GRAPHICAL PROGRAMMING AUTOMATES PRODUCTION RESEARCH WORK STATIONS

David B. Burnett
BP Research
Houston

Recent conversion to graphic programming allows BP Research's Westport laboratory, Houston, better control of experiments investigating oil and gas production problems.

Graphical programming gives more consistent and accurate experimental results, with fewer mistakes and repetitions. The software requires less operator attention and lessens errors in reading and analyzing data.

For process model apparatus, this software eliminates the long-standing gap between the state of the art and actual practice in using computers.

Custom programs that previously were written in cumbersome general-purpose languages have been replaced by modular languages that provide more functions and shorter development times.

WELL BORE MODEL

One application of graphic programming at Westport is a well bore model for studying well completion practices, such as the use of lost circulation materials (LCMs). In the model, different LCMs can be tested by various methods under conditions much like those in a real well.

As shown in Figs. 1 and 2, the apparatus centers on a 2 ft section of 7-in. casing. Different layers of the producing formation are simulated by four cylindrical core samples mounted in special holders.

Well perforation tunnels are simulated by 3/4-in. diameter holes running through the casing wall and dead-ending in the cores. Thermostatic mantles heat the cores and casing to the downhole temperature.

The system includes a variable-speed pump, remotely adjustable back pressure regulators, flow and pressure transmitters, remotely controlled valves, and four storage tanks. Three of the tanks are heated by submerged thermostatic elements.

All this equipment is mounted on two rolling stands connected by high pressure, high-temperature hoses (Fig. 3).

When an experimental run begins, the stands are rolled into a large, ventilated fume hood. In the start of a typical run, brine in one of the heated tanks is circulated through the cores, while temperatures and pressures are brought up to the desired levels.

Then an LCM slurry might be pumped in from another tank, following a prescribed pattern of flow rates and pressures. Other materials may be introduced, such as acid for removing LCM residue.

By switching of valves, flow is reversed and oil production measured with flow rate monitored through each core. Afterwards, the data are analyzed to calculate core permeability and determine possible formation damage.

AUTOMATING THE MODEL

The well bore model is representative of about half a dozen kinds of research stations in use at the Westport laboratory.

Before computerization, as in other companies since the 1930s, the only automation consisted of manually adjusted process controls such as pressure regulators and thermostatic heater controls. Data were collected by strip-chart recorders and manual logging of instrument readings.

For computer analysis and presentation, data had to be keyed in manually.

In most production research laboratories, computer control and data acquisition had been an expensive luxury, rarely available to the average researcher. Software was typically developed with a general-purpose, text-based language, incorporating assembly language modules to process any especially high-speed data.

One can easily imagine the shortcomings of such methods. Development costs were high, the resulting systems were inflexible, and scientists faced nonuniformity among the many workstations within a company. In 1987, a consultant started to investigate possible changes at Sohio's production laboratory in Dallas. In late 1988, the laboratory was moved to Houston after British Petroleum acquired Sohio.

Apple Computer Inc.'s Macintosh was chosen as a standard for model automation as well as for general office use at the Westport laboratory. The Mac provides an efficient and appealing user-interface style.

A search for laboratory-oriented control and data acquisition packages running on the Mac turned up two candidates. Labview software from National Instruments Corp. was adopted as a standard for the Westport laboratory. The software uses a super-high-level language designed specifically for purposes that match a production laboratory's requirements.

National Instruments also provides a wide selection of I/O (input/output) boards for the Mac. The graphics includes software drivers for interfacing the computer with various laboratory instruments through those boards.

Late in 1987, eight "ruggedized" AI-90 Macintosh II systems were purchased for automating existing laboratory models. Additional nonruggedized Mac II and IIx systems have been added since then, expanding the number of models automated to 11. Each computer has 5 megabytes of memory, 100-megabyte hard disk, 3 1/2-in. diskette drive, and 13-in. color monitor.

For I/O purposes, each contains four National Instruments circuit boards. Two NB-MIO-16 boards provide a total of 32 analog inputs, four analog outputs, 16 discrete I/O lines, and six independent 16-bit counter/timers. An additional NBDIO-32 board carries 32 discrete I/O lines, and an NB-GPIB board allows communication with IEEE-488-compatible instruments.

One of the keys to success of the automation program has been the small team of specialists working together to develop all of the laboratory's work stations. The team consists of a consultant programmer, an electronics technician, and a few key scientists.

Using off-the-shelf hardware, they designed all of the automated systems to meet the users' needs. The result is that Westport now has shorter development times, lower automation costs, and simplified maintenance programs.

LEARNING CURVE

In contrast to text-based, general-purpose programming languages, which typically involved a hefty learning curve, the consultant programmed the graphical software after attending just one introductory 3-day training course. Because a graphical language allows program building on a type of flow sheet or block diagram on the screen, programs can often be planned without pencil and paper.

The blocks representing program elements in the diagram are pictorial icons known as "virtual instruments" (VIs) chosen from a library list. Besides using standard Vis, software engineers can create their own and add them to the library (Fig. 4).

Time savings with graphics programming has lowered to about 3 months the average time needed to automate a model. By comparison, it might have taken more than 2 years to develop similar automation applications with text-based programming languages.

OPERATOR INTERACTION

During an experimental run, the operator interacts with a virtual control panel generated on the screen. This interaction involves reading meters and strip-chart traces, scanning indicator lights, flipping switches, turning knobs, and so forth.

If a control panel is larger than the screen, hidden elements can be reached by horizontal and vertical scrolling. The panel configuration provided by the programmer can be modified by the operator to suit the particular experiment.

For instance, strip charts which are not used might be moved off the screen, and those in use might be enlarged (Fig. 5).

Experimental runs consist of steps followed as for a programmable logic controller (PLC). Using various options, without changing programs, operators can configure their own sequences for various experiments.

For each step, an operator can specify the position of control valves, the setpoints of controllers (constant or ramped up or down), and so forth.

Conditions at which the program goes to another step can also be set. During execution, the operator can manually terminate any step, override any variable, and turn data recording on or off. Data points selected by the operator are recorded at specified time intervals in a standard tabular format compatible with spreadsheet software.

The length of the data file might be anywhere from 100 to 6,000 lines. For data reduction, scientists can use any of the programming aids provided by Labview, including time and frequency analysis, curve fitting, and graphic charts. Files can also be exported directly into other applications including Lotus 1-2-3 and Excel. Report-quality charts and tables can be prepared in minutes rather than days.

An additional software package provides remote operator access to laboratory computers via telephone on a dial-up basis or through a local network.

Models running automatically under the graphical programming software can be monitored and controlled by scientists in their offices, or even at their homes.

EQUIPMENT AUTOMATED

Some of the equipment controlled by the software includes:

Core flow stations to determine porosity and flow characteristics of rock or unconsolidated sand with various fluids at reservoir conditions under simulated overburden pressure.
Shale tester to investigate the degradation of shale layers, through which a well bore passes, under various prospective treatments of the well.
Viscometer to measure the viscosity of fluids at downhole pressures and temperatures.
Gas-migration tester for investigating the displacement of water by gas in various types of cement around the casing of a well.
Lubricity tester for evaluating the effectiveness of various downhole liquids in reducing frictional resistance to drill pipe rotation.

So far, 4 years into computer automation at the Westport laboratory, the consensus is that the work has been done in the right way. There have been a few surprises, but virtually no complaints or regrets.

Next up for the BP team is the construction of an automated, portable core flow station for Prudhoe Bay, Alaska. The device will be used to control reservoir core flow tests using Prudhoe Bay field's produced brine. With the data links in place, an engineer in Houston can monitor in real time the effectiveness of new biocides and corrosion inhibitors in one of the world's most remote and hostile locations.