Use
of a novel simulator to reduce the front-end design time
to increase productivity in the petrochemical industry
Steve
Burchell, Cris Fells (BP), Frances
Collins, Helen Cook, Mark Elder, James Love and Andrew
Rooney (SIMUL8
Corp)
This project has made it possible
for Operation Research, in the form of simulation, to be
used on a much larger percentage of the occasions where it
can benefit the client. The client had been using simulation
to help with the design of petrochemical production systems
for many years but the process of using simulation had been
so time consuming that it was not possible to use it as much
as needed. In this paper we describe how we changed the use
of simulation at this client away from “major project” towards “nimble
calculator” that can quickly check ideas for the configuration
of huge production systems during the design process.
The client in this case is BP. The system
they need to design and optimise is the “production system” that
moves oil / gas from wells (typically undersea) to the point
where it either reaches one of their refineries, or some point
where it boards a ship taking it to a refinery or bulk customer.
This system is constrained by the
reliability of a wide variety of equipment (pumps, compressors
etc) that will typically be installed in very severe environments
and used over several decades of the life of the oil or gas
field. There are often thousands of these items, each of
which may have many potential causes of failure. This equipment
needs to be maintained (so it can be out of commission for
planned periods) and it may not be required for the entire
life of the field. The total cost of the equipment varies
by field size but is commonly more than a billion dollars.
The problem is complicated further many interacting rules
and regulations about how the equipment can be operated.
The impact of a breakdown of a piece of
equipment is different depending on the current oil or gas
flow rate. This varies over the age of the field as the content
of the field reservoir starts to decline. Alternatively, an
equipment breakdown may sometimes have no impact (because some
other item has already stopped flow in that part of the system
or because spare equipment is available to take over the work).
Hence the process to be modelled is very
complex and simulation is an ideal weapon. The client was aware
of this and had been using simulation on some of their cases
for many year. However the process used to create the simulations
meant that many weeks were required to build and validate each
model, and many hours were required to run them. Each time
a model was created a standard, generic, simulation software
was used.
The implemented methodology described in
this paper uses a domain specific discrete event simulator
created on top of a commercially available generic simulation
product. By creating a series of drag-and drop components,
most of which represent each of the different types of equipment,
it has been possible to reduce model build and validate time
to a few hours (for very large-scale models) and run time to
a few seconds. Building a simulation from components means
that all the (simulation-related) thinking work has been done
by the time the building of the actual model starts.
This radically increased speed means many
benefits to the client including; more cases analysed, better
results because more scenarios for each case can be examined,
more accurate results because longer/more runs are made, better
results because non-simulation skilled design engineers can
do the simulation work at the same time as considering other
design factors (like oil flow dynamics), more confidence in
results because validation can be done on the components ahead
of the model creation, more comparable results because all
models start with a standard set of reliability data embedded
in the components.
In the full paper we describe the project
to create this simulator, how it is now being used and show
the simulator in action. |
