czech aerospace - Výzkumný a zkušební letecký ústav

Transkript

obal32006CLKV.qxd
15.11.2006
11:56
StrÆnka 2
Contents / Obsah
Experimental Test System for Fibrous
Thermosetting Composites Breakdown
Program SPAD and its Use in Noise
Abatement of Propeller Driven Airplanes
Composite Airplane Control Rod with
Metel End Joint
The Aerodynamic Design
of Cold Jet
Preliminary Testing of Friction
Stir Welding
Evaluation Methodology of Research and
Development Projects
Fatigue Testing and Analysis of VUT 100
Aircraft Landing Gear
FMECA of MAC-03 Electronic Equipment
Optimization Methods for 2D Flow Passage
Design of Axial Compressor
Pokročilá detekce, izolace a přizpůsobení
chybných údajů od senzorů pomocí
umělé neuronové sítě
Experimentální zkušební systém pro
pyrolýzní vláknových termosetických
kompozitních materiálů
LETECK Ý
zpravodaj
In this issue:
Program SPAD a jeho užití při snižování
hluku vrtulových letadel
Czech Aerospace
Research Centre
Zkoušky kompositního táhla řízení
letounu s kovovou koncovkou
CLKV
Návrh aerodynamického řešení proudové
cesty studeného propulsoru
Proceedings of the
6th Annual
Workshop held at
Prague, Czech
Republic
Úvodní zkoušky frikčního
svařování
Hodnoticí metodika projektů
výzkumu a vývoje
November 2 to 3,
2006
Únavové zkoušky a výpočty životnosti
podvozku letounu VUT 100
Centrum leteckého
a kosmického
výzkumu
FMECA elektronického vybavení MAC-03
Optimalizační metody pro 2D proudění
axiálním kompresorem
Composite Propeller Spinner Nose Cone
Made by LF-Technology
Výroba kompositového krytu vrtulové
hlavy technologií LF
Optimalization of Stiffened Panel with the
Help of Mathematical Programming
Optimalizace vyztuženého potahu pomocí
matematického programování
Verification Ground Frequency Tests
on HPH G-304 CZ Sailplane
Proceedings
Ověřovací frekvenční zkoušky na
kluzáku HPH G-304 CZ
Sborník vybraných
referátů
přednesených na
6. ročníku
semináře CLKV
Prague
2006
Praha, 2. — 3.
listopadu 2006
E
CK
Ý Ú
S
TA
V
© C Z E C H AE R O S PAC E
M A N U FAC T U R E R S A S S O C IAT I O N
CLKV
V
U
O
Advanced Detection, Isolation and
Accomodation of Sensor Failures by
Means of Artificial Neural Network
Měření letových veličin
v akrobatických manévrech
CZECH
AEROSPACE
T
Measurement of Aerobatic
Flight Characteristics
CFD simulace kvality mikroklima
v kabinách malých dopravních letounů
LE
CFD Simulation of Quality of Environment
in Small Transport Airplane Cabins
Slovo úvodem k semináři CLKV 2006
N o v e m b e r
Introductory Lecture to the ARC 2006
2 0 0 6
ISSN 1211—877X
T
B
RN
No. 3 / 2006
obal32006CLKV.qxd
15.11.2006
11:56
StrÆnka 4
Composite Airplane Control
Rod with Metal End Joint
Colour illustrations to the article published on pages 19-22.
Figure 3 — Loading
machine INOVA ZUZ 200
and loading fixture
CZECH
AEROSPACE
P r o c e e d i n g s
J OU R N A L
F O R
C Z E C H
AE RO S PAC E
R E S E A R C H
Figure 2 — Specimen of the VL-3 plane
control rod and bolt test assembly
LETECK Ý
Figure 7 — Test specimen with first filament layout ripped off
zpravodaj
VÝZKUMNÝ A ZKUŠEBNÍ LETECKÝ ÚSTAV, a.s.
Editorial address:
Aeronautical Research and Test Institute / VZLÚ, Plc.
Beranových 130, 199 05 Prague 9, Letňany
Czech Republic
Phone.: +420-225 115 223, Fax: +420-869 20 518
Editor-in-Chief:
Editor & Litho:
Ladislav Vymětal (e-mail: [email protected])
Stanislav Dudek (e-mail: [email protected])
Editorial Board:
Chairman:
Vice-Chairman
Members:
Publisher:
Printing:
Milan Holl, President ALV, Managing Director VZLÚ
Vlastimil Havelka, ALV
Jan Bartoň, Tomáš Bělohradský, Vladimír Daněk, Jiří Fidranský,
Luboš Janko, Petr Kudrna, Pavel Kučera, Oldřich Matoušek,
Vojtěch Nejedlý, Zdeněk Pátek, Antonín Píštěk
Czech Aerospace Manufacturers Association / ALV, Prague
Studio Winter Ltd. Prague
Figure 6 — Ripped test specimen
The Aerodynamic Design of a Cold Jet
Figure 1 — Angles of
attack of the rotor
blades, first geometry settings
Figure 2 — Angles of attack of
the rotor blades,
twisted duct
Published with the assistance of Czech Ministry of Education, Youth and Sports (MŠMT).
Subscription and ordering information available at the editorial address. Legal liability for published
manuscripts’ originality holds the author. Manuscripts contributed are not returned automatically to
authors unless otherwise agreed. Notes and rules for the authors are published at our Internet pages
http://www.vzlu.cz/.
Czech AEROSPACE Proceedings
Letecký zpravodaj
3/2006
© 2006 ALV / Association of Aviation Manufacturers, All rights reserved. No part of this publication may
be translated, reproduced, stored in a retrieval system or transmitted in any form or by any other means, electronic,
mechanical, photocopying, recording or otherwise without prior permission of the publisher.
ISSN 1211 - 877X
Figure 4 — Contours of tangential velocity immediately behind the
rotor stage
Figure 5 —
Angles of
attack of the
rotor blades,
twisted duct
with stator
vanes
1
L E T E C K Ý Z P R AV O D A J
3/2006
Contents / Obsah
2
Aerospace Research Centre — Opening Speech at the 2006 ARC Workshop
Úvodní přednáška na Semináři CLKV 2006
Prof. Ing. Antonín Píštěk, CSc.
3
CFD Simulation of Quality of Environment in Small Transport Airplane Cabins
CFD simulace kvality mikroklimatu v kabinách malých dopravních letounů
Ing. Jan Fišer, Prof. Ing. Miroslav Jícha, CSc.
6
Measurement of Aerobatic Flight Characteristics
Měření letových veličin v akrobatických manévrech
Doc. Ing. V. Daněk, CSc., Ing. M. Kouřil, PhD., Ing. R. Šošovička, PhD.
8
Advanced Detection, Isolation and Accomodation of Sensor Failures by Means of
Artificial Neural Network
Pokročilá detekce, izolace a přizpůsobení chybných údajů od senzorů pomocí umělé neuronové sítě
Ing. Jaromír Lamka, CSc.
12
Experimental Test System for Fibrous Thermosetting Composites Breakdown
Experimentální zkušební systém pro pyrolýzní rozklad vláknových termosetických kompozitních materiálů
Miroslav Valeš, Ing. Bedřich Štekner
14
Program SPAD and its Use in Noise Abatement of Propeller Driven Airplanes
Program SPAD (ver. 0.6) a jeho užití při snižování hluku turbovrtulových letadel.
Ing. Tomáš Salava, Dr.Sc.
19
Composite Airplane Control Rod with Metal End Joint
Zkoušky kompositního táhla řízení letounu s kovovou koncovkou
Ing. Tomáš Marczi, Ing. Jiří Stehlík, Ing. Karel Blahouš
22
Návrh aerodynamického řešení proudové cesty studeného propulsoru
Ing. Erik Ritschl, Ing. Robin Poul
24
Preliminary Tests of Friction Stir Welding
Úvodní zkoušky frikčního svařování
Ing. Petr Bělský
26
Evaluation Methodology of Research and Development Projects
Hodnoticí metodika projektů výzkumu a vývoje
Ing. Klára Grammetbauerová
34
Fatigue Testing and Analysis of VUT 100 Aircraft Landing Gear
Únavové zkoušky a výpočty životnosti podvozku letounu VUT 100
Ing. Petr Augustin, Ph.D., Ing. Martin Plhal, Ph.D., Ing. Jan Šplíchal
36
Ing. Milan Merkl, CSc
40
Optimization Methods for 2D Flow Passage Design of Axial Compressor
Optimalizační metody pro 2D proudění axiálním kompresorem
Ing. Jan Tůma
42
Composite Propeller Spinner Nose Cone Made by LF-Technology
Výroba kompositového krytu vrtulové hlavy technologií LF
Ing. L. Křípal
45
Optimalization of Stiffened Panel with the Help of Mathematical Programming
Optimalizace vyztuženého potahu pomocí matematického programování
Ing. Miroslav Pešák, Prof. Ing. Antonín Píštěk, CSc.
50
Verification Ground Frequency Tests on HPH G-304 CZ Sailplane
Ověřovací frekvenční zkoušky na kluzáku G-304 CZ
Ing. Karel Weigel
2
C Z E C H A E R O S PA C E P R O C E E D I N G S
Aerospace Research
Centre
Opening Speech at the 2006 ARC
Workshop
Prof. Ing. Antonín Píštěk, CSc, Director, Institute of Aerospace Engineering,
Brno University of Technology
Ladies and Gentlemen, Dear Colleagues,
Photo: Evektor
This is already the second workshop we have had this year
under the heading of the new Aerospace and Research Centre, which started its activities in 2005. The history, achievements and appreciation of the previous Centre have proved
the unique and efficient composition of its working teams,
bringing together the research and human resources of technical universities, an aeronautical research body and the aviation industry. My precedent opening lectures were aimed at
stressing the Centre’s role in the development of aerospace
and assessing its technical, economic and personnel aspects.
Today I would like to draw your attention to the results the
Centre has achieved in the process of taking on projects of
the Czech aviation industry and European research programmes. Being successful is among other things linked with
a new organisation of the research projects into a so called
matrix structure. It is easier to work with, featuring better
communication between ARS workplaces and more accommodation to requirements of the aviation industry and projects organized by the EU. To increase the efficiency of the
new organisational structure, an ARC coordinating team of
young workers has been set up, showing active and dynamic
approaches. Enlargement of the Centre’s council to adopt
representatives from the industry has also been an important
move. Discussions taking place in this so called Big Council
proved to be very effective, encouraging integration of ARC
programmes into R&D industrial projects and programmes of
the EU. In this way the Centre has become a partner to the
industry that has something to say to its problems. In the
2005 ARC workshop we waited eagerly for the first European project to be coordinated by the Czech Republic. Today
we can say that the CESAR project (Cost Effective Small
AiRcraft) has got much support and green light to get started.
In addition to VZLU, the project coordinator and a member
ARC research team, there are other 40 organizations from all
of Europe, including the Institute of Aerospace Engineering
of Brno University of Technology. CESAR’s importance has
been mentioned many times recently. Apart from R&D point
of view, this project also raises prestige of the Czech Republic in one of its traditional industries - the category of general
aviation. The Brno Institute of Aerospace Engineering was
also successful, along with other industrial firms, in other
projects of the third call of 6th Framework Programme.
These projects are very complicated and difficult to handle
and the Centre is expected to use the most of its knowledge
and skill to master them. Let us mention e.g. a project ENFICA-FC aimed at designing a propulsion system based upon
hydrogen fuel cells. This shows that the Centre has qualified
as an equal partner to the Czech aviation industry, gaining
reputation in the European territory. Last but not least, there
are projects co-funded by the Ministry of Industry and Trade.
With the first flight of the VUT 100 airplane taking place
recently, powered by a 300 hp engine, the VUT 100 project
is drawing near to certification. The EV-55 airplane, a project
of EVEKTOR Company, represents a key programme in the
Czech aeronautical industry and it is my pleasure to say that
ARC participates in a number of R&D tasks involved.
Project Staffing
Age Diagram of ARC Employees: [21 to 30 years — 30; 31 to 40 years
— 18; 41 to 50 years — 6; 51 to 60 years — 8; 61 years and over — 9]
60- let - 9
51-60 let - 8
21-30 let -30
41-50 let - 6
31-40 let -18
Vìkový prùmìr 38 let, celkem 71
Conclusion
I cannot help repeating once more that aerospace research as
indispensable support for the aerospace industry exerts
a major influence on the development of technology and industrial innovation. This is evidenced by a close link of ARC
work with the European projects that highlight the role of science in aerospace research. The existence and results of ARC
provide the guarantee of getting funds from European resources. The labs with modern equipment, the use of research
results in industrial applications, team work of young and
high-spirited researchers make one believe that
ARC’s successful activities will continue till 2009 and on.
3
3/2006
CFD simulace kvality mikroklimatu v kabinách malých
dopravních letounů
Ing. Jan Fišer, Prof. Ing. Miroslav Jícha, CSc, Brno University of Technology
This paper deals with CFD simulation of chosen environment parameters in small airplane cabins (ARC task A6 - Prediction of
inner environment in airplane cabin). Newly - designed airplane EV-55 falls in this category, so the cabin geometry of the airplane
was chosen as a referential model of the interior. Two groups of cases were studied, group A - empty cabin, group B - completely
occupied cabin. A simplified model of human body was created and placed into models (group of cases B). Indoor environment
quality indexes Age of Air (AA) and Draft Risk (DR) were calculated from airflow pattern and temperature field. The model was
created using Star-CD ver. 3.26 (pre/postprocessor, solver) software package.
Příspěvek se zabývá CFD simulací vybraných mikroklimatických parametrů v kabinách malých dopravních letadel (výzkumný
úkol A6 - Predikce vnitřního prostředí v kabinách letadel). Do této kategorie spadá i nově projektovaný letoun EV-55, který byl
autory zvolen jako referenční model uspořádání interiéru kabiny. Byly vytvořeny dvě skupiny modelových situací, skupina A kabina bez cestujících, skupina B - kabina plně obsazena cestujícími. Pro skupinu modelových situací B byl vytvořen geometrický
model se zjednodušenými figurínami reprezentujícími cestující. Na základě znalosti teplotních a proudových polích byla určena
pole rozložení indexů AA (Age of Air) a DR (Draft Risk) popisujících kvalitu mikroklimatu a tepelnou pohodu v kabině.
Modelování a simulace byly provedeny pomocí softwaru Star CD ver. 3.26 (solver, automasher, pre/postporcessing).
Keywords: Cabin environment quality, CFD simulation, ECS, Air-conditioning, Age of Air, Draft Risk
Introduction
Modern commercial airplanes can change their altitude by thousands of
meters in tens of minutes. Parameters of ambient environment (e.g.
temperature, pressure, relative humidity etc.) vary rapidly with these
changes. Human is adapted to environment on earth surface but not to
such quick changes of environmental parameters. So cabins of airplanes must be equipped with a complex system to control the environment inside cabins. This system is called environmental control system
— ECS. The cabin of an airplane is also an environment with limited
volume, in which passengers directly influence the air quality [1]. The
ECS must provide enough fresh air to sufficient ventilation and control
heating/cooling of cabin to maintain thermal comfort. Indoor air quality and thermal comfort are two basic conditions offering a comfortable
environment (without health risk) and enabling intensive mental work.
Indoor environment quality indexes
The operation target of an ECS is complete replace of the internal
cabin air within a prescribed time. For this purpose bleed air is fed into
the cabin from outside. It is generally recognized that the perception
of cabin air quality depends on parameters such as relative humidity,
temperature, convection velocity of the air (draft risk), age of air, diffusion of gases (CO2, CO, O3) and presence of foreign particles. Also
actual airflow patterns will influence the passenger's perception of
thermal comfort [1].
Mean age of air
One of the key parameters in the determination of quality of environment is the mean age of air. The mean age of air τB is defined as the
average time for all air molecules to travel from the supply inlets to
the point B (see Fig. 1). It can be derived from the measured transient history of tracer-gas concentration.
∞

C (t ) 
τ B = ∫ 1 − B dt
C B (∞ ) 
0
The mean age of air is governed by a transport equation [2]
∂
∂
∂ 
∂ 
Γτ
( ρτ ) +
( ρu jτ ) =
+ρ

∂t
∂x j
∂x j  ∂x j 
Draft risk
One of the most critical factors is draft. Many people at low activity
levels (seated/standing) are very sensitive to air velocity, and therefore draft is a very common cause of occupant complaints in ventilated
and air-conditioned room. Fluctuations of the air velocity have a significant influence on a person's sensation of draft. The fluctuations
may be expressed either by standard deviation of the air velocity or by
the turbulence intensity Tu, which is equal to standard deviation divided by the mean air velocity w.
∑ (w −w )
2
i
i
Fig. 1 — Age and residence time in an enclosure
s
Tu = =
w
w=
n −1
w
w
∑ i
i
n
The percentage of people feeling draft may be estimated from the
equation:
DR = (34 − Ta )( w − 0,05) 0, 62 (0,37 ⋅ w ⋅ Tu + 3,14)
Forw m.s-1 usew = 0,05 m.s-1. For DR > 100 % use DR = 100 %.
The model applies to people at light, mainly sedentary activity whit
a thermal sesation for the whole body neutral [3], [4].
4
Studied cases
Two groups of cases were studied, group A — empty cabin, group B
— fully occupied cabin, each group containing three cases (see Tabs.
1 and 2). Case A1 represents ground operation and extreme design
climate conditions. Air temperature is taken from the aircraft operational envelope and altitude of a typical flight profile. Case A2 represents ground operation with standard design climate conditions and
case A3 represents flight operation. Air temperature was taken from
international standard atmosphere (ISA). Cases in group B are for the
same climate conditions and operation conditions as those of group
A but the cabin is completely occupied.
Tab. 1 — Cases description, group A — empty cabin
Fig. 2— Model of cabin
with manikins
Tab. 2 — Cases description, group B — full cabin
CFD model description
Dimensions of model geometry were based on real cabin dimensions.
The whole volume of the geometry model was sub-divided into a grid
of cells. Some parts of the grid were divided more, because higher
gradients of temperature and velocities were supposed there (e.g. near
inlets, near walls etc.). This enabled us to acquire better results in
these parts and also supported convergence of solution. A simplified
model of human body was created and placed in geometry of models
of group B (see Fig. 2). Turbulence was modeled by k-ε low Reynolds number turbulence model so that eight layers (thickness of
layer 1.88 mm) of cells were generated near the walls. Thickness of
layer result from y+ (dimensionless thickness of laminar sub layer),
which was c. 1 for all models and this value is necessary for correct
function of chosen turbulence model [5]. Final CFD model A contains 1,800,000 cells and model B 3,200,000 cells. All simulations were
computed as static simulations. The CFD model was created using
Star-CD 3.26 software package.
Fig. 3 — Boundary condition placement. Description is in Tab. 3
Tab. 3 — Description of Boundary conditions
Boundary conditions
Right specification of appropriate boundary conditions is a very
important part of every CFD simulation especially in case of indoor
environment simulation. The main goal in our simulation was to find
right specification of thermal properties of cabin constructions, volume and temperature of supplied air and surface temperatures of manikin parts. Description of boundary conditions used in models is in
Table 3 and Figure 3. Surface temperatures of manikin parts are in
Table 4, for placement of boundaries see Figure 4.
Discussion of results
An example of the age of air pattern on right side of the cabin is in
Figure 5. The lowest value of age of air is in front of the cabin and
rises in front-aft direction (see Figs. 5, 6, 7 and 8). It is effect of frontaft airflow which can carry pollutants and particles from front to aft
Tab. 4 — Boundary conditions on manikin [6]
section of the cabin. This front-aft airflow is due to configuration of
air distribution system (nine inlets and just one outlet [7]). The draft
risk pattern is in Figures 9 (left side of the cabin) and 10 (third row,
cross section). The percentage of dissatisfied passengers due to draft
is low in space near the seats but rises in region near cabin walls (all
5
3/2006
Fig. 4 — Boundary conditions on
manikin. Description is in Tabs. 3
and 4
Fig. 8 — Age of air [s], cases B, aisle, front-aft direction
Fig. 5 — Age of air [s], right side of the cabin, case A2
Fig. 9 — Draft risk [%], head high, case A1
Fig. 6 — Age of air [s], right side of the cabin, case B3
cases — see Fig. 9), floor (case A2, B2 - see Fig. 10) and in aisle (all
cases — see Fig. 9).
Conclusion and future work
The results of the study indicate that the draft risk index in most part
of the cabin is very low. A higher draft risk index is just in parts which
are close to the walls (all studied cases) and floor (in cases when temperature of distributed air is similar to the cabin temperature). The age
of air index is a very useful index to visualize efficiency of ventilation
in the cabin. The results of study indicate that age of air index is rising
in front-aft direction. It is the effect of front-aft airflow which can
carry pollutants and particles from front to aft section of the cabin.
This front-aft airflow is effect of configuration of air distribution system and the possible way of eliminating this problem is to redesign the
air distribution system (more outlets).
Fig. 10 — Draft risk [%], third row, cross section, case B2
Future work will be focused on application of novel air distribution
system as displacement ventilation or personalized ventilation in small
airplane cabins and also on application of a comfort prediction diagram.
References:
[1]
[2]
[3]
[4]
[5]
[6]
[7]
Fig. 7 — Age of air [s], cases A, aisle, front-aft direction
Kok J. C., Van Muijden J., Spekreijse S. P.: Enhancement of Aircraft
Cabin Comfort Studies by Coupling of Human Thermoregulation
With Radiation and Turbulent Convection; EUROMECH Colloquium 471, Göttingen 2005
Li X., Jiang Y.: Calculation of age-of-air with velocity field; PostIAQ 96 Seminary, Beijing, 1996
Goodfellow H., Tähti E.: Industrial Ventilation, Academic Press, San
Diego, 2001
Fanger O. P., Melikov A. K. , Hanzawa H., Ring J.: Turbulence and
Draft; ASHRAE Journal 1989; 31(7)
Pennecot J., Ruetten M., Bosbach, Wagner C.: Numerical Simulations of Mixed Turbulent Convection in a Generic Aircraft Cabin;
EUROMECH Colloquium 471, Göttingen, 2005
Mlčák R. , Pavelek M., Tománková K. , Dýrová J.: Možnosti využití termovizní kamery v posilovně; Fluid Mechanics and Thermomechanics, Slovak University of Technology in Bratislava, Bratislava,
2006
Fišer J.: Computer Simulation of Cabin Environment in the Evektor
EV-55 Aircraft; Letecký zpravodaj, Vol. 2005, No. 3, Czech Aerospace Manufacturers Association, Praha 2005
6
Measurement of Aerobatic Flight
Characteristics
Měření letových veličin v akrobatických manévrech
Doc. Ing. V. Daněk, CSc., Ing. M. Kouřil, PhD., Ing. R. Šošovička, PhD.,
Institute of Aerospace Engineering, Brno University of Technology
The article describes progress, i.e. preparation, execution and evaluation of flight measurements of aerobatic manoeuvres of the
L-13AC Blaník glider.
Článek popisuje průběh, tj. přípravu, provedení a vyhodnocení, letových měření akrobatických manévrů kluzáku L-13AC Blaník.
Keywords: flight measurement, aerobatic flight, flight test, mechanics of flight.
Introduction
Determination of flight parameters in non-steady turns (e.g. aerobatic
manoeuvres, stalls, spins) is one of the most complicated tasks within
the flight measurement, from the point of view of character of performance and aircraft equipment. Limitary parameters of flight for
components of aerobatic manoeuvres are integral part of technical
documentation of corresponding aircraft category. According to the
Airworthiness standards, aerobatic and training aircraft classes have to
have the ability to make safe aerobatic turns as specified in the type
certificate.
There are measurements, taken within the flight measurements at
the Institute of Aerospace Engineering in Brno, that made it possible
to monitor the whole aerobatic configuration of a two-seat, all-metal
glider L-13AC Blaník (Fig. 1). During the test flights, basic aerobatic
elements such as loop, stall turn, inverted flight, roll, etc., were tested
on the sailplane. It was necessary to equip the aircraft with the measurement technology capable of scanning a large number of flight
quantities for example attitude angles, angle of attack, sideslip angle,
airspeed and flight altitude, etc.
pilot place. Aircraft control system was taken to be rigid. The data
logging and control unit along with static pressure transducer (flight
altitude determination), differential pressure transducer (flight speed
determination) and optical gyroscope (scanning of load factor, attitude angles and angular rate) were put behind the back pilot seat (Fig.
3). The power supply to optical gyroscope and data logging and control unit were hitched in the glider baggage room. An atmosphere temperature sensor was placed underneath the left wing (no sunlight
influence on the sensor).
The whole measurement set was activated, its functionality was
checked and the calibration of some sensors was done (surface
deflection [Fig. 4], temperature sensor, etc.). Before the flight itself,
the data logging and control unit along with the optical gyroscope was
set for aerobatic elements regime. Recording frequency of all monitored quantities was set on the data logging and control unit for 100
samples per second.
Preparation of measurement
It was essential to equip the glider with a number of apparatuses and
sensors for the recording of flight characteristics of aerobatic manoeuvres. One of the first tasks was to find the position and install a pitotstatic probe with vanes to scan total and static pressures, angle of
attack and side slip angle. The location of the pitot-static probe was
conditioned by the least impact of the glider on read quantities (glider
impact on flow field in its surroundings). In the end, after consideration of possible location, the pitot-static probe was fitted to the glider
nose.
Its attachment required an assembly fixture (a fiberglass cone), which
would enable easy installation and dismantling of the pipe (Fig. 2).
The sensors of aileron, rudder and elevator deflection were installed on longitudinal and lateral-directional control system in the back
Figure 1 — L-13AC Blaník glider (OK-1730)
Figure 2 — Production of the ”cone“
7
Figure 3 — Measuring installation
3/2006
Figure 4 — Calibration of aileron deflection
Figure 5 — The aerobatic manoeuvres
Loop
Stall turn
Roll
Spin
Measuring process
Figure 6 — Data and functional checking
Two measurement flights (27 minutes and 24 minutes) were made.
Aerobatic elements were active for 6.5 min and 5.5 min of flight.
There was only one pilot in the glider because of the measurement
equipment installation. The first aerobatic set of manoeuvres included
especially loops and stall turns. The second flight included right spin,
roll, half-roll-inverted flight-half-loop and loops (Fig. 5). After the
first flight, the measured data and the function of measurement set
were rechecked (record setting, etc.), (Fig. 6). During the flight,
ground personnel communicated with the pilot over the radio because of recording the sequence and time periods of individual aerobatic
elements. During the measured data processing, the notes were used
for easier identification of a given aerobatic element in measurement
record.
Measurements results
After both flights, the gained data were checked to see whether it was
necessary to repeat the flights (technical defect, etc.). Relevant software is added to the data logging and control unit not only for preflight timing but also for a possibility of data processing. The first step
to primary data processing was their calibration according to the calibrating curves and required units conversion. The calibrating curves
added to software were obtained from the maker of apparatuses and
sensors, or we did the calibration ourselves (ambient temperature, surface deflection, etc.). These processed data (manoeuvre interval) were
checked visually for possible non identifiable mistakes in the record.
The next step was to use filters (part of the software unit) that helped
250
150
100
200
true airspeed [km.h -1]
altitude [m]
50
0
-50
-100
-150
150
100
50
-200
0
-250
0
50
100
150
200
250
300
350
400
450
horizontal distance [m]
Figure 7 — Trajectory of loops
500
550
600
0
5
10
15
20
25
30
35
time [sec]
Figure 8 — Course of true airspeed
40
45
50
8
4
nz
3,5
nx
3
2,5
load factor [-]
to eliminate coincidental mistakes of measurement and possible rustling of records.
Data evaluation was aimed mainly at obtaining quality time behaviour of individual flight quantities in aerobatic manoeuvre, which
were used for the quality consideration of piloting and determination
of flight trajectory. Later on, the measurement results will be also
compared with demanding calculations of flight characteristics manoeuvres. Record illustration (Figs. 7-11.) represents the course of flight
data during three loops and stall.
2
1,5
1
0,5
0
-0,5
-1
0
5
10
15
20
25
30
35
40
45
50
40
45
50
time [sec]
Conclusion
Figure 9 — Course of load factors
100
80
60
40
pitch angle [°]
Described measurements, first of all, verified the functionality of the
whole measuring equipment, from the individual sensors, apparatuses, connection, data logging and control unit to the software. New
data knowledge and valuable experience, which will be used for next
planned experiments, were gained during preparation, the measurement itself, and data processing. The conducted experiments suggested new ideas to improve the techniques and avoid mistakes. The measurement met the targets and the results obtained will be used in next
research activities at the Institute of Aerospace Engineering.
20
0
-20
-40
-60
-80
-100
0
5
10
15
20
25
30
35
time [sec]
Figure 10 — Course of pitch angle
References:
[4]
10
180
5
]
160
140
0
120
-5
100
80
-10
60
-15
40
20
-20
0
-25
true airspeed
-20
elevator angle [°]
aileron angle [°]
[3]
Figure 11 — Course of flight data during stall
200
-1
[2]
Letová příručka L-13AC Blaník (OK-1730), Letecké závody a.s., Kunovice, Czech Republic 2001
Kouřil, M., Šošovička, R., Daněk, V., Jebáček, I.: The Dynamic Measurement Unit Application for The Measurement of
Light Aircraft Flight Characteristics; International conference Transport means 2003, ISBN 9955-09-511-3, Kaunas
Lithuinia 2003
Kouřil, M., Šošovička, R.: Optické gyro v aplikaci pro měření dynamické stability letounu; Conference PhD 2003, ISBN
80-7043-246-2, Srní, Czech Republic, 2003, pp. 37
Kouřil, M., Šošovička, R.: Experimentální určování polohových úhlů letounu v prostoru; grant Fondu vědy FSI, VUT
v Brně, FP 330056, Brno, Czech Republic.
pitch angle [°], pitch rate [°/s]
roll angle [°], true airspeed [km.h
[1]
pitch rate
-40
pitch angle
-60
roll angle
-80
elevator angle
-30
-35
aileron angle
-100
0
2
4
6
8
10
12
14
16
18
-40
20
time [sec]
Accomodation of Sensor Failures by Means
of Artificial Neural Network
Pokročilá detekce, izolace a přizpůsobení chybných údajů od
senzorů pomocí umělé neuronové sítě
Ing. Jaromír Lamka, CSc. / VZLÚ, Plc., Prague
This paper sets out application of Artificial Neural Networks to provide a fast and accurate diagnostic tool for
identification of sensor faults. The network is also able to provide information on which sensor signal is degraded.
Several architectures for networks were assessed to find the optimum design for the application. The engine
performance was simulated by a computer program, yielding data sets for training and validation of the networks.
V práci je vysvětleno použití umělých neuronových sítí jako diagnostického nástroje pro zjišťování chybných dat od
snímačů. Síť rovněž umožňuje poskytnout informaci, který snímač dává nesprávné hodnoty. Pro nalezení optimálního
uspořádání sítě bylo hodnoceno několik možných řešení. Chod motoru byl simulován programem poskytujícím
datové soubory pro trenink a validaci sítě.
Keywords: Gas Turbine Engine, Sensor, Diagnostics, Artificial Neural Network.
9
Nomenclature
a . . . . . .network response (network output)
A . . . . .universal measured parameter
ANN . . .Artificial Neural Network
AUNS . . .universal parameter calculated by neural network
FFNN . .Feed Forward Neural Networks
gk . . . . .vector of current gradient
MSE . . .Mean Square Error
nG . . . . .relative revolution speed of the gas generator spool
P . . . . . .input vector of neural network
p0C . . . .barometric pressure
Q . . . . .number of training patterns
SCG . . .Scaled Conjugate Algorithm
t . . . . . .target value
T . . . . .target vector of neural network
T0C . . . .atmosphere temperature
T4C . . . .temperature between turbines
xk . . . . .vector of current weights and biases
αk . . . . .the learning rate
Introduction
In-flight sensor fault detection and isolation is critical to maintaining
reliable engine operation during flight. The propulsion system is operated at demanded conditions by the aircraft engine control system
which computes control commands based on sensor measurements.
Any undetected sensor faults therefore may cause the control system
to drive the engine into an undesirable operating condition. If a sensor
fails, it is crucial to detect and isolate the fault as soon as possible so
that such scenarios can be avoided. A sensor fault detection and isolation system which is capable of doing so with high reliability is indispensable for flight safety enhancement. A challenging issue in developing reliable sensor fault detection and isolation systems is to make
them robust to other faults, besides sensor faults, that can occur
during flight. Engine component performance can degrade gradually
due to usage, and abruptly due to fault events such as foreign or
domestic object damage. Likewise, errors can exist between the commanded and actual actuator positions. Such anomalies result in shifts
in sensed engine variables from their nominal values.
Gas turbine components, such as compressors and turbines, usually operate in harsh environments and are bound to degrade during
their operation. The component degradation will lead to overall engine performance change, which causes the engine to operate in a noneconomical conditions or even results in a total failure. Of about 70 to
80 per cent of gas turbine engine performance loss accumulated
during operation is attributed to compressor fouling, turbine erosion
and corrosion. Although the degradations have different causes, they
all result in a loss of engine performance.
Effective operation of an engine diagnostic system relies on correct
information from sensors but sensors may fail within a hostile working environment. Different sensor fault detection techniques have
been developed in the past, such as those using neural networks, Gas
Path Analysis and Genetic Algorithms.
Artificial neural networks (ANN) have been introduced into gas
turbine diagnostics since the late 1980's and different types of neural
networks have been used for diagnostic purposes. The most popular
neural networks are feed forward back propagation neural networks.
Due to the complexity of the problem, nested neural networks have
been introduced in gas turbine diagnostics by many researchers.
When a sensor fails, the engine diagnostic system should respond
to the change in order to continue the diagnosis correctly. To do this,
the engine diagnostic system either has the capability to detect the failed sensor and only use the measurements from the sensors that are
3/2006
not failed, or still use the same measurements but the measurement
corresponding to the failed sensor must be replaced with a recovered
measurement approximated with other healthy measurements.
Obviously, the engine diagnostic system with the former method is
more complicated than that with the latter method.
In this research, a number of decentralised neural networks are
used to recover the measurement of any failed sensor, once it is detected, by using other healthy measurements. These decentralised neural
networks are then embedded into a neural network gas turbine diagnostic system. Therefore, the whole diagnostic system has the capability of component fault diagnosis, sensor fault diagnosis and recovery. The proposed system is applied to a model gas turbine engine.
Discussions and analysis for some cases of typical sensor and engine
component degradations are provided.
Gas Turbine Engine Model
The gas turbine engine model used for purpose of this work is special M-601 software. Walter engines of the M-601 family are two-shaft
turboprop aircraft engines with a free power turbine and reverse flow
configuration. Engine layout is presented in APPENDIX I.
Object of this study was proposing, training and validating an artificial neural network for calculation of temperature T4C between Walter M-601 turbines depending on relative revolution speed of the gas
generator spool nG and on weather conditions, i.e. atmosphere temperature T0C and barometric pressure p0C. It can be expressed through
a functional relation:
(1)
T 4C = f nG ,T 0C , p0C
The function f was realised by specially developed neural network.
(
)
Neural Network Diagnostic System
A single neural network, whatever its type is, cannot cope with the
complexity of real problems such as gas turbine component and sensor diagnosis. Therefore, a nested neural network system is needed.
Such a system consists of a number of different neural networks; each
one of them is trained only for one measured variable.
Figure 1 — Diagnostic system with neural network
The configuration of the nested neural network diagnostic system of
the sensor and how the system works is shown in Figure 1. Independent variables which define operating point of the engine, i.e. relative
rotational speed of gas generator spool nG and barometrical conditions, temperature T0C and pressure p0C, is fed to NET. This artificial
neural network is capable of approximating the correct output of the
sensor when it works properly. AUNS parameter calculated by means
of NET is then compared with a measured parameter A. In subsequent step the relative deviation must be calculated for the faulty sensor
isolation. The equation used for the faulty isolation is shown below:
∆A =
A − AUNS
AUNS
× 100
(2)
If NET does not detect sensor fault, the measurement data is fed
directly into engine diagnostic system that is able to identify if there is
10
a component fault. If there is no component fault, the engine is declared being clean then the diagnostic process finishes. If there is
a component fault the measurement data is used for detection which
component has the fault, i.e. the compressor, the compressor turbine
or the power turbine.
When the faulty sensor is detected, the recovered sensor value
AUNS is then fed into auxiliary display in cockpit.
Artificial Neural Networks
The most commonly used artificial neural networks for gas turbine
diagnosis are the Feed Forward Neural Networks (FFNN) and they
are used in this study. FFNNs have been used by many researchers for
the detection, isolation and quantification of gas turbine component
faults. FFNNs for solutions of specific problems successfully use training function called Error Back-propagation of gradient.
Back-propagation was created by generalizing the Widrow-Hoff
learning role to multiple-layer networks and nonlinear differentiable
transfer functions. Input vectors and the corresponding target vectors
are used to train a network until it can approximate a function, associate input vectors with specific output vectors, or classify input vectors in an appropriate way as defined by designer. Networks with biases, a sigmoid layer, and a linear output layer are capable of approximating any function with a finite number of discontinuities.
The simplest implementation of back-propagation learning updates
the network weights and biases in the direction in which the performance function decreases most rapidly, the negative of the gradient.
One of iteration of this algorithm can be written
(3)
x k +1 = x k − α k g k
where xk is a vector of current weights and biases, gk is the current
gradient, and αk is the learning rate.
There are two different ways in which this gradient descent algorithm can be implemented: incremental mode and batch mode. In incremental mode, the gradient is computed and the weights are updated
after each input is applied to the network. In batch mode, all the inputs
are applied to the network before the weights are updated.
There are generally four steps in the training process:
— assemble the training data,
— create the network object,
— train the network,
— validate the network, i.e. simulation of the network response
to new inputs.
A typical back propagation neural networks that consists of a hidden layer is shown in Figure 2. A FFNN normally consists of an input
layer of source nodes, one or more hidden layers and an output layer.
The function of the hidden layers is to intervene between the input
layer and the output layer. Non-linear activation functions, such as the
hyperbolic tangent sigmoid function, are used to take into account the
non-linear relationship between the input and the output of the neural
network.
Figure 2 — A typical back-propagation neural networks
In this research, it is assumed that only one sensor may fail at a time.
Such a network system configuration gives the system the ability to
approximate the measurement of a failed sensor by using the information from other sensors that are working correctly. Usually the
same of number of neural networks to that of the sensors is used so
that the measurement of any failed sensor can be recovered.
Training data assembly
The neural networks should be trained with training samples and validated with validation samples before they can be used for engine and
sensor diagnosis. The training and validation samples, it means correct
data, were simulated by special M-601 software.
The functional relation of Eq.(1) can be written:
T = f (P),
(4)
where T is target vector defining T4C temperatures and P is input training matrix of the neural network that is created by three vectors, i.e.
nG, T0C and p0C. The M-601 turboprop engine can be operated in wide
range conditions:
1. 60% ≤ nG ≤ 98.5%, T0C = 288.15K, p0C = 101.325 kPa,
2. 223,15 K ≤ T0C ≤ 323,15 K, p0C = 101.325 kPa,
3. 25 kPa ≤ p0C ≤ 115 kPa T0C = 288.15K.
Above mentioned conditions determine the target vector T and input
training matrix P. In APPENDIX II three curves representing all three
conditions are presented. Points on each of curves show where the
curve was calculated. Now equation (4) can be expressed:
T 1 
 P1 
 = f 
(4)
T
 2
 P 2
T 3 
 P 3
Vector T was simply carried out by adding T2 behind vector T1 and so
on.
Network architecture creation
The first step in training a feed-forward network is to create the network architecture. For this study three neurons will be in input layer
and one neuron is necessary in output layer (see Eq. 1). Three layers
with different neurons are in hidden layer. The hidden layer has ”bottle-neck“ shape [1] i.e. second layer is created by smaller number of
neurons than first and third layer.
The activation function used for all neurons is the hyperbolic tangent sigmoid transfer function (tansig). It squashes all inputs to give
output in the range {-1, 1}. Therefore Linear activation function was
used in output layer.
Neural network training
A significant part of using a neural network is training. Training is
achieved through an iterative process where the training data is repeatedly fed into the network and it incrementally improves interconnection weights to match the network data to desired targets. However,
matching network outputs with desired targets do not say that the network has trained well. Therefore, testing a trained network with
a validation test is essential while a validation test does not include training data. The MATLAB's Neural Network toolbox ver. 5.0 is employed for this research to investigate the application of neural networks for diagnostics of the gas turbine.
The networks were trained using batch training and supervised
learning. The training algorithm was the Scaled Conjugate Algorithm
(SCG). The SCG training algorithm is chosen as it performs well over
a wide variety of problems and has relatively modest memory requirements compared to other algorithms such as Lavenberg-Marquardt
criterion.
The basic backpropagation algorithm adjusts the weights in the
steepest descent direction (negative of the gradient), the direction in
which the performance function is decreasing most rapidly. It turns
out that, although the function decreases most rapidly along the negative of the gradient, this does not necessarily produce the fastest con-
11
vergence. In the conjugate gradient algorithms a search is performed
along conjugate directions, which produces generally faster convergence than steepest descent directions. A full description of the SCG
training algorithm can be found in [2].
Having decided on all the above, the next step is the actual training
and validation of the neural networks. Neural networks with different
architectures were trained and validated until acceptable results were
obtained. The performance of the networks is assessed by two criteria, the Mean Square Error (MSE) and the generalization of the validation data set. The MSE function is given by equation
MSE =
1 Q
2
∑ (t (k )− a (k )) ,
Q k =1
(5)
where Q is the number of training patterns, t are target values and a is
network response (network output). In theory the training process
should reduce MSE.
The generalization of a network is defined as the ability of the network to produce the outputs that are close to their target values.
The training of the proposed neural networks was stopped when
the MSE reached value of 0.01 or a maximum number of 106 epochs.
The training of the networks was carried out with 63 different sets.
Some of the best successful neural networks are presented in Tab.1.
Network Structure of MSE
identifier Hidden layers
ANN 32
70-30-70
0.014
ANN 33
70-20-70
0.2
ANN 41 200-30-200
2.1
ANN 45
50-20-50
0.1
ANN 46
50-25-50
0.01
3/2006
Conclusions
An approach based on artificial neural network was investigated for
the development of an aircraft sensor fault detection and isolation system. Artificial neural networks can first alarm the operator,
consequently replace the faulty data with a close approximation to the
other users such as control or fault diagnostic system. A single feedforward neural network is trained for engine fault diagnosis purposes
and tested against untrained data, where results of more than 99.89
percent correct fault precision is achieved. The engine component
diagnostic system get rid of the impact of the sensor bias and ensure
the prediction accuracy of the system.
References:
[1]
Ghoreyshi, M., Singh, R.: Using Neural Network for Diagnostics of
an Industrial Gas Turbine; Proceedings of ASME Turbo Expo 2004,
Power for Land, Sea, and Air. June 14-17, 2004, Vienna, Austria.
[2]
Neural Network Toolbox for use with MATLAB. User's Guide, Version 5.
[3]
Lamka, J.: Inteligentní validace údajů senzoru pomocí umělé neuronové sítě; Metodický postup. VZLÚ Report, R-3911, 2006
APPENDIX II — Curves for neural network training
Condition 1
1000
970
940
Table 1 — Some of the best successful neural networks.
910
APPENDIX I — Walter M-601 Turboprop
Engine Layout
880
850
820
790
760
730
700
60
65
70
75
80
85
90
Relative revolution speed n
G
95
100
Mathematical relation
T4C = f1(nG) represents
the relation T1 = f1(P1)
for training of the neural network.
[%]
Condition 2
1020
980
G
[%]
1000
Temperature T 4C [K] and Relative revolution speed n
Validation test of the trained network does not include training data.
The network was validated using twelwe sets of parameters.
The results of validation of each trained networks is evident in
APPENDIX III. For example the network named ANN 32 has very
good value of MSE but great number of neurons in hidden layers gave
rise to great relative deviation of the network outputs. The ANN 41
has much more neurons in hidden layers than previous network, but
there is very high value of MSE, i.e. value of 2.1. It appears from this
that it is not good to use more neurons than is optimal number of. This
neural network tends to fabricate nonsense [3].
The ANN 46 is the best of all. It consists of 50 neurons in first hidden layer, 25 neurons in second hidden layer and again 50 neurons in
third hidden layer. Range of relative deviations of network outputs is
wide of 0.11%. It is very good result. The ANN 45 has very good
results too but output points are not situated symmetrically around
required values.
It is possible say that optimal number of neurons in hidden layers
is about of 120 neurons. On the assumption that number of neurons in
first and third layer is equal so it is suitable choose the number of neurons from 25 to 30 in second layer.
Temperature T 4C [K]
Validation of neural networks
960
940
920
T4C [K]
900
Ng/10 [%]
880
860
840
820
800
220
240
260
280
300
Atmospheric temperature T
0C
320
[K]
T4C = f2(T0C, nG) represents the relation
T2 = f2(P2) for training
of the neural network.
12
Condition 3
Temperature T 4C [K] and Relative revolution speed n
G
[%]
990
985
980
T4C [K]
Ng/10 [%]
I n t en t io n a l l y l eft vo id
975
970
965
20
40
60
80
100
Barometric pressure p
0C
120
[kPa]
T4C = f3(p0C, nG) represents the relation
T3 = f3(P3) for training
of the neural network.
APPENDIX III — Deviations
0,12
3,5
ANN 32: 70-30-70
MSE 0.014
ANN 41: 200-30-200
MSE 2.1
ANN 32: MSE 0.014
2,5
ANN 41: MSE 2.1
ANN 46: MSE 0.01
ANN 46: 50-25-50
MSE 0.01
2,0
1,5
1,0
0,5
0,0
700
750
800
850
900
-0,5
-1,0
950
1000
ANN 33: 70-20-70
MSE 0.2
0,10
Relative deviation: Validation data vers. Network ouput [%]
Relative deviation: Validation data vers. Network ouput [%]
3,0
ANN 45: 50-20-50
MSE 0.1
ANN 33: MSE 0.2
ANN 45: MSE 0.1
ANN 46: MSE 0.01
0,08
ANN 46: 50-25-50
MSE 0.01
0,06
0,04
0,02
0,00
700
750
800
850
900
950
1000
-0,02
-0,04
-0,06
-1,5
-2,0
-0,08
Temperature T4C [K]
Temperature T4C [K]
Experimentální zkušební systém pro pyrolýzní rozklad
vláknových termosetických kompozitních materiálů
Miroslav Valeš, Ing. Bedřich Štekner / VZLÚ, Plc., Prague
This paper deals with a new experimental test system for thermal breakdown of fibrous thermosetting composites.
The system is determined for processing of composite materials in connection with research on methods to recycle
composite waste arising during production, like chippings, snippings, etc., or after the end of a product life.
Příspěvek se týká nového experimentálního zkušebního systému pro teplotní rozklad vláknových termosetických
kompozitních materiálů. Systém je určen pro zpracování kompozitního materiálu, prováděného v rámci výzkumu a
řešení recyklace kompozitního odpadu, vznikajícího z výroby jako odřezky, odstřižky, apod., nebo po ukončení
životnosti výrobků.
Keywords: waste, recycling, composite material, thermoset, decomposition, experimental system.
13
One of the topics solved within the frame of the
Aerospace Research Centre Project is the area of
recycling of fibrous thermosetting composite
materials. The common types of composite materials are structures based on epoxy, phenolic or
polyester resins reinforced by carbon, Kevlar or
glass etc… There are several reasons why to deal
with this problem. One of the most important is
the growing usage of these composite materials in
comparison to conventional materials whether in
aeronautical or in other industries, another one is
the worldwide trend of change in waste treatment
based on re-using or usage of composites in another way rather than waste storage or disposal.
Pursuant to research done in 2005, when suitable techniques of treatment and recycling of
these materials were determined and some simple
and basic trials were performed, it has been decided to design an experimental system, which will
enable to test the processing of fibrous thermosetting composites and on basis of which it would be
possible to continue research on optimal processing technology with
respect to feasibility, optimization and determination of characteristics
of outputs — recycled materials.
The experimental system itself was designed and realized during
2006 in cooperation with the company Elektrické pece Svoboda, Světice u Říčan. The base of the system consists of a thermal reactor,
where, under conditions of high temperatures and normal or inert
atmosphere, decomposition of composite materials takes place. The
results of the process are remaining fibrous or other reinforcement
and gasified components of matrix, which are conducted into cooling
and filtration parts. These parts enable to obtain additional components from material breakdown, depending on the type of material
and processing technology.
The whole system is fully programmable and variably adjustable
as to temperature or inert media flow. The movement of gaseous products can also be controlled by several restrictors, suck-in valves and
flaps. The system also allows for monitoring of many parameters of
the process, in particular the temperature inside the furnace and inside the charge, temperature at cooler and filter unit outlets; furthermore the mass decrement of the material processed and, last but not
least, gaseous products of breakdown are monitored, especially O2,
CO, CO2, (H2), NO, NO2, SO2, or CXHX contents. Of course, it is possible to transmit measured values to PC for further analyses.
Some of the experimental system characteristics are given in the
Table No. 1 below.
It can be stated that by the launching of the above mentioned experimental system an important step has been made towards the solution of recycling of fibrous thermosetting composite materials. The
experimental system (not yet in serial production) is currently optimized in terms of its function, separate modes of technological processing etc. In the next stage the system will be utilized for the following
research on recycling of the fibrous thermosetting composite materials, processing technologies and characteristics of the resulting products.
Tab. 1 — basic technical characteristics of the system
3/2006
Fig.1 — view of experimental test system for
fibrous thermosetting
composite breakdown
Fig. 2 — Sample of treatment characteristics of composites
The pictures below show a real composite product before and after
processing; graph presents some characteristics from process of
decomposition.
Above Figs. 3 & 4 — Composite
product before and after processing in experimental system
Right Fig. 5 — Residual glass
structure after thermal treatment
of composite detail
14
Program SPAD (ver. 0.6) a jeho užití při snižování hluku
vrtulových letadel
Ing. Tomáš Salava, DrSc. / VZLÚ, Plc., Prague
The program is intended for fast analysis of the noise of propeller driven airplanes. It works basically on the principle
of spectral decomposition as it has already been explained in [1]. In this article a more advanced experimental version
of this program is described. It allows continuous decomposition and separate evaluation of decomposed noise
components in real time, or faster. Its adjunct procedures make it possible to back-transform or re-synthesize the
separated noise components, and to assess by listening the effects of possible noise control measures. Limitations of
spectral decomposition and some ways of their overcoming are discussed, too.
Program SPAD je určen pro rychlou analýzu hluku vrtulových letadel. Pracuje na principu spektrální dekompozice a
umožňuje vyhodnocovat odděleně hluk vrtulí a zbývajících zdrojů hluku. V tomto programu se nyní zkouší pokročilejší
procedury rozpoznání, separace a extrakce hlukových komponent generovaných vrtulemi, případně ještě také dalších
periodických složek hluku generovaných např. pístovým motorem. Poslední verze umožňují takto provádět selektivní
analýzu hluku vrtulového letadla kontinuálně a v reálném čase za různých letových podmínek. Separované hlukové
složky je možné zpětně rekonstruovat nebo simulovat pro doplňkové hodnocení poslechem.
Keywords: Noise analysis, noise abatement, propeller-driven aircraft.
Introduction
Propeller propulsion is still taken for more economical than fan and
other propulsion systems. A specific disadvantage of propeller propulsion is its characteristic, low-frequency noise. Though a considerable progress has been reached in propeller design and understanding the physical grounds of their noise emission, propellers are
still mostly dominant noise sources at propeller driven airplanes.
The intensity of the noise generated by propellers and its share in
the overall noise of an airplane may however differ broadly.
Noise power emitted by a propeller increases rapidly with the
helical speed of the blade tips and with the torque and thrust the
propeller is creating. [2], [3], [4]. Noise of propellers is therefore
a serious problem mainly in larger airplanes with higher cruise
speed, as are e.g. regional airliners, and very heavy aircraft. However, propellers are mostly the dominant noise sources also with
small airplanes. The low-frequency tonal noise created by propellers propagates well in the atmosphere, and can be annoying over
large distances [2].
A specific problem of larger, especially two-engine propeller
driven airplanes, is the interior noise as the propellers are placed
close to fuselage. In recent years this problem has been solved
quite successfully by active noise and vibration control systems [5],
[6], [7]. This solution is, however, not quite without some problems, mostly in adjusting and stability. Not negligible is also the
electric power consumptions of these systems. They are also not
maintenance free. Further solutions are therefore being found e.g.
in active, frequency-selective increasing dynamic rigidity of the
fuselage and fuselage walls etc. [8].
With small, light airplanes, which are driven by piston engines,
the propeller need not always be the dominant noise source. Sometimes a better exhaust muffler may bring more perceptible quieting
of a small airplane than experiments with propellers. If both the
engine and propeller noise components are at the same frequencies, and approximately equally strong, than changing the mutual
phasing of the propeller against the engine may bring some improvement.
In noise abatement of propeller driven airplanes a high resoluti-
on spectral analysis and spectral decomposition is a very effective
tool. It can be very helpful especially in identification and separate
evaluation of the main noise components, and also in evaluation of
the results of accomplished noise control measures. The SPAD
software set should manage even a bit more than only high-resolution spectral analysis, with spectral decomposition.
SPAD concept and main procedures
The basic concept of program SPAD (spectral analysis and decomposition) has already been explained in [1], where also its first working, experimental version was introduced. The version which is
described in this article (ver. 0.6) makes possible continuous, highresolution spectral analysis, which is followed by spectral decomposition, and separate evaluation of the decomposed components of
the analyzed noise. It can work in real time or faster even on contemporary low-cost computers.
Most procedures in program SPAD have been remake and
improved during its development. The adjunct procedures has also
been added into the SPAD SW-set, which make it possible to backtransform or re-synthesize the separated noise components for subsequent listening tests. Readjusting and remixing the decomposed
components of an airplane noise allow assess by listening the perceptual effects of different possible noise control measures.
Program SPAD computes repeatedly the short-time high-resolution spectra of the analyzed signal, in which it traces salient discrete spectral components. Next it finds the components, which could
belong to a frequency equidistant series of so called harmonics,
which could belong to or are created by a specific cyclic process.
Finally for the recognized set of equidistant spectral components it
computes the following basic values:
● fundamental frequency of the recognized set of harmonic
components (Hz)
● equivalent sound pressure level of the recognized and extracted harmonic spectral components (dB)
● equivalent sound pressure level of the rest noise without the
extracted spectral components (dB)
● level difference of the previous two values (dB)
15
Fig. 1 The SPAD ”one second“ basic evaluation window; shown are
the results of the interior noise analysis of a turboprop airliner during
its climbing flight phase
Such values can certainly be computed ”by hand“, too, just from
usual spectrograms, which would, however, be rather time consuming. This program can work in three run modes: (1) continuous
fast, (2) continuous standard and (3) break on mode. The last one
makes it possible to stop executing after any of the subsequent partial procedures and watch the particular results.
Input procedures
Version 0.6 works in post-processing mode as yet. The input data
are supposed to be noise signal recordings preferably in the Microsoft *.wav format. In the default setting the sampling frequency is
22 050 Hz with 16bit liner quantization. The input procedure in version 0.6 has been equipped by a decoder of the wav file headers,
which reads the header data, as sampling frequency, quantization,
number of channels, or length of the recording and can adjust the
proper settings. The recordings can be of nearly unlimited length,
however, they are processed in successive 30-second or one minute data blocks. In the ”break on“ run mode the input procedures can
also display the time course of the time segments just analyzed.
Transform procedure
The transform procedure computes short-time, discrete, time-averaged spectra using a standard FFT algorithm [9]. In default setting
the Hannig time window is used, and the averaged spectra are computed over one second intervals with the window time shift step 0.2
sec. By time averaging the tonal spectral components are enhanced,
whereas the noise components, which are random in time, are smoothed. In the default setting FFT is computed over 8192 samples,
so that about 4000 usable discrete spectral values are obtained.
With sampling frequency 22050 Hz the frequency resolution is 2.7
Hz and the time window length is 0.371 sec. Both the complex and
power spectra are computed and stored for the following processing and evaluation.
Frequency spectra processing
The next task is to find the spectral components which are generated by periodic processes, as is typically the rotation of propellers,
or e.g. pulsation form exhausts of piston engines. If they are strong
enough, they manifest themselves in frequency spectra as salient
narrow local maxima. These can be found even by quite simple
algorithms. If the tonal components in the aircraft noise are less
prominent or even buried in the non-periodic noise, the tonal components could be enhanced, beside the time-averaging, also by
more advanced methods [9]. In our case, however, weak tonal
components in aircraft noise are less interesting.
With turbo-propeller aircraft the task of finding the propeller
3/2006
Fig. 2 An example of the SPAD basic evaluation of a 30 second time
segment of a turbo-propeller airplane interior noise (for details see
text)
noise components is relatively easy, as mostly no other prominent
periodic noise components are present. As a rule, the fundamental
spectral component of the tonal propeller noise is the most prominent one. The higher harmonics can be found on the multiples of
the fundamental frequency. If enough strong spectral components
of more than one periodic process are present, the task is less simple. This is the case of the noise of propeller airplanes with piston
engines. In this case spectral components of the two periodic signals may partly or entirely merge. This case is discussed later.
In program SPAD ver. 0.6, the first step of the short-time spectra processing is founding the most prominent narrow local maximum, after which up to 8 narrow maxima are being found on the
upper harmonic frequencies. Two different, relatively simple algorithms are used for this purpose. Then the remaining prominent
maxima are being found. If they are found, they are tested to determine which of them can create a series of harmonics, and can
belong to another periodic process. Different possible combinations
of the detected maxima are also tested.
Spectral decomposition and basic evaluation
When the first series of the most prominent harmonic components
is found, the spectral decomposition is executed. The corresponding
spectral components are extracted and their characteristic data are
computed and stored together with the same data of the remaining
rest spectra. Then, in the same way the rest spectrum is reprocessed to found the spectral components of the second cyclic process.
Finally a possible frequency merging of the components of the two
sets of harmonics is tested and the stored data at the merging frequencies are readjusted in the way, which is outlined later.
In the default setting, the program executes the basic evaluation
of the processed noise in every second. Then it displays the basic,
one-second results of the analysis in combined graphical and numeric mode, as it is shown in Fig. 1. In this figure there is an example
of the SPAD basic evaluation of a one second of an interior noise in
a regional category turbo-propeller airplane, which was recorded
during its fast climbing flight phase. In Fig. 1 are displayed the
decomposed spectra (the propeller noise — light curve, the rest
noise — dark curve), and in the right upper corner there is added
a table wit the basic results. They are from the top down:
1. the total averaged sound pressure level (tot)
2. equivalent sound pressure level of the decomposed periodic
components (per)
3. sound pressure level of the rest of the signal without decomposed components (rst)
4. difference of the two former values (dif)
16
All the sound pressure levels are given with C and A frequency
weighting. On the right side is a table of the eigen frequencies of
the extracted harmonic spectral components (Fmx) in Hz.
The next Fig. 2 shows an example of basic evaluation of a 30
second sample of an interior noise in the same regional category
turbo-propeller airplane, which was recorded after the take-off of
the airplane. Among the fifteen and twenty five second in this noise
sample the airplane ends the fast climbing flight phase and is in
transition to reduced power climbing flight phase . In Fig. 2 are
plotted from the top down:
1. fundamental frequency of the propeller noise — F1 (Hz)
2. overall sound pressure level of the processed noise — Ltot (dB)
3. equivalent sound pressure level of the extracted propeller noise
— Lprop (dB)
4. equivalent sound pressure level of the rest noise — Lrst (dB)
5. level difference Lprop-Lrst (dB)
As it is printed in the right upper corner, all the sound pressure
levels are plotted with frequency weighting C (which is nearly linear). Other possible frequency weightings are A and D.
In the fast climbing phase the propeller noise has been approximately 25 dB(C) above the all other noise. It decreases to about 18
dB (C) during the transition and next flight phase. In Fig. 2 there is
also well visible e.g. that the fundamental frequency of the propeller noise changes only during the transition phase, from 121 Hz to
nearly exactly 100 Hz. Otherwise it remains constant, as it corresponds to the controlled rotation of the constant speed propellers.
SPAD analysis and spectrograms
As it has already been mentioned, program SPAD performs basically a fast analysis of high-resolution frequency spectra. It traces,
separates and independently evaluates the periodic and non-periodic noise components, and is especially designed for fast analysis
of the noise of turbo-propeller airplanes. Let us compare the SPAD
concept of airplane noise analysis with description and evaluation
of turbo-propeller airplane noise by contemporary spectrograms.
Fig. 3 An example of a 3-D spectrogram of the interior noise in a regional category turbo-propeller airliner, at the beginning of its take-off to
a regular flight
In Fig. 3 there is first an advanced 3-D spectrogram of an interior noise in a turbo-propeller aircraft. The noise sample was recorded inside still the same regional category airplane, however,
during its start and take-off to a regular flight, this time. Originally
the spectrogram is a color 3-D one, with also the color sound pressure level scale. In its gray scale reproduction, regrettably, much of
the color information is lost. In this spectrogram time elapses from
the front to the back, in other words, the time scale actually begins
at the back of the 3-D graph.
The noise recording begins by a last few seconds of an idle state,
after which engines go to the full take-off power. At the beginning,
Fig. 4 An example of a SPAD basic evaluation of the same noise recording as for which the 3-D spectrogram in shown in Fig. 3. (from top to
bottom: F1, Ltot, Lprop, Lrest, Ldif)
the propeller noise traces are nearly invisible however. They suddenly increase very strongly in their level, together with their increasing eigen frequencies. Beside the fundamental component of
the propeller noise at about 120 Hz, also three upper harmonic
components are visible. The third and fourth harmonics diminishes
slightly after a few seconds, whereas the fundamental and the
second ones remain strong and dominant
Spectrograms, especially the contemporary color 3-D ones, are
very synoptic and impressive. However, its direct analysis and evaluation is not just easy and mostly can not be much accurate. Consecutive short-time spectra evaluation remains much better in this
respect. A ”hand-made“ analytical evaluation of larger series of
shot-time spectra over a longer time is obviously very time-consuming. Beside this, in aircraft noise evaluation, the many details in
spectral signatures, as they are visible in spectrograms, are mostly
less significant.
The SPAD basic analysis and evaluation should conform much
better to the needs of the current practice in noise abatement of propeller driven aircraft. The SPAD analysis results are easily understandable, and transparent, as it has already been illustrated by
Figs. 1 and 2. For further comparison there is in Fig. 4 the SPAD
basic evaluation of the same noise recording of which the 3-D
spectrogram is in Fig. 3.
On the graph in Fig. 4 the same result values as in Fig. 2 are plotted. They are from the top down: (1) fundamental frequency F1 of
the separated propeller noise, (2) the overall sound pressure level
Ltot, (3) the equivalent sound pressure level Lprop of the extracted
propeller noise, (3) the equivalent sound pressure level Lrest of the
rest noise and (4) the level difference Lprop-Lrest, which also tells
us how high is the propeller noise level above the other noise components.
During the idle period the propeller noise components are weak
and the program is not able to trace reliably even its fundamental
component. Therefore also the decomposition is not executed, and
in the graph only the overall sound pressure level is plotted. (about
60 dBA). During the fast transition to the full take-of power of the
engines, the propeller noise increases rapidly, their components can
be traced easily and the decomposition starts. In Fig. 4 the full
SPAD evaluation begins since the ninth second of the plot.
During the ninth and tenth second the revolutions of the propellers reach the nominal take-of value and are held constant. Correspondingly, the fundamental frequency of the propeller noise increases to 121 Hz and then it also remains constant. The sound pres-
17
Fig. 5 Sound pressure levels of decomposed three harmonic components of a propeller noise
sure level of the propeller noise inside the aircraft reaches 80
dB(A), and is about 20 dB above the other noise. The rest noise
does not increase much yet. As it can be seen from fig. 2 the propeller noise remains strong and dominant during all climbing flight
phase, even when the aerodynamic noise level increases.
Program SPAD (ver. 6) makes it possible to store many of the
partial computed data and print them into individual data files for
further processing. As an example of using the stored partial data
there is in Fig. 5 shown a plot of the sound pressure levels of the
three first harmonic components of the propeller noise for the same
case of which the basic evaluation is shown in Fig. 4. They were
plotted into a standard Excel graph after some processing, together
with the fundamental frequency (F1) and the total sound pressure
level (Ltot).
From Fig. 5 it is evident, that even during the take-off the first
harmonic of the propellers noise was the most dominant one. From
this finding it may be e.g. concluded that the noise control measures, which would work only in a narrow, frequency-limited band
(around 100-120 Hz in this case) could bring significant quieting.
The last example in Fig. 6 shows the SPAD evaluation of the interior noise sample, which was recorded in the same airplane during
its steady cruise flight phase. The propeller noise level is significantly lower in this case, however, still dominant and more then 10
dB above the other noise.
Propeller airplanes with piston engines
Spectral decomposition in its basic principle cannot be used if the
discrete spectral components of two cyclic processes merge in frequency. This is, however, often the case in the noise of small propeller airplanes with piston engines, especially with not much efficient exhaust mufflers. As it is quite evident the following simple
relation holds for the fundamental, basic frequency of a propeller
noise:
F1p = mb . n / 60
[Hz],
where mb is number of blades and n is number of revolutions of the
propeller per minute. Basic or fundamental frequency of a piston
engine exhaust pulsation can be computed e.g. from the following
simple relation:
F1e = n . pc . ks / 60 [Hz]
In this relation n is number of revolutions of the crank or output
shaft per minute (rpm), pc is number of cylinders and ks = 1/2 for
four-stroke and 1 for two-stroke engines. In most common case of
no gearbox, all or at least some harmonics of the propeller and
engine noise can merge in frequency.
In case of partial frequency merging a simple interpolation procedure, based also on the knowledge of propeller noise spectra pro-
3/2006
Fig. 6 SPAD basic evaluation of a sample of the interior noise of
a turbo-propeller regional category airplane during its steady cruise
flight (from top to bottom: F1, Ltot, Lprop, Lrest, Ldif)
perties, may help to estimate correctly the partial levels of the two
merged components. Such procedure is now under development,
using also algorithms which has been described in [10].
In Fig. 7 is an example of a partially decomposed noise spectra
with merged spectral components. In the first step of processing
program SPAD found and extracted the spectral components of the
harmonic series, with the most prominent component at frequency
121 Hz. This is also the fundamental frequency and fundamental
component of the dominant tonal noise.
The equivalent sound pressure level of the decomposed components, as it has been computed by the program is 63.8 dB(A), which
is 5.8 dB(A) above the rest noise. The basic procedure can now be
repeated on the rest spectra. The second periodic process generates
the series of the harmonics with the fundamental frequency 88.7
Hz. The first merged components are at frequency 242 Hz, at which
is the second harmonics of the first and the third one of the second
harmonic series. The next merged components are at frequency 480
Hz.
The levels of the components of the first harmonic series are
enough higher then those of the second one. In this case it is unlikely that the first merged component in the first series could be significantly different from the found common level of the two merged components. Therefore also for the particular level of the merged component in the second series can be taken the value
interpolated from the levels of the two adjacent components in this
series.
To determine which series of the dominant spectral components
is generated by propeller and which by e.g. engine exhaust, it is
Fig. 7 An example of noise spectra with partly merged spectral components
18
necessary to know (1) number of the propeller blades, (2) number
of cylinders and (3) whether the engine is two or four stroke. In Fig.
7 there are shown the results of the first decomposition of a sample
of the outer noise of an older two-seater with two-blade propeller
and four-stroke six-cylinder engine. In this case dominant is the
noise of the engine, which had no exhaust muffler in this case.
From the fundamental frequencies can be derived, that the propeller and the engine was working at 2661 RPM.
In case of total frequency merging the SPAD analysis and basic
evaluation may still be helpful, though not as straightforward in
use. In this case e.g. one of the possible solutions can be in comparing the airplane noise high-resolution spectra which were measured for different exhaust mufflers of known properties or with
and without an exhaust muffler, otherwise under the same conditions. A relatively simple arithmetic operation applied on the measured power spectra can yield well useable results to assess the
share of the propeller and engine noise on the overall noise of an
airplane.
Auralization in spectral decomposition
Auralization in spectral decomposition means a possibility to listen
separately the decomposed components of an acoustic signal. In
noise control an auralization, and an auralization facility could
make possible to evaluate by listening or by listening tests the subjective effects of different feasible measures which could control
only some of the noise components. In practice such facility could
make it possible to hear e.g. the effect of using a quieter propeller
a more efficient exhaust muffler or e.g. too, the effect of an active
noise control system which would suppresses only the fundamental component of the propeller noise.
In the SPAD program two possible solutions are tested. One possibility is the back transforming of the decomposed spectra to the
corresponding time representation by inverse Fourier transform.
The second one is re-synthesis of the separated noise components
from their spectral signatures. Though the first possibility looks
more exact or straightforward, it involves some specific problems
with the back transformation on short-time spectra.
An example of the back synthesis of the decomposed noise components is shown in Fig. 8. As no auralization is possible in a printed article, in Fig. 8 is shown the 3-D spectrogram of a short segment of the re-synthesed propeller noise followed by the rest noise.
The back synthesis was done from the decomposed propeller and
rest noise spectra which are shown in Fig. 1.
It should be added that quality of the auralization depends not
only on the digital procedures, which are used for digital backsynthesis, but also very much on the properties of the sound repro-
Fig. 8 Decomposed and separately re-synthesized propeller tonal
noise followed by the rest noise without the extracted propeller spectral components shown in 3-D spectrogram
duction equipment used. This especially means enough sound
power of the sound reproduction equipment at low frequencies.
Such equipment should possess any laboratory which would want
to deal with aircraft noise measurement and analysis seriously.
Summary
Program SPAD is intended for fast analysis and evaluation of the
noise of propeller driven airplanes. Basically, it repeatedly traces
the periodic noise components in high-resolution discrete short-time
spectra and separates them from the rest noise. Then it evaluates
independently the particular separated noise components.
In application on propeller driven airplanes the program makes
possible to separate the propeller noise from the rest noise and evaluate e.g. the share of propeller noise in the overall noise of an airplane. With some limitations also separation of propeller and piston
engine noise components is possible. Ways of overcoming these
limitations were tested in this program version (ver. 0.6), together
with the so called auralization.
The SPAD (spectral analysis and decomposition) concept is relatively simple and transparent, and the way how the propeller aircraft noise is analyzed and evaluated is easily understandable. Its use
could be very effective as it has already been illustrated by the
examples given in this article. This program is certainly just one of
many contemporary advanced tools for successful noise control of
propeller driven airplanes. It has arisen partly as a by-product of
more broadly aimed activities in aircraft noise measurement and
control.
References:
[1]
Salava T., Sloufova M.: Spectral Decomposition in Noise
Abatement of Propeller Driven Airplanes; Czech Aerospace
Proceedings. pp. 22-25, No. 3, 2005
[2]
Hubard H. H. (edit.): Aeroacoustics of Flight Vehicles, Theory and Practice, Vol. 1, 2; NASA Reference Publication
1258, Washington, D.C. [Springfield, National Technical
Information Service, 1994]
[3]
F. Farassat, Kenneth S. Brentner: The Acoustic Analogy and
the Prediction of the Noise of Rotating Blades; Theoretical
and Computational Fluid Dynamics, Vol. 10, Numbers 1-4 /
January, 1998
[4]
Campos L. M. B. C., Lau and F. J. P.: On Propeller Acoustics Design Synthesis Using Source Distributions Along
Blade Span; 12th AIAA/CEAS Aeroacoustics Conference
(27th AIAA Aeroacoustics Conference), 8-10 May 2006,
Cambridge, Massachusetts
[5]
Kallergis, M.: Possibility of Active Propeller-noise Suppression in Piston-engine Aircraft by Changing the Phase Relation Between the Propeller and Exhaust Signals; Journal d'Acoustique (ISSN 0988-4319), Vol. 2, Dec. 1989, pp 401-406
[6]
Gardonio P.: Review of Active Techniques for Aerospace
Vibro-Acoustic Control; J. Aircraft, Vol. 39, No. 2, MarchApril 2002
[7]
Ross Colin: Active Control of Interior Noise in Aircraft.
(Ultra Electronics, UK); 9th CEAS-ASC Workshop ”Active
Control of Aircraft Noise: Concept to Reality“, 10-11
November 2005, KTH Stockholm, Sweden
[8]
Preumont, A.: Vibration Control of Active Structures, an Introduction; Kluwer Academic Publishers, Dordrecht, The
Netherlands, 2002
[9]
Mitra S.K., Kaiser J.F.: Handbook for Digital Signal Processing; Viley Publishers, 1993
[10] A. P. Klapuri, ”Multiple Fundamental Frequency Estimation
Based on Harmonicity and Spectral Smoothness“, IEEE
Trans. Speech Audio Proc., 11(6), 804-816, 2003
19
3/2006
Metal End Joint
Zkoušky kompositního táhla řízení letounu s kovovou koncovkou
Ing. Tomáš Marczi, Ing. Jiří Stehlík / Department; Ing. Karel Blahouš /
Department of Mechanics, Division of Strength of Materials, Czech
Technical University in Prague
The paper presents a project focused on the use of composite materials in the airplane control system. The presented
paper describes the tests of a composite rod used in the Ultra Light airplane VL-3. The control rod consists of a conic
aluminum joint and a composite tube. The composite tube is made of carbon filament, tube and aluminum end joint
are glued together. Strength of the thread connection conic aluminum end joint vs. end lug bolt is investigated. Next
investigation is focused on the glued connection between conic aluminum end joint and composite tube. Problems of
such connection are highlighted and a solution proposed.
Článek popisuje zkoušky kompozitového táhla letounu VL-3. Táhlo sestává z uhlíkové kompozitové trubky a duralové
kovové koncovky. Zkoušky jsou zaměřeny na šroubové spojení koncového oka a duralové koncovky a dále na pevnost
lepeného spoje kompozitová trubka vs. duralová koncovka. V článku jsou rozebrány problémy lepeného spoje vzhledem
na jeho využití v konstrukci táhel systému řízení letounu.
Keywords: Control Rod, Composite, Carbon, Glass, Filament, Airplane Control System, Winding,
Gluing.
Introduction
Connection testing
The project is focused on the Ultra Light airplane control system. The
composite control systems are already used in a few types of Ultra
Light airplanes. First composite control bars are based on glass filament [1]. They have been used since 1996. These kinds of control
rods can be found in Czech UL airplane such as Lambada, Samba or
Seehawk.
Next composite control rod already used in the plane control system
is made of carbon filaments. This kind of control rod (Figure 1) is used
in the plane VL-3 and it was subjected to the tests of this project.
Tested composite control rod consists of conic aluminum end joint
glued into the carbon based composite tube. The glued or adhesively
bonded joint can distribute the required load over a larger area than
the pure mechanical joint, requires no holes, adds very little weight to
structures and has superior fatigue resistance. However, the adhesively bonded joint requires the careful surface preparation of the adherend, is affected by service environments, is difficult to disassemble
for inspection and repair [3] and it is difficult to get axial alignment in
required tolerance. However; gluing is the most common composites
connection technology used in the industry.
Two tests have been done on the real connecting rods (Figure 1)
used in the plane VL-3. Four specimens (Figure 2) have been tested.
The specimen consists of carbon based composite tube of 28mm
outer diameter, 1mm wall thickness and about 250mm length. Thread inside of the conic joint is M6, length of the thread is 10mm. Conic
end joint is hollow, made from aluminum and glued into the composite tube. Hydraulic testing machine INOVA ZUZ 200 has been used
for specimen loading (Figure 3).
The strength of the thread connection of the end lug bolt and conic
aluminum joint has been tested first. The test assembly is shown on
Figure 2.
Six different bolts have been tested. The high strength bolts have
been used. Specimens have been loaded by axial tensile force. Maximum applied load has been 2.6ton. After the tests, no damage on the
inner thread of the aluminum end joint has been noticed. Also the
glued connection of the composite tube and aluminum end joint has
been unaffected. Hence, assumption that the end lug bolt defines the
Figure 1 — 'Carbon' control rod used in planes VL-3
The 'carbon' composite control rods are made by 'winding' technology, so the rod cross section is circular, with constant rod wall thickness. The aluminum end lugs are glued into the composite tube. These
'carbon' composite control rods are made by Czech company CompoTech s.r.o. Great advantage of this control rod design is its filament
layout parallel to the rod longitudinal axis. From the flexural rigidity
and buckling resistance point of view, axial filament layout gives the
most advantageous mechanical properties of the control rod.
Control rods buckling behavior define resulting rigidity of the
whole airplane control system. End lug accuracy alignment is one of
the problems which can not be omitted in the composite control rod
design. Buckling resistance of the long slender rods is highly affected
by the axial misalignment of the applied compressive load. Allowable misalignment of the control rod end lugs are about 1mm [2].
However, any of the misalignment of the end lugs under the applied
compressive load causes the additional parasitic bending load. According to [1], maximum operational load of the UL airplane control system is about the ±3000N.
Figure 2— Specimen of the VL-3 plane control rod and bolt test
assembly
20
strength of the connection in this kind of control rod design has been
confirmed.
Figure 4 shows the test load running and ultimate strength of three
different bolts.
During the test the problems of axial misalignment of the composite tube and conic aluminum end joint has been observed. Origin of
these problems lies in the gluing technology of the conic aluminum
end joint and composite tube. Moreover, there is aluminum tube glued
on the composite tube on the other end of the specimen. This tube is
used for the specimen clamping into the loading fixture. This additional gluing connection increases the axial misalignment of both ends of
the specimen. Thus, the additional parasite bending load occurred
during the axial tensile loading.
After the thread connection tests, the strength of the glued connection of conic aluminum end joint and composite tube has been tested.
For the purpose of test the conic aluminum joint has been adjusted.
The top of the aluminum joint has been cut off and through the hole
fixed into the loading fixture.
Figure 3 — Loading machine
INOVA ZUZ 200
and loading fixture
Loading process
30000
jet turbine strength
bolt
25000
strength bolt
Force [N]
20000
15000
common bolt
10000
Figure 5 —
Adjusted conic
aluminum end
joint and loading
fixture
The bottom part of the loading fixture remained the same as in the
thread test. The clamping fixture of the adjusted conic end joint consists of two conic clamping devices. The inner part is a conic tap inserted into the conic aluminum end joint. The second part is a ”cup“ with
conic hole and it is placed from the top on the conic aluminum end
joint. Than, both clamping devices are tighten together. Conic aluminum end joint is squeezed between them.
The conic aluminum end joint is glued into the composite tube by
glue SPABOND. Contact gluing area of conic aluminum joint and
composite tube is about 2637.6mm2. If the value of ultimate shear
stress recommended by the glue manufacturer U = 8MPa is taking
under consideration, expected ultimate tensile force is about 2.1ton.
However, according to [4] the ultimate shear stress of glued connection of the same glue as is used for the testing specimen could be up
to 20MPa. It has to be mentioned, that tests performed by [4] were
done on the glass filament specimens and strength of gluing heavily
depend on the technology conditions.
As in the thread test, the problem of axial misalignment of glued
component from which the specimen consist occurred. Thus, conic
aluminum end joint has been loaded also by the parasite bending load.
The conic end joint of the first tested specimen has not been properly
tightened, which caused rupture of the conic aluminum end joint
(Figure 6).
The parasite bending load caused jamming of the conic aluminum
end joint in the composite tube. The end joint has not been pulled out
of tube axially, but it tries to break out the tube wall. The damage of
the tube wall can be also noticed on the Figure 6.
5000
0
0
100
200
300
400
Time [s]
500
600
700
800
Figure 4 — Bolt
loading running
and broken bolts
21
3/2006
Figure 8 — Loading of the
gluing connection
Loading Process
60000
50000
40000
Force [N]
During the next tests the parasite bending has been eliminated by
adjusting the specimen clamping into the fixture. Thus, performed
tests were successful, no damage occurred on the conic aluminum
joint neither on the outer wall of the composite tube. The conic aluminum joint has been clearly pulled out of the tube. The damage of
the inner wall of the composite tube occurred during the test (Figure
7).
Used composite tube itself consists of several filament layout layers. First layer is winded in 0°. This layer stiffens the tube cross section - tube cross section holds the circular shape. Then there are several filament layers parallel to the tube longitudinal axis. These layers
give to tube desirable flexural rigidity and mechanical properties. Last
layer is the ±30° covering filament layout.
30000
20000
10000
Figure 9 — Measured
sections (below)
Section 1
Figure 7 — Test specimen with first filament
layout ripped off
0
0
10
20
30
40
50
60
70
Time [s]
Section 2
Section 4
Section 3
resulting stress distribution in different way than in section 1. The strain gauges and clamping ring are attached to the same tube wall in the
section 4. In the section 1, gluing connection tube vs. aluminum end
joint is done on the inner tube wall, yet strain gauges are placed on the
outer tube wall. Probably this is a reason of the steeper deformation
running than in the rest of the sections. Running of the deformation in
the section 4 (Figure 10) shows lower max. than in the sections 2 and
3. Strain gauges of the section 1 are placed above the gluing connection of conic aluminum end joint and composite tube. Resulting tube
deformation (stress) in section 1 is distinctively lower than the stress
in the rest of the tube sections.
DEFORMATION RUNNING
4000
Bridge 4 CH=2
Bridge 3 CH=3
Bridge 2 CH=4
Bridge 1 CH=5
3000
2000
Def. [ m]
It is obvious the first filament layer winded in 0° has a low resistance in the applied tensile load (90°). From Figure 7 can be clearly
seen damage of the composite tube. First filament layer of the composite tube has been ripped off from the tube.
Figure 8 shows the loading running of two test specimen. It is seen
that sustained ultimate tensile force is about 5.4ton. It is two times
more than expected value.
One of the specimens has been used for investigation of the stress
distribution during the loading. The stress distribution has been measured via the strain gauges connected into the full bridges. Measurement has been done in four sections.
Logger SPIDER 8 and software CatmanEasy have been used for
measurement, data acquisition and evaluation. Measurement record of
this test is shown on Figure 10. Maximum tensile load has been
3.1ton. From the Figure 10 it can be assumed that sections 1 and 4 are
affected by the presence of glued aluminum joints. Especially section
1, where load is distributed into the aluminum end joint via the glued
connection. Max. deformation is lower than in the sections 2 and 3. In
the sections 2 and 3, the deformation has almost identical running.
Obtained results are similar to those presented in [4]. Stress is distributed into the tube via the glued connection and max. stress in the
tube wall occurs in the distance approximately equal to 1.5DTUBE
behind the glued connection. However, used specimen was too short;
therefore, section 4 has been affected by the clamping aluminum ring
glued on the end of the composite tube. More over; aluminum clamping ring is glued on the outer tube diameter, which may affect the
1000
0
244
245
246
247
248
249
250
251
252
253
254
255
-1000
-2000
-3000
time [s]
Figure 10 — Deformation running in four sections of the specimen
Conclusion
Tested specimens have been loaded by tensile force only. Ultimate
loading force carried by the specimen in both tests was more than
seven times greater than operational load of UL airplane control system. Also can be concluded that glued connection has sustained at
least 1.5 time greater tensile stress than glue manufacturer is guarantee. However, two specimens sustained more than 5ton of axial force
compare to the last one, which sustained ”only“ 3.2ton. This difference clearly indicates sensitivity on used gluing technology and careful surface preparation of the adherend.
Next conclusion from the performed test can be done about the axial
misalignment problems of glued connection composite tube vs. metal
22
end joint. In the serial production it is difficult to ensure precise accommodation and gluing of the metal end joint inside the tube. In the case
of tested control rod used in the Ultra Light airplane control system this
misalignment does not affect the resulting control system stiffness. It is
due to relative low load applied into the system [1]. Precise axial alignment of the metal end joint and composite tube can be controversial in
the case of the extremely long control rod. New airplanes build from
several pieces of composite materials shell structures requires control
rod of more thane 3m of its length. Such a control rod is accommodated into the airplane structure in more than two points. Thus, this long
control rod is bending according the operational deformation of the supporting shell structure. Of course, the function of the control rod has to
be unaffected by this operational deformation.
Advantage of the composite control rod is its ability to predefine
the filament layout and define the final control rod stiffness.
Important conclusion about the winding technology can be done
towards the tube manufacturer. According the experiences from the
performed tests, authors suggest use filament layout parallel to the
longitudinal axis as a first layer of the tube. Second layer could be the
0° oriented filament layout. This filament layout will prevent the rip
off of the inner filament layer of the tube during the end joint pulling
out of the tube see Figure 7.
Presented tests did not involve the fatigue behavior investigation of
the glued connection of conic aluminum metal end joint and carbon
based composite tube. Also the investigation of whole control rod
buckling rigidity can be considered as a future work.
References:
[1]
Marczi T., Kábrt M.: Composites in the Airplane
Control System; in Proceedings of Advance Engineering Design, Prague, 2006
[2]
Mikula J.: Konstrukce a projektování letadel, Podvozky, řízení letadla, rozpočty a náklady; Czech
Technical University in Prague, 2004
[3]
Lee D.G., Kim H. S., Kim J. W., Kim J. K.: Design
and manufacture of an automotive hybrid aluminum/composite drive shaft; Composite Structure
Journal 63, pp 87-99, 2004
[4]
Sedláček R.: Návrh a experimentálni ověrení nových
řešení spojů kompozitních a kovových dílů v rotačních pohonech dopravních a zemědělských strojů;
Ph.D. Thesys, ČVUT in Prague, 2005
Acknowledgement
Authors gratefully acknowledge support of this research carried out
by the Aerospace Research Center at The Czech Technical University in Prague. Authors also gratefully acknowledge cooperation of Ing.
O. Uher, Ph.D., Research and Development Director of CompoTech
Company.
Editorial Note: Further colour pictures to this article are printed on
the inner back-cover page.
Návrh aerodynamického řešení proudové cesty studeného
propulsoru
Ing. Erik Ritschl, Ing. Robin Poul / Department of Aerospace, Czech
Technical University in Prague
The paper presents an aerodynamic design of a cold jet propulsion unit. The main focus is on the shaping of the inlet
channel. The cold jet propulsion unit consists of a piston engine, a blower and the necessary air duct. This propulsion
unit can be used in small aircraft which may look like jets. The efficiency of the unit depends strongly on the inlet
channel shaping. The authors suggest two variants of inlet design and their comparison.
V příspěvku je přiblížen návrh proudové cesty pro studený propulsor s důrazem na tvarování vstupní části před
rotorem dmychadla. Studený propulsor je pohonná jednotka navržená pro pohon sportovních letadel, která mají
navozovat dojem proudového letounu, ale jsou cenově mnohem dostupnější. Takto navržená pohonná jednotka sestává
z pístového motoru a ventilátoru ukrytého uvnitř trupu. Pro ventilátor je třeba přivést dostatečné množství vzduchu
a to pokud možno beze ztrát. V další časti článku jsou nastíněna dvě řešení proudové cesty spolu s jejich rozborem.
Keywords: Propulsion Unit, Cold Jet, Aerodynamic Design, CFD.
Editorial Note: Further colour pictures to this article are printed on
the inner back-cover page for the sake of better clarity.
Introduction
The cold jet propulsion unit is directed to power small sports aircraft.
This category is usually driven by propeller, not cold jet. The cold jet
is propelled by a fan placed inside the fuselage and powered by a piston engine. [1]
It is a big dream of many pilots to pilot a jet plane. But only few
of them are able to convert their dream into a reality. The cold jet is
an alternative.
The air duct consists of inlet channel, rotor and stator stage and the
nozzle part. In the nozzle will be placed a by-pass channel where is
placed the radiator. In this study, this part is simplified and modelled
as a cylindrical part. The main attention is paid to the air duct before
the rotor stage. The dimensions of the duct are as follows:
Length of the inlet channel . . . . . . . . . . . . . . . . . 700 mm
Blower dimension: inner diameter . . . . . . . . . . . . 256 mm
Outer diameter . . . . . . . . . . . . . . . . . . . . . . . . . . . 560 mm
Diameter of the nozzle . . . . . . . . . . . . . . . . . . . . . 450 mm
Inlet channel
The inlets are placed on the both sides of the airplane fuselage. These
two air streams are mixed together in close to the blower. The first
choice of the blower design was put in this order: rotor stage first and
stator as a second. The design condition for the flow at the rotor leading edge was set as a strictly axial flow and zero angel of attack to
23
the rotor blades. But hence the inlet channel arrangement the left and
right half of channel form a tailing edge. In this mixing area the streams go against themselves. That has a negative influence on the rotor
operation. [2] The angle of attack changes strongly in this area. (Fig.
1) The tailing edges are situated at twelve and six clock positions.
Improvement on this flow form should be achieved by the extension of the common duct. This is not preferred because of the duct
length restriction. The other chance is to set the stator vanes in front
of the rotor inside the inlet duct. The second variant increases the pressure losses. There is a third possibility — to twist the duct and supply the stator. The flow through this twisted duct forms the following
map of angles of attack. (Fig. 2) There are the tailing edges situated
at the two and seven clock position.
This variant promises improvement in rotor work. The flow is twisted and substitutes the stator. For the right function of the blower it is
necessary to get the axial flow through the nozzle. The rotor blade
geometry is defined by boundary diameters and power supplement.
The axial flow creates the stator behind the rotor stage. Other variant
is to twist the flow before the rotor. In both cases is the flow behind
the blower axial. The stator vanes were placed in front of the rotor
inside the inlet channel to reach the best condition for the rotor. The
total length of the duct 700 mm was preserved. This geometry was
tested in software Fluent version 6.
Model setting
An aerodynamic efficiency test was made. The computation volume
(Fig. 3) consists of three base parts: inlets, blower and simplified duct.
The inlet and duct were lengthened. All thirteen rotor blades and fourteen stator wanes were modelled for better approaching the flow non
uniformity.
Fig. 3 — The computational volume
3/2006
The computation setting:
The geometry was three-dimensional with single rotating frame. The
angular velocity was changing, the design point is at 7.6 thousand
rpm. The air speed was modelled as increase in the total pressure and
total temperature at the flow inlets. The flow was modelled as compressible and viscous. K - ω model of turbulence was employed. The
equation of flow and energy were coupled. The flow was solved as
time steady.
Results
The flow simulation for the design point is depicted in next figures.
(Fig 4) The design point air speed is 40 meter per second and the
angular velocity is 7,600 rpm. In the tangential velocity field behind
the rotor stage, axial cores are visible. The radial lines with increased
tangential velocities correspond to the shear layers behind the rotor
blades. The map of angle of rotor blade attack shows that the angles
little differ from zero. (Fig. 5) It might be due to some influence of
the rotor blades. The number of the increased areas corresponds to the
number of rotor blades, not stator vanes. This influence is hard to
reduce because of the vicinity of rotor and stator.
Conclusion
The designed geometry with the twisted channel and the stator vanes
included in the inlet channel is suitable to restrict the negative influence
of the side-inlet connection. On the hand the computed efficiency for this
unit does not exceed 88 % of all computed regimes. Also the tangential
velocity in the nozzle is not negligible. There forms a rotating core of
flow passing through the nozzle. This could be caused by overestimating
the viscous effects. Further inaccuracy is caused by modelling the flow
as the time steady flow, but in fact it is not always true. These questions
should be answered after building up a testing stand. This stand is before its completion, enabling the data to be validated soon.
24
Preliminary Tests of Friction Stir Welding
Úvodní zkoušky frikčního svařování
Ing. Petr Bělský / VZLÚ, Plc., Prague
The article gives short information about preliminary tests of Hi-Tech joining technology FSW at VZLÚ, Plc. Realized
experiments were mainly focused on friction welding of AA2024-T3 sheets and extruded profiles made of AA6060-T6.
The effects of welding parameters on mechanical properties and temperature field of butt welds have been studied.
Maximum tensile strength of the welds reached as much as 85% of base metal. Finally, short information about next
planned research FSW activities at VZLÚ is presented.
Příspěvek stručnou formou informuje o úvodních zkouškách moderní technologie FSW ve VZLÚ. Realizované
experimenty byly zaměřeny na třecí svařování plechů z Al slitiny 2024-T3 a svařování extrudovaných profilů ze slitiny
6060-T6. Zkoumán byl vliv svařovacích parametrů na mechanické vlastnosti a rozložení teplotního pole tupých svarů.
Během tahových zkoušek bylo dosaženo až 85% pevnosti základního materiálu. Na závěr je podána stručná informace
o dalších plánovaných aktivitách VZLÚ v oblasti FSW.
Keywords: Friction Stir Welding, butt joint, tensile test, aluminium alloy 2024-T3.
Introduction
Friction stir welding (FSW) is a solid-state joining process enabling
fabrication of low-cost lightweight structures. This revolutionary HiTech technology was invented and patented in 1991 by The Welding
Institute (TWI) in Cambridge.
The basic form of the process uses a cylindrical, non-consumable
tool, consisting of a flat circular shoulder, with a smaller probe protruding from its centre. Plunging a rotating probe into the adjoining
plates joins material of work-piece and subsequent heating caused by
friction between the rotating tool and material elevates the temperature of the local weld region high enough to plasticize material of workpiece. Through mechanical forces the heated material is extruded
from the front of the probe to the back as the probe transverses the
length of the joint. The combination of the frictional heat and mechanical working produces a solid-phase joint. Because no macroscopic
melting takes place the weld is left in a fine grain wrought structure
and other problems associated with liquid to solid transformation,
porosity, solidification cracking, residual stresses are all eliminated.
The friction process is environmentally friendly, as it does not require
consumables (filler wire, flux or gas) and produces no fumes. FSW
makes possible joining speeds 6 times faster than automated riveting
or 60 times faster than manual riveting, with improved quality.
The benefits utilizable also for aircraft structures were main reason
why VZLÚ, Plc. started its own research activities in the area. First
official tests of the modern joining technology were performed in
VZLÚ, Plc. in September 2006 (see Fig.1). These experiments were
mainly focused on tests of basic principle and demonstration of general possibilities of FSW.
Fig. 1 — Preliminary tests of FSW
in VZLU, Plc
FSW Tools
Two different tool configurations were used. The first tool for welding
3 mm thick 2024-T3 sheets had shoulder diameter of 13 mm, shoulder concavity angle of 7 degrees, pin diameter of 5 mm, and pin
length of 3 mm (see Fig.2). For welding 1.36 mm thick 2024-T3 sheets and extruded 6060-T6 profiles was used tool with shoulder diameter of 8 mm, shoulder concavity angle of 6 degrees, pin diameter
of 3 mm, and pin length 1.35 mm. No threads or scrolls on pin and
Experimental procedure
Materials
The test materials used in these preliminary experiments were sheets
from AA 2024-T3 and extruded profiles from AA6060-T6. Sheet
thicknesses for 2024-T3 were 3 and 1.36 mm. Profile wall thickness
was 1.4 mm. The nominal chemical composition and basic mechanical properties are listed in Table 1.
Sheets and extruded profiles were degreased prior to welding using
a household cleaner. No mechanical means of oxide removal were
employed. However, scale on the sheets and profiles was very light.
Fig. 2 — FSW tool for 3 mm thick sheets
Table 1 - Chemical composition and mechanical properties of AA 2024 and AA6060
25
3/2006
shoulder were used. Both tools were made of maraging steel Vascomax C-350 (18.5%Ni, 12%Co, 4.8%Mo, 1.4%Ti). Tool-to-workpiece
angle was maintained at 1°10' deg and a shoulder plunge of 0.15 mm
was used for all welds.
Machines and fixtures
First tests focused on welding of 3 mm thick sheets were carried out
on a manual knee-type milling machine FGU40 with 11kW spindle
drive motor power and a table size of 1100x350 mm. This old milling
machine is characterized by good stiffness and sufficient power
(spindle speed range: 35-1800 rpm, max. torque: 2000 Nm). VZLU,
Plc. intends to use this machine also in the future for some experiments (welding of steels?).
The main "FSW machine" of VZLU is a new NC bed-type milling
machine FSG 80 from TOS Kuřim. This machine has large table
clamping surface 2000x800 mm, high main spindle motor output 19.3 kW (spindle speed range: 20-4000 rpm, max. torque: 1000 Nm)
and control system HEIDENHAIN-iTNC530. Maximal feed trust in
linear axes is 20 kN. The milling machine is also equipped by HorstWitte vacuum clamping system and special wireless 3D measuring
probes. At an early date new FSW monitoring system LOSTIR will
be integrated into the machine for force feedback control of FSW tool
(see below). All 1.36 mm thick 2024-T3 sheets were welded with the
milling machine FSG 80.
All experiments were carried out on a various-purpose fixture (see
Fig.1). It was designed so that all types of weld configurations (butt,
lap and T joint) could be carried out. The fixture is also equipped with
other elements for demonstration activities (e.g. welding of panels
from extruded profiles). Maximum length of welds is 200 mm.
Welding Procedures
In order to understand the basic principle of FSW technology, the following preliminary experiments were performed:
a) Butt welds of 2024-T3 sheets with thickness 3mm
The welding process was carried out with the tool rotating at 710 rpm
and at a feed rate of 80 mm/min, with a 1°10' tilt angle and a 0.15 mm
plunge. Only one welded panel was cut and tensile tests were performed.
b) Butt welds of 2024-T3 sheets with thickness 1.36 mm
All these welds were carried out on NC milling machine FSG 80. At
the beginning of each weld the rotating tool slowly plunged into the
joint line between two plates 200x100 mm at the plunge feed rate
10 mm/min. After attaining full plunge 0.15 mm, tool dwelled at the
place for 7 seconds. Then followed accelerating (40 mm/min to
distance 15 mm) and transition to planned welding speed. Table 2 lists
the welding parameters used for all welds of 1.36 mm thick sheets.
During the experiments the transient temperatures were measured
using an array of thermocouples (see below).
Table 2 — Welding parameters for 1.36 mm thick sheets
c) Demonstrations of general possibilities of FSW
Main objective of the demonstrations was testing of general possibilities of the technology for various types of weld configurations and
applications. The most interesting outputs of the experiments
are T-profiles welded from three 2024-T3 sheets and thin-wall panels welded from extruded profiles made of AA6060-T6. Welding parameters for all the demonstrators were the same (1100 rpm,
100 mm/min).
Fig. 3 — Time-transient temperature evolution
Temperature measuring
Transient temperatures were measured during welding of 1.36 mm
thick sheets. There were 6 thermocouples placed in the middle of the
welded panels. Five thermocouples were placed on the top surface of
the sheets outside the weld region (3 on the advancing side + 2 on the
retreating side) and 1 thermocouple was placed from below of the
sheets closely to joint line. The thermocouples were of the k-type
using wires ?0.19mm with a welded ball junction of about ?0.5mm;
the wires were coated with glass fibres, which have a claimed working
temperature of up to about 400°C. The diameter of the coated wires
was also about 0.5mm. Data was acquired and saved using the USB
external measurement system meM-Adfo BPL-16. Example of temperature courses recorded during welding of panel FSW-1.36-4 is
shown in Figure 3. The figure shows that temperature on the advancing side is a little higher in comparison with the retreating side (thermocouples 1-4 and 2-5). Maximum temperature measured in distance 2.5 mm from the joint line reached 387° C.
Results and discussion
First welding trials with 3mm thick 2024-T3 sheets (710rpm,
80mm/min) were carried out without any problems. Welds appeared
very well. No surface or internal defects were evident.
Markedly greater problems appeared during welding of 1.36 mm
thick sheets. It was caused mainly by too high welding speed (300,
150, 100 mm/min). Temperature of plasticized material on the backside of the probe was not enough high and surface defects appeared
on the advancing side of the welds.
All welded panels were cut about 180 hours after welding and tensile tests were carried out. Although in some cases some defects were
observed, all welds were tensile tested on specimens 25x200 mm perpendicular to the weld. All tensile tests were performed on a 500-kN
TIRAtest 28500 S testing machine at a constant crosshead displacement rate of 1 mm/min. The maximum (failure) load and failure loca-
Fig. 4 — Failure location and fracture area appearance for the specimens with minimum and maximum tensile strength
26
tion were recorded for each specimen. Table 3 lists failure location(s),
average and maximum tensile strength in percents of base metal.
Abbreviations used in the table are: A-advancing side, R-retreating
side, t.n.-through nugget, ad.-advancing, LOP-Lack of Penetration.
A maximum joint efficiency 85% of base metal was achieved for
welded panel FSW-1.36-4 that was welded with low welding speed
50 mm/min. Specimens made from the panel failed on the retreating
side. A minimum joint efficiency 34% of base metal was achieved for
welded panel FSW-1.36-2. Differences in failure location and appearance of fracture area are evident in Figure 4.
Planned FSW activities in VZLÚ, Plc.
a) Integration of FSW monitoring system LOSTIR into milling machine FSG 80
VZLÚ participated in EU project LOSTIR in last 2 years. The main
objective of the project was the development of a low cost torque/force
monitoring device for conversion milling machines to FSW. VZLÚ is
the first user of the system in the world. At the moment the system is
intended for passive monitoring of forces acting during welding. The
best welding results is always achieved with equipment having force
control. That is why VZLÚ is going to integrate the system into new NC
milling machine FSG 80 for force feedback control.
b) Optimalization of welding parameters for lap joints of FSW demonstrator
FSW technology is a relatively new process and there is not enough
experience with fatigue behaviour of whole parts of airframe structures produced by FSW. That is why it is very important to obtain
enough knowledge and experience to help formulate pre-normative
Table 3 — Failure locations and tensile test data
tructural safety design and certification criteria for small and medium sized commercial airplanes. Fatigue test of demonstrator representative section of wing produced by FSW will provide a lot of useful information and practical experience. VZLÚ now cooperates with
Belgian company CEWAC on optimalization of welding parameters
for the demonstrator.
References:
[1]
Beamish K., Lewis P., Cheetham P.: Development of a low
cost FSW monitoring system; Paper presented at the 6IFSW
Symposium, Saint-Sauveur, Canada, October 2006
[2]
Bělský P.: Friction Stir Welding of Aircraft Structures; Letecký zpravodaj, No. 3, pp 15-17, VZLÚ, Plc., Prague, 2003
[3]
Bělský P.: The First Eurostir Workshop: ”FSW — making it
work for you“; Report C-2713/02, VZLÚ, Plc., Prague,
2002
Hodnoticí metodika projektů výzkumu a vývoje
Ing. Klára Grammetbauerová / VZLÚ, Plc., Prague
The Czech Republic as an EU member state has to accept the basic principles that were specified in the scope of
technical development and innovations by the highest EU authorities. Suitably structured, sufficiently financed and on
innovation focused research should be a source of ever lasting renewal of product and services competitiveness in the
EU. It is the basic condition for ensuring economic development. Innovation projects evaluation reflects the criteria of
Frascati and Oslo Manuals; included is also the Real Option Pricing Approach. ESA and, NASA methodologies, and
information from DLR are introduced too. The present Czech R&D evaluation and its results as it was introduced in
October 2006 is concerned with program providers´ evaluation (not project evaluation) and the aim is to evaluate
"effectiveness" of sources provided by each budget chapter for R&D support. Four evaluation options have been
ntroduced, based on identified typical R&D projects. Further verification will follow in the next phase according to
another R&D project solved together with ESA, within the PECS and according to NASA aeronautical projects or
DLR.
Česká republika jako členský stát EU musí přijmout základní principy, které pro oblast technického rozvoje a inovací
byly deklarovány nejvyššími orgány EU. Vhodně strukturovaný, dostatečně financovaný a proinovačně zaměřený
výzkum a vývoj se má stát zdrojem trvalé obnovy konkurenceschopnosti výroby a služeb v EU coby základní podmínka
zajištění hospodářského růstu. Hodnocení projektů výzkumu a vývoje odráží kritéria Frascatiho a Oslo manuálu,
zohledněno je hodnocení reálných opcí. Vlastní metodiku hodnocení má ESA, NASA a informativně je představena
i metodika DLR. Současné hodnocení výzkumu a vývoje v České republice jak bylo představeno v říjnu 2006 je
zaměřeno na hodnocení programů (nikoliv projektů) jednotlivých poskytovatelů a cílem je ohodnotit efektivitu
prostředků poskytnutých jednotlivými kapitolami státního rozpočtu na podporu vývoje a výzkumu. Na základě
identifikovaných typových projektů byly uvedeny 4 varianty hodnocení projektů V&V. V následující fázi bude
provedeno testování podle dalších projektů V&V řešených spolu s ESA, v rámci PECS a podle leteckých projektů
NASA nebo DLR.
Keywords: methodology, research & development projects, R&D, evaluation of.
Since the 90s together with changes taking place in the society, it is
possible to perceive modern understanding of R&D projects at the
territory of the Czech Republic and new approach to innovation. The
last important milestone was entrance of the Czech Republic and
other economics from Middle and Eastern Europe into the European
Union (in 2004). This move among other things has also exerted influ-
27
ence on understanding R&D projects. The main targets of innovation
researches is (1) monitoring of the innovation activities — TPP1 in
companies and sectors and (2) measurement of the total expenditures
of technological innovations of products and processes Several manuals are used for evaluation at the European level.
Oslo Manual Evaluation
For evaluation of innovation projects the Oslo manual2 serves as
a main reference source.— the measurement of scientific and technological activities proposed by guideline for collecting and interpreting
innovation data. It is possible to divide the measurement indicators of
total technological innovations into two groups:
(i) The sum of means allocated to R&D — data for R&D are
obtained through the national statistics. According to many studies this data are considered to be valuable however it has two
important limits:
■ this data are input into R&D and even it is clearly connected
with the technological change it does not measure it;
■ the data of R&D do not include all the impacts on the companies and government at the territory as there are also another sources of the technological change such as learning-bydoing and it leaves the narrow definition.
(ii) Patent statistics — the limit is in fact that many of the innovations did not correspond with patents and on the other hand
many patents are connected with minimal technological or economic change and many patents also did not lead to any innovations al all (according to summary of Patent, innovations and
economic performance Conference Proceedings, OECD 2004).
Because of the mentioned limits additionally is for innovation projects
suggested (Oslo Manual, p. 55) to collect some general data of the
firm, for the beginning and end of the three-year period:
■ sales at year t and t-2;
■ exports at year t and t-2;
■ employees at year t and t-2;
■ operating margin at year t and t-2.
Frascati Manual Evaluation
Frascati Manual3 is a Proposed Standard Practice for Surveys on
Research and Experimental Development. According to this manual
the total R&D expenditures can be measured as:
■
The total expenditures for innovation activities of the firm for
given year = subjective approach or innovative budget approach. This approach represents expenditures for implemented,
potential and non-successful technological innovation activities.
Update R&D position corresponds with R&D expenditures: the
manual includes R&D expenditures that are not directly connected with specific innovation project.
■
Total expenditures for implemented innovations regardless of
the year in which the expenditures occur = objective approach.
The relevant amount consists of the total technological innovations or of the main TPP innovations that were implemented in
the relevant period. It excludes expenditures for TPP non-
3/2006
successful innovations or still ongoing innovations and general
expenditures for R&D that are not connected with any specific
product or process application. This approach is found as suitable especially for innovative research, innovative research of
successful TPP innovations or TPP innovations that were implemented.
It is stated that R&D is only one step in innovation process and therefore expenditures for R&D are only one part of financial input.
Expenditures evaluation for all over aspects of technological innovations can lead ROI (Return On Investment) in innovations.
Real Option4 Pricing Approach
The company has either one R&D project or portfolio R&D projects
and opportunities at the same time. When company keeps own portfolio of R&D projects it is supposed that the externality created within one project has no significant influence on the other projects in
portfolio and it is possible to evaluate each project R&D separately.
Prior to entrance to the project the company evaluates expected quality of final result together with the income adherent to product marketing. Beyond are evaluated the costs that are caused by R&D. The
company then decides whether the R&D project will be implemented.
The main attribute of value estimate based on the real options pricing is that the greater standard deviation of expected profits increases
the asset value (from the option perspective). The negative outputs
will be in the future eliminated e.g. by decision not to continue, the
positive outputs can be realized in case of positive development [Soustružník, 2004]. It is also one of the basic characteristics of R&D sector. Within one economy have each sector different option characteristics. This characteristic is possible to perceive as the level of fluctuation of its values. The R&D sector is typical with quite large
standard deviation of revenues. Similarly in new sectors it is quite difficult to estimate all of the risks.
The research and development area has call option characteristics5
— the amount spent on research and development is a cost of call
option. Projects or products as a result of the research and development represents option premium. When the products are tangible then
the premium is the difference between initial investment and the cash
flow from the realized investment. According to Damodaran the
results from application of the option theory are as follows:
(a) R&D expenditures should provide higher value for the companies that operate in more volatile technologies or businesses,
since the variance in product or project cash flows is positively
correlated with the value of the call option6
(b) The value of research and the optimal amount to be spent on
research will change over time according to business sector
maturity. The best example is the pharmaceutical industry —
pharmaceutical companies spent most of the 1980s by investing
substantial amounts in research and earning high returns on new
products, as the health care business expanded. In the 1990s,
however, as health care costs started leveling off and the business matured, many of these companies found that they were
not getting the same premium on research and started cutting
1
A technological product innovation is the implementation/ commercialization of a product with improved performance characteristics such as to deliver objectively new or improved services to the consumer. A technological process innovation is the implementation/adoption of new or significantly improved production or delivery methods. It may involve changes in equipment, human resources, working methods or a combination of these. See Oslo Manual p. 9.
2
According to Oslo Manual, further international standards and related concepts.
3
Frascati Manual, ISBN 80-8070-157-1, OECD 2002, http://www1.oecd.org/publications/e-book/9202081E.PDF
4
Note: The idea of options (usually financial options) is certainly not new. Ancient Romans, Grecians, and Phoenicians traded options against outgoing cargoes
from their local seaports. When used in relation to financial instruments, options are generally defined as a "contract between two parties in which one party
has the right but not the obligation to do something, usually to buy or sell some underlying asset". Having rights without obligations has financial value, so
option holders must purchase these rights, making them assets. This asset derives their value from some other asset, so they are called derivative assets. Call
options are contracts giving the option holder the right to buy something, while put options, and conversely entitle the holder to sell something.
5
In general an option contract that gives the holder the right to buy a certain quantity of an underlying security from the writer of the option, at a specified
price (the strike price) up to a specified date (the expiration date). also called call option.
6
Thus Minnesota Mining and Manufacturing (3M), which expends a substantial amount on R&D on basic office products, such as the Post-it pad, should
receive less value for its research than biotechnological companies or companies operating in aeronautical and space research.
28
back. Some companies moved research investments from conventional drugs to bio-technology products, where the uncertainty about future cash flows remains high and therefore expected premium is high as well. [Damodaran, p. 49].
As a good reason for evaluation via option pricing models it is also
taken the opportunity of evaluation of project abandon option. When
the project does not bring the expected cash flow then it is possible to
evaluate the finishing project option as a difference between the present value of the project in case of following the project till the end of
the project and present value of project disposal. The option in R&D
is then either to continue in the research in case of the positive development or to quit the project in the negative case. In case of increasing accumulation of capital and technologies there is increasing
number of the options for its usage for further investment projects.
As a real option pricing model is used Black-Scholes7 model option pricing however this model is assumed moreover for European
option, i.e. an option that can be exercised only at its expiration date.
On the other hand there is also the second type of options — American option i.e. an option that can be exercised at any time until its
expiration date. This second type reflects better the option of potential project abandon.
Nevertheless it is on discussion whether to use this model with respect to the prerequisites of its assumptions — e.g. premise of continual price changing of underlying asset that changes accidentally, etc.
At the real options pricing there is often used binomial model8 that
uses only elementary mathematics and that has minimum prerequisites.
ESA Evaluation
When evaluating the R&D projects in the machinery and aviation sector there are included ESA criteria coming from the specific requirements of the project. It is possible to sum up them to the following
groups:
1. Education and experiences of the project applicants (connected
in general with relevant research area) in dimension either at the
applicants or at the company as a whole together testing facilities adequacy, etc.)
2. Understanding of the requirements and targets and technical discussion of the solved areas.
3. Quality and eligibility of the offered work plan, engineer access
adequacy
Those 3 criteria usually have high percentage of weight valuation:
70%. Further 2 criteria have value 30%. Those criteria are as follows:
4. Management suitability and cost adequacy to the work that
should be done.
5. Consensus with the tender administrative conditions and contract conditions acceptance.
All of the criteria must be fulfilled at least at the minimum level
(50%). Otherwise the project proposal is rejected from further examination.
Project value scale according to ESA
Project is valuated by these values — the committee must agree with
one available value
Table 1 — ESA evaluating scale of project
Value
100
90
75
60
50
40
0
Verbally
perfect
excellent
very good
good
average
hardy acceptable
worthless
NASA9 Evaluation
It is used the methodic of the Technology Readiness Level (TRL).
Technology Readiness Levels (TRLs) represent a systematic
metric/measurement system that supports assessments of the maturity of a particular technology and the consistent comparison of maturity between different types of technology. The TRL approach has
been used on-and-off in NASA space technology planning for many
years and was recently incorporated in the NASA Management
Instruction (NMI 7100) addressing integrated technology planning at
NASA. However, to be most useful the general model must include:
(a) 'basic' research in new technologies and concepts (targeting identified goals, but not necessary specific systems), (b) focused technology development addressing specific technologies for one or more
potential identified applications, (c) technology development and
demonstration for each specific application before the beginning of
full system development of that application, (d) system development
(through first unit fabrication), and (e) system 'launch' and operations.
Technology Readiness Levels Summary
TRL 1 Basic principles observed and reported
TRL 2 Technology concept and/or application formulated
TRL 3 Analytical and experimental critical function and/or characteristic proof-of concept
TRL 4 Component and/or breadboard validation in laboratory environment
TRL 5 Component and/or breadboard validation in relevant environment
TRL 6 System/subsystem model or prototype demonstration in
a relevant environment (ground or space)
TRL 7 System prototype demonstration in a space environment
TRL 8 Actual system completed and ”flight qualified“ through test
and demonstration (ground or space)
TRL 9 Actual system ”flight proven“ through successful mission
operations
At the attachment there is the technology discussion for each level.
DLR10 Evaluation
The primary goal of avionic research activities is to ensure the competitiveness of Germany and the whole Europe in space and aeronautics and reach the government and society targets. Moreover to the
basic research is DLR orientated to the applied space research and
development. DLR set up itself the as a challenge to create rapidly
growing air transportation which is effective, environmental friendly
and sustainable in this aspects. DLR space technologies portfolio reflects European strategic vision ”Vision 2020“ a respective German
7
Black-Scholes model — The first complete mathematical model for pricing options, developed by Fischer Black and Myron Scholes. It examines market price,
strike price, volatility, time to expiration, and interest rates. It is limited to only certain kinds of options.
8
An options valuation method developed by Cox, et al, in 1979. The binomial option pricing model uses an iterative procedure, allowing for the specification of
nodes, or points in time, during the time span between the valuation date and the option's expiration date. The model reduces possibilities of price changes, removes the possibility for arbitrage, assumes a perfectly efficient market, and shortens the duration of the option. Under these simplifications, it is able to provide a
mathematical valuation of the option at each point in time specified. The binomial model takes a risk-neutral approach to valuation. It assumes that underlying
security prices can only either increase or decrease with time until the option expires worthless. For the compound options (combination of several type of options) is preferable the binomial model. See http://www.investopedia.com/terms/b/binomialoptionpricing.asp
9
According to John C. Nankine: TECHNOLOGY READINESS LEVELS, A White Paper, April 6, 1995, Advanced Concepts Office, Office of Space Access
and Technology, www.hq.nasa.gov/office/codeq/trl/trl.pdf
10
Deutsches Zentrum fur Luft und Raumfahrt (ie. German Center for Aeronautical and Space Research), http://www.dlr.de/en/Desktopdefault.aspx/tabid10/81_read-59/
29
response ”Luftfahrt 2020“. The main aims are:
■ tCost reduction of aeronautical transport of 30%
■ tReduction of percentage decrease of 80%
■ tIncrease of share of European aeronautic transport up to 16
mil. flights a year
■ tDecrease of carbon and nitrogenous emission to 50%, 80%
respectively
■ tDecrease level of noise of 50%
In the own research project evaluation in DLR it is used scenarios approach (see http://www.dlr.de/tt/system).
Evaluation of Type Projects — model options
The present Czech R&D evaluation and its results in 2006 as it
was introduced in October 2006 is handling with program provider's evaluation and the aim is to evaluate ”effectiveness“ of sources provided by each budget chapter for R&D support. The result
might be modified by each provider's panel. This methodology
does not have the aim to evaluate each project or research intention. For preparing the methodology of evaluation of R&D projects in the meaning of each project evaluation is therefore suitable to take into account existing methodologies of different
institutions.
At the first phase for model options of project evaluation it was
used type projects with different type of financing. Identified
types of project include European research project (CO2SINK),
Project CRAFT (cooperative research project), Project from
Eureka program — Pollutdegradcell and Project from machinery
and aeronautical sectors — ESA type project.
Option 1 — Project as a Target
The basic postulate of the R&D evaluation at the level of projects is
the most objective statement how the targets of offered/ solved project or other activity contribute to the target of research program and
its fulfillment. As a starting point is necessary to understand own project target whether it fulfills the SMART11 condition of targets.
■ Offered solution has specific description.
■ Offered solution is measurable — it is stated how to recognize when the solution is successful, solution verification should
be stated at the beginning.
■ The solution must meet the needs of recipient.
■ Offered solution must be realistic.
■ It is stated the timeframe.
The criteria should be taken into account especially in the phase of
decision making of financial means aimed for selected project. With
exception of ESA project all of the projects were accepted and
SMART condition was fulfilled.
Option 2 — Evaluation According to Criteria
— Project Scoring
R&D evaluation should be done not according to activities and inputs
but according to results. For project evaluation according to results
with respect to the existing methodologies it is found as the most suitable evaluation according to group of criteria (project scoring) rather
than evaluation according to one criterion only. Project scoring includes criteria that reflect how successful the project solution is and it
is possible to sum up several characteristics of successful projects. At
the typical projects were found following characteristics:
According to the share of all of the items in the project it will be stated the weight of each item in the total evaluation. For further specifying of the weights and type of criteria it is necessary to include in
testing another projects.
11
SMART = specific, measurable, aligned, realistic, timed.
3/2006
Option 3 — Project Evaluation with the Option of
Abandon the Project
This option includes opportunity to decrease the risk and the possibility to leave the project according to the progress in the project. The
criterion is the difference between the present value in case of following in the project and the present value of the project liquidation. It
is also possible to evaluate the option to leave the project.
For specifying and differentiation of each phase it is suitable to
have stated the measurements of successful termination, the check
points for project progress and to have ensured that there are specified risks in case of new technologies. Evaluation according to results it
should be included if the results are applicable and clear; and if there
is the difference between a quality and successful solution of the project.
The statement of discount rate for (net) present value is fundamental. Important is also to state the phases as it is shown in e.g. the
ESA project. The payment is here done after fulfillment of each
phase. The reason is to restrict the waste over the stated level. The criterion is the waste that will have the company in case of leaving the
research option. For using of this Project evaluation option it is necessary to know the company expenditures for innovation activities and
also correct stating of expenditures according to Oslo and Frascati
manuals.
Option 4 — Combination of Previous Options
The combination of option 2 and 3 uses as a selecting criterion the risk
level. E.g. in case that there are two sources of financing of the investment the risk is shared as the investment is shared. I.e. when 1/3 of the
investment project is paid by one side and 2/3 by the other partner then
the risk of abortion is divided to 1/3 for the first partner and 2/3 for the
second partner. The risk is possible to understand as a one of the evaluation criteria together with others mentioned in previous options.
Conclusion
The existing evaluation methodologies of R&D programs together
with the identified type projects can be understood as a starting point
for new evaluation model options for R&D projects. The offered methodology options refer to the identified projects. In the following
phase it will be done further verification according to another R&D
project solved together with ESA, within the PECS (Plan for European Cooperating States; http://www.czechspace.cz/cs/veda-a-vyzkum)
and according to NASA aeronautical projects or DLR.
Attachement
The Black and Scholes Model
Note: Since 1973, the original Black and Scholes Option Pricing
Model has been the subject of much attention. Many financial scholars have expanded upon the original work. In 1973, Robert Merton
relaxed the assumption of no dividends. In 1976, Jonathan Ingerson
went one step further and relaxed the assumption of no taxes or transaction costs. In 1976, Merton responded by removing the restriction
of constant interest rates. The results of all of this attention that originated in the autumn of 1969 are alarmingly accurate valuation models
for stock options.
The Black and Scholes Option Pricing Model
Fisher Black started out working to create a valuation model for stock
warrants. This work involved calculating a derivative to measure how
30
the discount rate of a warrant varies with time and stock price. The
result of this calculation held a striking resemblance to a well-known
heat transfer equation. Soon after this discovery, Myron Scholes joined Black and the result of their work is a startlingly accurate option
pricing model. Black and Scholes can't take all credit for their work;
in fact their model is actually an improved version of a previous model
developed by A. James Boness in his Ph.D. dissertation at the University of Chicago. Black and Scholes’ improvements on the Boness
model come in the form of a proof that the risk-free interest rate is the
correct discount factor, and with the absence of assumptions regarding investor's risk preferences.
4) No commissions are charged
Usually market participants do have to pay a commission to buy or
sell options. Even floor traders pay some kind of fee, but it is usually
very small. The fees that Individual investor’s pay is more substantial
and can often distort the output of the model.
5) Interest rates remain constant and known
The Black and Scholes model uses the risk-free rate to represent this
constant and known rate. In reality there is no such thing as the riskfree rate, but the discount rate on U.S. Government Treasury Bills
with 30 days left until maturity is usually used to represent it. During
periods of rapidly changing interest rates, these 30 day rates are often
subject to change, thereby violating one of the assumptions of the
model.
6) Returns are log normally distributed
This assumption suggests, returns on the underlying stock are normally distributed, which is reasonable for most assets that offer options.
Source: http://bradley.bradley.edu/~arr/bsm/pg04.html
Technology Readiness Level
In order to understand the model itself, we divide it into two parts.
The first part, SN(d1), derives the expected benefit from acquiring
a stock outright. This is found by multiplying stock price [S] by the
change in the call premium with respect to a change in the underlying
stock price [N(d1)]. The second part of the model, Ke(-rt)N(d2),
gives the present value of paying the exercise price on the expiration
day. The fair market value of the call option is then calculated by
taking the difference between these two parts.
Assumptions of the Black and Scholes Model:
1) The stock pays no dividends during the option’s life
Most companies pay dividends to their share holders, so this might
seem a serious limitation to the model considering the observation that
higher dividend yields elicit lower call premiums. A common way of
adjusting the model for this situation is to subtract the discounted
value of a future dividend from the stock price.
2) European exercise terms are used
European exercise terms dictate that the option can only be exercised
on the expiration date. American exercise term allow the option to be
exercised at any time during the life of the option, making American
options more valuable due to their greater flexibility. This limitation is
not a major concern because very few calls are ever exercised before
the last few days of their life. This is true because when you exercise
a call early, you forfeit the remaining time value on the call and collect the intrinsic value. Towards the end of the life of a call, the remaining time value is very small, but the intrinsic value is the same.
3) Markets are efficient
This assumption suggests that people cannot consistently predict the
direction of the market or an individual stock. The market operates
continuously with share prices following a continuous process. To
understand what a continuous process is, you must first know that
a Markov process is ”one where the observation in time period t
depends only on the preceding observation.“ This process is simply
a Markov process in continuous time. If you were to draw a continuous process you would do so without picking the pen up from the
piece of paper.
TRL 1
Basic principle observed and reported
This is the lowest ”level“ of technology maturation. At this level,
scientific research begins to be translated into applied research and
development. Examples might include studies of basic properties of
materials (e.g., tensile strength as a function of temperature for a new
fiber).
Cost to Achieve: Very Low 'Unique' Cost (investment cost is borne by
scientific research programs)
TRL 2
Technology concept and/or application formulated
Once basic physical principles are observed, then at the next level of
maturation, practical applications of those characteristics can be
'invented' or identified. For example, following the observation of high
critical temperature (Htc) superconductivity, potential applications of
the new material for thin film devices (e.g., SIS mixers) and in instrument systems (e.g., telescope sensors) can be defined. At this level,
the application is still speculative: there is not experimental proof or
detailed analysis to support the conjecture.
Cost to Achieve: Very Low 'Unique' Cost (investment cost is borne by
scientific research programs)
TRL 3
Analytical and experimental critical function and/or
characteristic proof-of-concept
At this step in the maturation process, active research and development (R&D) is initiated. This must include both analytical studies to
set the technology into an appropriate context and laboratory-based
studies to physically validate that the analytical predictions are correct. These studies and experiments should constitute ”proof-of-concept“ validation of the applications/concepts formulated at TRL 2. For
example, a concept for High Energy Density Matter (HEDM) propulsion might depend on slush or super-cooled hydrogen as a propellant: TRL 3 might be attained when the concept-enabling phase/temperature/pressure for the fluid was achieved in a laboratory.
Cost to Achieve: Low 'Unique' Cost (technology specific)
TRL 4
Component and/or breadboard validation in laboratory environment
Following successful ”proof-of-concept“ work, basic technological
elements must be integrated to establish that the ”pieces“ will work
together to achieve concept-enabling levels of performance for a component and/or breadboard. This validation must devise to support the
concept that was formulated earlier, and should also be consistent
with the requirements of potential system applications. The validation
is relatively ”low-fidelity“ compared to the eventual system: it could
31
3/2006
be composed of ad hoc discrete components in a laboratory. For
example, a TRL 4 demonstration of a new 'fuzzy logic' approach to
avionics might consist of testing the algorithms in a partially computer-based, partially bench-top component (e.g., fiber optic gyros)
demonstration in a controls lab using simulated vehicle inputs.
Cost to Achieve: Low-to-moderate 'Unique' Cost (investment will be
technology specific, but probably several factors greater than investment required for TRL 3)
logy and/or subsystem application is mission critical and relatively
high risk. Example: the Mars Pathfinder Rover is a TRL 7 technology demonstration for future Mars micro-rovers based on that system
design. Example: X-vehicles are TRL 7, as are the demonstration projects planned in the New Millennium spacecraft program.
Cost to Achieve: Technology and demonstration specific, but a significant fraction of the cost of TRL 8 (investment = ”Phase C/D to TFU“
for demonstration system)
TRL 5
TRL 8
Component and/or breadboard validation in relevant
environment
At this, the fidelity of the component and/or breadboard being tested
has to increase significantly. The basic technological elements must
be integrated with reasonably realistic supporting elements so that the
total applications (component-level, sub-system level, or systemlevel) can be tested in a 'simulated' or somewhat realistic environment.
From one to several new technologies might be involved in the
demonstration. For example, a new type of solar photovoltaic material promising higher efficiencies would at this level be used in an actual fabricated solar array 'blanket' that would be integrated with power
supplies, supporting structure, etc., and tested in a thermal vacuum
chamber with solar simulation capability.
Cost to Achieve: Moderate 'Unique' Cost (investment cost will be
technology dependent, but likely to be several factors greater that cost
to achieve TRL 4)
TRL 6
System/subsystem model or prototype demonstration
in a relevant environment (ground or space)
A major step in the level of fidelity of the technology demonstration
follows the completion of TRL 5. At TRL 6, a representative model
or prototype system or system - which would go well beyond ad hoc,
'patch-cord' or discrete component level bread boarding — would be
tested in a relevant environment. At this level, if the only 'relevant
environment' is the environment of space, then the model/prototype
must be demonstrated in space. Of course, the demonstration should
be successful to represent a true TRL 6. Not all technologies will
undergo a TRL 6 demonstration: at this point the maturation step is
driven more by assuring management confidence than by R&D requirements. The demonstration might represent an actual system application, or it might only be similar to the planned application, but using
the same technologies. At this level, several-to-many new technologies might be integrated into the demonstration. For example, a innovative approach to high temperature/low mass radiators, involving liquid droplets and composite materials, would be demonstrated to TRL
6 by actually flying a working, sub-scale (but scaleable) model of the
system on a Space Shuttle or International Space Station 'pallet'. In
this example, the reason space is the 'relevant' environment is that
microgravity plus vacuum plus thermal environment effects will dictate the success/failure of the system — and the only way to validate
the technology is in space.
Cost to Achieve: Technology and demonstration specific; a fraction of
TRL 7 if on ground; nearly the same if space is required
TRL 7
System prototype demonstration in a space environment
TRL 7 is a significant step beyond TRL 6, requiring an actual system
prototype demonstration in a space environment. It has not always
been implemented in the past. In this case, the prototype should be
near or at the scale of the planned operational system and the demonstration must take place in space. The driving purposes for achieving
this level of maturity are to assure system engineering and development management confidence (more than for purposes of technology
R&D). Therefore, the demonstration must be of a prototype of that
application. Not all technologies in all systems will go to this level.
TRL 7 would normally only be performed in cases where the techno-
Actual system completed and ”flight qualified“
through test and demonstration (ground or space)
By definition, all technologies being applied in actual systems go
through TRL 8. In almost all cases, this level is the end of true 'system development' for most technology elements. Example: this would
include DDT&E through Theoretical First Unit (TFU) for a new reusable launch vehicle. This might include integration of new technology into an existing system. Example: loading and testing successfully
a new control algorithm into the onboard computer on Hubble Space
Telescope while in orbit.
Cost to Achieve: Mission specific; typically highest unique cost for
a new technology
(investment = ”Phase C/D to TFU“ for actual system)
TRL 9
Actual system ”flight proven“ through successful mission operations
By definition, all technologies being applied in actual systems go
through TRL 9 in almost all cases, the end of last 'bug fixing' aspects
of true 'system development'. For example, small fixes/changes to
address problems found following launch (through '30 days' or some
related date). This might include integration of new technology into an
existing system (such operating a new artificial intelligence tool into
operational mission control at JSC). This TRL does not include planned product improvement of ongoing or reusable systems. For example, a new engine for an existing RLV would not start at TRL 9: such
'technology' upgrades would start over at the appropriate level in the
TRL system.
Cost to Achieve: Mission Specific; less than cost of TRL 8 (e.g., cost
of launch plus 30 days of mission operations)
Type project with different sources of financing
Example of European research project (CO2SINK)
Following the commitments made by the Kyoto Protocol, EU
countries are challenged to reduce their emissions of CO2 by 8%
during the period 2008 to 2012. There are several options for CO2
reduction in the power and heat sector. However, CO2 capture and
geological storage is the only one that has the potential to achieve substantial CO2 reductions at acceptable cost levels over the
next few decades. To address and alleviate potential public concerns about the safety and environmental impact of geological
storage, a better understanding of CO2 storage is needed
The CO2SINK integrated project, supported under the FP/6 framework by the EU commission, aims to develop the basis for this storage technique by injecting CO2 into a saline aquifer near the town of
Ketzin, west of Berlin. The project will develop an in situ laboratory
for CO2 storage to fill the gap between the numerous conceptual engineering and scientific studies on geological storage and a fully-fledged onshore storage demonstration. The project started in April 2004.
To characterize the underground and understand the processes
which happen there, detailed analysis will be made of samples of
rocks, fluids and micro-organisms from the underground. The project
involves intensive monitoring of the injected CO2 using a broad range
of geophysical and geochemical techniques, the development and
benchmarking of numerical models, and the definition of risk-assess-
32
ment strategies. These steps will all help to evaluate the reservoir’s stability and integrity.
Particular attention will be given to:
● The quality of the geological seals and the possibility of leakage through overlying strata
● Upward migration of gas along artificial pathways (such as
the metal casing of injection/observation boreholes)
● Migration of CO2 within reservoirs
● The rate at which CO2 dissolves in brine-filled reservoirs or
reacts with indigenous minerals.
● Understanding fate of CO2 and developing a risk assessment
for the long-term evolution of the CO2 storage.
The project is coordinated by the GeoForschungsZentrum (GFZ)
Potsdam.
The main cause of climate change or global warming effects is
believed to be the accumulation of the carbon dioxide (CO2) gas in the
atmosphere. This accumulation is a result of the extensive burning of
fossil fuels that began during the Industrial Revolution.
We can reduce the volume of CO2 emitted into the
atmosphere by collecting and storing it deep underground.
The concept is a simple one, but establishing whether the
technique can be applied safely, defining the necessary
effort for monitoring and selecting a suitable location in
which to test is challenging. On this website, you will find
details of experiments that will be conducted in a small
geological reservoir located in Germany.
The CO2SINK project started in April 2004. The project
centers on careful observation of the effects of injecting a significant amount of CO2 into a reservoir. The project team
must demonstrate that the project will be safe and environmentally acceptable before any injection begins.
Further information — www.co2sink.org
Example of CRAFT project (cooperative research
project)
Project ICACOST — Individually configurable automatic
cost calculation system for 3-d laser cutting COOPERATIVE (SMEs-Co-operative research contracts)
Project Value:
0.61Meuro, EC Contribution:
0.61Meuro
Partners (9): (Prime Contractor) LASER ZENTRUM HANNOVER
E V, GERMANY, FUNDACION ROBOTIKER, SPAIN, MARS
LASERTECHNIK GMBH, GERMANY, LASER TECH SPOL
S R O, CZECH REPUBLIC, WILCO WILKEN LASERTECHNIK
GMBH & CO, GERMANY, DISMODEL, S A, SPAIN, CLW
CLAUSTHALER LASER UND WERKSTOFFTECHNIK GMBH,
GERMANY, TUBECUT EINGETRAGENER KAUFMANN, GERMANY, GUALINI LAMIERE INTERNATIONAL SPA, ITALY.
Description: At present, small and medium-sized enterprises (SME's)
in the sheet metal industry have to invest much time in offer preparations, because of the low probability of getting the job. The offer calculation is mostly done manually as estimation. For example, the
length of the contours to be cut is extracted from drawings and summed up. Actual production time for work piece problem areas like
sharp angles and narrow radii can only be calculated by a post-processor simulation or by a comparison with a similar work piece that
was manufactured before. This complicates the cost calculation. Therefore, only experienced employees can estimate the costs for the cutting of 3-d work pieces. The aim of the proposed software ICACOST
is the quick and automatic cost calculation for 3d laser cutting. Less
experienced persons shall be able to use the configured tool. Therefore, characteristic numbers are generated on the basis of the work
piece geometry. They describe all necessary machine work to perform
and the work piece specific problem areas. As next step, the dynamic
machine behavior for these problem areas is projected to specific
machine parameters. By connecting the characteristic numbers with
the machine parameters, the machining time for a specific machine is
calculated. This machining time is an important factor for the cost calculation. The characteristic numbers are also used to find similar work
pieces within the planned offer database. This database is also implemented during the project and contains all created offers. As plausibility check the user can search for similar offers and compare them
with the new one. A complete cost calculation must include all costs,
which are important for the production, not only the machining time.
These diverse parameters come from various sources, and must be
combined with each other to get a precise result. This will be done by
standard formulas stored within the database.
Source:
http://pi.ijs.si/PiBrain.exe?Cm=Project&Project=ICACOST&Reference=508220
Example of Eureka project — Project Pollutdegradcell
Source: http://www.eureka.be/inaction/AcShowProject.do;jsessionid=7f00000122b83855f9f0f4654695a636d19085c8659d?id=1438
Example of ESA project — Project from the aeronautical sector
The project is settled as a tender; the interest party enlists and should
prove fulfilling of the stated requirements. The fulfillment should be
proved. Applicant will recognize from the entering conditions not
only general information related to the specific activity such as requirements for the technical abstract, the rough amount for what can be
the solution proposed, the type of R&D program, contact persons,
etc. The project must respond to all of the requirements that arise
from the tender. In other case is the project included neither in the
evaluation phase.
The proposal typically includes Introduction in stated format,
summary of the project, technical discussion and administrative
items of proposal. These items include management proposal,
financial proposal of the project, and optional comments to the
agreement. However in case of ESA agreement it is not supposed
any comment.
Potential applicants should prove the interest by the specialization field and indicate their ability of cooperation with the companies from the different sector as subcontractors or providers of the
services. Also it is suitable to prove the ISO certification with the
impact on the whole project for required quality reasons.
With respect to the fact that the project is usually only one of the
company activities it is proper to identify the position that will
handle with the project but also with the daily activities they will
33
3/2006
be included in the company hierarchy (the company hierarchy is includes building of facilities, wage costs, accessories from the
subcontractors and providers). The calculation includes also propart of the documents sent with the proposal).
The project includes description of each phase that was identifi- fit. As acceptable it is understood a profit of 8 % of the total offeed for the solved project. The phases are stated as they follow and red costs. Final calculated price is fixed.
The applicant should be clearly informed about actual terms of
also how are stated the check points. There is also included a financial calculation of the items necessary for the project (it includes payment. The total payment can be divided into a prepayment at
building of facilities, wage costs, accessories from the subcontrac- the beginning of the transaction - e.g. for facility building. Furttors and providers). The calculation includes also profit. As accep- her payment is for the progress in the solution and then is the
table it is understood a profit of 8 % of the total offered costs. Final finally payment, It is suitable to have some share of payment in
percentage, e.g. prepayment of 30% of the total amount based on
calculated price is fixed.
The applicant should be clearly informed about actual terms of agreement and signature by both contractual sides.
payment. The total payment can be diviTable 3 — Evaluation according to European Innovation Bulletin
ded into a prepayment at the beginning of
the transaction — e.g. for facility building. Further payment is for the progress
in the solution and then is the finally payment, It is suitable to have some share of
payment in percentage, e.g. prepayment
of 30% of the total amount based on
agreement and signature by both contractual sides.
Source:
http://www.eureka.be/inaction/AcShowProject.do;jsessionid=7f00000
122b83855f9f0f4654695a636d19085c8659d?id=1438
Source: NIP 2005 (Table 4),
http://www.cpkp.cz/regiony_old/rkv/nrp/v2_20050725_inovace.htm#
_Toc110071312
Sources:
Example of ESA project — Project from the aeronautical sector
[1]
http://www.solvaypharma.pl/index.php/ida/20
The project is settled as a tender; the interest party enlists and
should prove fulfilling of the stated requirements. The fulfillment
should be proved. Applicant will recognize from the entering conditions not only general information related to the specific activity such as requirements for the technical abstract, the rough
amount for what can be the solution proposed, the type of R&D
program, contact persons, etc. The project must respond to all of
the requirements that arise from the tender. In other case is the
project included neither in the evaluation phase.
The proposal typically includes Introduction in stated format,
summary of the project, technical discussion and administrative
items of proposal. These items include management proposal,
financial proposal of the project, and optional comments to the
agreement. However in case of ESA agreement it is not supposed
any comment.
Potential applicants should prove the interest by the specialization field and indicate their ability of cooperation with the companies from the different sector as subcontractors or providers of the
services. Also it is suitable to prove the ISO certification with the
impact on the whole project for required quality reasons.
With respect to the fact that the project is usually only one of
the company activities it is proper to identify the position that will
handle with the project but also with the daily activities they will
be included in the company hierarchy (the company hierarchy is
part of the documents sent with the proposal).
The project includes description of each phase that was identified for the solved project. The phases are stated as they follow
and also how are stated the check points. There is also included
a financial calculation of the items necessary for the project (it
[2]
http://www.optel.pl/ue/polska/ueproject.htm
[3]
http://www.rdaova.cz/eic_tendry_c.php?cid=
359http://www.eureka.be/inaction/
AcSearchProject.do?partnerSearch=on
[4]
http://www.ueapme.com/business-support/
Training%20tools/Czech%20Republik/
CZ-MECA%20Research%20and%20technology%20instruments.pdf
[5]
http://www.cpkp.cz/regiony_old/rkv/nrp/v2_20050725_inovace.htm#_Toc110071312http://
www.econtent.cz/docs/pracprog.doc
[6]
http://www.avo.cz/dotproj.htm
[7]
http://www.vlada.cz/1250/rvv/cep/cepfind.sqw?dba=CEP
[8]
http://www.cebre.cz/publikaceVaV/VaV.pdf
[9]
http://www.cdv.cz/text/publik/evropske-programy-provyzkum-a-vyvoj.pdf
[10] http://www.businessinfo.cz/cz/clanky/zdroje-financovani-zeu/program-eureka/1000522/38574/
[11] http://www.avo.cz/dotproj.htm
[12] http://www.vyzkum.cz/storage/att/
525E354752EFA0C10CCB02462AC8B40A/
Navrh%20hodnoceni%20schvaleny%20vladou%20644.doc
[13] Damodaran, A.: The Promise and Peril of Real Options,
http://pages.stern.nyu.edu/~adamodar/
14]
Soustružník, J.: Reálné opce, hodnota akcií a růst ekonomiky; 02.10.2004, www.patria.cz
15]
http://www.investopedia.com/terms/b/binomialoptionpricing.asp
16]
http://bradley.bradley.edu/~arr/bsm/pg04.html
34
Únavové zkoušky a výpočty životnosti podvozku letounu VUT 100
Ing. Petr Augustin, Ph.D., Ph.D., Ing. Martin Plhal, Ph.D., Ing. Jan Šplíchal /
Institute of Aerospace Engineering, Brno University of Technology
The paper deals with the fatigue evaluation of the VUT 100 Cobra airplane landing gear. Institute of Aerospace
Engineering BUT participates in the solution of this problem with fatigue calculation based on the results of FEM
stress analysis. IAE also provides generation of load sequence and data for the control electronic of servohydraulic
testing system for fatigue tests performed at the laboratory of the landing gear producer Technometra Radotin Plc.
The fatigue tests of both main and nose landing gears are currently under way Their results allow verification of
fatigue analyses and improvement in their methodology.
Příspěvek se zabývá hodnocením únavové životnosti podvozku letounu VUT 100 Cobra. Letecký ústav FSI VUT v Brně
se podílí na řešení této problematiky výpočty bezpečné životnosti založenými na výsledcích napěťové analýzy provedené
pomocí MKP. V experimentální oblasti zajišťuje přípravu zatěžovací sekvence a dat pro řídicí systém zařízení pro
únavové zkoušky, které jsou prováděny na zkušebně výrobce podvozku, a.s. Technometra Radotín. V současné době
probíhají únavové zkoušky hlavního i příďového podvozku. Jejich výsledky umožňují verifikaci únavových výpočtů
a zpřesnění jejich metodiky.
Keywords: aircraft landing gear, fatigue testing, load sequence generation, analysing.
Introduction
Institute of Aerospace Engineering collaborates on fatigue evaluation
of the VUT100 Cobra landing gear with EVEKTOR Ltd and Technometra Radotin Plc. IAE participates by fatigue calculations based on
the results of FEM stress analysis. In the experimental field, IAE provides generation of load sequences and data for the control electronic
of servohydraulic testing system for fatigue tests. After the initial fatigue analyses, fatigue tests of main and nose landing gear has been
started at the laboratory of the landing gear producer Technometra
Radotin (see Figure 1).
Load sequence generation
The fatigue analyses and tests are performed with similar sequences
of loads randomly generated using the in-house software FLTSIM
[4]. This program requires the definition of load cases and load spectra. Load cases represented by two types of landing impact, rebound,
taxiing and turning and ground load spectra shown in Figure 2 were
specified by the airplane manufacturer [5, 6]. Load sequences for fatigue calculation consist of 27 000 different landings. In the case of
Fig. 1 — Fatigue test of VUT 100 nose landing gear at the laboratory of Technometra Radotin
fatigue tests, the load history has to be stored on a hard disc of the testing system control electronics with a the sampling frequency of
1,000 Hz which is given by the period of computation in the regulator. It leads to large data files and subsequent problems during the
starting of fatigue test. This is why the reduction of number of landings in the sequence to one tenth had to be done. Since the load
sequence with reduced number of landings can’t include the highest
load factors, two modified versions of load sequence denominated S1
and S2 were defined [7, 8]. The load factors in the first one were truncated on the levels pertaining to 2700 landings for both types of
ground load spectra applied. During the fatigue tests, nine repetitions
of S1 sequences are always followed by one sequence S2. The major
part of the content of the S2 sequence is identical with the S1, but it
also involves the highest load factors belonging to 27 000 landings. It
means that the load spectra for fatigue calculations and fatigue tests
are identical but the sequences of loads are different. The highest load
factors are in the course of the fatigue tests located into the considerably smaller segment.
With respect to the application of load sequences for fatigue tests,
the small amplitude taxi loads were omitted and related fatigue damage was realized on one higher load level using essentially smaller
number of cycles. It allows acceptable time of fatigue test but also
reduces computing time in the case of detailed FEM model of the landing gear.
Fatigue analysis
Experimental evaluation was preceded by fatigue analyses started
during the early development phase of the landing gear [1-3]. The aim
of these calculations was to determine the acceptable stress levels
ensuring the requested undercarriage safe life. Fatigue calculations of
the landing gear are carried out by MSC. Fatigue software. They are
based on the results of FEM stress analysis done by MSC.Nastran.
Starting evaluation utilized the FEM models prepared originally for
the static analysis [9]. As the first step in the simulation of real loading
of undercarriage, the maximum principal stresses related to unit loads
defined in three directions were calculated using linear static analysis.
MSC.Fatigue uses the principle of linear superposition to combine
35
Fig. 2 — Ground load spectra
these load cases together and determine the stress variation in each
node of the FEM model. This can by expressed by formula:
σ
σ ij (t ) = ∑ Fk (t )  ij ,k
k
 Fk



3/2006
Fig. 3 — A few landings from the sequence S2 applied during
the fatigue test of nose landing gear
(1)
where the elastic stresses σij from each unit load case k are normalized by the load magnitude from the FE analysis Fk and then multiplied by the time variation of the loading Fk(t).Fk(t) load sequences pertaining to x, y and z axis are generated by the FLTSIM software.
Obtained stress-time histories are subsequently treated using the rain
flow counting procedure and extracted cycles are written into the rain
flow matrix.
Fatigue damage is calculated from the rain flow matrix occurrences against the number of cycles from the material S-N curves related
to the stress concentration factor α of 1 (see Figure 4). The local elastic FE stresses at the notches are directly applied for determining the
number of cycles from fatigue curves.
Verification of fatigue analysis
At the present time, fatigue tests of both landing gears are alternately
carrying out by Technometra Radotin. The nose landing gear has already accumulated a number of landings approving 43 percent of calculated safe life (considering the scatter factor of 6.98) without any
signs of fatigue cracks. This test was temporarily stopped and replaced in the test room by the main landing gear. During this test, the fatigue crack has occurred in early phase represented 12 percent of calculated safe life. The crack originated in the lubricator tapped hole
located in the highly loaded lug for the pivot joining the beam and the
fork of main landing gear (see Figure 6). The subsequent investigation has shown that the probable reason of unconservative calculation
was incorrect stress input into the fatigue calculation. It was due to the
application of the common presumption of only using of linear static
stress analysis in conjunction with fatigue calculations in MSC.Fatigue. This default approach is certainly true as far as the material nonlinearity is concerned, since if we want to obtain sufficient value of the
safe life, we have to ensure such dimensions of the part that lead to
the operation in the linear region of the stress-strain material relationship. Then the main principle of obtaining the stress spectra can be
multiplication of stress quantities calculated for single load case by
the load factors from the load spectra. However, in the case of cracked spot of main landing gear, the geometrical nonlinear solution had
to be used because of the requirement of realistic involving the contact between the pivot and the lug. Detailed FEM models of this area
(in Figure 6) and two others (see Figure 5) have been prepared for
nonlinear contact analyses. Results of the calculation for cracked lug
showed the stresses 85 percent higher then in the case of the linear
solution.
Fig. 4 — Material S-N curves [10, 11]
The methodology of fatigue calculation described in the previous paragraph was adapted for nonlinear stress input because the
linear superposition of stress values obtained from three individual solutions pertaining to particular one axis unit loadings according to formula (1) is in this case incorrect. Stress histories in the
nodes of FEM models were built as sequences of stresses calculated for complete load cases. This can be expressed by equation:
σ ij (t ) = ∑ nk (t ) σ ij ,k
(2)
k
where the stresses σij for each one g load case k were multiplied
by the time variation of the load factor nk(t). Load factor sequences nk(t) pertaining to all load cases applied were again generated
by the FLTSIM software in such a way that for a particular time
moment only one from the functions nk(t) has the value other than
zero.
Fig. 5 — Detailed FEM
model of main landing
gear fork for geometrical nonlinear analysis
36
Upon the modified methodology of fatigue analysis, a reinforcement of the lug and changes in the position of lubricators were
designed in order to obtain higher life value.
Conclusion
The fatigue evaluation of main and nose landing gear of the
VUT 100 airplane has entered into the experimental phase. Tens
of thousands of landings have already been carried out in the
laboratory of Technometra Radotin using the service simulation
load sequences prepared at the Institute of Aerospace Engineering. Results of these tests are useful for verification of FEM
based fatigue analyses and led to the redesign of one fatigue critical area of main landing gear.
(Right) Fig. 6 — Log of damage contour plot of failed area of main landing gear beam after redesign
References:
[6]
Vychopen, J., Belohradsky, T.: Load Cases for Safe Life Calculation
and Fatigue Tests of VUT 100 Airplane Landing Gear; Report
EVUT019.05-ST, Evektor Ltd (in Czech).
[7]
Augustin, P.: Load Sequence for Fatigue Test of VUT 100 Aircraft
Main Landing Gear, Report LU18/2006, IAE BUT (in Czech).
[8]
Augustin, P.: Load Sequence for Fatigue Test of VUT 100 Aircraft
Nose Landing Gear; Report LU27/2006, IAE BUT (in Czech).
[9]
Plhal, M.: Static Analysis of VUT 100 Aircraft Main and Nose-wheel
Landing Gear; Letecky zpravodaj, No. 3/2003, pp. 9-11.
Augustin, P.: Flight by Flight Fatigue Test with Random Application
of Loads; Letecky zpravodaj, No. 3/2003, pp. 3-6.
[10]
MIL-HDBK-5H Military Handbook, Metallic Materials and Elements for Aerospace Vehicle Structures.
Belohradsky, T.: Data for Safe Life Calculation and Fatigue Tests of
VUT 100 Airplane Landing Gear; Report EVUT991.10-ST, Evektor
Ltd (in Czech).
[11]
Becvarik, P.: Fatigue Properties of Czechoslovak Steels; SNTL Prague, 1985 (in Czech).
[1]
Augustin, P., Plhal, M., Splichal, J.: FEM Based Fatigue Life Analysis of Landing Gear; Letecky zpravodaj, No. 3/2005, pp. 6-8.
[2]
Augustin, P., Plhal, M., Splichal, J.: Safe Life Analysis of VUT100
Aircraft Main Landing Gear; Report LU64/2005, IAE BUT (in
Czech).
[3]
Augustin, P., Plhal, M., Splichal, J.: Safe Life Analysis of VUT100
Aircraft Nose Landing Gear; Report LUVY7/2006, IAE BUT (in
Czech).
[4]
[5]
Ing. Milan Merkl, CSc / VZLÚ, Plc., Prague
Methodological procedures of electronic circuit reliability analysis are presented on an example of Earth’s satellite
scientific equipment MAC-03 analyses. The analyses comprise in the first place FMEA/FMECA, next reliability
prediction, de-rating analysis and technical risk analysis. The worst case analysis was not performed in the progress
phase described in the published VZLU report. The difficulty of the analysis and means for its implementation were
only discussed. HSIA was not performed either since the requirement for it was not stated and groundwork for it was
not at our disposal. The methodology is worked out from the point of compliance with ECSS standards requirements.
These standards requirements are obligatory for problems of this discipline and the assessed product implementation
area. The referenced standards are part of the methodology in the scope presented. The analytical results validate
compliancy with requirements, e.g. with mean time to failure of MAC-03 microaccelerometer. The criticality matrix
graph indicates however an unfavourable relation of failure severities and frequency of their occurrence. It is much
desirable to direct design changes and component selection towards further improvements.
Na příkladu prováděných analýz mikroakcelerometru MAC-03 jsou uvedeny metodické postupy rozboru bezporuchovosti
elektronických obvodů vědeckého přístroje umělé družice Země. K prezentovaným analýzám patří především analýza
FMEA/FMECA, dále predikce bezporuchovosti, analýza odlehčení a analýza technických rizik. Ve fázi řešení, jejíž
výsledky budou publikovány jako zpráva VZLÚ, nebyla provedena analýza WCA, ale jen diskutována náročnost takovéto
analýzy a prostředky nutné k jejímu provedení. Nebyla též provedena analýza HSIA, která nebyla v této fázi požadována
a pro níž nebyly dodány podklady. Metodika je zpracována z pohledu dodržení požadavků norem ECSS, které jsou pro
problematiku tohoto oboru v předpokládané oblasti využití závazné. Citované normy jsou pak v uvedeném rozsahu
součástí této metodiky. Závěry provedených analýz potvrzují v tomto případě vyhovění stanovenému požadavku na
střední dobu do poruchy mikroakcelerometru MAC-03. Diagramy znázornění matice kritičnosti však ukazují nepříznivý
vztah závažnosti poruch a frekvence jejich výskytu. Na zlepšení tohoto stavu je nutno zaměřit v rámci možností jak
příslušné konstrukční úpravy tak zejména výběr použitých součástek.
Keywords: microaccelerometer, reliability analysis, criticality matrix, failure severity, frequency of
ocurrence.
37
Payload dependability assurance
Dependability assurance of scientific payload is part of product assurance requirements system. The system is framed by relevant ECSS
standards [1] to [5] and requirement documents. Presented article is
an attempt to describe a methodology used for confirmation of scientific equipment design success in the area of compliance with reliability requirements. The main stay of the methodology assessment is the
FMECA forming a system with another support analysis. The other
system parts are part stress analysis, derating analysis and worst case
analysis. The reliability analyses take their structural data both from
functional tree of the system and component failure database.
The procedure is demonstrated on an example of MAC-03 microaccelerometer electronic circuits analysis. Block diagram of
MAC-03 is known on various detail levels. One of them is presented in Figure 1.
3/2006
The next step is a method to assess unfortunate top events in function of MAC-03. FTA methods and software package best suit for the
description of reasons of their occurrence. At first we list these top
events without ranking of their criticality:
●
Loss of scientific data on MAC output
●
Sensing cube unlocking failure
●
Loss of cube stabilisation in working position
●
Non relevant reading of data by electronic circuits
●
Bad function of CPU in command/data transfer
Fault tree for one of top events from the above list is presented in
Figure 3.
We used reliability prediction method for initial analysis of
MAC-03 and the procedure and results of the analysis were described
in [6]. The procedure is in RELEX, the software package used for the
analysis, closely connected with other methods described below in
more detail, e.g. derating analysis or FMECA. It means, that saved
data and partial results from reliability prediction analysis could be
used for following analysis, in our case for FMECA. The saved data
shall be nevertheless completed with other requested inputs needed
for the FMECA analysis.
Acronyms and abbreviations
Figure 1 — Block diagram of MAC-03 function
The first step to the equipment analysis is to perform a functional analysis. We could proceed according to ECSS-E-10-05A [5] standard
and receive the structure in Figure 2.
Figure 2 — Functional analysis of MAC-03
FMECA Failure Modes, Effects and Criticality Analysis
FMEA Failure Modes and Effects Analysis
FTA
Fault Tree Analysis
PSA
Part Stress Analysis
WCA Worst Case Analysis
ECSS European Co-operation for Space Standardisation
MFR
Mode Failure Rate
FR
Failure Rate
FMR
Failure Mode Ratio
OT
Operating Time
MC
Mode Criticality
IC
Item Criticality
AC
Assembly Criticality
SC
System Criticality
SL
Severity Level
SYS
System
ASSY Assembly
FEP
Failure Effect Probability
Failure modes, effects and criticality analysis
FMECA is an analysis serving to categorisation of equipment parts
according to criticality of failure modes, which are associated with
system component failures. Such categorisation serves to aim our
effort to both design changes and parts selection useful for improving
equipment reliability.
We stated in first clause that FMECA follows-up an earlier reliability prediction. The tree structure of the system is developed to particular components. Failure rates of the components are predicted
using component libraries with data taken from MIL-HDBK-217F,
Notice 2 [7] for failure rates and FMD-97 [8] for failure mode ratios.
Functional FMECA starts with functional analysis of MAC. Reliability engineer receives a detailed description of individual block
functions from designer of MAC electronic circuits. The description
of the blocks has to be completed with hints to criticality of the blocks
for MAC function. The detail schemas of electronic circuit boards are
provided with block schemes. There are provided manufacturers catalogue data for important components. The procedure is documented.
FMEA procedure documents by lexical description systematically
in predefined and standardized tables the following attributes of components and their failures:
38
Figure 3 — Fault tree structure for ”Sensor
out of margins“ top event
●
●
●
●
●
●
●
●
●
function
failure mode
failure cause
mission phase
failure effects
▲ local effect
▲ end effect
severity
failure detection method / observable symptoms
compensation provisions
correction actions
FMECA additionally presents three numerical data:
●
●
●
severity number (SN)
probability / probability number (PN)
criticality number (CN)
Simplification of severity categories tables in [2] for system and subsystem level without safety consequences provides Table 1 classification.
Table 1 — Severity categories and severity numbers for system/subsystem without safety consequences
We need one another table for following calculation, mapping probability levels within specified limits to probability numbers — Table
2, which is valid for system level. We will use a rough assumption of
system breakdown to subsystem and parts with coefficients 10 and
100 and denote the probability numbers PN10 and PN100 respectively. The numbers will be used later in decision diagram serving for
criticality numbers calculation using auxiliary Excel macro program
— see Figure 4.
Table 2 — Probability levels, limits and numbers for system
Symbols from second clause can help in reading of following
equations for mode failure rate and mode criticality respectively:
MFR = FR x FMR
MC = OT x MFR
RELEX sw package determines criticality of item, assembly and system respectively in failure rate units following the equations:
IC =
∑ MC
SL
AC =
∑ IC
ASSY
SC =
∑ AC
SYS
We need transform the received numbers using the flow diagram of
Figure 4 for tailoring them to ECSS-Q-30-02A requirements. We
allocate probability numbers and criticality numbers relevant to the
received numbers above according to equation:
CN = SN x PN
where SN and PN are severity number and probability number of
appropriate level of product breakdown structure respectively.
The resulting criticality number in FMECA is a unique attribute of
technology risk for particular component and its failure modes.
Results of FMECA
The result of FMECA is a critical parts list sorted according to criticality of the parts failure modes (in addition to data, which are a part
of a FMEA table too).
Table 3 presents a sample of such data received in RELEX and
transformed using Excel macro program. In the table is used another
subdivision of criticality attribute. In the column Subcritical is 1 for
parts ranked as non-critical but whose CN1 or CN2 does not equal 0.
It corresponds to severity category catastrophic or critical.
A global view of component failure modes criticality in the whole
system is presented by a criticality matrix. Its 3D view received in
39
3/2006
Conclusions and suggestions
Figure 4 — Flow diagram of Excel macro for assigning
of criticality numbers
MAC-03 FMECA is shown in Figure 5. The matrix discussion reveals
very inconvenient distribution of failure modes with respect to severity and probability of their occurrence. The most severe failure modes
are placed in the most probable region of the matrix. The consequence of it has to be a stimulus for design improvements and particular
design changes.
Figure 5 - Criticality matrix view for MAC-03 analysis results
The ECSS-Q-30B standard lists reliability analyses starting with
FMEA/FMECA through FTA to the part derating analysis and
WCA. Presented methodology with emphasis on FMECA complies with ECSS standards. It is relatively simple and could be
performed with standard reliability software packages. The methodology could be performed by a reliability engineer with basic
equipment structure and performance knowledge and general
skills in the region of reliability analyses.
It does not comprise WCA because of its link to CAE software methods and packages. WCA requires, contrary to the above
stated, more comprehensive knowledge of the system, its functional blocks and work of components.
The results of the reliability analysis procedure should be used
in iterative process in close co-operation with engineering and
managerial processes serving for product reliability improvement.
Implementation of the analyses from the very beginning of product life cycle is very important for successful mission of the equipment.
Results of the work, which was funded by CLKV E1 project,
were published in detail in a VZLU report [9].
References:
[1]
ECSS-Q-30B Space product assurance — Dependability
[2]
ECSS-Q-30-02A Space product assurance — Failure
modes, effects and criticality analysis (FMECA)
[3]
ECSS-Q-60-01A Space product assurance — European
preferred parts list (EPPL) and its management
[4]
ECSS-Q-30-11A Space product assurance — Derating —
EEE components
[5]
ECSS-E-10-05A Space engineering — Functional analysis
[6]
Nová, M., Merkl, M.: Scientific satellite’s payload reliability analysis; Czech Aerospace Proceedings, No. 1/2006,
pp. 25-27
[7]
MIL-HDBK-217F, Notice 2
[8]
FMD-97 Failure Mode/Mechanisms Distribution; Reliability Analysis Center
[9]
Merkl, M., Nová, M.: Metodika rozboru bezporuchovosti
elektronických obvodů vědeckého přístroje umělé družice
Země na příkladu mikroakcelerometru MAC-03; VZLU
Report R-3913, VZLÚ, Plc. Prague, July 2006, 51 pp.+84
pp. of annexes (in Czech)
(Merkl, M., Nová, M.: Methodology of electronic circuits
reliability analysis of a scientific payload of an Earth's
satellite documented through an example of MAC-03 microaccelerometer analysis)
Table 3 — Selection of items
ranked as critical or subcritical
40
Optimization Methods for 2D Flow
Passage Design of Axial Compressor
Optimalizační metody pro 2D proudění axiálním
kompresorem
Ing. Jan Tůma / VZLÚ, Plc., Prague
Genetic Algorithm (GA) has been applied to the aerodynamic shape optimization of airfoil cascades. The goal of this
work has been to minimize the total pressure loss, with maintaining the same flow turning and mass flow rate. The
optimization was solved with the help of CFD software.
Byla provedena tvarová optimalizace profilové mříže pomocí genetického algoritmu. Cílem práce bylo minimalizovat
pokles celkového tlaku při průtoku mříží se současným zachováním stejného ohnutí proudu a hmotnostního toku.
Optimalizace byla provedena s využitím CFD programu.
Keywords: Optimization, CFD, genetic algorithm, aerodynamics.
Introduction
Optimization process
Methods for optimization concerning a flowing part of the turbomachine are useful for industrial companies in our country. Financial
resources in our companies are in most cases limited. In fact the firms
usually do not do any research of importance. They draw on their
older designs, where major effectiveness increase could be achieved
by optimization of the blade shapes. We will not deal especially with
pure centrifugal impellers (fig.1). Optimization and numerical methods can play a very important role with centrifugal machines, which
have properly measured and examined flow conditions. Recent developments in computers and above mentioned methods help to solve
3D geometry optimization, too. The method was checked on 2D level
of an axial compressor.
Today, genetic algorithm (GA) is widely used for solving the optimization problems. The GA was first published by Holland in 1979. GA
was inspired by evolutional natural principles of selection and survival of living organisms. The main advantage of GA is its ability to find
the global function optimum without the risk of stopping at local optimum. This principle is based on selection, mutation and crossing of
individuals.
Optimization processes are based on optimization algorithm and
solver. This process generates a certain number of individuals and an
individual represents one blade design, given by set of design variables (stagger angle and angle attack in first step).
The initial population of modified individuals of defined count is
coincidentally created from the original individual. Out of this population GA creates the basis of another population. Every individual is
rated by the fitness function. Due to this, the more successful individuals are more likely to proceed into the following population. But
this doesn't mean that the less successful individuals lack the chance
of survival. Chromosome representation is necessary to describe each
individual (e.g. bypass-junction) in the population. Each individual is
defined by chromosome which consists of a sequence of genes. In
presented GA the genes are real numbers. Each gene represents value
of one variable. In our case the variables (junction angle, graft diameter) is geometric parameter of bypass junction. The selection of
individuals from current population to produced next population is
very important feature. There are several methods how to do it.
All are based on probabilistic selection which is done according to
individual’s fitness. At the presented GA the method of tournament
was used. The selection method works by random selection of several individuals from current population and inserts the best one into
the new population. The procedure is repeated until the required number of individuals has been selected. The genetic operators created
new individuals based on selected predecessors. Two basics types of
operators exist. Crossover creates from two existing individuals a new
individual while mutation modifies one existing individual to produce
new one. There are a lot of different methods how crossover and
mutation may be done. The presented GA uses arithmetic and heuristic crossover and uniform mutation. For example arithmetic crossover
produces two linear combinations of the parents.
Each individual whit different stagger angle had different fitness
function, which was defined by total pressure loss coefficient.
Generation by generation, the fitness function of new stagger angle
setup is evaluated by the results of the numeric simulations.
Global Performance of the Compressor
The basic components of the performance of the compressor are:
The mass flow rate
The rotational speed
The total pressure ratio
Optimization method
Profile vertices from profile database NACA 65-410
Automatic creation of computational grid
Automatic set up of solver
Automatic generate of output value
Fitness function process — Genetic Algorithm (GA)
Stagger and attack angle change by automatic process
Fig. 1 — Centrifugal
impeller
41
3/2006
Geometrical parameterization
In the rotor shape parameterization, we have used only stagger angle
in this first step, but our optimization method was prepared for optimization of chord length, blade spacing and control points of NURBS
curves for pressure and suction side, too.
Geometry of the profile NACA 65-410 was rotated around its trailing edge. The NACA profile stagger angle was changed from 30° to
60°.
Range of stagger angle (°) . . . . . . . . . . . . . . . .30-60
Range of angle attack (°) . . . . . . . . . . . . . . . . .1-10
Middle Radius Chord (mm) . . . . . . . . . . . . . . .82
Thickness (mm) . . . . . . . . . . . . . . . . . . . . . . . .5.2
Number of rotor blades . . . . . . . . . . . . . . . . . .17
Number of rotor blades . . . . . . . . . . . . . . . . . .24
Computational Grid
Fig. 4 — Contours of velocity magnitude (stagger angle 45°;
angle of attack 48°)
The quality of mash plays a very important role for accuracy results
of solution. An example of the orthogonal structured grid used is
given in Figs. 2 and 3. It has used two type of grid. The Near-Wall
Model Approach was used for describing right velocity profile in
boundary layer including viscous sublayer. The maximum dimensionless distance from the wall was y+<= 3,5. Therefore the wall function
was necessary used in some region. The maximum value of this parameter was on shape where the pressure side passes to trailing edge.
The fineness mash doesn't affect the flow because of separation flow
at the blade passage. Average value in the computational region was
about y+<= 2,5 on blades. Wall function was used in next type of grid
where was y+>= 55.
Fig. 5 — Contours Static pressure (stagger angle 45°;
angle of attack 48° )
Compressor stage performance needed to compute the fitness
function is evaluated a Navier-Stokes analysis of the flow field. This
step is the most time consuming in the overall optimization process;
moreover, the accuracy of results is greatly affected by the grid size.
The loss coefficient is defined as the total pressure drop over the
outlet dynamic head.
ζ =
pc1 − pc 2
pc1 − pS 1
Drag and lift Coefficients was monitoring during process.
s  cos 3 α m 

= ζ p1 
1
c  cos 2 α1 
ρU m2 c
2
L
s
cL =
= 2 (tan α1 − tan α 2 )cos α m
1
c
2
ρU m c
2
The static pressure distributions near the suction surface are shown in
Figures 5 and 6.
cD =
Fig. 2 — Detail of Grid at area close to the leading edge
D
Fig. 3 — Example of computational Grid
Results
Numerical solution has been sensitive to different stagger angle, angle
attack and in the same time to output boundary condition. Useful
solution wasn't there for some profile set up, therefore outlet boundary condition have to be chosen very carefully. This condition hasn't
been satisfied in this first step of the presented paper.
Pressure outlet target for mass flow was set up like outlet boundary condition and initialization of numeric solver passed very slowly
from low to final back pressure.
Fig. 6 — Static pressure distribution for pressure and suction side
(stagger angle 45°; angle of attack 48°)
42
[1]
Emery, J. - Herrig, J. - Erwin, J. - Felix, R.: NACA Report
1368; Systematic two-dimensional cascade tests of NACA
65-series compressor blades at low speeds; 1957
[2]
Citavý, J. - Nožička, J.: Lopatkové mříže; 1957 (ČVUT,
Praha)
[3]
Whitley, D.: A Genetic Algorithm Tutorial; 1994(CSU, Fort
Collins)
0,075
0,1
Fig. 7 — Loss polars for
mid-span section
0,05
References:
0,025
In this paper, we have described optimization of a compressor stage
using a Navier-Stokes solver in order to minimize the total pressure
loss for different stagger angle. Method was prepared for optimization of profile shape but in this paper is presented only transformation
of stagger and angle of attack.
0,125
ζ
Conclusion
-60
-50
-40
∆α1
-30
Výroba kompositového krytu vrtulové hlavy technologií LF
Ing. L. Křípal / Institute of Aerospace Engineering, Brno University of
Technology
The present paper discusses an LF-Technology (Letoxit Foil) used for manufacturing 3D aircraft propeller nose cone
part and properties of unidirectional (UD) glass reinforcements. LF-Technology was introduced by 5M Ltd. of the
Czech Republic. LF-Technology is a modified RFI technology (Resin Film Infusion). The simple and fast process is
based on vacuum and/or press-assisted resin film impregnation of a dry reinforcement and core material in the
mould, with high-temperature curing (autoclave is not necessary). The process allows placing a required type and
a number of layers to different locations to achieve the exact resin content required (tailored properties) as a function
of the number of resin layers used. Only one type of resin film is necessary, which can be combined with almost any
type of reinforcement. The influences of a number of processing parameters on the quality of cured parts are
discussed along with previous flat sample testing results. LF Technology gives freedom to designers, helps push down
prices of composite products and increases properties and reliability, which is supported by favourable responses from
customers and quick and relatively easy development of new products. All these advantages promise a very good
future for LF Technology.
Tento článek pojednává o LF Technologii (Letoxit Foil) aplikovanou na výrobu krycí části propeleru (kužel), dále pak
o ohybových vlastnostech jednosměrných výztuží. LF-Technologie byla vyvinuta firmou 5M s.r.o. Česká Republika. LFTechnologie je modifikovaná RFI technologie (Resin Film Infusion). Jednoduchý a rychlý proces je založen na aplikaci
vakua (volitelně i vnějšího tlaku) na kladenou suchou výztuž kombinovanou s epoxidovou pryskyřicí ve formě folie
(použitelnost technologie i pro sendviče). Takto uspořádaný systém je poté vytvrzen zvoleným teplotním cyklem bez
nutnosti použít autokláv. Proces umožňuje zvolit požadovaný typ a počet vrstev a dosáhnout tak přesného množství
pryskyřice ("šití materiálu na míru"). Je možno použít pouze jeden typ pryskyřice, který se dá kombinovat s téměř
každým typem výztuží. Vlivy procesních parametrů na kvalitu vytvrzených kompozitů jsou zde diskutovány spolu
s předešlými výsledky testování rovinných jednosměrných vzorků. LF Technologie dává svobodu designérům, snižuje
výrobní náklady a zvyšuje vlastnosti a spolehlivost takto vyrobených kompozitů, což je ostatně podloženo ohlasy
zákazníků a díky rychlému a relativně jednoduchému způsobu jak vyrobit nové produkty. Všechny tyto výhody slibují
dobrou budoucnost pro LF technologii.
Keywords: LF-Technology, mechanical testing, RFI technology.
Introduction
The benefits of composite materials are now well known and a great
variety of composite applications ranging from ”industrial“ and
”sports and leisure“ to high performance aerospace components indicates that composite materials have a promising future [1].
Vacuum infusion technology called LF-Technology was applied to
manufacture of UD samples and a spinner cone part. One of RFI vari-
ety is LF-Technology (Letoxit Foil) provided by local Czech producer 5M Ltd. RFI (Resin Film Infusion) has been identified as an alternative cost-effective manufacturing technology to RTM and the conventional autoclave prepreg technique [2].
Resin Film Infusion (RFI) uses dry textile preforms with resin film,
consolidated and cured in a single step vacuum process. The large
number of material properties and processing parameters must be specified and controlled during resin infiltration.
43
Two different issues are presented herein. The first part of this article discusses design of a spinner step by step to the final shape
(LF-Technology verification — complexity of shape in composite part
processing). The second part describes result of wet out problem with
unidirectional glass fibre fabrics (solution to easier ”wet out problem“
for UD reinforcements).
LF-Technology
Just a few further details about our RFI (LF-Technology) setup are
presented in this paper since RFI process was discussed in Czech
Aerospace Proceedings No. 2/2006.
The production cycle is performed usually in the autoclave under
increased temperature and pressure (from 60kPa), where the fibre
infiltration and the composite consolidation occur in a single step
process. Typically, the RFI element consists of a thermoset resin
film (solid resin also signed as β-stage resin) placed between one
side of a metal tool and a dry textile fibre preform can be seen on
(see Fig. 1).
Institute of Aerospace Engineering’s own facility for composites
production by LF-Technology is shown in Figs. 2 and 3. It consists
of vacuum pressure source; ejector vacuum pump and intelligent
programmable hat air dryer (see Figs. 2 and 3).
LFX035 is a high performance 130°C cure matrix used for UD
glass reinforcements. It is generally known that UD textiles is hard
to wet-out but it was easy to get clear material after selecting ”the
right“ curing cycle thanks to low viscosity resin properties.
This resin can be used in a variety of production techniques
(Resin Film Infusion [RFI], prepreg and semipreg) which is making
it easy to process. Its low density contributes to increased weight
savings, and the flexibility in curing gives potential for composite
part cost savings.
3/2006
Part One — Propeller spinner and spinner
backplate manufacturing
Mould construction and processing
There is a need for mould for higher temperatures and it is under
development at the moment. Smaller metal tool was chosen for the
first pre-testing. The mould will consist of two parts. The first one is
female (cone shape part) and the second one is male (plug part). This
set up can be seen in the figures below. Mould for spinner backplate
will be also made.
Materials used
Epoxy resin film LFX023 together with plain weaved glass (carbon)
reinforcement will be used for fabrication of nose cone and backplate. Epoxy gel coat will be applied to the surface finish.
Figs. 4, 5 - Male (plug) and female (conic shape metal tool) parts
of mould for preliminary verification test
Fig. 1 — LF-Technology set up
Figs. 6, 7 — Preliminary concept of
SPINNER&BACKPLATE
Figs. 2, 3 — RFI Technology facility
Experimental
The whole composite spinner production cycle is described here,
together with flexural properties of unidirectional glass composites.
Processing parameters
For spinner and spinner backplate, a curing cycle with slow heat up
rate and dwell will be used. This curing cycle is shown in Fig. 9, signed as J. Curing temperature 120°C is recommended with 60 minutes
of curing time. Vacuum pressure about 80kPa will be applied.
44
Part Two — Unidirectional glass composites
Materialsu Used
Epoxy resin film LFX023 and LFX035 with glass unidirectional fabric (Interglas 92145 220 g/m2) were used to produce quality flat testing specimens.
Samples preparation
Dry reinforcement with solid resin films were laid into the mould
which was then closed by vacuum bag, evacuated (app. 80kPa) and
cured at 130°C for 1 hour.
Results
At the beginning of unidirectional glass fabrics testing, preliminary
”wet out“ samples were made. It was hard to wet out glass fabric well
with the aid of LFX023 epoxy resin. That is why another epoxy resin
(LFX035) was examined.
To get better clear effect, stripes of LFX023 epoxy resin were laid
on UD glass fabric. The system contained less resin and low void content thanks to the free space among the stripes [3].
This formation is similar to ZPREG technology where surface ply
consists of a lightweight (usually glass) fabric laminated to a dry,
medium weight carbon or glass fabric using stripes of resin.
The resin impregnates the lightweight fabric sufficiently to provide
light tack, whilst the heavier inner ply retains a dry surface. This format ensures that air is channelled away from the tool face before the
resin stripes close [4].
Comparison to hand lay-up technique is shown in Fig. 11.
Fig. 8 — Typical lay out
Processing parameters
The most suitable curing cycles for epoxy resin LFX035 are cycles
signed as J and X1 (shown in Figs. 9 and 10). They have different
ramp parameters. The most convenient curing cycle can be chosen
from these two cycles.
Fig. 11 — Maximum flexural stress — comparison of RFI technology
and R&G data
Application of external pressure during curing process can help to
obtain higher mechanical properties.
Figs. 9, 10 — Curing cycles
Fig. 12 — Maximum flexural stress LFX035
Fig. 13 — Flexural modulus LFX035
45
3/2006
These graphs indicate the highest mechanical properties in curing cycles signed J and X1. Different curing cycles were chosen to get optimal results. Figs. 14 and 15 are giving information about flexural properties of UD reinforcements in use with standard epoxy resin
LFX023.
Conclusions & Summary
Propeller nose cone can be fabricated from glass (cheap), carbon
(tough), basalt (cheap, impact resistance and heat resistant) or hybrid
(basalt-Kevlar, carbon-Kevlar [impact resistance]) fabrics used for
specific purpose, depending on customer needs. LFX (Letoxit Foil)
foils were selected for matrix of the composite material.
Testing specimens of UD fabrics were prepared with LFX023 and
LFX035 epoxy resin films. From mechanical results we can assume
that resin LFX035 has more stable and reproducible results. Therefore, this resin is recommended for use with UD fibre reinforcements in
thermoset polymer matrix composites.
Vacuum level from 70kPa is recommended because a unidirectional fabric is hard to wet-out. Lower heat-up rates and insertion of
dwell can be applied if needed. Resin has to be defrosted at 25°C for
5 or 6 hrs before use to remove residual humidity and it was found out
that 12 hrs is better. Single layers of the resin layers could be pre-melted by heat gun for better handling. Beware of resin overheating
which can lead to material properties reduction! Vacuum level and
pre-melting are not the most important factors in RFI UD fabrics processing.
Heat-up rate (”ramp“) is an important factor. The number of filaments in fibre bundle is another influential parameter of the process
and it is good to count quantity of resin needed for successful wet out
process in advance.
It is easygoing to wet out classic fabrics. Better wet out effect was
achieved after curing cycle correction of unidirectional fabrics.
Future work
Spinner and spinner backplate will be fabricated by RFI technology
during fall 2006. Application of external pressure during curing process is planned to obtain higher mechanical properties in the future.
The relationship between compaction pressure and fibre volume fraction is very important in the RFI process since fibre volume fraction
has a large influence on the mechanical properties of the composite
part.
Fig. 14 — Maximum flexural stress LFX023
Fig. 15 — Flexural modulus LFX023
References:
[1]
[2].
[3]
[4]
HEXCEL, Prepreg Technology, 1997 Publication No. FGU 017,
(Hexcel composites, prepreg handbook), accessible from: [<
www.formulaschools.com/curriculum/prepregtechnology.pdf >],
[Quoted 10/2006]
Křípal L., Mihalides D., Daniel M., Pavlica R.: Influence of matrix
systems on mechanical properties of composite test specimens made
by RFI technology; Reinforced Plastics 2005 (XXIII International
Conference), 24 to 26 May 2005, Karlovy Vary, Czech Republic —
conference proceedings.
Křípal L.: Mechanical testing of composite specimens made by RFI
Technology; 7th International seminar on RRDPAE, 11 to 12 October
2006, Tallinn, Estonia — article and oral presentation.
Available from WWW: [<http://www.advanced-composites.com/>]
— ZPREG [online], [Quoted 10/2006]
Optimalization of Stiffened Panel with
the Help of Mathematical Programming
Optimalizace vyztuženého potahu pomocí matematického
programování
Ing. Miroslav Pešák, Prof. Ing. Antonín Píštěk, CSc. / Institute of Aerospace
Engineering, Brno University of Technology
In modern aircraft structural design, the high accuracy calculation to obtain the highest efficiency of the structure is
possible through the use of computer analysis. A system of computer programs to optimize and analyze skin-stringer
panel for fuselage and lifting surfaces (wing and empennage structures) is being developed at the Institute of aerospace
engineering, FME, BUT now. This paper deals with the optimum design of the reinforced panel which means Skin —
stringer Panel. This form of construction is a logical development of the necessity of providing a continuous surface for
an airplane, combined with the requirement that the weight of structure should be as small as possible.
Ve VUT Brno se vyvíjí systém počítačových programů pro optimální konstrukci vyztužených potahů trupu a nosných
ploch (křídlo a ocasní plochy). Příspěvek uvádí postup optimálního návrhu vyztuženého panelu potah — výztuha
z hlediska minimální hmotnosti.
46
Introduction
The components of skin-stringer panel may be classified as follows:
Longitudinal reinforcing member:
This is represented by stringers and longerons of fuselage shells and
the spar flanges of wings. They are able to carry appreciable tensile
loads and, when supported, compressive loads as well. They can
carry small secondary bending loads, but their bending rigidity is
negligible. So it is customary to describe them as direct load carrying
members.
Skin:
Like all thin shells, this is best suited to carrying load in its own surface as membrane stresses. Tensile, compressive and shear loads can
be carried, but reinforcement (lateral support) is required for all but
the first. The thin skins used in aircraft can only sustain and transmit
normal pressure over very short distances by bending. Pressurization
loads in a circular section fuselage can, however, be taken by hoop
tension stresses.
Transverse reinforcing members:
These are the rings, frame, bulkheads or diaphragms of fuselages and
the ribs of wings. In the design of these members most attention is
paid to providing stiffness and strength in the plane of the member.
And therefore they are usually incapable of carrying much lateral
load.
Fig. 1 Stiffened panel
The stringers and rings or ribs are attached to the skins by lines of the
rivets, spot welds or perhaps bonding. These joints will be called
upon to transmit forces mainly along their length. Forces parallel to
the skin and directed at right angles to the stringers or rings or ribs
will be limited by the torsion flexibility of these members. Forces normal to the skin will be limited in magnitude by the small bending
strength of the skin and stringers. The primary function of these joints
is thus the transmission, by shear forces, of direct loads in the reinforcing members to the skin and vice versa. Their secondary functions
are indeed essential to the working of the structure but do not give rise
to such large loads.
of optimization theory to concrete engineering problems it is necessary to clearly define the boundaries. The system boundaries are simply
the limits that separate the system from the remainder of the universe.
The performance criterion
Given that we have selected the system of interest and have defined
its boundaries, next we need to select a criterion on the basis of which
the performance or design of the system can be evaluated so that the
best design or set of operating conditions can be identified.
The independent variables
The third key element in formulating a problem for optimization is the
selection of the independent variables that are adequate to characterize the possible candidate designs or operating condition of the system.
There are several factors to be considered in selecting the independent
variables.
First, it is necessary to distinguish between variables whose values
are amenable to change and variables whose values are fixed by
external factors lying outside the boundaries selected for the system
in question. Furthermore, it is important to differentiate between system parameters that can be treated as fixed and those that are subjected to fluctuations influenced by external and uncontrollable factors.
Second, it is important to include in the formulation all the important variables that influence the operation of the system or affect the
design definition.
Finally, another consideration in the selection of variables is the
level of detail to which the system is considered . While it is important to treat all key independent variables, it is equally important not
to obscure the problem by the inclusion of a large number of fine
details of subordinate importance.
The system model:
Once performance criterion and the independent variables have been
selected, the next step in problem formulation is to assemble the
model that describes the manner where the problem variables are
related and the way in which the performance criterion is influenced
by the independent variables. In general, the model will be composed
of the basic material and energy balance equations, engineering
design relations, and physical property equations that describe the
physical phenomena taking place in the system.
Objective functions and functions of constrains:
Design of optimally stiffened panel is mainly focused on selection
design parameters so that entire panel's weight will be minimal and all
boundary condition and requirement will be satisfied.
Objective function is expressed by design parameters.
W(x ) = ρ ⋅ (A ⋅ B ⋅ t + K FV ⋅ nV ⋅ tV ⋅ hV ⋅ A)
Objective function with labeling of design parameters used in
OPTPAN
W()
= ρ ⋅ (A ⋅ B ⋅ X 1 + K FV ⋅ X 2 ⋅ X 3 ⋅ X 4 ⋅ A)
x
Introduction to optimization
In the most general terms, optimization theory is a body of mathematical results and numerical methods for finding and identifying the
best candidate from a collection of alternatives. The process of optimization lies in the root of engineering, since the classical function of
the engineer is to design new, better, more efficient and less expensive systems as well as to devise plans and procedures for the improved
operation of existing systems.
The power of optimization methods to determine the best case without actually testing all possible cases comes through the use of
a modest level of mathematics and at the cost of performing iterative
numerical calculations using clearly defined logical procedures or
algorithms implemented on computing machines.
Define the system boundaries
In order to apply the mathematical results and numerical techniques
where:
W [kg] - panel weigh
ρ
- material density
A,B
- external panel dimensions
t
- sheet thickness
- stringer coefficient
KFV
- number of stringers
nv
- characteristic stringers thickness
tv
- characteristic stringers high
hv
The following functions define the problem of the optimum design of
a stiffened panel:
We can find two basic types of constrains. The first one is in the
form of equality, these constrains are not used in this software. The
second one included inequality constrains, which are listed below and
you can find them in their final form:
47
The skin- stringer construction can develop several separate type of
instability, which may be coupled to a greater or less degree.
1) Local skin buckling between two stringers is given by: g1(X)
Initial buckling (skin buckling) generally involves waving of the
skin between stringers in half-wavelength comparable with the stringer pitch. There will also be a certain amount of waving of the stringer web and lateral displacement of the free flange. For some proportions these may become larger than the skin displacements, and
the mode becomes more torsion or local in nature.
Torsion instability: The stringer rotates as a solid body about a longitudinal axis in the plane of the skin, with associated smaller displacements of the skin normal to its plane and distortion of the stringer
cross section. The half-wavelength is usually in the order of three
times the stringer pitch.
Fig. 4 Torsional instability
()
N y  N xy
N
−
g3 x = 1 − x −
Sx
S y  S xy
where:
[(
)
2
A ⋅ x 3⋅ 1 + α 2 + G
Sx = 1 1 2
α ⋅ (1 + ~t )
Sy =
S xy =
g1 (X ) = 1 −
where
Ny
Nx
−
m
A2 ⋅ k x
A2 ⋅ k y m
 N xy 

⋅
 A2 ⋅ kxy m 


A2 = A1 ⋅ ( X 2 + 1) ⋅ X 1
A1 =
α
where
Ny ⋅µ
X1
Av =
−
X 
Nx
+ Av ⋅  3 
X1
 X4 




2
]
(1 + G )]
(1 + G ) + 3 ⋅ (1 + G )+ 1
2
(
)
x 2 ⋅ K I ⋅ x3 ⋅ x 4 3 ⋅ 1 − µ 2 ⋅ κ
B ⋅ x1 3
~ K F ⋅ x 2 ⋅ x3 ⋅ x4
t =
B ⋅ x1
4) The strength is included in the constrain: g4(X)
2) Local stringer buckling is given by: g2(X)
Local instability: A secondary short wave length buckling may take
place where the stringer web and flange are displaced out of their planes in a half-wavelength comparable with the stringer depth. There
will be smaller associated movements of the skin and lateral displacements of the stringer free flange.
g 2 (X ) =
G=
3
π2 ⋅E
1
⋅ 2
2
12 ⋅ (1 − µ ) B
[
A1 ⋅ x1 3 ⋅2 ⋅ 1 +
α2
A1 ⋅ x1 3 ⋅2 ⋅ 4 (1 + G ) ⋅ 4 ⋅
Fig. 2 Local skin buckling
2
3/2006
2
g 4 (X ) = Rm −
1
2
2
⋅ N x + N y − N x ⋅ N y + 3 ⋅ N xy 2
X1
5) The limitation of the number of stringers is one of the design-technological limitations and it is given by: g5(X)
g5 (X ) = nv max − X 2
where nvmax - max. required number of stringers
6) The limitation of the minimum skin thickness is given by: g6(X)
g 6 (X ) = X 1 − t min
7) Analogous to point six is determined the maximum skin
thickness: g7(X)
g 7 (X ) = t max − X 1
π 2 ⋅ E ⋅ kv
12 ⋅ (1 − µ 2 )
3) Local stringer buckling is given by: g3(X)
Flexural instability : simple strut instability of the skin-stringer construction in a direction normal to the plane of the skin. There may be
small associated twisting of the stringers. The half-wavelength is
generally equal to the rib or frame spacing.
The thickness tmax, tmin are input data.
8) The stringer height is defined as follows: g8(X)
g 8 (X ) = hv max − X 4
9) Regularity of asymptotic values of coefficient is given by: g9(X)
g 9 (X ) = X 3 − X 1
Program OPTAN was created in software Borland Delphi Profession
version 7.0.
Introduction to optimization program OPTPAN:
Fig. 3 Flexural
instability
Main menu:
The main menu of this program is divided in to several parts.
a) Panel constrains
The biggest advantage of this optimizing method (It means non-
48
Fig. 5
Main menu of
OPTPAN
linear programming of Himmelblau) in comparison with analytical
optimizing method of stiffened panel is possibility to use panel's constrains. The basic constrains, which are used in this software, are sheet's thickness (minimal and maximal), maximal allowed number of
stringer. Due to this program we are able to solve concrete situations
and problems.
c) Optimizing parameters
Starting polygon size is value which is used for the first cycle of the
running through the program code. During each cycle this value
decrease until size of polygon is lower than required value of convergency.
Fig. 8
Optimizing
parameters
Fig. 6
Panel’s
constrains
b) Stringer’s parameter
In this dialog window is a user allowed to change some of stringer's parameter. Stringer area coefficient, moment of inertia coefficient,
coefficient of stringer stability and asymmetric influence is constant
for each type of stringer and belongs to so-called ”date information“
of stringer. Values of these example stringer's parameters are determined for Z-profile.
Fig. 7
Stringer’s
parameters
Stringer’s shape
Some typical skin-stringer constructions are shown on Fig. 9. These
constructions are used on existing aircraft structure.
Figs. 9a and 9b are the most popular shapes of stringer because
both of them have got a high structural efficiency and easy assembly.
J-stringer does not have so high structural efficiency as Z-stringer, but
its next positives are better fail-save characteristics due to the double
row of fasteners attached between stringer and skin. There are a lot of
other stringer's shapes used on aircraft. But some of them are not
accepted by commercial operators due to the possibility of corrosion
problem in the closed (not inspect able) area. Examples of this stringers Fig. c, e, g.
Fig. 9 Stringer’s shape
49
3/2006
Fig. 10
Result window
The database of profiles has not been done completely yet. Nowadays the program's user can decide between two basic profiles:
Z and L stringer. Panel loading and dimensions of stiffened panel are
in accordance with the figure of the panel which is shown in the main
menu of the OPTPAN.
Input properties:
Nx [N/mm] - load in the longitudinal (stringer) direction
Ny [N/mm] - load in the lateral direction
Nxy [N/mm] - shear load
Results window:
The main optimization criterion is minimal panel’s weight. For this
minimal values of the weight are determined other important parameters in accordance with required panel's boundary condition.
Main design parameters:
— Panel’s weight
— Sheet’s thickness
— Number of stringer
— Stringer’s dimensions
— Minimal rivet spacing
— Stringer spacing
— Stringer’s cross-section area
— Panel’s cross-section area
Examples of the calculation:
The input data are as follows :
A = 800 [mm]
B = 600 [mm]
Nx = 600 [N/mm] - load in the stringer direction
Ny = 0 [N/mm] - load in the lateral direction
Nxy =75 [N/mm] - shear load
KFV = 2.543 [-] - stringer moment of inertia coeff.
KV = 1.162 [-] - stringer stability coeff.
κ = 4 [-]
- influence of stringer asymmetry
Rm = 380[MPa] - allowed tension stress
Conclusions:
The design and calculation of stiffened panel were made in this paper
with the help of numerical method of mathematical programming.
This numerical method was developed by Himmelblau and Paviani
and used for optimizing program KOFLEX developed by A. Pistek.
The new optimizing program for designing skin-stringer panels with
Fig. 11 Relation between the panel optimum dimensions and number
of stringers
designation OPTPAN was created on the basis of this program. This
program is intended for preliminary design of stiffened panels. For
entered strength (stress limit, stringer buckling stress,…), technological (minimum and maximum thickness of sheet or stringer, number of
stringers,…) and construction (external panel dimension) requirements and loading which acts on the panel, OPTAN is able to find
such a solution that panel's weight is minimal and all other requirements are satisfied.
This program is not done completely yet. It is necessary to make
some slight changes. Next very important step of our effort will be the
verification of the program results. Several panels and also a testing
frame has been already made. This testing will begin soon.
Literature:
[1]
Píštěk, A.: Kandidátská disertační práce; Brno, 1980
[2]
Gallagher, R. H., Zienkiewicz, O. C.: Optimum Structural Design;
John Wiley and Sons, London, 1973
[3]
Reklaitis, G. V., Ravindran, A., Ragsdell, K. M.: Engineering Optimization-Methods and Applications; John Wiley and Sons, USA,
1983
[4]
Michael C. Y. Niu: Airframe Structural Design; second edition,
1999
[5]
Himmelblau, D. M.: Prikladnoje nelinejnoje projektirovanije; Mir,
Moskva, 1975
[6]
Píštěk, A., Hobza, P.: Optimum design of stiffened panel using the
method of mathematical programming; ICAS 2000 Congress
50
Ověřovací frekvenční zkoušky na kluzáku G-304 CZ
Ing. Karel Weigel / Department of Aerospace, Czech Technical University
of Prague
A Ground Vibration Test (GVT) on an HPH G-304 CZ sailplane was conducted in order to verify the functionality of
the TL-5424_CDD modal measuring unit, try on the test procedure and compare the test results with GVT conducted
on the same sailplane by N. Niedbal [1]. A GVT is the first step in further aeroelastic analysis, including flutter analysis.
The necessary steps in a GVT include sensor calibration, equipment setup, data acquisition, and frequency response
analysis. Electro-dynamic shakers were used to excite the structure with manually control sweep sine forces. The input
forces were measured by load cells, and the output motions were measured by accelerometers, which were attached to
the aircraft with an adhesive tape. Measurements were digitized by the measuring unit and recorded on PC. The modal
analysis was performed in Sigview and LabVIEW software to calculate the natural frequencies, and mode shapes from
the frequency response data at each discrete measurement point on the G-304 CZ sailplane.
Byly provedeny pozemní frekvenční zkoušky (GVT) kluzáku HPH G-304 CZ. Účelem této zkoušky bylo ověření
funkčnosti aparatury TL-54-24_CDD, postupů měření a porovnání získaných výsledků s měřením provedeným na
stejném typu letounu N. Nedbalem. GVT jsou prvním krokem v aeroelastické analýze, zahrnující i analýzu flutteru.
Postup provedení GVT zahrnuje kalibraci snímačů, přípravu, odměření dat a analýzu frekvencí. K buzení konstrukce
byly použity elektrodynamické budiče, budicí síla sinusového průběhu byla nastavována ručně. Snímána byla jak budicí síla, tak odezva konstrukce pomocí akcelerometrů. Digitalizovaná data byla ukládána pro následnou analýzu na PC.
Výpočty vlastních frekvencí a vlastních tvarů kmitů proběhly pro každý měřený bod letounu zvlášť pomocí programů
Sigview a LabVIEW.
Keywords: Ground vibration test, GVT, aeroelasticity, flutter, frequency response.
Introduction
Enhancing safety of small sport aircraft with a maximum take-off
weight of 500kg, which are certificated under simplified national aviation regulations, requires easy-to-manage and low-cost manufacture
on the one hand but also reliable and confirmative validation of the
flutter structure resistance on the other hand.
Easy critical flutter speed calculations via simple design criteria are
insufficient to provide dependable and safe aero-elastic certification.
The use of current progressive numerical methods (FEM) provides
great capabilities but these procedures depend on model accuracy and
there are not any experimental data to support the credibility of gained numerical results especially as far as modal structure characteristics Are concerned. Such methods are also very time-consuming.
Therefore aviation authorities in some EU countries accept as a proper flutter analysis regarding the small sport aircraft category a combination of experimental modal characteristics introduced into adequately simplified flutter equations.
The article deals with the TL-5424_CCD device for Ground Vibration Tests of small sport airplanes up to weight 650kg. This device has
been developed in cooperation with Czech company TL Electronic
Inc. Hradec Kralove in the frame of an ARC project. The facility enables one to excite harmonic forces by means of three independent
electrodynamics vibrators equipped with load cells. Control software
permits one to isolate natural modes on a base of the phase resonance and other assistance criteria. Up to twelve acceleration sensors on
the airplane can scan a structure response and evaluate natural shapes.
The system was tested on 17m sailplane G-304 CZ produced by the
HPH Ltd. Company in Kutna Hora.
Description of the test facility, methodology of measurements and
results of natural modes performed with the G-304 CZ sailplane are
presented and compared with Ground Vibration Test realized by
Prof. N. Niedbal [1].
Test Equipment
A variety of equipment is required to conduct a GVT. The amount and
type of equipment and the scope of the test itself depend on the size
of the vehicle or article being tested and desired results. The equip-
ment used at the Department of Aerospace CTU is designed for vibration tests on aircraft up to 650kg max. TOW. That means category of
Small and Light Sport Aircraft. The test equipment is designed according to requirements showed in [3] and respecting [2].
Data Logger
The modular 15-channel logger consists of the microprocessor central board and the cache memory developed by TL electronic comp.,
Hradec Kralove under the sign TL-5412-CCD.
The central unit is designed for 12 inputs from accelerometers and
3 force sensors inputs. The outputs consist of three microprocessor
vibrator control boards. The data
interface between the central
board and the control PC is made
by USB 2.0. The Data Logger
provides actuating of the frequencies of the structure. These frequencies are actuated and monitored for the range of app. 2 150 Hz for whole structures. Please mention that the max. aircraft
takeoff weight is only 650 kg.
The measuring unit is designed
according to requirements showed in [3] and respecting [2].
Excitation
Pic. 1 — The System Block Diagram
Electrodynamics shakers are the method of exciting an aircraft structure.
The system has a total of three shakers. Typically the test requires two shakers. used an ET-126B, (Labwork Inc. U.S.A.) types and together with
power amplifiers PA-141 make a system ensuring to generate variable
amplitude and force in range from 0.5 to 1 kHz and from 5 to 236 N.
Each shaker is attached to the aircraft structure by a telescoping stinger. A load cell is mounted atop the stinger and is attached to vacuum
cups. The stinger allows the transmission of axial loads. The vacuum
cups provide easy-to-mount and non destructive connection on composite structures. A vacuum pump is required to maintain vacuum in the
cup.
51
Sensors
Pic. 2 — Shaker with
Twelve measuring channels
stinger, load cell
with transducers and AD con& vacuum cup
verters for single axis accelerometers ADXL 190EM and typically two load cells HBM U9B
are used. Each sensors input
work with 16bit resolution and
8,000 samples per second sampling frequency. All channels
are sampled simultaneously.
The ADXL 190EM is a single axis low-cost micro mechanics accelerometer. The magnitude limitation of this accelerometer is 100 gs and the frequency range is
from 0 to 300 Hz.
The HBM U9B is tensile/compressive strain gauge force sensor. The
magnitude limitation of this load cell is 500 N and the natural frequency
is 15.5 kHz.
3/2006
frequency response functions. For bending shapes animation 3D
model of an aircraft was made in Unigraphics software.
The Object under Test
The HPH G-304 CZ is a single seat all-composite sailplane with
retractable landing gear; T-tail and wing extenders were used.
Technical data of the HPH 304 CZ sailplane (Tab. 1):
Pic. 5 — HPH G-304 CZ
Test Procedure
Free Support System
The free support system is used to approximate free-free boundary
conditions by reducing the frequency of the six rigid-body support
modes as much as possible. The free system is approximated by hanging-up the aircraft on the adjustable frame via fabric belts and soft
metal springs.
Pic. 3 — Sailplane in the
laboratory
Excitation Techniques
Structural excitation is an important point of ground vibration testing.
The structural responses caused by this excitation are analyzed to
determine frequency, damping, and mode shape information. Many
excitation waveforms have been used for both ground vibration testing. The data logger provided manually controlled sweep sine function in range of 1 - 150 Hz with 0.1 Hz steps. . The exciters can operate in symmetrical or asymmetrical (180 dg. phase shift) mode.
The aircraft was completely assembled; in the cockpit a 70 kg weight
was installed. Wing water tanks were empty and control surface was
constrained.
The aircraft was lifted up on springs by hanging frame. The electrodynamics shakers were positioned under wing, 3 m away from wing
roots. The accelerometers were positioned on top of the wing, along
the wing spar in 1 m distance. They are connected on the wing by a
two-layer adhesive tape. The load cell was connected to the vacuum
cup and the combination was connected to the top of the stinger on
each shaker.
The structure was excited by manually swept sine force in symmetrical or asymmetrical mode in range of 5 - 100 Hz; excitation force
was set manually too. On the control PC screen most important information about structure response was displayed. The control software
provides real-time representation on waveform data; time response
(excitation force and response acceleration) and phase shift due to
Lissajous pattern. The control software displays data from all force &
acceleration channels together. The main shapes are computed in
postprocessor but quick assessment of main frequency can be made in
real-time by phase shift displays.
Analysis
The requirements for simplified flutter analysis are determined in [2].
It is necessary to determine first four symmetrical and first three
asymmetrical modes of the wing. The measured data was stored in PC
as recorded time waveform. The natural frequencies and bending shapes were computed with Sigview and LabVIEW SVT software by
Pic. 4 — 3D model of
the sailplane
Pic. 6. — Control software window
Results
The object of the test was to check out the system for GVT on real
aircraft structure and compare the results with reference values. For
comparison, the main frequencies tabled were measured. The main
shapes were computed too, but comparison is only qualitative. Reference values were given by Ground Vibration Test of HPH G-304 CZ
52
Pic. 7 — Symmetrical wing bending shapes
Tab. 2. Comparing of the resonant frequencies
by N. Neidbal, Germany. 1st symmetrical and 1st asymmetrical mode
were not measured because main frequency of the suspension was 4.4
Hz
References:
[1]
Niedbal N.: Aeroelasic Investigation of the Sailplane 304 CZ with
Winglets; Technical Report, HpH Kutna Hora, Steinhagen, 2000
[2]
Stender W., Kiessling F.: Aeroelastic Flutter Prevention in Gliders
and Small Aircraft; DLR-Mitteilung 91-03, Institut für Aeroelastik
der DLR, Göttingen, 1991
[3]
Kousal P.: Laboratorní vibrační zkoušky frekvenčních charakteristik větroňů a lehkých letadel; Technická zpráva, AIR CONSUL
ZLIN, Zlín 2000
Conclusion
The functionality of the System was verified. There are no significant
problem with the hardware, without measuring under 5 Hz. Excitation by manually selecting frequencies is acceptable but takes longtime. Used software applications are good for a few of data. Analyzing
the whole aircraft took a lot of time.
Czech Aerospace Research Centre — Workshop 2006: / Seminář CLKV 2006
Addendum:
Directory of Papers not Published in this Issue
Seznam nepublikovaných příspěvků
To bring a complete overview of this year's proceedings the following list sums up all other
papers presented at the 2006 conference that
could not be published in this issue due to capacity restriction.
Šmíd M., Hanzal V. ([email protected], [email protected]):
Flow Simulation near Spinning Unit Rotor
Simulace proudění v okolí rotoru spřádací jednotky
Chvojka M. ([email protected]):
Microaccelerometer Control Circuits Optimization
Optimalizace řídicích obvodů MAC
Klínek P. ([email protected]):
Aerodynamic Analysis of Wind Power Plant Blade
Aerodynamická analýza vrtulového listu větrné elektrárny
Dostál J., Klínek P. ([email protected], [email protected]):
Propeller Aerodynamic Analyses by Vortex-Lattice Method, Part 1: Basic theory
Metoda aerodynamického výpočtu vrtule pomocí vírové
mříže, část 1: Základní teorie
Reček J., Fabián J. ([email protected], [email protected]):
Structural Proposal of the Frame of a Small Satellite
Konstrukční návrh tělesa malé družice
Patočka K., Jamróz T., Had J. ([email protected],
[email protected], [email protected]):
Optimizing Composite Parts
Optimalizace kompositových částí
Růžička P. ([email protected]):
Optimalization and Measurement of Ultra-light Transmission Shafts
Konstrukční optimalizace a pevnostní měření ultralehkých
transmisních hřídelů
Průcha P. ([email protected]):
Loss of Stability Analysis of Composite-sandwich Panels
under Shear Load
Analýza ztráty stability kompositových-sendvičových panelů
zatížených smykem, analytický a MKP výpočet
Čenský T. ([email protected]):
Calibration Line for Anemometer Probes
Kalibrační trať pro anemometrické sondy
Hlinka J. ([email protected]):
Reliability Analysis — Aircraft vs. Spacecraft
Analýzy spolehlivosti-rozdíly v zajištění spolehlivosti u konvenčních letadel a kosmické techniky
Šplíchal J. ([email protected]):
FEM-simulation of Dynamic Phenomena in ARC Projects
MKP simulace dynamických jevů v rámci projektů CLKV
Klement J. ([email protected]):
Mechanical Properties of Fiber-metallic Reinforced Plastics Affected by Aging
Vliv stárnutí na mechanické vlastnosti vlákno-kovových
laminátů
Doupník P. ([email protected]):
CFD-optimization of Undercarriage Wheel-wells
Optimalizace podvozkových gondol letounů kategorie commuter prostředky CFD
obal32006CLKV.qxd
15.11.2006
11:56
StrÆnka 4
Composite Airplane Control
Rod with Metal End Joint
Figure 3 — Loading
machine INOVA ZUZ 200
and loading fixture
CZECH
AEROSPACE
P r o c e e d i n g s
J OU R N A L
F O R
C Z E C H
AE RO S PAC E
R E S E A R C H
Figure 2 — Specimen of the VL-3 plane
control rod and bolt test assembly
LETECK Ý
Figure 7 — Test specimen with first filament layout ripped off
zpravodaj
VÝZKUMNÝ A ZKUŠEBNÍ LETECKÝ ÚSTAV, a.s.
Editorial address:
Aeronautical Research and Test Institute / VZLÚ, Plc.
Beranových 130, 199 05 Prague 9, Letňany
Czech Republic
Phone.: +420-225 115 223, Fax: +420-869 20 518
Editor-in-Chief:
Editor & Litho:
Ladislav Vymětal (e-mail: [email protected])
Stanislav Dudek (e-mail: [email protected])
Editorial Board:
Chairman:
Vice-Chairman
Members:
Publisher:
Printing:
Milan Holl, President ALV, Managing Director VZLÚ
Vlastimil Havelka, ALV
Jan Bartoň, Tomáš Bělohradský, Vladimír Daněk, Jiří Fidranský,
Luboš Janko, Petr Kudrna, Pavel Kučera, Oldřich Matoušek,
Vojtěch Nejedlý, Zdeněk Pátek, Antonín Píštěk
Czech Aerospace Manufacturers Association / ALV, Prague
Studio Winter Ltd. Prague
Figure 1 — Angles of
attack of the rotor
blades, first geometry settings
Figure 2 — Angles of attack of
the rotor blades,
twisted duct
Published with the assistance of Czech Ministry of Education, Youth and Sports (MŠMT).
Subscription and ordering information available at the editorial address. Legal liability for published
manuscripts’ originality holds the author. Manuscripts contributed are not returned automatically to
authors unless otherwise agreed. Notes and rules for the authors are published at our Internet pages
http://www.vzlu.cz/.
Czech AEROSPACE Proceedings
Letecký zpravodaj
3/2006
© 2006 ALV / Association of Aviation Manufacturers, All rights reserved. No part of this publication may
be translated, reproduced, stored in a retrieval system or transmitted in any form or by any other means, electronic,
mechanical, photocopying, recording or otherwise without prior permission of the publisher.
ISSN 1211 - 877X
Figure 4 — Contours of tangential velocity immediately behind the
rotor stage
Figure 5 —
Angles of
attack of the
rotor blades,
twisted duct
with stator
vanes
obal32006CLKV.qxd
15.11.2006
11:56
StrÆnka 2
Contents / Obsah
Metel End Joint
The Aerodynamic Design
of Cold Jet
Preliminary Testing of Friction
Stir Welding
Optimization Methods for 2D Flow Passage
Design of Axial Compressor
Pokročilá detekce, izolace a přizpůsobení
chybných údajů od senzorů pomocí
umělé neuronové sítě
Experimentální zkušební systém pro
pyrolýzní vláknových termosetických
kompozitních materiálů
LETECK Ý
zpravodaj
In this issue:
Program SPAD a jeho užití při snižování
hluku vrtulových letadel
Czech Aerospace
Research Centre
Zkoušky kompositního táhla řízení
letounu s kovovou koncovkou
CLKV
Návrh aerodynamického řešení proudové
cesty studeného propulsoru
Proceedings of the
6th Annual
Workshop held at
Prague, Czech
Republic
Úvodní zkoušky frikčního
svařování
Hodnoticí metodika projektů
výzkumu a vývoje
November 2 to 3,
2006
Únavové zkoušky a výpočty životnosti
podvozku letounu VUT 100
Centrum leteckého
a kosmického
výzkumu
Optimalizační metody pro 2D proudění
axiálním kompresorem
Výroba kompositového krytu vrtulové
hlavy technologií LF
Optimalization of Stiffened Panel with the
Help of Mathematical Programming
Optimalizace vyztuženého potahu pomocí
matematického programování
Proceedings
Ověřovací frekvenční zkoušky na
kluzáku HPH G-304 CZ
Sborník vybraných
referátů
přednesených na
6. ročníku
semináře CLKV
Prague
2006
Praha, 2. — 3.
listopadu 2006
E
CK
Ý Ú
S
TA
V
© C Z E C H AE R O S PAC E
M A N U FAC T U R E R S A S S O C IAT I O N
CLKV
V
U
O
Accomodation of Sensor Failures by
Means of Artificial Neural Network
Měření letových veličin
v akrobatických manévrech
CZECH
AEROSPACE
T
Measurement of Aerobatic
Flight Characteristics
CFD simulace kvality mikroklima
v kabinách malých dopravních letounů
LE
Slovo úvodem k semináři CLKV 2006
N o v e m b e r
Introductory Lecture to the ARC 2006
2 0 0 6
ISSN 1211—877X
T
B
RN
No. 3 / 2006

czech aerospace - Výzkumný a zkušební letecký ústav

Transkript

Podobné dokumenty

Tchaj-wan

No. 07100 - Woodcomp

czech aerospace - Výzkumný a zkušební letecký ústav

Here - Technical University of Liberec