Closed-loop active turbine tip clearance control methods are being sought by industry to improve the efficiency and longevity of modern-day gas turbine engines. Efforts are underway at the NASA Glenn Research Center to develop rapid-response actuation systems that regulate a segmented shroud ring that is free to move radially relative to the turbine blades. When used in conjunction with a tip clearance proximity sensor, the system enables tight clearances to be maintained throughout the operating envelope, even when rapid changes in the rotor diameter are experienced during power-up events.
Targeted reductions in specific fuel consumption and peak exhaust gas temperatures, which will negatively impact engine service life, are in excess of 1 percent and 10 °C, respectively. As progress continues on developing fast tip-clearance actuators and high-temperature proximity sensors, there is a strong need to develop control laws that can protect the engine from damaging blade rubs at the tight clearances required to achieve the targeted reductions in fuel consumption and exhaust temperatures. NASA researchers have identified model predictive control (MPC) as offering the most promise for attaining this challenging objective. With this method, a model of the engine is used to compute a trajectory of the clearance evolution through time; the controller uses this trajectory as a basis for optimizing future actuator commands. The advantage is that the controller can actively avoid operating limits, such as a minimum acceptable clearance to prevent the actuator from moving the shroud to within an unsafe proximity to the blades. MPC, therefore, improves the robustness of the system, allowing tighter clearance setpoints than those that are used with controllers designed with conventional methods.
There are inherent limitations to MPC when it is applied to a highly nonlinear turbofan engine. Linear parameter-varying models used to describe such systems lose fidelity when high-magnitude transients occur, such as takeoff or reacceleration, resulting in degraded MPC performance and loss of robustness to blade rubs, and therefore, reduced tip-clearance performance. In this work, a rate-based linear parameter-varying model was used instead to greatly reduce the model’s dependence on statically derived parameters, thereby extending fidelity to rapid transients. As shown in the block diagram below, this rate-based model was used in conjunction with standard quadratic programming optimizers to derive a new MPC with enhanced performance throughout the flight trajectory.
Rate-based MPC. The “Estimator” block reconstructs the model states at the current time step on the basis of sensed output data. The “QP solver” block (quadratic programming) uses the estimated states at the current step to generate a model-predicted trajectory and computes an optimal rate-based controller output, 
. In the rate-based framework, this output is integrated to form the actuator command, uk. The d/dt block represents signal differentiation.
Long description of figure 1.
Extensive simulation studies of a tip-clearance system and high-fidelity engine simulation were used to verify the improved performance achievable with this novel MPC algorithm. In the evaluation, both thermal- and servohydraulic-type actuators were evaluated; the results for three transient event scenarios using the servohydraulic are shown in the following time traces and are summarized in the table. Note that a zero-clearance setpoint is used here for illustration; in actual systems, negative values would correspond to clearances smaller than the setpoint. It is shown that, using this MPC controller, the minimum clearances in response to takeoff, thrust reversal, and airplane stall events were significantly smaller than those generated by a conventional control with no constraints. Applying worst-case analysis on the results reveals that the achievable setpoint of an MPC-based controller may be 2.55 mils tighter than that of a conventional controller, improving engine efficiency by approximately 0.25 percent. This improvement is considerable when compared with the 1.0-percent gain in efficiency realized by replacing open-loop thermally activated clearance control with a closed-loop mechanically actuated device. Efforts are underway to implement and evaluate the controller in a nonrotating turbine clearance control test rig.
Time histories of the high-pressure turbine clearance disturbance rejection with the rate-based MPC and a conventional linear controller. Both controllers have zero-clearance setpoints, but the MPC imposes a lower limit (also set at zero clearance) in order to realize tight clearances without blade rubs.
Long description of figure 2.
| Mission event | Minimum clearance, mils | ||
|---|---|---|---|
| Conventional control |
Model predictive control |
Delta improvement |
|
| Takeoff | -0.80 | -0.18 | 0.62 |
| Thrust reversal | -2.29 | -0.50 | 1.79 |
| Airplane stall | -3.26 | -0.71 | 2.55 |
DeCastro, J.: Rate-Based Model Predictive Control of Turbofan Engine Clearance. AIAA-2006-5107 (NASA/CR--2006-214419), 2006. http://gltrs.grc.nasa.gov/Citations.aspx?id=219
ASRC Aerospace contact: Jonathan A. DeCastro, 216-433-3946, Jonathan.A.DeCastro@nasa.govLast updated: December 28, 2007
Responsible NASA Official:
Gynelle.C.Steele@nasa.gov
216-433-8258
Point of contact for NASA Glenn's Research & Technology reports:
Cynthia.L.Dreibelbis@nasa.gov
216-433-2912
SGT, Inc.
Web page curator:
Nancy.L.Obryan@nasa.gov
216-433-5793
Wyle Information Systems, LLC
