The Development of Control Engineering in Britain
and the Cambridge Contribution
Malcolm C. Smith
At Caius there are two Engineering Fellows who specialise
in control engineering (GV and MCS). The latter was recently asked
to prepare an article for the American Control Systems Society
(IEEE) on educational practice in the Britain in the area of control,
as part of a survey of several countries around the world. In
fact the U.K. has been quite strong in this area historically,
so it was quite nice for it to be represented in the survey.
A few extracts from the article are sewn together below, specially
abridged and prepared for the Caius Engineer at the request of
the editor (Dr T.P. Bligh).
Historically, applications of feedback control date
back to antiquity (e.g. Egyptian water clocks), though one of
the most famous examples in technology is the centrifugal governor
for steam engines invented by the Scottish engineer James Watt.
It was this application that brought about the birth of control
theory in a paper for the Royal Society of London by James Clerk
Maxwell---a Scot, a Cambridge man, and perhaps most famous for
his equations of electromagnetics. Another Cambridge man, who
was contemporary with Maxwell, Edward J. Routh, was responsible
for an important stability criterion which bears his name and
which is taught in control courses all over the world to this
day (see the end of the article for an anecdote on the Tripos
performance of Maxwell and Routh). Routh was one of the most
famous and successful coaches at Cambridge in the 19th century,
being a private tutor to more than six hundred pupils between
1855 and 1888, and preparing the Senior Wrangler (top mathematics
student) on 27 occasions---an unparalleled feat. Routh was himself
Senior Wrangler in 1854.
It seems that the stability analysis of governors
and related mechanisms was taught from this time at Cambridge,
at first in the mathematical tripos and later on in mechanical
sciences (after the beginning of this tripos in 1894). The question
below is the last of 10 on the "Heat Engines and Dynamics
of Machines'' examination paper in the Cambridge Mechanical Sciences
Tripos part II of 1897 --- students were instructed to answer
not more than eight (!) such questions in three hours:
"Investigate the equations of motion of the
governing masses in some form of Watt's centrifugal governor,
and find an expression for the period of the oscillations about
the state of steady motion. Point out in what circumstances the
oscillations will degenerate into hunting, and shew how
this fault may be prevented.''
Questions related to regulation, control, and later
feedback in amplifiers, continued to be set occasionally for Tripos
exams on the initiative of individual lecturers. However, the
treatment and appreciation of automatic control as a subject in
its own right did not begin until after World War II.
The emergence of Control Engineering as a discipline,
and the realisation that there were common techniques and methods
which could be used in a wide variety of applications (ranging
from control of power plants, chemical processes, to aircraft
stability, engines, even applications in economics, biology),
happened in the late 1930s and 1940s. A substantial part of what
came to be called "classical control'' originated in the
work on feedback amplifiers at the Bell Laboratories in America
in the late 1920s and 1930s. Experience of engineers in Britain
involved in key war-time applications (such as radar fire control)
helped to establish a strong basis of expertise in the area. In
1942, a very influential committee was formed to coordinate work
and to facilitate an exchange of information: the Interdepartmental
Committee on Servomechanisms (the Servo Panel). Several members
of this committee played an important role in disseminating control
system design techniques after the war and in helping universities
to establish courses. Due partly to the Servo Panel, Britain was
the first country to hold a major international conference on
automatic control (the Cranfield Conference of 1951).
An influential figure in Britain at this time was
Arnold Tustin, who worked for Metropolitan Vickers until 1945
and became Professor at Birmingham University (1947) and later
at Imperial College (1955). In a lecture to the Servo Panel in
1942, Tustin described the design methods developed at Metro-Vick
for gun control, which used the frequency response locus. The
connection with the Nyquist stability criterion and work going
on in America was later drawn to Tustin's attention by P.J. Daniell
of the Mathematics Department at the University of Sheffield.
(Incidentally Daniell was one of several people to independently
develop the describing function method in the 1940s---this is
one of the topics taught in the I3 module in the fourth year of
the Engineering Tripos). Another influential member of the Servo
Panel was A.L. Whiteley who developed the use of inverse Nyquist
diagrams for feedback design in 1943 at British Thomson-Houston
Company (in parallel with H.T. Marcy in America). It seems that
the experience gained by early British designers in the use of
frequency response methods had a strong effect on the subsequent
development of control in Britain. It is possible that this influence
extends to later research on multivariable control design and
H-infinity control.
Probably the first set of lectures on automatic control
engineering (as a subject in its own right) at Cambridge were
given by G.D.S. MacLellan (Pembroke) in 1946-7. By the late 1940s,
the basic theory of closed loop systems was being given in a course
on "mechanics of machines'', taken by all engineering undergraduates,
and was treated as an extension of the elementary theory of engine
governors. For those who went on to Part II of the mechanical
sciences tripos (roughly the top 20% of students) there were more
advanced lectures, for example, on the applications of control
to the governing of prime movers (for mechanical engineers) and
the theoretical background common to servo-mechanisms and feedback
amplifiers (for electrical engineers). The principle of teaching
the basics of classical control to all engineering undergraduates
in their second year has continued until the present day, with
the material being moved from the mechanics paper to a new information
engineering paper in 1986. The teaching of control in the third
year underwent a gradual unification (from separate mechanical/electrical
treatments) which was fully achieved in 1987. This principle
has continued after the introduction of the four-year M.Eng course
in 1992.
The control research group at Cambridge was established
in 1947 by R.H. Macmillan and was later expanded by J.F. Coales
after 1952. A post-graduate course in control engineering was
started in 1954 which was intended mainly for graduates who had
spent some time in industry. This course continued until it was
replaced by the M.Phil in control engineering and operational
research in 1977, which itself continued until 1986. Research
by the control group was carried out first in the areas of mechanical
control systems, and later in nonlinear and optimal control with
A.T. Fuller, and applications to industrial processes. After 1974,
multivariable frequency response methods became a prominent research
theme with the appointment of A.G.J. MacFarlane. From the mid
1980s the group was recognized for work in the major new approach
of H-infinity control with the work of Professor K. Glover now
the Deputy Head of Engineering (Research) at Cambridge. These
techniques were recently implemented successfully on a Harrier
aircraft at Bedford with the control systems design work being
carried out at Cambridge by R. Hyde and K. Glover.
To conclude, here is the promised anecdote on Maxwell
and Routh. It was already noted above that Routh was Senior Wrangler
in 1854. In fact Maxwell sat for the mathematical tripos in the
same year and was placed second. Apparently Maxwell was so confident
of achieving first place himself that he did not trouble to rise
early to hear the list of successful candidates read out in the
Senate House, but sent his servant instead. (Interesting, isn't
it, that undergraduates had servants in those days!) On his return,
Maxwell is said to have enquired, "Well, tell me who's second!'',
and was somewhat taken aback to receive the reply, "You are,
sir''.