|Volume 7, Issue 1 -
USING THE PAGOSA SIMULATOR TO STUDY OIL WELL PERFORATION
Bart Daly, John Hopson, Doug Kothe, Matt Maltrud, Martin Torrey, John Baumgardner, Rick Smith, Mike Hall, Tom Adams, John Bolstad, John Cerutti, Kathy Holian, Jay Mosso, Jim Painter, Larry Schwalbe, Tom Bennion, Gary Dilts, Ed Idar, Bob Shea, Jean Marshall, Ed Kober, Chuck Hansen, Paul Hinker, Los Alamos National Laboratory; Robert Ferrell, Dan Fraser, Wayne Weseloh, Thinking Machines Corporation; Dave Gardner, Courtenay Vaughn, Sandia National Laboratories
PAGOSA is a parallel modeling system for three-dimensional high-speed fluid flow and high-rate material deformation. The modeling capabilities of PAGOSA are illustrated with the simulation of an important oil industry problem known as oil well perforation. Well holes are typically lined with steel pipe and/or concrete casing that usually must be perforated with tiny high-explosive charges ("oil well perforators") prior to pumping. Perforation allows production of oil from specific depths predetermined from logging data. The perforators are inserted into the well hole inside "carrier tubes" and then detonated when the tube has been lowered to the prescribed depth. They are designed to make clean holes in the casing and to penetrate several inches outward into the surrounding oil-bearing strata. By modeling the perforation process, PAGOSA can be used to study perforator performance, i.e., hole size and penetration depth, as a function of perforator design and layout, tubing/casing geometry, rock formation, and other design parameters.
Consider two perforator charges aimed horizontally in opposite directions inside a steel carrier tube that has been inserted down-hole in a main oil well casing. The perforators are very similar to a current industrial design, with a conical copper liner surrounded by a high explosive material and a steel case. The carrier tube is positioned flush against one side of the well casing. Each charge is point- detonated at the apex of the high-explosive layer surrounding the conical copper liner. Energy released in the detonated explosive then causes the liners to converge and form shaped-charge jets that perforate the steel carrier tube and casing, and penetrate into the surrounding rock.
Inner diameters and thicknesses of the carrier tube and casing are 1.22/0.175 inches and 4.89/0.30 inches, respectively. The conical copper liners in the tiny perforators are 75 mils thick, having a base- to-apex height of 0.75 inch and a base radius of 0.34 inch. The oil- bearing rock is modeled as quartz because of its similar mass density, and the high explosive in the perforators is cyclotol. A simple elastic- plastic material response model is used for the copper and steel. All materials are assumed to be described by a Mie-Gr "uneisen equation of state," except for the perforator high explosive, which follows a JWL form.
The quantitative accuracy needed for this complex simulation is dictated in part by adequate resolution of all relevant time and length scales. This resolution places CPU speed and memory capacity requirements that would be excessively prohib- itive were it not for today's massively parallel supercomputers. The PAGOSA oil well perforator calculation was performed in 10.3 CPU hours on 512 processing nodes of the Thinking Machines CM-5 at Los Alamos National Laboratory. Reliable modeling of the jet formation process dictates a 0.5 mm zone dimension, and results in a total of 1.9 x 106 cells for the computational domain and a total memory requirement of 3 gigabytes.
Visualizing the data from the simulation proved to be a challenging task. Complex computational simulations can produce gigabytes of data, making it increasingly difficult to analyze the results that these simulations produce. To help solve this problem, surface extraction and optimization techniques have been developed to take advantage of the computational power of the massively parallel hardware available. The well-known iso-surface extraction algorithm called Marching Cubes was modified to run efficiently on both the CM-200 and CM-5 architectures. Benchmarks have shown that the parallel version of the algorithm is capable of extracting millions of polygons per second from three- dimensional volume data.
The evolving list of potential areas to where PAGOSA can be applied include compressible hydrodynamics, realistic equations of state, elastic-plastic material deformation, reactive burn of energetic materials, neutron transport, and turbulence effects. PAGOSA particularly has potential applications in modeling physical phenomena for problems important to industry, including crash simulations, machining processes, and impact scenarios such as the consequence of turbine blade failure in large electric generators.
Table of Contents