over time to
meet a specified offtake rate. GMANSIM is used as a subprogram within GMANOPT.
An input file of economic data is prepared with GMANOPT’s Windows-based input
GMAN is the combined result of over 20
years extensive research, development and application by Dr E. L. Dougherty
(Professor Emeritus of Chemical & Petroleum Engineering, University of
Southern California) through his company Maraco, Inc., where he is ably assisted
by his partner Dr. Jincai Chang and their colleagues. Dr Dougherty has
specialized throughout his career in the study and application of large scale
engineering and economic computer systems for the oil and gas industry, and has
written numerous papers on the subject (see Publications). The practicality and
effectiveness of this approach to optimal planning is well demonstrated as
previous versions of the program have been in heavy use for over 18 years by
Santos Limited in Australia. This usage has now been extended to include
operators in Europe, USA, and the Middle East.
E.L, Juber, F. and Chang, J.: “Prediction and Analysis of Gas Wells Producing
from Reservoirs Containing Several Non-communicating Layers,” SPE 25908
presented at Rocky Mountain Regional and Low Permeability Reservoirs Symposium
(Denver) April 28,1993.
Dougherty, E.L., et. al.: “A Method for Simulating Pressure/Production
Performance of Volumetric Dry Gas Reservoirs,” JPT (Nov. 1985) 2059-70.
Dougherty, E.L., Lombardino, E., Zagalai, B., O’Donnell, B., Goode, P. &
Hollis, R.,”A New Systems Approach to Optimizing Investments in Gas Production
and Distribution”, Proc. 1983 Hydrocarbon Economics and Evaluation Symposium,
Dallas (March 1985) p. 33-48.
Dougherty, E.L., Lombardino, E., & Hutchinson, P., “Impact of Investment
Criteria, and Reservoir Characteristics, Taxes, Royalty and Price on Optimal
Plans”, Proc. 1985 Hydrocarbon Economics and Evaluation Symposium, Dallas
(March 1985) p. 21-32.
Dougherty, E.L., Lombardino, E., & Hutchinson, P. & Goode, P.A., “Use
of Mathematical Decomposition to Optimize Investments in Gas Production and
Distribution”. J. Pet. Tech. (Jan. 1986) p.70-84.
Dougherty, E.L., Dare, D., Lombardino, E., & Hutchinson, P., “Optimizing
SANTOS’s Gas Production and Processing Operations in Central Australia Using
the Decomposition Approach”, INTERFACES 17:1 (Jan/Feb 1987) p. 65-93.9
GMAN is designed for, and operates very
efficiently on a PC. Minimum requirements are a Pentium 1600 computer with a
minimum of 128 MB RAM, Windows NT, 2000 and XP operating systems.
MARACO, Inc. has successfully performed
consulting projects for major operators and governments throughout the world.
Major assignments include gas and oil field* optimizations in Holland,
Australia and Kuwait. Services offered include full project management, field
optimization and oil and gas simulation training, and contract software
*Oil field optimization is performed using
the GOMAN program, which was originally developed for the Kuwait Oil Company.
This program is now available as a commercial product and Maraco is seeking oil
company participation to further develop this software.
To download a copy of this information in a PDF format, right click here and save the file.
Send mail to firstname.lastname@example.org with
questions or comments about this web site.