Document
Proceedings of ASME Turbo Expo 2017: Turbomachinery Technical Conference and Exposition GT2017 June 26-30, 2017, Charlotte, NC, USA
GT201 7 - 651 28
A DYNAMIC MODEL FOR THE EVALUATION O F AIRCRAFT ENGINE IC ING DETECTION AND CONTRO L - BASED MITIGATION STR ATEGIES Donald L. Simon Aidan W. Rinehart Scott M. Jones NASA Glenn Research Center Vantage Partners, LLC NASA Glenn Research Center 21000 Brookpark Road 3000 Aerospace Parkway 21000 Brookpark Cleveland, OH, 44135 Brook Park, OH 44142 Cleveland, OH 44135 ABSTRACT engine thrust output . Follow - on work will assess the feasibility A i rcraft flying in regions of high ice crystal concentrations and effectiveness of such control - based mitigation strategies.
are susceptible to the buildup of ice within the compression system of their gas turbine engines. This ice buildup can restrict INTRODUCTION engine airflow and cause an uncommanded loss of thrust, also Airborne ice crystals can pose an operational hazard to known a s engine rollback, which poses a potential safety aircraft gas turbine engines. Since 1990, there have been over hazard. The aviation community is conducting research to 100 reported cases of engine power loss due to ice accretion understand this phenomena, and to identify avoidance and within the engine’s compression system [ 1 ,2,3]. When flying in mitigation strategies to address the concern. To support this high ice water content conditions, frozen ice crystals can be research, a dynamic turbofan e ngine model has been created to ingested by the engine and then impinge on warm engine enable the development and evaluation of engine icing surfaces, causing the particles to partially melt and begin to detection and control - based mitigation strategies. This model accrete. This can lead to the buildup of ice within the core of captures the dynamic engine response due to high ice water the engine, blocking airflow and resulting in an uncommanded ingestion and the buildup of ice blockage in the engine’s low loss of thrust, also known as an engine rollback event.
pressure compressor. It includes a fuel control system allowing Within the aviation community, much work is ongoing to engine closed - loop control effects during engine icing events to characterize the environmental conditions under which engine be emulated. The model also includes bleed air valve and icing can occur [4,5,6,7,8,9,10] and understand the mechanisms horsepower extraction actuators that, when modulated, cha nge by which ice particles can accrete on compressor components overall engine operating performance. This system - level model [11,12,13,14,15]. Example collaborative efforts include the has been developed and compared against test data acquired Engine Icing Working Group (EIWG), a joint committee of from an aircraft turbofan engine undergoing engine icing international government and industry representatives studies in an altitude test facility and also against outputs from coordinating research in the area of engine ice crystal icing, and the man ufacturer’s customer deck . This paper will describe the the Ice Crystal Consortium (ICC), a group of engine and model and show results of its dynamic response under open - airframe manufacturers formed to combine resources to address loop and closed - loop control operating scenarios in the the problem. Notable engine testing has also recently been presence of ice blockage buildup compared against engine test conducted at the NASA Glenn Research Center (GRC) cell data. Planned follo w - on use of the model for the Propulsion System Laboratory (PSL) [16], an altitude test development and evaluation of icing detection and control - facility that has been modified to enable engine testing in based mitigation strategies will also be discussed. The intent is simulated high ice water content conditions [17]. Flight test to combine the model and control mitigation logic with an campaigns have also been conducted to understand ice crystal engine icing risk calculation tool capable o f predicting the risk icing conditions and to assess the ability of radar and of engine icing based on current operating conditions. Upon meteorological probes to detect such conditions detection of an operating region of risk for engine icing events, [18,19,20,21,22]. The experimental results acquired from the control mitigation logic will seek to change the engine’s engine testing and flight testing have been key in advancing the operating point to a region of lower risk thro ugh the modulation understanding of the engine icing phenomena.
of available control actuators while maintaining the desired This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 1 Avoidance of flight through high ice crystal concentration CON TROL - BASED ICING RISK MIT IGATION atmospheric conditions and re-design of compressor hardware Figure 1 shows the envisioned control - based icing risk to make engines less susceptible to icing are potential solutions, mitigation architecture. The objective of this architecture is to and much of the current research is focused on these areas. provide early onset detection of engine icing initiation or icing However, additional benefit can be gained by taking a systems risk, and then , through the modulation of available control level approach to understand how engine system-level actuators, adjust the engine’s operating point to a condition of performance changes when exposed to icing conditions and lower ice accretion risk. Integral to this architecture is icing risk evaluating what mitigation steps can be taken to reduce the risk assessment logic. Resear chers at NASA GRC have developed a of engine icing. Towards this objective, this paper will present computational tool to estimate the risk of ice accretion and its a dynamic aircraft engine model created to facilitate the effect on turbofan engine performance [ 23 ]. This tool includes development and evaluation of aircraft engine icing detection an engine system thermodynamic cycle code, coupled with a and control-based icing risk mitigation strategies. The compressor flow analysis code, and a n ice particle code that has remaining sections of this paper will present the vision for a the capability of determining the rate of sublimation, melting, control-based mitigation approach. Next, a description of the and evaporation of ice particles as they pass through the model will be provided. This will be followed by a comparison compressor. This tool, or a modified version of it suitable for of the model against engine test data acquired from NASA real - time implementation, will be included in the active control GRC PSL testing and outputs from the manufacturer’s icing risk mitigation architecture shown in Figure 1 . A customer deck. The paper will conclude with a discussion of simplistic solutio n is desired in implementing this architecture , the next steps and a summary. striving to use available sensors and actuators if possible, and arriving at a computationally efficient implementation.
NOMENCLATURE Solutions that minimize impact on engine operating c Specific heat of ice performance are also required, where the requested thrust ice c Specific heat of water vapor output of the engine is not compromised.
steam c water Specific heat of water COMDES Mean - line compressor design code • Atmospheric conditions, including ice crystals • Engine operating condition dTamb Delta between actual and standard temperature EIWG Engine Icing Working Group Aircraft GRC Glenn Research Center Engine Actuator H Heat of fusion for ice f commands H v Heat of vaporization for water Aircraft & engine HPC High Pressure Compressor Engine sensor measurements HPT High Pressure Turbine Control ICC Ice Crystal Consortium Icing detection logic ISA International Standard Atmosphere LPC Low Pressure Compressor Ice accretion LPT Low Pressure Turbine risk calculation Activate MVD Median Volumetric Diameter control No N1 Fan speed mitigation N2 Core speed strategy Icing NPSS Numerical Propulsion System Simulation risk?
P25 Low pressure compressor exit pressure Yes P3 High pressure compressor exit pressure PLA Power Lever Angle Figure 1 . Active Control Icing Risk Mitigation Architecture PSL Propulsion System Laboratory Q Heat loss rate in the low pressure compressor LPC This paper will focus on the development of a dynamic Q HPC Heat loss rate in the high pressure compressor engine model, which can serve as a surrogate for the actual T2 Fan hub and LPC inlet temperature aircraft engine in this architecture when conducting initial T25 Low pressure compressor exit temperature development and evaluation of icing detection and control - T3 High pressure compressor exit temperature based mitigation simula tion studies. This model is T45 Exhaust gas temperature representative of a n ALF502 - R5 twin - spool high bypass Toolbox for the Modeling and Analysis of turbofan engine. The ALF502 - R5 engine, which was originally T - MATS Thermodynamic Systems produced by Lycoming, and then later by Allied Signal and now W2 Fan hub and LPC inlet air flow Honeywell , is designed for regional aircraft applica tions and W25 Low pressure compressor exit air flow produces approximately 7,000 pounds of thrust. NASA and W3 High pressure compressor exit air flow Honeywell collaborated on ALF502 - R5 engine ice crystal icing Wf Fuel flow testing in the NASA GRC PSL facility in 2013 and 2015 This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 2 [ 24 , 25 ]. The ALF 5 02 engines used during PSL testing were which case the user supplies commanded fuel - flow as an input, experimental engines , with significant differences from or closed - loop control mode, where the user supplies a production engines. Honeywell’s participation in this testing commanded engine throttle setting (or power lever angle was made possible through funding provided by the ICC. (PLA)).
DYNAMIC MODEL DESCRIPTION Engine Model The dynamic model of the ALF502 - 5R engine is a 0D Bypass Fan tip component level model code d using the Toolbox for the nozzle Fan hub Modeling and Analysis of Thermodynamic Systems ( T - MATS ) Ambient HPC with Core Inlet & LPC w/ LPT Splitter Combustor HPT conditions heat loss nozzle [ 26 ] , which is a NASA - developed open source graphical heat loss HP thermodynamic simulation package built in shaft MATLA B/Simulink (The MathWorks, Inc) . The developed T - LP shaft MATS model was derived from a Numerical Propulsion System Sensed feedback measurements Fuel flow Simulation (NPSS) model of the ALF502 - 5R engine using the Ice Particle Ambient Ice Fuel flow (open-loop) techniques described in Ref. [ 27 ] . The NPSS ALF502 - 5R Engine Density Inputs Blockage -or- Control engine model was developed to simulate engine performance PLA (closed-loop) during a n engine icing rollback event. The NPSS mode l include d constructs to capture the thermodynamic effects of ice Figure 2 . T - MATS ALF502 - 5R d ynamic e ngine m odel b lock particles melting and vaporizing within the engine and a d iagram reduction in the LPC flow path area due to ice blockage buildup in the LPC . The NPSS model can be run in steady - state off Description of Model Turbomachinery Components design mode over a series of analysis points to simulate engine The dynamic engine model is constructed in behavior during a rollback event.
MATLA B/Simulink using T - MATS library building blocks. The Like the NPSS model described in Ref. [ 27 ], the T - MATS blocks are based on basic thermodynamic equations and model includes const ructs to capture the thermodynamic effects principles, and use a series of user - supplied maps that define of ice water ingestion by the engine as well as ice blockage the characteristics of the system being modeled. The developed buildup within the engine’s LPC . A difference between the T - MATS model uses the component maps applied within the NPSS model and the T - MATS model presented in this paper is NPSS model described in Ref. [ 27 ]. N umerical methods , that the T - MATS model adds the capa bility to simulate dynamic including a Newton - Raphson iterative solver and Jacobian or transient engine behavior. It includes a fuel control system calculations , are also included to allow the model to be with a user selectable option to run the model in either open - balance d to a given operating point. First order lags are added loop or closed - loop control mode. In addition to fuel flow, the to the model’s sensed temperature outputs to provide better T - MATS model also includes several auxiliary actuators, such matching with the sensor measurements acquired from the as customer bleeds and horsepower extraction, that can be used actual ALF502 - 5R engine during trans ient operation .
to modulate engine operating performance. A high - level block diagram of the T - MATS ALF502 - 5R engine model is shown in Heat Extraction Due to Ice Particle Ingestion Figure 2 . Major blocks of the model include ambient A key feature of the model is the inclusion of heat conditions, inlet, fan tip, fan hub and low pressure compressor (enthalpy) extraction effects due to ice particle ingestion , which (LPC), high pressure compressor (HPC), combustor, high was also done for the NPSS model described in Ref. [ 27 ] .
pressure tu rbine (HPT), low pressure turbine (LPT), and bypass Modeling of heat extraction accounts for the phase transition and core nozzles.
that ingested ice particles undergo as they transition from ice - to - water - to - vapor when passing through the engine’s Model Inputs compression system (fan hub, LPC and HPC). For modeling Required T - MATS ALF502 - 5R model inputs include purposes, this heat extraction is modeled to occur both within ambient conditions consisting of altitude, Mach number, and the LPC and the HPC as shown in Figure 3 . A t the exit of the dTamb (difference between actual ambient temperature and LPC , heat is removed at a rate equal to the amount required to Intern ational Standard Atmosphere (ISA) temperature for the raise the ice temperature from the inlet temperature condition to given altitude) , as well as icing conditions consisting of ice 32˚F , melting the ice, and then further raising the water particle density and the percentage of flow blockage in the temperature to the LPC exit temperature . Here, the heat loss LPC. Although the true percentage of LPC flow blockage that rate, Q , is calculated as LPC occurred during PSL testing of the ALF502 - R5 engine was not measurable, a technique for calculating this m odel input
F T c w H w T F c w Q 32 25 2 32 ( 1 ) water ice f ice ice ice LPC parameter to achieve desired model - to - engine matching during rollback events has been developed, as will be described later in where w is the ice mass flow rate entering the core of the ice this paper. The user has two op tions for controlling the power engine, c is the specific heat of ice, c is the specific heat ice water setting of the engine. This includes open - loop control mode, in This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 3
Engine Model
Bypass Fan tip nozzle Fan hub Ambient HPC with Core Inlet & LPC w/ Combustor HPT LPT Splitter conditions heat loss nozzle heat loss HP shaft LP shaft Fan hub & HPC LPC 1 1 1 1 1 1 1 1 QHPC QHPC QHPC QHPC QHPC QHPC QLPC QHPC QHPC 8 8 8 8 8 8 8 8 Fan hub heat HPC heat HPC heat HPC heat HPC heat HPC heat HPC heat HPC heat HPC heat stg1 loss stg3 loss stg5 loss stg6 loss stg7 loss stg8 loss & LPC loss loss stg2 loss stg4 Figure 3 . Implementation of heat loss in compression system of water, H is the heat of fusion of ice, T2 is the inlet transfer through metal components are also neglected in the f temperature, and T25 is the LPC exit temperature. model.
The HPC of the ALF502 - 5R engine is an eight - stage design consisting of seven axial stages and one centrifugal Flow Blockage Due to Ice Buildup in the LPC stage. As shown in Figure 3 , within the model the HPC is The ALF502 - R5 T - MATS model includes an “LPC ice implemented as eight individual T - MATS compressor blocks, blockage” input, a lumped parameter that captures LPC each representing a single stage o f the HPC. A performance performance changes occurring from ice accretion. At the cycle characteristic map for each HPC stage similar to a single stage level, the impact of this input is on LPC pressure ratio , air flow, fan was used, and each stage was assumed to have the same and efficiency. In the model, ice blockage build up within the polytropic efficiency and specific work. After each stage, a LPC of the engine is captured through a series of modified LPC portion of the total heat extraction within the HPC is maps , each represent ing a different amount of flow blockage in performed. The model first calculates the heating rate required the LPC. These maps are generated using a NASA - developed to raise the liquid from the HPC inlet temperature to boiling mean - line compressor design code (COMDES) [ 28 ] , where a temperature, vaporizing it, and then further heating the vapor to reduction in the LPC flow path area was added, to simulate the the HPC exit temperature. This is given as effect of ice blockage as described in Ref. [ 29 ]. Using this code, the size of the blockage can be changed and a new compressor map developed. This results in a series of maps that can be
H w T F c w Q 25 212
v ice water ice HPC ( 2 ) “stacked” and interpolated between to enable the simulation of
F T c w 212 3
steam ice an arbitrary ice blockage level. By changing the “ LPC ice blockage ” input parameter the user can simulate the accretion where c and c are the specific heat of water and steam, water steam of ice crystals and flow blockage within the engine.
respectively, and H is the heat of vaporization of water. This v heat loss is then removed from the HPC stage - by - stage in eight Open - Loop versus Close d - Loop Control of Engine equal in crements, with the total heat loss rate summing to the Model amount given in Eq. ( 2 ) .
The user has two options for controlling the thrust output For the engine icing rollback events demonstrated during of the ALF502 - R5 T - MATS model. The first option is to run the PSL testing, the mass flow r ate of ice ingested into the engine model in open - loop control mode while supply ing a time was relatively small — approximately 0.5% that of air. This history of fuel flow, W f , which is directly fed in to the model ’s additional mass of ingested ice is not accounted for in the combustor block . The second option is to run the model in model calculations. Other parameters such as heat transfer from closed - loop control mod e , consistent with how the engine was the air to the water vapor past the HP C component and heat This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 4 operated during PSL testing . Here, as opposed to a fuel flow engine operating point, and how that change in operating point input, the user supplies a t ime history of power lever angle effect s the engine’s risk of ice accretion and thrust output.
( PLA ). Within the model’s closed - loop control logic, the Although not evaluated in this paper, follow - on work will provided PLA input is converted to a co mmanded engine power couple the engine model with an icing risk calculation tool setting . In the T - MATS ALF502 - 5R engine model ’s control developed by NASA GRC (see Ref. [ 23 ]) to enable this logic , this is implemented by converting the supplied PLA input evaluation.
into a corresponding Wf/P3 vs. N2 characteristic line, known as N2 governor an N2 governor droop line [ 30 ] . Here, Wf/P3 is the ratio of fuel droop line flow to HPC exit pressure , P3 , and N2 is core speed. A notional Increasing depiction of the N2 governor droop line plotted in Wf/P3 vs.
PLA N2 parameter space is shown in Figure 4 a . The location of the N2 governor droop line changes in response to changes in PLA.
Wf Also plotted in this figure is the engine’s own characteristic P3 Decreasing operating line . The location of the engine’s operating line is PLA Nominal engine dependent on various factors such as the engine’s deterioration operating line level, bleed extractions, heat transfer due to ice particle Nominal engine ingestion, and ice buildup in the engine. The intersection of the operating point N2 governor droop line and the engine operating line defines a N2 point of steady - state operation, and the model’s fuel control logic applies a proportional - integral controller to adjust fuel a) Movement of N2 governor droop line with changing PLA flow to run the model to this operating point. As shown in the figure, the engine model can be run to higher or lower po wer settings on the engine operating line by increasing or “Iced” engine “Iced” engine decreasing the provided PLA input . As prev iously noted, ice operating line operating point buildup within the compression system chang es the flow N2 governor droop line characteristics of the compressor , shifting the engine’s Shift in operating line due to ice ingestion and operating line . Figure 4 b shows a notional depiction of how the increasing ice blockage engine’s operating line shifts upward due to ice particle Wf ingestion and ice blockage , which results in a corresp onding P3 control adjustment of fuel flow to maintain engine operation on the N2 governor droop line . As described in Ref. [ 30 ], during Nominal engine ice blockage induced engine ro llback events, the engine operating line operating point will move upwards along the N2 governor Nominal engine operating point droop line. While the actual ALF502 - 5R engine includes additional control limit logic, these limits are not encountered N2 until a significant amount of ice buildup has occurred and the b) Movement of engine operating line caused by heat transfer engine is relatively far into an icing rollback event . Since the due to ice ingestion and increasing ice blockage intent of the model presented in this paper is for the Figure 4 . Engine N2 governor droop line control development and evaluation of approaches for the incipient detection and mitigation of engine icing events, such limit logic Engine Model is not included in the T - MATS model presented in this paper.
Fan Customer Bleed X Auxiliary Actuators Bypass X Core Customer Bleed Fan tip nozzle In addition to fuel flow, several additional actuators are Fan hub Ambient HPC with Core included in the T - MATS engine model. This includes two Inlet & LPC w/ Combustor HPT LPT Splitter conditions heat loss nozzle heat loss customer bleeds, an anti - ice bleed, and horsepower extraction.
X HP Enthalpy Figure 5 shows the location of each actuator within the engine shaft change LP model. When opened, the fan customer bleed and core shaft Horse Power customer bleed extract a fraction o f the airflow aft of the fan tip Extraction Anti-Ice Bleed and the HPC, respectively. The anti - ice bleed also extracts Sensed feedback Icing Ambient Fuel flow measurements airflow aft of the HPC exit. Additionally, the anti - ice bleed Condition Inputs Fuel flow (open-loop) Engine Inputs causes an enthalpy change in the airflow entering the fan hub -or- Control PLA (closed-loop) and LPC. Horsepower extraction results in a reduction in the torque of the model’s high pressure shaft.
Figure 5 . Dynamic Engine Model – Auxiliary Actuator The intent of including these additional auxiliary actuators Locations in the model is to evaluate how their modulation changes the This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 5 MODEL COMPARISON TO EXPERI MENTAL RESULTS Recently, two experimental (non - production) ALF502 - 5R engine s underwent ice crystal icing tests in the PSL at the NASA Glenn Research Center. The first test, conducted on engine serial number LF01, occurred in 2013 [ 24 ], and the second test, conducted on engine serial number LF11, occurred in 2015 [ 25 ]. During this testing, the engines were subjected to ice ice crystal ingestion and multiple engine rollback events were cloud demonstrated. In this paper, e xperimental data recorded from on th e LF01 engine test will be compared aga inst the developed T - MATS engine model.
NASA GRC PSL Facility and Ice Crystal Engine Testing The NASA GRC PSL is an altitude simulation fac ility for experimental research on air - breathing propulsion systems [ 31 ] .
The re are two test cells within the facility, PSL - 3 and PSL - 4, which are capable of providing simulated flight conditions to Figure 7 . Normalized e ngine measurement parameters recorded altitudes in excess of 90,000 feet. In 2012 , PSL - 3 was modified during LF01 Run 193 rollback event to include a water spray nozzle array system to produce ice crystal clouds at simulated altitudes during aircraft engine cloud turning on. During the test, engine power setting was first testing [ 17 , 32 ] . Photos of the spray nozzle system and the stabilized at the target operating cond ition and then the ice installation of the LF01 engine in PSL - 3 are shown in Figure 6 . cloud was turned on. This occurs at 10 seconds, and is evident from the step change in several of the parameters that occurs at this time. As ice blockage builds up in the engine, the uncommanded rollback event ensues. This is most a pparent in N1, P3, and Wf, which exhibit a noticeable reduction relative to their pre - ice cloud readings.
Comparison of Engine Model to LF01 Engine Experimental Data Experimental measurement data acquired during LF01 Figure 6 . Water injection spray nozzles and LF01 engine engine testing was used to evaluate th e T - MATS ALF502 - 5R installation in PSL - 3 engine model’s ability to capture engine dynamic behavior during icing events. The engine model was run under both The 2013 LF01 engine test in PSL - 3 marked the inaugural open - loop and closed - loop control operating modes, and the engine icing test conducted in the facility. Objectives of this test model’s output was compared against sensed measurement were to validate that the newly modified test cell could be outputs acquired from the engine. Recorded parameters of calibrated and operated at target flight test conditions, and to altitude, Mach, dTamb, Wf (open - loop only), and PLA (closed - duplic ate documented thrust rollback events that occurred loop only) were supplied as inputs to the model. Additional during revenue service. During LF01 testing in PSL - 3, multiple model input parameters of ice particle concentration and LPC engine rollback events were successfully induced. Figure 7 flow blockage were determined based on the experimental data.
shows engine performance parameter measurements recorded The subsections below will present the steps for calculating and during the first engine rollback event demonstrated during supplying these inputs to the T - MATS ALF502 - 5R engine * LF01 PSL - 3 engine testing, denoted as Run 193 . This test was model . A comparison of model and experimental results will conducted at an operating point of approximately 28,000 feet, first be shown for LF01 PSL - 3 Run 1 93, followed by a 0.5 Mach, and +28ºF dTamb. Shown are the first 200 s econds comparison of the results for additional LF01 engine rollback of recorded measurement s for fan speed, N1, core speed, N2, events .
HPC exit pressure, P3, HPC exit temperature, T3, exhaust gas temperature, T45, fuel flow, Wf , and the ratio of fuel flow to Calculation of Ice Particle Concentration . The HPC ex it pressure, Wf/P3 . Due to the proprietary nature of the fundamental research conducted during the LF01 PSL - 3 testing data, here and throughout the remainder of this document , y - included assessing engine response when subjected to ice c loud axis units are normalized to show fractional units relative to conditions of varying total water content and particle mean each parameter’s stabilized operating condition prior to the ice volumetric diameter (MVD). Throughout the test, the water flow rate supplied to the spray nozzle array and the airflow rate * Run number denotes the facility data acquisition system recording number for entering the engine was recorded. This information was used to a given LF01 PSL - 3 engine test run.
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 6 calc ulate the ice particle density input supplied to the T - MATS Calculation of the amount of LPC Ice Blockage .
ALF502 - 5R engine model. The e ngine rollback events demonstrated during LF01 PSL - 03 Figure 8 a show s an example of the calculated ice particle testing confirm that s ome amount of ice buildup was occurrin g concentr ation for the LF01 PSL - 3 Run 193 engine rollback within the engine’s compression system . However , there w a s no event . It should be noted that i ce particle density is affected by means of directly measuring the actual amount , or percentage, the direct connect nature of the PSL test set - up [ 24 ] . While the of LPC flow blockage. In order to estimate the amount of LPC total water flow rate supplied to the spray nozzles was held ice blockage, which is a necessary model input, an automated constant wh en an ice cloud was on, the total airflow into the technique was developed and applied . This technique estimates engine would reduce as the engine rolled back. This caus es an the ice blockage buildup time history that will produce the increase in ice particle concentration as the engine progresses same rollback rate in N2 within the model as was observed further into rollback . within the actual N2 measurement data recorded from the Figure 8 b shows the initial drop in LF01 HPC exit engine. To make this calculation , a two - dimensional table temper ature, T3, for the same PSL test run when the ice cloud lookup model was constructed that produces an estimate of the turned on . As previously described, this temperature drop is amount of LPC ice blockage given inputs of fuel flow and the hypothesized to be due to heat loss effects caused by the N2 error between the engine a nd the model run with no LPC melting and vaporization of ice particles as they pass through ice blockage . The table lookup model was constructed by the engine’s compr ession system. Also shown in the figu re is running the model in open - loop control mode at fixed altitude, the T - MATS model produced T3 output under two scenarios. Mach number, dTamb, and i ce particle concentration inputs First, when the model is supplied the calculated ice particle while varying fuel flow and % LPC ice blockage ove r a range density input parameter given in Figure 8 a , and then when this of discrete settings and storing the corresponding change in N2.
same model input parameter is set to zero. Setting this input to The resulting data is applied within the table lookup model, and zero is equivalent to running the model without heat loss linear interpolation is used to retrieve values between grid effects. The results show that including heat l oss effects is points. Once the table lookup model was created, the en gine necessary in order to accurately capture the initial drop in T3 . model was run in a three - step process. First, the model was run In evaluating different run case s from t he LF01 PSL - 3 open - loop and provided all necessary inputs except for the testing , it was found that the model’s ability to accurately match amount of LPC flow blockage, which is set to zero. Then, the the magnitude of the measured T3 drop that occurred when the resulting error between the model’s N2 output and the desired ice cloud was turned on varied by run case. The testing did N2 to produce the same N2 rollback rate as that observed in the evaluate engine response to different water spray patterns and engine was calculated. This N2 error along with the measured MVD conditions, which possibly contributed to these fuel flow from engine testing were then supplied as inputs to variations. Given this finding, the calculated ice particle the lookup table to retrieve an LPC ice blockage history .
concentratio n input supplied to the model was adjusted by a Finally, this bl ockage, along with the other original inputs, was scale factor that result s in the model producing the same drop in supplied as inputs to the model. Figure 9 a shows an example of T3 as was observed in the LF01 test engine T3 measurement the calculated amount of LPC ice blockage for the LF01 PSL - 3 data for a given run case . The magnitude of the ice particle Run 193 rollback event, and Figure 9 b shows the corresponding concentration scale fac tors applied for the various engine N2 response for the engine and the model run with and without rollback events evaluated in this study ranged from 0.86 to this LPC ice blockage input . The case without blockage input is 1.05. For Run 193 specifically, a scale factor of 0.88 was equivalent to the initial calibration run conducted to determine applied. the N2 value without any ice blockage (i.e., first step in the Ice cloud off Ice cloud on Ice cloud on a) Calculated LPC ice blockage b) Engine and model N2 response a) Calculated ice particle density b) Engine and model T3 response Figure 8 . Calculated ice particle density and T3 response during Figure 9 . Calculated LPC ice blockage and N2 response during the Run 193 engine rollback event the Run 193 engine rollback event This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 7 three - step process of simulating a rollback event). Without ice blockage input, the model’s N2 response rolls off more rapidly than what occurs in the actual engine. It is emphasized that the calculated amount of LPC ice blockage should not be interpreted as a true measure of the amount of blockage within the LPC. Instead, this is simply the amount of blockage input necessary for the model’s N2 output to follow the same rollback rate as that observed in the engine’s N2.
Modeling of Run 193 Engine Rollback Event . T he Run 193 rollback event was simulated by running the T - MATS ALF502 - 5R engine model in both open - loop and closed - loop control mode . The applied m odel i nput parameters are shown in Figure 10 a . These inputs include ice co ncentration and LPC ice blockage inputs ( calculated as described in the previo us subsections ) plus a n additional model input ( based on PSL recorded data ) of either fuel flow or PLA dependent on whether the model is run in open - loop or closed - loop control mode, respectively . Figure 10 b compares model - produced outputs agains t the engine sensed outputs recorded during the PSL testing. These outputs include N1, N2, P3, T3, T45, and Wf.
For the fuel flow parameter, open - loop fuel flow is omitted as this parameter is identical to PSL recorded fuel flow. Also shown in Figure 10 b is the T25 output produced by the model.
While T25 was not recorded during the test, this model - produced output is shown to illustrate it’s response dur in g the simulated rollback event.
The model is found to perform reasonably well in matching the response of PSL recorded engine data. This includes capturing the initial step change s in engine outputs that occur at t he time the ice cloud turns on , and the engine rollback event that ensues. Here , data are only plotted up to the point in Figure 10 . I nput and output parameters for Run 193 engine time when the model fails to converge and terminates execution rollback event due to operation moving outside defined component map boundaries. The LF01 engine tested in PSL actually ran fu rther Modeling of Additional Engine R ollback Events . In into rollback, but that data is not plotted here. Furthermore, the addition to Run 193, four other LF01 rollback events current capability of the model to capture the initial onset of demonstrated in PSL - 3 were selected for comparison against rollback events is deemed sufficient for its intended purpose of the developed T - MATS ALF502 - 5R engine model. These enabling the development of approaches for the incipie nt events include Run 199, Run 206, Run 570, and Run 940. Like detection and mitigation of engine icing events. The ability of Run 193 , these rollback events were each conducted at an the model to accurately capture engine behavior further into a operating point of approximately 28,000 feet, 0.5 Mach, and rollback event is not essential.
+28ºF dTamb. Model i nput parameters for the five run cases are Good agreement between the initial T3 step decrease in the shown in Figure 11 . Here, ice particle concentration and engine and model data is shown, wh ich is expected as the ice calculated ice bl ockage inputs are calculated following the particle concentration input is scaled to produce an equivalent procedure described earlier in the paper . The scale factor s response in T3. Similarly, good agreement in the N2 slope applied to the model’s ice particle concentration inputs for th e decrease of the engine and the model was s hown, which is also five cases were 0.88 (Run 193), 1.05 (Run 199), 0.90 (Run expected since the LPC % ice blockage is defi ned to produce 206), 0.99 (Run 570), and 0.86 (Run 940). In comparing the ice this result. The encouraging result is that the remaining model particle concentrations, i t can be observed that the magnitude outputs, including N1, P3, T45, and Wf, all track engine data and the duration of ice particle concentrations varied by run reasonably well. The agreement between the model and the data case. Ru n s 193 and 940 had the highest ice particle in Wf is particularly noteworthy as it confirms that the concentrations, followed by run 206 at an intermediate developed model’s closed - loop controller is able to reasonably concentration level , and run s 199 and 570 at the lowest levels .
emulate the actual engine’s fuel controller. The step decrease in T he input parameters for the various run cases were model - produced T25 outputs that occurs when the ice cloud supplied as inputs to the model while operating the model in turns is due to heat removal at the LPC exit (see Eq. ( 1 ) ) .
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 8 both open - and closed - loop control mode. Figure 12 compares mitigating the risk of engine icing. This will entail modulation model produced outputs against corresponding engine sensed of the model’s auxiliary actuators and assessing the out puts recorded during the PSL testing . Better agreement corresponding impact on icing risk. With the exception of the between engine and model outputs are found for runs 193, 199, anti - ice bleed valve, the auxiliary actuators included in the T - and 206, while runs 570 and 940 show noticeable bias offsets MATS ALF502 - 5R engine model were not modulated during between the engine and model. As described in Ref [ 24 ], LF01 the LF01 PSL engine testing. Given that engine test data was flow path hardware did experience damage during PSL testing. not available for assessing the correct implementation of these This is suspected to be a result of ice sheds impacting rotating actuators in the model, the authors opted to use outputs from a hardware. Runs 193, 199, and 206, which were the first three Honeywell provided custom er deck of the ALF502 - 5R engine.
engine icing rollback events demonstrated during the LF01 PSL The Honeywell customer deck produces steady - state engine test, occurred prior to the occurrence of this shedding damage outputs at user specified flight conditions and engine operating while Runs 570 and 940 were conducted after the damage had settings, and also permits the modulation of the four auxiliary occurred. For each of the test runs, the test commenced at the actuators included in the T - MATS mode l. To conduct this sa me target N1 speed, but a higher PLA setting and increased comparison, the customer deck was first run to a flight fuel flow (see Figure 11 ) was required to achieve this same condition of 28,500 feet, 0.5 Mach number, and dTamb = 28ºF t arget N1 speed once the engine had experienced damage. The while fuel flow was held constant. Next, the customer deck was developed T - MATS ALF502 - 5R model does not currently have re - run at the same operating condition and fixed fuel flow any inputs that allow for directly adjusting component setting while individually modulating the four auxiliary performance to reflect such degradation, and consequently the actuators. The corresponding percent change in select engine model does not match th e engine as well for runs 570 and 940. outputs due to the modulation of each actuator was recorded.
The T - MATS model was then run to the same flight condition and its actuators are individually modulated while operating the Comparison of Engine Model to Customer Deck In follow - on studies, the developed engine model will be model in open - loop control mode and holding fuel flow used to evaluate the feasibility of control - based strategies for constant. The T - MATS model and the customer deck were then Figure 11 . Model input parameters for five engine rollback events . Ice concentration and LPC ice blockage are provided as model inputs for both the open - loop and close - loop run cases, while Wf is only provided as a model input for the open - lo op run case and PL A is only provided as a model input for the closed - loop run case.
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 9 Figure 12 . Engine and model output parameters for five engine rollback events re - run under closed - loop control , and the modulation of icing risk reduction achievable through modulation of available actuators was repeated. For this evaluation, the T - MATS model control actuators.
was operated in closed - loop control mode with a fixed PLA input. For the customer deck, which does not provide direct DISCUSSION closed - loop control operating capability, closed - loop operation The T - MATS ALF502 - 5R engine model has been shown to was emulated by manually a djusting fuel flow to achieve emulate engine system - level behavior during ice crystal icing operation on the N2 governor droop line. Results comparing the tests conducted in the NASA GRC PSL facility and the steady - percent change in m odel outputs for the four actuators are state outputs of the manufacturer’s customer deck. This model shown in Figure 13 . Output parameters shown include rotor will be used as part of follow - on research focused on the speeds, fuel flow, and select pressures, temperature s, and air development and evaluation of engine icing risk detection and flow parameters at various stations of the model. The results control - based mitigation strategies. Detection strategies based show that the T - MATS model implementation of the actuat ors on both conventional and advanced sensor measurements will provides good agreement with the customer deck produced be considered. This includes monitoring for ambient conditions outputs , confirming the implementation of the actuators wi thin of known engine icing risk coupled wi th an observed change in the T - MATS model . In particular, t he closed - loop results reflect engine measurements indicative of ice particle ingestion (see the change in engine performance that can be expected when Figure 7 ). Advanced sensors may include aircraft forward modulating the various actuators while the engine is operating look ing sensors capable of detecting airborne ice crystals under closed loop - control, as it would be during actual flight and/or engine mounted sensors capable of detecting engine operation. As such, this provides information that will be used metal temperature changes or the accretion of ice on engine as part of planned follow - on studies that will assess the engine This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 10 Figure 13 . Model steady - state change to actuator modulation components. Detection techniques that offer high reliability SUMMARY (limited false alarms and missed detections) and incipient A dynamic model of the ALF502 - 5R engine has been detection capability prior to any appreciable ice accretion are developed and shown t o emulate engine system - level behavior key requirements. The mitigation research will couple the during ice crystal icing test cell evaluations as well as the model with an icing risk assessment tool to allow the benefit of steady - state outputs produced by the manufacturer’s customer control - based icing mitigation strategies to be assessed. This deck . Heat extraction effects are included in the compression work will include sensitivity studies to quantify the icing risk system of the model to reflect the heat loss the engine reducti on offered by the modulation of available engine experiences as ingested ice crystals transition from ice, to actuators as well as the development and evaluation of control - water, and vapor. Additionally, the model matches system - level mitigation logic . In developing control - based icing risk engine effects as i ce buildup occurs within the engine’s low mitigation approaches, simplistic solutions will be sought pressure compressor. The model will be used in follow - on which do not compromise the engine’s ability to deliver studies to develop and evaluate icing risk control - based requested thrust output. As part of the follow - on work , mitigation strategies.
a dditional fidelity will be added to the LPC of the model to capture the stage - by - stage operating characteristics of th e ACKNOWLEDGMENTS module and allow the risk of ice accretion within e ach stage to This work was conducted under the NASA Advanced Air be shown . The follow - on assessment will initially consider the Vehicle s Program , Advanced Air Tra nsportation Technologies baseline N2 droop line governor fuel controller and the current Project . The authors graciously acknowledge the support of suite of actuators included in the model. Additionally, Ryan May and Jeff Chapman in developing the model. The alternative fuel control strategies, such as fan speed contr ol, author s also wish to thank Honeywell Engines and the Ice will be evaluated for their robustness or susceptibility to icing Crystal Consortium for their supp ort of the engine testing that risk, and additional actuators will be evaluated for their benefit enabled this work.
of reducing the risk of engine icing.
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 11 REFERENCES [ 13 ] Feulner, M., Liao, S., Rose, B., and Liu, X. , (2015), “ Ice Crystal Ingestion in a Turbofan Engine, ” SAE 2015 [ 1 ] Mason, J. G., Strapp , J. W., Chow, P., (2006), “The Ice International Conference on Icing of Aircraft, Engines, Particle Threat to Engines in Flight,” AIAA 2006 - 206, and Structures, SAE Te chnical Paper 2015 - 01 - 2146 .
44th AIAA Aerospace Sciences Meeting and Exhibit [ 14 ] Grift, E., Norde, E., Va n der Weide, E., and Hoeijmakers, January 9 - 12, Reno, Nevada. H. , (2015), “ Computational Method for Ice Crystal [ 2 ] Mason, J. G., Grzych, M. L., Chow, P., (2008), “ Current Trajectories in a Turbofan Compressor, ” SAE 2015 P erspectives on J et E ngine P ower L oss in I ce C rystal International Conference on Icing of Aircraft, Engines, C onditions: Engine Icing,” i n Seventh AIRA Research and Structures, SAE Te chnical Paper 2015 - 01 - 2139 .
Implementation Forum Boeing. [ 15 ] Rigby, D. L. , Veres, J., and Bidwell, C. , (2015), “ Three [ 3 ] Mason, J. G. , (2007), “ Engine Power Loss in Ice Crystal Dimensional Simulation of Flow in an Axial Low Conditions, ” AERO . Vol. Qtr. 4, The Bo eing Company, Pressure Compressor at Engine Icing Operating Points, ” Seattle, WA, p gs . 12 - 17 . SAE 2015 International Conference on Icing of Aircraft, [ 4 ] Bravi n, M., Strapp, J. W., and Mason, J. G. , (2015), “ An Engines, and Structures, SAE Te chnical Paper 2015 - 01 - Investigation into Location and Convective Lifecycle 2132 .
Trends in an Ice Crystal Icing Engine Event Database, ” [ 16 ] Soeder, R.H., (1994), “NASA Lewis Propulsion Systems SAE 2015 International Conference on Icing of Aircraft, Laboratory Customer Guide Manual,” NASA - TM - Engines, and Structures, SAE Te chnical P aper 2015 - 01 - 106569, August.
2130 . [ 17 ] Griffin, T.A., Dicki, D.J., Lizanich, P.J., (2014), “PSL th [ 5 ] Grzych, M., Tritz, T., Mason, J., Bravin, M., and Icing Facility Upgrade Overview,” AIAA - 2014 - 2896, 6 Sharpsten, A. , (2015), “ Studies of Cloud Characteristics AIAA Atmospheric and Space Environments Conference, Related to Jet Engine Ice Crystal Icing Utilizing Infrared AIAA Aviation Forum, June, Atlanta, GA.
Satellite Imagery, ” SAE 2015 International Conference on [ 18 ] Grandin, A., Merle, J., Weber, M., Strapp, J. W., Protat, Icing of Aircraft, Engines, and Structures, SAE Te chnical A., and King, P. , (2014), “ AIRBUS Flight Tests in Hi gh Paper 2015 - 01 - 2086 . Total Water Content Regions,” 6th AIAA Atmos pheric [ 6 ] Grzych, M.L., “Avoiding Convective Weather Linked to and Space Environments Conference, AIAA - 2014 - 2753 .
Ice - Crystal Icing Engine Events,” AERO . Vol. Qtr. 1 , The [ 19 ] Leroy, D., Fontaine, E., Schwarzenboeck, A., Strapp, J.
Boeing Company, Seattle, WA, 20 10 . p. 23 - 28 . W., Lilie, L., Delanoe, J., Protat, A., Dezitter, F., and [ 7 ] Nilamdeen, S., Habashi , W. G., Aubé, M. S., and Baruzzi, Grandin, A. , (2015), “ HAIC/HIWC Field Campaign - G. S. , (2009), “ FENSAP - ICE: Modeling of W ater Specific Findings on PSD Micro physics in High IWC Droplets and Ice Crystals,” 1st AIAA Atmospheric and Regions from In Situ Measurements: Median Mass Space Environments Conference, AIAA - 2009 - 4128 . Diameters, Particle Size Distribution Characterist ics and [ 8 ] Liao , S., Liu, X., and Feulner, M., (2011), “ An Analysis Ice Crystal Shapes,” SAE 2015 International Conference of Turbofan Inlet Water and Ice Concentration Effects in on Icing of Aircraft, Engines, and Structures, SAE Icing Conditions, ” SAE 2011 International Conference on Te chnical Paper 2015 - 01 - 2 087 .
Aircraft and Engine Icing and Ground Deicing SAE [ 20 ] Leroy, D., Coutris, P., Emmanuel, F., Schwarzenboeck, Te chnical Paper 2011 - 38 - 0050 . A., Strapp, J. W., Lilie, L. E., Korolev, A., McFarquhar, [ 9 ] Mazzawy, R., (2011), “ Critical Point Evaluation of Ice G., Dezitter, F., and Grandin, A. , (2016), “ HAIC/HIWC Crystals in Turb ofan Engines, ” SAE Technical Paper field campaigns - Specific findings on ice crystals 2011 - 38 - 0049, SAE 2011 International Conference on characteristics in high i ce water content cloud regions,” Aircraft and Engine Icing and Ground Deicing. 8th AIAA Atmospheric and Space Environments [ 10 ] Tsao, J., Struk, P. M., and Oliver, M. J. , (2014), “ Possible Conference, AIAA - 2016 - 4056 .
Mechanisms for Turbofan Engine Ice Crystal Icing at [ 21 ] Strapp, J. W., Lilie, L. E., Ratvasky, T. P., Davison, C. R., High Al titude, ” 6th AIAA Atmospheric and Space and Dumont, C. , (2016), “ Isokinetic TWC Evaporator Environments Conference, AIAA - 201 4 - 3044 . Probe: Development of the IKP2 and P erformance Testing [ 11 ] Bidwell, C. , (2011), “ Particle Trajectory and Icing for the HAIC - HIWC Darwin 20 14 and Cayenne Field Analysis of the E3 Turbofan Engine Using LEWICE3D Campaigns,” 8th AIAA Atmospheric and Space Version 3, ” SAE 2011 International Conference on Environments Conference, AIAA - 2016 - 4059 .
Aircraft and Engine Icing and Ground Deicing SAE [ 22 ] Wolde, M., Nguyen, C., Korolev, A., and Bastian, M. , Te chnical Paper 2011 - 38 - 0048 . (2016), “ Characterization of the Pilot X - band ra dar [ 12 ] Bidwell, C. , (2012), “ Ice Particle Transport Analysis with responses to the HIWC environment during the C ayenne Phase Change for the E3 Turbofan Engine Using HAIC - HIWC 2015 Campaign,” 8th AIAA Atmospheric LEWICE3D Version 3.2, ” 4th AIAA Atmospheric and and Space Environments Conference, AIAA - 2016 - 4201 .
Space Environments Conference, AIAA - 2012 - 3037 . [ 23 ] Jorgenson , P.C.E., Veres, J.P., Wright, W.B., Struk, P.M., (2013), “A Model to Assess the Risk of Ice Accretion Due This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 12 to Ice Crystal Ingestion in a Turbofan Engine and its Effects on Performance,” NASA - TM - 2013 - 218094, November.
[ 24 ] Oliver, M.J., (2014), “Validation Ice Crystal Icing Engine Test in the Propulsion Systems Laboratory at NASA Glenn Researc h Center,” AIAA - 2014 - 289 8 , 6th AIAA Atmospheric and Space Environments Conference, AIAA Aviation Forum, June, Atlanta, GA.
[ 25 ] Flegel, A.B., Oliver, M.J., (2016), “Preliminary Results from a Heavily Instrumented Engine Ice Crystal Icing Test in a Ground Based Altitude Test Facility,” AIAA - 201 6 - 3894 , 8 th AIAA Atmospheric and Space Environments Conference, AIAA Aviation Forum, Jun e, Washington, DC .
[ 26 ] Chapman, J.W., Lavelle, T.M., May, R.D., Litt, J.S., Guo,T - H., (2014), “Toolbox for the Modeling and Analysis of Thermodynamic Systems (T - MATS) User’s Guide,” NASA/TM — 2014 - 216638, January.
[ 27 ] Veres, J.P., Jorgenson, P.C.E, Jones, S.M., (2014), “Modeling the Deterioration of Engine and Low Pressure Compressor Performance During a Roll Back Event Due to Ice Accretion,” AIAA 2014 - 3842 , 50th AIAA/ASME/SAE/ASEE Joint Propulsion Conference Cleveland, OH, July.
[ 28 ] Veres, J.P., (2009), “Axial and Centrifugal Compressor Mean Line Flow Analysis Method,” AIAA - 2009 - 1641, 47th Aerospace Sciences Meeting, Orlando, Florida, January 5 – 8.
[ 29 ] Jorgenson, P.C.E., Veres, J.P., May, R.D., Wright, W.B., (2011), “En gine Icing Modeling and Simulation (Part I): Ice Crystal Accretion on Compression System Components and Modeling its Effects on Engine Performance,” 2011 - 38 - 0025, SAE International Conference on Aircraft and Engine Icing and Ground Deicing, Chicago, IL, Ju n 13 - 17. doi:10.4271/2011 - 38 - [ 30 ] Goodwin, R.V., Dischinger, D.G., (2014), “ Turbofan Ice Crystal Rollback Investigation and Preparations Leading to Inaugural Ice Crystal Engine Test at NASA PSL - 3 test Facility ,” AIAA - 2014 - 2895, 6th AIAA Atmospheric and Space Environments Conference, AIAA Aviation Forum, June, Atlanta, GA.
[ 31 ] Soeder, R.H., (1994), “NASA Lewis Propulsion Systems Laboratory Customer Guide Manual,” NASA - TM - 106569, August 1.
[ 32 ] Thomas, Q.P., (2015), “NASA Glenn Propulsion Systems Lab (PS L) Icing Facility Update,” SAE 2015 International Conference on Icing of Aircraft, Engines, and Structures, June 22 – 25, Prague, Czech Republic.
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited. 13