TOPIC: LPA-01-93

CALL: H2020-CS2-CFP11-2020-01

PROJECT Improved ejector models for higher-performance engines

Description

Ejectors (jet pumps) convert the pressure energy of a fluid/stream into kinetic energy to entrain and compress a lower pressure fluid/stream via a mixing process between both streams. The first goal of the EU-funded EJEMOD project is to set up and run steady-state and transient experiments and computational fluid dynamics simulations of different ejector geometries. Based on the generated data, the second goal is the development and validation of new 0D/1D on the Dymola platform – a commercial modelling and simulation environment based on the open Modelica modelling language. Steady-state models will describe on-design and off-design modes thermal performance. Transient models will be used to optimize the system control, through the analysis of the transitions between two operating points or during transitions towards abnormal functioning. Surrogate modelling methodologies based on deep learning will also be used as a complementary approach. Such robust 0D/1D models covering the full operating envelope will allow future optimization of the ejector implementation in the aircraft bleed-air system, thus improving the net performance of the engine.

Objective

EJEMOD project focuses on the study of the physics and operation of new ejector prototypes proposed by the Topic Manager. The main objective of the project is to acquire knowledge on the functioning of these jet pumps, both in their steady (on and off-design modes) and transient behaviour (transition between operating points after a (sudden) change of boundary conditions). The wide working range analysed and the focus on the transitional phases will serve to acquire knowledge and transfer it into accurate and robust Dymola libraries, applicable (for various purposes) to the modelling of ejector devices whose function may also be different to those previously described. Hence, the specific objectives of EJEMOD are the following:
• Perform a CFD analysis of the proposed ejectors, studying all the on-design and off-design modes indicated by the Topic Manager. The analysis will be performed by means of steady-state simulations (WP 1.1) covering on-design (double-choked) and off-design (single-choked or unchoked, malfunction) operation. Moreover, a transient model will be employed to analyse in detail transitions between two operating points (WP 2.1). Both steady-state and transient CFD models will be validated by comparison with experimental results.
• Perform an experimental characterization of the ejectors, following a design of experiments (DoE) approach which covers the operational envelope indicated by the Topic Manager. Testing will be carried out in static conditions (WP 2.1), by setting primary and secondary pressure and temperatures, as well as back-pressure and in time-varying conditions (WP 2.2), by setting time-dependent functions for primary, secondary and back-pressures.
• Development and validation of 0D-1D numerical models on the Dymola platform for the simulation of ejectors, to be used in both static (Development WP 1.3, Validation WP 3.1) and dynamic scenarios (Development WP 2.3, Validation WP 3.2). Surrogate modelling methodologies based on deep learning will be assessed as well.

CONSORTIUM MEMBERS

WORK PERFORMED / RESULTS

The work performed during the project has covered the analysis of the ejectors first under static operation, and in a second stage focusing on the operating transitions. On both stages, a combination of CFD and experimental studies have constituted the methodology to get a detailed and consolidated insight on the ejectors behaviour.

Figure 1 – Ejector operational map and operating modes identification.

During the static analysis, a new operational map (Figure 1) has been generated, which
enabled to categorize different situations of interest (modes and transitions), while covering the physics involved (choked, un-choked and reverse flows). Following these reference Cases, a very intensive static CFD simulation campaign (Figures 2 and 3) has been carried out, which has revealed interesting physics and supported the development and calibration of the 0D static model (Figure 4).

Figure 2 – CFD mixing zone analysis for different operating modes.
Figure 3 – Full ejector CFD Mach contour plot showing the characteristic shock train.
Figure 4 – Static 0D model comparison vs. CFD for normal (left) and close secondary abnormal (right) scenarios.

In parallel to the numerical activities, an experimental set-up has been built (Figure 5), capable of providing air at the expected mass flow, pressure and temperature ranges. The ejector prototypes have also been manufactured and instrumented to gather temperature, pressure and mass flow at each port, together with a very detailed pressure profile in the mixing chamber. An extensive static test campaign has been performed, collecting data for the referred Cases of interest for the three ejectors of the project. Using this data, a validation activity has been done comparing successfully the 0D model predictions with the experimental results.

Figure 5 – Overall view of the ejector experimental set-up. Left: flow control and temperature section. Right. Ejector with implemented temperature and pressure sensors.

For the second stage of the project, the focus was reoriented towards dynamic analysis of the ejectors. A CFD transient campaign has allowed to improve the understanding of the physics and time response characteristics for the situations of interest at system level. Different scenarios have been studied, like step or ramp-like evolution of values at the ports at different pressure levels.

These transient phenomenological aspects have also been verified by specific experimental studies, while describing the transitions of interest. Adaptation of the test bench has been done accordingly by adding fast valves and hot- wire velocity measurement on the different streams (Figure 6). From all this knowledge and data, a 0D tool has been developed, validated (Figure 7) and integrated in Dymola (Figure 8) for static and transition scenarios. A parallel action has generated a surrogate model trained with the 0D model which can be implemented in Modelica as a black box with an extended robustness perimeter.

Figure 6 – Dynamic test bench particularities: fast valves, hot-wire sensors.
Figure 7 – Dynamic/transition test (closing/opening secondary) and comparison with the 0D model prediction.
Figure 8 – Modelica model set-up: ejector model and integration within a representative system composed of three valves and a downstream heat exchanger.

As a summary of the results, the project has generated a huge amount of new scientific data and knowledge on the behaviour of air-air ejectors under a wide range of situations, both from static and dynamic perspective. The final 0D model has been progressing continuously to finally reach a transversal capacity to cover the whole operational envelope with a reasonable accuracy after an extensive calibration and validation campaign. As commented, the successful generation of the surrogate model is acting as a robustness-oriented twin.

Publications / Dissemination

  1. Air Ejector Analysis in Normal and Abnormal Modes, Oriented to Control Purposes in Aircraft Systems
    SCHILLACI Eugenio, OLIET Carles, VEMULA Jagadish Babu, DUPONCHEEL Matthieu, BARTOSIEWICZ Yann, PLANQUART Philippe
    EUCASS Conference 2022 Proceedings

  2. Air Ejector Analysis in Extended Operational Perimeter for Control Oriented Simulation of Bleed-Air Aircraft Systems
    Schillaci Eugenio, Oliet Carles, Vemula Jagadish Babu,, Duponcheel Matthieu, Bartosiewicz Yann, Planquart Philippe
    12th EASN International Conference on Innovation in Aviation & Space for opening New Horizons

  3. NUMERICAL STUDY OF THE COMPLETE OPERATING MAP OF EJECTORS IN ULTRA HIGH BYPASS RATIO ENGINE BLEED SYSTEMS
    Jagadish Babu Vemula, Matthieu Duponcheel, Yann Bartosiewicz
    8th Thermal and Fluids Engineering Conference (TFEC)

  4. CALIBRATION OF A 1D EJECTOR MODEL FROM FULL-SCALE EXPERIMENTS
    Jan Van den Berghe, M. Delsipee, Philippe Planquart, Yann Bartosiewicz, Miguel A. Mendez
    8th Thermal and Fluids Engineering Conference (TFEC)

  5. Numerical investigation of an air-air ejector behavior in the transient operating conditions of Ultra High Bypass Ratio (UHBR) engine bleed systems
    Vemula Jagadish Babu, Duponcheel Matthieu, Bartosiewicz Yann
    22nd Computation Fluids Conference (CFC2023)

Contact

Dr. Carles Oliet

Telf: 34 93 739 87 77

e-mail: carles.oliet@upc.edu

Dr. Joaquim Rigola

Telf: 34 93 739 80 66

e-mail: joaquim.rigola@upc.edu

Centre Tecnològic de Transferència de Calor (CTTC)

Departament de Màquines i Motors Tèrmics
Universitat Politècnica de Catalunya (UPC)
Escola Superior d’Enginyeries Industrial, Aeroespacial i Audiovisual de Terrassa (ESEIAAT)
C/ Colon, 11 E-08222 Terrassa (Barcelona), Spain

 web: http://www.cttc.upc.edu