The 2nd International Conference on Computer Application and System Modeling (2012)

A Method for Correcting The Error in Indicated Normal Acceleration Due to G-

Sensor Location

Mi Yi Chen Mingtai
Shanghai Aircraft Design and Research Institute Shanghai Aircraft Design and Research Institute

Abstract-Particularly for structural purposes, it is important to aircraft described by the

 body axes system consists of three
obtain accurate measurements of the normal acceleration at linear
 velocities
 ( u , v and w ) and three
 angular velocities
the Center of Gravity of an aircraft. For most airliners and ( p , q and r ). The linear velocity u , isdefined as the
fighters, the g-sensors are located away from the Center of component of the resultant linear velocity VT , which lies in

Gravity (C.G.). Theoretical mechanics shows that, depending the X direction. Similarly, the rolling velocity p is defined
on the distance between the g-sensor and the C.G., aircraft as the component of the total angular velocity ω , which is
angular rates and accelerations will induce a difference about the X axis. Positive directions are: X , forward along
between the actual normal acceleration at the C.G. and the
the fuselage; Y , out the right wing; Z , down through the
normal acceleration indicated by the g-sensor. A method was
devised to correct the normal acceleration indicated by the g-
bottom of the aircraft [1].
sensor so as to obtain the actual normal acceleration at the B. Acceleration
aircraft C.G. Based on the method presented in this paper, Acceleration is a vector quantity, and its unit is
tests were conducted on the B737-800 simulator and the traditionally referred to as g (load factor), because it
corrections determined by the method were compared with the indicates the apparent acceleration of gravity sensed on
test data obtained from the simulator. board. Normal acceleration (Nz) is the component of the
linear acceleration of an aircraft along the Z axis. A normal
Keywords-Normal Acceleration, Correction, g-sensor, C.G., acceleration of one, or 1 g, represents conditions in straight
Aircraft and level flight, where the lift is equal to the weight.
Assume that the acceleration obtained by a g-sensor
I. INTRODUCTION 
located away from the C.G. position is ag − sensor which
Normally, in order to obtain the accurate normal
acceleration of the aircraft, g-sensors should be located as contains three components along three axes:
   
close to the Center of Gravity (C.G.) as possible. However, ag − sensor = ag − sensor . x i + ag − sensor . y j + ag − sensor . z k
for most airliners and fighters, g-sensors are usually located (1)
in the Electrical/Electronic cabinet as components of the 
Attitude Heading Reference System (AHRS) or the Inertial Assume that the acceleration at C.G. position is aC .G . :
   
Navigation System (INS). Furthermore, the C.G. position of aC .G. = aC .G. x i + aC .G . y j + aC .G. z k (2)
the aircraft changes in flight due to fuel consumption. It is 
Define that the correction of acceleration is Δa :
important to determine the correction of the acceleration    
indicated by a g-sensor located away from the C.G. of the Δa = Δax i + Δa y j + Δaz k (3)
aircraft. Compared with the accelerations along X and Y The acceleration at the position of the g-sensor is equal
axes, the normal acceleration is more important due to its to the sum of the acceleration at C.G. position and the
effect on structural design. correction caused by the g-sensor location away from the
C.G.. As a component of the linear acceleration of the
II. DEFINITIONS aircraft, normal acceleration is:
A. Body Axes ag − sensor . z = aC .G . z + Δaz (4)
Figure.1 shows the body axis notation and the velocities
along and about each of the axes. III. REDUCTION

Figure 1. Body Axis System

X is along the longitudinal axis, Y is along the lateral Figure 2. Location of g-sensor
axis and Z is along the normal axis. The motion of the

Published by Atlantis Press, Paris, France.

© the authors
0531
 The 2nd International Conference on Computer Application and System Modeling (2012)


Assume that the aircraft is at rest and level. The  d R  
location of the g-sensor is shown in Figure.2. ω× = ω × ΔV
dt
The C.G. position is ( x , y , z ) . Here, the C.G. is the   
0 0 0
i j k
origin of the body axes, so,
= p q r
( x0 , y0 , z0 ) = (0, 0, 0) (5)
Δu Δv Δw
Assume that the g-sensor location is ( xg , yg , z g ) , so the 
= (q ⋅ Δw − r ⋅ Δv)i (12)
distance between the C.G. and the g-sensor 
 + (r ⋅ Δu − p ⋅ Δw) j
is R : ( Rx , Ry , Rz ) . 
+ ( p ⋅ Δv − q ⋅ Δu )k
Rx = xg − x0 = xg
The resultant acceleration is the sum of tangential
R y = y g − y0 = y g (6) acceleration and radial acceleration.
    
Rz = z g − z0 = z g  d ΔV d (ω × R) d ω   d R
Δa = = = × R +ω×
The acceleration of the g-sensor consists of the dt dt dt dt

acceleration at the C.G. and the correction caused by the g- = (qR  y ) + (q ⋅ Δw − r ⋅ Δv)  i
 z − rR (13)
sensor location away from the C.G. Here, only the 
correction was considered in following reduction. + [ (rR  z ) + (r ⋅ Δu − p ⋅ Δw)] j
 x − pR

A. Linear Velocity of g-sensor + ( pR  x ) + ( p ⋅ Δv − q ⋅ Δu )  k
 y − qR
Assume that the airplane has an instantaneous angular
 The resultant normal acceleration is:
velocity ω ( p, q, r ) . The linear velocity [2] of g-sensor due to   
Δaz = ( pR  y − qR  x )k + ( p ⋅ Δv − q ⋅ Δu )k (14)
the arm and rotation around C.G. is:
   According to equation(8):
i j k Δaz = ( pR
 y − qR
 x)
  
ΔV = ω × R = p q r (15)
(7) +  p (rRx − pRz ) − q ( qRz − rRy ) 
Rx Ry Rz
   Therefore, the correction caused by the location is:
= (qRz − rRy )i + (rRx − pRz ) j + ( pRy − qRx )k Δaz = ( pr − q ) Rx
(16)
The three components of linear velocity are: + ( p + qr ) Ry − ( p 2 + q 2 ) Rz
Δu = qRz − rRy
Δaz is the acceleration increment caused by the g-
Δv = rRx − pRz (8) sensor location away from the C.G. of the aircraft and it
Δw = pRy − qRx must be subtracted from the normal acceleration obtained
B. Acceleration Caused by the Location of the g-sensor from the g-sensor to obtain the normal acceleration at the
The resulting linear acceleration due to the location of C.G..
the g-sensor is: IV. INFLUENCING FACTORS ANALYSIS
  
 d ΔV d (ω × R )
Δa = = Based on the model presented above, the correction is
dt dt determined by the distance between the g-sensor and the
  (9)
dω 
  dR C.G. and the angular acceleration and angular velocity of
= × R +ω× the aircraft. Tests were conducted on the B737-800
dt dt
The first term is tangential acceleration and the second simulator which gave the normal acceleration, pitch rate,
term is radial acceleration. Firstly, the angular velocity of yaw rate and roll rate of the aircraft. For the purpose of
 verifying the model, the aircraft was programmed to
the aircraft is ω ( p, q, r ) , so the differential of angular
velocity can be written as: perform an unusual manoeuver. In this manoeuver, the yaw
 rate and roll rate were as large as the pitch rate [3]. The
dω pitch rate, yaw rate, roll rate and normal acceleration at C.G.
= ( p , q , r) (10)
dt position are shown in Figure.3.
Then, the tangential acceleration is: A. Distance Effect
   In order to verify the effect of the distance between the
 i j k
dω 
 g-sensor location and the C.G. position, assume that there
× R = p q r are two g-sensors located at 10meters and 15meters along
dt (11)
Rx Ry Rz the X axis, respectively. Based on the pitch rate, yaw rate,
   roll rate and normal acceleration presented above, the
= (qR
 z − rR
 y )i + (rR x − pR z ) j + ( pR  y − qR
 x )k correction to the normal acceleration indicated by the two g-
Secondly, the radial acceleration is: sensors was calculated and is shown in Figure.4.

Published by Atlantis Press, Paris, France.

© the authors
0532
 The 2nd International Conference on Computer Application and System Modeling (2012)

Observation of Figure.4 shows that, when the distance According to equation (16):
between the g-sensor and the C.G. was 10meters, the az = ( pr − q ) Rx + ( p + qr ) Ry − ( p 2 + q 2 ) Rz (18)
maximum correction of normal acceleration was 0.32g, and
the minimum correction was -0.36g. When the distance was So, in this case, az can be simplified as:
changed to 15meters, the maximum correction of normal az = ( pr − q ) Rx − ( p 2 + q 2 ) Rz (19)
acceleration was 0.66g, and the minimum correction was - Furthermore, for an airliner, compared with the pitch
0.67g. Without changing angular velocity, the correction of rate and pitch acceleration, the roll and yaw rate are small in
normal acceleration changed substantially due to the normal flight, so they can be neglected, therefore:
distance variation. It can be concluded that the correction of pr ≈ 0
normal acceleration increases with the distance between the (20)
g-sensor and the C.G. while keeping other variables p2 ≈ 0
constant. Finally, equation (16) can be simplified as:
B. Angular Velocity and Angular Acceleration Effects az = −qR  x − q 2 Rz (21)
The time history in Figure.5 shows the variation of the
normal acceleration correction with pitch rate and pitch Consequently, just three variables are needed to make
acceleration. the correction, as follows: pitch rate, q , and the distance
As shown in Figure.5, the correction of normal between the g-sensor and the C.G. Rx and Rz (the pitch
acceleration has a relationship with angular velocity and acceleration q can be calculated from q ).
angular acceleration. When the angular acceleration is small
(which means the angular velocity changes slowly), the VI. SUMMARY
correction of normal acceleration was nearly zero g which
can be neglected in practical flight. However, when the It has been shown that the correction of the normal
angular acceleration is large (which means the angular acceleration as indicated by an “away from the C.G.” g-
velocity changes rapidly), the correction of normal sensor is determined by the distance between the g-sensor
acceleration was 0.36g which cannot be neglected in and the C.G. position, angular velocity and angular
practical flight. It can be concluded that when the angular acceleration. The correction model given in this paper was
velocity changes more rapidly (the angular acceleration is not tested on an actual aircraft due to time constraints and
large), the correction of normal acceleration will be the availability of test aircraft. Validating tests should be
substantial while keeping the distance variables constant. conducted using test aircraft.

V. FURTHER RESEARCH REFERENCES

According to the model, there are eight parameters [1] Dr. Robert C. Nelson. Flight Stability and Automatic Control [M].
McGraw-Hill Book Company, 1989.
required to determine the correction of normal [2] JIANG Yong-zheng, Simulator Motion System and Software Dissect,
acceleration: Rx , Ry , Rz , p, q, r , p , q . However, based on the Journal of UEST of China [J], 2002,8, PP374-378.
specific characteristics of large aircraft such as airliners, the [3] YOU Jun-sheng; YOU Yong; XU Xu-yao; MA Yue-xin. Research for
Modeling and Simulation of Pilot Three-shaft Acceleration, Journal
model can be simplified. Here, assume that the g-sensor is
of System Simulation [J], 2006.8, PP28-30.
located close to the X-Y plane, then,
Rx ≠ 0
Ry ≈ 0 (17)
Rz ≠ 0

Published by Atlantis Press, Paris, France.

© the authors
0533
The 2nd International Conference on Computer Application and System Modeling (2012)

Figure 3. Variation of Velocity and Normal Acceleration with Time

Figure 4. Nz Correction with Time

Figure 5. Nz Correction due to Pitch Rate and Pitch Acceleration

Published by Atlantis Press, Paris, France.

© the authors
0534