Shishavan R.A.,University of Houston |
Hubbell C.,Brigham Young University |
Perez H.D.,Brigham Young University |
Hedengren J.D.,Brigham Young University |
And 2 more authors.
SPE Journal | Year: 2016
With the recent advance in high-speed data communication offered by wired-drillpipe (WDP) telemetry, it is now possible to design automated control systems that directly use downhole data (e.g., pressure) to optimize drilling procedures. This research couples drilling hydraulics, rate of penetration (ROP), and rotationalspeed (rev/min) control into a single controller for managed-pressure- drilling (MPD) systems. This novel multivariate controller improves drilling performance during normal drilling operations and enhances safety during abnormal drilling conditions such as unwanted gas-influx situations. New advances in drilling automation have made the closedloop control of downhole weight on bit (WOB) and drillstring rotational speed (rev/min) possible. This study uses two feedback controllers that control the downhole WOB and rev/min by use of surface data. A multivariate nonlinear model-predictive controller (NMPC) uses downhole and surface measurements to simultaneously regulate the bottomhole-assembly (BHA) pressure and maximize the ROP. For this purpose, NMPC provides the necessary set points for the WOB and rev/min feedback controllers and manipulates the choke-valve opening and pump-flow rates. Controller performance is enhanced by means of a nonlinear estimator that works continuously online with the NMPC and provides the necessary estimated parameter values (such as annulus density, friction factor, and gas influx) for precise and efficient drilling control. The designed NMPC controller has a multipriority approach that is described in the following three scenarios: during unexpected gas influx, the NMPC gives priority to BHA pressure control and attenuates the influx effectively by means of a novel kickattenuation method that switches the control objective from BHA pressure to choke-valve pressure; during connection procedures when adding a new stand, ROP is stopped and the NMPC focuses on maintaining the BHA pressure constant; and during normal drilling operation, which involves changes in the rock formation and differential pressures, NMPC gives priority to ROP maximization while maintaining rev/min, WOB, and BHA pressure within specified bounds. Preliminary results suggest that this multivariate controller for ROP and BHA-pressure control decreases drilling costs, reduces operator workload, and minimizes risk significantly. Specific improvements in drilling performance include higher ROP, effective kick attenuation, and more-uniform cuttings. The use of a multivariate NMPC allows for better ROP optimization and BHA-pressure control than is possible with the use of two independent controllers. These benefits are demonstrated across the three scenarios mentioned previously. In simulation, this technology delivers significant performance improvements during MPD and furthers the development of automated-driller systems. ©2016 Society of Petroleum Engineers. Source
Brunner K.M.,BrighamYoung University |
Duncan J.C.,BrighamYoung University |
Harrison L.D.,BrighamYoung University |
Pratt K.E.,BrighamYoung University |
And 3 more authors.
International Journal of Chemical Reactor Engineering | Year: 2012
A trickle fixed-bed reactor model for the Fischer-Tropsch synthesis applicable to both cobalt and iron catalysts which accounts for gas and liquid recycle is described. A selection of kinetic models for both iron and cobalt catalysts (4 each) is included in the reactor model and their effect on model predictions is compared. While the model is 1-D and reaction rates are determined for quasi-average radial bed temperatures, a correlation is used to account for radial thermal conductivity and radial convective heat transfer. Traditional pressure drop calculations for a packed column were modified with a correlation to account for trickle-flow conditions. In addition to describing the model in detail and showing validation results, this paper presents results of varying fundamental, theoretically-based parameters (i.e. effective diffusivity, Prandtl number, friction factor, etc.). For example, the model predicts that decreasing effective diffusivity from 7.1E-09 to 2.8E-09 m2/s results in a lower maximum temperature (518 K vs. 523 K) and a longer required bed length to achieve 60% conversion of CO (8.5 m vs. 5.7 m). Using molar averages of properties to calculate the Prandtl number for the gas phase (recommended by the authors) results in average bed temperatures up to 10 K higher and reactor lengths 17-45% shorter than assuming a Prandtl number of 0.7. Using the Tallmadge equation to estimate friction losses, as recommended by the authors, results in a pressure drop 40% smaller than using the Ergun equation. Validation of the model was accomplished by matching published full-scale plant data from the SASOL Arge reactors. Copyright © 2012 De Gruyter. Source