Chapitre précédent
Chapter 041

virtually powered by


AUTOPILOT ENGAGED

OK, let's set aside for a while photography, vacation, French, seasons, recipies, songs, and crazy comments, to focus on the most important issue of the very moment (well, at least as far as I'm concerned) : «How aircrafts' autopilot works ?»

Why the («--beep--») do I deal with that here and now?
Just because I spent my whole week preparing for a major job interview at CEA, for a "Flight Simulator Autopilot Systems Engineer" position.
Thus, as I definitely can't rely on my C++ programming skills
(how would A380 pilots appreciate a "Hello World!" message on their Head Up Display?), I'd rather focus on my avionics knowledge. And of course, the best way to ensure I've understood something is to try and teach it! So here we go!

Well, I appologize in advance to those of yours who definitely don't give a inch of care about these messy technical & materialistic concerns...


Chapitre suivant

 


1. AUTOPILOT PURPOSE :

Basically, autopilot aims to help flight crew in 3 main ways :
- maintaining aircraft's attitude
- following a given flight track
- performing approach and landing

From a Beechcraft Bonanza to an Airbus flight panel, the aim is the same, but the amount of possible modes (and their complexity) deeply changes :

 

Beechcraft Bonanza F33 F-BNEC and it's autopilot switches

 

AIRBUS A320 cockpit and autopilot panel

 

 

2. DIFFERENT MODES :

So, from simply lightening a pilot's workload, to a complete automatic landing management in the fog, autopilot has many functions (called "modes"), between which the flight crew can select the appropriate(s) :

BASIC MODES :
- Maintaining General Attitude (ATT)
- Maintaining a pre-set Roll Angle
- Following a given Heading (HDG)

UPPER MODES :
- Maintaining Altitude (ALT)
- Acquiring Altitude (ACQ)
- Maintaining Speed (SPD) through surfaces and / or Autothrottle
- Radioelectric track slaving (VOR)
- Way-points track slaving (WPT)

COMMON MODES :
- Automatic approach (APP, APR or ILS)
- Automatic landing (LAND)

 

Airbus FCU (Flight Control Unit)

 

 

3. ACTUATING :

Autopilot controls the aircraft just as a pilot would :
- through surfaces actuators (on pitch, roll and yaw axis)
- through engine levers (with autothrust)

It can be quickly disengaged with a switch, or simply by firmly moving the stick. Any of the following event can also disengage the autopilot :
- electrical failure detection
- hydrolic pressure failure
- vertical gyro failure
- roll angle > 35°
- yaw damper out of order
- load factor > a given threshold

When disengaged, the autopilot returns in simple monitoring mode, called "synchronisation mode". Thus, each surface position is recorded, in order to prevent any violent step when it will take back control of the aircraft.

 

 

4. CONTROL LOOPS :

Autopilot actions depend on control laws, which are specific to each mode, and are always based on imbricated slaving loops.
Here is a diagram which shows how a parameter (roll angle for example) can be controlled by a slaving loop.


Of course, a really more complex derivative / intergral comparator is used in real models to improve both response time and precision.
The same kind of control loop manages also yaw and pitch parameters. Thus a 3 loops parallel structure, known as "surface loop", must be controlled by the AP computer with a global procedure.

 

 

5. COMPUTING PROCEDURE :

The AP computer determines the right movements to perform using the basic following procedure : (example given for NAV mode)

1. Given a cross-track (angle between heading set-point and real aircraft heading), the AP computes the heading to reach. That's the track control law.

2. The AP computes the roll parameters needed to turn in order to reach the right heading. That's the heading control law.

3. Then the AP computes the surface actuators signals to ensure a proper turn previously defined, which implies, in the following order :
- roll parameters to properly adapt heading
- yaw compensation to keep the track-ball centered (no sideslip)
- pitch compensation to prevent any altitude loss
- throttle adjustment to compensate for rise in drag

4. Actuation of surfaces and throttle using servo-units, and depending on their own former position as acquired through a constant monitoring procedure.

In fact, it exactly reproduces with computed algorithms what any (good) pilot naturally does to follows his navigation plan.

 


6. HIGHER SCALE INTEGRATION :

The "surface loop" described above is in fact only the more inner loop of a complete system, headed by the AP computer through various level control laws. The following diagram illustrates this loop integration for the case of NAV mode autopilot, also called VOR/LOC, INS :


 

7. AUTOLAND EXEMPLE :

The following scheme illustrates how autopilot handles the whole automatic landing procedure. You can see, in the bottom of the picture, how and when each main parameter is computed.
And you should now understand how inner and outter loops are used to securely and smoothly bring the plane down to runway tarmac.
The set-points are the following ones :
- LOC axis (radioelectric axis aligned with the runway)
- GLIDE axis (radioelectric axis defining a 3° descend plan)
- preset SPEED (in accordance with ATC parameters)

Note that in approach phase, jet engines always need to be computer controlled, because of power instability.

 

To ensure a high security level, autopilots are often duplicated, or even triplicated. This redundency allows cross-checks and inhibition of one defaulting AP.
Moreover, approach and autoland aircraft capabilities are ranked in different precision levels, as determined by the Decision Height (DH), where the pilot has to chose if he can complete the procedure by himself (visibility suitable), or if a go-around is necessary.
It's called CAT I, CAT II, CAT IIIA ("see to land"), CAT IIIB ("see to taxi") and CAT IIIC ("see if you haven't forgotten anything under you seat!").
Autoland (CAT IIIB and IIIC) is the ultimate capability, but only few airports have the required device to manage it.

 

 

 

8. LET'S HAVE A BREAK!

Congratulations you brave fellows! If you've really read and understood everything until now (end of this 1st part), you definitively deserve a little break. ('cause so do I ! )
Enjoy the comic strip... and fly directly to http://www.lapin.org to have more, much more.

http://www.lapin.org

 

 

9. AUTOFLIGHT SYSTEM

Let's get back to our lesson.
Autopilot, as described bellow, is in fact only one element of a much wider system call AFS (AutoFlight System), or even AFCS (Automatic Flight Control System), which provides flight crew a fully integrated solution to manage the flight.
So AFCS not only handles autopilot functions, but also :
- aircraft stabilisation
- surfaces autotrim (to equilibrate them)
- automaticaly keeps the aircraft in an authorized flight domain
   (in order to prevent overspeed and stall)
- climb or cruise trajectory optimisation

Thus AFS also includes Autothrottle (through FADEC) and Flight Director (FD) management.

 

 

 

10. VISUALIZATION

Pilot and copilot can observe AP modes and AFS behavior directly on their respective Primary Flight Display (PFD) by selecting the Auto Flight System status screen, as shown below.

In addition to flight and AP parameters, Flight Director (FD) bars appear in that screen. Basicaly, FD and AP share the very same computing device.
FD role is to give the pilot necessary information to meet or keep a preset situation. This information is simply given by the position of the 2 yellow bars.
Thus, FD can be seen as an AP from which surface loop and actuation device have been unplugged. It just suggests to the pilot courses of action, based on AP computing, instead of doing it automaticaly.
(In the above photo, it is suggested that the pilot descend a bit and make a little right turn to center the FD bars)

 

 

11. AVIONICS INTEGRATION

Since the autoflight system is a fully communicating system, avionics integration is fundamental to allow each device to exchange information between each other. In addition to a large amount of sensors and other device, the main FMS (Flight Management System) architecture is the following one :

Flight Management and Guidance Computers (FMGC) are the core of the whole FMS. Autopilot computers are inside them.
Flight Control Unit (FCU) and Navigation Display Control Panels allow flight crew to set desired parameters to manage the flight.
More complicated data, such as way-points coordonates, must be selected on MCDU (Multipurpose Control and Display Units).
FMGCs compute all navigation data (and much more), and manage aircraft controls and engines, though a specific computer, depending on flight crew wishes.
Appropriate information are displayed on respective PFD and ND (main screens pilots should watch), and on MCDUs screen.

 

 

12. HIGH-TECH RUSSIAN DOLLS

To put it into a nutshell, the whole FMS system is a multi-level, highly integrated system, designed to perform a very complex flight management with the highest security level.
Autopilot is only a part of it. But yet a rather complex one! It is mainly due to :
- the large amount of I/O parameters to handle
- specific and optimised control laws algotithms


But think I'd better stop it here.
You and I have had enough for the moment.

I hope I can soon bother you much more about those concerns in a couple of weeks from now. Anyway, I now feel ready to show CAE managers my motivation and knowledge in this field!
That was the one, and really only, aim of this chapter, wasn't it?

«AUTOPILOT DISENGAGED»

 

 

Chapitre précédent

Table des matières
Chapitre suivant