Aerospace and Electronic Systems Magazine July 2018 - 6

EGT Prediction
stream turbines have to operate with a lower mass flow and if the
speed is kept constant, the turbine has to produce the same amount
of work required by the compressor. This results in a higher turbine
entry temperature and therefore higher EGT [21]. In addition, a
higher ambient temperature increases EGT, but higher mechanical
efficiency results in lower EGT [21].
Rahman et al. [22] did a parametric study of thermodynamic
performance on a gas turbine power plant. They investigated variation of operating conditions, such as turbine inlet and exhaust temperature on gas turbine performance. They found that gas turbine
performance is affected by component efficiencies and turbine
working temperature. They concluded that in general for every 56°
C increase in temperature, work output increases approximately
10%, efficiency increases about 1.5% [22].
Based on previous studies [20]-[22], EGT and its effect on the
performance of gas turbine engines is a worthy area of investigation.
According to the abovementioned research, studying the thermodynamic and gas dynamic parameters of gas turbine engine can
be a very useful research activity field, especially considering the
high capability of ANN and MPR data mining methods for investigating the performance parameters of gas turbine engine.
Thus, the purpose of the present work is investigating EGT
variations as the output parameter of variations of engine rotational speed in an AMT Olympus HP E-Start gas turbine engine,
using ANN and MPR data mining methods. For this purpose, correspondence between predicted and measured values of EGT was
quantified by calculating Root Mean Square of Error (RMSE) and
coefficients of determination (R2) between predicted and measured
values. Based on the calculated root mean squares of errors and
values of coefficients of determination, prediction capability of
these two data mining approaches (i.e., ANN and MPR) was compared. Also, EGT is studied in the experimental work because it is
a key parameter in a gas turbine engine performance, so investigating and analyzing the relation between EGT and other parameters
especially engine rotational speed can be very useful and can give
a better perception from engine performance.
To clarify the distinctions between the present work and the
past ones, it should be said that based on the author's knowledge
no one has done any research on gas turbine engine using both statistical and artificial intelligence together. Since these methods are
both capable of predicting the relation between different parameters, these methods are compared with each other in this study to
show which one is better. In addition, since the first step in the engineering design process is to take the advantage of past successful
design and manufacturing experiences in the use of statistical data,
the comparison between these two data mining methods could help
engineers choose the better or the best method for other activities.

EXPERIMENTAL APPARATUS
The AMT Olympus HP E-Start (Olympus HP) gas turbine engine,
which is widely investigated in other related researches [2]-[4] is
used. The schematic of the engine along with the details are shown
in Figure 1. The engine is used for experimental and research activities in different research centers and universities and is capable
of producing a maximum thrust of 230 Newton [11]. The Olympus
6

Figure 1.

Schematics of the used experimental micro gas turbine engine [23].

HP has a single radial compressor and an axial flow turbine. The
Olympus HP owes much of its excellent performance and superb
power/weight ratio to its turbine wheel, which is designed by AMT
staff. The time required for the turbojet to spool up and down is
also positively influenced by the low mass of the axial turbine
wheel, taking less than 4 seconds to reach from minimum to maximum rotations per minute (RPM) and only 4 seconds to reach from
maximum to minimum RPM. The combustion chamber is of the
annular type, which is fitted with a unique low pressure fuel system
[23]. Both the front and the rear hybrid bearings are also lubricated
and cooled by the fuel system, and therefore the motor requires
no separate lubrication system or oil tank. The engine is protected
from misuse and accidental damage by means of an electronic
control unit (ECU) which regulates the maximum performance
within programmed software limits. The ECU is fully automatic
and needs no adjustment by the operator. The ECU controlled by
a microprocessor, which is powered by a fuel pump battery. The
control unit has two inputs, called throttle and switch channel that
can be changed by the operator. Other components like fuel pump
and control valves are also controlled by the ECU [23]. The ECU
also has inputs for the EGT and the RPM of the motor. These are in
place to make sure that the motor can't exceed the maximum design
RPM or EGT. The fuel pump of the engine is a type built from two
gear-wheels running in a high-precision chamber. Therefore, using
absolutely clean and pure fuel, in order to prevent blockage in the
fuel system, is very important. Fuel used in the engine is Kerosene.
The Olympus HP also uses the fuel for lubrication, so the fuel must
be premixed with 4.5% Aeroshell 500 turbine oil before use. This
oil takes care of the lubrication during start-up and power-down
sequences. When a power-down is activated, the fuel flow stops
and the fuel will vaporize in the hot turbine. Remaining oil also
lubricates the turbine during the next start-up sequence [23]. The
ignition system is developed around a "Webasto" 12-volt ceramic
igniter with the "Webasto part #84906B". Under normal condi-

IEEE A&E SYSTEMS MAGAZINE

JULY 2018

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine July 2018